Differential Expression and Phenotypic Activities of Osterix in Equine Mesenchymal Stem Cell Populations

DIFFERENTIAL EXPRESSION AND PHENOTYPIC ACTIVITIES OF OSTERIX IN EQUINE MESENCHYMAL STEM CELL POPULATIONS

KALYN K. HERZOG

DISSERTATION

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in VMS – Comparative Biosciences in the Graduate College of the University of Illinois at Urbana-Champaign, 2018

Urbana, Illinois

Doctoral Committee:

Professor Matthew C. Stewart, Chair Professor Jodi A. Flaws Associate Professor Prabhakara P. Reddi Clinical Professor Brian M. Aldridge

ABSTRACT

Bone is one of the few tissues in vertebrates that possess the innate capacity for authentic regeneration, yet approximately 10% of fractures fail to heal adequately. There is considerable overlap in the cellular and molecular events occurring during embryological bone development and fracture repair. Mesenchymal stem cell (MSC) precursors undergo osteogenesis to give rise to osteoblasts, which are responsible for bone matrix synthesis and mineralization. Osteogenesis is regulated by two master transcription factors, runt-related transcription factor 2 (RUNX2) and Osterix (OSX), with interplay from signaling pathways such as the Bone Morphogenetic Protein (BMP) network. MSCs isolated from bone marrow (BM) and adipose tissue (ADI) are most commonly utilized in human and veterinary regenerative medicine applications. BM-MSCs have shown superior osteogenic capacity to ADI-MSCs in many studies, yet sample collection and culture techniques often vary between species. The overall objective of this dissertation was to determine factors that influence the in vitro osteogenic capacity differences of equine adult ADI- and BM-MSCs in order to standardize techniques and provide insights into avenues for improving osteogenesis in less-osteogenic MSC populations. The first study investigated the hypotheses that in vitro cell viability, proliferation, and osteogenic differentiation of MSCs collected from horses under general anesthesia and after euthanasia would be impaired compared to MSCs collected from horses during sedation. The second set of experiments characterized the RUNX2 and OSX response axes of ADI- and BM-MSCs during osteogenesis to test the hypotheses that 1) the relative osteogenic capacities are consistent with resting and inducible RUNX2 and OSX transcript levels, and 2) increasing OSX expression in ADI-MSCs improves their osteogenic capacity. The third series of experiments evaluated the comparative BMP signaling activities in ADI- and BM-MSCs to determine whether 1) osteogenic BMP ligand expression differs, 2) BMP inhibition in BM-MSCs impairs osteogenesis, and 3) BMP supplementation to ADI-MSCs improves osteogenesis. The results indicate that donor status during MSC collection does not affect in vitro cell viability, proliferation, and osteogenic capacity. BM-MSCs have higher OSX expression and intrinsic BMP signaling in both resting and osteogenic conditions than ADI-MSCs. Endogenous BMP activity is necessary for osteogenesis in BM-MSCs, and exogenous BMP supplementation, but not OSX over-expression, was sufficient to improve the osteogenic capacity of ADI-MSCs. In summary, OSX expression and BMP signaling significantly influence the in vitro osteogenic capacity of MSC populations. ii

To Mike

iii

ACKNOWLEDGEMENTS

There are numerous people and groups I would like to thank that have greatly influenced my educational and career journeys thus far and provided invaluable support along the way. To my ‘horse family’, George and Karen Favorite, Andy and Mary Lindblad, and many others; thank you for welcoming me into your lives 25 years ago and sharing your incredible horses, starting with Shotgun, my first love, through Junior Mint and Merlin, our wedding steeds. Without them, I may have never aspired to become an equine veterinarian devoted to research. Thank you to the veterinary professionals that allowed me to volunteer and work for them, especially Dr. Roberta Drell at Morton Grove Animal Hospital, Drs. Margaret MacHarg and Jon Vacek at Kendall Road Equine Hospital, and the amazing technicians and staff I interacted with along the way. Your mentorship was invaluable into shaping my veterinary education and subsequent career choices. I would also like to acknowledge all of the faculty, house officers, staff, and students at the University of Illinois Veterinary Teaching Hospital that served as colleagues during my veterinary internship training in the Large Animal Clinic. I was fortunate to gain an interest in biomedical research during my undergraduate training in Dr. David Miller’s lab under the guidance of Dr. Ana Vieira. I would like to acknowledge the other groups I had the pleasure of working with, especially the labs of Drs. Dale “Buck” Hales and Janice Bahr, Dr. Indrani Bagchi, Dr. Stuart Clark-Price, and the Summer Research Training Program with Dr. Lois Hoyer. My veterinary internship research project with my PhD advisor, Dr. Matthew Stewart, greatly influenced my decision to re-pursue a graduate degree. Thank you to the Animal Sciences department and my initial research advisor, Dr. Matthew Wheeler, for giving me the opportunity to start a PhD program, even though I chose to leave. I am indebted to the Comparative Biosciences and Veterinary Clinical Medicine departments and Dr. Stewart for providing the support and guidance needed to complete my PhD degree. Lastly, thank you to Holly Pondenis, Dr. Sushmitha Durgam, Dr. Annette McCoy, and my committee members for the technical assistance, advice, and moral support over the years. None of my accomplishments would matter or have been possible without the unwavering love, support, and encouragement of my husband, mom, and family members. I am eternally grateful for the amazing life you have all helped me build and look forward to an even better future with you. Finally, thank you to my past and current pets, all of whom have kept me afloat countless times by always reminding me to smile, relax, and enjoy the little moments! iv

TABLE OF CONTENTS

LIST OF ABBREVIATIONS ...... vi

CHAPTER 1: Overview ...... 1

CHAPTER 2: Literature Review ...... 5

CHAPTER 3: The Effects of General Anesthesia and Euthanasia on Equine Mesenchymal Stem Cell Viability, Proliferation, and In Vitro Osteogenic Capacity ...... 46

CHAPTER 4: Characterization of the RUNX2 and OSX Regulatory Axis in Equine Mesenchymal Stem Cells During In Vitro Osteogenesis ...... 86

CHAPTER 5: The Role of Bone Morphogenetic Protein Activity in Equine Mesenchymal Stem Cell In Vitro Osteogenesis ...... 114

CHAPTER 6: Conclusions and Future Directions ...... 139

LIST OF ABBREVIATIONS

ADI-MSC Adipose-derived mesenchymal stem cell AKT Protein kinase B (PKB) ALP Alkaline phosphatase ANOVA Analysis of Variance BCS Body Condition Score BM-MSC Bone-marrow derived mesenchymal stem cell BMP Bone morphogenetic protein BSP Bone sialoprotein / integrin-binding sialoprotein (IBSP) c-AMP Cyclic adenosine monophosphate CD Cluster of differentiation cell surface marker COL1A1 Collagen, type I, alpha 1 CREB cAMP response element-binding protein Cx Control / growth media ddH2O Double-distilled water DLX-5 Distal-less homeobox 5 DMEM Dulbecco’s Modified Eagles Medium EF1α Elongation Factor 1α (housekeeping gene) EKLF Krueppel-like factor 1 ER Endoplasmic reticulum FBS Fetal bovine serum FGF Fibroblast growth factor GH Growth hormone Hh Hedgehog IGF Insulin-like growth factor Ihh Indian hedgehog IRE1a Inositol-requiring enzyme 1 alpha MAPK Mitogen-activated protein kinase / MKP-1 MEM Eagle’s Minimal Essential Medium MMP Matrix metalloproteinase

MSC Mesenchymal stem cell MSX-2 Muscle segment homeobox protein 2

NF H2O Nuclease-free water OIM Osteogenic induction media OSC Osteocalcin / bone gamma-carboxyglutamic acid-containing protein (BGLAP) OSE2 Osteoblast-specific element 2 OSM Osteomodulin / osteoadherin proteoglycans (OMD) OSN Osteonectin / secreted protein acidic and rich in cysteine (SPARC) OSP Osteopontin / secreted phosphoprotein 1 (SPP1) OSX Osterix / transcription factor SP7 P/S Penicillin/Streptomycin PDGF Platelet-derived growth factor PI3K Phosphatidylinositol-4,5-bisphosphate 3-kinase PKC Protein kinase C PTH Parathyroid hormone PTHrP Parathyroid hormone-related protein receptor qPCR Quantitative (Real-Time) Polymerase Chain Reaction RT Reverse Transcription RUNX2 Runt-related transcription factor 2 / core-binding factor subunit alpha-1 (Cbfa1) SBE Smad-Binding Element Shh Sonic hedgehog SOX-9 SRY-like box 9 transcription factor SP1 Transcription factor Sp1 (specificity protein 1) TAK1 Transforming growth factor-beta-activated kinase-1 TAZ Transcription Adaptor putative Zinc finger TGF-β Transforming growth factor beta VDRE Vitamin D response element VEGF Vascular endothelial growth factor Wnt ‘Wingless/Integrated’ XBP1 X-box binding protein 1

vii

CHAPTER 1 Overview

Bone is one of the few tissues in the vertebrate body with the capacity for authentic regeneration, yet approximately 10% of fractures fail to heal due to a variety of physiological and mechanical factors. Bone grafting is currently considered the “gold standard” therapy in fracture non-union and defect management, but there are numerous limitations that warrant investigation into alternative modalities. Regenerative medicine typically utilizes a multi-modal approach, incorporating combinations of bone grafting materials, bio-scaffolds, cell-based therapeutics or transplantation, gene therapy, and exogenous growth factors. To date, there is no “one-size-fits-all” strategy to ensure uneventful fracture repair.

MSCs have shown considerable promise in treating a variety of orthopedic conditions, both in human and veterinary medicine, due to their extensive proliferative ability, multi-lineage differentiation potential, and trophic activities. MSCs demonstrate tissue-specific lineage predilections, making particular sources more or less valuable for bone regenerative applications. Several studies have demonstrated that bone marrow-derived (BM) MSCs show superior in vitro and in vivo osteogenic capacity compared to adipose tissue-derived (ADI) and other MSCs. However, the techniques used in MSC collection, propagation, and in vitro differentiation often differ between species, making cross-study comparisons difficult to interpret. Therefore, there is a need for research into 1) optimizing MSC collection and in vitro techniques used within and across species, 2) identifying factors that influence the differing osteogenic potentials of MSCs from various sources, and 3) determining how to best enhance osteogenesis in MSC populations.

The overall objective of this dissertation was to identify factors that influence the in vitro osteogenic capacity differences between equine adult ADI- and BM-MSCs. The experiments were designed to 1) determine the effect of donor status during sample collection on the viability, proliferation, and in vitro osteogenic differentiation of ADI- and BM-MSCs, and 2) characterize the influence of intrinsic cellular characteristics, such as expression of the master osteogenic transcription factors runt-related transcription factor 2 (RUNX2) and osterix (OSX) and Bone Morphogenetic Protein (BMP) signaling activity, on the osteogenic abilities of ADI- and BM- MSCs. The results of the experimental studies will help to standardize sample collection and

1 culture techniques, and provide insights into approaches for improving osteogenesis in less osteogenically-inclined MSC populations to optimize their use in bone regenerative medicine.

The literature review (Chapter 2) first provides an overview of skeletal development, osteogenic differentiation, and bone regenerative medicine approaches. The remainder of the literature review focuses on MSC characteristics, cross-species ADI- and BM-MSC collection and isolation techniques, in vitro osteogenic differentiation models, and assays for measuring osteogenesis in vitro. Finally, the literature review outlines comparative osteogenesis studies in various species, provides an overview of the current understanding of the BMP-RUNX2-OSX regulatory axis, and then evaluates approaches for improving osteogenesis with a specific focus on BMP signaling enhancement and RUNX2 and OSX over-expression.

The first experimental study (Chapter 3) addressed a probable cause of inter-donor variability in stem cell biology research by testing the hypothesis that in vitro cell viability, proliferation, and osteogenic differentiation of MSCs collected from donors under general anesthesia and after euthanasia would be impaired compared to MSCs collected from donors during sedation. Equine adult ADI- and BM-MSCs were collected from four healthy adult horses sequentially during 1) standing sedation, 2) general anesthesia, and 3) shortly after euthanasia. Isolated MSCs were evaluated for initial primary cell yields and population doubling times through three passages. Passage 3 cells were subjected to in vitro osteogenic differentiation and evaluated at days 5 and 10 post-osteogenic induction for multi-cellular aggregation, Alizarin Red staining of mineralized matrix, mRNA expression of osteogenic genes RUNX2, OSX, and alkaline phosphatase (ALP), and ALP enzymatic activity. Overall, MSCs collected during general anesthesia and after euthanasia behaved equivalently to MSCs collected during sedation. There was a slight effect of donor status during collection on some earlier time point analyses, but the differences equilibrated at later time points. This study established that in vitro MSC viability, expansion, and osteogenesis were not adversely affected by general anesthesia or short- term euthanasia during tissue collection; however, both inter- and intra-donor variability influenced many of the outcomes measured, and the previous finding that equine adult BM- MSCs are more osteogenic than ADI-MSCs was re-confirmed.

The second experimental study (Chapter 4) characterized the RUNX2 and OSX regulatory axes of equine adult ADI- and BM-MSCs during osteogenesis to determine the effect of these genes on determining osteogenic differentiation capacity. The hypotheses were that 1) the relative osteogenic capacities of ADI- and BM-MSCs are consistent with resting and inducible RUNX2 and OSX expression levels, and 2) increasing OSX expression in ADI-MSCs improves their osteogenic capacity. Expression of RUNX2 and OSX under control and osteogenic conditions was evaluated at days 0, 2, 4, and 7 post-osteogenic induction. OSX, but not RUNX2, was significantly higher in resting BM- compared to ADI-MSC cultures. RUNX2 up-regulation was similar in both cell types in response to osteogenic media, but only BM-MSCs significantly induced OSX by day 7. Next, OSX was over-expressed in ADI-MSCs using an adenovirus, and osteogenic responses were assessed at days 4 and 7 post-infection. OSX over- expression did not consistently or significantly increase multi-cellular aggregation, Alizarin Red staining, ALP mRNA expression or enzymatic activity, or downstream mRNA expression of a panel of osteoprogenitor marker genes. This study established that differences in the RUNX2- OSX response axes might partly influence the osteogenic capacities of BM- and ADI-MSCs, but over-expression of OSX was not sufficient to substantially improve osteogenesis in ADI-MSCs.

The third experimental study (Chapter 5) evaluated the comparative BMP signaling activities in equine adult ADI- and BM-MSCs. The hypotheses were that 1) osteogenic BMP ligand expression differs in ADI- and BM-MSCs, 2) BMP inhibition in BM-MSCs impairs osteogenesis, and 3) BMP addition to ADI-MSCs improves osteogenesis. MSCs were first evaluated for resting and osteogenesis-induction expression of BMP ligands -2, -4, -6 and -7. BM-MSCs had significantly higher BMP-2 and -4 transcript levels than ADI-MSCs. BM-MSCs were then subjected to BMP inhibition under control and osteogenic conditions by using the small molecule BMP receptor kinase inhibitors DMH-1 and K-02288 and the extra-cellular antagonist Noggin. OSX, but not RUNX2, mRNA expression was significantly and dose- dependently down-regulated by BMP inhibition in both control and osteogenic conditions. BMP inhibition in BM-MSC cultures impaired multi-cellular aggregation, Alizarin Red staining, and ALP mRNA expression and enzymatic activity in a dose-dependent manner. Finally, recombinant human BMP-2 was supplemented to ADI-MSCs in both control and osteogenic conditions at 100 ng/ml, and samples were evaluated at days 2, 4, 7, and 10 post-osteogenic induction. BMP-2 supplementation significantly increased OSX expression and slightly 3 increased RUNX2 and ALP expression, while multi-cellular aggregation and Alizarin Red staining were improved over controls. This study first established that differences in baseline and osteogenesis-induced BMP signaling influence the osteogenic capacities of ADI- and BM- MSCs, next demonstrated that endogenous BMP activity is required for BM-MSC osteogenesis, and finally showed that exogenous BMP supplementation in ADI-MSCs improves osteogenesis.

Overall, the results of these studies indicate that MSCs collected from anesthetized and recently deceased donors show equivalent in vitro cell viability, proliferation, and osteogenic differentiation capacities as MSCs collected from awake but sedated donors. BM-MSCs show superior osteogenic differentiation to ADI-MSCs in part due to enhanced OSX induction in response to osteogenic cues and higher intrinsic BMP signaling in both resting and osteogenic conditions. Endogenous BMP activity is necessary for osteogenesis in BM-MSCs. Exogenous BMP supplementation, but not OSX over-expression by itself, was sufficient to improve the osteogenic capacity of ADI-MSCs. Therefore, the factors that influence the in vitro osteogenic capacity differences between equine adult BM- and ADI-MSCs are OSX expression and BMP signaling, while donor status during MSC collection does not impact in vitro cell behavior.

The dissertation’s “Conclusion and Future Directions” (Chapter 6) first summarizes the results from each of the experimental study chapters, next provides an outline for follow-up studies, and finally describes the more global implications and recommended future research direction from the entire body of work. The studies globally re-confirmed that equine adult BM- MSCs show superior osteogenic capacity to ADI-MSCs. The results of the first study indicate that MSCs from recently deceased donors can serve as a viable option for research and allogeneic stem cell use clinically. Donor variability significantly influenced outcomes, suggesting that donor inclusion criteria need to be established and MSC screening methods improved. The results of the second and third studies suggest that OSX induction and BMP signaling activity, but not RUNX2 levels, significantly influence the osteogenic capacity of different MSC populations. Enhancing BMP signaling had a more significant effect on improving osteogenesis in a less-osteogenic MSC population than OSX over-expression. In conclusion, MSCs from many sources can be utilized for bone repair and regenerative medicine, but there are certain modalities that are more effective at improving osteogenic capabilities based on the intrinsic differences in cell populations from different source tissues and donors.

CHAPTER 2 Literature Review

Abstract Bone serves many diverse functions in the body. Although the cells on and within bone constitute a very small percentage of the tissue bulk (approximately 5%), surface osteoblasts and intra-osseous osteocytes are vital for ensuring appropriate matrix homeostasis and skeletal structural integrity. The development of bone during skeletogenesis and its regeneration during fracture repair utilize similar cellular and molecular processes, collectively termed ‘endochondral ossification’. Osteoblasts line the surface of bone, on the inner layers of the periosteum and endosteum. They are derived from mesenchymal stem cell (MSC) precursors through osteogenic differentiation and responsible for bone synthesis. MSC osteogenesis is regulated by the master transcription factors RUNX2 and Osterix (OSX), with interplay from several signaling pathways, most importantly TGF-β/BMP signaling. Downstream targets of RUNX2 and OSX include genes necessary for bone matrix deposition and mineralization. Although bone possesses the inherent ability to regenerate, repair failure can occur due to multiple factors, including ischemia, excessive instability, extensive gap formation, and infection. Bone regenerative therapies often utilize multi-modal approaches that can include bone grafting, growth factor delivery and other biologic therapeutics, biomaterials, gene therapy, and/or MSC implantation. MSCs are a promising therapeutic option due to their accessibility from several adult tissues such as bone marrow (BM) and adipose (ADI), high proliferative ability, and capacity for multi-lineage differentiation. This review first provides an overview of skeletal development, osteoblast differentiation, and bone regenerative medicine approaches. The remainder of the review focuses on MSC characteristics, cross-species BM- and ADI-MSC collection and isolation techniques, in vitro osteogenic differentiation models, and assays for measuring osteogenesis in vitro. Comparative osteogenic potentials of BM- and ADI-MSCs are discussed, with a focus on methods for improving osteogenesis based on the BMP-RUNX2-OSX regulatory axis, specifically BMP signaling enhancement and directed RUNX2 and OSX over-expression.

Bone Structure & Function Bone is a dense, mineralized connective tissue that is formed and maintained by osteoblasts and osteocytes and resorbed by osteoclasts during remodeling and normal turnover. 5

Due to the structural rigidity of bone, its primary functions include support and protection of soft tissues and locomotion [reviewed in 1]. Bone is also vital for endocrine regulation of calcium and phosphorus, production of erythrocytes, platelets, and white blood cells within the bone marrow compartment, and metabolic processes such as fat storage and acid-base balance [reviewed in 2]. Bone contains three main cell types responsible for initial skeletogenesis, skeletal maintenance, and bone remodeling: osteoblasts, osteocytes, and osteoclasts. Bone is composed of approximately 50 – 70% bound minerals, 20 – 40% flexible matrix, 5 – 10% water, and < 3% lipids [reviewed in 3]. Structurally, bone is classified as ‘cortical’ or ‘trabecular’ [1 – 3]. Cortical or compact bone is dense and solid, and it surrounds bone marrow spaces. Trabecular or cancellous bone is less dense and is composed of a porous network of plates and rods. Trabecular bone is interspersed with bone marrow spaces, near the ends of long bones, and in the vertebrae. Both types of bone are composed of ‘osteons’, which consist of concentric layers or ‘lamellae’ of bone matrix interspersed with osteocytes and their canaliculi, surrounding a central Haversian canal [3]. The organic phase of bone extracellular matrix consists of 85 – 90% elastic collagen fibers, predominantly collagen type I, and trace amounts of other collagens, also known as “ossein”. Collagen fibers are typically arranged into cylindrical lamellae to form lamellar bone; an arrangement that is mechanically superior to woven or fibrous bone that contain randomly- organized collagen fibers [4]. The remaining matrix composition includes serum proteins, glycosaminoglycan-containing proteins, glycoproteins, and many other factors collectively known as “ground substance” [3]. The mineral content of bone is predominantly calcium hydroxyapatite, with small amounts of carbonate, magnesium, and acid phosphate [3]. Together, the mineral provides mechanical rigidity, allowing for load-bearing strength, while the organic matrix provides limited elasticity and flexibility, which improves fracture resistance.

Skeletal Development Overview The formation of bone during fetal development (skeletogenesis) occurs via two processes: intramembranous and endochondral ossification [reviewed in 5]. Intramembranous ossification involves direct mineralization of mesenchymal tissue, without an intermediate cartilaginous phase. Intramembranous ossification is responsible for formation of the flat bones

6 of the skull, jaw bones, and clavicles. In contrast, endochondral ossification involves the formation of bone from a pre-existing cartilage model, and it occurs during development of the remaining bones in the body. Endochondral ossification is also the process that drives most fracture repair, through callus maturation.

Initiation & Patterning Skeletogenesis begins in the vertebrate embryo when multipotent mesenchymal stem cells (MSCs) originating from the ectoderm and mesoderm migrate and condense at specific locations in the embryonic body, and then commit to a skeletal fate [6]. The MSCs give rise to common osteo-chondro progenitor cells that are capable of differentiating along both chondroblastic and osteoblastic lineages. Specific fate commitment is largely determined by expression and activities of the master transcription factors SRY-like box 9 (SOX-9) and Runt- related transcription factor 2 (RUNX2), respectively [7]. Morphogen and signaling networks, driven by bone morphogenetic proteins (BMPs), fibroblast growth factors (FGFs), retinoic acid, Sonic Hedgehog (Shh), and Wnt ligands, control the expression and activity of various patterning transcription factors [reviewed in 8].

Cartilage Anlagen The initial skeletal framework is pre-figured through cartilaginous models, also referred to as ‘anlagen’. Anlaga form the template for both axial and appendicular skeletal elements. Initially, osteo-chondro progenitor cells aggregate, forming condensations, undergo chondroblastic differentiation, and begin producing cartilage extracellular matrix [5]. Chondrocyte proliferation and hypertrophy drive tissue elongation, which begins to shape the bone diaphysis and epiphyses. Terminally differentiated hypertrophic chondrocytes at the primary and secondary centers of ossification and, at later time points, the growth plates, secrete angiogenic factors and begin to express osteoblast markers, initiating the replacement of cartilage matrix with bone [5]. Several ligand-mediated pathways regulate chondrocyte maturation during endochondral ossification, including BMPs and transforming growth factor betas (TGF-β), growth hormone (GH), insulin-like growth factors (IGFs), thyroid hormone, retinoic acid, testosterone, and estrogen [8, 9].

Endochondral Ossification The eventual replacement of the cartilage by bone at primary centers of ossification and, later, in growth plates occurs through endochondral ossification [reviewed in 5, 9]. Terminal chondrocytes undergo apoptosis, triggering chondro-osteoclastic resorption of the cartilaginous matrix, vascularization, and osteoblast migration to replace the cartilage with bone matrix [5]. Pre-hypertrophic chondrocytes secrete Indian hedgehog (Ihh), which induces osteoblast differentiation in the adjacent periosteum, leading to bone matrix formation [10]. Hypertrophic chondrocytes also secrete vascular endothelial growth factor (VEGF) to trigger blood vessel development in the forming bone [11]. Matrix metalloproteinases (MMPs) are secreted by terminal chondrocytes, degrading the surrounding cartilage to facilitate osseous substitution [12]. The close interactions between cartilage and bone cells during endochondral ossification result in the formation of primary ossification centers in the diaphyses and secondary ossification centers in the epiphyses and apophyses of long bones, and in developing cuboidal bones.

Intramembranous Ossification Intramembranous ossification occurs when osteogenic progenitors form compact nodules within the developing skull bones [reviewed in 5, 13]. The cells become osteoblastic, without any chondrogenic intermediate, and produce an organic matrix that is mineralized from an ossification center in the center of each bone [13]. The skull bones remain connected by fibrous interfaces called “sutures” that serve as growth centers; they maintain a pool of undifferentiated, osteogenic cells that are recruited into ossification fronts until fusion of the sutures during postnatal growth [5].

Osteoprogenitor Differentiation Osteoblasts Osteoblasts are responsible for both intramembranous and endochondral ossification. They comprise 4 – 6% of the total bone cell population, reside along the surfaces of bone, and predominantly function as bone-forming cells [reviewed in 13]. They are derived from MSCs through osteogenic differentiation. Osteoblasts synthesize bone matrix by first secreting a non- mineralized, organic matrix known as ‘osteoid’ that contains abundant collagen type I [reviewed in 5, 15]. As osteoblasts mature, they secrete bone-specific proteins, such as bone sialoprotein /

8 integrin-binding sialoprotein (BSP / IBSP), osteocalcin / bone gamma-carboxyglutamic acid- containing protein (OSC / BGLAP), osteonectin / secreted protein acidic and rich in cysteine (OSN / SPARC), osteomodulin / osteoadherin proteoglycan (OSM / OSAD), osteopontin / secreted phosphoprotein 1 (OSP / SPP1), and other proteoglycans [reviewed in 14 – 16]. These osteogenic genes are often used to identify and monitor the osteogenic phenotype, experimentally. Matrix is then mineralized by osteoblasts, initially through a ‘vesicular phase’, which involves release of matrix vesicles containing calcium, enzymes that degrade proteoglycans, and alkaline phosphatase (ALP) to secrete phosphate ions from the vesicles. Ultimately, the calcium and phosphate from the osteoblast-derived vesicles form hydroxyapatite crystals, which are released during the ‘fibrillar phase’ of mineralization and integrate throughout the surrounding matrix [17]. The differentiation of osteoblasts and their functional processes are predominantly governed by the transcription factors RUNX2 and osterix (OSX), and multiple signaling pathways [reviewed in 15, 18].

RUNX2 Runt-related transcription factor 2 (RUNX2), also known as core-binding factor subunit alpha-1 (Cbfa1) is considered a master osteogenic transcription factor as its inactivation in developmental murine models blocks osteoblast differentiation and results in a complete absence of mineralized bone [19, 20]. RUNX2 contains a ‘Runt domain’ near the N-terminal region that is responsible for binding DNA through an alpha subunit and hetero-dimerization with a non- DNA binding core binding factor beta subunit [21]. RUNX2 targets downstream genes by binding to an osteoblast-specific element 2 (OSE2) in the promoter region of genes such as OSC, BSP, collagen type I α1 (COL1A1), and OSP [22]. While RUNX2 is necessary for osteogenesis, it is not sufficient for osteoblast differentiation from chondrocyte precursors [23].

Osterix Osterix (OSX), also known as transcription factor Sp7, is also necessary for osteogenesis, as its deletion also blocks osteoblast differentiation and mineralized skeleton formation, similarly to RUNX2 [24]. OSX expression is dependent on and downstream of RUNX2: RUNX2-null mice do not express OSX, while OSX-null mice express RUNX2 with wild type spatial and distribution profiles [24]. The OSX promoter contains response elements for RUNX2 and several

9 other factors, such as DLX-5, SOX-9, and SP1 [reviewed in 26]. Downstream targets of OSX include COL1A1, BSP, OSC, OSN, OSP, and ALP [24, reviewed in 27]. OSX-null cells exhibit a pre-hypertrophic chondrocyte phenotype, suggesting that OSX is required to suppress chondrocytic differentiation of osteo-chondro progenitor cells and enable osteoblast differentiation of RUNX2-positive cells [24].

Signaling Pathways Osteoblast differentiation is regulated by several signaling pathways, with the TGF-β / BMP networks arguably playing the most significant roles [reviewed in 28 – 29]. TGF-β ligand or receptor-deficient mice models show various defects ranging from reduced bone growth and mineralization or deformed bones to early embryonic lethality [28]. TGF-β autocrine and paracrine signaling is important in maintaining and expanding MSCs [30], and the sensitivity of osteoblasts to TGF-β’s mitogenic activity is directly correlated with developmental processes [reviewed in 31]. TGF-β signaling regulates osteoprogenitor proliferation, early differentiation, and osteoblast commitment through canonical Smad 2/3 networks, non-canonical MAPK pathways, and through interplay with Wnt, Hedgehog, Notch, FGF, and parathyroid hormone (PTH) signaling [28]. RUNX2 is a primary nuclear target of both canonical and non-canonical TGF-β signaling [28]. BMPs are vital for both in vivo bone development and in vitro osteoblast differentiation [32]. BMPs were first discovered in 1965, following the observation that demineralized bone matrix implanted into muscular tissue induced ectopic bone formation [33]. BMP-deficient mice models often die too early in development to establish their skeletal phenotypes, and individual BMP ligands can compensate for each other or have antagonistic effects on bone formation [reviewed in 28, 32]. In vitro, recombinant BMP proteins applied to osteoblastic lineage cells enhance expression of bone-specific markers such as ALP, collagen type I, osteocalcin, and parathyroid hormone-related protein receptor (PTHrP) [32]. BMPs are able to stimulate osteoblast differentiation of several pluripotent cell lines while suppressing differentiation down myogenic [35] and chondrogenic [36] lineages. Similar to TGF-β signaling, BMPs regulate osteoblast differentiation through canonical Smad 1/5/8 networks, non-canonical MAPK pathways, and via interactions with Wnt, Notch, FGF, and hedgehog (Hh) signaling [28]. The primary nuclear targets of both canonical and non-canonical BMP signaling are RUNX2 and

OSX, with OSX also a secondary target through DLX-5 [28, 29]. The BMP regulatory axis will be discussed in further detail later in this review in the context of its control over RUNX2 and OSX activity, and utilizing BMPs to enhance osteogenesis. The canonical Wnt / β-catenin signaling pathway is important in the fate-determining decision of osteo-chondro progenitor cells to undergo osteogenesis instead of chondrogenesis [37]. Wnt signaling is involved in both endochondral and intramembranous ossification processes during development [38]. Over-expression of Wnt9a during later stages of cartilage development promotes endochondral bone formation, while targeted β-catenin removal in osteo- chondro progenitor cells, which blocks Wnt signaling, significantly reduces embryonic bone formation in both the calvarium and long bones [37, 39]. The timing of Wnt signaling activity during endochondral ossification is crucial to regulate early chondrocyte differentiation and promote late stage osteoblast differentiation [37, 39]. Hedgehog (Hh) family member signaling is involved in the regulation of skeletal development and homeostasis [reviewed in 37], with Ihh specifically required for osteoblast differentiation during endochondral ossification [10]. Ihh signaling was found to be upstream of Wnt / β-catenin during osteoblast differentiation in multiple mutant mice models. Ihh regulates the transition of pre-osteoblasts to RUNX2+ and OSX+ immature osteoblasts, while Wnt / β- catenin functions in osteoblast maturation and is required after OSX expression [37]. Fibroblast growth factor (FGF) signaling is involved in several processes during both intramembranous and endochondral ossification, with multiple FGF ligands showing specific spatiotemporal expression patterns during skeletogenesis [40]. FGF ligands -2, -4, and -6 were found to be inducers of osteoblast proliferation [41]. There are several gain- and loss-of-function FGF ligand and receptor mutations that result in skeletal diseases [reviewed in 40]. Many of the FGF receptor mutations appear to affect all cell types involved in skeletal development, including chondrocytes, osteoblasts, and osteoclasts. FGF signaling exerts its effects through Ras-MAP Kinase, protein kinase C (PKC), and PI3K / AKT networks [40], as well as cross-talk with previously mentioned signaling pathways. In addition to the significant cross-talk between the signaling pathways mentioned above, there are other important pathways involved in both positively and negatively regulating osteogenesis, including calcium (Ca2+) flux, insulin-like growth factor (IGF), platelet-derived growth factor (PDGF), and Notch signaling [reviewed in 42]. Many of these signaling networks

11 are common targets for therapeutic interventions in bone regenerative medicine, and BMP use will specifically be covered further in detail later in this review.

