Directed Adult Neural Stem/Progenitor Cell Fate in Microsphere-Loaded Chitosan Channels

Howard Kim

A thesis submitted in conformity with the requirements for the degree of Doctorate of Philosophy Institute of Medical Science University of Toronto

Directed Adult Neural Stem/Progenitor Cell Fate in Microsphere- Loaded Channels

Howard Kim

Doctor of Philosophy

Institute of Medical Science University of Toronto

2011 Abstract

Spinal cord injury (SCI) is a devastating condition characterized by the loss of neuronal pathways responsible for coordinating motor and sensory information between the brain and the rest of the body. The mammalian spinal cord is limited in its ability to repair itself, so treatments devised to replace damaged tissue and promote regeneration are essential towards developing a cure. This work describes the development of a guidance channel strategy for spinal cord transection. Chitosan guidance channels were designed as a delivery vehicle for neural stem/progenitor cell (NSPC) transplants and drug-eluting poly(lactic-co-glycolic acid) (PLGA) microspheres. PLGA microspheres were embedded into chitosan channels by a spin-coating method. These microsphere-loaded channels demonstrated the ability for controlled short-term bioactive release of the small molecule drug dibutyryl cyclic-AMP (dbcAMP) and long-term bioactive release of the protein alkaline phosphatase. NSPCs were shown to be responsive to dbcAMP delivery, which results in greatly enhanced differentiation into neurons. The effect of directed neuronal differentiation was investigated after spinal cord transection in rat, resulting in a dramatic increase in NSPC transplant survival. Guidance channels containing NSPCs treated with dbcAMP resulted in robust tissue bridge formation after SCI, demonstrating extensive axonal regeneration and promoting functional recovery. ii

Acknowledgments

My motivation for entering the field of spinal cord research begins first and foremost from my younger brother Jonathan, who suffered a spinal cord injury during high school. His perseverance and independence has been a great inspiration for me throughout my degree, and has served as a constant reminder of the importance of translational research.

I would also like to deeply thank my supervisors Dr. Molly Shoichet and Dr. Charles Tator, who both have been instrumental in my growth as a scientist, and without whose guidance this work would not be possible. Thank you also to lab members, past and present, of the Shoichet and Tator labs, who are too numerous to name individually. I would also like to express gratitude to my committee member Dr. Cindi Morshead, who provided valuable insights and direction for this work.

Finally, I would also like to acknowledge my friends and family, particularly my parents Sue and Vince Kim, who all have been so supportive of me during my graduate career.

Many thanks, Howard Kim

iii

Table of Contents

Acknowledgments ...... iii

Table of Contents ...... iv

List of Tables ...... viii

List of Figures ...... ix

List of Appendices ...... xv

1 Background and Introduction ...... 1

1.1 Rationale ...... 1

1.2 Hypothesis and Objectives ...... 2

1.3 Anatomy of the Spinal Cord ...... 3

1.4 Pathology of Spinal Cord Injury ...... 5

1.5 Animal Models of Spinal Cord Injury ...... 7

1.6 Treatment of Spinal Cord Injury ...... 8

1.7 Cell-Based Therapies for SCI ...... 9

1.7.1 Overview ...... 9

1.7.2 Neural Stem Cells ...... 10

1.7.3 Directed Differentiation of NSPCs ...... 12

1.8 Biomaterials for SCI ...... 13

1.8.1 Overview ...... 13

1.8.2 Biomaterial Use in SCI ...... 14

1.8.3 Chitosan ...... 17

1.8.4 Fibrin ...... 19

1.8.5 Poly(lactic-co-glycolic acid) ...... 20

1.9 Drug Therapies for SCI ...... 21

1.9.1 Overview ...... 21

1.9.2 Dibutyryl Cyclic-AMP ...... 23

1.9.3 Drug Delivery Systems for the Spinal Cord ...... 25

1.10 The Entubulation Strategy ...... 26

2 Evaluation of Chitosan in the Spinal Cord ...... 29

2.1 Abstract ...... 29

2.2 Introduction ...... 29

2.3 Materials and Methods ...... 31

2.3.1 Material Processing ...... 31

2.3.2 In vivo Implantation ...... 32

2.3.3 Tissue Preparation ...... 33

2.3.4 Explant Analysis ...... 33

2.3.5 Staining of Paraffin-Embedded Tissue ...... 34

2.3.6 Statistics ...... 34

2.4 Results ...... 34

2.4.1 Intrathecal Implantation ...... 34

2.4.2 Intramedullary Implantation ...... 38

2.5 Discussion ...... 41

3 Incorporating a Drug Delivery System into Chitosan Guidance Channels...... 45

3.1 Abstract ...... 45

3.2 Introduction ...... 45

3.3 Materials and Methods ...... 47

3.3.1 Materials ...... 47

3.3.2 Microsphere Preparation ...... 48

3.3.3 Chitosan Guidance Channel Preparation ...... 48

3.3.4 Microsphere-loaded Channels ...... 49

3.3.5 In Vitro Protein Release ...... 50 v

3.4 Results and Discussion ...... 50

3.5 Conclusion ...... 55

4 Evaluating Directed Differentiation of NSPCs into Neurons with Dibutyryl Cyclic-AMP .... 56

4.1 Abstract ...... 56

4.2 Introduction ...... 57

4.3 Materials and Methods ...... 58

4.3.1 Chitosan Films and Channels ...... 58

4.3.2 Dibutyryl Cyclic-AMP PLGA Microspheres ...... 59

4.3.3 Adult Neural Stem/Progenitor Cells ...... 60

4.3.4 Chitosan Film Studies ...... 61

4.3.5 Chitosan Channel Studies ...... 61

4.3.6 Immunocytochemistry ...... 62

4.3.7 Quantitative RT-PCR ...... 62

4.3.8 Animal Studies ...... 63

4.3.9 Tissue Preparation ...... 64

4.3.10 Tissue analysis ...... 64

4.3.11 Statistics ...... 65

4.4 Results ...... 65

4.4.1 Dibutyryl Cyclic-AMP Promotes Neuronal Differentiation of NSPCs ...... 65

4.4.2 PLGA Microspheres for dbcAMP Release ...... 67

4.4.3 NSPC Differentiation in Microsphere-Loaded Channels ...... 68

4.4.4 In vivo NSPC Transplantation ...... 68

4.5 Discussion ...... 75

4.6 Conclusion ...... 79

5 Discussion ...... 80

5.1 Chitosan Guidance Channels for SCI ...... 80 vi

5.2 Drug Delivery in PLGA Microspheres and Microsphere-Loaded Channels ...... 81

5.3 Fibrin Scaffolds for Cell Delivery ...... 84

5.4 Dibutyryl Cyclic-AMP and NSPCs ...... 87

5.5 NSPC Transplant Optimization ...... 89

5.6 Translational Considerations ...... 91

5.6.1 Drug Delivery ...... 91

5.6.2 Stem Cell Source ...... 92

5.6.3 Injury Model ...... 93

5.7 The Entubulation Strategy for SCI ...... 94

6 Conclusions ...... 97

6.1 Achievement of Objectives ...... 97

6.2 Major Contributions ...... 98

7 Recommendations for Future Work ...... 100

7.1 Further Investigation dbcAMP-Mediated Survival ...... 100

7.2 Utilizing the Versatility of the Drug Delivery System ...... 102

7.3 Directed Differentiation of Other Cell Types ...... 102

7.4 Application in Other SCI Models ...... 103

References ...... 104

Appendix A: Abbreviations ...... 128

Appendix B: Additional Figures ...... 129

vii

List of Tables

Table 1.1: Common biomaterials used in SCI research (Adapted from [39]) ...... 15

Table 4.1: In vivo treatment groups ...... 63

viii

List of Figures

Figure 1.1: Gross anatomy of a spinal cord segment in cross section (Copyright Michael Corrin)...... 3

Figure 1.2: Detrimental pathological mechanisms resulting from initial trauma to the spinal cord...... 6

Figure 1.3: Experimental models of spinal cord injury (Copyright Michael Corrin)...... 7

Figure 1.4 Neural stem cells have the ability to self-renew and proliferate as neurospheres, which can be expanded through passaging (dissociation and reformation of neurospheres). Neural stem cell can also give rise to the three main cell types of the CNS: astrocytes, oligodendrocytes, and neurons...... 11

Figure 1.5: The structures of chitin and chitosan. The material is generally considered chitosan when the ratio of amine to acetylamide groups (ie. the degree of deacetylation, DD) is greater than 0.6. The DD affects many biological properties of chitosan including cell-adhesiveness and degradation...... 18

Figure 1.6: Reaction scheme of thrombin-mediated polymerization of fibrinogen into fibrin. ... 19

Figure 1.7: Chemical structure of PLGA. The ratio of lactic acid to glycolic acid monomers (x:y) affect many physical properties of PLGA, most notably the degradation kinetics. PLGA degrades by hydrolysis and its degradation products, lactic acid and glycolic acid, are naturally occurring metabolites...... 21

Figure 1.8: Structures of cyclic-AMP and dibutyryl cyclic-AMP. The dibutyryl modification gives the molecule greater stability and enhances cell membrane permeability...... 23

Figure 1.9: Pathways of intracellular cAMP associated with promoting neuronal differentiation. Increases in cyclic-AMP result in activation of protein kinase A (PKA) which in turn phosphorylates the transcription factor CREB. CREB initiates gene expression of proteins related to neuronal differentiation and maturation...... 24

Figure 2.1: A) Intrathecal implantion where sheets of chitosan, Gore-Tex, or Vicryl were placed on the dorsal surface of the spinal cord on either side of the durotomy. B) Intramedullary implantations are performed by longitudinal durotomy and myelotomy, followed by placement of the material into the spinal cord parenchyma...... 32

Figure 2.2: Measurement of weight change of implants removed after one and six months. Vicryl is the only material that exhibited degradation as assessed by weight loss. (n = 4, mean ± standard deviation, ***p<0.001)...... 35

Figure 2.3: Scanning electron micrographs of intrathecal Chitosan (DD=85%), Gore-Tex, and Vicryl sheets prior to implant and one and six months after implant. Chitosan and Gore-Tex show no significant signs of degradation. There was no Vicryl sheet remaining at six months. All images are shown at the same magnification...... 35

Figure 2.4: Representative images of intrathecal implants. LFB/H&E staining of chitosan, Vicryl, and Gore-Tex implants at (A-C) one month and (D-F) six months. Fibrous capsules formed around all three materials and persist up to six months. Chitosan stains mainly with eosin (pink) at one month and hematoxylin (blue) at six months. At six months, Vicryl is degraded whereas chitosan and Gore-Tex implants show no discernable evidence of mass loss. Symbols: SC-spinal cord, FC-fibrous capsule, -implant, D-dura...... 36

Figure 2.5: Comparison of fibrous capsule thickness between chitosan, Vicryl, and Gore-Tex implants at one and six months. Vicryl elicited a stronger fibroblastic response versus chitosan at one month and Gore-Tex at one and six months. No significant differences were seen between chitosan and Gore-Tex. (n = 4, mean ± standard deviation, *p<0.05, **p<0.01)...... 37

Figure 2.6: Representative images of ED1 staining at one month. Very few activated macrophages (arrowheads) interacting with chitosan or Gore-Tex, but high activation against Vicryl. It should be noted that chitosan becomes brittle upon fixation and sectioning, resulting in a fractured appearance. Symbols: SC-spinal cord, FC-fibrous capsule, -implant, D-dura...... 37

Figure 2.7: Representative GFAP staining of intrathecal implants at six months. GFAP reactivity (arrowheads) on the dorsal surface of the cord is seen adjacent to each implant. Symbols: SC- spinal cord, FC-fibrous capsule, -implant...... 38

Figure 2.8: Representative images of intramedullary implants with H&E/LFB and Masson’s Trichrome staining. (A-D) One month, (E-H) six month, and (I-L) twelve months images with H&E/LFB show fibroblast infiltration and fibrous encapsulation of implants. No signs of direct cellular engagement are observed with chitosan or Gore-Tex, whereas foreign body giant cells are seen engulfing portions of Vicryl (C,G arrowheads). Both 78% DD and 85% DD chitosan exhibit greater hematoxylin staining (blue) over time. (M-P) Masson’s staining of the implant sites at one twelve months show fibrous capsules to consist mainly of collagen (green). It should be noted that chitosan becomes brittle upon fixation, resulting in its fractured appearance upon sectioning. Symbols: FC-fibrous capsule, -implant...... 39

Figure 2.9: Comparison of fibrous capsule thickness between chitosan, DD=78%, chitosan, DD=85%, and Gore-Tex after intramedullary implantation. Chitosan (78%) had a significantly thicker fibrous capsule than either chitosan (85%) or Gore-Tex, which were not significantly different from each other. Data is pooled from one, six, and twelve month timepoints. (n = 6, mean ± standard deviation, *p<0.05)...... 40

Figure 2.10: Representative images of ED1 staining of intramedullary implants at (A-D) one month and (E-H) six months. Macrophage activity is elevated at one month in the spinal cord parenchyma but subsides by six months. Direct macrophage interaction with the material is only observed in the case of Vicryl (C, arrowhead). All images were taken at the same magnification. Symbols: FC-fibrous capsule, -implant...... 40

Figure 2.11: Representative images of GFAP staining of intramedullary implants at one month. Reactive astrocytes (arrowheads) were located in the spinal cord parenchyma, separated from the implanted materials by the fibrous capsule. All images were taken at the same magnification. Symbols: SC-spinal cord, FC-fibrous capsule, -implant...... 41

Figure 3.1: A) Chitosan channels are prepared by adding acetic anhydride to chitosan then injecting the mixture into a glass mold to prepare chitin channels. The channels are converted back into chitosan by high-temperature alkaline hydrolysis. B) Microsphere-loaded channels are prepared by spin-coating the interior of a chitosan channel with a chitosan solution containing microspheres...... 49

Figure 3.2: A) PLGA microspheres visualized using scanning electron microscopy have smooth exterior surfaces with sporadic pores. B) The size distribution of PLGA particles as measured by light scattering. The average volume weighted diameter of the particles was 26.5 µm as measured by static light scattering...... 51

Figure 3.3: A,B) Light micrographs of the microsphere-loaded chitosan channels. The thickness of the channels when hydrated is approximately 200 µm, of which the secondary chitosan layer (indicated by arrows) contributes about 20 µm. C) Scanning electron microscopy shows microspheres (arrowheads) embedded by the secondary chitosan coating...... 53

Figure 3.4: Cumulative release profile of (A) total and (B) bioactive alkaline phosphatase from free floating PLGA microspheres (FreeMS), microsphere-loaded channels (MLC), and dispersed directly into the channel walls (FreeAP) (n=3 for all groups). The release profile of MLCs closely mirrored that of free microspheres. Alkaline phosphatase was released in bioactive form throughout the study, with 87.6 ± 0.4% (avg ± s.d.) of released protein remained active from free microspheres compared to 78.7 ± 1.1% from MLCs...... 54

Figure 4.1: A) Dose response curve of dbcAMP on neuronal differentiation of NSPCs after 7 days in culture. B) Differentiation profile of NSPCs after 7 days. NSPCs were treated with media containing 1mM dbcAMP for 0, 1 or 7 days. Only sustained exposure to dbcAMP resulted in increased number of BetaIII-positive neurons. C-J) Representative images of NSPCs after 7 days in culture for markers of progenitors cells (nestin), neurons (BetaIII), oligodendrocytes (RIP), and astrocytes (GFAP). Scale bar represents 100 μm. K) Cell numbers at 1, 3, and 7 days in culture with or without 1 mM dbcAMP. L) Ki67 staining for proliferating cells after 3 days. M) Cell differentiation over time with or without 1 mM dbcAMP treatment. Data represented as mean ± standard (n=3 to 9)...... 66

Figure 4.2 (preceding page): A) Cumulative release profiles of dbcAMP from microspheres and microsphere-loaded channels. The process of embedding microspheres into channel walls is likely responsible for early degradation of PLGA and faster drug release from channels. B) Schematic of the entubulation strategy. NSPCs are seeded on fibrin scaffold within a chitosan channel. Drug-loaded PLGA microspheres release dibutyryl cyclic-AMP causes NSPCs to preferentially differentiate into neurons. C) Viability of NSPCs in a three-dimensional fibrin scaffold. Simultaneous staining of CalceinAM (green) and Ethidium homodimer (red) for live xii and dead cells respectively show good cell viability of NSPCs in fibrin scaffolds at 1 week. Scale bar represents 100 μm. D-G) Immunostaining of NSPCs for DAPI-nuclear stain and betaIII-tubulin with various dbcAMP treatments. Scale bar represents 100 μm. H) Quantification of betaIII-tubulin immunostained NSPCs with various dbcAMP treatments. I,J) Quantitative RT- PCR data for (I) betaIII tubulin and (J) nestin mRNA expression with various dbcAMP treatments, normalized to housekeeping gene HPRT. Data represented as mean ± standard (n=3 to 6)...... 70

Figure 4.3: A) Photograph of the surgical implantation of fibrin-filled chitosan channels. B) Tissue bridges obtained from animals 2 weeks after implantation. C,D) Longitudinal section of tissue bridge demonstrating NSPC survival after 6 weeks in an animal receiving dbcAMP pre- treatment (dbcAMP, 4div). Boxed area in (C) is magnified in (D). E) NSPC survival after 2 and 6 weeks for various treatment groups. F) Assessment of functional recovery using the BBB locomotor scale. After 6 weeks, rats receiving transplants of dbcAMP-pre-treated NSPCs show a statistically significant increase in hindlimb function relative to untreated animals (*, p<0.05). Mean ± standard deviation shown for n=4 to 6...... 71

Figure 4.4 (preceding page): A-L) Representative images of tissue samples demonstrating NSPC differentiation profile of (A-C) nestin-positive progenitor cells, (D-F) BetaIII-positive neurons , (G-H) CC1-positive oligodendrocytes, and (J-L) GFAP-positive astrocytes. Scale bar represents 50 μm. M) Quantification of NSPC differentiation profile for the various treatment groups. Mean ± standard deviation are plotted, n= 3 to 5; significant differences noted with an asterisks, p<0.05. N) Deconvoluted confocal image of betaIII-positive NSPC-derived neurons (arrows) 6 weeks post-transplantation. Scale bar represents 50 μm...... 73

Figure 4.5: A) Representative image of endogenous axonal regeneration into the tissue bridge based on betaIII tubulin staining. B) Evidence of association between betaIII-positive endogenous axons with surviving GFP-positive NSPCs at six weeks. Synaptophysin staining is observed at the interface (inset). C) RECA1 staining for endothelial cells show blood vessel formation throughout the tissue bridge at 2 weeks. D) Prolyl-4-hydroxylase (rPH) staining of bridge tissue indicates that the majority of cells are collagen producing fibroblasts...... 74

Figure B.1: In vitro degradation of different preparation of chitosan in 1mg/ml lysozyme solution at 37oC...... 129 xiii

Figure B.2: Different designs of microsphere-loaded channels. A) Sandwich design published by Goraltchouk et al. (J Controlled Release, 2006) where microspheres are held between an inner chitosan channel and outer chitin channel. B) Microspheres are embedded within chitin channels, which are then dried around an inner chitosan channel. C) Microspheres are cast into chitosan films, which can be cut and placed inside chitosan channels. All of these designs were not suitable due to issues of thickness (A) or loss in drug bioactivity (B,C)...... 130

Figure B.3: A) Release of dbcAMP from PLGA microspheres. B) Neuronal differentiation of NSPCs after one week shows that dbcAMP released from microspheres as effective as control dbcAMP media at creating neurons. C-E) Immunofluorescent images of betaIII-positive staining of NSPCs after one week...... 131

Figure B.4: In vitro measurement of cell viability as assessed by double stranded DNA content. Fibrin supports the proliferation and survival of NSPCs over one week in culture...... 132

Figure B.5: A-H) Immunofluorescence images comparing NSPCs treated with 1mM dbcAMP for one week with initial plating in proliferation media (A-D; ‘primed’) or differentiation media (E-H; ‘unprimed’). I) The ‘unprimed’ protocol results in diminished effect of dbcAMP on neuronal differentiation compared to the ‘primed’ protocol. Data represented as mean ± standard deviation, n=6 to 9...... 133

Figure B.6: Longitudinal sections showing endogenous axonal regeneration (betaIII-tubulin, red) from the rostral spinal cord stump through the bridge and aborting at the caudal bridge/spinal cord interface. This pattern was consistent in all treatment groups...... 134

xiv

List of Appendices

Appendix A: Abbreviations ...... 128

Appendix B: Additional Figures ...... 129

xv 1

1 Background and Introduction 1.1 Rationale

The spinal cord is responsible for communicating and coordinating motor and sensory information between the brain and the rest of the body. Traumatic spinal cord injury (SCI) occurs when the spinal cord is physically crushed, impinged, severed, or otherwise damaged, resulting in irreversible disruption to the neuronal pathways responsible for motor and sensory function. Approximately 1.2 million North Americans suffer from paralysis caused by SCI [1]. SCI not only imposes major quality of life issues on the individual, but also results in considerable financial burden, estimated at nearly $2 million in lifetime costs for a young adult with low tetraplegia [2]. No clinical cure for SCI currently exists and most trials in progress or on the horizon focus mainly on neuroprotection to minimize the extent of damage upon initial injury or remyelination of spared tissue. Ultimately, treatment strategies that emphasize regeneration and/or cell replacement are necessary to achieve a true cure for SCI.