Bone Regenerative Medicine Approaches Overview Bone is one of the few tissues with the inherent ability to authentically regenerate; fracture healing mimics the cellular and molecular events of embryonic skeletogenesis [reviewed in 43]. Approximately 10% of fractures fail to repair adequately, often due to factors such as insufficient blood supply, infection, systemic disease, excessive bone loss, and inadequate stabilization [43]. Bone regenerative medicine often utilizes a multi-modal approach and can include techniques involving bone grafting materials, growth factors, blood- and bone marrow- based therapeutics, biomaterial scaffolds, gene therapy, and cultured MSC transplantation. Each of these modalities have been extensively reviewed elsewhere [44 – 52], so the following sections will contain only a brief overview of bone grafting, growth factor use, gene therapy, and MSCs. The latter part of this review will discuss in vitro MSC techniques in more detail and then focus on BMP growth factor use and RUNX2 and OSX gene therapy approaches specifically.

Bone Grafts In situations where fracture repair is likely to be compromised, the current “gold standard” approach is to use autogenous cancellous bone and/or bone marrow grafting to augment repair [45]. Bone grafts in humans are typically harvested from the iliac crest or, less commonly, the distal femur or proximal tibia [53]. Bone grafts contain osteogenic, osteoconductive, and osteoinductive properties [45]. Graft materials are osteogenic in that they contain osteoblasts, osteoconductive due to promoting capillary ingrowth and MSC migration, and osteoinductive in their promotion of mitogenesis of MSCs through release of BMPs and other osteogenic proteins [45, 53]. However, there are several limitations to autogenous bone grafting, including the limited availability of healthy donor tissue, loss of bone stock, and donor- site morbidity issues [45, 53 – 54]. Allogeneic grafts provide an alternative, but screening of cadaveric donors, preparation, and storage of graft materials are highly complex and expensive processes, with the attendant risk of recipient rejection [45]. Ongoing research is focusing on alternative options to autografting, including xenografts and synthetic bone graft substitutes [44].

Growth Factors Growth factor applications for bone repair capitalize on signaling proteins that are prevalent during normal bone development and remodeling processes, with the most common agents being BMPs [46]. BMPs 2 and 7 already have FDA-approved applications for spinal fusion and tibial non-union treatment [56, 57]. However, they may not be beneficial over control treatments in all types of fracture healing or non-union repairs [58], and their use carries risk of excessive or heterotopic bone formation [57 – 59]. Therefore, there is still ongoing research into the efficacy and strategic application (dose, carrier, etc.) of BMPs in various sites for bone healing [59]. The success of recombinant BMP use has expanded research into other growth factors, such as TGF-βs, FGFs, IGFs, VEGF, PDGF, and PTH, all of which show promise for the potential use in bone regeneration [reviewed in 60 – 65]. However, direct growth factor delivery to repair sites is often limited by insufficient local retention, high protein concentration requirements, and cost-inefficiency, thus warranting investigation into strategies to optimize their use by coupling with biodegradable carriers, MSCs, and/or biomaterials [66].

Gene Therapy Gene transfer technologies have the potential to improve bone healing by delivering local and sustained expression of osteogenic genes to sites of repair. There are numerous strategies for modifying gene expression, such as viral vectors (retrovirus, lentivirus, adenovirus, adeno- associated virus) and expression plasmids, with either ex or in vivo incorporation techniques [reviewed in 69 – 72]. Ex vivo approaches involve genetically engineering MSCs in vitro prior to implantation. In vivo strategies include direct injection of vectors, implantation of gene-activated matrices, or biomaterial-coupled delivery. There are numerous ‘osteogenic’ genes that have been investigated in this context, with the most common being BMPs, TGF-βs, and VEGF, due to their importance in stimulating osteogenesis and angiogenesis, or the osteogenic transcription factors RUNX2 and OSX [70, 71]. Clinical trials in animal models have been mostly promising, but the results have been inconsistent [70, 71]. For example, direct injection of BMP-2 or -6 adenoviruses into defects in rats, rabbits, and horses resulted in improved defect healing, but in a commonly used sheep tibial osteotomy model, the results have been poor [70]. Traditional ex vivo approaches have been more successful in large animal models, including horses, goats, and pigs. The differing success between animal model studies can likely be attributed to

13 inconsistency in the gene delivery methods and doses, the type of bone defect used and extent of healing evaluated, and species-specific differences in osteogenic responses to a given gene target. There are numerous safety issues associated with gene delivery, such as possible DNA recombination with integrating retro- and lentiviruses, excessive gene over-expression, immune reactions, and tumor development [72]. A few human deaths in early clinical trials, which were mostly attributed to non-vector related factors, significantly impaired the continued advancement of the field [71]. Further research is needed to determine which gene(s), delivery technologies, and implantation approaches are most appropriate before human clinical trials can commence. Data is lacking in vital areas such as large animal studies using clinically-relevant bone defect models, pharmacological and toxicological safety trials, and optimizing which MSC source and/or scaffold material is most appropriate for ex vivo gene therapy approaches.

Cell-Based Therapies Cell-based therapies seem to hold the most promise in replacing autogenous bone grafting, and extensive research has been performed investigating treatments ranging from direct injection of cells into bone defects to implanting MSC-seeded biomaterial scaffolds in bone repair sites [reviewed in 52]. MSCs can be isolated from a variety of adult tissues and fluids, and they are involved in multiple roles during fracture repair and bone regeneration [73 – 75]. MSCs contribute directly in these processes through osteogenic differentiation and indirectly through paracrine signaling and immune response modulation [74]. In addition to primary bone marrow aspirate-based cell therapies that do not require culture, there are multiple other cell-based therapy modalities that utilize in vitro expansion and selection, and implantation with and without biomaterial matrices [75]. Cell culture techniques allow for expansion and screening of the MSC population prior to in vivo use, but can be both time- and cost-consuming, and in the context of human clinical applications, create substantial ‘chain of custody’ requirements that add cost and complexity to the procedures. There has been a multitude of studies and clinical trials that utilized cell-based therapies for various bone healing or regeneration purposes with generally successful outcomes [reviewed in 74, 75]. The use of allogeneic stem cells could prove beneficial in patients with non-unions that lack sufficient or healthy osteoprogenitor cells themselves [74], although safety and regulatory requirements have yet to be established for allogeneic stem cell banking [76]. Limitations associated with cell-based therapy alone include

14 inadequate delivery or survival of appropriate numbers of cells at the site of repair, infection risks, and donor morbidity associated with tissue harvesting [77]. The remainder of this review will focus on MSC characteristics, comparative in vitro techniques across species, in vitro osteogenesis approaches, and methods for improving osteogenesis with a focus on the BMP, RUNX2, and OSX regulatory axes.

Mesenchymal Stem Cells Characteristics MSCs are self-renewing and multi-potent cells, although their exact definition and roles vary within the literature [78]. MSCs were originally identified by their in vitro characteristics, which include their ability to adhere to plastic culture substrates, capacity for substantial clonal expansion, and differentiation ability along several mesenchymal tissue lineages [79, reviewed in 80]. Adherence to cell culture plastic was a criterion originally established for bone marrow- derived MSCs as it excluded hematopoietic lineage cells also found in marrow aspirates [79]. Clonal expansion eliminates terminally-differentiated cell types, allowing for a population of predominantly stem cells to develop over time, through continued cell proliferation [80]. However, MSCs are prone to senescence, which is associated with a loss of multi-potency [81]. Multi-potency of MSCs is typically tested as a tri-lineage capacity, since in vitro adipogenic, chondrogenic, and osteogenic differentiation protocols and outcome assessments are well- established [80]. An additional criterion for defining MSCs, based on cell surface marker profiling, has been proposed more recently and includes identifying positive and negative markers [82]. Surface markers are well-characterized in human MSC populations [reviewed in 83], but surface marker criteria in other species is highly variable and non-specific [84].

Sources & Uses MSCs were initially isolated from bone marrow [79], but numerous other adult tissues and fluids contain MSC populations, including but not limited to adipose tissue, umbilical cord blood and Wharton’s jelly, periosteum, muscle, synovium, tendon, and circulating blood [78, 80, 82, 85, 86]. Stem cells reside in specialized local environments known as “niches” within their tissue of origin, which provide support and appropriate stimuli to maintain “stemness” in MSC populations. Stem cell niches are generally located close to vasculature to allow rapid

15 mobilization of MSCs in response to local and systemic signals. Self-evidently, accessing niches is crucial to the efficient harvesting of MSCs, regardless of the specific source [86]. The widespread use of MSCs in tissue regeneration studies is a consequence of several stem cell characteristics, including homing to sites of inflammation, direct contributions to healing processes, their ability to differentiate into various cell types, local effects on tissues and other cells, and anti-inflammatory and immuno-modulatory properties [reviewed in 87, 88]. These beneficial activities have been documented in preclinical animal model studies [reviewed in 87], and human clinical trials using MSCs for multiple target conditions are extensive and ongoing [reviewed in 88].

MSC Collection & Isolation Techniques Overview The following sections will focus on the two most readily available sources of MSCs in most species, bone marrow (BM) and adipose tissue (ADI). The variability in bone marrow and adipose tissue collection approaches complicates cross-species study comparisons due to differences in donor status during collection and potential site-specific variation in the bone marrow and adipose MSC niches. Sample processing techniques are relatively similar across species with regard to isolation of MSCs from bone marrow aspirates and adipose tissue samples. However, there does not appear to be a consensus as to the best type of cell culture media or additives to use, and there are subtle differences in recommended protocols for culturing and expanding MSCs both across and within species.

Bone Marrow Collection & Isolation Bone marrow aspiration from live patients requires aseptic site preparation and use of a Jamshidi bone biopsy needle to penetrate the medullary cavity of the bone being harvested from. Common sites for bone marrow aspiration in humans are the iliac crest, sternum, and femoral head, and the procedure is commonly performed while the donor is under anesthesia [89, 90]. In animal models, the source and method of harvest is highly variable. In rodents, bone marrow is collected from recently euthanized animals by flushing the femoral marrow cavity [91]. In dogs, bone marrow can be collected from the wing of the ilium, proximal medial tibia, and proximal humerus, typically under general anesthesia [92]. Cats usually have bone marrow aspiration

16 performed from the proximal humerus [93]. In horses, bone marrow is routinely harvested from the sternum or iliac crest under standing sedation and restraint [94]. In many food / production animal species, it is common to perform MSC research on bone marrow harvested from abattoir- derived samples, with considerable variability in the time lapse between death and stem cell collection [95]. There are numerous reported uses of bone marrow-derived MSCs for treating a variety of clinical conditions in veterinary medicine [96]. Bone marrow aspirates contain a relatively low density of MSCs (< 0.01% of the total cell population), so there are several techniques that involve varying degrees of “concentrating” the sample prior to plating [94]. Classic techniques involve transferring the entire aspirate directly into tissue culture flasks or first centrifuging the sample to condense cellular contents and remove liquid components [91, 94]. These techniques rely on the assumption that only MSCs will selectively adhere to the culture vessel plastic, thus eliminating non-MSC cells. Additionally, aspirates can be treated with red blood cell-lysing agents, such as 0.8% ammonium chloride. Bone marrow aspirate can be placed in a density gradient solution, such as Ficoll or Percoll, to separate the mononuclear cell fraction from other blood and bone marrow components [90, 94]. Depending on the processing method, it may be possible to count mononuclear cells prior to plating, although this does not provide a direct indication of stem cell numbers.

Adipose Tissue Collection Adipose tissue is usually isolated from subcutaneous fat stores, thus necessitating removal of surrounding connective tissue and parenchyma. For this reason, adipose tissue collection often involves minor surgical procedures [89, 92, 94]. In humans, the most common collection technique for adipose tissue is lipo-aspiration [97]. In rodents, adipose tissue is harvested from recently euthanized animals and typically taken from intra-abdominal or epididymal / testicular locations [98]. In dogs, common collection sites for adipose tissue are the falciform ligament or omental tissue, which have been shown to be superior sources of MSCs over subcutaneous fat stores due to higher cell yields regardless of body condition, but require general anesthesia and invasive surgical approaches [99]. Cats can have adipose tissue harvested under sedation via punch biopsy of the ventral abdomen due to abundant fat stores located caudal to the umbilicus [93]. In horses, adipose tissue is typically collected from subcutaneous stores adjacent to the tail head, which is performed on sedated animals using local anesthetic [94].

There are many reported uses of adipose-derived MSCs in veterinary medicine for orthopedic and non-orthopedic conditions [100]. Adipose tissue samples also contain a relatively low density of MSCs compared to adipocytes, fibroblasts, and other parenchymal cells within the tissue. Therefore, MSCs must be liberated from the tissue by various digestion and agitation techniques. Samples are typically minced into smaller pieces, separated into phases, and digested using a 0.1% collagenase solution in PBS or culture media [89, 94, 98]. Agitation can be performed by vortexing [98] or incubation at 37oC in an orbital shaker [89, 94] to improve collagenase digestion efficiency. Digested adipose tissue can additionally be filtered through 70 and 40 μm filters to remove poorly digested debris. The suspension is centrifuged to pellet the vascular-stromal fraction containing the MSCs, which is then washed several times prior to counting, to exclude the more buoyant adipocytes [89]. Some sources recommend different initial cell seeding densities based on the source of the adipose tissue, but they generally range from 1 – 10,000 cells/cm2 [89].

MSC Culture MSCs are typically cultured and expanded in “growth media” with the addition of 10 – 20% fetal bovine serum (FBS). The most commonly used base media formulations are variations of Eagle’s Minimal Essential Medium (MEM), such as α-MEM and Dulbecco’s MEM (DMEM) [89, 91, 94], although DMEM-Ham’s F12 is also occasionally used [89, 94, 98]. Antibiotics are routinely added, most commonly penicillin (100 U/ml) and streptomycin (100 μg/ml) [89]. Antifungals such as amphotericin B (25 μg/ml) are sometimes also included [98]. A comparative study using human bone marrow MSCs found that α-MEM + 10% FBS + Glutamax or L-glutamine resulted in the fastest in vitro expansion while preserving multi-potency [101]. The study also found that lower initial cell seeding densities of 1,000 cells/cm2 resulted in higher numbers of passage 0 (P0) adherent cells and greater MSC-enrichment than with densities higher than 25,000 cells/cm2 [101]. Collectively, it appears that there is no steadfast rule regarding growth media formulations or seeding densities for MSCs, suggesting that they respond reasonably consistently across species during in vitro expansion. MSCs will adhere within the first 72 hours of culture [79]. Within 1 – 2 weeks, MSCs will appear in clusters as spindle-shaped, fibroblast-like cells, reflective of clonogenic expansion [89]. Since such a small proportion of adherent cells are ultimately MSCs, homogeneity of the

18 culture can only be achieved after prolonged culture of at least 2 – 3 weeks and up to passages 3 and 4 [79], to enrich for the persistently proliferative stem cell sub-population. Culture medium is typically changed 2 – 3 times per week, with some sources recommending the removal of only 40 – 50% of the existing volume to preserve soluble factors released by the cells [89]. Cell contact inhibition occurs when culture reach near confluence (70 – 80%), upon which detachment via 0.05% trypsin-EDTA is recommended [101]. MSCs have higher sensitivity to trypsin than other adherent cell types [102]; thus, it is recommended to limit trypsinization to 5 – 7 minutes at 37oC [89]. Replating trypsinized MSCs at a lower seeding density of 1 – 8 x 104 cells/cm2 has been shown to improve their proliferation rate and preferentially enhance their expansion [89, 102]. However, care must be taken not to over-passage MSC cultures as there is a clear link between increased passage number or time in culture and senescence [103, 104]. It has been shown that bone marrow MSCs reach senescence faster than adipose- and umbilical cord- derived MSCs [104]. MSCs can be assessed for “stemness” characteristics, such as self-renewal and clonogenicity, by performing a colony-forming unit (CFU) assay or evaluating cluster of differentiation (CD) cell surface markers. CFU assays are carried out by seeding very low density of cells (such as 18 cells/cm2), and then observing the culture regularly to identify individual colonies of cells originating from a single parent cell. The number of colonies present reflects the self-renewal potential of the initial cell population [105]. CD markers on the cell surface can be evaluated through immunocytochemistry or flow cytometry [94]. Human MSCs should express CD73, CD90, and CD105, but they should not express CD11b, CD14, CD19, CD34, CD45, CD79, and major histocompatibility complex II (MHC-II) [89]. Other markers of human MSCs include CD29, CD44, CD106, and CD166 [79]. The International Society for Cellular Therapy defines the surface markers for human MSCs, but there is not a consensus regarding markers for other species. This is partially due to a lack of species-specific antibodies and lack of cross-reactivity with available antibodies for human targets [94]. MSCs are also routinely evaluated for multi-lineage differentiation, typically for tri-lineage adipogenic, chondrogenic, and osteogenic capabilities, the last of which will be discussed in detail in the following section.

In Vitro Osteogenic Differentiation Osteogenic Induction Media In vitro osteogenic differentiation is induced in monolayer cultures of MSCs, typically passage 3 or 4 due to presumed homogeneity of cells and a more ‘pure’ MSC population. Growth media, as previously described, is supplemented with ascorbic acid, β-glycerophosphate, and dexamethasone to produce osteogenic “induction” media (OIM) [80, 89]. Additional additives to OIM sometimes include 1,25-Dihydroxyvitamin D3 and recombinant BMP-2 [106]; these are shown to have added benefit in improving osteogenic outcomes over OIM alone in less osteogenic progenitor cell populations [107]. At least three weeks of continuous treatment with OIM is required for complete osteogenic differentiation, but clear indications of phenotypic commitment are usually evident by day 7 in responsive cultures [108]. Ascorbic acid, β- glycerophosphate, and dexamethasone are necessary over the entire differentiation process [reviewed in 109]. It has also been shown that allowing cultures to concentrate their soluble products in the media by performing partial media replacement improves osteogenic outcomes, likely through retention of autocrine and paracrine signaling [110]. The following section will elaborate on the roles of each additive in further detail.

Ascorbic Acid Ascorbic acid (Vitamin C) is used in several formulations, most commonly L-ascorbic acid, L-ascorbic acid-2-phosphate, and L-ascorbic acid phosphate magnesium salt n-hydrate, with the phosphate analogs being more stable in culture conditions [109]. Its activity is primarily in collagen biosynthesis as it serves as an essential cofactor for prolyl lysil hydrolase, but it also possesses antioxidant functions [111]. Production of collagen in the extracellular matrix is necessary for osteoblast differentiation, as it results in integrin-mediated activation of the MAPK signaling pathway, with downstream targets such as RUNX2 being upregulated [112]. The optimal concentration of ascorbic acid-2-phosphate for osteogenic differentiation of human bone marrow MSCs was found to be 0.05 mM [110], yet up to 0.3 mM is used in some species [94].

β-glycerophosphate β-glycerophosphate is typically used as disodium salt hydrate formulations. It primarily serves as a source of phosphate for hydroxyapatite deposition during bone formation, but it may

20 also influence intracellular signaling molecules through liberation of inorganic phosphate [109]. Phosphate activates MAPK/ERK signaling pathways that induce downstream osteogenic genes such as OSP and BMP-2 [109]. BMP-2 expression is also increased by phosphate through the c- AMP/PKA pathway [109]. The optimal concentration of β-glycerophosphate for osteogenic differentiation of human bone marrow MSCs was determined to be 10 mM [110], but concentrations greater than 2 mM can result in non-apatitic or “dystrophic” mineralization as a product of cell death in populations lacking osteogenic differentiation capacity [113].

Dexamethasone Dexamethasone is a synthetic glucocorticoid that can preferentially induce MSCs toward adipogenic, chondrogenic, and osteogenic lineages depending on the concentration and other factors present in the media formulation [106]. Its exact mechanisms of action during differentiation aren’t well defined, but it enhances the responsiveness of MSCs to other differentiation stimuli since it does not define any specific lineages by itself. Dexamethasone directly induces RUNX2 expression and activity through the Wnt/β-catenin signaling pathway [114], activating TAZ [115], and de-phosphorylating RUNX2 through MAPK phosphatase (MKP-1) [116]. Dexamethasone induces apoptosis in MSC populations that have poor intrinsic differentiation capacity [117], prevents apoptosis in confluent cultures [108], and promotes MSC proliferation [118]. The optimal concentration of dexamethasone for osteogenic differentiation of human bone marrow MSCs was determined to be 100 nM [110], but concentrations as low as 10 nM are commonly used with similar efficacy [119].

Assays for In Vitro Osteogenesis Overview There are numerous assays for in vitro osteogenesis that are routinely used to assess the quality and degree of differentiation. These include observing and counting multi-cellular aggregate formations over time, determining expression profiles of osteoblast-specific gene transcripts, measuring alkaline phosphatase enzymatic activity through staining and colorimetric assays, staining for mineralized matrix with Alizarin Red or Von Kossa dyes, and, in late stage cultures, quantifying calcium deposition in the matrix. The following sections describe the methodologies and evaluate the utility of each assay for determining in vitro osteogenesis.

Cellular Aggregation In the early stages of in vitro osteogenesis, permissive cell monolayers initially adopt a ‘herringbone’ alignment, before coalescing into dense, multi-cellular aggregates. Aggregate counting is not a commonly used osteogenic differentiation outcome measure, but it serves as a straightforward and self-evident quantitative assay for early cellular responses to osteogenic stimuli [120]. Aggregation is not specific for osteogenic differentiation as it also occurs during adipogenic differentiation, and it is often considered necessary for chondrogenic differentiation since cultures are typically pelleted or grown in micro-masses [80]. The concept of counting aggregates over time to assess osteogenic capability is based on the observation that cultures showing little or no aggregation also do not tend to have other positive osteogenic assay outcomes [120]. As previously reviewed, MSC migration in vivo is necessary for multiple processes during development and bone repair/remodeling, and it is responsive to many systemic cytokines and local paracrine cues. Cell-cell contact is essential for osteogenic differentiation, due in part to specific interactions mediated by cadherin-11 [121] and gap junctions [122], and other contact-dependent signaling pathways [108]. Although counting cellular aggregation during in vitro osteogenesis is an easy way to monitor early differentiation, it is neither sensitive nor specific for osteogenesis as cellular aggregation is influenced by multiple factors other than differentiation cues alone, such as chemotaxis in response to paracrine signaling and direct cell contact-mediated signaling [122, 123].

Osteogenic Gene Expression Gene expression profiling is one of the more specific assays for osteogenesis. There are multiple genes that are routinely evaluated, specifically the master transcription factors RUNX2 and OSX, their upstream regulators (transcription factors DLX-5 and MSX-2, plus several additional signaling pathway effectors), and their common downstream targets. A large-scale study evaluated multiple sources of literature and bioinformatics databases to classify the major genes involved in osteogenesis into four categories: cell migration, skeletal development, calcium ion binding, and ossification. The “Leader” genes involved in ossification were found to be the common osteoblast markers RUNX2, BMP-2, and OSN/SPARC, with “class B” genes including OSP/SPP1, OSM, BMP-4, MSX-2, and several others involved in secondary processes [125]. A microarray-based study found that ALP, BSP/IBSP, OSP/SPP1, OSC/BGLAP, and

COL1A1 were all important markers of osteogenesis and reflected osteoprogenitor commitment and osteoblast maturation [126]. It is well-established that RUNX2 is not osteoblast-specific but instead serves as markers for osteo-chondro progenitors; therefore, it is most appropriately used in evaluation of pre- or immature osteoblasts, especially since RUNX2 expression inhibits later stage osteoblast maturation [127]. OSX expression peaks in pre-osteoblasts and remains high through most of osteoblast differentiation [reviewed in 128], but it too can also inhibit late stage osteoblast maturation [129]. ALP and COL1A1 peak in pre-osteoblasts and remain high throughout differentiation [130]. OSP precedes expression of BSP and OSN, while OSC is expressed in fully differentiated osteoblasts [130]. In summary, there are multiple marker genes for osteoblast differentiation, and it is important to be aware of their roles and temporal expression profiles when evaluating osteogenesis in a cell population, regardless of the methodology used.

Alkaline Phosphatase ALP activity is commonly evaluated through monolayer staining and/or bioactivity assays. Histochemical staining for ALP activity typically relies on phosphatase enzymatic activity to dephosphorylate molecules under alkaline conditions, which is monitored through the generation of a colorimetric dye. A common substrate is napthol AS-BI phosphate, with color development produced by fast red violet LB or fast blue BB salt compounds. Fluorescent and immunostaining methods are able to identify ALP activity, but these do not allow for easy quantification. ALP activity assays utilize p-nitrophenyl phosphate as a substrate that is hydrolyzed to produce p-nitrophenol, resulting in a yellow end product measurable at an absorbance of 405 nm wavelength [131]. ALP is a downstream target gene of both RUNX2 and

OSX, and its expression is also induced by dexamethasone, BMP-2, and vitamin D3 [132]. ALP activity increases during in vitro osteogenesis, with low levels apparent by 1 week and a 2 – 6 fold increase at later time points [132]; however, its expression decreases when mineralization is advanced [130]. There is often a poor correlation between ALP gene expression and enzymatic activity levels in stem cell osteogenesis studies. One study found that levels of ALP activity are not proportional to observed mineralization levels, which is considered the ultimate endpoint of in vitro osteogenesis [132]. ALP is necessary, but not sufficient, to produce mineralized matrix, as some bone marrow MSC cultures in growth media can exhibit high ALP levels but show a

23 delay in mineralization compared to dexamethasone-treated cultures [133]. Other bone marrow MSC cultures produce high levels of ALP without ever mineralizing [134]. Therefore, the correlation between ALP activity and mineralization is dependent on the source and quality of the MSCs regarding their proportion of primary osteoblasts, and not their inherent ability to express ALP.

Alizarin Red Alizarin Red staining is commonly used as a means to visualize mineralization as an endpoint of in vitro osteogenesis. Alizarin Red is an anthraquinone derivative that forms a complex with calcium, manganese, barium, strontium, and iron ions through chelation [135]. In MSC cultures, calcium ions are the only substrates present at a sufficient levels for detection with staining. Alizarin Red stain can be extracted from fixed cells using a protocol involving acetic acid and ammonium hydroxide, which is then quantified by absorbance at 405 nm wavelength [136]. A positive stain results in calcium salts appearing deep orange or red under plain or compensated polarized light microscopy. The main benefit of Alizarin Red staining is that it is a simple and fast protocol that works on fixed tissue, fluid, and cell culture samples. However, there is a high rate of false positives [137, 138], and quantitative measurements of stain uptake are not particularly sensitive as clear increases in Alizarin Red binding are generally only detectable in very late stages (14 – 21 days) of osteogenic cultures. There is also a high amount of background and sequestered stain sometimes present in cellular aggregates, making subjective interpretation of the “positivity” of the staining results difficult. Alizarin Red staining can also be strongly “positive” in cultures undergoing cell death due to release of intracellular calcium ions that can be absorbed into the surrounding matrix.

Von Kossa Von Kossa staining also detects matrix mineralization by binding to phosphate in the matrix. The staining reaction relies on silver nitrate reacting with phosphate ions in an acidic environment to produce a precipitate [139]. Photochemical degradation of the silver phosphate precipitate results in a black, metallic silver product. While the stain does not directly detect calcium, the silver essentially replaces the calcium in the calcium phosphate complex. There have been several modifications of the original protocol to better optimize its use [140, 141].

Although von Kossa staining has been increasingly used in osteogenic cultures to detect “calcium” in mineralized matrix, it has been found to be insufficient to confirm that mineralization in vitro actually reflects in vivo bone matrix-forming capacity [113]. Bonewald et al showed that cultured cells failed to form hydroxyapatite mineral even after 30 days in culture using Fourier-transform infrared spectroscopy (FT-IR) and electron microscopy (EM) analyses, despite being strongly positive for von Kossa staining. Similar to Alizarin Red, there is a high likelihood of obtaining false positive results [142].

Calcium Quantification Matrix calcium quantification can be performed using o-cresolphthalein complexone, which binds to calcium ions to form a chromophore in alkaline medium [143]. The compound 8- hydroxyquinoline is then added to reduce any interference by magnesium, and the absorbance is measured at 570 nm wavelength [135]. Other methods of quantifying calcium deposition include fluorescent dyes, such as calcein, coelenterazine, dehydrocalcein fluo-3, fura-2, indo-1, and rhod-2 [144], or colorimetric kit assays [145]. The major concern with calcium quantification assays is that they are not specific for the type of calcium detected, and the quantification of calcium content in a cell layer does not allow for accurate evaluation of mineralization [145]. As with mineral staining procedures, false positives can exist from things such as apoptotic bodies producing non-apatitic calcification [146]. The determination of the ‘calcium to phosphate’ ratio is more important than calcium quantification alone, since human bone has a ratio of 1.63 [147]. Analysis of the matrix structure through advanced imaging methods, such as x-ray diffraction, FTIR, and EM, is necessary to truly confirm hydroxyapatite secretion and exclude other calcium phosphate-based compounds in the matrix [148].

Improving Osteogenic Differentiation of MSCs Comparative Osteogenic Potentials Bone marrow MSCs (BM-MSCs) were the first source of MSCs derived from human samples shown to possess multi-lineage differentiation potential [79], yet their cell yield, long expansion time, and earlier senescence risk are disadvantageous compared to MSCs from other sources, such as adipose tissue (ADI-MSCs) [149]. However, there is extensive literature documenting the superior osteogenic capabilities in BM-MSCs compared to other donor-

25 matched progenitor populations [149 – 156]. This is hardly surprising, given the location of BM- MSCs within the actively remodeling osseous environment of the medullary cavity. ADI-MSCs showed delayed collagen type I deposition, ALP activity, and osteogenic gene and protein expression compared to BM-MSCs [151]. In another study, BM-MSCs showed superior ALP activity and von Kossa staining compared to ADI-MSCs collected from non-donor matched samples in older patients [152]. BM-MSCs have also shown accelerated and/or superior osteogenic differentiation capabilities compared to ADI-MSCs in rats [153], pigs [154], and horses [155, 156], while results in dogs have been mixed [57, 158]. In animal model studies, the results are largely inconclusive regarding which cell source has superior osteogenic potential in vivo [150]. Excluding donor-specific differences in cell proliferation and differentiation capabilities, ADI- MSCs on their own likely cannot replace BM- MSCs for osteogenic purposes. Therefore, several approaches can be employed to improve the osteogenic capacity of MSCs from less osteogenic sources by capitalizing on normal osteogenesis regulatory processes.

The BMP-RUNX2-OSX Regulatory Axis The critical importance of BMP signaling and activities of the master osteogenic transcription factors RUNX2 and OSX make them ideal targets for improving bone formation both in vitro and in vivo. Both canonical and non-canonical BMP signaling activities are necessary for bone formation [28]. Canonical BMP signaling involves BMP ligand binding to receptor type II and type I, which results in formation of a heteromeric complex of serine/threonine kinase receptors. Activated type I receptors initiate intracellular signaling through phosphorylation of Smad proteins (1/5/8). Activated Smads form a complex with Smad- 4, translocate into the nucleus, and bind to Smad-Binding Elements (SBE) on target genes to direct transcriptional responses. Two direct transcriptional targets of canonical BMP signaling in osteogenesis are RUNX2 and DLX-5 [28]. In contrast, non-canonical BMP signaling is not Smad-dependent, but involves transforming growth factor-beta-activated kinase-1 (TAK1), a component of a MAP kinase pathway [159]. The TAK1/MAPK pathway mediates phosphorylation of RUNX2, increasing the association with its coactivator, CREB-binding protein, and regulating expression of downstream osteoblast-specific genes [160]. The ligands BMP-2, -4, -5, -6, and -7 all have strong osteogenic capabilities, and BMP-2 especially has been found to be necessary and sufficient for osteo-chondral differentiation of MSCs [161].

BMP autocrine signaling is required for RUNX2-dependent osteogenesis, as the pathways work synergistically to stimulate osteoblast gene expression [162, 163]. In addition to being a direct transcriptional target of BMP signaling, RUNX2 expression is also indirectly induced through DLX-5 [164]. RUNX2 acts upstream of the other master regulator of osteogenesis, OSX, [24], but OSX induction by BMP signaling is not mandatorily dependent on RUNX2 [165 – 167 ]. RUNX2-independent induction of OSX by BMPs is mediated through DLX-5 and MSX-2, as blocking of either factor results in complete absence of BMP-induced OSX expression [165 – 167]. In RUNX2-deficient cells, OSX was able to regulate expression of genes not controlled by RUNX2, suggesting that at least some of their osteogenic activities and target genes are distinct [167]. Another OSX regulatory factor, protein kinase B (PKB / AKT), phosphorylates OSX and DLX-5, which results in increased protein stability and osteogenic transcriptional activity [168, 169]. In summary, there is a complex interplay between BMP signaling, RUNX2, and OSX regulation during osteogenesis. While all three factors have been found to be essential for complete osteogenesis, it appears that BMP signaling, being the furthest upstream, influences the osteogenic capabilities of a cell most comprehensively and therefore is an ideal target for therapeutic approaches.

BMP Signaling Enhancement Given the importance of BMP signaling in both bone formation and repair [170], a common approach for improving osteogenic capacity of cells, both in vitro and in vivo, is potentiating the BMP signaling cascade. Two widely used methods to accomplish this include supplementing cells with exogenous recombinant BMP ligand or increasing BMP expression levels through gene delivery [171]. There are other strategies that help activate and maintain the BMP signaling response, such as neutralizing extracellular BMP antagonists, inactivating endogenous inhibitors, and regulating Smad protein modifications with small molecule inhibitors [172]. The Food and Drug Administration (FDA) has approved two recombinant human BMP products for clinical applications: InFuse/BMP-2 (Medtronic) and OP-1/BMP-7 (Stryker Biotech). They were initially approved for spinal fusions and fracture non-unions but have since been used extensively off-label for many orthopedic procedures [173, 174]. However, their use in spinal fusion procedures has been associated with complication rates of 10 – 50%, depending

27 on the surgical approach, which includes adverse events such as ectopic bone formation, osteolysis, implant displacement, and infection [175]. Exogenous BMP administration in vitro variably improves osteogenic differentiation capabilities of ADI-MSCs, depending on the specific ligand, the concentration used, differentiation media additives, and administration time point [175]. Very high concentrations of BMP-2 (300 – 500 ng/ml) can induce osteogenic differentiation without additional media additives in ADI-MSCs, yet mid-range doses (5 – 200 ng/ml) require ascorbic acid and β- glycerophosphate in the media [176]. The optimal effect of BMP-2 in ADI-MSC in vitro osteogenesis was seen at a dose of 100 ng/ml [176]. Gene technology approaches to increasing BMP signaling attempt to avoid the costs and side effects associated with recombinant BMP use [177]. Common methods include viral or non-viral vector expression of BMP ligands and other key signaling components, and gene knockdown of BMP antagonists. There have been many in vivo and ex vivo gene therapy studies performed in animal models utilizing various BMP ligand over-expression modalities, and the results are mostly favorable in their ability to improve bone regeneration [reviewed in 178 – 179]. However, the issue regarding the cost-effectiveness of recombinant BMP ligand application under clinical conditions is far less certain. Ongoing research is focusing on temporally and spatially regulated gene expression systems to control the timing, localization, and extended delivery of BMP activity, which should further improve safety and efficacy over current recombinant protein-based therapies [180].

Direct RUNX2 & OSX Expression Both RUNX2 and OSX are necessary for osteoblast differentiation [19, 20, 24]. MSCs are directed toward osteoblast progenitor differentiation initially by RUNX2 and Ihh, while OSX and canonical Wnt signaling direct RUNX2-positive osteoblast progenitors into an immature osteoblastic phenotype [181]. The roles of RUNX2 and OSX in the maturation of osteoblasts are less well-established, but both can inhibit late osteoblast differentiation and transformation of mature osteoblasts to osteocytes [127, 129, 181]. Therefore, sustained over-expression of RUNX2 and OSX may not be beneficial for stimulating complete osteogenesis, and inducible or transient expression strategies are likely to be more effective approaches [182]. The following summarizes studies that have addressed RUNX2 and OSX over-expression, alone or in combination with other factors.

RUNX2 adenoviral-directed over-expression in MSCs was found to enhance osteogenic differentiation and upregulate osteoblast-specific in vitro and improve healing in vivo in murine calvarial [183] and femoral [184] critical-size defects. RUNX2 adenovirus injected directly into femoral bone marrow induced osteogenesis, suggesting that direct approaches could be advantageous over ex vivo gene therapy protocols [185]. Retroviral-directed over-expression of both RUNX2 and mitogen-activated protein kinase (MAPK / MKP-1) effectively trans- differentiated pre-adipocytes into fully differentiated osteoblasts [186]. Compared to BMP-2 over-expression, transplanted fibroblasts expressing RUNX2 were not as effective at differentiating into osteoblasts in vivo [187]. However, co-transfection of BMP-2 and RUNX2 into ADI-MSCs increased osteoblast differentiation in vitro and improved bone regeneration in vivo over BMP-2 transfection alone [188]. In contrast, RUNX2 over-expression in transgenic murine models has been associated with bone loss due to increased osteoclast differentiation [189] and osteoblastic MMP expression [190]. There is also concern over the role of RUNX2 as an oncogene in osteosarcoma and bone metastases [191]. Therefore, while the majority of research shows promise utilizing RUNX2 over-expression to improve osteogenic outcomes, caution must be taken as to the methodology and timing of the applications used. OSX over-expression in murine embryonic stem cells was able to induce osteoblast differentiation by upregulating osteoblast-specific genes, and, in combination with osteogenic media additives, increase bone nodule formation [192]. Retroviral OSX over-expression in murine BM-MSCs enhanced osteoblast proliferation, differentiation, and mineralization in vitro [193]. A follow-up study used ex vivo OSX transfection of BM-MSCs implanted into a mouse calvarial critical defect model; the OSX vector stimulated five times more bone formation compared to empty vector controls [194]. In cells that are not predisposed to osteoblast lineage differentiation (NIH3T3 fibroblasts), OSX over-expression enhanced proliferation, but the expression of most osteoblast-specific genes and mineralization were not increased, suggesting that OSX alone is insufficient to establish osteogenic lineage commitment in cells not already ‘inclined’ towards osteogenesis [195]. In an in vivo rabbit model of mandibular lengthening, BM-MSCs transfected with recombinant OSX plasmids enhanced callus formation and bone deposition during distraction osteogenesis compared to non-transfected BM-MSCs [196]. OSX over-expression in transgenic models increased osteoblast proliferation, yet it inhibited their late stage differentiation [129]. Of particular concern, the same study found that the OSX transgenic

29 mice developed osteopenia and increased fracture risk, due to collagen fiber orientation abnormalities resulting from impaired osteoblast differentiation [129]. As with RUNX2, further research is required to establish optimal timing and, perhaps as importantly, cessation of OSX over-expression in bone regeneration applications.

Conclusions There are many modalities utilized in bone regenerative medicine research that attempt to mimic and enhance the natural processes occurring during osteogenesis. Mesenchymal stem cells are the necessary cell source responsible for osteoblast differentiation, which must occur successfully in order to coordinate healing and repair of bone. The process of osteogenesis requires a tightly regulated orchestration by master transcription factors RUNX2 and OSX and multiple signaling pathways, most importantly through the TGF-β/BMP network. MSCs show strong potential for use in bone regenerative medicine, but their collection, propagation, and in vitro processing can vary significantly between species, making comparative analyses and clinical translation difficult to interpret accurately. Methods to improve the osteogenic capacity of MSC populations commonly include growth factor(s) supplementation and gene therapy techniques, both ex vivo and in vivo. There has been considerable success with exogenous BMP supplementation in various orthopedic surgical conditions, and other modalities of BMP signaling enhancement that are becoming more prevalent. Direct over-expression of RUNX2 or OSX in MSCs has yielded some improvement in osteogenic outcomes, but this approach appears to be less effective overall compared to BMP supplementation and can have detrimental effects depending on the technique used. Combinatorial approaches that enhance both BMP signaling and RUNX2 or OSX expression in MSCs may yield the best results, but optimization of methodology regarding doses and coordinated timing of effects are required prior to clinical use.

References 1. Robling, A.G., Castillo, A.B., Turner, C.H. (2006) Biomechanical and molecular regulation of bone remodeling. Annu Rev Biomed Eng. 8: 455-498.

2. Datta, H.K., Ng, W.F., Walker, J.A., et al (2008) The cell biology of bone metabolism. J Clin Pathol. 61(5): 577-587.

3. Clarke, B. (2008) Normal bone anatomy and physiology. Clin J Am Soc Nephrol. 3(Supp): S131-S139.

4. Currey, J.D. (2003) Role of collagen and other organics in the mechanical properties of bone. Osteoporos Int. 14 (Supp5): S29-S36.

5. Lefebvre, V., Bhattaram, P. (2010) Vertebrate skeletogenesis. Curr Top Dev Biol. 90: 291- 317.

6. Shum, L., Coleman, C.M., Hatekeyama, Y., Tuan, R.S. (2003) Morphogenesis and dysmorphogenesis of the appendicular skeleton. Birth Defects Res C Embryo Today. 69(2): 102-122.

7. Eames, B.F., Sharpe, P.T., Helms, J.A. (2004) Hierarchy revealed in the specification of three skeletal fates by Sox9 and Runx2. Dev Biol. 274(1): 188-200.

8. Butterfield, N.C., McGlinn, E., Wicking, C. (2010) The molecular regulation of vertebrate limb patterning. Curr Top Dev Biol. 90: 319-341.

9. Mackie, E.J., Ahmed, Y.A., Tatarczuch, L., et al (2008) Endochondral ossification: how cartilage is converted into bone in the developing skeleton. Int J Biochem Cell Biol. 40(1): 46-62.

10. St-Jacques, B., Hammerschmidt, M., McMahon, A.P. (1999) Indian hedgehog signaling regulates proliferation and differentiation of chondrocytes and is essential for bone formation. Genes Dev. 13(16): 2072-2086.

11. Zelzer, E., Mamluk, R., Ferrara, N., et al (2004) VEGFA is necessary for chondrocyte survival during bone development. Development. 131(9): 2161-2171.

12. Stickens, D., Behonick, D.J., Ortega, N., et al (2004) Altered endochondral bone development in matrix metalloproteinase 13-deficient mice. Development. 131(23): 5883- 5895.

13. Opperman, L.A. (2000) Cranial sutures as intramembranous bone growth sites. Dev Dyn. 219(4): 472-485.

14. Capulli, M., Paone, R., Rucci, N. (2014) Osteoblast and osteocyte: games without frontiers. Arch Biochem Biophys. 561: 3-12.

15. Jensen, E.D., Gopalakrishnan, R., Westendorf, J.J. (2010) Regulation of gene expression in osteoblasts. Biofactors. 36(1): 25-32.

16. Florencio-Silva, R., Sasso, G.R., Sasso-Cerri, E., et al (2015) Biology of Bone Tissue: Structure, Function, and Factors That Influence Bone Cells. Biomed Res Int. 2015: 421746.

17. Anderson, H.C. (2003) Matrix vesicles and calcification. Curr Rheumatol Rep. 5(3): 222- 226.

18. Karsenty, G., Kronenberg, H.M., Settembre, C. (2009) Genetic control of bone formation. Annu Rev Cell Dev Biol. 25: 629-648.

19. Komori, T., Yagi, H., Nomura, S., et al (1997) Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell. 89(5): 755-764.

20. Otto, F., Thornell, A.P., Crompton, T., et al (1997) Cbfa1, a candidate gene for cleidocranial dysplasia syndrome, is essential for osteoblast differentiation and bone development. Cell. 89(5): 765-771.

21. Bruderer, M., Richards, R.G., Alini, M., Stoddart, M.J. (2014) Role and regulation of RUNX2 in osteogenesis. Eur Cell Mater. 28: 269-286.

22. Ducy, P., Zhang, R., Geoffroy, V., et al (1997) Osf2/Cbfa1: a transcriptional activator of osteoblast differentiation. Cell. 89(5): 747-754.

23. Takeda, S., Bonnamy, J.P., Owen, M.J., et al (2001) Continuous expression of Cbfa1 in nonhypertrophic chondrocytes uncovers its ability to induce hypertrophic chondrocyte differentiation and partially rescues Cbfa1-deficient mice. Genes Dev. 15: 467–481.

24. Nakashima, K., Zhou, X., Kunkel, G., et al (2002) The novel zinc finger-containing transcription factor osterix is required for osteoblast differentiation and bone formation. Cell 108(1): 17-29.

25. Komori, T. (2005) Regulation of skeletal development by the Runx family of transcription factors. J Cell Biochem 95(3): 445-453.

26. Nishio, Y., Dong, Y., Paris, M., et al (2006) Runx2-mediated regulation of the zinc finger Osterix/Sp7 gene. Gene. 372: 62–70.

27. Sinha, K.M, Zhou, X. (2013) Genetic and molecular control of osterix in skeletal formation. J Cell Biochem. 114(5): 975-984.

28. Chen, G., Deng, C., Li, Y.P. (2012) TGF-β and BMP signaling in osteoblast differentiation and bone formation. Int J Biol Sci. 8(2): 272-288.

29. Nishimura, R., Hata, K., Matsubara, T., Wakabayashi, M., Yoneda, T. (2012) Regulation of bone and cartilage development by network between BMP signaling and transcription factors. J Biochem. 151(3): 247-254.

30. Derynck, R., Akhurst, R.J. (2007) Differentiation plasticity regulated by TGF-beta family proteins in development and disease. Nat Cell Biol. 9(9):1000-1004.

31. Centrella, M., McCarthy, T.L., Canalis, E. (1988) Skeletal tissue and transforming growth factor beta. FASEB J. 2(15): 3066-3073.

32. Katagiri, T., Takahashi, N. (2002) Regulatory mechanisms of osteoblast and osteoclast differentiation. Oral Dis. 8(3): 147-159.

33. Urist, M.R. (1965) Bone: formation by autoinduction. Science. 150(3698): 893-899.

34. Yamaguchi, A., Komori, T., Suda, T. (2000) Regulation of osteoblast differentiation mediated by bone morphogenetic proteins, hedgehogs, and Cbfa1. Endocr Rev. 21(4): 393- 411.

35. Yamaguchi, A., Katagiri,T., Ikeda, T., et al (1991) Recombinant human bone morphogenetic protein-2 stimulates osteoblastic maturation and inhibits myogenic differentiation in vitro. J Cell Biol. 113(3): 681-687.

36. Rosen, V., Nove, J., Song, J.J., et al (1994) Responsiveness of clonal limb bud cell lines to bone morphogenetic protein 2 reveals a sequential relationship between cartilage and bone cell phenotypes. J Bone Miner Res. 9(11): 1759-1768.

37. Day, T.F., Yang, Y. (2008) Wnt and hedgehog signaling pathways in bone development. J Bone Joint Surg Am. 90 (Suppl 1): 19-24.

38. Day, T.F., Guo, X., Garrett-Beal, L., et al (2005) Wnt/beta-catenin signaling in mesenchymal progenitors controls osteoblast and chondrocyte differentiation during vertebrate skeletogenesis. Dev Cell. 8: 739-750.

39. Hill, T.P., Taketo, M.M., Birchmeier, W., Hartmann, C. (2006) Multiple roles of mesenchymal beta-catenin during murine limb patterning. Development. 133: 1219-1229.

40. Du, X., Xie, Y., Xian, C.J., Chen, J. (2012) Role of FGFs/FGFRs in skeletal development and bone regeneration. J Cell Physiol. 227(12): 3731-3743.

41. Bosetti, M., Boccafoschi, F., Leigheb, M., Canna, M.F. (2007) Effect of different growth factors on human osteoblasts activities: a possible application in bone regeneration for tissue engineering. Biomol Eng. 24(6): 613-618.

42. Hayrapetyan, A., Jansen, J.A., van den Beucken, J.J. (2015) Signaling pathways involved in osteogenesis and their application for bone regenerative medicine. Tissue Eng Part B Rev. 21(1): 75-87.

43. Einhorn, T.A., Gerstenfeld, L.C. (2015) Fracture healing: mechanisms and interventions. Nat Rev Rheumatol. 11(1): 45-54.

44. Oryan, A., Alidadi, S., Moshiri, A., Maffulli, N. (2014) Bone regenerative medicine: classic options, novel strategies, and future directions. J Orthop Surg Res. 9(1): 18.

45. Khan, S.N., Cammisa Jr., F.P., Sandhu, H.S., et al (2005) The biology of bone grafting. J Am Acad Orthop Surg. 13(1): 77-86.

46. Fisher, D.M., Wong, J.M., Crowley, C., Khan, W.S. (2013) Preclinical and clinical studies on the use of growth factors for bone repair: a systematic review. Curr Stem Cell Res Ther. 8(3): 260-268.

47. Roffi, A., Di Matteo, B., Krishnakumar, G.S., et al (2017) Platelet-rich plasma for the treatment of bone defects: from pre-clinical rational to evidence in the clinical practice. A systematic review. Int Orthop. 41(2): 221-237.

48. Virk, M.S., Lieberman, J.R. (2012) Biologic adjuvants for fracture healing. Arthritis Res Ther. 14(6): 225.

49. Agarwal, R., García, A.J. (2015) Biomaterial strategies for engineering implants for enhanced osseointegration and bone repair. Adv Drug Deliv Rev. 94: 53-62.

50. Crowley, C., Wong, J.M., Fisher, D.M., Khan, W.S. (2013) A systematic review on preclinical and clinical studies on the use of scaffolds for bone repair in skeletal defects. Curr Stem Cell Res Ther. 8(3): 243-252

51. Tarassoli, P., Khan, W.S., Hughes, A., Heidari, N. (2013) A review of techniques for gene therapy in bone healing. Curr Stem Cell Res Ther. 8(3): 201-209.

52. Ambikaipalan, A., Wong, J.M., Khan, W.S. (2013) Preclinical and clinical studies on the use of stem cells for bone repair: a systematic review. Curr Stem Cell Res Ther. 8(3): 210-216.

53. Jakoi, A.M., Iorio, J.A., Cahill, P.J. (2015) Autologous bone graft harvesting: a review of grafts and surgical techniques. Musculoskelet Surg. 99(3): 171-1788.

54. Goulet, J.A., Senunas, L.E., DeSilva, G.L., Greenfield, M.L. (1997) Autogenous iliac-crest bone graft. Complications and functional assessment. Clin Orthop Relat Res. 1997(339): 76- 81.

55. Fisher, D.M., Wong, J.M., Crowley, C., Khan, W.S. (2013) Preclinical and clinical studies on the use of growth factors for bone repair: a systematic review. Curr Stem Cell Res Ther. 8(3): 260-268.

56. Carreira, A.C., Zambuzzi, W.F., Rossi, M.C., et al (2015) Bone Morphogenetic Proteins: Promising Molecules for Bone Healing, Bioengineering, and Regenerative Medicine. Vitam Horm. 99: 293-322.

57. Bibbo, C., Nelson, J., Ehrlich, D., Rougeux, B. (2015) Bone morphogenetic proteins: indications and uses. Clin Podiatr Med Surg. 32(1): 35-43.

58. Garrison, K.R., Shemilt, I., Donell, S., et al (2010) Bone morphogenetic protein (BMP) for fracture healing in adults. Cochrane Database Syst Rev. 2010(6): CD006950.

59. Begam, H., Nandi, S.K., Kundu, B., Chanda, A. (2017) Strategies for delivering bone morphogenetic protein for bone healing. Mater Sci Eng C Mater Biol Appl. 70(Pt 1): 856- 869.

60. Poniatowski, L.A., Wojdasiewicz, P., Gasik, R., Szukiewicz, D. (2015) Transforming growth factor Beta family: insight into the role of growth factors in regulation of fracture healing biology and potential clinical applications. Mediators Inflamm. 2015: 137823.

61. Fei, Y., Gronowicz, G., Hurley, M.M. (2013) Fibroblast growth factor-2, bone homeostasis and fracture repair. Curr Pharm Des. 19(19): 3354-3363.

62. Kawai, M., Rosen, C.J. (2012) The insulin-like growth factor system in bone: basic and clinical implications. Endocrinol Metab Clin North Am. 41(2): 323-333.

63. Hu, K., Olsen, B.R. (2016) The roles of vascular endothelial growth factor in bone repair and regeneration. Bone. 91: 30-38.

64. Shah, P., Keppler, L., Rutkowski, J. (2014) A review of platelet derived growth factor playing pivotal role in bone regeneration. J Oral Implantol. 40(3): 330-430.

65. Takahata, M., Awad, H.A., O’Keefe, R.J., et al (2012) Endogenous tissue engineering: PTH therapy for skeletal repair. Cell Tissue Res. 347(3): 545-552.

66. Vo, T.N., Kasper, F.K., Mikos, A.G. (2012) Strategies for controlled delivery of growth factors and cells for bone regeneration. Adv Drug Deliv Rev. 64(12): 1292-1309.

67. Roffi, A., Di Matteo, B., Krishnakumar, G.S., et al (2017) Platelet-rich plasma for the treatment of bone defects: from pre-clinical rational to evidence in the clinical practice. A systematic review. Int Orthop. 41(2): 221-237.

68. Virk, M.S., Lieberman, J.R. (2012) Biologic adjuvants for fracture healing. Arthritis Res Ther. 14(6): 225.

69. Tarassoli, P., Khan, W.S., Hughes, A., Heidari, N. (2013) A review of techniques for gene therapy in bone healing. Curr Stem Cell Res Ther. 8(3): 201-209.

70. Evans, C.H. (2010) Gene therapy for bone healing. Expert Rev Mol Med. 12: e18. Evans, C.H. (2012) Gene delivery to bone. Adv Drug Deliv Rev. 64(12): 1331-1340.

71. Evans, C.H. (2012) Gene delivery to bone. Adv Drug Deliv Rev. 64(12): 1331-1340.

72. Marie, P.J. (2011) Cell and gene therapy for bone repair. Osteoporos Int. 22(6): 2023-2036.

73. Wang, X., Wang, Y., Gou, W., et al (2013) Role of mesenchymal stem cells in bone regeneration and fracture repair: a review. Int Orthop. 37(12): 2491–2498.

74. Watson, L., Elliman, S.J., Coleman, C.M. (2014) From isolation to implantation: a concise review of mesenchymal stem cell therapy in bone fracture repair. Stem Cell Res Ther. 5(2): 51.

75. Rosset, P., Deschaseaux, F., Layrolle, P. (2014) Cell therapy for bone repair. Orthop Traumatol Surg Res. 100(1 Suppl): S107-S112.

76. Mamidi, M.K., Dutta, S., Bhonde, R., et al (2014) Allogeneic and autologous mode of stem cell transplantation in regenerative medicine: which way to go? Med Hypotheses. 83(6): 787- 791.

77. Fei, Y., Xu, R.H., Hurley, M.M. (2012) Stem cell-based bone repair. Am J Stem Cells. 1(2): 106-113.

78. Bianco, P. (2014) "Mesenchymal" stem cells. Annu Rev Cell Dev Biol. 30: 677-704.

79. Pittenger, M.F., Mackay, A.M., Beck, S.C., et al (1999) Multilineage potential of adult human mesenchymal stem cells. Science. 284(1999): 143-147.

80. Stewart, M.C. & Stewart, A.A. (2011) Mesenchymal Stem Cells: Characteristics, Sources, and Mechanisms of Action. Vet Clin North Am Equine Pract. 27(2): 243-261.

81. Javazon, E.H., Beggs, K.J., Flake, A.W. (2004) Mesenchymal stem cells: paradoxes of passaging. Exp Hematol, 32(5): 414-425.

82. Dominici, M., Le Blanc, K., Mueller, I., et al (2006) Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy. 8(4): 315-317.

83. Lv, F.J., Tuan, R.S., Cheung, K.M., Leung, V.Y. (2014) Concise review: the surface markers and identity of human mesenchymal stem cells. Stem Cells. 32(6): 1408-1419.

84. De Schauwer, C., Meyer, E., Van de Walle, G.R., & Van Soom, A. (2011) Markers of stemness in equine mesenchymal stem cells: a plea for uniformity. Theriogenology. 75(8): 1431-1443.

85. da Silva Meirelles, L., Chagastelles, P.C., Nardi, N.B. (2006) Mesenchymal stem cells reside in virtually all post-natal organs and tissues. J Cell Sci. 119 (Pt 11): 2204-2213.

86. Kolf, C.M., Cho, E., Tuan, R.S. (2007) Biology of adult mesenchymal stem cells: regulation of niche, self-renewal and differentiation. Arthritis Res Ther. 9(1): 204

87. Otto, W.R., Wright, N.A. (2011) Mesenchymal stem cells: from experiment to clinic. Fibrogenesis Tissue Repair. 4: 20.

88. Sharma, R.R., Pollock, K., Hubel, A., McKenna, D. (2014) Mesenchymal stem or stromal cells: a review of clinical applications and manufacturing practices. Transfusion. 54(5): 1418-1437

89. Mosna, F., Sensebe, L., & Krampera, M. (2010) Human bone marrow and adipose tissue mesenchymal stem cells: a user’s guide. Stem Cells Dev. 19(10): 1449-1470.

90. Penfornis, P. & Pochampally, R. (2011) Isolation and expansion of mesenchymal stem cells/multipotential stromal cells from human bone marrow. Methods Mol Biol. 698: 11-21.

91. Nardi, N.B. & Camassola, M. (2011) Isolation and culture of rodent bone marrow-derived multipotent mesenchymal stromal cells. Methods Mol Biol. 698: 151-60.

92. Sullivan, M.O., Gordon-Evans, W.J., Fredericks, L.P., et al (2016) Comparison of Mesenchymal Stem Cell Surface Markers from Bone Marrow Aspirates and Adipose Stromal Vascular Fraction Sites. Front Vet Sci. 2: 82.

93. Webb, T.L., Quimby, J.M., Dow, S.W. (2012) In vitro comparison of feline bone marrow- derived and adipose tissue-derived mesenchymal stem cells. J Feline Med Surg. 14(2): 165- 168.

94. Taylor, S.E. & Clegg, P.D. (2011) Collection and Propagation Methods for Mesenchymal Stromal Cells. Vet Clin North Am Equine Pract. 27(2): 263-274.

95. Cortes, Y., Ojeda, M., Araya, D., et al (2013) Isolation and multilineage differentiation of bone marrow mesenchymal stem cells from abattoir-derived bovine fetuses. BMC Vet Res. 9: 133.

96. Fortier, L.A., Travis, A.J. (2011) Stem cells in veterinary medicine. Stem Cell Res Ther. 2(1): 9.

97. Zuk, P.A., Zhu, M., Ashjian, P., et al (2002) Human adipose tissue is a source of multipotent stem cells. Mol Biol Cell. 13(12): 4279-4295.

98. Lopez, M.J., Spencer, N.D. (2011) In vitro adult rat adipose tissue-derived stromal cell isolation and differentiation. Methods Mol Biol. 702: 37-46.

99. Bahamondes, F., Flores, E., Cattaneo, G., et al (2017) Omental adipose tissue is a more suitable source of canine Mesenchymal stem cells. BMC Vet Res.13(1): 166

100. Marx, C., Silveira, M.D., & Beyer Nardi, N. (2015) Adipose-derived stem cells in veterinary medicine: characterization and therapeutic applications. Stem Cells Dev. 24(7): 803-813.

101. Sotiropoulou, P.A., Perez, S.A., Salagianni, M. et al (2006) Characterization of the optimal culture conditions for clinical scale production of human mesenchymal stem cells. Stem Cells. 24: 462–471.

102. Lennon, D.P., Caplan, A.I. (2006). Isolation of human marrow-derived mesenchymal stem cells. Exp Hematol. 34: 1604–1605.

103. Gruber, H.E., Somayaji, S., Riley, F., et al (2012) Human adipose-derived mesenchymal stem cells: serial passaging, doubling time and cell senescence. Biotech Histochem. 87(4): 303-311.

104. Vidal, M.A., Walker, N.J., Napoli, E., Borjesson, D.L. (2012) Evaluation of senescence in mesenchymal stem cells isolated from equine bone marrow, adipose tissue, and umbilical cord tissue. Stem Cells Dev. 21(2): 273-283.

105. Castro-Malaspina, H., Gay, R.E., Resnick, G., et al (1980) Characterization of human bone marrow fibroblast colony-forming cells (CFU-F) and their progeny. Blood. 56(2): 289- 301.

106. Lavrentieva, A., Hatlapatka, T., Neumann, A., et al (2013) Potential for osteogenic and chondrogenic differentiation of MSC. Adv Biochem Eng Biotechnol. 129: 73-88.

107. Song, I., Kim, B.S., Kim, C.S., Im, G.I. (2011) Effects of BMP-2 and vitamin D3 on the osteogenic differentiation of adipose stem cells. Biochem Biophys Res Commun. 408(1): 126–131.

108. Song, I.H., Caplan, A.I., Dennis, J.E. (2009) Dexamethasone inhibition of confluence- induced apoptosis in human mesenchymal stem cells. J Orthop Res. 27: 216–221.

109. Langenbach, F., Handschel, J. (2013) Effects of dexamethasone, ascorbic acid and β- glycerophosphate on the osteogenic differentiation of stem cells in vitro. Stem Cell Res Ther. 4(5): 117.

110. Jaiswal, N., Haynesworth, S.E., Caplan, A.I., Bruder, S.P. (1997) Osteogenic differentiation of purified, culture-expanded human mesenchymal stem cells in vitro. J Cell Biochem. 64(2): 295-312.

111. Schwarz, R.I., Kleinman, P., Owens, N. (1987) Ascorbate can act as an inducer of the collagen pathway because most steps are tightly coupled. Ann N Y Acad Sci. 498: 172–185.

112. Xiao, G., Gopalakrishnan, R., Jiang, D., et al (2002) Bone morphogenetic proteins, extracellular matrix, and mitogen-activated protein kinase signaling pathways are required for osteoblast-specific gene expression and differentiation in MC3T3-E1 cells. J Bone Miner Res.17: 101–110.

113. Bonewald, L.F., Harris, S.E., Rosser, J., et al (2003) von Kossa staining alone is not sufficient to confirm that mineralization in vitro represents bone formation. Calcif Tissue Int. 72(5): 537-547.

114. Hamidouche, Z., Hay, E., Vaudin, P., et al (2008) FHL2 mediates dexamethasone- induced mesenchymal cell differentiation into osteoblasts by activating Wnt/beta-catenin signaling-dependent Runx2 expression. Faseb J. 22: 3813–3822.

115. Hong, D., Chen, H.X., Xue, Y., et al (2009) Osteoblastogenic effects of dexamethasone through upregulation of TAZ expression in rat mesenchymal stem cells. J Steroid Biochem Mol Biol. 116(1-2): 86-92.

116. Phillips, J.E., Gersbach, C.A., Wojtowicz, A.M., García, A.J. (2006) Glucocorticoid- induced osteogenesis is negatively regulated by Runx2/Cbfa1 serine phosphorylation. J Cell Sci. 119(Pt 3): 581-591.

117. Oshina, H., Sotome, S., Yoshii, T., et al (2007) Effects of continuous dexamethasone treatment on differentiation capabilities of bone marrow-derived mesenchymal cells. Bone. 41(4): 575-583.

118. Wang, H., Pang, B., Li, Y., et al (2012) Dexamethasone has variable effects on mesenchymal stromal cells. Cytotherapy. 14(4): 423-430.

119. Walsh, S., Jordan, G.R., Jefferiss, C., et al (2001) High concentrations of dexamethasone suppress the proliferation but not the differentiation or further maturation of human osteoblast precursors in vitro: relevance to glucocorticoid-induced osteoporosis. Rheumatology. 40: 74–83.

120. Piccione, M., Durgam, S., Stewart, M. (2016) Impact of bisphosphonates on osteoprogenitor motility and differentiation. J Stem Cell Res Regen Med. 3(2): 1-9.

121. Kii, I., Amizuka, N., Shimomura, J., et al (2004) Cell-cell interaction mediated by cadherin-11 directly regulates the differentiation of mesenchymal cells into the cells of the osteo-lineage and the chondro-lineage. J Bone Miner Res. 19(11): 1840-1849.

122. Tang, J., Peng, R., Ding, J. (2010) The regulation of stem cell differentiation by cell-cell contact on micropatterned material surfaces. Biomaterials. 31(9): 2470-2476.

123. Agnew, D.J., Green, J.E., Brown, T.M., et al (2014) Distinguishing between mechanisms of cell aggregation using pair-correlation functions. J Theor Biol. 352: 16-23.

124. Fagotto, F., Gumbiner, B.M. (1996) Cell contact-dependent signaling. Dev Biol. 180(2): 445-454.

125. Orlando, B., Giacomelli, L., Ricci, M., et al (2013) Leader genes in osteogenesis: a theoretical study. Arch Oral Biol. 58(1): 42-49.

126. Piek, E., Sleumer, L.S., van Someren, E.P., et al (2010) Osteo-transcriptomics of human mesenchymal stem cells: accelerated gene expression and osteoblast differentiation induced by vitamin D reveals c-MYC as an enhancer of BMP2-induced osteogenesis. Bone. 46(3): 613-627.

127. Liu, T.M., Lee, E.H. (2013) Transcriptional regulatory cascades in Runx2-dependent bone development. Tissue Eng Part B Rev. 19(3): 254-263.

128. Zhang, C. (2012) Molecular mechanisms of osteoblast-specific transcription factor Osterix effect on bone formation. Beijing Da Xue Xue Bao Yi Xue Ban. 44(5): 659-65.

129. Yoshida, C.A., Komori, H., Maruyama, Z., et al (2012) SP7 inhibits osteoblast differentiation at a late stage in mice. PLoS One. 7(3): e32364.

130. Madras, N., Gibbs, A.L., Zhou, Y., et al (2002) Modeling stem cell development by retrospective analysis of gene expression profiles in single progenitor-derived colonies. Stem Cells. 20(3): 230-240.

131. Sabokbar, A., Millett, P.J., Myer, B., Rushton, N. (1994) A rapid, quantitative assay for measuring alkaline phosphatase activity in osteoblastic cells in vitro. Bone and Mineral. 27(1): 57-67.