In contrast to neuroprotection strategies, repair and regeneration strategies are aimed towards growing or replacing the damaged area with functional tissue. It was historically thought that CNS neurons had little to no inherent capacity to regrow after injury in the adult. However, a landmark study done by David and Aguayo in 1981 showed that CNS neurons could, if given the right environment, extend axons across an injury site [3]. It has since been verified that the adult CNS is intrinsically inhibitory, containing many signals that repress axonal growth. Regeneration of the spinal cord will require manipulation of the local environment, whether by drugs, biomaterial implants, or cellular transplants, to overcome the lack of intrinsic recovery and inhibitory environment after SCI.

An ongoing collaboration between the laboratories of Dr. Molly Shoichet and Dr. Charles Tator focuses on the use of polymeric guidance channels to study mechanisms of regeneration in the spinal cord. This ‘entubulation strategy’ initially investigated synthetic, non-degradable polymeric channels to bridge the rat spinal cord after complete transection [4]. It was observed that these channels promote tissue bridge formation across the previously separated stumps, whereas untreated transection injuries only develop fibrous tissue connections. While the presence of new tissue was encouraging, the channel material caused undesirable side effects

2 such as calcification [4] in many cases. Recent work with channels made from the natural biopolymer chitosan has shown much more promise, both in its capacity to promote bridge formation and axonal regeneration, but also as a cell-delivery vehicle. Chitosan channels seeded with adult brain-derived neural stem/progenitor cells (NSPCs) resulted in high survival of transplanted cells that contributed to the newly formed tissue [5]. Notably, in this study NSPCs rarely differentiated into neurons.

The next evolution in the entubulation strategy was the addition of drug delivery capabilities. Local and sustained release of factors to the spinal cord can be used to mitigate effects of inflammation, digest scar tissue, neutralize the inhibitory environment, or promote endogenous axonal outgrowth. Drug delivery can also be used to influence the fate of transplanted cells. The overarching goal of this project was to develop a method for incorporating local and sustained release of soluble factors within the guidance channels, and to explore the impact of directed differentiation on NSPCs using this system.

1.2 Hypothesis and Objectives

The stated hypothesis governing this body of work is:

Local sustained release of a neuronal-differentiating factor from chitosan channels to transplanted neural stem/progenitor cells will lead to improved cell differentiation, integration, and tissue repair in a transected spinal cord injury model.

To test this hypothesis, the following objectives were set:

1. Assess the long-term biocompatibility of chitosan channels. Characterize the acute and chronic host immune/inflammation response to different chitosan formulations in the non-injured and injured spinal cord. Evaluate degradation of chitosan formulations in the non-injured and injured spinal cord

2. Design a drug delivery vehicle capable of sustained release within chitosan channels. Incorporate biodegradable polymer microspheres into chitosan channels.

Demonstrate sustained delivery of a bioactive factor from the drug-delivery channels.

3. Evaluate the effects of directed neuronal differentiation caused by the bioactive factor on transplanted neural/stem progenitor cell fate. Identify a suitable drug candidate for promoting NSPC differentiation. Identify a suitable biomaterial scaffold to improve cell seeding encapsulation and distribution within channels. Test the efficacy of drug delivery channels to direct NSPC fate in vitro. Test the efficacy of drug delivery channels to direct NSPC fate in vivo.

1.3 Anatomy of the Spinal Cord

The central nervous system (CNS), consisting of the brain and spinal cord, serves as the main processing and control center for nerve signals. The main role of the spinal cord is as a conduit for communication between the brain and the peripheral nervous system (PNS), but the spinal cord is also very important in many brain-independent pathways including activating the motor reflex after a pain stimulus, controlling walking and standing movements, and regulating the muscles of several internal organs.

Figure 1.1: Gross anatomy of a spinal cord segment in cross section (Copyright Michael Corrin).

The spinal cord runs from the brainstem down the center of the spine terminating in the higher lumbar area. The spinal cord is divided into segments based on where the spinal roots interface the CNS and PNS, and loosely corresponds to the levels of the vertebral column. The gross

4 anatomy of a spinal cord segment is illustrated in Figure 1.1. The spinal cord parenchyma is divided into white matter and gray matter. When viewed in cross-section, the grey matter forms a butterfly-shape in the interior and contains cell bodies of interneurons responsible for local signal processing, and motor neurons that project out to the spinal roots. The white matter consists mainly of rostral-caudal oriented long-tract axons going to and from the brain. The white color comes from myelin. The spinal cord communicates with the PNS through the spinal roots that extend at each segment. The ventral roots consist of central motor axon projections to the periphery, while dorsal roots consist of peripheral axon projections to the spinal cord. The dorsal root ganglia, which contain the cell bodies of the sensory axons of the dorsal roots, are located where the two roots converge.

On a cellular level, the spinal cord is composed mainly of neurons, oligodendrocytes, and astrocytes. Neurons are the core component of the CNS and their primary role is to process and transmit electrical signals. They are composed of a cell body, dendrites that receive incoming signals, and an axon that projects to signaling targets. Oligodendrocytes are the myelinating cell of the CNS. Myelin sheaths insulate axons and aid in the efficient signal conduction in neuronal signaling. A single oligodendrocyte can myelinate several axonal targets. Astrocytes are the main supporting cell type important for nutrient exchange and sealing the blood-brain barrier. After injury, astrocytes are responsible for sealing off the injury site by forming the glial scar. Other cell types present in the spinal cord include microglia, the resident immune cell, as well as cells associated with the vasculature including endothelial cells.

The extracellular matrix of the spinal cord consists mainly of glycoaminoglycans (GAGs), linear polysaccharide chains made from repeating disaccharide units. GAGs vary by composition of sugar residues, linkage chemistry, and degree of sulfation, but all are hydrophilic and highly charged. The most prominent GAGs in the CNS are hyaluronan, and also include chondroitin sulfate, dermatan sulfate, heparin, heparin sulfate, and keratin sulfate. With the exception of hyaluronan, GAGs are bound to a core protein, and together are termed proteoglycans. Proteoglycans in the spinal cord include aggrecan, agrin, brevican, reelin, decorin, neurocan, phosphacan, versican, and tenascin. These molecules vary in size depending on the number of GAGs attached to the core protein. For instance, decorin only contains one GAG side chain whereas aggrecan contains about 130 GAG chains. Fibrous ECM proteins, such as collagen, are largely absent from the spinal cord, apart from collagen IV which is associated with the vascular

5 system. Basement membrane proteins such as laminin and fibronectin are also present in limited quantities in CNS tissue.

The spinal cord is surrounded by three membranes; the pia, arachnoid, and the dura. The pia mater is a thin, highly vascular membrane surrounding the white matter of the spinal cord. Cerebrospinal fluid (CSF) flows between the pia and the dense connective tissue membranes of the arachnoid and dura mater. CSF is an acellular solution that is involved in nutrient exchange and provides physical protection to the spinal cord as an insulating cushion. The spinal cord is encased within a bony vertebral column for protection.

1.4 Pathology of Spinal Cord Injury

Spinal cord injury is heterogeneous in cause and severity. Clinically, the most common type of injury is contusion or compression of the spinal cord following a fracture-dislocation of the vertebral column. This primary damage results in significant mechanical nerve and tissue damage, typically of the central grey matter, and also leads to local swelling, hemorrhaging, and blood vessel constriction. Next follows the secondary phase of injury which arises due to several mechanisms including further swelling and blood flow constriction, release of high concentrations of glutamate and other excitotoxic molecules, lipid peroxidation, free radical production, and an inflammatory response. This secondary phase is responsible for much of the spinal cord degeneration and associated loss of function. Wallerian degeneration, or ‘dieback’, also occurs during this time and is characterized by retraction of the swollen proximal end of damaged axons [6]. Eventually, a large fluid-filled cavity or cyst is formed in the center of the cord, surrounded by a rim of white-matter axons. Many of these spared axons are compromised, as the myelin sheaths are stripped or otherwise disorganized, disrupting efficient signaling. Surrounding the cyst is a glial scar, a bed of hypertrophic reactive astrocytes which act as both a physical and chemical barrier to regeneration. These processes are summarized in Figure 1.2.

Figure 1.2: Detrimental pathological mechanisms resulting from initial trauma to the spinal cord.

The other relevant injury type observed clinically is the transection, where the spinal cord is partially or completely severed. These injuries comprise a relatively small percentage of SCI cases and typically originate from laceration wounds. Much of the pathology regarding secondary injury mechanisms in contusion/compression injuries applies, including axonal dieback and glial scarring. However, due to the lack of an internal cavity, the lesion epicenter is more accessible. In cases of full transection injuries, no spared axonal pathways remain.

Spinal cord injuries are often classified as acute, sub-acute, or chronic depending on the time elapsed from initial injury. These stages mark the pathological features that are characteristic of the host response to injury at different times. The acute injury phase encompasses the primary and initial injury stages and last for roughly one to two weeks in human and rat. The sub-acute phase refers to the days to weeks post-injury where the inflammation response is diminished but prior to significant glial scar formation. An injury is classified as chronic when the body’s responses are stabilized, which usually occurs a few months after initial trauma in humans and around four weeks in rat.

1.5 Animal Models of Spinal Cord Injury

The research effort into spinal cord repair has been quite extensive, with many different approaches being taken to target the various features of SCI. In designing potential treatment strategies, it is important to first consider the injury model. Factors such as animal species and age, injury type, anatomical level, injury severity, and the therapeutic window of treatment can have a great bearing on host response, and can ultimately determine the effectiveness of a given treatment and the interpretation of the results.

Figure 1.3: Experimental models of spinal cord injury (Copyright Michael Corrin).

In rodent models, contusion or compression-type injuries are induced using a calibrated clip- compression or direct impact methods (eg. weight drop). These injuries are characterized by lesion epicenter at the central grey matter which subsequently forms a cavity surrounded by a rim of spared tissue. Contusion/compression injuries are popular in animal models due to their similarity to SCI presentation in humans. These injuries are generated using defined forces which can be tuned, allowing researchers to investigate a range of injury severities. However, these injury models can have relatively high variability, and depending on the treatment, it can be difficult to distinguish between regenerated and spared axons. Nevertheless, compression/contusion injury models are the most widely used models and are particularly important for showing pre-clinical efficacy of a treatment.

The other popular category of animal model is transection or ablation. These injuries are induced by cutting or removing sections of spinal cord. These models are most useful in regeneration studies, where there is less ambiguity between regenerated and spared axons in the lesion area. These models also lend themselves well to anterograde and retrograde tracing studies, where dyes can be placed in the brain or caudal to the injury site, respectively, to track axonal continuity across the lesion. Moreover, studies aimed at investigating specific spinal tracts such as the corticospinal tract (CST) are often simpler to design in these models.

Transection models are often used in biomaterial implantation studies in part due to practical issues of defined lesion volume, size, and access.

The timing of treatment must also be considered. Treatment paradigms generally fall into the categories of acute (time of injury), subacute (7-14 days post-injury) or chronic (28 days or greater post-injury). Acute injuries are most appropriate for neuroprotection strategies as these are aimed at minimizing secondary damage. Lesion expansion and the hostile local environment generally preclude cell or biomaterial-based strategies in the acute model. Sub-acute models are generally preferred for cell transplantation studies when the exictotoxic and inflammatory response have settled, creating a more permissible milieu for transplantation. Chronic injuries present an added challenge of the glial scar but may be the most relevant for treatments aimed at the existing SCI population.

1.6 Treatment of Spinal Cord Injury

The mammalian body has insufficient capability to regenerate neural tissue in the CNS. Current clinical practices for treatment of the acute stages of SCI can include decompression, spinal fusion and fixation, local hypothermia, and intravenous methylprednisolone. These treatments are limited to the prevention or minimization of further injury. There are currently no clinically approved surgical interventions that are aimed at spinal cord regeneration for either new victims of SCI or chronic patients.

There are many therapeutic paradigms in the treatment of SCI. Neuroprotection strategies target the acute stages of injury and aim to minimize the extent of secondary injury. Attenuating the excitotoxic environment, mitigating the inflammation response, and promoting neuronal survival are examples of neuroprotection. As this type of intervention focuses on the acute and sub-acute phases of injury, neuroprotection strategies are limited to new SCI victims and are generally not applicable to the existing SCI population. Nonetheless, neuroprotection strategies will have the most short-term impact on how SCI treatments may be advanced clinically. Completed or in- progress trials include drug treatments such as the anti-inflammatory agents methylprednisolone and minocycline, or neuroprotective agents such as erythropoietin and or ion channel blockers such as Riluzole (sodium) and nimodipine (calcium) (reviewed in [7,8,9]).

Remyelination strategies target spared but dismyelinated axons in the injury penumbra. After injury, the myelin sheath is compromised due to lost or damaged oligodendrocytes, resulting in poor signal propagation in spared axons. Improving the conduction properties of these axons can re-establish dysfunctional signaling pathways, resulting in recovery of function. The Fehlings laboratory at the University of Toronto has shown that replenishing the oligodendrocyte population after SCI through exogenous stem cell transplantation can result in proper re- organization of myelin sheaths in relation to axonal ion channels (ie. Nodes of Ranvier) [10]. Notably, the first North American embryonic stem cell-based therapy for SCI currently in progress is based on a remyelination strategy [11,12]. This therapy is based on the expansion of a purified oligodendrocyte precursor cell (OPC) population derived from human embryonic stem cells, which has been shown to promote remyelination and functional recovery in rat [13]. This phase I trial is investigating safety and potential neurological improvement following transplantation in spinal cord injured patients between one and two weeks post-injury.

Lastly, regeneration strategies aim to promote the formation of new functional connections. The endogenous regenerative response is very limited due to the inhibitive environment of the adult CNS towards axonal growth [14]. Regenerative strategies aim to stimulate endogenous axonal outgrowth or plasticity, neutralize inhibitory signaling, or replace damaged tissue with exogenous cells to network with existing pathways. The following sections will provide a review of the relevant regeneration strategies for SCI treatment, including both pre-clinical research and current or planned clinical trials. These strategies are often centered around cell- based therapies, biomaterial-based therapies, drug delivery therapies, or their combination.

1.7 Cell-Based Therapies for SCI

1.7.1 Overview

Cell transplantation therapy has become one of the favored strategies to induce regeneration after spinal cord injury. Exogenous cells can replace damaged tissue and provide trophic or cell- contact mediated support for neuroprotection and regeneration. Examples of cells that have been transplanted in animal models of SCI include microglia, activated macrophages, olfactory ensheathing glia, Schwann cells, bone marrow stem cells, hematopoietic stem cells,

10 mesenchymal stem cells, umbilical cord blood stem cells, embryonic stem cells, adult neural stem/progenitor cells (NSPCs), and glial restricted precursors (reviewed by [15,16]). In many cases, these attempts have shown glimpses of potential, demonstrating cell survival and integration, accompanied by little or modest gains in function.

Of these different cell types, neural stem/progenitor cells are particularly appealing for their therapeutic potential [17]. Our lab has extensive experience with both spinal cord and brain- derived adult rat NSPCs. We have established protocols for maintaining and expanding these cells in culture, staining and characterizing their differentiation profiles, studying their interactions with different materials, and controlling their cell fate decisions through the use of growth factors [18]. We also have experience with transplanting NSPCs in vivo, in both compression and transection models of spinal cord injury.

1.7.2 Neural Stem Cells

Neural stem cells are multipotent cells capable of self-renewal, allowing for expansion in culture, and given proper environmental cues can differentiate into the three main cell types of the CNS; neurons, astrocytes, and oligodendrocytes (Figure 1.4). In the field, there are some distinctions between ‘true’ neural stem cells and neural progenitor/precursor cells, which have limited self- renewal capabilities. These cells will be collectively referred to as neural stem/progenitor cells (NSPCs). In culture, NSPCs are typically grown as free-floating spheres (neurospheres) in the presence of epidermal growth factor (EGF) and/or fibroblast growth factor-2 (FGF2). NSPCs can be expanded by dissociating the neurospheres into single cells and re-incubating them in growth media, a process termed passaging.

The isolation of adult neural stem cells was first reported in 1992 by Weiss and Reynolds from mouse brain striatum [19]. NSPCs have since been identified as originating from the subventricular zone (SVZ) lining the lateral ventricles [20,21]. Adult NSPCs have also been isolated from dentate gyrus of the hippocampus [22] and from cells adjacent to the central canal of the spinal cord [23]. Alternatively, NSPC can be derived from embryonic stem (ES) cells [24] or induced pluripotent stem (iPS) cells [25]. NSPCs derived from these pluripotent sources tend to be more neurogenic at early passages and more gliogenic with subsequent passages [26], similar to the order in which the CNS is populated during development. Also, for both ES and

11 iPS-derived NSPCs, proper differentiation or sorting protocols are necessary as teratoma formation is a great concern in cell transplantation studies.

The literature surrounding the transplantation of NSPCs in SCI models is quite extensive and highlights the potential of these cells as a therapy for SCI. Tetzlaff et al. recently published an extensive review of pre-clinical transplantation studies in the injured spinal cord. They found that 17 of 20 published articles (up to summer 2008) reported improved behavioural outcome with NSPC transplantation [27], demonstrating the therapeutic potential of these cells. Several mechanisms have been proposed for how NSPCs contribute to SCI treatment. These include reducing cavity size, enhancing plasticity, providing neuroprotection, reducing axonal dieback, and replacing lost cells [28]. Notably, the majority of NSPC transplantation studies report primarily glial differentiation, with neuronal differentiation rates typically less than 1% [27].

Clinical trials of stem cells are currently in progress in humans. ES-derived oligodendrocyte progenitor cells (OPCs) are being investigated in sub-acute SCI in a North American multicenter phase 1/II study. Clinical trials using human fetal brain-derived NSPCs are in progress for the treatment of two separate fatal brain disorders in children, Pelizaeus-Merzbacher Disease and Neuronal Ceroid Lipofuscinosis (ie. Batten disease) (reviewed in [9,28]).

Although there are many similarities between NSPCs derived from different sources (adult, embryonic, fetal, or induced), it would be an oversimplification to assume they are the same. Biologically, these cells can differ in their growth and differentiation properties, as well as their response to environmental cues. From a translational view, each pose their own significant challenges. Adult NSPCs may be obtained from tissue biopsy or post-mortem donor tissue, but these cells are difficult to expand in clinically relevant numbers. Fetal tissue is limited in source and is tied to ethical scrutiny. ES- and iPS cells can be expanded in great numbers, but tumor formation is of great concern. With the exception of possibly iPSCs, immune-suppression of the patient will be a key concern for translation.

1.7.3 Directed Differentiation of NSPCs

Differentiation of NSPCs is largely a response to the local environment, and thus the phenotypic fate of these cells can be manipulated if provided the proper signals. At this time it is not entirely certain how beneficial each individual NSPC progeny will be towards regenerating the spinal cord and restoring function. Indeed, it is even unclear whether obtaining highly purified NSPC phenotypes is desirable, or if an optimal ratio of phenotypes exists. While it has been demonstrated that oligodendrocytes can remyelinate spared axons [13], their utility in a full transection model (where there is no axonal sparing) may be limited. Indeed, one can imagine that the presence of myelin associated inhibitors made by oligodendrocytes may even impede endogenous axonal growth. Likewise, astrocytes can contribute by providing nutrient and growth factor release, but may also contribute to glial scarring. Cao et al have demonstrated that embryonic mouse NSPCs differentiate mainly into astrocytes with no reported benefit to function [29]. Lastly, neuronal differentiation may provide a relay across the injured site, but would still require overcoming significant barriers such as axonal penetration into target tissue, synapse formation at both ends, and training of new pathways. Even if successful, it is hard to predict the potential benefits of random reconnections, which may cause side effects such as pain or

13 uncontrolled spasms. To date, in vivo neuronal differentiation of NSPCs has been difficult to achieve, even with genetic manipulations that promote this phenotype in vitro [30]. A major part of the proposed research is to investigate the role of directed differentiation of NSPCs after SCI, neurons in particular.

1.8 Biomaterials for SCI

1.8.1 Overview

Biomaterials can provide a physical substrate to replace native ECM after tissue injury and are more recently being investigated as cell and drug delivery devices. The term biomaterial can be applied to a diverse set of natural and synthetic materials with a wide range of physical and chemical properties. In general, biomaterials are non-cytotoxic and should not elicit significant local or systemic immune responses, particularly when used in the spinal cord.

Natural biomaterials are derived and purified from biological sources, mammalian or otherwise, and are appealing because they are often found in native mammalian tissue and include collagen, hyaluronan, fibrin, laminin, and fibronectin. Collagen is a fibrillar protein and the main structural component of most connective tissues in the body. Glycosaminoglycans, such as hyaluronan, chondroitin sulfates, and heparin sulfates, are highly hydrated molecules and ubiquitous in the CNS [31]. Other natural biomaterials are derived from plant and non- mammalian sources. Chitosan, a deacetylated form of the polysaccharide chitin (a major component of fungi and arthropod exoskeletons), has been used in the development of guidance channels for spinal cord regeneration [32,33]. The degree of deacetylation of chitosan influences degradation rate, cell adhesion, and amount of neurite extension [34].

Synthetic polymeric hydrogels are also of interest because they can be uniformly produced and their properties easily tuned. Examples of synthetic biomaterials include derivatives of both poly(acrylate) and poly(acrylamide), and notably poly(2-hydroxyethyl methacrylate) (pHEMA) and poly(2-hydroxypropyl methacrylate), p(HPMA), which have been studied as nerve guidance channel materials with controlled structure and wall porosity [35]. Since surgical removal of non-degradable materials is undesirable, several resorbable synthetic polymers have been investigated for nerve guidance materials including polyesters, polycarbonates, and

14 polyurethanes. Most of the synthetic polyesters used in these applications are based on only a few monomer units, used alone or in combination, including poly(glycolide) (PGA), poly(L- lactide) (PLLA), poly(D,L-lactide) (PDLLA), and poly(ε-caprolactone) (PCL). The mechanical and biological properties of these materials can vary widely based on changes in the copolymer composition and fabrication method.

1.8.2 Biomaterial Use in SCI

Biomaterial implants can be used to provide physical support or guidance to regenerating tissue. Important criteria to consider in biomaterial choice may include cost and availability, ease of fabrication, mechanical properties, degradability, and cell-adhesiveness. A wide array of materials, natural and synthetic, has been used in spinal cord applications in the form of hydrogel scaffolds, fibers, and channels. These are summarized in Table 1.1.

The primary role of biomaterials in neural tissue engineering is to provide a physical foundation for tissue regeneration. In spinal cord injury, one of the biggest obstacles for regeneration is the cavity or physical gap created at the lesion site. Scaffolds can serve to bridge these gaps, providing a substrate for regenerating axons to grow. A large number of studies have shown that a diverse set of biomaterials are capable of being effective in this role (reviewed in [36,37,38]). To enhance bridging, the implant can also be designed to include physical guidance cues.

Fibers and micropatterned substrates are some examples that demonstrate axonal growth by contact guidance [61,62,63,64]. Axons from sympathetic neurons were guided along the fibers, where the diameter of the fiber was shown to affect guidance [64]. In another study, when these microfibers were embedded in an agarose gel, axons grew on the fiber until it reached the end of the fiber. The axon then detached from the fiber and continued to grow in the agarose gel [65]. Axonal outgrowth was also influenced by nano-imprinted patterns on polymer substrates such as poly(methyl methacrylate) (PMMA) [62]. These axons preferred to grow on ridge edges and elevations rather than in grooves. Using a fiber-templating technique, physical channels inside a scaffold can be created and may provide an increased surface area for nerve regeneration, as well as guiding and promoting fasciculation of the regenerating cables [66].

Table 1.1: Common biomaterials used in SCI research (Adapted from [39]) Material Description Device Designs Natural Materials Agarose Thermogelling polysaccharide that is non Hydrogel [40] cell-adhesive and non-degradable. Alginate Polysaccharide that undergoes ionic Hydrogel [41] crosslinking in the presence of calcium. Collagen Fibrillar cell-adhesive ECM protein. Hydrogel [42], Fibers [43] Chitosan Polysaccharide derived from chitin. Channel [5], Hydrogel Degradation and cell adhesion properties [44] can be tuned [34]. Fibrin Enzymatically polymerized fibrinogen. Can Hydrogel [45] potentially be derived from autologous sources. Hyaluronic Acid ECM glycosaminoglycan that is injectable Hydrogel [46,47,48] and non cell-adhesive. Matrigel Derived from mouse tumor cells, limited Hydrogel [49] clinical relevance. Highly cell adhesive. Methylcellulose Reverse thermogelling, degradable Hydrogel [46,47] polysaccharide that is non-cell adhesive. Synthetic Materials (degradable) PLA/PGA/PLGA Poly α-hydroxyacids of lactic and glycolic Channel [50], scaffold acids. Degradation occurs via hydrolysis [51], fibers [52] and can be tuned based on composition. PCL A degradable polyester of ε-caprolactone. Channel [53], fibers [54] Degrades by hydrolysis, but more slowly than PLA. PHB Polyhydroxybutyrate is a biocompatible Channel [42], fibers [55] polyester that degrades by surface erosion, but more slowly than PLA. Ampiphillic peptides Can be made to express cell-adhesive Fibers [56,57] peptide sequences, and can self-assemble into nanofibers. Synthetic Materials (non-degradable) PHEMA/PHEMA-MMA Poly(2-hydroxyethyl methacrylate) and co- Channel [4], scaffold polymers are synthetic hydrogels whose [58] mechanical properties can be matched to the spinal cord. PHPMA (NeuroGel) Porous crosslinked hydrogel used in spinal Hydrogel [59] cord and brain applications. PAN/PVC Copolymer of polyacrylonitrile and Channel [60] polyvinylchloride can be used to form stable and non-toxic channels.

Guidance channels for SCI are also based on the principle of physical guidance, albeit on a more macroscopic scale. Popularized by their clinical success in treating peripheral nerve injuries [67], guidance channels have been widely investigated in spinal cord injury repair. The channel provides a substrate for directed growth while potentially providing an isolated local environment that can be manipulated with regenerative cues. Guidance channels can be made from synthetic or naturally occurring polymers, and can be degradable or non-degradable.

Early studies with guidance channels focused on the use of non-degradable materials such as silicone, poly(acrylonitrile-co-vinyl chloride) PAN/PVC, and poly(2-hydroxyethyl methacrylate- co-methyl methacrylate) (PHEMA-MMA) to ensure stability of the channels; however, biodegradable materials such as PLGA and chitosan have become more popular for potential clinical translation. These materials can be tuned to degrade on the order of months to years, allowing for the relatively long time thought to be required for neural regeneration. Several groups have incorporated intricate engineering designs to provide greater surface area for tissue in-growth. This includes matrix-filled channels [49,68], channels within scaffolds [51,69,70], and aligned fiber networks [43]. A more in-depth history of guidance channel work as it pertains to this project is presented in a later section.

Apart from physical guidance, biomaterials can also serve as delivery vehicles for cell transplantation and drug treatments. When NSPCs are injected into an acute spinal cord lesion, poor survival is generally observed [71,72,73]. Moreover, the spinal cord is not inductive towards neurogenesis, so very few NSPCs differentiate into neurons after transplantation. Biomaterial and tissue engineering strategies for cell transplantation aim to provide a more conducive microenvironment for stem cells to survive, migrate, differentiate, and integrate with host tissue. Part of the rationale for using biomaterials-based delivery of cells is to ease the transition from optimal cell culture conditions to the hostile injury lesion site. Anoikis (apoptosis due to lack of cell adhesion molecules) may play a role in the poor survival of transplanted NSPCs after SCI [74]. Indeed, NSPCs perform very poorly when injected as dissociated cells compared to neurospheres [73], where local cell-cell interactions may positively influence survival.

In our own studies, we have shown that adult rat NSPCs show great survival when seeded onto chitosan guidance channels after rat spinal cord transection [5,75]. As early as five weeks post-

17 transplantation, a robust tissue bridge forms between two previously separated stumps. The NSPCs contribute greatly to this newly formed tissue, and also promote endogenous axonal regeneration. Other groups have shown benefits when cells are delivered in conjunction with a biomaterial scaffold. Itosaka et al. found MSCs seeded on a fibrin scaffold survive and migrate significantly further compared to MSCs delivered via direct injection [76]. Teng et al. showed NSC-seeded PLGA scaffolds performed better than either scaffold alone or NSCs alone in promoting functional recovery after rat spinal cord hemisection [77]. NSPC-seeded scaffolds have also shown potential in canine [78] and primate [79] models of SCI as well.

Biomaterials can also serve as an effective means of drug delivery. In many cases, local and sustained delivery of factors is required, and typical methods of drug administration such as repeated bolus injection and intravenous delivery are not ideal in the CNS. Biomaterials can act as reservoirs in a drug delivery system to achieve prolonged drug release to the injury site. These are typically in the form of injectable hydrogels for short-term drug release, or micro- or nano-particles for moderate to long-term drug release.

The biomaterials of primary importance to the present study are introduced in greater detail. These include: chitosan, used to make nerve guidance channels; fibrin, used as a scaffold for NSPCs; and PLGA, a polymer used for microsphere-based drug delivery.

1.8.3 Chitosan

Chitosan is a naturally occurring polysaccharide found in some fungi, but is more commonly derived from the deacetylation of chitin, one of the most abundant biopolymers in nature found most notably as the primary component in insect and crustacean exoskeleton. Chitosan is linear β(1-4)-linked polysaccharide composed of D-glucosamine and N-acetylglucosamine residues (Figure 1.5) with ratio of glucosamine units often referred to as its degree of deacetylation (DD). In general, the material is referred to as chitosan when DD > 60%. The conversion of chitin to chitosan (ie. deacetylation) is typically performed through high-temperature hydrolysis in sodium hydroxide solution, with repeated hydrolysis cycles yielding chitosan product with greater degrees of deacetylation.

Chitosan is a particularly promising material for biomedical applications. Chitosan is generally considered biocompatible and degradable, but these properties can vary greatly depending on

18 molecular weight and DD, as well as implant shape, size, and crystallinity. Clinical use of chitosan is mainly focused on wound dressings, in which chitosan excels due to its hemostatic [80] and anti-fungal and microbial properties [81,82]. This latter property has made chitosan is an important polymer for industrial use in agriculture, horticulture, and water filtration [83].

In nature, natural chitinases and chitosanases exist that specifically catalyze the degradation of these polysaccharides [84]. However, these enzymes are not present in mammals. Lysozyme, while not specific for chitin and chitosan, is the main enzyme responsible for chitosan degradation in mammals. Lysozyme catalyzes the hydrolysis of the β(1-4) linkage, but its ability to bind and act on chitosan is dependent on monomer distribution. Specifically, it has been hypothesized that at least three consecutive N-acetylglucosamine residues are required for lysozomal activity to occur [85]. Indeed, chitosan with DDs greater than 85-90% have shown low rates of degradation in vitro and in vivo [5,34,86].

Tissue engineering applications are typically defined structures such as films [34], particles [87], scaffolds [88], fibers [89], and channels [90] as chitosan is insoluble at neutral pH. Chitosan is however soluble in acidic conditions (the pKa of the primary amine is 6.3) so chitosan is typically prepared and manipulated in acidic solutions, for example in dilute acetic or hydrochloric acid. Common crosslinkers used to create chitosan structures such as particles or scaffolds include glutaraldehyde and genipin [91,92]. Depending on preparation method, strong interchain hydrogen bonding can create stable structures such as films or channels without the use of chemcial cross-linkers [34,90]. Alternatively, physical cross-linking of chitosan at neutral pH can be achieved through polyol salts such as glycerol phosphate [93,94] or long chain fatty acid surfactants [95]. Chemical modification of chitosan has also been investigated to produce

19 water-soluble chitosan derivatives. For example, methacrylamde chitosan is soluble at neutral pH and can be photo-crosslinked in the presence of cells to form hydrogels for cell encapsulation [96]. In experimental spinal cord injury models, chitosan has been shown to reestablish neuronal membrane integrity after damage [97], support axonal regeneration [98] and stem cell transplant survival [99].

1.8.4 Fibrin

Fibrin is the polymerized form of fibrinogen, and is a naturally occurring biopolymer whose primary physiological role is to form clots as the final step of the coagulation cascade. This natural role in hemostasis has led to the clinical use of fibrin as a hemostatic agent and tissue sealant. Fibrin hydrogels are formed by two primary components, fibrinogen and thrombin. Fibrinogen is a 340 kDa plasma protein composed of three pairs of polypeptide chains (Aα, Bβ, and γ). Thrombin initiates the cleavage of fibrinopeptide A and fibrinopeptide B, which normally block the polymerization sites of fibrinogen. With these removed, fibrinogen monomers are able to cross-link with each other to form insoluble fibrin networks. Thrombin further aids in the formation of fibrin by activating the transglutaminase Factor XIIIa, which catalyzes the cross-linking of lysine and glutamine of fibrinogen γ chains. This process is summarized in Figure 1.6. The kinetics of this reaction can be altered by varying concentrations of fibrinogen, thrombin, and calcium [100], which ultimately affects the physical properties of the resultant fibrin scaffold, most notably fiber diameter and porosity [101]. Degradation of fibrin occurs relatively quickly by enzymes present in the bloodstream or secreted by cells, notably plasmin. Degradation of fibrin scaffolds can be slowed by the incorporation of fibrolytic inhibitors such as aprotinin.

Figure 1.6: Reaction scheme of thrombin- mediated polymerization of fibrinogen into fibrin.

Fibrin is highly compliant, degradable, and biocompatible, and thus has been investigated in a wide range of tissue applications including cardiovascular, adipose, muscle, liver, skin, bone,

20 cartilage, and nervous tissue (reviewed in [102]). In nervous tissue, fibrin has shown to be an effective matrix in promoting peripheral nerve regeneration, particularly in relation to guidance channel studies [103,104,105,106,107]. In the CNS, fibrin scaffolds have demonstrated the ability to attenuate the astrocytic glial response [45,108] and promote axonal sprouting after SCI [49,109]. Fibrin has also been utilized as a drug delivery vehicle to localize various neurotrophic factors to the lesion site, often through affinity binding systems. In these systems, the transglutaminase activity of Factor XIIIa is utilized to covalently cross-link heparin binding domain to the γ chains of fibrin gels. Heparin binding domain-functionalized fibrin is then able to sequester proteins for drug delivery including as neurotrophin-3 (NT3), basic fibroblast growth factor (bFGF), platelet-derived growth factor BB (PDGF-BB), nerve growth factor (NGF) and glial-derived neurotrophic factor (GDNF) [110,111,112,113]. Fibrin has also been investigated clinically as an injectable hydrogel for drug delivery to localize the Rho antagonist BA-210 to the spinal cord in the Cethrin clinical trials for acute SCI [114].

Fibrin scaffolds are also useful as cell delivery vehicle for SCI transplants, and have been used to investigate cell transplants of NSPCs [115] and BMSCs [76]. Fibrin scaffolds are advantageous due to their ability to encapsulate cells during gel formation. Fibrin scaffolds have been shown support neuronal differentiation of embryonic stem cells [116] and NSPCs [117], which can be enhanced with growth factor delivery [118,119].

1.8.5 Poly(lactic-co-glycolic acid)

Poly(lactic-co-glycolic acid) (PLGA) is a random copolymer typically prepared commercially by a ring opening polymerization of cyclic dilactide and diglycolide monomers. PLGA is a synthetic polyester that degrades by bulk hydrolysis, yielding lactic acid and glycolic acid as its final degradation products which are naturally occurring metabolites in the body. These chemical structures are shown in Figure 1.7.

The ratio of lactic acid (LA) to glycolic acid (GA) monomers can play a large role in the polymer properties. PLGA degrades faster than the respective pure polymers PGA and PLA, with PLGA50:50 (LA:GA) the fastest degrading copolymer blend. This is the result of differences in crystallinity, as PLGA50:50 is largely amorphous in structure. In general, hydrolysis occurs more readily in amorphous regions compared to crystalline regions, attributed to differences in water uptake [120]. PLGA50:50 typically degrades between weeks to months, where polymer

21 molecular weight, device design (ie. size, surface area, porosity), and local environment (ie. temperature, pH) also affect degradation. PLGA degradation is often self-catalyzed in late stages based on a build-up of local lactic acid and glycolic acid, which lowers pH.

PLGA is considered to be biocompatible, but it elicits moderate inflammatory and immune responses when placed into the body. PLA, PGA, and PLGA are clinically approved in biomedical applications such as sutures, fixation devices, and drug delivery devices (reviewed in [121,122,123]) . In the spinal cord, PLGA has been used extensively as a scaffold material to bridge the lesion site [50,124,125] and as microparticle drug delivery devices for temporal release of bioactive factors [126,127]. Several studies have shown PLGA to be tolerated in CNS tissue [128,129].

1.9 Drug Therapies for SCI

1.9.1 Overview

Delivery of pharmaceutical agents has been one of the most frequently used treatments for studying spinal cord injury. Experimental drugs have been investigated to address various individual detrimental aspects related to spinal cord injury. These pharmaceutical agents include: neuroprotective drugs to attenuate secondary injury mechanisms (eg. anti-inflammatory, anti-apoptotic, etc); neurotrophic factors to stimulate axonal growth;, proteins or antibodies to

22 neutralize endogenous growth inhibition signals; enzymes to breakdown scar tissue; growth factors to promote angiogenesis; stimulation of endogenous stem cells; and molecules to support the survival, proliferation, migration, or differentiation of transplanted cells (reviewed in [130,131]).

Conceptually, one of the simplest paradigms for inducing regeneration is by shifting the environmental balance from growth inhibiting to growth stimulating. Neurotrophic factors such as NGF, brain-derived neurotrophic factor (BDNF), and NT3, among others, are known to have neural-promoting effects including enhanced survival, proliferation, maturation, and differentiation [132]. These growth factors are important during development and are thought to be important during regeneration as well. Other growth promoting molecules target intracellular pathways of growth cone behaviour. For example, the intracellular Rho pathway is involved in actin depolymerization and growth cone collapse. Rho antagonists have been shown to increase axonal growth and functional recovery after SCI in mice [133]. Phase IIA testing of Cethrin (BA210), a Rho antagonist, suggest that it is safe and may improve motor recovery after SCI [9].

Alternatively, neutralizing the many growth inhibitory molecules naturally present in the adult spinal cord has also been shown to greatly increase axonal growth. It is now known that certain cell-surface receptors present on CNS myelin can cause neuronal growth cone collapse. These inhibitory molecules include Nogo-A, oligodendrocyte myelin glycoprotein (OMgp), and myelin associated glycoprotein (MAG). The scar tissue that forms after injury also presents an inhibitory barrier, both physically and chemically. Antibodies and other molecules that block these signals have been developed against these receptors to neutralize their binding ability, resulting in extensive axonal growth both in vitro and in vivo [133]. Clinical trials of these agents include: ATI355, an antibody against Nogo-A for SCI; and GSK249320, an antibody against MAG for stroke [7].

Another application of drug therapy is to enhance the effectiveness of cell transplantation treatments. This is done in an effort to promote cell survival, migration, differentiation, and integration with the host tissue. For example, the Fehlings lab has shown that NSPC supplemented with growth factor cocktail, including FGF2, PDGF-AA, and EGF, results in remyelination whereas poor transplant survival was found when no growth factor was given [10]. Other studies have shown that NSPCs engraft better in chronic models of SCI with the co-

23 delivery of chondroitinase ABC, which degrades the glial scar [134]. Drug delivery can also be effective for promoting stem cell differentiation. Johnson et al. demonstrated that controlled release of NT3 and PDGF-AA resulted in enhanced neuronal differentiation of ES-derived neural precursor cells after transplantation into a mouse model of SCI [119]. Adapting this paradigm, Ishii et al. co-delivered antibodies against ciliary neurotrophic factor (CNTF) to neutralize the endogenous CNTF signaling on transplanted rat spinal cord-derived NSPCs, resulting in less astrocyte differentiation and greater oligodendrocyte differentiation after SCI in rat. In the present project, the small molecule dibutyryl cyclic-AMP was investigated for its potential in promoting neuronal differentiation of NSPCs.

1.9.2 Dibutyryl Cyclic-AMP

Dibutyryl cyclic-adenosine monophosphate (dbcAMP) is a membrane-permeable analogue of cyclic-adenosine monophosphate (cAMP), an important second messenger signaling molecule involved in a variety of intracellular pathways. The dibutyryl tag confers the small molecule (Mw = 491 Da) with some hydrophobic character, allowing it to penetrate cell membranes. Moreover, the dibutyryl modification also serves to prevent hydrolytic inactivation of cAMP.