132. Hoemann, C.D., El-Gabalawy, H., McKee, M.D. (2009) In vitro osteogenesis assays: influence of the primary cell source on alkaline phosphatase activity and mineralization. Pathol Biol (Paris). 57(4): 318-323.

133. Chang, P.L., Blair, H.C., Zhao, Y.W., et al (2006) Comparison of fetal and adult marrow stromal cells in osteogenesis with and without glucocorticoids. Connect Tissue Res. 47: 67- 76.

134. Evans, J.F., Yeh, J.K., Aloia, J.F. (2000) Osteoblast-like cells of the hypophysectomized rat: a model of aberrant osteoblast development. Am J Physiol Endocrinol Metab. 278: E832- E838.

135. Bruedigam, C., Driel, M., Koedam, M., et al (2011) Basic techniques in human mesenchymal stem cell cultures: differentiation into osteogenic and adipogenic lineages,

genetic perturbations, and phenotypic analyses. Curr Protoc Stem Cell Biol. Chapter 1: Unit1H.3.

136. Gregory, C.A., Gunn, W.G., Peister, A., Prockop, D.J. (2004) An Alizarin red-based assay of mineralization by adherent cells in culture: comparison with cetylpyridinium chloride extraction. Anal Biochem. 329(1): 77-84.

137. Gordon, C., Swan, A., Dieppe, P. (1989) Detection of crystals in synovial fluids by light microscopy: sensitivity and reliability. Ann Rheum Dis. 48(9): 737-742.

138. Rosenthal, A.K., Fahey, M., Gohr, C., et al (2008) Feasibility of a tetracycline-binding method for detecting synovial fluid basic calcium phosphate crystals. Arthritis Rheum. 58(10): 3270-3274.

139. Puchtler, H., Meloan, S.N. (1978) Demonstration of phosphates in calcium deposits: a modification of von Kossa's reaction. Histochemistry. 56(3-4): 177-185.

140. Meloan, S.N., Puchtler, H. (1985) Chemical mechanisms of staining methods: von Kossa’s technique. What von Kossa really wrote and a modified reaction for selective demonstration of inorganic phosphate. J Histotechol. 8: 11-13.

141. Rungby, J., Kassem, M., Eriksen, E.F., Danscher, G. (1993) The von Kossa reaction for calcium deposits: silver lactate staining increases sensitivity and reduces background. Histochem J. 25(6): 446-451.

142. Puchtler, H., Meloan, S.N., Terry MS (1969) On the history and mechanism of alizarin and alizarin red S stains for calcium. J Histochem Cytochem. 17(2): 110-124.

143. Connerty, H.V., Briggs, A.R. (1966). Determination of serum calcium by means of orthocresolphthalein complexone. Am. J. Clin. Pathol. 45: 290‐296.

144. Davis, L.A., Dienelt, A., zur Nieden, N.I. (2011) Absorption-based assays for the analysis of osteogenic and chondrogenic yield. Methods Mol Biol. 690: 255-272.

145. Schäck, L.M., Noack, S., Winkler, R., et al (2013) The Phosphate Source Influences Gene Expression and Quality of Mineralization during In Vitro Osteogenic Differentiation of Human Mesenchymal Stem Cells. PLoS One. 8(6): e65943

146. Proudfoot, D., Skepper, J.N., Hegyi, L., et al (2000) Apoptosis regulates human vascular calcification in vitro: evidence for initiation of vascular calcification by apoptotic bodies. Circ Res. 87(11): 1055-1062.

147. Cassella, J.P., Garrington, N., Stamp, T.C., Ali, S.Y. (1995) An electron probe X-ray microanalytical study of bone mineral in osteogenesis imperfecta. Calcif Tissue Int. 56(2): 118-122.

148. Boskey, A.L., Roy, R. (2008) Cell culture systems for studies of bone and tooth mineralization. Chem Rev. 108(11): 4716-4733.

149. Dmitrieva, R.I., Minullina, I.R., Bilibina, A.A., et al (2012) Bone marrow- and subcutaneous adipose tissue-derived mesenchymal stem cells: differences and similarities. Cell Cycle. 11(2): 377-383.

150. Liao, H.T., Chen, C.T. (2014) Osteogenic potential: Comparison between bone marrow and adipose-derived mesenchymal stem cells. World J Stem Cells. 6(3): 288-295.

151. Mohamed-Ahmed, S., Fristad, I., Lie, S.A., et al (2018) Adipose-derived and bone marrow mesenchymal stem cells: a donor-matched comparison. Stem Cell Res Ther. 9(1): 168.

152. Im, G.I., Shin, Y.W., Lee, K.B. (2005) Do adipose tissue-derived mesenchymal stem cells have the same osteogenic and chondrogenic potential as bone marrow-derived cells? Osteoarthr Cartil. 13(10): 845–853.

153. Hayashi, O., Katsube, Y., Hirose, M., et al (2008) Comparison of osteogenic ability of rat mesenchymal stem cells from bone marrow, periosteum, and adipose tissue. Calcif Tissue Int. 82(3): 238-247.

154. Monaco, E., Bionaz, M., Rodriguez-Zas, S., et al (2012) Transcriptomics comparison between porcine adipose and bone marrow mesenchymal stem cells during in vitro osteogenic and adipogenic differentiation. PLoS One. 7(3): e32481.

155. Vidal, M.A., Kilroy, G.E., Lopez, M.J., et al. (2007) Characterization of equine adipose tissue-derived stromal cells: adipogenic and osteogenic capacity and comparison with bone marrow-derived mesenchymal stromal cells. Vet Surg. 36(7): 613-622.

156. Toupadakis, C.A., Wong, A., Genetos, D.C., et al (2010) Comparison of the osteogenic potential of equine mesenchymal stem cells from bone marrow, adipose tissue, umbilical cord blood, and umbilical cord tissue. Am J Vet Res. 71(10): 1237-1245.

157. Kang, B.J., Ryu, H.H., Park, S.S., et al (2012) Comparing the osteogenic potential of canine mesenchymal stem cells derived from adipose tissues, bone marrow, umbilical cord blood, and Wharton's jelly for treating bone defects. J Vet Sci. 13(3): 299-310.

158. Chung, D.J., Hayashi, K., Toupadakis, C.A., et al (2012) Osteogenic proliferation and differentiation of canine bone marrow and adipose tissue derived mesenchymal stromal cells and the influence of hypoxia. Res Vet Sci. 92(1): 66-75.

159. Hoffmann, A., Preobrazhenska, O., Wodarczyk, C., et al (2005) Transforming growth factor-beta-activated kinase-1 (TAK1), a MAP3K, interacts with Smad proteins and interferes with osteogenesis in murine mesenchymal progenitors. J Biol Chem. 280(29): 27271-27283.

160. Greenblatt, M.B., Shim, J.H., Zou, W., et al (2010) The p38 MAPK pathway is essential for skeletogenesis and bone homeostasis in mice. J Clin Invest. 120: 2457-2473.

161. Noël, D., Gazit, D., Bouquet, C., et al (2004) Short-term BMP-2 expression is sufficient for in vivo osteochondral differentiation of mesenchymal stem cells. Stem Cells. 22(1): 74- 85.

162. Lee, K.S., Kim, H.J., Li, Q.L., et al (2000) Runx2 is a common target of transforming growth factor beta1 and bone morphogenetic protein 2, and cooperation between Runx2 and Smad5 induces osteoblast-specific gene expression in the pluripotent mesenchymal precursor cell line C2C12. Mol Cell Biol. 20(23): 8783-8792.

163. Phimphilai, M., Zhao, Z., Boules, H., et al (2006) BMP signaling is required for RUNX2- dependent induction of the osteoblast phenotype. J Bone Miner Res. 21:637-646.

164. Lee, M.H., Kim, Y.J., Yoon, W.J., et al (2005) Dlx5 specifically regulates Runx2 type II expression by binding to homeodomain-response elements in the Runx2 distal promoter. J Biol Chem. 280(42): 35579-35587.

165. Lee, M.H., Kwon, T.G., Park, H.S., et al (2003) BMP-2-induced Osterix expression is mediated by Dlx5 but is independent of Runx2. Biochem Biophys Res Commun. 309(3): 689- 694.

166. Ulsamer, A., Ortuño, M.J., Ruiz, S., et al (2008) BMP-2 induces Osterix expression through up-regulation of Dlx5 and its phosphorylation by p38. J Biol Chem. 283(7): 3816- 3826.

167. Matsubara, T., Kida, K., Yamaguchi, A., et al (2008) BMP2 regulates Osterix through Msx2 and Runx2 during osteoblast differentiation. J Biol Chem. 283(43): 29119-29125.

168. Choi, Y.H., Jeong, H.M., Jin, Y.H., et al (2011) Akt phosphorylates and regulates the osteogenic activity of Osterix. Biochem Biophys Res Commun. 411(3): 637-641.

169. Jeong, H.M., Jin, Y.H., Kim, Y.J., et al (2011) Akt phosphorylates and regulates the function of Dlx5. Biochem Biophys Res Commun. 409(4): 681-686.

170. Salazar, V.S., Gamer, L.W., Rosen, V. (2016) BMP signaling in skeletal development, disease and repair. Nat Rev Endocrinol. 12(4): 203-221.

171. Fischer, J., Kolk, A., Wolfart, S., et al (2011) Future of local bone regeneration - Protein versus gene therapy. J Craniomaxillofac Surg. 39(1): 54-64.

172. Lowery, J.W., Brookshire, B., Rosen, V. (2016) A Survey of Strategies to Modulate the Bone Morphogenetic Protein Signaling Pathway: Current and Future Perspectives. Stem Cells Int. 2016: 7290686.

173. Ong, K.L., Villarraga, M.L., Lau, E., et al (2010) Off-label use of bone morphogenetic proteins in the United States using administrative data. Spine (Phila Pa 1976). 35(19): 1794- 1800.

174. Lowery, J.W., Rosen, V. (2018) Bone Morphogenetic Protein-Based Therapeutic Approaches. Cold Spring Harb Perspect Biol. 10(4). pii: a022327.

175. Carragee, E.J., Hurwitz, E.L., Weiner, B.K. (2011) A critical review of recombinant human bone morphogenetic protein-2 trials in spinal surgery: emerging safety concerns and lessons learned. The Spine J. 11(6): 471-491.

176. Zhang, X., Guo, J., Zhou, Y., Wu, G. (2013) The roles of bone morphogenetic proteins and their signaling in the osteogenesis of adipose-derived stem cells. Tissue Eng Part B Rev. 20(1): 84-92.

177. Panetta, N.J., Gupta, D.M., Lee, J.K., et al (2010) Human adipose-derived stromal cells respond to and elaborate bone morphogenetic protein-2 during in vitro osteogenic differentiation. Plast Reconstr Surg. 125(2): 483-493.

178. Kimelman Bleich, N., Kallai, I., Lieberman, J.R., et al (2012) Gene therapy approaches to regenerating bone. Adv Drug Deliv Rev. 64(12): 1320-1330.

179. Alden, T.D., Varady, P., Kallmes, D.F., et al (2002) Bone morphogenetic protein gene therapy. Spine (Phila Pa 1976). 27(16 Suppl 1): S87-S93.

180. Wilson, C.G., Martín-Saavedra, F.M., Vilaboa, N., Franceschi, R.T. (2013) Advanced BMP gene therapies for temporal and spatial control of bone regeneration. J Dent Res. 92(5): 409-417.

181. Komori, T. (2018) Runx2, an inducer of osteoblast and chondrocyte differentiation. Histochem Cell Biol. 149(4): 313-323.

182. Gersbach, C.A., Le Doux, J.M., Guldberg, R.E., Garcia, A.J. (2006) Inducible regulation of Runx2–stimulated osteogenesis. Gene Ther. 13(11): 873–882.

183. Zhao, Z., Wang, Z., Ge, C., et al (2007) Healing cranial defects with AdRunx2- transduced marrow stromal cells. J Dent Res. 86(12): 1207-1211.

184. Wojtowicz, A.M., Templeman, K.L., Hutmacher, D.W., et al (2010) Runx2 overexpression in bone marrow stromal cells accelerates bone formation in critical-sized femoral defects. Tissue Eng Part A. 16(9): 2795-2808.

185. Bhat, B.M., Robinson, J.A., Coleburn, V.E., et al (2008) Evidence of in vivo osteoinduction in adult rat bone by adeno-Runx2 intra-femoral delivery. J Cell Biochem. 103(6): 1912-1924.

186. Takahashi, T. (2011) Overexpression of Runx2 and MKP-1 stimulates transdifferentiation of 3T3-L1 preadipocytes into bone-forming osteoblasts in vitro. Calcif Tissue Int. 88(4): 336-347.

187. Hirata, K., Tsukazaki, T., Kadowaki, A., et al (2003) Transplantation of skin fibroblasts expressing BMP-2 promotes bone repair more effectively than those expressing Runx2. Bone. 32(5): 502-512.

188. Lee, S.J., Kang, S.W., Do, H.J., et al (2010) Enhancement of bone regeneration by gene delivery of BMP2/Runx2 bicistronic vector into adipose-derived stromal cells. Biomaterials. 31(21): 5652-5659.

189. Geoffroy, V., Kneissel, M., Fournier, B., et al (2002) High bone resorption in adult aging transgenic mice overexpressing cbfa1/runx2 in cells of the osteoblastic lineage. Mol Cell Biol. 22(17): 6222-6233.

190. Schiltz, C., Prouillet, C., Marty, C., et al (2010) Bone loss induced by Runx2 over- expression in mice is blunted by osteoblastic over-expression of TIMP-1. J Cell Physiol. 222(1): 219-229.

191. Li, N., Luo, D., Hu, X., et al (2015) RUNX2 and Osteosarcoma. Anticancer Agents Med Chem. 15(7): 881-887.

192. Tai, G., Polak, J.M., Bishop, A.E., et al (2004) Differentiation of osteoblasts from murine embryonic stem cells by overexpression of the transcriptional factor osterix. Tissue Eng. 10(9-10): 1456-1466

193. Tu Q1, Valverde P, Chen J. (2006) Osterix enhances proliferation and osteogenic potential of bone marrow stromal cells. Biochem Biophys Res Commun. 2006 Mar 24;341(4):1257-65. Epub 2006 Jan 30.

194. Tu, Q., Valverde, P., Li, S., et al (2007) Osterix overexpression in mesenchymal stem cells stimulates healing of critical-sized defects in murine calvarial bone. Tissue Eng. 13(10): 2431-2440.

195. Kim, Y.J., Kim, H.N., Park, E.K., et al (2006) The bone-related Zn finger transcription factor Osterix promotes proliferation of mesenchymal cells. Gene. 366(1): 145-151.

196. Lai, Q.G., Yuan, K.F., Xu, X., et al (2010) Transcription factor osterix modified bone marrow mesenchymal stem cells enhance callus formation during distraction osteogenesis. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 111(4): 412-419.

CHAPTER 3 The Effects of General Anesthesia and Euthanasia on Equine Mesenchymal Stem Cell Viability, Proliferation, and In Vitro Osteogenic Capacity

Abstract Multi-potent mesenchymal stem cells (MSCs) isolated from adult tissues, such as adipose (ADI) and bone marrow (BM), are promising sources for regenerative medicine therapies in both human and veterinary medicine. Tissues are typically collected from patients while they are under sedation or general anesthesia; however there is potential for collecting stem cells from recently deceased donors, somewhat analogous to organ donor programs, which would substantially increase the donor pool for allogeneic applications, in both veterinary and human medicine. The purpose of this study was to test the hypotheses that cell viability, proliferation, and in vitro osteogenic capacities of equine adult ADI- and BM-derived MSCs collected from anesthetized horses and shortly after euthanasia would be compromised compared to sedation- collected controls. The results demonstrated that donor status during collection did not impact cell viability or in vitro proliferation. Further, osteogenic differentiation was not negatively impacted overall, as assessed by osteogenic gene expression, alkaline phosphatase (ALP) enzymatic activity, and Alizarin Red mineralized matrix staining. Some effects of donor status during collection on gene expression were evident at early time points, but this influence did not persist. In contrast to donor status during collection, all outcomes were strongly influenced by both inter- and intra-donor variability. Overall, equine adult MSCs collected during general anesthesia and shortly after euthanasia performed similarly to MSCs collected under sedation, thus supporting the collection of cadaver-sourced MSCs for allogeneic therapeutic applications.

Introduction Mesenchymal stem cells (MSCs) can be harvested and isolated from several adult tissues and body fluids [1 – 3]. They possess the ability to differentiate along multiple lineages [2, 4, 5], reduce inflammation at injury sites [6], and modulate host immune cell activities [7], thus making them an appealing source for regenerative medicine therapies [8]. In human medicine, MSCs for musculoskeletal applications are commonly isolated from bone marrow while the patient is anesthetized [9], but adipose tissue is also an abundant and easily harvestable alternative [10]. In veterinary medicine, MSC isolation techniques vary between species [1, 11, 46

12]. Horses are the veterinary species that utilize MSC-based therapies most extensively, due to their use as high-performance athletes and associated propensity for orthopedic injury [13, 14]. Equine adipose (ADI) tissue and bone marrow (BM) aspirates can be collected from sedated horses [1], and the MSCs isolated from these sources have been successfully utilized in multiple clinical research scenarios for the healing of various musculoskeletal injuries [2, 11, 15 – 17]. There has been an interest in utilizing MSC-based therapies for targeting bone repair [17 – 18], especially in species such as horses that are prone to high morbidity and mortality rates due to fracture [13]. Optimizing MSC collection and isolation approaches, as well as in vitro osteogenic potential screening methods, can improve the use of MSCs in fracture healing. In both veterinary and human medicine, stem cell therapy is most often autologous in nature, meaning the donor is also the eventual recipient of the cells. This approach avoids the risk of immuno-rejection of the donor cells [19, 20]. However, there are many disadvantages surrounding the collection and use of autologous MSCs. Sample collection from live donors carries many risks for the patient, such as reaction to anesthesia, excessive bleeding, infection, disfigurement, and discomfort at the harvest site [21]. Human bone marrow collection can cause co-morbidities that may outweigh the benefits of any consequent MSC-based therapies [22, 23]. There is also considerable risk for the harvester working with veterinary species such as horses that often respond unpredictably during collection procedures. Further, the process of utilizing autologous MSCs involves extensive in vitro isolation and expansion to generate sufficient viable cells for clinical applications [9]. This can be problematic when therapy is necessary immediately to optimize clinical benefit, such as during a fracture repair scenario. Finally, MSCs from any given donor may be negatively altered by disease status, age, pharmaceutical agents, and other factors [24], thus making autologous MSC use a less than ideal option for some patients. Therefore, while autologous MSC may be the gold standard, there are many compelling reasons to explore allogeneic alternatives. The use of allogeneic MSCs banked from other donors provides for immediate therapeutic delivery that can be administered in conjunction with other procedures at the time of primary intervention. Current ethical and technical standards related to stem cell collection for allogeneic stem cell banking are not clear [25]. Recently deceased individuals have been proposed as a source for tissues and cells [26, 27], which could provide a large pool of donors while also avoiding many of the concerns related to collection morbidities [21 – 23], sample

47 quantity limitations [28], and regulatory protocol oversight at the institution level. Although there are known immuno-incompatibility issues with allogeneic grafts [29], these cells could be prospectively screened to minimize recipient rejection. However, it is uncertain whether MSCs collected while the donor is under general anesthesia or recently deceased behave equivalently to MSCs collected from conscious individuals, and whether experimental data derived from cells collected from anesthetized or recently deceased donors can be extrapolated to activities present in conscious donor cell populations [28, 30]. The aims of this study were to compare the cell viability, proliferation, and osteogenic capacities of equine adult ADI- and BM-MSCs harvested from the same donor horses while under 1) standing sedation, 2) general anesthesia, and 3) shortly after euthanasia. The experiments were designed to test the hypotheses that equine MSCs collected during general anesthesia and after euthanasia would have decreased cell viability, proliferation, and osteogenic capacity, when compared to sedation-collected controls.

Materials & Methods Research Animals All procedures were approved by the University of Illinois Institutional Animal Care and Use committee (IACUC). Four healthy, adult horses (Table 3.1) were donated to the Veterinary Teaching Hospital for teaching and/or research purposes. Adipose tissue and bone marrow were collected sequentially from each horse while they were: 1) sedated but standing, 2) under general anesthesia in lateral recumbency, and 3) shortly after euthanasia.

Sedation, Anesthesia, and Euthanasia Protocols All horses were fasted overnight prior to the collection procedures. Horses were sedated with xylazine (1 – 2 mg/kg), and a 16-gauge intravenous catheter was placed in the jugular vein for administration of additional sedation and anesthetic drugs. For collection under sedation, horses were placed in stocks for restraint and given additional xylazine as required. The collection sites were cleaned and prepared once the horse showed signs of sedation and did not respond adversely to hair clipping and skin scrubbing. After collection of the “sedated” samples, the horses were anesthetized with a combination of ketamine (2.2 mg/kg) and midazolam (0.1 mg/kg) administered IV through the

48 jugular catheter. Horses were placed in lateral recumbency, and an endotracheal tube was inserted for mechanical ventilation and anesthesia maintenance with isoflurane. Vital parameters were monitored using electrocardiogram (ECG), pulse oximetry, and direct arterial blood pressure and blood gas sampling via a 20-gauge catheter placed in the transverse facial artery. Horses under general anesthesia were allowed to stabilize until an appropriately deep anesthetic plane was reached and maintained for at least 5 minutes. After collection of the “anesthesia” samples, the horses were euthanized while still under general anesthesia using an IV bolus of pentobarbital (10 – 15 ml / 100 lb body weight). Death was confirmed by cardiac auscultation, ECG, and a negative corneal reflex response. “Euthanasia” samples were collected approximately 5 – 10 minutes after death was confirmed.

Adipose Tissue Collection & Processing For adipose tissue collection, the skin adjacent to the tail head was clipped and cleaned with chlorohexidine and ethanol. The collection site was desensitized using 1% lidocaine subcutaneously (10 – 20 ml) until the horse did not respond to pressure from a hemostat. A 5 – 10 cm skin incision was made into the subcutaneous adipose depot lateral to each tail head. Approximately 8 – 10 grams of adipose tissue was removed and immediately placed into sterile cell culture growth media [1X Dulbecco’s modified Eagle’s medium (DMEM), 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 streptomycin μg/ml (1% P/S)]. The skin incision was closed with monocryl suture material. Additional samples collected under general anesthesia and after euthanasia were harvested from sites adjacent or contralateral to the initial site. Adipose tissue samples were diced into small pieces and digested in cell culture media (10 ml per 1 gram of tissue) containing 0.2% collagenase type II (Worthington Biochemical Corporation) for 3 hours in an orbital shaker at 37oC. The digest was triturated via pipette during the incubation, to facilitate cell dispersion. After digestion, the cells were filtered through 70 μm and 40 μm mesh filters, and then centrifuged at 300 x g for 10 minutes (Sorvall ST 16R Centrifuge, Thermo Scientific). The number of primary adipose cells was determined using a hemocytometer with cellular viability assessed via Trypan Blue exclusion. The isolated adipose (ADI) MSCs were then seeded in 100-mm dishes at 5 x 103 cells/cm2 in growth media and o maintained in an incubator at 37 C in 5% CO2 and 90% humidity. Media was changed every 2 – 3 days until cells reached 80 – 90% confluence.

Bone Marrow Collection & Processing For bone marrow aspiration, the skin of the ventral sternum region was clipped and aseptically prepped. The 5th sternebra was chosen as the collection site during standing sedation, which is located approximately 8 cm caudal to the point of the elbow. The area was blocked using 1% lidocaine subcutaneously and near the periosteum (5 – 10 ml). A small stab incision was made through the skin using a #11 scalpel blade. A 12-gauge Jamshidi needle was introduced into the stab incision and advanced approximately 1 – 2 cm through the cortical bone and into the medullary cavity of the sternebra. A 12-ml syringe preloaded with approximately 5000 IU of heparin was connected to the Jamshidi needle and used to slowly aspirate 10 – 15 ml of bone marrow. The syringe was inverted several times, and then the bone marrow aspirate was transferred asceptically to a 50-ml sterile conical tube. Additional samples collected under general anesthesia and after euthanasia were harvested from sternabrae adjacent to the initial site. Bone marrow aspirates were mixed and suspended in PBS containing 1% P/S, then centrifuged at 1000 x g for 15 minutes (Sorvall ST 16R Centrifuge, Thermo Scientific) to pellet the cellular contents. All but approximately 5 ml of the supernatant was aspirated, and the pellet was resuspended in growth media and transferred to two adherent T-75 cm2 cell culture flasks. Additional media was added to a total volume of 15 – 20 ml per flask. Flasks were placed in an o incubator at 37 C in 5% CO2 and 90% humidity. To allow adequate time for bone marrow (BM) MSCs to settle out of the suspension and adhere to the cell culture flask, only a small portion of the media was changed every 2 – 3 days for the first week. The process involved rocking the flask back and forth several times to loosen red blood cells and other non-adherent debris, then tilting the flask on its side for 5 minutes to allow heavier contents to settle toward the bottom. About half of the supernatant was removed and replaced with 10 ml fresh growth media. After the first week, the remaining media/cell suspension was completely removed and the flask surface washed with PBS + 1% P/S before refeeding with growth media, which was changed every 2 – 3 days until the adherent cells reached 80 – 90% confluence.

Cellular Expansion & Proliferation Assays When MSCs reached near-confluence, they were passaged via trypsinization. Growth media was first removed, and the culture vessel rinsed with 1X PBS + 1% P/S. An appropriate

50 volume of 0.5% trypsin-EDTA was added to the vessel, which was then placed in an incubator at 37oC for 5 minutes. The vessel was briefly agitated to assess cell detachment, and then growth media (1X DMEM, 10% FBS, 1% P/S) was added to deactivate the trypsin. A small volume of the cell suspension was used to make a 1:10 dilution in cell counting solution (10% Trypan blue in 1X PBS). A volume of 10 μl was added to each side of the hemocytometer, and the total number of viable cells (i.e., those that do not absorb Trypan blue) was counted:  Total number of live cells/ml = (# of live cells counted / # of squares) x 10 (cell dilution factor) 104 (hemocytometer surface area factor)  Total number of live cells = # of cells/ml x volume of cell suspension  Cell viability = # of live cells / (# dead cells + # live cells)

All primary MSCs were expanded through 3 passages. The number of cells isolated from the initial collections, if known, the numbers seeded and recovered at each passage, and the time required for expansion to confluence were recorded. These values were used to calculate the population doublings and population doubling times [31]:

 Population doublings (PDs) during each passage = Log2 (harvested cell # / seeded cell #)  Population doubling time (PDT) = time between passages / PDs during each passage

In Vitro Osteogenesis Third passage (P3) MSCs were seeded at 2,000 cells/cm2 in cell culture plates and maintained in growth media (1X DMEM, 10% FBS, 1% P/S) until approximately 50% confluence was reached. Control (Cx) cultures remained in growth media, while osteogenic cultures were transferred to osteogenic induction media (OIM), which consisted of a growth media base supplemented with 50 μg/ml L-Ascorbic Acid Phosphate Magnesium Salt n-Hydrate (Wako, Japan), 10 mM β-Glycerophosphate (Sigma-Aldrich), and 100 nM Dexamethasone (Sigma-Aldrich). ADI-MSCs in OIM were also supplemented with 100 ng/ml recombinant human bone morphogenetic protein (BMP)-2 (Gemini Biosciences) based on preliminary results. Cultures were maintained for up to 10 days. Media was replaced every 2 – 3 days, and representative images of the cultures were recorded via digital light microscopy. Cells were collected for Alizarin Red staining, RNA isolation for gene expression analyses, and alkaline phosphatase (ALP) enzymatic assay at days 5 and 10 post-osteogenic induction.

Alizarin Red Staining

A 2% Alizarin Red-S (Sigma-Aldrich) solution was prepared in ddH2O and adjusted to a pH of 4.1 – 4.3 using an UltraBASIC UB-5 pH meter (Denver Instrument). The stain was filter- sterilized to remove any debris and stored at 4oC when not in use. Cells in 6-well plates were washed in 1X PBS + 1% P/S, and then fixed with 4% paraformaldehyde for 30 minutes at room temperature. Wells were washed three times with ddH2O, and 1 ml of 2% Alizarin Red-S stain was applied. Plates were incubated for 20 minutes at room temperature on a plate shaker to distribute stain evenly. Excess stain was removed with repeated ddH2O rinses, and stained cells were imaged dry using a DMIL microscope, PFC320 digital camera, and supporting Leica Application Suite (LAS Version 2.6.R1) software (LEICA). A positive Alizarin Red stain was indicated by a red color uptake into the cellular matrix, while a negative stain maintained a dull brownish color, even when sequestered in cellular aggregates.

RNA Isolation & Quantification Cells in 2 – 3 wells of a 6 well plate were collected and pooled for RNA isolation using the phenol-based dissociation agent, TRIzol® (Life Technologies). Media was removed, and each well was washed with 1X PBS + 1% P/S. TRIzol® reagent was applied directly to the cell monolayers, and the lysates were scraped from the vessel surface and transferred to 15 ml conical tubes. Tubes were placed on ice, and then homogenized for 30 seconds using an Ultra- Turrax T25 homogenizer (Janke & Kunkel, IKA Labortechnik). Samples were stored at -80oC. RNA isolation was performed according to the recommended TRIzol® reagent protocol. Lysates were first transferred to a 1.5 ml microcentrifuge tube. For every 1 ml TRIzol® reagent used, 200 μl of chloroform (Sigma-Aldrich) was added. Samples were vortexed for 15 seconds, and then allowed to separate at room temperature for 2 – 3 minutes. Following centrifugation at 12,000 x g at 4oC for 15 minutes (Centrifuge 5415 R, Eppendorf), the upper, aqueous portion (approximately 400 – 500 μl) was transferred to a new 1.5 ml microcentrifuge tube. The interface and lower portions, containing insoluble proteins, lipids, and DNA, were discarded. RNA within the aqueous phase was precipitated by adding 500 μl isopropanol and incubating at -20oC overnight. The next day, the samples were centrifuged as before, resulting in the precipitation of a translucent pellet. The supernatant was removed, and the pellet was washed

52 with 1 ml of 75% ethanol to reduce the salt content of the pellet. The samples were briefly vortexed to detach the pellet from the tube wall, and then centrifuged at 7,500 x g for 5 minutes at 4oC. The ethanol wash supernatant was removed, and the pellet was allowed to air-dry for approximately 5 – 10 minutes. Any remaining liquid was carefully removed, and the RNA pellet was resuspended in nuclease-free water (NF-H2O) at a volume appropriate for the estimated pellet size (typically 50 – 100 μl). The tubes were vortexed, spun down briefly, and then incubated in a 65oC water bath for 10 minutes. RNA samples were immediately quantified using a NanoDrop 2000 spectrophotometer (Thermo Scientific), and then stored at -80oC.

Reverse Transcription RNA was reverse transcribed into cDNA using the SuperScriptTM IV First-Strand Synthesis System kit (Invitrogen). One microgram of each RNA sample was brought up to 11 μl total volume with diethyl pyrocarbonate (DEPC)-treated water. A master mix containing 1 μl of

50 μM oligo(dT)20 and 1 μl of 10 mM dNTP mix per sample was made, and 2 μl of this master mix was added per sample. Tubes were vortexed and quick spun down, and then incubated in a 65oC water bath for 5 minutes. Tubes were removed and chilled on ice for at least 1 minute. A second master mix was made, which contained 4 μl of 5X SSIV Buffer, 1 μl of 100 mM DTT, 1 μl of 40 U/μl Ribonuclease inhibitor, and 1 μl of 200 U/μl SuperScript IV reverse transcriptase (RT) enzyme. Seven microliters of the second master mix were added per sample, and the tubes were briefly spun down. For the reverse transcription reaction, tubes were incubated in a 55oC water bath for 20 minutes. The RT reaction was inactivated by placing the samples in an 80oC water bath for 10 minutes. RNA was removed by treating each sample with 2 units of E. coli RNase H and incubating in a 37oC water bath for 30 minutes. Final cDNA samples were diluted o 1:10 by adding 189 μl NF-H2O, and the samples were stored at -20 C until qPCR.

Primers & Gradient PCR Samples were assessed for their relative expression of major osteogenic genes: runt- related transcription factor 2 (RUNX2), osterix (OSX), and alkaline phosphatase (ALP). Elongation factor 1 alpha-1 (EF1α) was used as a reference gene. Primer sequences were designed using NCBI-BLAST software and purchased from Integrated DNA Technologies (IDT). See Table 3.2 for primer sequences and optimal annealing temperatures.