Figure 1.8: Structures of cyclic-AMP and dibutyryl cyclic-AMP. The dibutyryl modification gives the molecule greater stability and enhances cell membrane permeability.

In neuroscience applications, upregulation of intracellular cyclic-AMP has been shown to promote neurite extension in culture [135,136], which can even overcome growth inhibitory molecules present on myelin [137,138]. Indeed, cAMP levels are known to be higher during development when axonal growth occurs and is even promoted by myelin components [139]. Intracellular cAMP is generated from ATP by the transmembrane enzyme adenyl cyclase. In

24 vivo cAMP upregulation can be achieved through direct application of cAMP or its analogues, most often dbcAMP, or through phosphodiesterase inhibitors such as Rolipram which reduces intracellular inactivation of cAMP. In vivo studies using these methods confirm the stimulation of axonal outgrowth following injury [140,141,142,143].

Moreover, cAMP activation has been implicated in promoting neuronal differentiation of various stem/progenitor cells. Neuronal induction via cAMP has been shown in rat hippocampal progenitor cells [144] and human neuroblastoma cells [145]. More recently, work from our lab has demonstrated that dbcAMP exposure to adult rat brain-derived NSPCs results in extensive generation of betaIII-tubulin positive neurons [18]. The primary intracellular pathway dictating cAMP-mediated neuronal differentiation appears to be primarily through the activation of Protein Kinase A (PKA) [144,145,146], which in turn phosphorylates cAMP-Responsive

Element Binding (CREB) protein, an important family of transcription factors in neuronal differentiation and maturation [147]. The CREB transcription factor family consists of CREB itself, cAMP responsive element modulator (CREM), and activating transcription factor 1 (ATF1). Homologous or heterologous dimerization of these factors forms a functional transcriptional factor unit. CREB target genes that may be important in its neurogenic role include BDNF, prolactin, bcl-2, NGF, PSA-NCAM, NGF, and cyclin D2 [148].

A PKA-independent pathway involving MEK/ERK may also contribute to cAMP-mediated neuronal differentiation [146,149] although this pathway is more implicated in the neuritogenesis promoting-role of cAMP [150]. These signaling pathways are illustrated in Figure 1.9.

1.9.3 Drug Delivery Systems for the Spinal Cord

Drug delivery to CNS tissue faces many challenges, particularly if localized and sustained delivery is required. The vertebral column encasing the spinal cord limits simple access to the tissue, and safety is a concern due to the sensitivity of the tissue to mechanical injury. Moreover, the presence of the blood-brain barrier can prevent the effective diffusion of many molecules from the blood stream to the spinal cord parenchyma.

There are many methods currently available to deliver soluble factors to the injured spinal cord. However, traditional drug delivery methods such as bolus and intravenous injections leave much to be desired as they involve non-localized or invasive procedures. Non-localized delivery methods such as intravenous injections limit the targeting ability of the drug and can lead to higher dosing requirements and systemic side effects, while continuous methods such as intrathecal drug delivery through a catheter/osmotic mini pump system has been reported to contribute to scar formation and further compression to the spinal cord [151]. Indeed in the human Phase I clinical trials of anti-Nogo-A for SCI, concerns over infection potential of indwelling catheters necessitated the switch to multiple intrathecal bolus injections for four week delivery [9].

Biomaterial strategies for drug delivery can address the need for localized and prolonged release. Injectable hydrogel systems such as collagen [152] or hyaluronan/methylcellulose [47] can be applied to the intrathecal space, allowing entrapped growth factors to diffuse to the spinal cord.

Alternatively, the hydrogel can be inserted into the spinal cord tissue itself, as in the case of fibrin [153], lipid microtubules in agarose [40,154], and collagen [155]. Drug release kinetics in these systems can be controlled by degradation, diffusion, affinity-binding systems, and immobilization [156].

Degradable polymer devices such as nano- or mircrospheres are also effective as drug delivery systems. These are typically fabricated from synthetic polyesters such as PLGA, PLA, or PCL, but other synthetic polymer classes such as polyanhydrides or polyorthesters, or natural polymers such as chitosan have also been used (reviewed in [157,158,159,160]). Microsphere systems are versatile in have been used to deliver a wide range of molecules, from small molecules to proteins, both hydrophobic and hydrophilic, to the brain and spinal cord (reviewed in [160]). Moreover, they can be designed to release their drug contents over days to years. Drug delivery particle systems can be applied adjacent to [161] or directly into nervous tissue [129,162]. They can also be incorporated into hydrogels [126,127] and guidance channels [163,164].

1.10 The Entubulation Strategy

Guidance channels have been traditionally studied in the context of peripheral nerve injuries, where it is well established that these channels can successfully promote functional nerve regeneration [165]. Studies done by our lab and others demonstrate that this strategy may hold promise in a spinal cord model as well [4,49,60,68,166,167,168]. Guidance channels provide a physical substrate for directed regeneration and are typically utilized in transection models where the rostral and caudal spinal cord stumps are inserted into the channel. Alternatively, these channels can be adapted for placement into a spinal cord cavity by myelotomy. Importantly, the implantation surgery was shown not to have any functional detrimental effects [98].

Initial studies using guidance channels typically made use of non-degradable synthetic materials such as silicone [169], PAN/PVC [60,170], or pHEMA-MMA [4,166] . These studies showed guidance channels to be effective at bridging transection gaps and encouraging the formation of tissue bridges between the two separated ends of the spinal cord, which does not occur if transection is left untreated. These bridges often contained host axons, although these tended to be limited in number. More recent work in our lab has focused on the use of channels made

27 from chitosan. As noted above, chitosan is a natural polymer that forms strong but elastic channels and the biological properties of chitosan can be tuned based on the degree of deacetylation. This has been shown to affect degradation, cell-adhesion, and stimulation of neurite outgrowth [34]. Chitosan channels in combination with NSPCs seeded along the inner walls demonstrated much more robust tissue bridge formation compared to channel alone following spinal cord transection in rat [5]. Transplanted NSPCs were shown to be present throughout the bridge. However, the NSPCs did not seem to contribute greatly to the quality of the bridge as there were no differences seen the number or the thickness of axonal fibers, or in myelination in the bridge when comparing channels with NSPCs to channels alone. The majority of NSPCs differentiated into astrocytes or oligodendrocytes, with little to no neuronal differentiation.

The aim of the current work is to investigate NSPC fate control through soluble factor delivery. For this work, neuronal differentiation of NSPC was targeted for several reasons. First, in vivo differentiation of NSPCs after transplantation in the injured spinal cord is rare, typically less than 1% [27]. As such, demonstrating enhanced neuronal differentiation provided a clear and defined endpoint by which to assess the efficacy of drug treatment. Secondly, neurons are the major functional cell type of the CNS but also one of the most vulnerable to injury. The loss of neurons after SCI accounts for much of the functional deficit. Finally, replacement of this cell type by NSPCs is under-explored in the literature and, if successful, may provide insight into mechanisms of recovery by neuronal relay systems. This is particularly appealing in transection injuries where there is no intact host architecture to rely on for signaling across the injury gap.

Several molecules have been shown to enhance neuronal differentiation of NSPCs including dbcAMP [18], interferon gamma [18], retinoic acid [171], and NT3 with PDGF [119]. DbcAMP was chosen for this project primarily for its robust effectiveness in promoting neuronal differentiation. It also has other advantages in that it is a relatively stable small molecule drug compared to proteins which can often denature and lose bioactivity. As well, dbcAMP has axonal growth promoting effects in the spinal cord independent on its action on NSPCs.

In order to achieve control of stem cell fate in the channels, a drug delivery system based on the inclusion of PLGA microspheres is proposed. Material-based treatments such as the entubulation strategy offer a unique opportunity to integrate drug release technology as part of

28 the implant itself, obviating the need for external pumps or the need of a removal surgery. Chitosan guidance channels have previously been designed to incorporate drug-loaded PLGA microspheres to provide localized and sustained delivery and have shown bioactive release of protein for up to 14 days [164]. However, this design of PLGA microspheres layered between two concentric channels was designed primarily for use in peripheral nerve injury, and practical considerations such as implant thickness made it unsuitable for in vivo testing in the spinal cord. Nevertheless, the strategy of PLGA microsphere-based drug delivery is appealing due to its versatility in drug types and temporal release windows, and would provide a platform for testing multiple drug combinations in the future to address the various challenges that SCI presents.

The larger goal of this research is to develop the entubulation strategy as a platform for studying the regenerative process after SCI. Guidance channels offer the ability to manipulate the local environment of the injury site with various cellular and molecular treatments, and to investigate their effects on the host response. The following chapters will explore the suitability of chitosan as a biomaterial for spinal cord research, the development of PLGA microsphere-loaded chitosan channels, and the use of these channels for NSPC and dbcAMP delivery to the spinal cord.

2 Evaluation of Chitosan in the Spinal Cord Chitosan Implants in the Rat Spinal Cord: Biocompatibility and Biodegradation*

*This chapter has been peer-reviewed and published in the Journal of Biomedical Materials Research Part A with the following authors: Howard Kim, Charles Tator and Molly Shoichet [172]

2.1 Abstract

Biomaterials are becoming increasingly popular for use in spinal cord repair, but few studies have investigated their long-term biocompatibility in central nervous system tissue. In this study, chitosan was compared with two commercial materials, degradable polyglycolide (Vicryl, polyglactin 910) and non-degradable expanded poly(tetrafluoroethylene) (GoreTex, ePTFE), in terms of the host tissue response and biodegradation in the rat spinal cord in two different spinal cord implantation models. In an uninjured model, implants were placed in the spinal cord intrathecal space for up to six months. At one month Vicryl implants elicited an elevated macrophage/microglia response compared to chitosan and GoreTex, which subsided in all groups by six months. Fibrous encapsulation was observed for all three materials. At six months, the in vivo degradation of Vicryl was complete, while GoreTex showed no signs of degradation, as assessed by mass loss and SEM. Chitosan implants showed evidence of chain degradation at six months as demonstrated by differential hematoxylin and eosin staining; however, this did not result in mass loss. In the second model, implants were placed directly into the spinal cord for up to twelve months. This resulted in increased immune and inflammatory responses, but did not alter degradation profiles. The same trends observed for the materials in the intrathecal space were mirrored in the spinal cord tissue. These results demonstrate that chitosan is a relatively inert biomaterial that does not elicit a chronic immune response and is suitable for long-term applications for repair of the spinal cord.

2.2 Introduction

Spinal cord injury (SCI) presents a significant challenge for regeneration due to major losses to native cellular and extracellular matrix architecture. In particular, the formation of cysts,

30 cavities, or gaps in the spinal cord at the injury site results in the lack of a physical substrate for regeneration. Biomaterials are increasingly popular as a potential strategy for treatment of SCI in that they can serve to replace the extracellular matrix at the site of injury. A wide range of biomaterials, both of natural and synthetic origin, are being investigated for potential applications in the spinal cord [36,37,173,174]. These materials can support endogenous tissue regeneration [56,175], promote directed axonal growth [43,69], enhance cell transplant survival and engraftment [76,77], deliver drugs locally [47,153,156], and seal damaged dura mater [176]. It is important to ensure that these materials are safe and well-characterized.

One of the key criteria of biomaterial design is biocompatibility. Biomaterials designed for spinal cord repair should provoke minimal chronic inflammation and immune responses when implanted in the body [177,178]. These responses depend not only on the inherent properties of the material itself, but can also be affected by the form in which the material is presented, for example implant shape [179], size [180], and porosity [181]. Degradable materials in particular are important to monitor over time because the degradation products can elicit different inflammatory responses than those of the parent material.

Chitosan is derived from the deacetylation of chitin, the primary polysaccharide component of crustacean shells. It is an attractive material because its degradation rate can be tuned based on its degree of deacetylation (DD), where fully deacetylated (DD=100%) chitosan is non- degradable [34,86] and partially deacetylated (DD=70%) is fully degradable [86,182]. Chitosan is a versatile material currently in clinical use in wound dressings, primarily for its hemostatic property [80]. We have previously reported that chitosan channels (DD=90%) promote extensive tissue bridge formation following spinal cord transection in rats [5,75]. In these reports, chitosan channels remained structurally intact for 6 months in vivo and showed no evidence of degradation. In other tissues, as well as our own in vitro screening, chitosan is degradable when DDs are less than 85-90% [34,86,183].

In this study, chitosan samples of DD=78% and 85% were investigated in terms of the in vivo foreign body response and degradation profile in the spinal cord and compared to two well- established commercial biomaterials – degradable polyglycolide (Vicryl, polyglactin 910), which is used clinically as absorbable sutures and meshes, and non-degradable expanded poly(tetrafluoroethylene) (ePTFE, GoreTex™), which is used clinically to minimize tissue

31 adhesion. 78% DD chitosan is the lowest DD limit for synthesized channels to have acceptable mechanical properties for application in the spinal cord.

The host tissue response and degradation profile of three biomaterials materials were compared in two complementary studies. In the first set of experiments, chitosan (DD=85%), Vicryl, and GoreTex were separately implanted in the intrathecal space, between the spinal cord and dura mater, and characterized over a six month period. Implants were characterized for degradation by mass loss and SEM, and biocompatibility by fibrous capsule formation, activated macrophage and reactive astrocyte responses. In the second set of experiments, the same materials were tested, with the addition of DD=78% chitosan, in the intramedullary space, directly in the spinal cord tissue parenchyma, over a twelve month period.

2.3 Materials and Methods

2.3.1 Material Processing

For chitosan sheet implants, chitosan channels were first processed as previously described [184]. Chitosan chloride (Protosan UP CL213; NovaMatrix, Drammen, Norway) was dissolved as a 1% (w/v) solution in water and precipitated with 4% NaOH solution, washed and lyophilized. The dried chitosan was made into a 3% (w/v) solution in 2% acetic acid, followed by a 50/50 (v/v) dilution in ethanol and stored at 4 °C.

Tubes were prepared in 15 cm long cylindrical glass moulds, made by inserting an inner glass rod (OD=4 mm) into a larger glass tube (ID=7 mm). The inner rod was fixed in place at both ends by rubber septa. The chitosan solution was used to form chitin tubes by adding 18.2 µl acetic anhydride per 1 ml chitosan solution, mixed for 30 s at 5000 rpm (SpeedMixer DAC 150 FVZ; Hauschild Engineering, Hamm, Germany), then injected into the moulds. After 24 h, the chitin tubes were removed from the outer mould, and washed in distilled water for an additional 24 h. The chitin tubes were converted back into chitosan by hydrolysis in 40 wt% NaOH solution at 110 °C, first for 2 hours followed by an additional 15 or 25 minutes to achieve different degrees of deacetylation (78% and 85% respectively, as determined by 1H-NMR) [185]. After another 24 h wash, the chitosan tubes were removed from the glass rods and air dried over stainless steel cylindrical cores (OD=3.7 mm). Tubes were rehydrated in water, removed from

32 the steel core and cut into 1 mm by 2 mm sheets. These sheets were air-dried and sterilized by gamma irradiation at 2.5 MRad and rehydrated in sterile saline prior to use.

Vicryl (Polyglactin 910) woven mesh (Ethicon, Somerville, NJ) and GoreTex Preclude PDX Dura Substitute (Gore, Flagstaff, AZ) were received sterile and cut into 1 mm by 2 mm sheets prior to use.

2.3.2 In vivo Implantation

All animal work was approved by the Animal Care Committee of the University Health Network. Adult female Sprague-Dawley rats (250-350 g, Charles River, St. Constant, QC) were anesthesized with 4% halothane and an oxygen/nitrous oxide (2:1) mixture, and then the halothane concentration was maintained at 2% during the operation. Following incision of the dorsal skin, a laminectomy was performed at the T8 vertebral level to expose dorsal dura overlying the spinal cord.

Intrathecal implantation: The dura was lifted and incised resulting in a 2 mm durotomy perpendicular to the midline. Each animal received two sheets (2 mm x 1 mm) of either chitosan (DD=85%), GoreTex, or Vicryl, and the sham control received only the durotomy. These sheets were inserted over the dorsal surface of the cord approximately 1 mm from the durotomy, one sheet rostral and one sheet caudal (Figure 2.1A).

Intramedullary implantation: Longitudinal incisions were made into the dura and the underlying spinal cord 1 mm lateral to either side of the midline. 2 mm x 1 mm sheets of chitosan, Vicryl, or GoreTex were inserted into the incision site (Figure 2.1B). For this experiment we used two different formulations of chitosan (DD=85% and 78%). Figure 2.1: A) Intrathecal implantion where sheets of chitosan, Gore-Tex, or Vicryl were placed on the dorsal surface of the spinal cord on either side of the durotomy. B) Intramedullary implantations are performed by longitudinal durotomy and myelotomy, followed by placement of the material into the spinal cord parenchyma.

Following implantation in both models, the dural openings were overlaid with 20 µl of fibrin glue (Beriplast P; CSL Behring, King of Prussia, PA) and the overlying muscle and skin were closed with Vicryl sutures and metal clips, respectively. Rats were given buprinorphrine post- surgery, and every 8-12 h for the next 48 h. In the intrathecal implantation study, animals were kept for either one month or six months (n=4 per group per timepoint). The animals receiving intramedullary implants were kept for one, six, or twelve months (n=3 per group per timepoint).

2.3.3 Tissue Preparation

At the specified times after implantation, rats were transcardially perfused with neutral buffered formalin as previously described [166] and the entire segment of the spinal cord adjacent to or containing the implanted materials was removed. A 1 cm portion of the spinal cord encompassing the implantation site (ie. the caudal sheet in the intrathecal model and both sheets in the intramedullary model) was harvested and post-fixed for up to one week in formalin followed by paraffin embedding.

For the intrathecally implantated animals, the rostral implanted sheet was removed, washed in saline, and dried for explant analysis as indicated below. Sections of the spinal cord were cut at 8 µm thickness and mounted on Superfrosted Plus slides (Fisher Scientific, Markham, ON). The spinal cords were sectioned parasagittally or as cross-sections for the intrathecal and intramedullary studies, respectively.

2.3.4 Explant Analysis

For the intrathecal experiment, the rostral sheet samples were removed at one and six months and analyzed as follows. For chitosan and GoreTex, the fibrous capsules could be easily separated from the implant and removed. Due to the interwoven nature of Vicryl, it was not possible to separate the fibrous capsule. The explanted materials were washed and air dried. Mass measurements were taken and compared to those pre-implantation. Samples were then gold- sputter coated and imaged by scanning electron microscopy (Hitachi S2500) at an acceleration voltage of 20 kV.

2.3.5 Staining of Paraffin-Embedded Tissue

Every sixth section was stained with Luxol fast blue and hematoxylin and eosin (LFB/H&E) or Masson’s trichrome. LFB/H&E was used to examine general tissue architecture including fibrous encapsulation and Masson’s trichrome was used to stain collagen including the extent of fibrous encapsulation.

Immunohistochemical staining with the following antibodies was also performed as previously described [166]: mouse anti-glial fibrillary acidic protein (GFAP; 1;200, Chemicon, Temecula, CA) to visualize reactive astrocytes; and mouse anti-rat monocytes/macrophages antibody (ED- 1; 1:200, Serotec, Raleigh, NC) to visualize activated macrophages.

2.3.6 Statistics

Comparisons involving one independent factor (eg. material type) or two independent factors (eg. time and material type) were analyzed using one-way and two-way ANOVA, respectively, followed by Bonferroni post-hoc test. P-values less than 0.05 were used as the criteria for statistical significance. Statistical analysis was performed using GraphPad Prism Software. All values are represented as mean ± standard deviation.

2.4 Results

2.4.1 Intrathecal Implantation

Chitosan (DD=85%), GoreTex, and Vicryl were characterized for their biocompatibility and degradation properties in the intrathecal space of healthy adult rats. Degradation was assessed by explanting the material after sacrifice and measuring mass loss and imaging structural integrity with SEM. Figure 2.2 shows the dry mass measurements of the explanted materials over the course of the study. Vicryl was significantly degraded at one month and completely degraded at six months. Neither GoreTex nor chitosan (DD=85%) showed any mass loss over the six month period. SEM images shown in Figure 2.3 are consistent with these findings, as neither chitosan nor GoreTex show evidence of degradation at either one or six months. Both dorsally and ventrally facing surfaces appeared identical to their respective pre-implantation controls, with no signs of cellular infiltration or material breakdown. Conversely, Vicryl showed

35 significant breakdown of structure under SEM at one month and it could not be identified at six months, indicating complete degradation.

Histological characterization of the implants and underlying spinal cord tissue was also performed. Figure 2.4 shows sections stained with H&E/LFB at one month (Figure 2.4A-C) and six months (Figure 2.4D-F). It should be noted that chitosan becomes very brittle upon fixation, resulting in fracturing of chitosan into shards as an artefact of sectioning, as seen in Figure 4A.

At one month, relatively thin fibrous encapsulation of chitosan and GoreTex implants was observed whereas cells had thoroughly infiltrated the Vicryl meshwork. The average fibrous capsule thickness was measured on the dural side of the implant and plotted in Figure 2.5. Vicryl had a significantly higher encapsulation thickness compared to both chitosan at one month (p<0.05) and GoreTex at both time-points (p<0.01). No significant differences were seen between chitosan and GoreTex. The fibrous capsule thickness did not change significantly between one and six months of implantation.

Notably, a shift occurs in the H&E/LFB staining of chitosan over time, with the chitosan being more eosinophilic (pink) at one month and transitioning to more basophilic (blue) at six months (Figure 2.4A, D). This shift in H&E staining pattern has been previously observed during chitosan degradation [186] and its significance is discussed more thoroughly below.

The inflammatory response against these materials is shown in Figures 2.6. Activated (ie. phagocytic) macrophages and/or microglia were characterized by their round morphology and

37 positive staining for ED1 [187]. ED1 staining was highest with Vicryl at one month, with evidence of foreign giant body cells. Chitosan and GoreTex both elicited minimal phagocytic activity at one month. At six months ED1 activity at the implant site subsided in all cases (data not shown), indicating that there was no chronic inflammatory response associated with any of these materials. No ED1 positive cells were observed in the sham control animals at either timepoint.

Figure 2.5: Comparison of fibrous capsule thickness between chitosan, Vicryl, and Gore-Tex implants at one and six months. Vicryl elicited a stronger fibroblastic response versus chitosan at one month and Gore- Tex at one and six months. No significant differences were seen between chitosan and Gore-Tex. (n = 4, mean ± standard deviation, *p<0.05, **p<0.01).

GFAP staining was also conducted to visualize reactive astrocytes in the adjacent spinal cord. Reactive astrocytes were present at both one month (Figure 2.7) and six months, and were mainly localized to the pial surface underlying the implants. Moderate reactivity was seen in all groups, and was slightly elevated in Vicryl groups.

2.4.2 Intramedullary Implantation

The intramedullary implantation model (ie. direct insertion into the spinal cord) was used to determine if the more robust inflammatory and immune response associated with the tissue injury would alter the inflammatory reaction or degradation profile of the materials. The lack of observed physical degradation of 85% chitosan after intrathecal implantation prompted the addition of a second formulation of chitosan (DD=78%), which we hypothesized would be more susceptible to degradation.

Implants inserted intramedullary were histologically characterized at one, six, and twelve months by H&E/LFB (Figure 2.8A-L). Vicryl degraded completely by six months, while both chitosans (DD=78% and DD=85%) and GoreTex remained intact and unaltered over twelve months. The increased inflammatory and immune responses elicited by the trauma induced by intramedullary insertion did not result in accelerated degradation of chitosan. Likewise, lowering the degree of deacetylation to 78% did not affect the degradation of chitosan implants over 12 months as the implants appeared to be intact. It is again notable that a change in staining was seen with both chitosan formulations as evidenced in the color shift in the H&E/LFB staining from pink to blue over time (Figure 2.8).

Fibrous capsules were formed around the different materials and were comprised mainly of fibroblasts and collagen, as assessed by Masson’s trichrome stain (Figure 2.8M-P). Measurement of the capsule thickness showed that chitosan (DD=78%) elicited a greater fibroblast response compared to chitosan (DD=85%) and GoreTex (p<0.05), as measured by average capsule thickness (Figure 2.9). Vicryl was omitted from quantification due to its marked degradation.

ED1 staining (Figure 2.10) peaks at 1 month but ED1 positive cells did not penetrate the fibrous capsule of either chitosan 85%, 78% or GoreTex. However, many ED1 positive cells were seen closely associated with Vicryl (Figure 2.10C). At six months, ED1 reactivity subsided in the surrounding spinal cord parenchyma, indicating no chronic inflammatory response generated by any of the tested materials. GFAP staining showed reactive astrocytosis surrounding the fibrous capsules of each implant at one month (Figure 2.11) and this persisted at twelve months (data not shown).

2.5 Discussion

The biocompatibility of chitosan was compared with two commercially available materials: Vicryl, an absorbable synthetic polymer mesh made primarily from poly(glycolic acid), and GoreTex, a flexible but inert ePTFE membrane. The materials were tested both in uninjured and injured spinal cord settings.

The results of the intrathecal implantation study for both Vicryl and GoreTex were consistent with similar studies in which these two materials were investigated as dural substitutes [188,189,190], and showed a minimal inflammatory response, the formation of a thin membrane with GoreTex, and a slightly elevated inflammatory response to Vicryl. Chitosan (DD=85%) performed similarly to GoreTex, without signs of chronic inflammatory or immune response. The results with chitosan show that when it is presented as a pure composition, non-porous sheet, it is an acceptable material in spinal cord tissue. Although positive GFAP staining was observed on the pial surface of the underlying spinal cord, much of this was attributed to the sustained pressure on the spinal cord caused by the implants and not necessarily a reaction to the chemical properties of the material itself. Indeed, there was some indentation of the spinal cord caused by both GoreTex and chitosan implants.

In the six month intrathecal study, chitosan did not show any physical signs of degradation. Unlike Vicryl, which degrades due to hydrolysis, the degradation of chitosan requires enzymatic catalysis. Lysozyme catalyzes the hydrolysis of glycosidic bonds between N-acetylmuramic acid and N-acetylglucosamine [191], the latter of which is the acetylated component of chitosan. It

42 has been suggested that lysozyme requires at least three consecutive acetylated monomer units to recognize the cleavage site [85], consistent with the slower degradation rates of highly deacetylated chitosans.

Lysozyme concentrations vary largely, from ~1 mg/ml in tears [192] to ~10 μg/ml in serum [193] and ~ 1 μg/ml or lower in cerebrospinal fluid [194]. The lysozyme level in normal rat spinal cord is very low, but is upregulated in the spinal cord after injury, mainly localized to microglia and macrophages [195]. Accordingly, a second study was performed to investigate whether activating the immune response with a mild injury would accelerate the degradation process of chitosan. For this study, a second formulation of chitosan was also tested, with a DD of 78%, and the duration of the implantation was extended to 12 months. Chitosan with greater D-acetylglucosamine content has also been shown to stimulate a stronger macrophage response [196], which should also accelerate the degradation response.

Intramedullary implantations of chitosan (DD=85% and 78%), Vicryl, and GoreTex resulted in fibrous encapsulation, and notably, the 78% DD chitosan resulted in a two-fold increase in capsule thickness compared to the 85% chitosan. This may be attributed to the different surface chemistries, where 85% chitosan carries a more positive charge due to increased amine content. One of the first and most influential events in the acute response to the material is protein adsorption. Differences in adsorption of immunoglobulins, complement system proteins, and adhesion molecules can dictate the strength or selective recruitment of leukocytes, and consequently, the severity of the ensuing inflammatory events. It has been reported that there are differences in protein binding to chitosan depending on the degree of deacetylation. In particular, lowering the degree of deacetylation leads to stronger activation of the complement system [197], which would explain the more robust inflammatory response to 78% chitosan compared with 85% chitosan.

Another measure of the immune response to the intramedullary implanted materials was the activated microglia/macrophages response. ED1-positive cells were found at one month in all cases, largely a result of the injury to the spinal cord during implantation. However, only in the case of Vicryl were they seen to interact directly with the material. ED1 staining subsided in the surrounding spinal cord tissue at six and twelve months, suggesting that there were no detrimental long-term or chronic effects induced by these materials. GFAP-positive reactive

43 astrocytes were also present surrounding the fibrous capsules of each material, and remained for the duration of the study. GFAP is considered a persistent signal, so this was not necessarily an indication of a chronic response to the materials. The majority of the GFAP reactivity was attributed to the injury caused by implantation, and was not a direct effect of the materials themselves given that the implants are sequestered from the surrounding spinal cord tissue by the fibrous capsule.

Although both 78% and 85% chitosan implants were unaltered physically after one year in vivo, the shifts in the H&E/LFB staining from eosinophilic to basophilic suggest some degradation in the chitosan [186]. Hematoxylin staining, used mainly to identify cell nuclei, is the result of oxidized hematoxylin (hematein) combining with aluminum ions to form a hemalum dye-metal complex. This complex is then able to label nucleic acids via interaction with the aluminum [198]. Upon chitosan degradation by lysozyme, the hydrolysed β1-4 glycosidic bond results in a free anomeric hydroxyl group on the cleaved residue. We hypothesize that this free anomeric hydroxyl group, along with the C2 amine group, is able to complex with aluminum-containing hemalum dye. Indeed, glucosamine residues have been shown to form complexes with other metal ions in this manner [199].

The lack of physical degradation of chitosan implants suggest that although scission of the chitosan chains may be occurring, these chains are sufficiently long that they are insoluble and do not diffuse away. Chitosan degradation does not immediately result in mass loss, but requires the degraded fragments to have a low enough molar mass to dissolve in the surrounding environment [183]. Our preparation of chitosan results in a very packed, non-porous structure [34]. Strong intermolecular hydrogen bonding between chitosan chains results in higher crystallinity and can lead to reduced swelling [183] and solubility [200], limiting access of lysozyme to the bulk material. This is evidenced here by a color shift in H&E staining initiating at the edges of the chitosan implants and migrating inwards over time (Figure 2.8A, E, I and B, F, J). Porosity [201] and cross-linking [91] have also previously been shown to affect chitosan degradation rate.

In conclusion, chitosan has been shown to be a safe and relatively inert material in the spinal cord. It elicits a minor foreign body response similar to that of GoreTex, including thin fibrous encapsulation and early, yet non-persisting, activation of microglia/macrophages. Changes in the

44 charge profile of chitosan occur over time, suggesting degradation of polymer chains; however, this level of degradation does not result in mass loss or changes in the physical integrity of chitosan implants up to one year in vivo. Chitosan is a relatively inert biomaterial that does not elicit a chronic inflammatory or immune response making it suitable for long-term spinal cord applications.

3 Incorporating a Drug Delivery System into Chitosan Guidance Channels Design of Protein-Releasing Chitosan Channels*

*This chapter has been peer-reviewed and published in Biotechnology Progress with the following authors: Charles Tator and Molly Shoichet [184]

3.1 Abstract

After traumatic injury to the spinal cord, the neural tissue degenerates resulting in lost function below the site of injury. Promoting axonal regeneration after injury remains a challenge; however, guidance channels have demonstrated some success when combined with cellular and protein therapies. One of the limitations of current guidance channels is the inability to deliver therapeutically relevant molecules in situ, within the guidance channel, to enhance regeneration. In an effort to provide a system for local and sustained drug release, poly(lactide-co-glycolide) (PLGA) microspheres were embedded into chitosan guidance channels by a novel spin-coating technique. The method was designed to create guidance channels with the appropriate dimensions for implantation into the spinal cord, with special attention paid to the wall thickness. The release and bioactivity of a model protein, alkaline phosphatase, was followed from the channels and compared to those from free-floating microspheres over a 90 day period. Since chitosan formulations often require the use of acidic solutions, careful attention was paid to re- design the process to minimize exposure of PLGA microspheres to acid. This was achieved as demonstrated by release and bioactivity data where alkaline phosphatase released from chitosan/microsphere channels followed a similar profile and bioactivity to those of free floating microspheres.

3.2 Introduction

Axons in the adult spinal cord fail to regenerate after injury, translating into permanent loss of motor and sensory function. Although the clinical treatment of spinal cord injury has improved over the past several decades, promotion of regeneration of nerve tissue still eludes scientists [202]. Complete transection of the spinal cord is the most severe experimental model of spinal

46 cord injury. While relatively little success has been achieved with this model in terms of functional recovery, it serves as a good model for studying axonal regeneration because there is no ambiguity that axons crossing the injury site have regenerated.

One promising strategy for studying and promoting axonal regeneration after spinal cord transection is entubulation by guidance channels. Guidance channels, popularized by their success in treating long-gap peripheral nerve injuries [203,204], function as a physical substrate for directed growth of regenerating tissue and provide a permissive environment. Our laboratory and others have demonstrated that entubulation of the transected spinal cord results in tissue bridging across the injury site and supports the survival of cellular transplants using a variety of guidance channels including: poly(hydroxyethylmethacralyate-co-methylmethacrylate) (pHEMA-MMA) [4,49], polylactic acid (PLA) [50], polyacrylonitrile/polyvinyl chloride (PAN/PVC) [170] and chitosan [75]. Unfortunately, meaningful recovery of function was not achieved in these studies.

Combining the guidance channels with growth promoting signals is a logical extension of the regenerative strategy. Therapeutic molecules such as neurotrophic factors [205,206,207], agents that neutralize growth inhibition [208,209,210], and enzymes that break down scar tissue [211] are promising candidates for stimulating axon growth. Many potential drug therapies for nerve regeneration will require prolonged delivery to allow the relatively slow process of axonal regeneration to occur. Moreover, localized delivery is desired because many of these agents have limited ability to cross the blood brain barrier and may also result in unwanted side effects if delivered systemically. Minipumps/catheters have been used to achieve sustained delivery to the spinal cord; however, the release is not localized, there is no pathway on which axons can regenerate and there are complications associated with catheter placement, such as infection and/or compression of the cord [212,213]. Our hypothesis is that the guidance channel implant can be adapted to serve a dual role as drug delivery vehicle and regenerative pathway to enhance axonal regeneration.

Several groups, our own included, have reported various designs of drug delivery channels. Typically, these have involved filling channels with a loose hydrogel matrix containing drug [49,214], direct drug encapsulation into the channel walls [215,216], or incorporating separate drug delivery devices such as rods [217,218] or degradable microspheres [163,219]. Of these

47 designs, microsphere-based systems are the most desirable because they allow for the greatest flexibility in controlling release rates. Recently, our group has reported the successful release of bioactive epidermal growth factor (EGF) from poly(lactide-co-glycolide) (PLGA) microspheres entrapped between concentric chitin and chitosan channels [164]. However, the large wall thickness of this three-layered tube was problematic when used in the spinal cord: the tube wall occupied too much space in the spinal canal, requiring the spinal cord tissue to be raised in order to be inserted into the tube. This thick, three-layered tube design led to further injury and compression of the cord and was thus inappropriate for use in the spinal cord. This concern was reported by Nomura et al., who showed that thick-walled channels (0.6 mm) resulted in cavity formation (syringomyelia) inside the inserted spinal cord stump [166].

The present study describes the design and evaluation of a drug delivery channel specifically for use in the spinal cord. Figure 3.1 summarizes the design strategy. Protein-loaded PLGA microspheres were embedded into chitosan channels by a spin-coating technique, resulting in an even distribution of microspheres along the interior channel wall. These microsphere-loaded channels were tested by monitoring the release and bioactivity of a model protein, alkaline phosphatase, over a three month period.

3.3 Materials and Methods

3.3.1 Materials

Poly(lactide-co-glycolide) (PLGA) 50/50 with an inherent viscosity of 0.37 dL/g in HFIP was purchased from Absorbable Polymers Incorporated (Pelham, NJ, USA). Poly(vinyl alcohol) (PVA), magnesium carbonate, and alkaline phosphatase were purchased from Sigma Aldrich (Oakville, ON, CA). Purified chitosan chloride (Protosan UP CL213) was obtained from NovaMatrix (Drammen, Norway). Glass and stainless steel tubing were purchased from McMaster-Carr (Dayton, NJ, USA). A Millipore Milli-RO 10 Plus and Milli-Q UF Plus (Billerica, MA, USA) water purification unit provided distilled and deionized water for all experiments. All other reagents were purchased and used as received from Caledon Laboratories (Georgetown, ON, CA) unless otherwise stated.

3.3.2 Microsphere Preparation

Microsphere mean diameter and size distribution were measured via static light scattering using a Malvern Mastersizer 2000 laser diffraction particle sizer, using refractive indices of 1.33 and 1.59 for water and PLGA respectively. Reported values are the average of three consecutive measurements.

Microspheres were imaged under high magnification by scanning electron microscopy (SEM). Samples were gold sputtered twice for 45 s and analyzed on a Hitachi S2500 SEM at 20 kV acceleration voltage.

The loading of alkaline phosphatase in the microspheres was measured by the microBCA protein quantification assay (Pierce; Rockford, IL, USA). Briefly, an accurately measured weight of particles was dissolved in 0.8 ml of dichloromethane then mixed with 0.8 ml of water. The solution was mixed vigorously for 60 s then centrifuged at 10,000 rpm for 5 min. The aqueous portion was removed and the process repeated two more times. The pooled aqueous portion was assayed against a standard curve of known alkaline phosphatase concentration. Encapsulation efficiency was defined as experimental weight percent of protein in the microsphere compared to the theoretical weight percent.

3.3.3 Chitosan Guidance Channel Preparation

Chitosan guidance channels were prepared by a chitosan-chitin-chitosan conversion process as previously described [90], with minor alterations. Chitosan chloride was dissolved in distilled water then precipitated with 4% NaOH solution, filter washed, and lyophilized. The dried chitosan was dissolved as a 3% solution in 2% acetic acid, followed by 50/50 dilution in ethanol and stored at 4oC.

Channels were prepared in 15 cm length cylindrical glass molds, made by inserting an inner glass rod (OD=4mm) into a larger glass tube (ID=7mm). The inner rod was fixed in place at both ends by rubber septa (Sigma Aldrich). The first step in the process was to convert chitosan into chitin hydrogel in order to control the shape of cast channels. 67 µl of acetic anhydride (2- fold molar excess with respect to available amines) was added to 4 ml chitosan/ethanol solution, vortexed, evacuated to remove air bubbles, and injected into the molds. After 24 h, the chitin channels were removed from the outer mold, and washed in deionized water for an additional 24

49 h. The chitin channels were converted back into chitosan by two consecutive hydrolysis cycles (2h, 110oC) in 40wt% NaOH solution, with rinsing between cycles. This conversion process allows for control over the final degree of deacetylation of the chitosan channels [90], which affects many biological properties of chitosan [34]. After another 24 h wash, chitosan channels were removed from the glass rods and air dried over stainless steel cylindrical cores (OD=3.7 mm). Channels were rehydrated in water, removed from the steel core, and cut to 8 mm lengths. The degree of deacetylation of the chitosan channels was 95% as determined by proton nuclear magnetic resonance (1H NMR) spectroscopy [220].

Figure 3.1: A) Chitosan channels are prepared by adding acetic anhydride to chitosan then injecting the mixture into a glass mold to prepare chitin channels. The channels are converted back into chitosan by high-temperature alkaline hydrolysis. B) Microsphere- loaded channels are prepared by spin-coating the interior of a chitosan channel with a chitosan solution containing microspheres.

3.3.4 Microsphere-loaded Channels

Microspheres were embedded into the channel walls by a spin-coating method. PLGA particles were suspended at a concentration of 75 mg/ml in a 2:1 mixture of 75 mM phosphate buffer (pH7) and 2% chitosan in 1% acetic acid. 20 µl of microsphere/chitosan solution was added to a horizontally mounted chitosan channel, held inside a glass mold. The channels were rotated at 2500 rpm for approximately 30 min, until the inner coating became dry. Microsphere loading was 1.5 mg per 8 mm channel.

Channels with alkaline phosphatase dissolved directly into the secondary chitosan layer (75 µg/channel) and channels loaded with blank (non-protein containing) PLGA microspheres were used as controls for the release study.

Microsphere-loaded channels were imaged by light microscopy (Leica MZ6) using ImagePro software. High magnification images of the microsphere-embedded channel wall were taken by SEM (Hitachi S2500). Channels were gold sputter coated in two cycles of 45 s and imaged at an acceleration voltage of 20 kV.

3.3.5 In Vitro Protein Release

The release of alkaline phosphatase was monitored over time and assayed for bioactivity. Free microspheres, microsphere-loaded channels, and control channels (n=3 per group) were placed in 2 ml maximum recovery Eppendorf vials and suspended in release media (PBS with 0.01% sodium azide) and incubated at 37 oC under mild agitation. At various time points, release media was collected and replaced. Samples containing free-floating microspheres were centrifuged at 8000 rpm for 3 min before media collection and vortexed to re-suspend the microspheres after media replacement. Channel groups had full media collection and replacement at each time point.

The collected release media was analyzed for protein content using the Pierce microBCA protein quantification assay and following the manufacturer’s protocol. Alkaline phosphatase was quantified by assessing the enzyme’s ability to act on a substrate para-nitrophenylphosphate (pNPP). One pNPP (Sigma Aldrich) tablet was dissolved in 5 ml of 100 mM Tris-HCl buffer, pH 8.6 with 10mM MgCl2. 50µl of pNPP solution was added to 50 µl for 10 min at room temperature. The reaction was stopped with 50 µl 4% NaOH and the absorbance of free p- nitrophenyl was measured at 405 nm and compared against a standard curve. Both assays were read on a Molecular Devices (Sunnyvale, CA, USA) VERSAmax plate reader.