Lyophilized primers were reconstituted in 1X TE (10 mM Tris-HCl, 1 mM EDTA, pH

8.0) buffer to generate a 100 μM stock solution, which was further diluted 1:10 in NF-H2O to yield a 10 μM working stock. Optimum annealing temperatures were determined by gradient PCR and subsequent gel electrophoresis to assess product band quality and intensity. Approximately 100 ng of a positive control cDNA sample expected to express detectable levels of the primer target was used. Each PCR reaction contained 2.5 μl of 10X PCR Buffer, 1 μl of 10 mM MgCl2, 0.5 μl of 10 mM dNTP mix, 0.3 μl of Taq polymerase (Life Technologies), and NF-

H2O to a final volume of 23 μl. A master mix of each 1:10 diluted primer set contained equal volumes of sense and anti-sense primer, which was added at 2 μl per reaction. Gradient PCR amplification was performed using an iCycler iQTM thermocycler (BIO- RAD) at various annealing temperatures ranging from 50 – 65oC. Following PCR amplification, 10 μl of each sample was mixed with 2 μl of Blue/Orange 6X loading dye (Promega), and 10 μl of this mixture was added to a well in a 2% agarose (Fisher BP160-100) gel in 1X TAE buffer (1:50 diluted 50X stock solution: 242 g Tris base, 57.1 ml glacial acetic acid, 100 ml of 500 mM

EDTA, pH 8.0, to a final volume of 1L in ddH2O) containing 0.5 mg/ml ethidium bromide (Sigma) submersed in 1X TAE running buffer. A volume of 2 – 3 μl TrackItTM 1Kb Plus DNA ladder (Invitrogen) was added per lane. Gels were run in a Midi-Horizontal Electrophoresis System (Fisher Biotech, FB-SB-1316) at 100 V constant using a Power Station 300 (Labnet) power source for approximately 1 – 2 hours or until desired distance was reached. Upon completion, gels were imaged using the Molecular Imager ChemiDocTM XRS+ (BIO-RAD) and corresponding Image Lab software (BIO-RAD). Band intensity was compared across temperatures to determine which produced a single product and most-intense band.

Primer Concentrations & Efficiency Primers utilized in the qPCR reactions were tested for optimal reaction concentration by testing several dilutions of primers with a known quantity of cDNA at an optimum annealing temperature determined from gradient PCR. The optimal primer concentration in the qPCR reaction was determined based on which dilution yielded the lowest CT values for that primer set. Primer efficiency was evaluated by testing the single best primer concentration on serially-diluted cDNA samples of known concentration. The CT values obtained were plotted

2 against the Log10 of the cDNA dilution factor. A linear fit equation (y = mx + b, R ) was determined using Microsoft Excel. Results were analyzed based on the following criteria:  PCR efficiency of 100% = slope of -3.32 o A “good” reaction has an efficiency between 90 – 110%, which corresponds to slope values between -3.58 and -3.10.  Amplification Efficiency = 10(-1/slope) – 1  % Efficiency = Amplification Efficiency x 100

Quantitative Real-Time PCR Quantitative real-time PCR (qPCR) was performed using 5 μl of 1:10 diluted cDNA samples added to 20 μl of a master mix comprised of 12.5 μl iQTM SYBR® Green Supermix (BIO-RAD), 2 μl each of 1:10 diluted sense and antisense primer (0.4 μM final concentration), and 3.5 μl NF-H2O. A “no template control” reaction was performed for each primer, containing TM 5 μl of NF-H2O instead of cDNA. The qPCR reactions were performed using the iCycler iQ thermocycler (BIO-RAD) and iCycler Optical Module (BIO-RAD), which was run and subsequently analyzed using iCycler iQTM software (BIO-RAD). The presence of a single PCR amplicon was determined by melting curve analysis, and in select cases, gel electrophoresis. Amplicons were sequenced during the initial optimization protocols to ensure correct targeting. -ΔΔCT Quantitative PCR data were analyzed using the 2 method [32]. The CT values for the constitutively reference gene, EF1α, were used to normalize CT values of the experimental transcripts. The corrected CT value of the sedation-collected control sample from each collection time point was then subtracted from each sample to produce the comparative ΔCT value. The final relative fold change in expression levels was determined by 2(- comparative ΔCT value). This equation results in the reference control sample yielding a value of 1, while each treatment sample’s value will vary depending on the relative change in expression of the target gene. For primer sets with a less than 90% efficiency, the equation was adjusted to 1.99(- comparative ΔCT value).

Alkaline Phosphatase Enzymatic Assay Cells were lysed in 1 ml of 2% Triton X-100 in 1X TE Buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0). In some cases, samples were from a single well of a 6-well plate, pooled from 2 wells of a 12-well plate, or collected as individual replicates from a 12-well plate. Surface area of

55 sample collection and dilution adjustments were taken into account during the assay analyses. Cell lysates were transferred to 2 ml microcentrifuge tubes and stored at -20oC until processing. Samples were transferred to glass tubes and homogenized for 15 – 30 seconds using an Ultra-Turrax T25 homogenizer (Janke & Kunkel, IKA Labortechnik). Homogenate was transferred into a microcentrifuge tube, left on ice for 30 minutes, and then centrifuged at 2,500 rpm and 4oC for 15 minutes (Centrifuge 5415 R, Eppendorf). The supernatant was transferred to a new microcentrifuge tube and stored at -20oC until the assay was performed. ALP activity was assessed using the LabAssayTM ALP kit (Wako) following the manufacturer’s instructions. The assay uses a p-nitrophenyl phosphate substrate, which is hydrolyzed into p-nitrophenol and phosphoric acid in the carbonate buffer when alkaline phosphatase is present in the sample. Working assay solution was made by dissolving a p- nitrophenyl phosphate tablet in 5 ml Buffer Solution (2.0 mmol/L MgCl2, 0.1 mol/L Carbonate Buffer, pH 9.8), producing a final concentration of 6.7 mmol/L. A standard curve was produced by serial dilutions of the Standard Solution (0.5 mmol/L p-nitrophenol). Twenty microliters of each standard and sample was added to a transparent 96-well microplate (Corning), and then 100 μl of working assay solution was added per well. The plate was briefly agitated to mix contents, and then placed in an incubator at 37oC for 15 minutes. After incubation was complete, 80 μl of Stop Solution (0.2 mol/L Sodium Hydroxide) was added to each well to deactivate the reaction. The reaction generated a yellow color change, which was optically measured at 405 nm wavelength using a FLUOstar OPTIMA microplate reader and supporting software (BMG). To determine the amount of p-nitrophenol product in each sample, a standard curve and linear fit equation was generated. The concentration was initially calculated as mmol/L, which is equivalent to nm/μl. The activity was calculated as follows:  ALP Activity (units/μl) = (C / RT) x DF o C = concentration of p-nitrophenol released relative to standard curve (nm/μl) o RT = reaction time, which was 15 minutes o DF = dilution factor, which was considered ‘1’ unless the sample was pre-diluted

The ALP activity of each sample was normalized to its DNA content by dividing the activity (units/μl) by the DNA concentration (μg/μl), as determined in a separate assay. The final data was analyzed and presented in ALP Activity (units/μg DNA) divided by 100.

DNA Quantification Assay Samples for DNA quantification were the same as those processed and used for the ALP activity assay above. DNA quantification was performed using the Quant-iTTM PicoGreen® dsDNA kit (Invitrogen) based on the manufacturer’s instruction manual. A working solution of 1:200 diluted Quant-iTTM PicoGreen® reagent in 1X TE Buffer

(provided in kit at 20X and diluted with NF-H2O) was made and kept protected from light. The DNA standards were prepared by using the lambda (λ) DNA standard provided in the kit and making a 12-point serial dilution in 1X TE Buffer. A volume of 100 μl of each standard and sample were added to a black-sided, clear-bottom 96-well plate (Corning). The working solution was added at a volume of 100 μl per well, the plate was covered to protect from light, and the plate was mixed briefly and incubated for 5 minutes at room temperature. Fluorescence was measured at 485 nm excitation wavelength using a FLUOstar OPTIMA microplate reader and supporting software (BMG). To determine the amount of DNA in each sample, a standard curve and linear fit equation were generated. The concentration was initially calculated as μg/ml, which was divided by 1000 and converted to μg/μl. This value was then used to normalize the ALP activity, as described.

57 post-hoc tests were performed to compare anesthesia- and euthanasia-collected samples to sedation-collected controls. The effect of culture treatment and donor status during collection on osteogenic gene expression and ALP activity levels were evaluated at each culture time point using 2-Way ANOVA. Bonferonni post-hoc tests were performed to compare anesthesia- and euthanasia- collected samples to sedation-collected controls. The effect of time post-osteogenic induction on osteogenic gene expression and ALP activity level outcomes within each culture treatment group was analyzed via 2-Way ANOVA.

Results Donor Horses The four horses used in this study consisted of three intact females and one castrated male, and they varied in age between 2 and 9 years and Body Condition Scores (BCS) of 3 to 7 out of 9 (Table 3.1). There were no systemic health concerns in any of the four horses. Three samples were lost from the study. Horse 1 could not have an adipose sample collected under sedation due to technical difficulties. Horse 3 lost the euthanasia-collected BM-MSC sample during cell culture due to senescence. Horse 4 lost the sedation-collected BM-MSC sample due to contamination that could not be treated with additional antimicrobials.

Sample Collection Adipose tissue specimens ranged from 7.41 to 17.91 grams. The mean (SEM) weight in grams of tissue collected was 8.603 (0.6077) for sedation-, 6.250 (1.640) for anesthesia-, and 11.39 (3.298) for euthanasia-collected adipose tissue. There was no statistically significant difference between the collected adipose tissue weights (data not shown). Bone marrow aspirate volumes ranged from 6 to 12 ml. The mean (SEM) volume in milliliters (ml) was 10.25 (1.346) for sedation-, 10.00 (1.225) for anesthesia-, and 8.250 (2.175) for euthanasia-collected aspirates. There was no statistically significant difference between the collected bone marrow aspirate volumes (data not shown).

Primary Cell Counts ADI-MSC primary cell yields were calculated as ‘cells/gram of tissue’. The mean (SEM) yields were 1.26 x 106 (441,966) for sedation-, 2.34 x 106 (895,473) for anesthesia-, and 2.02 x 106 (792,783) for euthanasia-collected samples (Figure 3.1). Donor variability significantly impacted ADI-MSC primary cell yields (p = 0.0002), but donor status during collection did not (Figure 3.1). Horses 1 and 4 differed significantly from Horses 2 & 3 across all collection statuses (p < 0.05). Horses 1 and 4 had lower BCSs, while Horses 2 & 3 had higher BCSs. Lower BCSs resulted in higher primary cell yields from the adipose tissue, but this was observation not statistically evaluated. BM-MSC primary cell yields were determined by counting the cells after they had reached near confluence, prior to the first passage, and were adjusted for the initial bone marrow aspirate volume (ml) and days to confluence. The mean (SEM) of the primary cells / ml of aspirate for each collection status was 1.27 x 106 (82,550) for sedation-, 1.65 x 106 (75,182) for anesthesia-, and 1.88 x 106 (66,154) for euthanasia-collected samples (Figure 3.2). Donor variability significantly impacted BM-MSC primary cell yields (p = 0.0034), but donor status during collection did not (Figure 3.2). Horse 1 differed significantly from Horse 3, and Horse 4 differed significantly from Horses 2 & 3 (p < 0.05). In summary, donor variability significantly impacted primary cell yields in both ADI- and BM-MSCs, while donor status during collection did not.

Days to Confluence & Cell Yields In ADI-MSC, days to confluence were highest in primary cell cultures and decreased in passages 1 and 2, but these differences were not statistically significant (data not shown). Cell yields generally increased with passage number. Passage significantly influenced cell yield (p = 0.0329), but donor status during collection did not (data not shown). In BM-MSCs, days to confluence were also highest with the primary cell cultures and decreased in passages 1 and 2. Passage significantly impacted days to confluence (p < 0.0001), but donor status during collection did not (data not shown). End of passage cell counts increased with passage number in sedation-collected samples only. Neither passage nor donor status during collection significantly impacted cell yields (data not shown).

Population Doubling Times Population doubling time (PDT) was calculated for each passage within each donor status during collection (Figure 3.3). In ADI-MSCs (Figure 3.3, top), the mean (SEM) of PDTs for all passages was 2.72 (0.4540) for sedation-, 1.964 (0.2470) for anesthesia-, and 1.987 (0.2653) for euthanasia- collected samples. PDT was significant affected by passage (p = 0.0002) and donor variability (p = 0.0164), but not donor status during collection. In BM-MSCs (Figure 3.3, bottom), the mean (SEM) of the PDT for all passages was 1.812 (0.2133) for sedation-, 2.371 (0.3615) for anesthesia-, and 1.942 (0.2586) for euthanasia- collected samples. Primary cell PDTs could not be calculated due to an unknown starting number of primary cells in the bone marrow aspirate. There were no significant effects of passage, donor variability, or donor status during collection on PDT. In summary, passage number and donor variability significantly affected PDTs in ADI- MSCs only. Donor status during collection did not affect PDTs in either cell type.

Multi-Cellular Aggregation & Alizarin Red Staining Observations ADI-MSCs were generally less responsive to osteogenic induction compared to BM- MSCs; they showed less multi-cellular aggregation and minimal Alizarin Red stain uptake, even at later time points. Only one ADI-MSC OIM + BMP-2 culture treatment group showed multi- cellular aggregation into large nodules that stained positive for Alizarin Red by Day 10 (data not shown). Conversely, in ADI-MSC cultures that did not show distinct multi-cellular aggregation or stain positively for Alizarin Red in osteogenic conditions, BMP-2 supplementation had little to no effect over osteogenic media alone (Figure 3.4). BM-MSCs were more responsive to osteogenic induction compared to ADI-MSCs, consistent with previous data. Of the four horses, only one had BM-MSCs survive in osteogenic media to the Day 10 collection time point (Figure 3.5). This horse showed dense cellular aggregation and strong Alizarin Red stain uptake in OIM compared to Cx cultures. Alizarin Red staining even leached out of aggregates to create a weak pink background in blank areas of the plate well. The BM-MSC cultures that did not survive in osteogenic media showed accelerated cellular aggregation and cell death and detachment by days 7 – 8, but no evidence of

60 mineralization (data not shown). Alizarin Red staining was not performed at the later time point in these samples, as all plate wells were pooled for RNA sample collection. Allowing for the somewhat subjective nature of these observations, there was an obvious effect of donor variability, but not donor status during collection, on the subsequent multi- cellular aggregation behavior and Alizarin Red staining observed in either ADI- or BM-MSCs undergoing osteogenesis that survived to the later Day 10 time point.

Osteogenic Gene Expression: ADI-MSCs RUNX2 expression in ADI-MSCs (Figure 3.6) was significantly affected by culture treatment at both Day 5 (p < 0.0001) and Day 10 (p = 0.0080), with increased transcript levels present in the OIM and OIM + BMP-2 culture treatment groups compared to the Cx culture treatment group. At Day 5, RUNX2 was also significantly impacted by donor status during collection (p = 0.0292), with a significant decrease in RUNX2 transcript levels observed in the euthanasia-collected group compared to sedation-collected control group within the OIM + BMP-2 culture treatment (p < 0.05). However, the effect of donor status during collection was no longer significant at Day 10. RUNX2 expression did not significantly differ within any of the culture treatment groups between Day 5 and Day 10. OSX expression in ADI-MSCs (Figure 3.7) was not significantly affected by culture treatment or donor status during collection at either time point. OSX expression did not significantly differ within any of the culture treatment groups between Day 5 and Day 10. ALP expression in ADI-MSCs (Figure 3.8) was significantly affected by culture treatment at both Day 5 (p = 0.0095) and Day 10 (p = 0.0124), with increased transcript levels in the OIM and OIM + BMP-2 culture treatment groups compared to the Cx culture treatment group. Donor status during collection did not significantly affect ALP expression at either time point. ALP expression did not significantly differ within any of the culture treatment groups between Day 5 and Day 10. In summary, donor status during collection only significantly affected RUNX2 expression at the earlier Day 5 time point, with a significant decrease in RUNX2 transcript levels seen in euthanasia-collected samples. Culture treatment significantly affected RUNX2 and ALP, but not OSX, at both time points. Time post-osteogenic induction did not result in a significant difference in RUNX2, OSX, or ALP expression within any culture treatment group. Collectively,

61 the results also indicate that exogenous BMP ligand is required to stimulate osteogenic gene expression in ADI-MSCs.

Osteogenic Gene Expression: BM-MSCs RUNX2 gene expression in BM-MSCs (Figure 3.9) was significantly affected by culture treatment at both Day 5 (p = 0.0131) and Day 10 (p = 0.0278), with increased transcript levels present in the OIM compared to Cx culture treatment groups. Donor status during collection did not significantly affect RUNX2 expression at either time point. RUNX2 expression did not significantly differ within either culture treatment group between Day 5 and Day 10. OSX expression in BM-MSCs (Figure 3.10) was significantly affected by culture treatment only at the later Day 10 time point (p = 0.0248), with increased transcript levels present the OIM compared to Cx culture treatment groups. Donor status during collection did not significantly affect OSX expression at either time point. OSX expression did not significantly differ within either culture treatment group between Day 5 and Day 10. ALP expression in BM-MSCs (Figure 3.11) was significantly affected by culture treatment at both Day 5 (p = 0.0223) and Day 10 (p = 0.0050), with increased transcript levels present the OIM compared to Cx culture treatment groups. Donor status during collection did not significantly affect ALP expression at either time point. ALP expression was significantly increased within the OIM culture treatment group at Day 10 compared to Day 5 (p = 0.0330). In summary, donor status during collection did not significantly affect any of the osteogenic genes at either time point in BM-MSCs. Culture treatment significantly affected RUNX2 and ALP at both time points and OSX at the later time point. Time post-osteogenic induction resulted in a significant increase in ALP expression in the OIM culture treatment group, but not RUNX2 or OSX expression.

Alkaline Phosphatase (ALP) Bioactivity Assay ALP activity in ADI-MSCs (Figure 3.12) was not significantly affected by culture treatment or donor status during collection at either time point. Euthanasia-collected samples showed higher and more variable ALP activity levels at the Day 10 time point, but the difference was not statistically significant. ALP activity levels did not significantly differ within any of the culture treatment groups between Day 5 and Day 10.

ALP activity in BM-MSCs (Figure 3.13) was significantly affected by culture treatment only at the later Day 10 time point (p < 0.0001), with increased ALP activity seen in OIM compared to Cx culture treatment groups. There was no significant effect of donor status during collection on ALP activity at either time point. ALP activity was significantly increased in the OIM culture treatment group at Day 10 compared to Day 5 (p = 0.0002). In summary, ALP activity was not impacted by culture treatment, donor status during collection, or time post-osteogenic induction in ADI-MSCs. ALP activity was significantly impacted by culture treatment and time post-osteogenic induction in BM-MSCs, but not donor status during collection.

Discussion This study compared the in vitro cellular viability, proliferation, and osteogenic differentiation capacity of equine adult ADI- and BM-MSCs collected during standing sedation, general anesthesia, and post-euthanasia. The hypothesis was that general anesthesia and euthanasia would negatively impact the in vitro behavior of MSCs compared to sedation- collected controls. The results do not support the hypothesis, indicating that general anesthesia and euthanasia do not negatively impair MSC behavior in vitro and can be considered equivalent methods for collection of MSCs from adipose tissue and bone marrow. The cell viability and proliferation outcome measures used in this study were not influenced by donor status during collection of the sample. There is growing suspicion that certain anesthetic drugs impair neurogenesis, which may contribute to post-anesthetic cognitive dysfunction, especially in older individuals [34]. Studies evaluating the effects of prolonged anesthesia with various anesthetic agents in young and aged rats have shown mixed results as to the effect on neuronal cell proliferation [35 – 37]. Therefore, it is likely that any deleterious effects of general anesthetic agents on cell proliferation are localized to neuronal tissues due to their mechanisms of action. This would explain why effects were not seen in adipose or bone marrow samples. Alternatively, the time of general anesthesia in this study may not have been long enough to allow significant compromise from the anesthetic agents or physiological effects following general anesthesia. Viable MSCs have been successfully isolated from diseased donors, such as human cadaver tissues within 24 hours after death [38], bovine fetal bone marrow from an abattoir [39],

63 and equine bone marrow within 30 minutes post-euthanasia [40]. In all three studies, the cells displayed in vitro behavior characteristics of and appropriate cell surface markers for multipotent MSCs [41, 42]. The outcomes here are consistent with these published findings and extend the assessment to the process of osteogenic differentiation. Clearly, the impact of anesthesia and donor death on other lineage differentiation capacities was not addressed. In light of the recognized impact of anesthetic agents on neuronal cell functions, the effects of anesthesia and euthanasia on neuronal progenitors warrant further investigation. Some cell proliferation parameters were significantly impacted by passage number. It has long been laboratory dogma that trypsinization can help “boost” cell cultures into increasing proliferation rates initially, whereas extending time to passage can negatively impact population doubling times and long-term cell viability [43]. In this context, it was not surprising to see that ‘days to confluence’ tended to decrease while ‘end of passage cell counts’ tended to increase with each subsequent passage. Had the experiment been taken out to longer passages, these trends may have regressed due to cell senescence [44]. It is also not surprising that ‘population doubling times’ also tended to decrease with passage given that this parameter is a function of both days to confluence and end of passage cell counts. The initial primary cell numbers and viability likely influenced subsequent proliferative behavior. For example, BM-MSCs from Horse 1 had the shortest days to confluence across all passages compared to Horses 2 – 4, which would suggest that either the starting primary cell counts were higher or the cell proliferative ability was superior. It is also important to note that Horse 1 was the youngest donor, at 2 years of age, and increasing age has been shown to negatively impact number of viable MSCs and cell proliferative capacities in human donors [45 – 47]. Some studies have evaluated age-related effects on MSC behavior in veterinary species [48 – 50], but more investigation would be necessary in order to establish criteria for eligible donors to allogeneic stem cell banks. Osteogenic capacity of both ADI- and BM-MSCs was not impacted by donor status during collection, which was not surprising given that cell viability and proliferation were similarly unaffected. In general, BM-MSCs were more osteogenic than ADI-MSCs in osteogenic media alone, consistent with previous studies. However, cells from different donors and statuses during collection often had different osteogenic responses, and it was not consistent within donors and between cell types. For example, Horse 1’s BM-MSCs had a robust osteogenic

64 response, while Horse 2 had a single ADI-MSC sample that responded well to osteogenic media supplemented with BMP-2. These results graphically demonstrate that both inter- and intra- donor variability greatly influence MSC differentiation outcomes, with factors such as primary cell number, proliferative capacity [51], and donor-dependent responsiveness [52] likely contributing to this apparently stochastic phenomenon. The results of this study suggest that post euthanasia-collected tissue samples are a viable source of MSCs and behave similarly in vitro to within-donor matched samples collected under sedation and general anesthesia. Although allogeneic stem cell therapy faces many clinical barriers [53, 54], cadaver tissues serve as a promising source of MSCs. Future research is needed to determine any toxicity effects of systemic drugs (such as sedation and anesthetic agents and euthanasia solution) on MSCs, define the ideal timing of tissue collection after death, and optimize in vitro screening methods for characterizing MSC differentiation potentials from any given donor or tissue source.

Conclusions In conclusion, equine adult ADI- and BM-MSCs collected during standing sedation, general anesthesia, and post-euthanasia behaved equivalently in vitro in regards to cellular viability, proliferation, and osteogenic differentiation capacity outcome measures. Factors such as sample-specific osteogenic responsiveness, donor age, initial numbers of viable primary cells, and cell seeding density likely impacted in vitro cell behavior and should be considered when evaluating donor cell responses. The large degree of inter- and intra-donor variability seen in this study further supports optimization of in vitro multi-potential differentiation screening methods for selecting viable MSC sources for use in clinical therapeutic applications and cell banking.

Figures & Tables

Table 3.1: Donor Horse Information

Age Horse Gender Breed BCS Missing Sample (years) 1 2 Female intact Quarterhorse 3 Sedation ADI-MSC 2 8 Female intact Quarterhorse 7 None 3 8 Female intact Warmblood 6 Euthanasia BM-MSC 4 9 Male castrated Warmblood 4 Sedation BM-MSC

Table 3.1: Donor horse demographics, including age (years), gender, breed, body condition score (BCS), and notes pertaining to any missing samples.

Figure 3.1: Adipose Tissue Primary Cell Counts

***

Figure 3.1: ADI-MSC primary cell counts expressed as cells per gram of adipose tissue. Results are pooled across donors (top) and presented as individual donors (bottom). Asterisks (*) indicate a significant effect of donor variability on ADI-MSC primary cell counts (*** p < 0.001). Evaluated with 1-Way ANOVA, n = 3 – 4, p < 0.05.

Figure 3.2: Bone Marrow Aspirate Primary Cell Counts

Figure 3.2: BM-MSC primary cell counts expressed as cells per ml of aspirate and adjusted for days to confluence. Results are pooled across donors (top) and presented as individual donors (bottom). Asterisks (*) indicate a significant effect of donor variability on BM-MSC primary cell counts (** p < 0.01). Evaluated with 2-Way ANOVA, n = 3 – 4, p < 0.05.

Figure 3.3: Population Doubling Times

∆ ***

Figure 3.3: Population doubling times (PDTs) expressed as days for ADI-MSCs (top) and BM- MSCs (bottom). “P” = passage number. Asterisks (*) indicate a significant effect of passage on PDTs (*** p < 0.001). Deltas (Δ) indicate a significant effect of donor variability on PDTs (Δ p < 0.05). Evaluated with 2-Way ANOVA, n = 3 – 4, p < 0.05

Figure 3.4: ADI-MSC Alizarin Red Stain Images

Horse 3 Cx OIM OIM + BMP-2 ADI-MSC

Sedation Collected Sample

Anesthesia Collected Sample

Euthanasia Collected Sample

Figure 3.4: Representative 4X magnification images of Day 10 post-osteogenic induction multi- cellular aggregation and Alizarin Red staining in ADI-MSCs samples from Horse 3. “Cx” = control media, “OIM” = osteogenic media, and “BMP-2” = recombinant human bone morphogenetic protein 2 supplemented at 100 ng/ml.

Figure 3.5: BM-MSC Alizarin Red Stain Images

Horse 1 Cx OIM BM-MSC

Sedation Collected Sample

Anesthesia Collected Sample

Euthanasia Collected Sample

Figure 3.5: Representative 4X magnification images of Day 10 post-osteogenic induction multi- cellular aggregation and Alizarin Red staining in BM-MSCs samples from Horse 1. “Cx” = control media, and “OIM” = osteogenic media.

Figure 3.6: ADI-MSC RUNX2 mRNA Expression

∆ ** ****

Figure 3.6: RUNX2 mRNA expression in ADI-MSCs evaluated at Day 5 (left) and Day 10 (right) post-osteogenic induction. “Cx” = control media, “OIM” = osteogenic media, and “BMP-2” = recombinant human bone morphogenetic protein 2 supplemented at 100 ng/ml. Asterisks (*) indicate a significant effect of culture treatment on RUNX2 expression (**** p < 0.0001, ** p < 0.01). Deltas (∆) indicate a significant effect of donor status during collection on RUNX2 expression (∆ p < 0.05). Carets (^) indicate significant post- hoc test difference from the sedation-collected sample within the culture treatment group (^ p < 0.05). Evaluated with 2-Way ANOVA, n = 3 – 4, p < 0.05. Y-axes ranges are equivalent to the corresponding BM-MSC RUNX2 mRNA expression figure (Figure 3.9).

Figure 3.7: ADI-MSC OSX mRNA Expression

Figure 3.7: OSX mRNA expression in ADI-MSCs evaluated at Day 5 (left) and Day 10 (right) post-osteogenic induction. “Cx” = control media, “OIM” = osteogenic media, and “BMP-2” = recombinant human bone morphogenetic protein 2 supplemented at 100 ng/ml. Evaluated with 2-Way ANOVA, n = 3 – 4, p < 0.05. Y-axes ranges are equivalent to the corresponding BM-MSC OSX mRNA expression figure (Figure 3.10).

Figure 3.8: ADI-MSC ALP mRNA Expression

* **

Figure 3.8: ALP mRNA expression in ADI-MSCs evaluated at Day 5 (left) and Day 10 (right) post-osteogenic induction. “Cx” = control media, “OIM” = osteogenic media, and “BMP-2” = recombinant human bone morphogenetic protein 2 supplemented at 100 ng/ml. Asterisks (*) indicate a significant effect of culture treatment on ALP expression (** p < 0.01, * p < 0.05). Evaluated with 2- Way ANOVA, n = 3 – 4, p < 0.05. Y-axes ranges are equivalent to the corresponding BM-MSC ALP mRNA expression figure (Figure 3.11).

Figure 3.9: BM-MSC RUNX2 mRNA Expression

* *

Figure 3.9: RUNX2 mRNA expression in BM-MSCs evaluated at Day 5 (left) and Day 10 (right) post-osteogenic induction. “Cx” = control media, and “OIM” = osteogenic media. Asterisks (*) indicate a significant effect of culture treatment on RUNX2 expression (* p < 0.05). Evaluated with 2-Way ANOVA, n = 3 – 4, p < 0.05. Y-axes ranges are equivalent to the corresponding ADI-MSC RUNX2 mRNA expression figure (Figure 3.6).

Figure 3.10: BM-MSC OSX mRNA Expression

Figure 3.10: OSX mRNA expression in BM-MSCs evaluated at Day 5 (left) and Day 10 (right) post-osteogenic induction. “Cx” = control media, and “OIM” = osteogenic media. Asterisks (*) indicate a significant effect of culture treatment on OSX expression (* p < 0.05). Evaluated with 2-Way ANOVA, n = 3 – 4, p < 0.05. Y-axes ranges are equivalent to the corresponding ADI-MSC OSX mRNA expression figure (Figure 3.6).

Figure 3.11: BM-MSC ALP mRNA Expression

** *

Figure 3.11: ALP mRNA expression in BM-MSCs evaluated at Day 5 (left) and Day 10 (right) post-osteogenic induction. “Cx” = control media, and “OIM” = osteogenic media. Asterisks (*) indicate a significant effect of culture treatment on ALP expression (** p < 0.01, * p < 0.05). Evaluated with 2-Way ANOVA, n = 3 – 4, p < 0.05. Y-axes ranges are equivalent to the corresponding ADI-MSC ALP mRNA expression figure (Figure 3.8).

Figure 3.12: ADI-MSC ALP Enzymatic Activity

Figure 3.12: ALP enzymatic activity in ADI-MSCs evaluated at Day 5 (left) and Day 10 (right) post-osteogenic induction. “Cx” = control media, “OIM” = osteogenic media, and “BMP-2” = recombinant human bone morphogenetic protein 2 supplemented at 100 ng/ml. Evaluated with 2-Way ANOVA, n = 3 – 4, p < 0.05. Y-axes ranges are equivalent to the corresponding BM-MSC ALP enzymatic activity figure (Figure 3.13).

Figure 3.13: BM-MSC ALP Enzymatic Activity

****

Figure 3.13: ALP enzymatic activity in BM-MSCs evaluated at Day 5 (top) and Day 10 (bottom) post-osteogenic induction. “Cx” = control media and “OIM” = osteogenic media. Asterisks (*) indicate a significant effect of culture treatment on ALP enzymatic activity (**** p < 0.0001). Evaluated with 2-Way ANOVA, n = 3 – 4, p < 0.05. Y-axes ranges are equivalent to the corresponding ADI-MSC ALP enzymatic activity figure (Figure 3.12).

Table 3.2: Primers Utilized in the qPCR Reactions

Table 3.2: Sense (S) and antisense (A) sequences for primers utilized in the qPCR reactions, including optimum annealing temperature ranges as determined by gradient PCR.

References

1. Taylor, S.E. & Clegg, P.D. (2011) Collection and Propagation Methods for Mesenchymal Stromal Cells. Vet Clin North Am Equine Pract. 27(2): 263-274.

2. Stewart, M.C. & Stewart, A.A. (2011) Mesenchymal Stem Cells: Characteristics, Sources, and Mechanisms of Action. Vet Clin North Am Equine Pract. 27(2): 243-261.

3. Mosna, F., Sensebe, L., & Krampera, M. (2010) Human bone marrow and adipose tissue mesenchymal stem cells: a user’s guide. Stem Cells Dev. 19(10): 1449-1470.

4. Dominici, M., Le Blanc, K., Mueller, I., et al (2006) Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy. 8(4): 315-317.

5. Pittenger, M.F., Mackay, A.M., Beck, S.C., et al (1999) Multilineage potential of adult human mesenchymal stem cells. Science. 284(1999): 143-147.

6. Bouffi, C., Bony, C., Courties, G., et al (2010) IL-6-dependent PGE2 secretion by mesenchymal stem cells inhibits local inflammation in experimental arthritis. PLoS One. 5(2010): e14247.

7. Prockop, D.J. (2013) Concise review: two negative feedback loops place mesenchymal stem/stromal cells at the center of early regulators of inflammation. Stem Cells. 31(2013): 2042-2046.

8. Mahla, R.S. (2016) Stem cells applications in regenerative medicine and disease therapeutics. Int J Cell Biol. Article ID: 6940283.

9. Penfornis, P. & Pochampally, R. (2011) Isolation and expansion of mesenchymal stem cells/multipotential stromal cells from human bone marrow. Methods Mol Biol. 698: 11-21.