3.4 Results and Discussion

PLGA is a well characterized polymer used in many drug delivery applications due to its ease of preparation, biocompatibility, and tunable degradation kinetics. Here, PLGA microspheres were

51 successfully fabricated using a standard double emulsion technique. The resultant microspheres were confirmed to be discrete spherical particles under visualization with SEM. As shown in Figure 3.2, particles had mostly smooth exterior surfaces with very few surface pores and had a mean volume weighted diameter of 26.5 µm with 80% of the spheres in the range of approximately 15 to 44 µm.

The described protocol for fabricating these microspheres was the result of several iterations where a series of process parameters (e.g. sample volumes and concentrations, homogenization speed) were adjusted to achieve high protein encapsulation efficiency and limit initial burst release. Encapsulation efficiency of alkaline phosphatase-loaded microspheres was 80.3 ± 1.3%, which is comparable to other optimized PLGA double emulsion systems [221,222,223,224]. High encapsulation efficiencies result in greater drug content in the microsphere product and minimize losses during the manufacturing process. Figure 3.2: A) PLGA microspheres visualized using scanning electron microscopy have smooth exterior surfaces with sporadic pores. B) The size distribution of PLGA particles as measured by light scattering. The average volume weighted diameter of the particles was 26.5 µm as measured by static light scattering.

Burst release, which we define as the percentage of encapsulated protein released after 24 hours in aqueous suspension, was another important factor in microsphere preparation. About 20% of

52 encapsulated alkaline phosphatase was released in the burst phase, a value that is typical for PLGA systems [164,225,226]. Burst release is typically attributed to protein at or near a porous surface or adsorbed to the outside of the particle [227]. High burst values, which in other systems are as high as 50% or greater [222,228,229,230], are problematic in most drug delivery applications because dose dumping is both inefficient and potentially harmful depending on the drug and target tissue. In our case, limiting burst release is especially important because the microspheres are exposed to additional aqueous processing during the embedding procedure. Low burst will minimize losses during this step, not only from diffusion-mediated release but also because protein close to the microsphere surface may be more susceptible to denaturation from the extra processing required. To gain a greater understanding of the acidic denaturing conditions possible with PLGA [231] and chitosan formulations, we chose to work with alkaline phosphatase as a model protein because of its known sensitivity to acidic conditions [232].

Our primary focus on the spinal cord presents unique requirements not satisfied by previous drug delivery channels designed for peripheral nerve injury. These include scale-up in channel diameter and a more stringent limit on wall thickness, which is important due to the space limitations of the spinal canal. Another important consideration is a simple manufacturing procedure that minimizes potential drug loss due to harsh conditions or extended exposure to aqueous environments.

The spin-coating procedure was devised as a simple method of stably incorporating microspheres into a chitosan channel without significant changes in the original channel dimensions or materials. Microspheres were suspended in a dilute chitosan solution and spun to dryness onto the interior of a pre-formed chitosan channel. The secondary chitosan solution was buffered such that the pH of the solution was as high as possible (~pH 6) without causing precipitation of the chitosan. This was done to minimize acidic exposure to both the PLGA polymer and the protein encapsulated within. As the microparticle/chitosan mixture dried, an even coating of microspheres embedded into the tube wall by the secondary chitosan coating resulted, as shown in Figure 3.3 under both light microscopy and SEM. This coating was stable under normal handling conditions throughout the three month study and did not delaminate even under vigorous shaking conditions.

From a design perspective, this method of incorporating microparticles improves many aspects of our previously published drug delivery channel, which was made by sandwiching microspheres between a chitosan and chitin channel [164]. The sandwich channels were primarily designed as a peripheral nerve guide, where the small inner diameter was not a limiting factor. The current design has thinner walls of approximately 200 µm (Figure 3.3). From our experience, a 300 µm thickness is the upper limit to prevent spinal cord distortion upon placement inside channel in an adult rat model (unpublished observation). Other advantages over the previous design include fewer material components, ability to predetermine microsphere loading, and localization of microspheres to the inner wall of the channel which should result in preferential release to target tissues in the central lumen.

The in vitro release of alkaline phosphatase was tracked and the results plotted in Figure 3.4. The complete release of alkaline phosphatase from free floating microspheres took approximately 90 d. The cumulative release profile of microsphere-loaded channels mirrored that of free microspheres, with a consistent difference between the two curves of approximately 7%, which represents protein loss during the microsphere embedding procedure. In contrast, when alkaline phosphatase was dissolved directly into the secondary chitosan coating, 80% was released in the first 24 h. Thus the secondary chitosan layer does not act as a significant barrier to protein diffusion and the release of protein from channels is primarily dependent on the properties of the microspheres themselves.

Figure 3.4: Cumulative release profile of (A) total and (B) bioactive alkaline phosphatase from free floating PLGA microspheres (FreeMS), microsphere- loaded channels (MLC), and dispersed directly into the channel walls (FreeAP) (n=3 for all groups). The release profile of MLCs closely mirrored that of free microspheres. Alkaline phosphatase was released in bioactive form throughout the study, with 87.6 ± 0.4% (avg ± s.d.) of released protein remained active from free microspheres compared to 78.7 ± 1.1% from MLCs.

Release from microspheres and channels followed the classic PLGA tri-phasic profile [233]. The initial stage is attributed to burst release, which is mainly diffusion mediated. The secondary phase is characterized by a steady release rate and is attributed to a balance between increased drug mobility due to hydrolysis of PLGA chains, and the increasing distance that drugs farther from the surface must diffuse [233]. The final accelerated release phase is explained by the local build-up of lactic and glycolic acid, the PLGA degradation byproducts, which have an autocatalytic effect on polyester degradation [234]. Eventually the PLGA chains breakdown to a critical state such that the microparticle structure starts to disintegrate [235]. Although the timeframe for release in this study was three months, the release rates for specific drug applications can be tuned to be shorter or longer by modifying various process parameters during microsphere preparation. These include changing the polymer lactide/glycolide ratio [236] or molecular weight [237], introducing copolymers such as polycaprolactone [238] or additives such as polyethylene glycol [224], or by simply altering microsphere size [239]. PLGA microsphere systems have been shown to be capable of degradation over timescales of a few weeks [237,240] to several months [241,242].

Protein stability is always one of the biggest concerns in drug delivery systems. As shown in Figure 3.4B, the majority of alkaline phosphatase remained bioactive throughout the study. Comparing the release data shows that 87.6 ± 0.4% (average ± standard deviation) of the alkaline phosphatase released from PLGA microspheres was active. The bioactivity ratio dropped to 78.7 ± 1.1% in microsphere-loaded channels, indicating that the embedding process did not greatly denature the encapsulated protein. The concern over acidic exposure was attenuated by buffering both the chitosan solution used in the embedding procedure, and by buffering the interior of the PLGA microspheres themselves with magnesium carbonate [243]. Together, these precautionary measures resulted in a high retention of bioactivity of an acid-sensitive protein released from the drug delivery channels.

Guidance channels provide a unique opportunity to influence the local environment, which we have exploited through the integration of a drug delivery system. The microsphere-based system allows great flexibility in both the type and number of therapeutic factors to be delivered, and also the timeframes during which these drugs are released. Concurrent drug therapies can be achieved by incorporating each drug separately into microspheres, then embedding these different microspheres into the same channel. The addition of a drug delivery system capable of local and sustained release is a major and potentially significant evolution of the guidance channel, and will serve as an important tool to test the regenerative capacity of combination therapy for spinal cord injury repair. Future directions of this work will focus on incorporating relevant growth factors for promoting regeneration after spinal cord injury.

3.5 Conclusion

PLGA microspheres were formed with high encapsulation efficiency and embedded into pre- formed chitosan guidance channels. Spin-coating of chitosan channels resulted in the stable entrapment of PLGA microspheres along the inner channel surface. The current design results in thin-walled channels with appropriate dimensions for implantation in the spinal cord. These channels are capable of long-term release of bioactive protein and are promising for the delivery of therapeutic agents to the regenerating spinal cord.

4 Evaluating Directed Differentiation of NSPCs into Neurons with Dibutyryl Cyclic-AMP Effects of dibutyryl cyclic-AMP on survival and neuronal differentiation of neural stem/progenitor cells transplanted into spinal cord injured rats*

*This chapter has been peer-reviews and accepted for publication in PLoS One with the following authors: Tasneem Zahir, Charles Tator and Molly Shoichet [244]

4.1 Abstract

Neural stem/progenitor cells (NSPCs) have great potential as a cell replacement therapy for spinal cord injury. However, poor control over transplant cell differentiation and survival remain major obstacles. In this study, we asked whether dibutyryl cyclic-AMP (dbcAMP), which was shown to induce up to 85% in vitro differentiation of NSPCs into neurons would enhance survival of transplanted NSPCs through prolonged exposure either in vitro or in vivo through the controlled release of dbcAMP encapsulated within poly(lactic-co-glycolic acid) (PLGA) microspheres and embedded within chitosan guidance channels. NSPCs, seeded in fibrin scaffolds within the channels, differentiated in vitro to III-tubulin positive neurons by immunostaining and mRNA expression, in response to dbcAMP released from PLGA microspheres. After transplantation in spinal cord injured rats, the survival and differentiation of NSPCs was evaluated. Untreated NSPCs, NSPCs transplanted with dbcAMP-releasing microspheres, and NSPCs pre-differentiated with dbcAMP for 4 days in vitro were transplanted after rat spinal cord transection and assessed 2 and 6 weeks later. Interestingly, NSPC survival was highest in the dbcAMP pre-treated group, having approximately 80% survival at both time points, which is remarkable given that stem cell transplantation often results in less than 1% survival at similar times. Importantly, dbcAMP pre-treatment also resulted in the greatest number of in vivo NSPCs differentiated into neurons (37 ± 4%), followed by dbcAMP- microsphere treated NSPCs (27 ± 14%) and untreated NSPCs (15 ± 7%). The reverse trend was observed for NSPC-derived oligodendrocytes and astrocytes, with these populations being highest in untreated NSPCs. This combination strategy of stem cell-loaded chitosan channels implanted in a fully transected spinal cord resulted in extensive axonal regeneration into the injury site, with improved functional recovery after 6 weeks in animals implanted with pre- differentiated stem cells in chitosan channels.

4.2 Introduction

Repair after traumatic spinal cord injury (SCI) remains a significant challenge. Extensive cell death and tissue disorganization, including demyelination, coupled with an inherent aborted and lack of spontaneous regeneration results in permanent disruption of signalling pathways. Although resident endogenous neural stem cells exist lining the central canal of the adult spinal cord [245], their recruitment is limited in response to injury [246]. This is in large part due to the inhibitory environment of adult central nervous system (CNS) tissue, which is further exacerbated after injury.

Cell transplantation therapy is a promising approach to replace the damaged or lost cells after SCI. In particular, adult neural stem/progenitor cells (NSPCs) are attractive because of their ability to form the three major CNS cell types exclusively. However, in practice, NSPCs transplanted into the spinal cord have a much greater tendency to differentiate into either astrocytes or oligodendrocytes, and rarely undergo neuronal differentiation in vivo [10,72,247]. Indeed, many attempts to program NSPCs to preferentially promote neuronal differentiation in vitro have not succeeded in vivo in the spinal cord [248,249,250]. It is important to investigate the potential of neuronal replacement strategies for SCI as neurons are the core functional component of spinal cord signalling. Neuronal replacement after SCI is important to areas rich in neurons such as the cervical or thoraco-lumbar segments of the spinal cord.

The advantages of transplanting NSPCs in combination with implants of biomaterial guidance channels have been described [5,75,99]. After complete spinal cord transection injury in the rat, new tissue formed between the two spinal cord stumps as early as five weeks post injury [75]. Transplanted NSPCs were found to survive and integrate into this tissue bridge, and were distributed throughout the length of the bridge. However, very few NSPCs differentiated into neurons and limited functional recovery occurred.

Directed differentiation of NSPCs can be accomplished through exposure to specific soluble factors. For example, platelet-derived growth factor has been shown to promote oligodendrocyte fate-specification [251] while bone morphogenetic protein-2 promotes astrocyte differentiation [252]. Zahir et al., screened several factors for promoting neurogenesis from adult rat subventricular zone (SVZ)-derived NSPCs and reported that dibutyryl cyclic-AMP (dbcAMP), a membrane-permeable analogue of cyclic-AMP, and the cytokine interferon-gamma both

58 enhanced neuronal differentiation after one week exposure in culture [18]. Several other groups have used dbcAMP to differentiate a variety of neural stem or progenitor cells into neurons [144,145,146,253]. This is thought to be primarily through the PKA pathway [144], which causes upregulation of cAMP-responsive element binding (CREB), an important transcription factor in regulating neuronal activity [148]. Moreover, dbcAMP may have beneficial effects independent of its action on NSPCs, particularly with respect to promoting axonal regeneration [140]. It is a downstream molecule in neurotrophin signaling [137,138], which has been shown to allow neurite outgrowth in the presence of myelin inhibitors in vitro [137] and in vivo [141].

In the present study, directed neuronal differentiation of NSPCs with dbcAMP was characterized first in vitro on 2D chitosan film surfaces and 3D fibrin scaffolds, and then in vivo in a rat spinal cord transection model. Specifically, NSPCs were transplanted in chitosan guidance channels in which the walls were loaded with dbcAMP-eluting poly(lactic-co-glycolic acid) (PLGA) microspheres [184]. We have recently shown that chitosan channels exhibit excellent biocompatibility in the spinal cord over a one year period [254]. Microsphere-based release of dbcAMP facilitates local and sustained drug release, providing a potential method for in situ differentiation. The effect of directed neuronal differentiation of NSPCs with dbcAMP was evaluated for its effect on survival, differentiation, and integration into the rat spinal cord following SCI. As well, the timing of differentiation was also investigated: NSPCs were either pre-differentiated in vitro prior to transplantation, or transplanted with dbcAMP-releasing microspheres for in vivo differentiation. Finally, fibrin was used as a 3D scaffold to entrap/suspend the cells within the channels. Fibrin hydrogel has previously shown to be useful for stem cell transplantation to the spinal cord [255].

4.3 Materials and Methods

4.3.1 Chitosan Films and Channels

High quality chitosan (Protosan UP B 80/20; NovaMatrix) was dissolved as a 3 wt% solution in 2% acetic acid, and then diluted with an equal volume of ethanol. Chitosan films were prepared by allowing 10 ml of chitosan solution to air dry in a 10 cm Petri dish. Once dry, the film was soaked in neutralization solution (90% methanol, 7% water, 3% ammonium hydroxide) for 15

59 min then washed thoroughly with water. Films were cut into 8 mm discs and disinfected in 70% ethanol prior to use in cell culture studies.

Chitosan channels were prepared as previously described [34]. Briefly, chitosan/ethanol solution was reacted with acetic anhydride and injected between concentric cylindrical glass molds, resulting in chitin hydrogel channels with an inner diameter of 4 mm and outer diameter of 8 mm. Chitin channels were washed for 24 h then were hydrolyzed in 40 wt% NaOH at 110oC for 2 h, rinsed, then hydrolyzed again for another 25 min resulting in chitosan with a degree of deacetylation of 85% as assessed by 1H-NMR [220]. Upon washing, channels were air-dried on 3.7 mm stainless steel rods. Channels were cut into 8 mm lengths.

PLGA microspheres were embedded along the inner walls of chitosan channels as previously described [184]. Briefly, 5 mg of blank or dbcAMP-containing PLGA microspheres (described later), was suspended in 25 μl of a buffered chitosan solution (1:5, v/v mixture of 2% chitosan in 1% acetic acid and 75 mM phosphate buffer, pH 7). The microsphere solution was introduced to the interior of channels under a rotation of 2500 rpm and spun until dry. The microsphere- embedded channels were then neutralized in a solution of 90% methanol, 7% water, and 3% ammonium hydroxide for 1 min, rinsed thoroughly, and air dried. Channels were then sterilized by gamma irradiation at a dose of 2.5 MRad prior to use in subsequent in vitro or in vivo cell culture studies.

4.3.2 Dibutyryl Cyclic-AMP PLGA Microspheres

Poly(lactic-co-glycolic acid) (PLGA) 50/50 of both 0.20 and 0.37 g/dl inherent viscosity in hexafluoroisopropanol (HFIP) were used as received from Lactel. PLGA microspheres were fabricated using a W/O/W double emulsion procedure. The inner aqueous phase consisted of 75 µl of ddH20 with or without 40 mg dibutyryl cyclic-adenosine monophosphate (Sigma-Aldrich). 130 mg of PLGA (80 wt% 0.20 g/dl, 20 wt% 0.37 g/dl) was dissolved in 600 µl of dichloromethane/acetone (3:1). The two solutions were emulsified under sonication (Vibracell VCX 130, Sonics and Materials) for 45 s using a 3 mm probe at 30% amplitude. This emulsion was added to 25 ml of 2.5% polyvinyl alcohol (PVA)/10% NaCl solution and homogenized at 4000 rpm for 60 s. The resultant solution was poured into a 250 ml bath of 0.25% PVA/10% NaCl solution under magnetic stirring. In a typical batch, three identical preparations were added to the final bath. Four hours later, the hardened microspheres were collected and washed

60 by centrifugation, then washed again over a 0.2 µm filter, collected, lyophilized and stored at - 20oC until use.

The amount of encapsulated dbcAMP was measured by dissolving a known quantity of microspheres in 800 μl of dichloromethane. An equal volume of water was added and the solution was vortexed and centrifuged for 3 min at 10 000 rpm. The water phase containing drug was collected. This process was repeated two more times. The collected water phase was then measured for dbcAMP content using a Nanodrop UV spectrometer at 273 nm against a calibration curve of known dbcAMP solution concentrations. Encapsulation efficiency (EE) was calculated as the percentage of actual loading compared to the maximum theoretical loading based on preparation parameters.

To measure the release of dbcAMP from microspheres or microsphere-loaded channels, approximately 3 mg of microspheres (accurately measured) or a single microsphere-loaded channel was suspended in 1.8 ml PBS and stored under gentle agitation at 37oC. At various times, 2 μl of the release buffer was sampled for measurement by Nanodrop UV spectroscopy at 273 nm. Known concentrations of dbcAMP were incubated alongside release samples, and a small but measurable decrease in dbcAMP signal due to hydrolysis was observed over time and was accounted for in drug release calculations. At the end of the release period, mass balance measurement of the remaining dbcAMP in microspheres was quantified by phase extraction in dichloromethane.

4.3.3 Adult Neural Stem/Progenitor Cells

Primary NSPCs were isolated from the subventricular zone of adult Wistar rats, both GFP and non-GFP, and propagated as previously described [18]. The transgenic rats expressing GFP (Wistar-TgN(CAG-GFP)184ys) were the gift of Dr. Armand Keating (University of Toronto, Toronto) and have been previously described [256,257]. NSPCs were passaged as neurospheres in growth media: Neurobasal media (Gibco-Invitrogen), B27 neural supplement (Gibco- Invitrogen), 2mM L-glutamine (Sigma-Aldrich), 100 μg/ml penicillin-streptomycin (Sigma- Aldrich), 20 ng/ml epidermal growth factor, 20 ng/ml basic fibroblast growth factor (Sigma- Aldrich), and 2 μg/ml heparin (Sigma-Aldrich). NSPCs were used after the fourth or fifth passage for in vitro and in vivo studies.

4.3.4 Chitosan Film Studies

Chitosan films were placed in 48-well plates and soaked in media prior to use. NSPCs were spun down at 1500 rpm for 5 min and the resultant pellet was re-suspended and dissociated in fresh growth media, with or without dbcAMP. The cell solution was then added so that each well contained 30 000 cells, and the final volume of media in the wells was 500 μl. After 24 h, the media was removed and replaced with differentiation media - Neurobasal media containing B27 supplement, L-glutamine, penicillin-streptomycin, and 1% FBS - with or without dbcAMP. Half-volume media changes were performed every 48 h thereafter.

4.3.5 Chitosan Channel Studies

For cell culture studies in channels, NSPCs were seeded in three-dimensional fibrin scaffolds. Fibrinogen (F3879; Sigma) was dissolved as a 22.2 mg/ml solution in 0.03 M Tris buffered saline (TBS), pH7, filtered through a 0.22 μm filter, and then stored at -80oC. Thrombin (T6884; Sigma) and aprotinin (A4529; Sigma) were similarly dissolved in TBS at concentrations of 24 NIHU/ml and 5 mg/ml respectively, filtered and stored at -80oC.