10. Zuk, P.A., Zhu, M., Ashjian, P., et al (2002) Human adipose tissue is a source of multipotent stem cells. Mol Biol Cell. 13(12): 4279-4295.

11. Marx, C., Silveira, M.D., & Beyer Nardi, N. (2015) Adipose-derived stem cells in veterinary medicine: characterization and therapeutic applications. Stem Cells Dev. 24(7): 803-813.

12. Bahamondes, F., Flores, E., Cattaneo, G., et al (2017) Omental adipose tissue is a more suitable source of canine Mesenchymal stem cells. BMC Vet Res. 13(1): 166

13. Clegg, P.D. (2011) Musculoskeletal disease and injury, now and in the future. Part 1: Fractures and fatalities. Equine Vet J. 43(6): 643-649.

14. Clegg, P.D. (2012) Musculoskeletal disease and injury, now and in the future. Part 2: Tendon and ligament injuries. Equine Vet J. 44(3): 371-375.

15. Reed, S.A. & Leahy, E.R. (2013) Growth and Development Symposium: Stem cell therapy in equine tendon injury. J Anim Sci. 91(1): 59-65.

16. Ortved, K.F. & Nixon, A.J. (2016) Cell-based cartilage repair strategies in the horse. Vet J. 208: 1-12.

17. Govoni, K.E. (2015) HORSE SPECIES SYMPOSIUM: Use of mesenchymal stem cells in fracture repair in horses. J Anim Sci. 93(3): 871-878.

18. Milner, P.I., Clegg, P.D., & Stewart, M.C. (2011) Stem Cell–based Therapies for Bone Repair. Vet Clin North Am Equine Pract. 27(2): 299-314.

19. Prasongchean, W. & Ferretti, P. (2012) Autologous stem cells for personalised medicine. N Biotechnol. 29(6): 641-650.

20. Jin, X., Lin, T., & Xu, Y. (2016) Stem Cell Therapy and Immunological Rejection in Animal Models. Curr Mol Pharmacol. 9(4): 284-288.

21. Bosi, A. & Bartolozzi, B. (2010) Safety of bone marrow stem cell donation: a review. Transplant Proc. 42(6): 2192-2194.

22. Martí, J., Antón, E., Valentí, C. (2004) Complications of bone marrow biopsy. British Journal of Haematology. 124(2): 557-558.

23. Bain, B.J. (2005) Bone marrow biopsy morbidity: review of 2003. J Clin Pathol. 58(4): 406– 408.

24. Mamidi, M.K., Dutta, S., Bhonde, R., et al (2014) Allogeneic and autologous mode of stem cell transplantation in regenerative medicine: which way to go? Med Hypotheses. 83(6): 787- 791.

25. Feldmann Jr., R.E., & Mattern, R. (2006) The human brain and its neural stem cells postmortem: from dead brains to live therapy. Int J Legal Med. 120: 201-211.

26. Shikh Alsook, M.K., Gabriel, A., Piret, J., et al (2015) Tissues from equine cadaver ligaments up to 72 hours of post-mortem: a promising reservoir of stem cells. Stem Cell Res Ther. 6: 253-262.

27. Hodgetts, S.I., Stagg, K., Sturm, M., et al (2014) Long live the stem cell: the use of stem cells isolated from post mortem tissues for translational strategies. Int J Biochem Cell Biol. 56: 74-81.

28. Friedlis, M.F., & Centeno, C.J. (2016) Performing a Better Bone Marrow Aspiration. Phys Med Rehabil Clin N Am. 27(4): 919-939.

29. Booth, G.S., Gehrie, E.A., Bolan, C.D., & Savani, B.N. (2013) Clinical guide to ABO- incompatible allogeneic stem cell transplantation. Biol Blood Marrow Transplant. 19(8): 1152-1158.

30. Nardi, N.B. & Camassola, M. (2011) Isolation and culture of rodent bone marrow-derived multipotent mesenchymal stromal cells. Methods Mol Biol. 698: 151-60.

31. Bianchessi, M., Chen, Y., Durgam, S., et al (2016) Effect of Fibroblast Growth Factor 2 on Equine Synovial Fluid Chondroprogenitor Expansion and Chondrogenesis. Stem Cells Int. Article ID: 9364974.

32. Livak, K.J. & Schmittgen, T.D. (2001) Analysis of Relative Gene Expression Data Using Real-Time Quantitative PCR and the 2-ΔΔCT Method. Methods. 25: 402-408.

33. Gregory, C.A., Gunn, W.G., Peister, A., & Prockop, D.J. (2003) An Alizarin red-based assay of mineralization by adherent cells in culture: comparison with cetylpyridinium chloride extraction. Analytical Biochemistry. 329: 77–84.

34. Culley, D.J., Baxter, M., Yukhananov, T., & Crosby, G. (2003) The memory effects of general anesthesia persist for weeks in young and aged rats. Anesth. Analg. 96(4): 1004- 1009.

35. Tung, A., Herrera, S., Fornal, C.A., & Jacobs, B.L. (2008) The effect of prolonged anesthesia with isoflurane, propofol, dexmedetomidine, or ketamine on neural cell proliferation in the adult rat. Anesth Analg. 106(6): 1772-1777.

36. Erasso, D.M., Chaparro, R.E., Quiroga Del Rio, C.E., et al (2012) Quantitative assessment of new cell proliferation in the dentate gyrus and learning after isoflurane or propofol anesthesia in young and aged rats. Brain Res. 1441: 38-46.

37. Wu, Y.Q., Liang, T., Huang, H., et al (2014) Ketamine inhibits proliferation of neural stem cell from neonatal rat hippocampus in vitro. Cell Physiol Biochem. 34(5): 1792-1801.

38. Cavallo, C., Cuomo, C., Fantini, S., et al (2011) Comparison of alternative mesenchymal stem cell sources for cell banking and musculoskeletal advanced therapies. J Cell Biochem. 112(5): 1418-1430.

39. Cortes, Y., Ojeda, M,. Araya, D., et al (2013) Isolation and multilineage differentiation of bone marrow mesenchymal stem cells from abattoir-derived bovine fetuses. BMC Vet Res. 9: 133.

40. Schrӧck, C., Eydt, C., Geburek, F., et al (2017) Bone marrow-derived multipotent mesenchymal stromal cells from horses after euthanasia. Vet Med Sci. 3(4): 239-251.

41. Dominici, M., Le Blanc, K., Mueller, I., et al (2006) Minimal criteria for defining multipotent mesenchymal stromal cells. The International Society for Cellular Therapy position statement. Cytotherapy. 8(4): 315-317.

42. De Schauwer, C., Meyer, E., Van de Walle, G.R., & Van Soom, A. (2011) Markers of stemness in equine mesenchymal stem cells: a plea for uniformity. Theriogenology. 75(8): 1431-1443.

43. Lennon, D.P., Schluchter, M.D., & Caplan, A.I. (2012) The effect of extended first passage culture on the proliferation and differentiation of human marrow-derived mesenchymal stem cells. Stem Cells Transl Med. 1(4): 279-288.

44. Gruber, H.E., Somayaji, S., Riley, F., et al (2012) Human adipose-derived mesenchymal stem cells: serial passaging, doubling time and cell senescence. Biotech Histochem. 87(4): 303-311.

45. Baxter, M.A., Wynn, R.F., Jowitt, S.N., et al (2004) Study of telomere length reveals rapid aging of human marrow stromal cells following in vitro expansion. Stem Cells. 22(5): 675- 682.

46. Mareschi, K., Ferrero, I., Rustichelli, D., et al (2006) Expansion of mesenchymal stem cells isolated from pediatric and adult donor bone marrow. J Cell Biochem. 97(4): 744-754.

47. Choudhery, M.S., Badowski, M., Muise, A., et al (2014) Donor age negatively impacts adipose tissue-derived mesenchymal stem cell expansion and differentiation. J Transl Med.12: 8-21.

48. Volk, S.W., Wang, Y., & Hankenson, K.D. (2012) Effects of donor characteristics and ex vivo expansion on canine mesenchymal stem cell properties: implications for MSC-based therapies. Cell Transplant. 21(10): 2189-2200.

49. Guercio, A., Di Bella, S., Casella, S., et al (2013) Canine mesenchymal stem cells (MSCs): characterization in relation to donor age and adipose tissue-harvesting site. Cell Biol Int. 37(8): 789-798.

50. Carter-Arnold, J.L., Neilsen, N.L., Amelse, L.L., et al (2014) In vitro analysis of equine, bone marrow-derived mesenchymal stem cells demonstrates differences within age- and gender-matched horses. Equine Vet J. 46(5): 589-595.

51. Siddappa, R., Licht, R., van Blitterswijk, C., & de Boer, J. (2007) Donor variation and loss of multipotency during in vitro expansion of human mesenchymal stem cells for bone tissue engineering. J Orthop Res. 25(8): 1029-1041.

52. Vanhatupa, S., Ojansivu, M., Autio, R., et al (2015) Bone Morphogenetic Protein-2 Induces Donor-Dependent Osteogenic and Adipogenic Differentiation in Human Adipose Stem Cells. Stem Cells Transl Med. 4(12): 1391-1402.

53. Strom, T.B., Field, L.J., & Ruediger, M. (2004) Allogeneic stem cell-derived "repair unit" therapy and the barriers to clinical deployment. J Am Soc Nephrol.15(5): 1133-1139.

54. Harris, D.T. (2014) Stem Cell Banking for Regenerative and Personalized Medicine. Biomedicines. 2(1): 50-79.

CHAPTER 4 Characterization of the RUNX2 and OSX Regulatory Axis in Equine Mesenchymal Stem Cells During In Vitro Osteogenesis

Abstract Osteogenesis is a complex process involving the differentiation of mesenchymal stem cell (MSC) precursors into osteoblasts. Two master regulators of osteogenesis are the transcription factors runt-related transcription factor 2 (RUNX2) and osterix/SP7 (OSX). In developmental models, both RUNX2 and OSX were found to be necessary for skeletogenesis, with RUNX2 acting upstream of OSX. MSCs, by definition, are capable of multi-lineage differentiation, yet the osteogenic capacity of MSCs from different tissue sources varies considerably. Bone marrow-derived MSCs exhibit more potent osteogenic capacity than several other putative progenitor populations, including adipose-derived (ADI) MSCs. The purposes of this study were to 1) compare baseline and osteogenesis-induced RUNX2 and OSX expression levels in equine adult ADI- and BM-MSCs, and 2) determine the impact of OSX over-expression on the in vitro osteogenic capacity of ADI-MSCs. The hypotheses were that 1) the relative osteogenic capacities of ADI- and BM-MSCs are consistent with resting and inducible RUNX2 and OSX expression levels, and 2) increasing OSX over-expression in ADI-MSCs would improve their osteogenic capacity. Results show that BM-MSCs have significantly higher OSX, but not RUNX2, baseline mRNA expression levels compared to ADI-MSCs. Osteogenic induction of RUNX2 was similar between cell types, but only BM-MSCs significantly upregulated OSX. OSX over-expression in ADI-MSCs did not result in significant improvements in osteogenic outcomes. In conclusion, the osteogenic response of ADI- and BM-MSCs may be due in part to differences in their RUNX2-OSX response axes, but other factors also have a substantial influence on osteogenic capacity. Over-expression of OSX is not sufficient by itself for improving osteogenic differentiation.

Introduction Bone formation during embryonic development (skeletogenesis) and fracture repair requires specialized cells called osteoblasts that are derived through osteogenesis. Mesenchymal stem cells (MSCs) give rise to osteo-chondro progenitor cells that are capable of terminally differentiating into chondrocytes or osteoblasts. During osteogenesis, many internal and external 86 cues, such as growth factors, hormones, signaling pathways, and transcription factors, dictate the response of the cells to osteogenic stimuli at specific time points throughout the process [reviewed in 1]. Osteoprogenitor cells express two main transcription factors: runt-related transcription factor 2 (RUNX2), also known as core-binding factor subunit alpha-1, and osterix (OSX), also known as transcription factor Sp7. RUNX2 and OSX are considered to be “master regulators” of osteogenesis, as gene deletion studies in murine models have categorically demonstrated that both RUNX2 [2] and OSX [3] are required for skeletogenesis. Mice lacking either gene do not develop a mineralized appendicular skeleton. Further studies illustrated that OSX expression is dependent on RUNX2 induction during bone development [reviewed in 4], since RUNX2-null mice do not express OSX, while OSX-null mice do express RUNX2. Runt-related transcription factor genes encode for proteins with a common 128-amino acid long ‘Runt domain’ located at the N-terminal site, which is responsible for both the binding of DNA by the α subunit and hetero-dimerization with the non-DNA binding β subunit [reviewed in 5 – 7]. RUNX2 gene expression is transcriptionally regulated by a distal P1 and proximal P2 promoter, which results in two mRNA products (Type II and Type I, respectively) that differ in the 5’ region with identical 3’ ends [6, 7]. The RUNX2 gene spans approximately 200 kb and comprises eight exons: exons 2 through 8 encode the ATP binding site, glutamine/alanine-rich domain, runt homology domain, a nuclear localization signal, a proline/serine/threonine-rich domain, and a nuclear matrix targeting signal [6]. The Type II RUNX2 isoform is mainly associated with osteoblasts [8] and originates from an alternative translation start codon within exon 1 [9]. RUNX2 gene expression is regulated by multiple stimulatory and inhibitory transcription factors [reviewed in 5] and several signaling pathways such as transforming growth factor beta (TGF-β) / bone morphogenetic protein (BMP) signaling molecules, fibroblast growth factors (FGF), retinoic acid, and glucocorticoids [reviewed in 6]. RUNX2 protein primarily interacts with core binding factor beta (CBFβ), which enhances the DNA binding affinity of the Runt domain [10]. Downstream targets of RUNX2 include an osteoblast-specific element 2 (OSE2) binding element originally identified in the promoter region of the gene osteocalcin (OSC) [11]. The most common targets that are important or specific to osteoblast differentiation include bone sialoprotein (BSP), collagen type I alpha-1 (COL1A1), osteocalcin (OSC), and osteopontin (OSP) [reviewed in 5 – 7]. Other osteoblast

87 targets include ameloblastin, collagenase 3 / matrix metalloproteinase 13 (MMP-13), osteoprotegrin, and TGF-β type I receptor [reviewed in 6]. OSX is a zinc finger-containing transcription factor that is a member of the SP family of DNA-binding proteins [3]. The OSX gene contains only 2 exons: exon 1 encodes for 7 amino acids at the N-terminal region, while exon 2 contains the remaining open reading frame and 3’ untranslated region [3]. The OSX protein is approximately 428 amino acids long with a molecular mass of around 46 kDa [12]. The DNA binding domain of OSX consists of three C2H2-type zinc fingers at the C-terminus, which interact with specificity protein 1 (SP1) and Krueppel-like factor 1 (EKLF) consensus sequences and other G/C-rich sequences commonly found in promoters [reviewed in 12]. The OSX protein sequence also contains a proline- and serine-rich transactivation domain, close to the N-terminus, which is responsible for eliciting an inhibitory effect on the Wnt signaling pathway [13]. OSX gene expression is regulated by several RUNX2-dependent and -independent pathways [reviewed in 14, 15], and its expression is necessary to further differentiation precursor cells into mature, functional osteoblasts [3, reviewed in 16]. The upstream regulatory sequence of the OSX promoter contains response elements for distal-less homeobox 5 (DLX-5), RUNX2, SRY-like box 9 transcription factor (SOX-9), SP1, and vitamin D response element (VDRE) [15]. OSX is directly induced by RUNX2, DLX-5, the BMP, mitogen-activated protein kinase (MAPK), and insulin-like growth factor 1 (IGF-1) signaling pathways, and the endoplasmic reticulum (ER) stress pathway through inositol-requiring enzyme 1 alpha (IRE1a) / x-box binding protein 1 (XBP1) [reviewed in 15 – 18]. OSX can also regulate its own expression through positive feedback on its promoter [19]. Downstream targets of OSX that are partially or fully dependent on its expression and induction include alkaline phosphatase (ALP), BSP, COL1A1, OSC, osteonectin (OSN), and OSP [3, 16]. MSCs from adult tissues are a promising source of cell therapy for use in regenerative medicine applications. In veterinary medicine, MSCs are typically used in horses for treatment of orthopedic injuries or degenerative conditions and are commonly harvested via bone marrow (BM) aspiration or adipose (ADI) tissue collection [20]. There is an interest in optimizing MSCs for use in bone repair and regeneration [reviewed in 21], especially in species like the horse that often have increased morbidity and mortality from fracture [22]. The purpose of this study was to compare the RUNX2-OSX response axis in equine adult ADI- and BM-MSCs to determine its impact on their differing osteogenic capacities. The experiments were designed to first compare

88 baseline and osteogenesis-induced RUNX2 and OSX expression levels, and then determine the effect of adenovirus-directed OSX over-expression on the osteogenic capacity of ADI-MSCs. The hypotheses were that ADI-MSCs had lower baseline and induced RUNX2 and OSX expression levels than BM-MSCs, and therefore increasing expression of OSX would improve the osteogenic capacity of ADI-MSCs.

Materials & Methods Research Animals & Cell Samples All procedures were approved by the University of Illinois Institutional Animal Care and Use committee (IACUC). Healthy, adult horses between the ages of 2 – 10 years old were donated to the Veterinary Teaching Hospital for teaching and/or research purposes. Adipose tissue and bone marrow aspirates were collected and MSCs were isolated as outlined in Chapter 3. For the experiments outlined in this study, 4 – 6 different cell samples were used immediately after collection and isolation or were cryo-stored until required.

Cell Culture & Expansion Culture procedures are similar as those outlined in Chapter 3. Briefly, passage 2 – 3 cells were seeded in 100-mm dishes at approximately 5,000 cells/cm2 in growth media [1X Dulbecco’s modified Eagle’s medium (DMEM), 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 streptomycin μg/ml (1% P/S)]. Passage 2 – 3 cells were trypsinized at near- confluence and transferred to osteogenesis and adenovirus experiments. Cells were counted using the Trypan Blue exclusion method, as outlined in Chapter 3.

In Vitro Osteogenesis Third passage (P3) MSCs were seeded at 2,000 cells/cm2 in cell culture plates and maintained in growth media until approximately 50% confluence was reached. Control (Cx) cultures remained in growth media, while osteogenic cultures were be maintained with osteogenic induction media (OIM), which consisted of growth media base supplemented with 50 μg/ml L-Ascorbic Acid Phosphate Magnesium Salt n-Hydrate (Wako, Japan), 10 mM β- Glycerophospate (Sigma-Aldrich), and 100 nM Dexamethasone (Sigma-Aldrich). Cultures were maintained for up to 7 days, during which media was replaced every 2 – 3 days and

89 representative digital pictures were taken via light microscopy. Cells were collected for RNA isolation at multiple time points after osteogenic induction.

RNA Isolation & Quantification, Reverse Transcription, and Quantitative Real-Time PCR All procedures are outlined in detail in Chapter 3. Total RNA was isolated using the TRIzol® reagent protocol and quantified using a NanoDrop 2000 spectrophotometer. RNA was reverse transcribed into cDNA using the SuperScriptTM IV First-Strand Synthesis System kit protocol. Quantitative real-time PCR (qPCR) was performed using the iQTM SYBR® Green Supermix and iCycler iQTM thermocycler (BIO-RAD) protocol outlined in Chapter 3. Primers utilized in the qPCR reactions are listed in Table 4.1. Data were analyzed using the 2-ΔΔCT method by normalizing to the reference gene, EF1α, and an early time point control sample. Comparisons between ADI- and BM-MSCs were made by normalizing all data to BM control samples, since this has served as a reference or “positive” control in previous studies.

Adenovirus Optimization Adenoviral multiplicities of infection (MOI) were optimized in a dose-response observational trial. ADI-MSC samples were infected at 24 doses between 0.5 – 1000 MOI, maintained in either control (Cx) or osteogenic (OIM) media, and observed daily for 14 days to monitor cell viability and survival.

OSX Adenovirus Infection ADI-MSC samples maintained in Cx and osteogenic media were infected with an OSX/SP7 adenovirus (VectorBiolabs ADV-224052) to assess the impact of increased OSX expression on induction of downstream osteoprogenitor marker target genes and osteogenesis. The OSX adenovirus (Ad-OSX) contained a cDNA clone of a Homo sapiens OSX/SP7 mRNA (GenBank: BC113613.1) under control of a CMV promoter and co-expressing an enhanced green fluorescent protein (eGFP) tag. Comparisons between the adenovirus OSX transcript sequence and predicted Equus caballus mRNA sequences using BLAST® (NCBI) resulted in 92% mRNA homology and up to 98% protein sequence homology. An eGFP vehicle control adenovirus (Ad-GFP) was also used to determine any effect of adenovirus infection on cells. The eGFP product of both adenoviruses allowed for tracking of infection efficacy via fluorescent

90 microscopy. Based on adenovirus optimization results, both a mid-range MOI of 20 and high- range MOI of 100 were used to evaluate a dose-response effect of OSX over-expression on downstream gene targets and osteogenesis assay outcomes. Samples were maintained in Cx and osteogenic media and collected at days 4 and 7 post-infection for RNA isolation, ALP bioactivity, and Alizarin Red staining.

Osteogenesis Assays In addition to osteogenic gene expression via RT-qPCR described above, Alizarin Red mineralized matrix staining and ALP bioactivity assays were performed as outlined in Chapter 3.

Statistical Analyses All data were analyzed using GraphPad Prism (GraphPad Software, INC) and expressed as mean + the standard error of the mean (SEM). Data were evaluated for normality using the Kolmogorov-Smirnov, D’Agostino and Pearson, and Shapiro-Wilk normality tests, where appropriate. Normally distributed data was analyzed using the tests mentioned below for each comparison. If normality could not be determined due to small sample size or data did not pass any of the normality tests, a corresponding non-parametric test was also performed. Statistical significant was assigned at p < 0.05. Sample sizes were n = 4 horses per cell type for baseline and temporal gene expression, and n = 6 horses for OSX over-expression in ADI-MSCs data. Differences between ADI- and BM-MSC baseline mRNA expression of RUNX2 and OSX were evaluated using 2-Way Analysis of Variance (ANOVA). Comparisons between the two cell types were performed by normalizing to BM Day 0 samples. Bonferonni post-hoc testing was performed to compares between cell types at each collection time point. Temporal changes in mRNA expression of RUNX2 and OSX in control versus osteogenic media within each cell type over time were evaluated using 2-Way ANOVA. Bonferonni post-hoc testing was performed to compare control and osteogenic media values at each time point. Comparison between ADI- and BM-MSCs temporal changes in mRNA expression of RUNX2 and OSX in osteogenic media was evaluated at each time point using 2- Way ANOVA by normalizing to BM Day 2 control samples. Bonferonni post-hoc testing was performed to compare ADI- and BM-MSC osteogenic responses at each time point.

The effect of OSX over-expression on mRNA expression and ALP activity was evaluated using 2-Way ANOVA. Bonferonni post-hoc testing was performed to determine significant comparisons between OSX adenovirus and control treatments at each time point. The effect of osteogenic media compared to control media for each outcome measure was evaluated in a separate 2-Way ANOVA at each time point.

Results Baseline RUNX2 & OSX mRNA Expression RUNX2 mRNA levels in basal/resting MSC cultures did not differ between ADI- and BM-MSCs at any of the time point evaluated (Figure 4.1). RUNX2 expression overall was not affected by cell type or time in culture. OSX mRNA levels in basal/resting MSC cultures were significantly higher (p = 0.0037) in BM- compared to ADI-MSCs across all time points evaluated (Figure 4.1). Post-hoc testing found a statistically significant increase in OSX levels in BM- compared to ADI-MSCs at Day 2 (p < 0.01). OSX expression was not affected by time in culture. In summary, baseline OSX, but not RUNX2, mRNA expression levels are significantly higher overall in BM-MSCs compared to ADI-MSCs.

Osteogenesis-Induced RUNX2 & OSX mRNA Expression RUNX2 mRNA expression levels (Figure 4.2 (A)) were significantly increased by osteogenic media over all time points evaluated in both ADI- (p = 0.0013) and BM- (p = 0.0020) MSCs. Post-hoc testing found a statistically significant increase in RUNX2 expression in osteogenic media compared to control media in BM-MSCs at the Day 7 time point (p < 0.05). OSX mRNA expression levels (Figure 4.2 (A)) were significantly increased in osteogenic media compared to control media in BM-MSCs at the Day 7 time point (p < 0.05). However, there was no induction of OSX in ADI-MSC cultures. There was a significant difference between cell types (Figure 4.2 (B)) in the induction of both RUNX2 (p = 0.0180) and OSX (p = 0.0348). Further, OSX expression was significantly increased in osteogenic BM- compared to ADI-MSC cultures at the Day 7 time point (p < 0.01). In summary, both ADI- and BM-MSCs significantly up-regulated RUNX2 mRNA expression in response to osteogenic media overall, but only BM-MSCs showed a significant

92 increase in OSX mRNA expression in response to osteogenic media by Day 7. Comparison between the two cell types showed that BM-MSCs had significantly higher induced levels of RUNX2 and OSX overall compared to ADI-MSCs, with significantly higher OSX expression seen in Day 7 BM- compared to ADI-MSC osteogenic samples.

Adenoviral Vector Optimization GFP was detectable via fluorescence microscopy by day 2 post-infection at the higher doses tested, with intensity peaking by Day 7 and then gradually decreasing over time (images not shown). In Cx media, cells survived up to 14 days at the highest MOI tested, 1000. However, cell survival in OIM was impaired by adenoviral MOIs greater than MOI 100 past Day 7.

Effect of OSX Over-Expression on ADI-MSCs in Control Conditions Cell culture, GFP fluorescence intensity, and Alizarin Red staining images for ADI- MSCs infected with OSX adenovirus (Ad-OSX) at MOIs of 20 and 100 in control media are represented in Figure 4.3. At Day 4 post-infection, the higher Ad-OSX dose slightly increased cellular aggregation compared to the lower MOI and control cultures. However, this effect was not apparent at Day 7. GFP fluorescence levels reflected the adenovirus infective dose and were equivalent between Ad-GFP and Ad-OSX MOI 100 cultures. Alizarin Red staining was negative in control cultures at both the Day 4 and Day 7 time points. ALP mRNA expression and enzymatic activity are represented in Figure 4.4. ALP mRNA expression transiently increased in the Ad-OSX treated groups compared to controls in a dose-dependent manner, but the magnitude was not considered significant. ALP expression was significantly decreased at the Day 7 time point compared to the Day 4 time point (p = 0.0414). ALP enzymatic activity levels were not significantly affected by Ad-OSX treatment or time post- infection in control cultures. Downstream osteoprogenitor markers mRNA expression for bone sialoprotein (BSP), collagen type I A1 (COL1A1), osteocalcin (OSC), osteomodulin (OSM), osteonectin (OSN), and osteopontin (OSP) are represented in Figure 4.5. BSP was the only gene significantly increased by Ad-OSX treatment (p = 0.0487), but the effect was not dose-dependent. None of the genes were significantly affected by time post-infection. Post-hoc testing did not find a significant

93 increase in gene expression with either Ad-OSX dose compared to controls at either time point for any of the genes evaluated. In summary, OSX over-expression in ADI-MSCs under control conditions resulted in a slight increase in cellular aggregation at Day 4 in the higher dose tested, a non-significant dose- dependent increase in ALP mRNA expression at Day 4, and a significant increase in BSP mRNA expression at both time points compared to controls.

Effect of OSX Over-Expression on ADI-MSCs in Osteogenic Conditions Cell culture, GFP fluorescence intensity, and Alizarin Red staining images for ADI- MSCs infected with OSX adenovirus (Ad-OSX) at MOIs of 20 and 100 in osteogenic media are represented in Figure 4.6. At both Day 4 and Day 7 post-infection, there was no apparent increase in cellular aggregation in Ad-OSX treated cultures compared to control cultures. GFP fluorescence levels reflected the adenoviral infective dose and were equivalent between Ad-GFP and Ad-OSX MOI 100 cultures. Alizarin Red staining was negative in osteogenic cultures at both the Day 4 and Day 7 time points. Alizarin Red staining was positive in a single experiment extended to Day 14, but there was no apparent difference between Ad-GFP and Ad-OSX infected cultures regarding Alizarin Red intensity (data not shown). ALP mRNA expression and enzymatic activity are represented in Figure 4.7. ALP expression increased slightly in the Ad-OSX MOI 100 group compared to controls, but the difference was not significant. ALP expression overall was significantly increased at the Day 7 time point compared to the Day 4 time point (p = 0.0271). ALP activity levels were not significantly affected by Ad-OSX treatment or time post-infection. However, Day 7 values were generally higher than Day 4 levels. ALP mRNA expression, but not ALP enzymatic activity, was significantly increased in osteogenic compared to control media samples (p = 0.0252). Downstream osteoprogenitor markers mRNA expression for bone sialoprotein (BSP), collagen type I A1 (COL1A1), osteocalcin (OSC), osteomodulin (OSM), osteonectin (OSN), and osteopontin (OSP) are represented in Figure 4.8. None of the genes were significantly affected by Ad-OSX treatment. COL1A1 was significantly decreased at Day 7 compared to Day 4 (p < 0.0001). Osteogenic media resulted in a significant increase in OSM (p = 0.0368), significant decrease in COL1A1 (p < 0.0001) and OSP (p = 0.0036), and did not significantly affect BSP, OSC, and OSN compared to control media.

In summary, OSX over-expression in ADI-MSCs under osteogenic conditions did not affect cellular aggregation or Alizarin Red staining, resulted in a non-significant dose-dependent increase in ALP mRNA expression at both Day 4 and Day 7, and did not consistently affect any of the downstream osteoprogenitor marker genes at either time point compared to controls

Discussion This study sought to characterize differences in the RUNX2-OSX regulatory axis in equine adult adipose (ADI) and bone marrow (BM) derived mesenchymal stem cells (MSCs), to determine whether differences in expression profiles influenced the osteogenic capacity of these cells. The goals were to first evaluate baseline and osteogenesis-induced expression levels of each, and then determine whether OSX over-expression improved osteogenic differentiation outcomes in ADI-MSCs. The hypotheses were that ADI-MSCs had lower baseline and induced RUNX2 and OSX expression levels than BM-MSCs, and therefore increasing expression of OSX would improve their osteogenic capacity. Results suggest that there are consistent differences in OSX, but not RUNX2, expression between the two cell types, but compensating for this difference in ADI-MSCs by directly over-expressing OSX was not sufficient to improve their osteogenic capacity to a level comparable to BM-MSCs. Several studies have compared the osteogenic potential of MSCs from different tissue sources, in both animal models and humans. BM-MSCs have been found to have superior osteogenic capacity to ADI-MSCs in humans [23], rats [24, 25], dogs [26, 27], goats [28], and horses [29, 30]. Other comparative studies performed in horses have found that MSCs from multiple sources are capable of osteogenic differentiation [31, 32], but these studies did not evaluate differences in osteogenic gene expression profiles. RUNX2 and OSX are essential master transcriptional regulators for osteogenesis, and this study showed that there is a difference in the baseline ‘set point’ of OSX: BM-MSCs have significantly higher resting levels than ADI- MSCs. OSX is also a downstream target of RUNX2, and results here suggest that BM-MSCs induced OSX more than ADI-MSCs in response to osteogenic induction. In summary, it is likely that the differential osteogenic capacities of ADI- and BM-MSCs are due, in part, to their difference in OSX expression profiles and responses. There have been many biological and genetic engineering-based approaches proposed to improve MSC use in bone repair/regeneration [reviewed in 33 – 36]. In this study, adenoviruses

95 were used to over-express OSX and evaluate the impact on downstream osteogenic gene targets and osteogenesis. Adenoviruses are highly efficient at delivering DNA to cells, are not dependent on cell replication for successful infection, and typically do not integrate with the host genome. The adenovirus infection process involves cells binding to viral proteins through various surface receptors for adenoviruses and αv integrins, which are widely distributed among different cell types [37]. Adenoviruses typically remain in the nucleus of infected cells as an episome, which will gradually degrade as cells divide [38]. The observation of decreasing GFP fluorescence at later time points post-infection supports this, which suggests that any potential benefit of adenovirus-directed gene over-expression may be short-lived. This could also explain why in the OSX over-expression experiments, the increase in OSX expression around the same time as osteogenic induction may not have been sufficient to improve osteogenesis since OSX is required for mid-stage osteoblast differentiation [reviewed in 12, 16]. In fact, sustained OSX over-expression may actually inhibit terminal osteoblastogenesis [19]. The difference in baseline and inducible OSX expression between ADI- and BM-MSCs may be due to several upstream pathways. Both RUNX2-dependent and -independent pathways regulate OSX expression, notably Bone Morphogenetic Protein (BMP) signaling [14 – 18]. Other factors involved in OSX regulation include MAPK-p38/Erk1-2 and Insulin-like Growth Factor 1 signaling, and inhibitory influences by Tumor Necrosis Factor α (TNFα) and p53 [reviewed in 16]. A multitude of epigenetic-level events have been shown to regulate OSX and its target genes, such as microRNAs [39], DNA methylation [40], and histone demethylases [41]. Further studies should evaluate difference in the epigenetic landscape in ADI- and BM-MSCs pertaining to osteogenic regulatory genes, especially with recent advances in epigenetic screening techniques and the potential for using epigenetic approaches in MSC-based bone regeneration [reviewed in 42, 43]. OSX over-expression was insufficient for improving ADI-MSC osteogenesis. Several studies have evaluated the effects of RUNX2 or OSX over-expression on osteogenesis. In murine bone marrow MSCs, RUNX2 over-expression enhanced the osteogenic activity both in vitro and in vivo [44]. However, RUNX2 over-expression impaired osteoblastogenesis at later stages and lead to osteopenia in murine models [45], and RUNX2 can contribute to development of osteosarcoma [reviewed in 46], thus emphasizing the temporal importance of RUNX2 activities [47]. OSX over-expression induces osteogenic differentiation in vitro in murine embryonic stem

96 cells [48] and murine bone-marrow derived MSCs [49]. OSX also stimulates proliferation in murine bone-marrow derived MSCs [49] and NIH-3T3 fibroblasts [50]. However, in NIH-3T3 fibroblast cells, OSX over-expression was insufficient to induce osteoblastic differentiation [50]. It is evident that even though RUNX2 and OSX are considered “master regulators” of osteogenesis, their expression profiles and activity are tightly regulated and timed during osteoblast differentiation [1, 5, 6, 16]. Given that over-expression of either transcription factor by itself is not sufficient for complete osteoblast differentiation, it is likely that a combinatorial approach of molecular targets would be most effective in enhancing the osteogenic capacity of MSCs. For example, it has been proposed that the use recombinant BMP with genetically- engineered cells capable of expressing RUNX2 and OSX could achieve complete and robust osteogenesis [reviewed in 34].

Conclusion Equine adult BM-MSCs were found to have higher baseline and inducible gene expression levels of OSX, but not RUNX2, compared to ADI-MSCs. These findings may provide a partial explanation as to why BM-MSCs show superior osteogenic capacity to ADI- MSCs. Over-expression of OSX in ADI-MSCs slightly increased expression of the downstream genes ALP and BSP, but other osteogenic differentiation outcomes such as cellular aggregation, Alizarin Red mineralized matrix staining, and ALP enzymatic activity were not improved. Over- expression of OSX alone is not sufficient for improving osteogenic differentiation capacity of ADI-MSCs out to Day 7. A preliminary study using sustained OSX over-expression out to Day 14 also did not significantly improve osteogenic outcomes. Further studies should focus on determining the precise timing of inducible RUNX2 or OSX over-expression to maximize their benefit during osteogenesis, or combinatorial approaches that also enhance the BMP signaling cascade since it serves as an upstream regulator of both RUNX2 and OSX.

Figures & Table

Figure 4.1: Baseline RUNX2 & OSX mRNA Expression

** ^^

Figure 4.1: Baseline/resting mRNA expression levels of RUNX2 (top) and OSX (bottom) in adipose-derived (ADI) and bone marrow-derived (BM) MSCs. Asterisks (*) indicate a significant effect of cell type on mRNA expression (** p < 0.01). Carets (^) indicate significant post-hoc test differences between cell types at a given time point (^^ p < 0.01). Evaluated with 2-Way ANOVA, n = 4, p < 0.05.

Figure 4.2 (A): Osteogenic Media Induction of RUNX2 & OSX mRNA Expression

** ^^ ^

Figure 4.2 (A): Osteogenesis-induced mRNA expression levels of RUNX2 and OSX in ADI- (top) and BM- (bottom) MSCs. “Cx” = control media, and “OIM” = osteogenic media. Asterisks (*) indicate a significant effect of culture treatment on mRNA expression (** p < 0.01). Carets (^) indicate significant post-hoc test differences between culture treatments at a given time point (^ p < 0.05, ^^ p < 0.01). Evaluated with 2-Way ANOVA, n = 4, p < 0.05.

Continued on next page

Continued from previous page

Figure 4.2 (B): Comparative Osteogenic Media Induction of RUNX2 & OSX mRNA Expression

* * ^^

Figure 4.2 (B): Comparative osteogenesis-induced mRNA expression levels of RUNX2 and OSX in ADI- and BM-MSCs pooled for Day 2 (D2), Day 4 (D4), and Day 7 (D7). “Cx” = control media, and “OIM” = osteogenic media. Asterisks (*) indicate a significant effect of cell type on mRNA expression (* p < 0.05). Carets (^) indicate significant post-hoc test differences between cell types at a given time point (^^ p < 0.01). Evaluated with 2-Way ANOVA, n = 4, p < 0.05.

100

Figure 4.3: Effect of OSX Over-Expression on ADI-MSCs in Control Media: Representative Culture Images

Media & Time Point No Treatment + Ad-GFP (MOI 100) + Ad-OSX (MOI 20) + Ad-OSX (MOI 100)

Cx – Day 4 Alizarin Red Stain 4X Magnification

Cx – Day 4 GFP Fluorescence 4X Magnification

Cx – Day 7 Alizarin Red Stain 4X Magnification

Cx – Day 7 GFP Fluorescence 4X Magnification

Figure 4.3: Representative 4X magnification images of Day 4 and Day 7 post-adenovirus infection multi-cellular aggregation and Alizarin Red staining in ADI-MSCs maintained in control (Cx) media. “Ad-OSX” = Osterix adenovirus, “Ad-GFP” = enhanced GFP vehicle control adenovirus, and “MOI” = multiplicity of infection.

101

Figure 4.4: Effect of OSX Over-Expression on ADI-MSCs in Control Media: ALP mRNA Expression & Enzymatic Activity

Figure 4.4: ALP mRNA expression (left) and enzymatic activity (right) in ADI-MSC samples collected at Day 4 and Day 7 post-adenovirus infection and maintained in control (Cx) media. “Ad-OSX” = Osterix adenovirus, “Ad-GFP” = enhanced GFP vehicle control adenovirus, and “MOI” = multiplicity of infection. Asterisks (*) indicate a significant effect of time post- infection (* p < 0.05). Evaluated with 2-Way ANOVA, n = 6, p < 0.05.

102

Figure 4.5: Effect of OSX Over-Expression on ADI-MSCs in Control Media: Osteoprogenitor Marker Genes mRNA Expression

Figure 4.5: Downstream osteoprogenitor marker genes mRNA expression of bone sialoprotein (BSP), collagen type I A1 (COL1A1), osteocalcin (OSC), and osteomodulin (OSM) in ADI- MSC samples collected at Day 4 and Day 7 post-adenovirus infection and maintained in control (Cx) media. “Ad-OSX” = Osterix adenovirus, “Ad-GFP” = enhanced GFP vehicle control adenovirus, and “MOI” = multiplicity of infection. Asterisks (*) indicate a significant effect of culture treatment (* p < 0.05). Evaluated with 2-Way ANOVA, n = 6, p < 0.05.

Continued on next page

103

Continued from previous page

Figure 4.5: Downstream osteoprogenitor marker genes mRNA expression of osteonectin (OSN) and osteopontin (OSP) in ADI-MSC samples collected at Day 4 and Day 7 post-adenovirus infection and maintained in control (Cx) media. “Ad-OSX” = Osterix adenovirus, “Ad-GFP” = enhanced GFP vehicle control adenovirus, and “MOI” = multiplicity of infection. Evaluated with 2-Way ANOVA, n = 6, p < 0.05.

104

Figure 4.6: Effect of OSX Over-Expression on ADI-MSCs in Osteogenic Media: Representative Culture Images

Media & Time Point No Treatment + Ad-GFP (MOI 100) + Ad-OSX (MOI 20) + Ad-OSX (MOI 100)

OIM – Day 4 Alizarin Red Stain 4X Magnification

OIM – Day 4 GFP Fluorescence 4X Magnification

OIM – Day 7 Alizarin Red Stain 4X Magnification

OIM – Day 7 GFP Fluorescence 4X Magnification

Figure 4.6: Representative 4X magnification images of Day 4 and Day 7 post-adenovirus infection multi-cellular aggregation and Alizarin Red staining in ADI-MSCs maintained in osteogenic (OIM) media. “Ad-OSX” = Osterix adenovirus, “Ad-GFP” = enhanced GFP vehicle control adenovirus, and “MOI” = multiplicity of infection.

105

Figure 4.7: Effect of OSX Over-Expression on ADI-MSCs in Osteogenic Media: ALP mRNA Expression & Enzymatic Activity

Figure 4.7: ALP mRNA expression (left) and enzymatic activity (right) in ADI-MSC samples collected at Day 4 and Day 7 post-adenovirus infection and maintained in osteogenic (OIM) media. “Ad-OSX” = Osterix adenovirus, “Ad-GFP” = enhanced GFP vehicle control adenovirus, and “MOI” = multiplicity of infection. Asterisks (*) indicate a significant effect of time post- infection (* p < 0.05). Evaluated with 2-Way ANOVA, n = 6, p < 0.05.

106

Figure 4.8: Effect of OSX Over-Expression on ADI-MSCs in Osteogenic Media: Osteoprogenitor Markers Gene Expression

****

Figure 4.8: Downstream osteoprogenitor marker genes mRNA expression of bone sialoprotein (BSP), collagen type I A1 (COL1A1), osteocalcin (OSC), and osteomodulin (OSM) in ADI- MSC samples collected at Day 4 and Day 7 post-adenovirus infection and maintained in osteogenic (OIM) media. “Ad-OSX” = Osterix adenovirus, “Ad-GFP” = enhanced GFP vehicle control adenovirus, and “MOI” = multiplicity of infection. Asterisks (*) indicate a significant effect of culture treatment (**** p < 0.0001). Evaluated with 2-Way ANOVA, n = 6, p < 0.05.

Continued on next page

107

Continued from previous page

Figure 4.8: Downstream osteoprogenitor marker genes mRNA expression of osteonectin (OSN) and osteopontin (OSP) in ADI-MSC samples collected at Day 4 and Day 7 post-adenovirus infection and maintained in osteogenic (OIM) media. “Ad-OSX” = Osterix adenovirus, “Ad- GFP” = enhanced GFP vehicle control adenovirus, and “MOI” = multiplicity of infection. Evaluated with 2-Way ANOVA, n = 6, p < 0.05.

108

Table 4.1: Primers Utilized in the qPCR Reactions

Gene Target Primer Sequence & Start Site Annealing (Product Size) S = Sense, A = Antisense Temp(s) ALP S: 5’ – TGGGGTGAAGGCTAATGAGG – 3’ (463) 56 – 60oC (222 bp) A: 5’ – GGCATCTCGTTGTCCGAGTA – 3’ (684) BSP S: 5’ – AGAACACCACCCTGTCTACG – 3’ (418) 54 – 60oC (118 bp) A: 5’ – TTGTTTCCTTTGTCCCCAGC – 3’ (535) COL1A1 S: 5’ – AGCCAGCAGATCGAGAACAT – 3’ (3834) 54 – 64oC (304 bp) A: 5’ – CGCCATACTCGAACTGGAAT – 3’ (4137) EF1α S: 5’ – CCCGGACACAGAGACTTCAT – 3’ (277) 54 – 64oC (329 bp) A: 5’ – AGCATGTTGTCACCATTCCA – 3’ (605) OSC S: 5’ – CGGTACCTGGATCATTGGC – 3’ (151) 56 – 60oC (93 bp) A: 5’ – GTCACAGTCTGGGTTGAGC – 3’ (243) OSM S: 5’ – TACATCCGTGTGGACCAAAA – 3’ (1000) 54 – 62oC (200 bp) A: 5’ – TCTGGGCCTTCATGAGAATC – 3’ (1199) OSN S: 5’ – AACCTTCTGACCGAGAAGCA – 3’ (591) 54 – 64oC (191 bp) A: 5’ – TGGGACAGGTACCCATCAAT – 3’ (781) OSP S: 5’ – GCCCTTCCAGTTAATCAGGC – 3’ (170) 54 – 60oC (127 bp) A: 5’ – GTGGTGCTAGGAGATTCTGC – 3’ (296) OSX S: 5’ – GGCTATGCCAATGACTACCC – 3’ (337) 56 – 60oC (188 bp) A: 5’ – GGTGAGATGCCTGCATGGA – 3’ (524) RUNX2 S: 5’ – CAGACCAGCAGCACTCCATA – 3’ (1135) 56 – 60oC (178 bp) A: 5’ – CAGCGTCAACACCATCATTC – 3’ (1312)

Table 4.1: Sense (S) and antisense (A) sequences for primers utilized in the qPCR reactions, including optimum annealing temperature ranges as determined by gradient PCR.

109

References

1. Karsenty, G., Kronemberg, H.M., Settembre, C. (2009) Genetic control of bone formation. Annu Rev Cell Dev Biol. 25: 629-648.

2. Komori, T., Yagi, H., Nomura, S., et al (1997) Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell. 89(5): 755-764.

3. Nakashima, K., Zhou, X., Kunkel, G., et al (2002) The novel zinc finger-containing transcription factor osterix is required for osteoblast differentiation and bone formation. Cell. 108(1): 17-29.

4. Komori, T. (2005) Regulation of skeletal development by the Runx family of transcription factors. J Cell Biochem. 95(3): 445-453.

5. Liu, T.M., Lee, E.H. (2013) Transcriptional regulatory cascades in Runx2-dependent bone development. Tissue Eng Part B Rev. 19(3): 254-263.

6. Bruderer, M., Richards, R.G., Alini, M., Stoddart, M.J. (2014) Role and regulation of RUNX2 in osteogenesis. Eur Cell Mater. 28: 269-286.

7. Vimalraj, S., Arumugam, B., Miranda, P.J., Selvamurugan, N. (2015) Runx2: Structure, function, and phosphorylation in osteoblast differentiation. Int J Biol Macromol. 78:202-208

8. Stewart, M., Terry, A., Hu, M., et al (1997) Proviral insertions induce the expression of bone-specific isoforms of PEBP2alphaA (CBFA1): evidence for a new myc collaborating oncogene. Proc Natl Acad Sci USA. 94(16): 8646-8651.

9. Xiao, Z.S., Thomas, R., Hinson, T.K., et al (1998) Genomic structure and isoform expression of the mouse, rat and human Cbfa1/Osf1 transcription factor. Gene. 214(1-2): 187-197.

10. Ogawa, E., Maruyama, M., Kagoshima, H., et al (1993) PEBP2/PEA2 represents a family of transcription factors homologous to the products of the Drosophila runt gene and the human AML1 gene. Proc Natl Acad Sci U S A. 90(14): 6859-6863.

11. Ducy, P., Karsenty, G. (1995) Two distinct osteoblast-specific cis-acting elements control expression of a mouse osteocalcin gene. Mol Cell Biol. 15(4): 1858-1869.

12. Zhang, C. (2012) Molecular mechanisms of osteoblast-specific transcription factor Osterix effect on bone formation. Beijing Da Xue Xue Bao Yi Xue Ban. 44(5): 659-665.

13. Zhang, C., Cho, K., Huang, Y., et al (2008) Inhibition of Wnt signaling by the osteoblast- specific transcription factor Osterix. Proc Natl Acad Sci U S A. 105(19): 6936-6941

110

14. Lee, M.H., Kwon, T.G., Park, H.S., et al (2003) BMP-2-induced Osterix expression is mediated by Dlx5 but is independent of Runx2. Biochem Biophys Res Commun. 309(3): 689– 94.

15. Nishio, Y., Dong, Y., Paris, M., et al (2006) Runx2-mediated regulation of the zinc finger Osterix/Sp7 gene. Gene. 372: 62–70.

16. Sinha, K.M, Zhou, X. (2013) Genetic and molecular control of osterix in skeletal formation. J Cell Biochem. 114(5): 975-984.

17. Matsubara, T. (2008) BMP2 regulates osterix through Msx2 and Runx2 during osteoblast differentiation. J Biol Chem. 283(43): 29119-29125.

18. Nishimura, R., Hata, K., Matsubara, T., Wakabayashi, M., Yoneda, T. (2012) Regulation of bone and cartilage development by network between BMP signaling and transcription factors. J Biochem. 151(3): 247-254.

19. Yoshida, C.A., Komori, H., Maruyama, Z., et al (2012) SP7 inhibits osteoblast differentiation at a late stage in mice. PLoS One. 7(3): e32364.

20. Taylor, S.E. & Clegg, P.D. (2011) Collection and Propagation Methods for Mesenchymal Stromal Cells. Vet Clin North Am Equine Pract. 27(2): 263-274.

21. Milner, P.I., Clegg, P.D., & Stewart, M.C. (2011) Stem Cell–based Therapies for Bone Repair. Vet Clin North Am Equine Pract. 27(2): 299-314.

22. Govoni, K.E. (2015) HORSE SPECIES SYMPOSIUM: Use of mesenchymal stem cells in fracture repair in horses. J Anim Sci. 93(3): 871-878.

23. Heo, J.S., Choi, Y., Kim, H.S., Kim, H.O. (2016) Comparison of molecular profiles of human mesenchymal stem cells derived from bone marrow, umbilical cord blood, placenta and adipose tissue. Int J Mol Med. 37(1): 115-125.

24. Hayashi, O., Katsube, Y., Hirose, M., et al (2008) Comparison of osteogenic ability of rat mesenchymal stem cells from bone marrow, periosteum, and adipose tissue. Calcif Tissue Int. 82(3): 238-247.

25. Lotfy, A., Salama, M., Zahran, F., et al (2014) Characterization of mesenchymal stem cells derived from rat bone marrow and adipose tissue: a comparative study. Int J Stem Cells. 7(2): 135-142.

26. Kisiel, A.H., McDuffee, L.A., Masaoud, E., et al (2012) Isolation, characterization, and in vitro proliferation of canine mesenchymal stem cells derived from bone marrow, adipose tissue, muscle, and periosteum. Am J Vet Res. 73(8): 1305-1317.

111

27. Alves, E.G., Serakides, R., Boeloni, J.N., et al (2014) Comparison of the osteogenic potential of mesenchymal stem cells from the bone marrow and adipose tissue of young dogs. BMC Vet Res.10: 190

28. Elkhenany, H., Amelse, L., Caldwell, M., et al (2016) Impact of the source and serial passaging of goat mesenchymal stem cells on osteogenic differentiation potential: implications for bone tissue engineering. J Anim Sci Biotechnol.7: 16.

29. Vidal, M.A., Kilroy, G.E., Lopez, M.J., et al. (2007) Characterization of equine adipose tissue-derived stromal cells: adipogenic and osteogenic capacity and comparison with bone marrow-derived mesenchymal stromal cells. Vet Surg. 36(7): 613-622.

30. Toupadakis, C.A., Wong, A., Genetos, D.C., et al (2010) Comparison of the osteogenic potential of equine mesenchymal stem cells from bone marrow, adipose tissue, umbilical cord blood, and umbilical cord tissue. Am J Vet Res. 71(10): 1237-1245.

31. Radtke, C.L., Nino-Fong, R., Esparza Gonzalez, B.P., et al (2013) Characterization and osteogenic potential of equine muscle tissue- and periosteal tissue-derived mesenchymal stem cells in comparison with bone marrow- and adipose tissue-derived mesenchymal stem cells. Am J Vet Res. 74(5): 790-800.

32. Barberini, D.J., Freitas, N.P., Magnoni, M.S., et al (2014) Equine mesenchymal stem cells from bone marrow, adipose tissue and umbilical cord: immunophenotypic characterization and differentiation potential. Stem Cell Res Ther. 5(1): 25.

33. Franceschi, R.T. (2005) Biological approaches to bone regeneration by gene therapy. J Dent Res. 84(12): 1093-1103.

34. Satija, N.K., Gurudutta, G.U., Sharma, S., et al (2007) Mesenchymal stem cells: molecular targets for tissue engineering. Stem Cells Dev. 16(1):7-23.

35. Gersbach, C.A., Phillips, J.E., García, A.J. (2007) Genetic engineering for skeletal regenerative medicine. Annu Rev Biomed Eng. 9: 87-119.

36. Marie, P.J. (2011) Cell and gene therapy for bone repair. Osteoporos Int. 22(6): 2023-2026.

37. Neumann, R., Chroboczek, J., Jacrot, B. (1988). Determination of the nucleotide sequence for the penton-base gene of human adenovirus type 5. Gene. 69(1): 153-157.

38. Oligino, T.J., Yao, Q., Ghivizzani, S.C., Robbins, P. (2000). Vector systems for gene transfer to joints. Clin Orthop Rel Res. 379(Suppl): S17-S30.

39. Wang, C., Liao, H., Cao, Z. (2016) Role of Osterix and MicroRNAs in Bone Formation and Tooth Development. Med Sci Monit. 22: 2934-2942.

112

40. Farshdousti Hagh, M., Noruzinia, M., Mortazavi, Y., et al (2015) Different Methylation Patterns of RUNX2, OSX, DLX5 and BSP in Osteoblastic Differentiation of Mesenchymal Stem Cells. Cell J. 17(1): 71-82.

41. Sinha, K.M., Yasuda, H., Zhou, X., deCrombrugghe, B. (2014) Osterix and NO66 histone demethylase control the chromatin of Osterix target genes during osteoblast differentiation. J Bone Miner Res. 29(4): 855-865.

42. Im, G.I., Shin, K.J. (2015) Epigenetic approaches to regeneration of bone and cartilage from stem cells. Expert Opin Biol Ther. 15(2): 181-193.

43. Fu, G., Ren, A., Qiu, Y., Zhang, Y. (2016) Epigenetic Regulation of Osteogenic Differentiation of Mesenchymal Stem Cells. Curr Stem Cell Res Ther. 11(3): 235-246.

44. Zhao, Z., Zhao, M., Xiao, G., Franceschi, R.T. (2005). Gene transfer of the Runx2 transcription factor enhances osteogenic activity of bone marrow stromal cells in vitro and in vivo. Mol Ther. 12(2):247–253.

45. Liu, W., Toyosawa, S., Furuichi, T., et al (2001) Overexpression of Cbfa1 in osteoblasts inhibits osteoblast maturation and causes osteopenia with multiple fractures. J Cell Biol. 155(1): 157-166.

46. Li, N., Luo, D., Hu, X., et al (2015) RUNX2 and Osteosarcoma. Anticancer Agents Med Chem. 15(7): 881-887.

47. Gersbach, C.A., Le Doux, J.M., Guldberg, R.E., Garcia, A.J. (2006) Inducible regulation of Runx2–stimulated osteogenesis. Gene Ther. 13(11): 873–882.

48. Tai, G., Polak, J.M., Bishop, A.E., et al (2004) Differentiation of osteoblasts from murine embryonic stem cells by overexpression of the transcription factor osterix. Tissue Eng. 10(9- 10): 1456–1466.

49. Tu, Q., Valverde, P., Chen, J. (2006) Osterix enhances proliferation and osteogenic potential of bone marrow stromal cells. Biochem Biophys Res Commun. 341(4): 1257– 1265.

50. Kim, Y.J., Kim, H.N., Park, E.K., et al (2006). The bone-related Zn finger transcription factor Osterix promotes proliferation of mesenchymal cells. Gene. 366(1): 145–151.

113

CHAPTER 5 The Role of Bone Morphogenetic Protein Activity in Equine Mesenchymal Stem Cell In Vitro Osteogenesis

Abstract Stem cell therapies primarily utilize cells from bone marrow (BM) and adipose tissue (ADI). Although adipose-derived stem cells are easily accessible, the biological activities and responses of ADI cells are less robust than those of BM cells. BMP ligands are potent osteo- stimulants, both developmentally and in clinical applications. The following study addressed the hypothesis that the differential osteogenic capacity of BM-MSC and ADI-MSCs cells is a consequence of differences in BMP ligand expression and/or induction. Bone marrow aspirates and adipose cells collected from the sternebrae and subcutaneous fat depots, respectively, were seeded at low density and expanded through two passages. Third passage cells were transferred to control or osteogenic cultures for up to 10 days. Basal osteogenic BMP ligand expression was determined in undifferentiated BM- and ADI-MSCs, and induced BMP ligand expression was measured after transfer to osteogenic medium. Under basal conditions, both BMP-2 and -4 ligand expression was consistently 4 – 7 times higher in BM cells than in ADI cells. In contrast, there was little change in expression of either BMP ligand in ADI cells maintained in osteogenic medium. The effect of BMP signaling on in vitro osteogenesis was assessed by monitoring multi-cellular aggregation and mineralized matrix formation detected by Alizarin Red staining, qPCR analysis of osteogenic genes runt-related transcription factor 2 (RUNX2), osterix (OSX), and alkaline phosphatase (ALP), and ALP enzymatic activity as measured by an assay. Antagonism of endogenous BMP signaling in BM-MSCs inhibited osteogenesis, while exogenous BMP-2 administration in ADI-MSCs significantly increased cell aggregation, matrix mineralization, and Osterix (but not RUNX2) expression. These findings support the hypothesis that the differential osteogenic capacity of BM-MSC and ADI-MSCs cells is a consequence of differences in BMP ligand expression and/or induction. However, given that exogenous BMP-2 administration was not sufficient to confer a ‘bone marrow MSC’ osteogenic capacity in ADI- MSCs, other factors are clearly active in this process. Accepting the abundance of adipose tissue in most human patients and the comparative ease of adipose tissue collection, exogenous BMP delivery will likely be necessary if ADI-MSCs are used to support bone repair.

114

Introduction In recent decades, mesenchymal stem cell (MSC) therapies have become increasingly prominent in the treatment of musculoskeletal conditions, in both people and veterinary species. The utility of MSCs for therapeutic applications is based on their capacity for extensive proliferation, supporting the generation of large numbers of cells from small number of progenitors, their ability to differentiate along several mesenchymal lineages, and their immuno- modulatory and other trophic effects on host cells [1, 2]. Routinely, the differentiation capacity of putative MSC populations is assessed through in vitro tri-lineage differentiation along chondrogenic, adipogenic, and osteogenic lineages [3]. The standard assay for in vitro osteogenesis is exceptional in that no extrinsic growth factors are required to stimulate robust osteogenesis in competent cell populations. Supplementing medium with ascorbic acid, beta glycerophosphate, and dexamethasone is sufficient to induce the osteogenic phenotype [4, 5]. Bone Morphogenetic Proteins (BMPs) were originally identified as an activity within demineralized bone matrix that stimulates mineralization of soft tissue [6]. Since then, approximately twenty BMP ligands have been identified, and their importance in the development and homeostasis of the skeleton is well established [7]. Although there is considerable redundancy in skeletal BMP signaling activity, murine models have clearly demonstrated a ‘critical mass’ requirement for BMP signaling in skeletal development, postnatal growth, and fracture repair [8 – 10]. Two ligands, BMP-2 and -7, have been commercially developed for orthopedic use in spinal fusion and non-union procedures [11 – 13]. In equine practice, stem cell therapies primarily utilize progenitor populations isolated from bone marrow (BM) and adipose tissue (ADI) [3]. Although adipose-derived stem cells offer a very large and easily accessible resource, the biological activities and responses of ADI cells are less robust than those of BM cells in many experimental contexts [14 – 16]. This is particularly evident in comparative in vitro osteogenesis assessments [17 – 24]. Given the potent osteogenic activity of BMPs and the ability of BM-MSCs, but not ADI-MSCs, to undergo osteogenic differentiation in standard osteogenic medium, we addressed the hypothesis that the differential osteogenic capacity of BM-MSC and ADI-MSCs cells is a consequence of differences in BMP ligand expression and/or induction. To test this hypothesis, the following four questions were addressed: 1. Is basal osteo-chondral BMP expression equivalent in BM and ADI cells?

115

2. Is BMP ligand induction under osteogenic conditions equivalent in BM and ADI cells? 3. Does BMP signaling blockade inhibit BM-MSC osteogenesis? 4. Does exogenous BMP-2 ligand supplementation stimulate ADI-MSC osteogenesis?

Materials & Methods Bone Marrow and Adipose Tissue Collection Bone marrow aspirates and adipose tissue samples were collected from healthy adult horses, under approval by the University of Illinois Institutional Animal Care and Use committee (IACUC). Donor horses were sedated prior to collections, and collection sites were aseptically prepared and desensitized with local anesthetic. Bone marrow aspirates were collected from the 5th or 6th sternebrae, using a ventral midline approach and a Jamshidi biopsy needle. Adipose tissue was collected from the fat depot adjacent the tail head, following local desensitization of the biopsy site [25]. Chapter 3 contains more details on sample collection and processing.

MSC Isolation and Culture Both the bone marrow aspirates and adipose cells isolated by collagenase digestion were seeded at low density in DMEM + 10% FBS, supplemented with antibiotics and expanded through two passages. Assessment of BMP ligand expression, BMP inhibition in BM-MSC cultures, and exogenous recombinant human BMP-2 (Gemini Bioscience) supplementation in ADI-MSC osteogenic cultures was carried out using third passage (P3) cells. Cells were maintained in control (as above) or transferred to osteogenic induction medium (1X DMEM + 10% FBS supplemented with ascorbic acid, beta glycerophosphate, and dexamethasone), for up to 10 days. Chapter 3 contains more details regarding osteogenesis induction media and assays. To assess the effects of BMP signaling blockade on BM-MSC osteogenesis, osteogenic medium was supplemented with the extracellular BMP inhibitor recombinant Noggin protein (10 – 1000 ng/ml; R&D Biosystems), and small molecule BMP receptor kinase inhibitors DMH-1 (10 – 500 nM; Calbiochem) and K-02288 (10 – 500nM; TOCRIS). A panel of phenotypic assays was used to assess differential BMP expression and induction, the effect of BMP inhibition on BM-MSC osteogenesis, and exogenous BMP supplementation on ADI-MSC osteogenesis, as described below.

116

Phenotypic Assays Expression of osteogenic BMP-2, -4, -6, and -7 ligand transcripts and of osteogenic genes [26 – 30] was assessed by quantitative PCR (qPCR). Induction of alkaline phosphatase (ALP) activity was measured by a commercial enzymatic assay, and generation of mineralized matrix was monitored by Alizarin Red staining [30]. See Chapter 3 for more details on methodology.

MSC Aggregation The formation of prominent multi-cellular aggregates is a characteristic early feature of osteogenic differentiation in competent cell populations. To monitor the effects of BMP inhibition on this process in BM-MSC cultures, aggregates in ten random low power (10X objective) microscopic fields were counted on days 3, 7, and 10 after transfer to osteogenic medium. Multi-cellular aggregates were observed, but not quantified, in ADI-MSC cultures supplemented with BMP-2 during osteogenesis.

Alizarin Red Staining To identify cellular matrix mineralization, a 2% Alizarin Red solution was prepared to stain the cell layers. Culture medium was aspirated from culture wells, the monolayers were washed with phosphate-buffered saline (PBS), and the cell layers were fixed with 4% formaldehyde at room temperature for 30 minutes. After fixation, the cell layers were washed three times with ddH2O. A freshly prepared 2% Alizarin Red (Sigma-Aldrich) solution (pH 4.1) was added to each well, ensuring that the fixed monolayers were completely submerged. After

20 minutes, the stain was aspirated and the wells were washed 3 – 6 times with ddH2O until the wash solution was clear. Representative microscopic images were obtained using dry, stained cells (Leica Microsystems, Leica Application Suite -LAS-version 4.0.R1).

RNA Isolation and Reverse Transcription-Quantitative PCR At appropriate collection times, culture medium was completely aspirated from the wells and the phenol-based dissociation agent, TRIZol® (Invitrogen Corporation, Carlsbad, CA), was added directly to the cell monolayers. The resultant lysates were scraped from the culture wells and transferred to 15 mL falcon tubes. The lysates were homogenized using an Ultra Turrax-T 25 homogenizer (Janke & Kunkel, IKA-Labortechnik, Staufen, Germany) to completely disrupt the

117 cells. Total RNA was isolated using the phenol-chloroform dissociation, and isopropanol precipitation protocol recommended by the manufacturer. One microgram of total RNA was reverse-transcribed using Superscript First-Strand Synthesis System for RT-PCR (Invitrogen) to generate cDNA. Quantitative PCR was performed using SYBR Green Supermix (Bio-Rad Laboratories, Hercules, CA) and a BioRad iCycler (Bio- Rad). Relative expression of BMPs-2, -4, -6, and -7 and osteoblast-specific genes alkaline phosphatase (ALP), runt-related transcription factor 2 (RUNX2), and Osterix (OSX) were derived from threshold cycle values. QPCR data were normalized to expression of the reference gene, elongation factor-1 alpha (EF1α). Expression for each target was calculated using the 2-ΔΔCt method to measure relative fold changes in gene expression [31]. See Chapter 3 for more details. Primer sequences used in the qPCR reactions can be found in Table 5.1.