On the day of use, fibrinogen, aprotinin, and laminin (Sigma) solutions were combined, resulting in a solution of 20 mg/ml fibrinogen, 500 μg/ml aprotinin, and 20 μg/ml laminin. This solution was mixed 5:3 with media containing dissociated NSPCs. The cell-fibrinogen mixture (48 μl) was then added by pipette into the lumen of pre-hydrated chitosan channels, followed by 12 μl of thrombin solution. After mixing, the fibrin scaffold containing NSPCs formed within thirty seconds. The initial cell number in each fibrin-filled channel was 250 000.

Cell-seeded channels were then placed in 24-well plates containing 1 ml of fresh growth media, with or without 1mM dbcAMP. For in vitro studies, the media was removed after 24 h and replaced with differentiation media, with or without 1 mM dbcAMP. Half-volume media changes were performed every 48 h thereafter.

NSPC-seeded channels for in vivo studies were prepared similarly using GFP-positive cells. Channels were implanted after either 1 or 4 days in vitro (1div and 4div respectively), depending on dbcAMP treatment (Table 1). The 1div channels did not receive a media exchange prior to transplantation. The 4div channels received media changes as described above, but FBS was removed from the differentiation media. Excess samples from each treatment group were

62 prepared and fixed on the day of transplant and analyzed for cell number and percentage of betaIII-positive neurons.

4.3.6 Immunocytochemistry

For cell culture experiments, fixed samples were permeabilized with PBS containing 0.3% Triton X-100 and 1% bovine serum albumin (BSA) for 15 min followed by 1 h blocking in 1% BSA. Samples were incubated for 2 h with primary antibodies including mouse anti-betaIII tubulin (1:1000, Covance) for neurons, mouse anti-GFAP (1:200, Millipore) for astrocytes, mouse anti- RIP (1:5, DSHB) for oligodendrocytes, mouse anti-nestin (1:100, BD Biosciences) for neural precursors, and mouse anti-Ki67 (1:100, Novocastra Laboratories Ltd) for proliferating cells. Following three 10 min washes in PBS, samples were incubated with goat anti-mouse IgG Cy3 (1:500, Jackson Immunoresearch) for 1 h. After another series of washes, samples were incubated with 10 μM Hoecsht 33342 (Invitrogen) nuclear dye and coverslipped with Vectashield mounting media. A minimum of five randomly selected fields per sample were photographed on an Olympus BX61 epifluorescent microscope to quantify cells co-labelled with nuclear dye and secondary antibody.

4.3.7 Quantitative RT-PCR

Cell culture samples were lysed after seven days and RNA was collected and purified using RNAqueous Micro kit (Ambion). Following RNA isolation and DNAse I treatment, total RNA concentration and purity was measured using Nanodrop ND-1000. RNA concentration was normalized across samples then built into cDNA libraries using a AffinityScript RT kit (Stratagene).

Quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) amplification was performed on a Light Cycler 480 (Roche) using SYBR Green II master mix kit (Roche). Following a 5 min activation at 95oC, fluorescent measurements were taken after temperature cycles of 10 s at 90oC, 15 s at 65oC, and 10 sec at 72oC. After 50 cycles, the temperature was held at 65oC for 1 min, followed by temperature ramping to 95oC over 8 min to generate a melt curve. All qRT-PCR samples were performed in triplicate. Gene expression for each sample was normalized to the housekeeping gene HPRT. Primer design for target genes of interest are

63 previously reported [258]. Data is expressed as fold-difference compared to undifferentiated (day 0) cells.

4.3.8 Animal Studies

The animal work was approved by the Animal Care Committee of the University Health Network. Adult female Sprague-Dawley rats (250-350 g, Charles River, St. Constant, QC) were anesthetized with 4% isofluorane and an oxygen/nitrous oxide (2:1) mixture. The isofluorane concentration was dropped to 2% during the operation. Following incision of the skin on the back, vertebral levels T6 to T10 were exposed. Laminectomy was performed on T7 to T9 exposing the spinal cord. The facets of the vertebrae at T7 to T9 were also removed. A midline longitudinal durotomy was performed, then cut laterally at the rostral and caudal ends to fully expose the underlying spinal cord. The spinal cord was completely transected at T8 with microscissors, as were the adjacent underlying dorsal and ventral roots.

Chitosan channels containing NSPCs in the fibrin scaffold were then implanted by placing the transected stumps into the channel openings. Treatment groups are described in Table 4.1. Channels were positioned symmetrically such that the caudal and rostral stumps were inserted equidistant inside the channel, approximately 2 mm, so that the gap between the stumps was approximately 4mm. The spinal cord stumps were inserted sufficiently far into the channel so that the stumps were in apposition to the fibrin scaffolds within the channels. A 4 mm x 3 mm sheet of Gore-Tex Preclude MVP membrane (Gore-Tex, Flagstaff, AZ) was placed at each end covering the channel-spinal cord interface. 30 μl of fibrin glue (Beriplast P, provided generously from CSL Behring) was then applied to secure the stumps to the implant. The overlying muscle and skin were closed with 3-0 Vicryl sutures (Ethicon) and metal clips, respectively.

Table 4.1: In vivo treatment groups Culture period 1 mM dbcAMP Cell Number prior to in media on Day of Animals for Animals for transplant during culture Microsphere Transplant two-week six-week Group (days) period? Content (x103) survival (n) survival (n) Untreated, 1div 1 No No drug 83 ± 4 5 5 Untreated, 4div 4 No No drug 46 ± 8 3 - dbcAMP-MS, 1div 1 No dbcAMP 78 ± 12 4 5 dbcAMP, 4div 4 Yes No drug 57 ± 10 4 6

Rats were given buprenorphrine post-surgery, and every 8-12 h for the next 48 h. Immunosuppression with cyclosporine-A (15 mg/kg) was administered daily via subcutaneous injection. Functional recovery was assessed using the Basso, Beattie, and Bresnahan (BBB) open field locomotor scale [259] once a week. The scores for the left and right hindlimb were averaged for each animal at each timepoint.

4.3.9 Tissue Preparation

At either 2 or 6 weeks post-implantation, animals were sacrificed via transcardial perfusion with 4% paraformaldehyde in 0.1M PBS, pH 7.4. Spinal cords were removed and cryoprotected in 30% sucrose for a minimum of 24 h. The length of cord encompassing the implant site, as well as 2-3 mm on either side, was removed for sectioning. Tissue was cut as 20 μm cryosections onto Superfrost Plus slides (Fisher Scientific, Markham, ON) in 1:10 series and stored at -80oC. For the 6 week animals, one sample from each treatment group was cut parasaggitally. All other 6 week samples, and all 2 week samples, were cut as cross-sections.

4.3.10 Tissue analysis

To quantify cell survival, every tenth section was stained with DAPI (Vectashield) and imaged under fluorescence to detect the GFP-positive transplanted cells. Cell counts were calculated by manually counting GFP-positive cells associated with DAPI-stained nuclei across the sample and extrapolating by a factor of ten. False positive signals associated with autoflourescence were eliminated by also imaging under the Cy3 filter.

For immunohistochemical analysis of tissue sections, slides washed with PBS then blocked in 2% goat serum and 0.3% Triton X-100 for 1 h. Primary antibodies were incubated in block or PBS overnight at 4oC. After three consecutive 10 min PBS washes, slides were incubated with secondary antibody in PBS for 1 h. Following another 3 washes, sections were coverslipped with Vectashield anti-fade mount containing DAPI. Some sections were double-labelled. In these cases the second primary-secondary antibody pairings was stained following the first pairing.

Primary antibodies used were mouse and rabbit anti-betaIII tubulin (1:1000, Covance), mouse anti-βIII tubulin (1:1000, Covance) for neurons, mouse anti-GFAP (1:200, Millipore) for astrocytes, mouse anti-CC1 (1:1000, Calbiochem) for oligodendrocytes, mouse anti-nestin

(1:100, BD Biosciences) for neural precursors, mouse anti-3CB2 (1:50, Hybridoma Bank) for radial glia, mouse anti-Reca1 (1:100, Serotec) for endothelial cells, mouse anti-prolyl-4- hydroxylase (1:100, Millipore) for fibroblasts, mouse anti-GFP (1:400, Invitrogen) for green fluorescent protein, and mouse anti-Ki67 (1:100, Novocastra Laboratories Ltd) for proliferating cells. Secondary antibodies included goat anti-mouse Alexa 568 (Invitrogen), goat anti-rabbit Alexa 568 (Invitrogen), goat anti-mouse Alexa 647 (Invitrogen), and goat anti-mouse Cy5 (Jackson Immuoresearch) all at 1:500. Images for differentiation analysis were taken on a Zeiss Observer Z1 confocal microscope using Volocity 4.3.2 software. A minimum of ten random fields containing the GFP-positive transplanted NSPCs were photographed over multiple tissue sections for each animal per stain. Phenotypic analysis of transplanted NSPCs was calculated based on identification of GFP-positive cells associated with a DAPI-stained nuclei and the immunohistological marker of interest.

4.3.11 Statistics

Comparisons involving one independent factor (eg. treatment) or two independent factors (eg. treatment and time) were analyzed using one-way and two-way ANOVA, respectively, followed by Bonferroni post-hoc testing. P-values less than 0.05 were used as the criteria for statistical significance. Statistical analysis was performed using GraphPad Prism software. All values are represented as mean ± standard deviation.

4.4 Results

4.4.1 Dibutyryl Cyclic-AMP Promotes Neuronal Differentiation of NSPCs

NSPCs were cultured on chitosan films and exposed to varying concentrations of dbcAMP for 7 d, then stained with the neuronal marker betaIII tubulin to quantify its effect on directed neuronal differentiation (Figure 4.1A). In the absence of dbcAMP, NSPCs produced very few neurons (1.4 ± 1.2%). At increasingly higher dbcAMP concentrations, the percentage of betaIII tubulin- positive cells also increased, reaching as high as 94.5 ± 0.5% with 4 mM dbcAMP. 1mM dbcAMP, which produced 84.8 ± 4.8% neurons, was selected as the target concentration for subsequent studies because it resulted in greater numbers of total cells compared to 4 mM.

A more complete differentiation profile was performed on NSPCs treated with dbcAMP (Figure 4.1B-J). NSPCs were cultured for 7 d with either no dbcAMP, 1mM dbcAMP for the first day only, or 1mM dbcAMP for all 7 days. All groups contained 1% FBS in the media. Only with prolonged exposure to dbcAMP was extensive neuronal differentiation seen. This was in contrast to 1 d dbcAMP exposure which resulted in only 7.7 ± 3.6% neurons. The increase in neurons with extended dbcAMP treatment was accompanied by a decrease in the percentages of RIP- positive oligodendrocytes (7.0 ± 3.4% vs. 21.1 ± 11.4% in FBS alone) and nestin-positive progenitor cells (12.7 ± 4.5% vs. 33.9 ± 9.6% in FBS alone). Very few GFAP-positive astrocytes were produced, with or without dbcAMP, as is typical for our population of brain- derived rat NSPCs [18,251], particularly on chitosan substrate [258]. One day treatment with dbcAMP had no effect on differentiation compared to FBS controls.

Cell density was notably lower in dbcAMP-treated wells after 7 d, suggesting either cytotoxic or anti-proliferative effects of the drug. To investigate this, NSPCs were characterized at 1, 3, and 7 d for total cell number, as well as for markers of proliferation and differentiation (Figure 4.1K,L). Prolonged exposure to dbcAMP in culture had no effect on total cell numbers at 3 d, but there was a clear anti-proliferative effect by 7 d where dbcAMP largely promoted post- mitotic neuron formation. Ki67 staining for proliferating cells was 4-fold lower for dbcAMP- treated cells (9.4 ± 1.4%) at 3 d than controls (37 ± 6.4%). DbcAMP appeared to have a direct effect on promoting neurons vs. being selectively cytotoxic to glial cells because oligodendrocyte and astrocyte populations did not decrease over time (Figure 4.1M).

4.4.2 PLGA Microspheres for dbcAMP Release

Based on the in vitro data where the greatest neuronal differentiation was observed in the presence of constant dbcAMP for 7 d, dbcAMP was encapsulated in PLGA microspheres and the release from which was tailored for 1 week. The encapsulation efficiency of dbcAMP was 80% and the loading was 19 wt%. Release curves of dbcAMP from PLGA microspheres, both free- floating in buffer and embedded within chitosan channels, are shown in Figure 4.2A. Drug release from native microspheres was linear over 11 days. Once incorporated into channels, the release of dbcAMP occurred over approximately 5 days, after which the drug contents are depleted. This discrepancy in release profiles is largely due to drug losses during the process of embedding microspheres into the channel. This was confirmed by mass balance which showed

68 less drug content in dbcAMP microsphere-loaded channels than expected based on microsphere quantity.

4.4.3 NSPC Differentiation in Microsphere-Loaded Channels

DbcAMP-loaded microspheres were incorporated into chitosan channels and tested for their efficacy in differentiating NSPCs in three-dimensional constructs, as depicted in Figure 4.2B. NSPCs were seeded into fibrin scaffolds and cultured within microsphere-loaded channels containing either no drug or dbcAMP. After seven days, NSPCs in three-dimensional fibrin scaffolds were well distributed and viable as assessed by calceinAM (live) / ethidium homodimer (dead) staining (Figure 4.2C).

NPSCs remained responsive to dbcAMP in 3D fibrin scaffolds (Figure 4.2D-H). NSPC cultured in media containing 1 mM dbcAMP for 7 days resulted in 54 ± 5% betaIII-positive neurons, while NSPCs cultured with dbcAMP-releasing microspheres resulted in approximately 37 ± 4% neurons. Both groups had significantly greater percentages of neurons than NSPCs cultured without dbcAMP, or dbcAMP for only the first 24 h, which resulted in only 7 ±3% and 11 ± 4% neuronal differentiation, respectively. Quantitative RT-PCR measurement of betaIII-tubulin mRNA showed a similar trend of expression between treatment groups, further supporting the immunostaining data (Figure 4.2I,J).

4.4.4 In vivo NSPC Transplantation

NSPCs were seeded in the fibrin scaffold within channels and incubated 1 or 4 days in vitro (div) prior to transplantation (see Table 4.1). The 1 div samples included NSPCs seeded in channels containing blank or dbcAMP-releasing microspheres. These groups were transplanted after only one day incubation to ensure that the majority of dbcAMP released from microspheres occurs in vivo. The 4 div groups included both NSPCs pre-differentiated (40% betaIII-tubulin-positive neurons) in 1 mM dbcAMP media prior to transplant and the corresponding untreated control. Cell numbers at the time of transplantation are listed in Table 4.1. The decrease in cell numbers for the 4 div groups was attributed to incubation in serum-free media.

Figure 4.2 (preceding page): A) Cumulative release profiles of dbcAMP from microspheres and microsphere-loaded channels. The process of embedding microspheres into channel walls is likely responsible for early degradation of PLGA and faster drug release from channels. B) Schematic of the entubulation strategy. NSPCs are seeded on fibrin scaffold within a chitosan channel. Drug-loaded PLGA microspheres release dibutyryl cyclic-AMP causes NSPCs to preferentially differentiate into neurons. C) Viability of NSPCs in a three- dimensional fibrin scaffold. Simultaneous staining of CalceinAM (green) and Ethidium homodimer (red) for live and dead cells respectively show good cell viability of NSPCs in fibrin scaffolds at 1 week. Scale bar represents 100 μm. D-G) Immunostaining of NSPCs for DAPI-nuclear stain and betaIII-tubulin with various dbcAMP treatments. Scale bar represents 100 μm. H) Quantification of betaIII-tubulin immunostained NSPCs with various dbcAMP treatments. I,J) Quantitative RT-PCR data for (I) betaIII tubulin and (J) nestin mRNA expression with various dbcAMP treatments, normalized to housekeeping gene HPRT. Data represented as mean ± standard (n=3 to 6).

The channels were implanted following spinal cord transection, with the spinal cord ends abutting the ends of the fibrin scaffold (Figure 4.3A). After two and six weeks, animals were sacrificed. Robust tissue bridges across the injury site formed as early as two weeks (Figure 4.3B). As shown in Figures 4.3C and D, GFP-positive transplanted NSPCs were found throughout the bridge tissue. Cell transplant survival was assessed by GFP-positive cell counts (Figure 4.3E). At 2 weeks, NSPCs pre-treated with free dbcAMP resulted in significantly higher cell survival, approximately 80%, compared to both untreated NSPCs (1 div or 4 div) and NSPCs implanted in chitosan channels with dbcAMP microspheres. Cell survival percentages did not differ between two and six weeks for any treatment group. Staining for the proliferative marker Ki67 was very low for NSPCs at two weeks (data not shown). There was no definite migration of NSPCs into the spinal cord stumps.

Six week animals were also monitored weekly for hindlimb function using the BBB motor function scale. A small but statistically significant increase in hindlimb movement at 6 weeks was observed in animals receiving NSPCs pre-treated with dbcAMP for 4 div compared to untreated NSPCs (Figure 4.3F).

Figure 4.3: A) Photograph of the surgical implantation of fibrin-filled chitosan channels. B) Tissue bridges obtained from animals 2 weeks after implantation. C,D) Longitudinal section of tissue bridge demonstrating NSPC survival after 6 weeks in an animal receiving dbcAMP pre-treatment (dbcAMP, 4div). Boxed area in (C) is magnified in (D). E) NSPC survival after 2 and 6 weeks for various treatment groups. F) Assessment of functional recovery using the BBB locomotor scale. After 6 weeks, rats receiving transplants of dbcAMP-pre-treated NSPCs show a statistically significant increase in hindlimb function relative to untreated animals (*, p<0.05). Mean ± standard deviation shown for n=4 to 6.

Figure 4.4 (preceding page): A-L) Representative images of tissue samples demonstrating NSPC differentiation profile of (A-C) nestin-positive progenitor cells, (D-F) BetaIII-positive neurons , (G-H) CC1-positive oligodendrocytes, and (J-L) GFAP-positive astrocytes. Scale bar represents 50 μm. M) Quantification of NSPC differentiation profile for the various treatment groups. Mean ± standard deviation are plotted, n= 3 to 5; significant differences noted with an asterisks, p<0.05. N) Deconvoluted confocal image of betaIII-positive NSPC- derived neurons (arrows) 6 weeks post-transplantation. Scale bar represents 50 μm.

Differentiation of NSPCs after two weeks was examined and the results are summarized in Figure 4.4. All four groups expressed low levels of nestin showing that few NSPCs remained as uncommitted progenitors. BetaIII-tubulin staining show about 15% neuronal differentiation for NSPCs not exposed to dbcAMP (both 1 div and 4 div). NSPCs pre-treated with dbcAMP resulted in 37.0 ± 4.1% neurons after two weeks, similar to time-of-transplant values. Treatment with dbcAMP microspheres resulted in an intermediate percentage of neurons, 26.7 ± 13.9%. CC1-staining for oligodendrocytes was highest for untreated NSPCs (27.9 ± 11.7% and 33.8 ± 16.5% for 1 div and 4 div, respectively) and lowest for dbcAMP pre-treated NSPCs (3.9 ± 3.8%). Similarly, GFAP-staining for astrocytes was also highest for untreated NSPCs (30.3 ± 12.8% and 29.3 ± 10.3% for 1 div and 4 div, respectively) and lowest for dbcAMP pre-treated NSPCs (3.5 ± 3.6%). DbcAMP-microsphere treated NSPCs resulted in intermediate percentages of oligodendrocytes and astrocytes (10.9 ± 13.2% and 17.7 ± 11.6%, respectively) which were not significantly different from either untreated or pre-treated groups. Staining for mature neuronal markers MAP2 and NeuN were negative across all groups. Transplanted NSPCs were also negative for the radial glia marker 3CB2.

At six weeks, despite similar cell survival numbers, the majority of transplanted NSPCs had lost expression of the typical CNS phenotypic markers used above, independent of treatment. Staining with an antibody against the GFP-antigen resulted in positive staining (data not shown), indicating that tissue preparation was not an issue. Large populations of unclassified NSPCs have been reported previously [5,10]. It is interesting to note that, particularly in the dbcAMP- pre-treated animals, the surviving NSPCs seemed to form continuous networks across the bridge, with highly ordered longitudinal orientation (Figure 4.3C,D). Cell morphology suggests a more mature phenotype with elongated cell bodies and processes. A small subset of the dbcAMP pre-

74 treated cells were betaIII-tubulin positive at six weeks (Figure 4.4N), although staining with mature neuronal markers MAP2 and NeuN were largely negative.

Figure 4.5: A) Representative image of endogenous axonal regeneration into the tissue bridge based on betaIII tubulin staining. B) Evidence of association between betaIII- positive endogenous axons with surviving GFP-positive NSPCs at six weeks. Synaptophysin staining is observed at the interface (inset). C) RECA1 staining for endothelial cells show blood vessel formation throughout the tissue bridge at 2 weeks. D) Prolyl-4-hydroxylase (rPH) staining of bridge tissue indicates that the majority of cells are collagen producing fibroblasts.