Alkaline Phosphate Enzymatic Activity Cells maintained in 12-well plates were harvested with 300 μL of 2% Triton X-100 in 1X TE Buffer per well and the samples were stored in -20°C until further processing. After thawing, each sample was homogenized using IKA Labortechnik T 25 basic homogenizer (Janke & Kunkel GmbH and Co., Staufen, Germany) and centrifuged at 2,500 rpm for 15 minutes at 4°C. A working assay solution was created by a substrate tablet containing p-nitrophenyl phosphate dissolved into 5 mL of buffer solution using an ALP assay kit (Wako, Japan). Twenty microliters of each sample supernatant were assayed for ALP activity in duplicates in a transparent 96-well plate. A 100 μL aliquot of working assay solution was transferred into each well. Along with the samples, a standard of p-nitrophenol solution was run in duplicate. Each sample and standard of the substrate p-nitrophenyl phosphate was hydrolyzed into p-nitrophenol by ALP [32]. The reactions were routinely run for 10 minutes, generating a yellow product that was optically measured at 405 nm wavelength (FLUOstar OPTIMA, BMG, Lab Technologies). ALP enzymatic activity was determined on the basis of release of 1 nmol of p-nitrophenol per minute. The ALP activity of each sample normalized to the corresponding DNA content. The means and standard errors of the mean were calculated for each duplicate sample.

118

Pico Green DNA Assay DNA content was measured using the Pico Green DNA kit (Quanti-iTTM PicoGreen dsDNA, Invitrogen). Samples were diluted 1:5 in 1X TE buffer (10mM Tris-HCl, 1mM EDTA, pH 7.5), and 100 μL of each sample was pipetted in duplicate into a black 96-well microplate. A standard curve was generated by serial dilutions of calf thymus DNA. One hundred microliters of Pico Green reagent (1 μL Pico Green reagent diluted in 200 μL of 1X TE buffer) was added to each sample and standard. Following a 5 minute incubation, protected from light to prevent reagent photo-degradation, the fluorescence was measured at 485 nm (FLUOstar OPTIMA, BMG, Lab Technologies).

Statistical Analyses All data were analyzed using GraphPad Prism (GraphPad Software, INC) and expressed as mean + the standard error of the mean (SEM). Data were evaluated for normality using the Kolmogorov-Smirnov, D’Agostino and Pearson, and Shapiro-Wilk normality tests, where appropriate. Normally distributed data was analyzed using the tests mentioned below for each comparison. If normality could not be determined due to small sample size or data did not pass any of the normality tests, a corresponding non-parametric test was also performed. Statistical significant was assigned at p < 0.05. Sample sizes were n = 3 – 4 horses for all experiments. Significance of pair-wise comparisons was determined using t tests, while multi-variable analyses were carried out by ANOVA. Differences in the baseline mRNA expression levels of BMP-2 and BMP-4 between ADI- and BM-MSC were evaluated using 2-Way Analysis of Variance (ANOVA). Comparisons between the two cell types were performed by normalizing to ADI-MSC values. Bonferonni post-hoc testing was performed to compares between cell types for each ligand. Differences in mRNA expression levels of BMP-2 and BMP-4 in control (Cx) and osteogenic (OIM) cultures was evaluated using 2-Way ANOVA. Bonferonni post-hoc testing was performed to compare between culture treatments for each ligand. The effect of a BMP inhibitor and on multi-cellular aggregate counts, ALP enzymatic activity levels, and osteogenic gene expression in BM-MSCs was evaluated at each collection time point using 2-Way ANOVA. Bonferonni post-hoc tests were performed to compare BMP inhibitor-treated samples to osteogenic (OIM) controls.

119

The effect of BMP-2 supplementation on osteogenic gene expression in ADI-MSCs was evaluated using 2-Way ANOVA. Bonferonni post-hoc tests were performed to compare BMP-2 supplemented samples to control (Cx) and osteogenic (OIM) cultures.

Results Osteogenic BMP Ligand Expression Under basal conditions, expression of BMP-2 was consistently 4 – 5 times higher in BM cells than in ADI cells, while BMP-4 expression was 6 – 7 times higher in BM cells (Figure 5.1). Transcript levels of BMPs-6 and -7, as reflected in Ct values at or above 30 cycles, were negligible in both cell types. After transfer into osteogenic medium, mean BMP-2 expression increased approximately four fold in BM cells by Day 7 and by approximately seven-fold by Day 14. Induction of BMP-4 expression followed a similar profile (Figure 5.2). In contrast, there was little or no change in expression of either BMP by ADI cells, which was comparable to basal expression in BM cells. In summary, BM-MSCs had significantly higher baseline and osteogenesis-induced mRNA expression levels of BMP-2 and BMP-4 ligands compared to ADI-MSCs.

BMP Inhibition in BM-MSC Osteogenic Cultures BMP signaling in differentiating BM-MSC cultures was inhibited by preventing interactions between secreted BMP ligand (recombinant human Noggin) and by inhibiting BMP receptor kinase function (DMH-1 and K-02288). All three inhibitors had similar effects on BM- MSC osteogenesis. Noggin dose-dependently and significantly inhibited BM-MSC aggregation and matrix mineralization at later stages (Figures 5.3 & 5.4). The BMP receptor kinase inhibitors had similar but more potent effects on aggregation: completely inhibiting aggregate formation at the 500 nM concentration (data not shown). All three BMP inhibitors significantly suppressed ALP activity in osteogenic cultures (Figure 5.5). As with the aggregation responses, the BMP receptor kinase inhibitors were more effective than Noggin in reducing ALP activity, particularly at the highest concentration used. BMP signaling inhibition had no significant effect on RUNX2 expression; however, OSX expression was significantly reduced by the 100 and 500 nM concentrations of K-02288 (Figure 5.6). BM-MSC responses to Noggin and DMH-1 were very similar (data not shown). BMP

120 signaling inhibition in BM-MSCs maintained in control media also produced the same effect on gene expression: significantly decreased OSX, but not RUNX2 (data not shown). In summary, BMP inhibition in BM-MSCs undergoing osteogenesis significantly decreased multi-cellular aggregation and Alizarin Red staining, ALP enzymatic activity, and OSX mRNA expression in a dose-dependent manner.

BMP-2 Supplementation in ADI-MSC Osteogenic Cultures The impact of supplementing osteogenic ADI-MSC cultures with exogenous BMP ligand was assessed by adding 100 ng/ml of recombinant human BMP-2 to culture medium. There was little or no indication that ADI-MSCs responded to osteogenic medium by aggregating or secreting a mineralized matrix (Figure 5.7). BMP-2 supplementation stimulated aggregate formation and mineralization of aggregate matrix by Day 10, analogously to BM-MSC cultures. ALP expression was increased modestly in osteogenic medium. Exogenous BMP-2 did increase ALP expression at later time points but, due to considerable inter-donor variability, this was not statistically significant (Figure 5.8). BMP-2 supplementation did not affect RUNX2 expression in osteogenic ADI-MSC cultures (Figure 5.9), but it did significantly increase OSX expression at all three time points analyzed compared to control or osteogenic media alone (Figure 5.10). In summary, BMP-2 supplementation in ADI-MSCs undergoing osteogenesis significantly increased Alizarin Red staining and OSX mRNA expression compared to osteogenic media alone. Both RUNX2 and ALP mRNA expression were increased in BMP-2 supplemented cultures, but not significantly more than osteogenic media alone due to inter-donor variability in responses.

Discussion This study was designed to answer four questions related to the role of intrinsic BMP signaling in equine MSC osteogenesis, using the highly osteogenic cells expanded from bone marrow aspirates and the relative non-osteogenic cells from adipose tissue as positive and negative biological controls, respectively, for this differentiation program. The first question addressed the relative expression of osteo-chondral BMP ligands in BM and ADI cells. Both BMP-2 and -4 were expressed at significantly higher levels in BM-MSCs than ADI-MSCs, prior

121 to exposure to any osteogenic stimulus. This differential expression persisted when third passage cells were transferred to osteogenic culture conditions. Both BMP-2 and -4 mRNAs were induced by osteogenic medium in BM-MSCs, whereas expression remained low and unresponsive in ADI-MSC cultures. The necessity for BMP signaling in BM-MSC osteogenesis was clearly confirmed, since competitive interference with BMP ligand binding to BMP receptor complexes through Noggin, and blockading BMP receptor kinase activity by DMH-1 and K-02288, had very similar and consistent inhibitory effects on BM-MSC aggregation, matrix mineralization (as detected by Alizarin Red stain retention), ALP activity, and OSX up-regulation. Of note, RUNX2 expression was not impacted, strongly suggesting that RUNX2 expression and induction during osteogenesis is not primarily regulated by the BMP pathway [33]. In several of the osteogenic outcomes assessed in these experiments, Noggin was less effective in antagonizing BMP activity than the small molecule inhibitors, despite being administered at supra-physiological concentrations (1000 ng/ml). This could be a consequence of restricted penetration of the recombinant protein from the culture medium into the dense, multicellular aggregates that form early in the process of osteogenesis, particularly given that BMP ligands secreted by BM-MSCs will be immediately available at the cell surface to act in autocrine and paracrine signaling before the exogenous ligand-binding antagonist can intervene. The final series of experiments addressed the possibility that exogenous BMP ligand delivery can compensate for the lack of intrinsic BMP expression in ADI-MSC cultures and impart a ‘bone marrow-like’ osteogenic capacity to these cells. BMP-2 supplementation did substantially increase expression of several osteogenic characteristics in ADI-MSCs, including cell aggregation and matrix mineralization, consistent with its known osteogenic activities [34, 35]. Although osteogenic medium itself increased ALP expression in ADI-MSCS, exogenous BMP did not consistently augment the primary response to osteogenic medium. Of note, BMP-2 had little detectable impact on RUNX2 expression but significantly up-regulated OSX, suggesting that BMPs mediate OSX expression via pathways independent of RUNX2, as occurs developmentally and in other experimental models [36, 37]. DLX-5 is a promising candidate for this RUNX2-independent regulatory pathway, based on findings in murine gene deletion models [38, 39]. Accepting that a single concentration of BMP-2 was used in these experiments, the results indicate that exogenous BMP-2 ligand does improve expression of the osteogenic

122 phenotype in ADI-MSCs but does not elevate the osteogenic capacity of these cells to ‘bone marrow MSC’ status. The requirements for ADI-MSC chondrogenic differentiation are somewhat analogous, in that adipose-derived stem cell chondrogenesis is significantly increased by adding BMP-6 to the chondrogenic medium [40]. Multi-lineage phenotypic potential is a requisite characteristic of mesenchymal stem cells. Although the equine adipose-derived cells used in this study qualitatively expressed some characteristics of the osteogenic phenotype (induction of ALP and RUNX2 expression), their ability to undergo osteogenesis was consistently far less than that of BM-MSCs, despite BMP-2 supplementation. Although adipose-derived stem cells have been strongly advocated for as an easily acquired and versatile resource for cell-based regenerative and tissue engineering applications [41, 42], their limited capacity for robust phenotypic differentiation will need to be supported by sustained co-delivery of appropriate lineage-specific growth factors.

Conclusions The results of this study support the hypothesis that the differential osteogenic capacity of equine BM-MSCs and ADI-MSCs is a consequence of differences in BMP ligand expression and/or induction. More generally, this outcome supports a regulatory mechanism by which BMP signaling determines mesenchymal stem cell lineage predispositions towards specific musculoskeletal lineages. The findings strongly suggest that BMP signaling acts directly through OSX expression, rather than through the RUNX2-OSX regulatory axis, in osteoprogenitors.

123

Figures & Table

Figure 5.1: Baseline BMP-2 and BMP-4 mRNA Expression in ADI- and BM-MSCs

Figure 5.1: Relative mRNA expression of BMP-2 and BMP-4 in undifferentiated adipose (ADI) and bone marrow (BM) derived MSCs. Asterisks (*) indicate a significant difference between cell types (* p < 0.05). Evaluated with 2-Way ANOVA, n = 4, p < 0.05.

124

Figure 5.2: Osteogenesis-Induced BMP-2 and BMP-4 mRNA Expression in BM-MSCs

Figure 5.2: BMP-2 and BMP-4 induced mRNA expression in BM-MSCs evaluated at Day 7 and Day 14 post-osteogenic induction. “Cx” = control media, and “OIM” = osteogenic media. Dashed lines represent the approximate relative expression in ADI-MSC cultures. Asterisks (*) indicate a significant difference between media types (* p < 0.05). Evaluated with 2-Way ANOVA, n = 4, p < 0.05.

125

Figure 5.3: Effect of BMP Inhibition in BM-MSCs in Osteogenic Media: Representative Culture Images

Figure 5.3: Representative 4X magnification images of Day 3, 7, and 10 post-osteogenic induction multi-cellular aggregation and Alizarin Red staining in BM-MSCs maintained in control (Cx) or osteogenic (OIM) media, +/- recombinant human Noggin (rhNoggin) at 10, 100, or 1000 ng/ml.

126

Figure 5.4: Effect of BMP Inhibition in BM-MSCs in Osteogenic Media: Multi-Cellular Aggregate Counts

Figure 5.4: Effect of BMP inhibition with recombinant human Noggin on multi-cellular aggregate formation at Day 7 and Day 10 post-osteogenic induction in BM-MSCs. “Cx” = control media, and “OIM” = osteogenic media. The number of aggregates per low power field (LPF) was assessed at ten random locations. Asterisks (*) indicate a significant decrease in aggregate counts compared to OIM alone (* p < 0.05). Evaluated with 2-Way ANOVA at each time point, n = 10, p < 0.05.

127

Figure 5.5: Effect of BMP Inhibition in BM-MSCs in Osteogenic Media: Alkaline Phosphatase (ALP) Enzymatic Activity

Figure 5.5: Effect of BMP inhibition with recombinant human Noggin, DMH-1, and K-02288 on alkaline phosphatase (ALP) enzymatic activity in osteogenic BM-MSCs. “Cx” = control media, and “OIM” = osteogenic media. Asterisks (*) indicate a significant decrease in ALP enzymatic activity compared to OIM alone (* p < 0.05). Evaluated with 2-Way ANOVA for each inhibitor, n = 3, p < 0.05.

128

Figure 5.6: Effect of BMP Inhibition in BM-MSCs in Osteogenic Media: Osterix mRNA Expression

Figure 5.6: Effect of BMP inhibition with K-02288 on Osterix (OSX) mRNA expression at Day 3 and Day 7 post-osteogenic induction in BM-MSCs. “Cx” = control media, and “OIM” = osteogenic media. Asterisks (*) indicate a significant decrease in OSX expression compared to OIM alone (* p < 0.05). Evaluated with 2-Way ANOVA at each time point, n = 3, p < 0.05.

129

Figure 5.7: BMP-2 Supplementation in ADI-MSCs: Representative Culture Images

Cx OIM OIM + BMP-2

Day 5 Alizarin Red Stain 4X Magnification

Day 10 Alizarin Red Stain 4X Magnification

Figure 5.7: Representative 4X magnification images of Day 5 and Day 10 post-osteogenic induction multi-cellular aggregation and Alizarin Red staining in ADI-MSCs. “Cx” = control media, “OIM” = osteogenic media, and “BMP-2” = recombinant human bone morphogenetic protein 2 supplemented at 100 ng/ml.

130

Figure 5.8: BMP-2 Supplementation in ADI-MSCs: ALP mRNA Expression

Figure 5.8: ALP mRNA expression in ADI-MSCs evaluated at Days 4, 7, and 10 post-osteogenic induction. “Cx” = control media, “OIM” = osteogenic media, and “BMP-2” = recombinant human bone morphogenetic protein 2 supplemented at 100 ng/ml. Asterisks (*) indicate a significant effect of culture treatment on ALP expression (* p < 0.05). Evaluated with 2-Way ANOVA, n = 4, p < 0.05.

131

Figure 5.9: BMP-2 Supplementation in ADI-MSCs: RUNX2 mRNA Expression

Figure 5.9: RUNX2 mRNA expression in ADI-MSCs evaluated at Days 4, 7, and 10 post- osteogenic induction. “Cx” = control media, “OIM” = osteogenic media, and “BMP-2” = recombinant human bone morphogenetic protein 2 supplemented at 100 ng/ml. Evaluated with 2- Way ANOVA, n = 4, p < 0.05.

132

Figure 5.10: BMP-2 Supplementation in ADI-MSCs: OSX mRNA Expression

**** ^ ^^

Figure 5.10: OSX mRNA expression in ADI-MSCs evaluated at Days 4, 7, and 10 post- osteogenic induction. “Cx” = control media, “OIM” = osteogenic media, and “BMP-2” = recombinant human bone morphogenetic protein 2 supplemented at 100 ng/ml. Asterisks (*) indicate a significant effect of culture treatment on OSX expression (**** p < 0.0001). Carets (^) indicate a significant post-hoc test differences in OSX expression compared to controls (^ p < 0.05, ^^ p < 0.01). Evaluated with 2-Way ANOVA, n = 4, p < 0.05.

133

Table 5.1: Primers Utilized in the qPCR Reactions

Gene Target Primer Sequence & Start Site Annealing (Product Size) S = Sense, A = Antisense Temp(s) ALP S: 5’ – TGGGGTGAAGGCTAATGAGG – 3’ (463) 56 – 60oC (222 bp) A: 5’ – GGCATCTCGTTGTCCGAGTA – 3’ (684) BMP-2 S: 5’ – TAACCACGCCATTGTTCAGA – 3’ 56 – 60oC (160 bp) A: 5’ – ACAACCCTCCACAACCATGT – 3’ BMP-4 S: 5’ – TGAGCCTTTCCAGCAAGTTT – 3’ 56 – 60oC (162 bp) A: 5’ – TCCTATGGGAGGGACACAAG – 3’ BMP-6 S: 5’ – AGCTGGGAACACTCAGGAGA – 3’ 56 – 60oC (223 bp) A: 5’ – ACTGCAACCAGGGAAAACAC – 3’ BMP-7 S: 5’ – CAGCCTGGCTCGTTTTAGAC – 3’ 56 – 60oC (158 bp) A: 5’ – ACGGAACAGACAGACCCAAC – 3’ EF1α S: 5’ – CCCGGACACAGAGACTTCAT – 3’ (277) 54 – 64oC (329 bp) A: 5’ – AGCATGTTGTCACCATTCCA – 3’ (605) OSX S: 5’ – GGCTATGCCAATGACTACCC – 3’ (337) 56 – 60oC (188 bp) A: 5’ – GGTGAGATGCCTGCATGGA – 3’ (524) RUNX2 S: 5’ – CAGACCAGCAGCACTCCATA – 3’ (1135) 56 – 60oC (178 bp) A: 5’ – CAGCGTCAACACCATCATTC – 3’ (1312)

Table 5.1: Sense (S) and antisense (A) sequences for primers utilized in the qPCR reactions, including optimum annealing temperature ranges as determined by gradient PCR.

134

References 1. Pittenger, M.F., Mackay, A.M., Beck, S.C., et al (1999) Multilineage potential of adult human mesenchymal stem cells. Science. 284: 143-147.

2. Caplan, A.I. (2017) Mesenchymal Stem Cells: Time to change the name! Stem Cells Transl Med. 6(6): 1445-1451.

3. Stewart, M.C., Stewart, A.A. (2011) Mesenchymal Stem Cells: Characteristics, sources, and mechanisms of action. Vet Clin Equine. 27(2): 243-261.

4. Langenbach, F., Handschel, J. (2013) Effects of dexamethasone, ascorbic acid and β- glycerophosphate on the osteogenic differentiation of stem cells in vitro. Stem Cell Res Ther. 4(5): 117.

5. Jaiswal, N., Haynesworth, S.E., Caplan, A.I., et al (1997) Osteogenic differentiation of purified, culture-expanded human mesenchymal stem cells in vitro. J Cell Biochem. 64(2): 295-312.

6. Urist, M.R. (1965) Bone: formation by autoinduction. Science. 150(3698): 893-899.

7. Salazar, V.S., Gamer, L.W., Rosen, V. (2016) BMP signaling in skeletal development, disease and repair. Nat Rev Endocrinol. 12(4): 203-221.

8. Bandyopadhyay, A., Tsuji, K., Cox, K. et al (2006) Genetic analysis of the roles of BMP2, BMP4, and BMP7 in limb patterning and skeletogenesis. PLoS Genet. 2: 2116-2130.

9. Katagiri, T., Boorla, S., Frendo, J. L., et al (1998) Skeletal abnormalities in doubly heterozygous Bmp4 and Bmp7 mice. Dev Genet. 22: 340-348.

10. Tsuji, K., Bandyopadhyay, A., Harfe, B.D., et al (2006) BMP2 activity, although dispensable for bone formation, is required for the initiation of fracture healing. Nature Genet. 38: 1424- 1429.

11. Carreira, A.C., Zambuzzi, W.F., Rossi, M.C., et al (2015) Bone Morphogenetic Proteins: Promising Molecules for Bone Healing, Bioengineering, and Regenerative Medicine. Vitam Horm. 99: 293-322.

12. Bibbo, C., Nelson, J., Ehrlich, D., Rougeux, B. (2015) Bone morphogenetic proteins: indications and uses. Clin Podiatr Med Surg. 32(1): 35-43.

13. Garrison, K.R., Shemilt, I., Donell, S., et al (2010) Bone morphogenetic protein (BMP) for fracture healing in adults. Cochrane Database Syst Rev. 2010(6): CD006950.

135

14. Dmitrieva, R.I., Minullina, I.R., Bilibina, A.A., et al (2012) Bone marrow- and subcutaneous adipose tissue-derived mesenchymal stem cells: differences and similarities. Cell Cycle. 11(2): 377-383.

15. Liao, H.T., Chen, C.T. (2014) Osteogenic potential: Comparison between bone marrow and adipose-derived mesenchymal stem cells. World J Stem Cells. 6(3): 288-295.

16. Webb, T.L., Quimby, J.M., Dow, S.W. (2012) In vitro comparison of feline bone marrow- derived and adipose tissue-derived mesenchymal stem cells. J Feline Med Surg. 14(2): 165- 168.

17. Mohamed-Ahmed, S., Fristad, I., Lie, S.A., et al (2018) Adipose-derived and bone marrow mesenchymal stem cells: a donor-matched comparison. Stem Cell Res Ther. 9(1): 168.

18. Im, G.I., Shin, Y.W., Lee, K.B. (2005) Do adipose tissue-derived mesenchymal stem cells have the same osteogenic and chondrogenic potential as bone marrow-derived cells? Osteoarthritis Cartil. 13(10): 845-853.

19. Hayashi, O., Katsube, Y., Hirose, M., et al (2008) Comparison of osteogenic ability of rat mesenchymal stem cells from bone marrow, periosteum, and adipose tissue. Calcif Tissue Int. 82(3): 238-247.

20. Monaco, E., Bionaz, M., Rodriguez-Zas, S., et al (2012) Transcriptomics comparison between porcine adipose and bone marrow mesenchymal stem cells during in vitro osteogenic and adipogenic differentiation. PLoS One. 7(3): e32481.

21. Vidal, M.A., Kilroy, G.E., Lopez, M.J., et al. (2007) Characterization of equine adipose tissue-derived stromal cells: adipogenic and osteogenic capacity and comparison with bone marrow-derived mesenchymal stromal cells. Vet Surg. 36(7): 613-622.

22. Toupadakis, C.A., Wong, A., Genetos, D.C., et al (2010) Comparison of the osteogenic potential of equine mesenchymal stem cells from bone marrow, adipose tissue, umbilical cord blood, and umbilical cord tissue. Am J Vet Res. 71(10): 1237-1245.

23. Kang, B.J., Ryu, H.H., Park, S.S., et al (2012) Comparing the osteogenic potential of canine mesenchymal stem cells derived from adipose tissues, bone marrow, umbilical cord blood, and Wharton's jelly for treating bone defects. J Vet Sci. 13(3): 299-310.

24. Chung, D.J., Hayashi, K., Toupadakis, C.A., et al (2012) Osteogenic proliferation and differentiation of canine bone marrow and adipose tissue derived mesenchymal stromal cells and the influence of hypoxia. Res Vet Sci. 92(1): 66-75.

136

25. Taylor, S.E., Clegg, P.D. (2011) Collection and Propagation Methods for Mesenchymal Stromal Cells. Vet Clin North Am Equine Pract. 27(2): 263-274.

26. Komori, T., Yagi, H., Nomura, S., et al (1997) Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell. 89(5): 755-764.

27. Otto, F., Thornell, A.P., Crompton, T., et al (1997) Cbfa1, a candidate gene for cleidocranial dysplasia syndrome, is essential for osteoblast differentiation and bone development. Cell. 89(5): 765-771.

28. Ducy, P., Zhang, R., Geoffroy, V., et al (1997) Osf2/Cbfa1: a transcriptional activator of osteoblast differentiation. Cell. 89(5): 747-754.

29. Nakashima, K., Zhou, X., Kunkel, G., et al (2002) The novel zinc finger-containing transcription factor osterix is required for osteoblast differentiation and bone formation. Cell. 108(1): 17-29.

30. Gregory, C.A., Gunn, W.G., Peister, A., et al (2004) An Alizarin red-based assay of mineralization by adherent cells in culture: comparison with cetylpyridinium chloride extraction. Anal Biochem. 329(1): 77-84.

31. Livak, K.J. & Schmittgen, T.D. (2001) Analysis of Relative Gene Expression Data Using Real-Time Quantitative PCR and the 2-ΔΔCT Method. Methods. 25: 402-408.

32. Bessey, O.A., Lowry, O.H., et al (1946) A method for the rapid determination of alkaline phosphates with five cubic millimeters of serum. J Biol Chem. 164: 321-329.

33. Yamaguchi, A., Komori, T., Suda, T. (2000) Regulation of osteoblast differentiation mediated by bone morphogenetic proteins, hedgehogs, and Cbfa1. Endocr Rev. 21(4): 393- 411.

34. Noël, D., Gazit, D., Bouquet, C., et al (2004) Short-term BMP-2 expression is sufficient for in vivo osteochondral differentiation of mesenchymal stem cells. Stem Cells. 22(1): 74-85.

35. Nishimura, R., Hata, K., Matsubara, T., et al (2012) Regulation of bone and cartilage development by network between BMP signaling and transcription factors. J Biochem. 151(3): 247-254.

36. Nishio, Y., Dong, Y., Paris, M., et al (2006) Runx2-mediated regulation of the zinc finger Osterix/Sp7 gene. Gene. 372: 62-70.

137

37. Sinha, K.M, Zhou, X. (2013) Genetic and molecular control of osterix in skeletal formation. J Cell Biochem. 114(5): 975-984.

38. Lee, M.H., Kwon, T.G., Park, H.S., et al (2003) BMP-2-induced Osterix expression is mediated by Dlx5 but is independent of Runx2. Biochem Biophys Res Commun. 309(3): 689- 694.

39. Ulsamer, A., Ortuño, M.J., Ruiz, S., et al (2008) BMP-2 induces Osterix expression through up-regulation of Dlx5 and its phosphorylation by p38. J Biol Chem. 283(7): 3816-3826.

40. Estes, B.T., Wu, A.T., Guilak F. (2006) Potent induction of chondrocytic differentiation of human adipose-derived adult stem cells by bone morphogenetic protein 6. Arthritis Rheum. 54(4): 1222-1232.

41. Zuk, P.A., Zhu, M., Ashjian, P., et al (2002) Human adipose tissue is a source of multipotent stem cells. Mol Biol Cell. 13(12): 4279-4295.

42. Guilak, F., Estes, B.T., Diekman, B.O., et al (2010) Multipotent adult stem cells from adipose tissue for musculoskeletal tissue engineering. Clin Orthop Relat Res. 468(9): 2530- 2540.

138

CHAPTER 6 Conclusions and Future Directions

The first study presented in this dissertation (Chapter 3) determined if donor status at the time of tissue harvest influenced subsequent in vitro viability, proliferation, and/or osteogenesis of adipose-derived (ADI) and bone marrow-derived (BM) MSCs collected from adult horses during standing sedation, general anesthesia, and shortly after euthanasia. Donor status during tissue collection did not impact primary cell isolate counts or cell viability/proliferation as evaluated by population doubling times. Of the osteogenic outcomes measured, only RUNX2 mRNA expression in ADI-MSCs was significantly decreased at the day 5 post-induction time point in euthanasia-collected samples, but the effect did not persist at the later day 10 time point. Multi-cellular aggregation during osteogenesis, Alizarin Red staining, OSX mRNA expression, ALP mRNA expression, and ALP enzymatic activity were not adversely affected by donor status during MSC collection but differed significantly between cell types and donors.

The overall result of this study indicates that MSCs harvested from recently deceased donors can serve as a viable option for allogeneic stem cells. Further research in animal models is necessary to identify any toxic effects of the systemic drugs administered for sedation, general anesthesia, and euthanasia on MSCs, and determine the ideal collection ‘window’ after death. This study also revealed a large degree of inter- and intra-donor variability, thus warranting further investigation into how various in vitro screening methods can be optimized to better characterize the proliferative and multi-lineage differentiation abilities of MSCs from a given donor and tissue source. Donor-specific factors such as age, body condition, and clinical condition(s) likely impacted the initial numbers of viable MSCs isolated and cell proliferative capacity. Therefore, determining exclusion criteria is vital when screening for potential donors.

The second study presented in this dissertation (Chapter 4) first characterized the differences in the RUNX2 and OSX response axis during osteogenesis in ADI- and BM-MSCs, and then determined whether OSX over-expression was sufficient to improve osteogenesis in ADI-MSCs. Resting/baseline levels of OSX, but not RUNX2, were significantly higher in BM- compared to ADI-MSCs. Both cell types upregulated RUNX2 in response to osteogenic cues, but only BM-MSCs induced OSX. BM-MSCs showed significantly higher RUNX2 and OSX

139 induction during osteogenesis than ADI-MSCs. Over-expression of OSX in ADI-MSCs was not sufficient to improve multi-cellular aggregation, Alizarin Red staining, ALP mRNA expression and enzymatic activity, or consistently induce expression of a panel of downstream osteoprogenitor marker genes in either control or osteogenic conditions.

The outcome of these experiments suggests that OSX induction influences the osteogenic potential of MSCs, yet OSX over-expression in less osteogenic MSC sources is not sufficient to improve osteogenesis. Follow-up studies should focus on characterizing epigenetic regulatory events that may impact differences in OSX expression profiles in MSCs from different sources. A comparison of the responses to RUNX2 versus OSX over-expression should also be performed, to identify critical non-OSX targets of RUNX2 activity. However, over-expression of RUNX2 or OSX alone has produced mixed results in various cell line and animal studies, likely because their roles during osteogenesis are precisely timed and regulated. Sustained over- expression of RUNX2 and OSX can inhibit terminal osteoblast differentiation, so future studies should focus on optimizing the timing of inducible RUNX2 and/or OSX expression. There are also several limitations and safety concerns with using adenoviral over-expression in clinical scenarios, thus warranting further investigation into alternative methodologies for enhancing gene expression in MSCs in both in vitro cell culture and in vivo animal models.

The third study presented in this dissertation (Chapter 5) established the importance of BMP signaling in determining the osteogenic potentials of MSCs. BM-MSCs were found to have higher baseline and osteogenesis-induced BMP-2 and -4 ligand expression compared to ADI- MSCs, and BMP ligand secretion appears to be higher in BM-MSCs undergoing osteogenesis. BMP signaling was found to be necessary for osteogenesis in BM-MSCs and sufficient for improving osteogenic differentiation in ADI-MSCs. BMP signaling had a significant effect on OSX, but not RUNX2, expression in both cell types.

This study confirmed that BMP signaling has significant influence in determining the osteogenic potential of MSCs from different tissue sources. The results also explain, in part, why OSX expression levels differed between the two cell types, and why OSX over-expression by itself was not sufficient for improving the osteogenic capacity of ADI-MSCs, as presented in Chapter 4. The BMP signaling pathway directly targets both RUNX2 and OSX, but there is

140 complex interplay from other network components and several additional signaling pathways that also impact osteogenic differentiation. Potential follow-up studies should include microarray analyses of BMP signaling network genes, characterizing epigenetic differences in BMP ligand and/or receptor genes between the two cell types, and quantifying how much of the OSX and RUNX2 expression differences were due to BMP signaling effectors using additional gene expression analyses and knock-down models. Recombinant BMP-2 and -7 proteins are already approved for use in several human orthopedic applications, yet their use is cost-prohibitive and can result in severe complications. Further research is therefore needed to develop controlled and predictable BMP signaling enhancement using gene therapy modalities and/or implantation of osteoinductive biomaterials.

Globally, these studies confirmed that equine BM-MSCs have superior in vitro osteogenic capacity than ADI-MSCs, and identified some of the factors that play crucial roles in determining this difference. Adipose tissue is much more abundant and easily-accessible compared to bone marrow, thus making ADI-MSCs an appealing alternative to BM-MSCs for use in regenerative medicine. However, ADI-MSCs are not inherently equal to BM-MSCs in their ability to undergo osteogenic differentiation. The research presented in this dissertation investigated potential interventions that could prove useful for enhancing osteogenic differentiation of ADI-MSCs and, by extension, other progenitor populations. The ideal outcome of follow-up studies would be to first improve in vitro MSC screening approaches to better identify more robust and higher responding cell populations, and then optimize strategies for improving ADI-MSC osteogenesis to levels comparable to BM-MSCs. The finding that donor tissue from recently deceased horses yielded viable osteogenic MSCs supports the possibility of using cadaveric samples in research and for allogeneic stem cell banking. In conclusion, there are many sources available to obtain viable MSCs from adult tissues and several promising methodologies that can optimize their use in bone repair and regenerative medicine.

141