Axonal regeneration across the bridge was also investigated. Interestingly, all treatment groups at six weeks resulted in numerous betaIII tubulin-positive axons penetrating the bridge from the rostral stump but stopping as they approached the caudal stump, as illustrated in Figure 4.5A. These axons were not GFP-positive, indicating that they originated from the host. Double labelling of betaIII-tubulin with synaptophysin, a neural synaptic marker, shows that there is

75 some association between the endogenous axons and the transplanted GFP-positive cells (Figure 4.5B). Also present in the bridge are RECA1 positive endothelial cells, which form discrete blood vessels as early as two-weeks (Figure 4.5C). No apparent association was found between blood vessel formation and treatment, or blood vessel vicinity to surviving NSPCs. Apart from the GFP-positive transplanted NSPCs, the tissue bridge was composed mainly of prolyl-4- hydroxylase-positive, collagen-producing fibroblasts at 6 weeks (Figure 4.5D). GFP-positive transplanted NSPCs were negative for the fibroblast marker.

4.5 Discussion

In this study, we aimed to understand the role of prolonged dbcAMP exposure on neuronal differentiation and the impact of pre-differentiated vs. in situ differentiated neurons on their survival and integration in vivo. By using the fully transected spinal cord injury model, we could clearly follow the effect of differentiated stem cells on tissue regeneration and locomotor functional recovery. The entubulation strategy, with which we have significant experience [4,5,49,75], is most suitable for these full-transection studies. The incorporation of dbcAMP- releasing microspheres in the tube walls allowed us to investigate the timing of differentiation on survival.

Interestingly, our in vitro studies showed that enhanced neuronal differentiation was obtained only with prolonged dbcAMP exposure and thus necessitated the inclusion of a prolonged release strategy in the chitosan tubular construct design. PLGA microspheres were fabricated and tailored for short-term release of dbcAMP and embedded into chitosan channels. In order to facilitate 7 d in vitro cultures, NSPCs were encapsulated in fibrin scaffolds that were incorporated into the channels. While previous studies have had NSPCs seeded along the interior walls of the channels [5,75], this was impractical for prolonged in vitro culture where cell proliferation resulted in cells dislodging from the channel surface. Fibrin scaffolds not only provided better entrapment of NSPCs, but also resulted in improved consistency in cell number and distribution. Fibrin scaffolds form quickly from fibrinogen and thrombin, and have been shown to be safe for cells in culture [116] and in vivo [109]. After spinal cord injury, fibrin scaffolds have been shown to attenuate glial scar formation [45]. Moreover, we have previously shown that guidance channels filled with fibrin matrix promote axonal regeneration and

76 functional recovery after spinal cord transection [49]. The cell differentiation data of NSPCs cultured in fibrin-filled, microsphere-loaded channels for 7 days in vitro confirmed that NSPCs are still dbcAMP-responsive in 3D fibrin scaffolds. More importantly, the NSPCs preferentially differentiated into neurons when cultured in channels where microspheres provided the only source of dbcAMP, validating microsphere-loaded channels as an effective drug delivery system in vitro. Although the neuronal differentiation percentage with dbcAMP on fibrin scaffolds was not as high as that on 2D chitosan films (~50% vs. ~80%), the effect was still a marked increase over non-dbcAMP treated cells (7%). It is possible that a higher target does of dbcAMP may compensate for this decrease in neuronal differentiation, but this was not investigated. Cell fate determination is a complex process that not only depends on presence of soluble factors and time, but is also influenced by surface chemistry/adhesion [260] and mechanical properties [258,261], so it was important to confirm that NSPCs would respond to dbcAMP in the three- dimensional fibrin matrix.

To better understand the benefit of pre-differentiated neurons vs. in situ-differentiated neurons (via dbcAMP-releasing microspheres), transplant survival and integration were studied in vivo in a fully-transected SCI rat model. Untreated NSPCs served as controls. Unexpectedly, NSPCs pre-differentiated with dbcAMP prior to transplantation resulted in a striking increase in cell survival of 80% whereas both untreated and dbcAMP-microsphere-treated NSPCs had survival rates of approximately 15%. There were no significant changes in survival of transplanted NSPCs between two and six weeks within a given treatment. Paired with the low level of staining with Ki67 proliferative marker, this suggests that the NSPC population had stabilized between these two timepoints. The enhanced survival effect associated with the dbcAMP 4 div group is attributed directly to pre-treatment with dbcAMP, as this was not observed with in situ dbcAMP delivery (dbcAMP-MS 1 div) nor in matching untreated controls (untreated 4 div). Therefore variables that may be associated with the extended in vitro incubation time prior to transplantation such as scaffold integrity or cell distribution were not significant factors.

Survival effects have been previously associated with dbcAMP pre-treatment, which can provide neuroprotection against excitotoxicity [136]. Rolipram, a phophodiesterase-IV inhibitor that works primarily by increasing intracellular cAMP [262], has been shown to be neuroprotective in the acute phase of SCI [263,264]. Moreover, downstream pathways of cyclic-AMP are important for the survival of newly generated neurons during development [147]. Recently,

Nout et al. used combined rolipram and dbcAMP drug treatment with glial restricted precursor cell transplants and observed reduced transplant survival with dbcAMP treatment [265]. However, the concentration of dbcAMP used in that study was 50 mM. Concentrations of 1 mM, as used in this study, are well-tolerated in the spinal cord whereas high doses can be damaging [266]. Rolipram-based elevation of cAMP has been shown to increase survival of transplanted olfactory ensheathing cells [264], and in combination with Schwann cell grafts, was implicated in axonal sparing [143]. The present study shows no effect on transplant survival with in situ dbcAMP delivery compared to no dbcAMP treatment, suggesting that enhanced transplant survival resulted from either a priming or pre-differentiation effect.

Pre-differentiation or pre-commitment of stem cells down a specific lineage may account for greater survival. Tarasenko et al. reported marginally higher survival when neuronally committed progenitor cells versus undifferentiated cells were transplanted in the injured CNS [267]. Olstorn et al. transplanted two preparations of human NSPCs in ischemic brain, demonstrating that both undifferentiated and neuronally-committed NSPCs were able to migrate towards the lesion site [268]. However, no differences in survival between the two preparations were reported. Davies et al. pre-treated glial restricted precursors with bone morphogenic protein-4 (BMP4) to derive immature astrocytes and explored the differences between the two populations after transplantation into lesioned rat spinal cord [269]. They found that pre- differentiated astrocytes better integrated with host regenerating tissue by supporting axonal regeneration and reducing astrogliosis. Interestingly, this group found that pre-differentiated astrocytes from ciliary neurotrophic factor (CNTF)-treated precursors had a detrimental effect [270] versus the beneficial effect observed with BMP4, indicating that subtle differences in the pre-differentiated progeny can significantly alter in vivo behaviour. Committed oligodendrocytes also performed better than undifferentiated NSPCs, showing better ability to remyelinate denuded spinal cord axons [271]. However, pre-differentiation of NSPCs to neurons is not always beneficial. For example, neuronal pre-differentiation resulted in poorer engraftment and survival than that achieved with undifferentiated NSPCs in hippocamppal transplant studies for Parkinson’s disease [272,273]. The extent of differentiation or perhaps maturation down a specific lineage may be of great importance. The 4 div with dbcAMP exposure ensured differentiation down the neuronal lineage; however, the differentiated progeny are unlikely fully mature, thereby promoting their survival and integration with the host tissue.

This committed, but not mature, phenotype for enhanced survival in vivo may allow greater integration.

Pre-treatment of dbcAMP resulted in the highest number of betaIII-tubulin positive neurons after two weeks in vivo, while the lowest percentage of neurons was associated with cells not exposed to dbcAMP, whether cultured for 1 or 4 div prior to transplant. The increased percentage of neurons in dbcAMP pre-treated NSPCs came at the expense of oligodendrocyte and astrocyte differentiation, which were both higher in untreated NSPCs. Although not statistically significant, in situ delivery of dbcAMP via microspheres consistently resulted in differentiation values that were between dbcAMP pre-treated and untreated groups. This suggests that dbcAMP-microspheres were likely able to influence NSPC fate in vivo, but that certain parameters of the delivery might be lacking, for example drug concentration or length of administration. Indeed, release parameters of PLGA microspheres can differ between an in vitro and in vivo environment. Using radio-labelling, de Boer et al. showed linear but more rapid release when comparing in vitro and in vivo release profiles of nerve growth factor from PLGA microspheres [274]. Technical restraints limited that study to a subcutaneous model, and indeed even more rapid degradation and release would be expected in injured tissue where acidic pH would cause accelerated degradation of the PLGA.

The tissue bridge had a high density of cells, the majority of which were collagen-producing fibroblasts, likely of meningeal origin from the disrupted pia, arachnoid, and dura. Angiogenesis was apparent as early as two weeks throughout the bridge, as evidenced by RECA1-postive staining of blood vessels. Notably, endogenous axonal regeneration into the bridge was observed in all treatment groups, and all showed initiation from the rostral stump of the spinal cord. However, the axons encountered a barrier at the caudal end of the bridge from the glial scar which prevented penetration into the caudal stump of the spinal cord.

Complete spinal cord transection is the most severe injury model that results in zero hindlimb movement immediately after injury. Typically rats will spontaneously recover only a very limited range of motion in the lower limbs over time. In this study, we showed that animals receiving dbcAMP pre-treated NSPCs showed a small but statistically significant (p<0.05) improvement in hindlimb function resulting in some movement of hindlimb joints. A strong histological basis for this improvement is not apparent, as endogenous axonal regeneration

79 across the tissue bridge did not appear to be qualitatively different among treatment groups. It is possible that some of the surviving NSPCs which spanned the entire length of the bridges, contributed to the observed enhanced functional recovery. Indeed, these transplanted cells appear to associated closely with each other to form a continuous network (Figure 4.3C,D). Directed differentiation of NSPCs into neurons has been reported by others to improve functional outcome after experimental SCI [30,275]. We did observe synapse formation between host axons and surviving NSPCs, which may underlie this minimal recovery if the axons from these surviving transplanted cells reached any of the motor neurons in the distal spinal cord. However, the apparent loss of neuronal phenotype of our transplanted NSPCs at six weeks suggests that a neuronal relay in the traditional sense may not be the mechanism responsible for the observed enhanced function.. Additional locomotor recovery would have likely continued in all groups at longer timepoints as BBB scores had not yet plateaued after six weeks. Thus, it is possible that dbcAMP pre-treatment does not enhance long-term functional outcome, but only accelerates the rate of recovery.

4.6 Conclusion

Cell transplant strategies offer great potential for replacing lost and damaged tissue. Drug treatments and biomaterial scaffolds further improve the efficacy of transplant cell survival and differentiation, as well as promote endogenous tissue regeneration. Our entubulation strategy combines these three aspects of treatment and has shown to be efficacious in promoting NSPC survival and host axonal regeneration.

5 Discussion 5.1 Chitosan Guidance Channels for SCI

Our previous work with guidance channels indicated that the choice of biomaterial impacts the quantity and quality of regeneration after spinal cord transection. First generation guidance channels used non-degradable synthetic polymers such as pHEMA-MMA. In these initial studies, pHEMA-MMA channels of varying mechanical strengths (elastic moduli of 177 and 311 kPa) successfully resulted in bridging the spinal cord transection [4]. The stronger channel resulted in greater bridge area and axonal growth, demonstrating the importance of the physical properties of the channel during regeneration. Stiffer, reinforced pHEMA-MMA channels were also tested in a separate study but were found to be potentially detrimental as syringomyelic cysts formed in the stumps [166]. Ultimately, pHEMA-MMA was deemed to be not suitable as channel material due to instances of channel collapse and calcification.

More recent work has shown chitosan (DD=90%) to be a suitable material for guidance channels in terms of mechanical strength and ability to promote tissue bridge formation. These studies were carried out to a maximum of 14 weeks, after which chitosan channels remained structurally intact with no overt signs of degradation [5]. Due to the multi-component nature of these studies, which included the use of NSPCs [5,75], as well as the complexity of the host tissue reaction in a full transection model, it was difficult to isolate and characterize the inflammatory and immune responses to chitosan itself in these initial chitosan channel studies.

The desire to characterize the long-term biocompatibility and degradation of chitosan prompted the investigation described in Chapter 2. In this context, biocompatibility was defined as not eliciting a chronic inflammation or immune system reaction. Chitosan channels were prepared with lower degrees of deacetylation (78% and 85%), as it was already established that 90% DD chitosan channels did not degrade appreciably after 14 weeks. Evidence from the literature also suggested that chitosan with DD greater than 85-90% are minimally degrading in the body [86], and this was also confirmed in our in vitro degradation testing of chitosan in the presence of 1 mg/ml lysozyme (Appendix Figure B.1). Sheets were of 78% DD and 85% DD chitosan were cut from channels to ensure preparation methods of chitosan remained consistent with earlier studies. This preparation resulted in pure compositions of chitosan (ie. no crosslinking agents)

81 that were densely packed and non-porous. The 78% DD chitosan was determined to be the lowest DD where channels were deemed to possess enough structural integrity for practical use. As described in Chapter 2, it was concluded that 85% DD chitosan was a relatively inert material which elicited neither significant acute inflammatory nor immune responses, nor any measurable chronic responses. The 78% DD resulted in a greater fibrous encapsulation response compared to 85% DD when implanted directly into the spinal cord, but also did not elicit a chronic host response. In both cases, chemical evidence of degradation of chitosan was observed between one, six, and twelve months. However, chain degradation of chitosan did not appear to result in mass loss or structural compromise of the implants, indicating that the degraded chains were yet unable to leave the bulk material. It was not established whether this was due to strong interchain interactions (eg. hydrogen bonds) or to the degraded chains not yet having a low enough molar mass to diffuse away.

It should be noted that degradation of these channels, particularly at early time points, is not necessarily desired. Degradation, in fact, may be undesirable in guidance channels if it results in mechanical failure causing injury to the cord. Ideally, the channel would undergo gradual degradation that does not result in collapse of the tubular structure. Thus, the slow degradation of chitosan, which is not likely to substantially compromise the structural integrity of the channels within one year, is a potentially positive outcome. Indeed, Li et al. investigated the use of collagen-filled chitosan tubes (85% DD) in a rat spinal cord lesion resulted in significant enhancement of axonal regeneration and functional recovery, with BBB scores continuing to rise even after one year [68]. In that study, the chitosan tube was still intact at one year, with the authors commenting that complete degradation of the tube does indeed occur by one and a half years. Long-term studies (eg. one or two years) will be necessary to assess if and how degradation affects channel integrity over time. Overall, chitosan was shown to be compatible with CNS tissue, and is a promising biomaterial for spinal cord applications.

5.2 Drug Delivery in PLGA Microspheres and Microsphere- Loaded Channels

The incorporation of drugs into the channels allows greater manipulation of the local environment during the regenerative process. In choosing a drug delivery system, several

82 criteria were considered. First, the system needed to be capable of sustained delivery over a tunable time window, ideally from weeks to months. Second, the system needed to be versatile in terms of the types of drugs it could deliver, whether they were small molecules or proteins, or hydrophobic or hydrophilic. Third, the system needed to be made of a biocompatible material, so as not to elicit or exacerbate local immune or inflammatory responses.

Several methods to incorporate drug delivery into guidance channel implants have been previously devised. Examples include direct addition of drug molecules into channel walls [215,217], incorporation of a degradable drug-releasing internal scaffold [104] or microspheres [163,164,219]. Of these systems, microsphere-based delivery was chosen based on its satisfaction of the above criteria. Moreover, the modularity of the microsphere system would allow for simultaneous delivery of multiple factors, each with its own tailored temporal release profile.

Microsphere drug release systems have been made from a range of different biomaterials including synthetic polymers, most commonly the polyesters (PLA, PGA, PCL and blends [158,159,160,276]) or natural polymers, most commonly chitosan [182,277,278,279]. The polyester PLGA was chosen for this work primarily due to its commercial availability, extensive historical use and research, established microsphere fabrication protocols, ability to encapsulate both small molecules and proteins, and tunable degradation kinetics. In particular, PLGA50:50 was chosen for its relatively fast degradation rates, which can vary from weeks to months primarily depending on molecular weight. These properties allowed the successful release of the bioactive protein alkaline phosphatase over a three month period, as well as bioactive release of the small molecule dibutyryl cyclic-AMP over ten days.

Guidance channels have previously incorporated PLGA microspheres for the release of nerve growth factor (NGF) [163] and epidermal growth factor (EGF) [164], demonstrating the ability of channels to effectively release bioactive proteins. However, these designs were not ideal for use in the spinal cord. The first design was based on pHEMA-MMA channels and incorporated PLGA microspheres directly in the channel walls during the polymerization reaction. Unfortunately, this methodology for embedding microspheres was not applicable to chitosan channels due to differences in fabrication procedure. The preparation of chitosan channels requires high temperature hydrolysis in concentrated base solution, which would destroy the

83 microspheres and its drug contents. Therefore, embedding microspheres directly into chitosan channel walls was not an option. Modifications required retrofitting pre-formed chitosan channels with microspheres. A sandwich design was proposed by Goraltchouk et al. [164] in a three layer design where microspheres were held in place between concentric inner chitosan channels and outer chitin channels. This design was based on the syneresis (expulsion of water from a gel) effect of chitin channels, which is used to entrap microspheres as the outer chitin jacket constricts upon drying. Despite demonstrating the bioactive release of EGF, this sandwich design has many disadvantages including inconsistent loading and potential preferential release of drug to the exterior of the implant, due to the higher swelling properties of chitin. Moreover, these triple-layered channels resulted in significant increase in wall thickness, which became a practical issue when implanting these constructs into the spinal canal of the vertebral column. Thicker walls result in greater flexion of the spinal cord during insertion into the channel, which can cause further damage.

Several alternative designs were attempted to incorporate PLGA microspheres into the chitosan channels. These included embedding microspheres into the chitin channels which could act as a sleeve around an inner chitosan channel, or casting microspheres into chitosan films which could be cut and inserted as an inner lining (Appendix Figure B.2). These designs were found to be unsatisfactory as they greatly diminished the bioactivity of the model protein alkaline phosphatase upon release. The method described in Chapter 3 was developed and used in subsequent studies. In this design, PLGA microspheres were suspended in dilute chitosan solution and placed into already prepared chitosan channels. These channels were rotated at high speeds (2500 rpm) along its longitudinal axis, forcing the chitosan/microsphere solution to coat the inner lumen. Upon drying, this created the stable embedding of microspheres into a secondary chitosan layer along the inner channel wall. This process was relatively quick (approximately 30 min), promoted preferential drug delivery to the inner lumen, and only minimally increased the wall thickness of the implant.

The bioactive release of the protein alkaline phosphatase was demonstrated over 90 days from both microspheres and microsphere-loaded channels. The release profiles from microspheres and microsphere–loaded channels were nearly identical. In contrast, significant differences in release profiles for dbcAMP were observed between free microspheres versus microsphere- loaded channels. The release of dbcAMP from channels occurred over a 5 day period, much

84 more rapidly compared to free microspheres which occurred over 14 days. The key difference between the dbcAMP and alkaline phosphatase microspheres was the use of lower molecular weight PLGA. Since dbcAMP required a shorter release window, a faster degrading PLGA was used. However, the process of embedding the microspheres into the channels involves short- term exposure to slightly acidic conditions. This process did not greatly affect the release of alkaline phosphatase, where PLGA with a Mw of approximately 15 kDa was used. For dbcAMP microspheres, PLGA with the lower Mw of around 5 kDa was used. These microspheres were less resistant to acidic conditions during channel embedding, resulting in the deviation of release curves when comparing free-floating microspheres to microsphere-loaded channels. Moreover, dbcAMP is a small molecule whereas alkaline phosphatase is a large protein, so diffusion out of the microspheres is much more easily facilitated by degradation.

For both alkaline phosphatase and dbcAMP PLGA microspheres and microsphere-loaded channels, the released factor was shown to be bioactive. In the case of dbcAMP, this was anticipated due to the relative stability of the small molecule. The bioactivity of dbcAMP released from microspheres was confirmed by cell culture assays that tested its ability to differentiate NSPCs into neurons (Appendix Figure B.3). Alkaline phosphatase, on the other hand, is a large protein which is generally more sensitive to denaturation compared to small molecules or peptides which do not depend on tertiary and quaternary (ie. dimerization) structure for activity. Alkaline phosphatase activity was assayed for its ability to dephosphorylate the substrate para-nitrophenylphosphate (pNPP). Only a moderate decrease in bioactivity was observed for alkaline phosphatase released from microspheres versus microsphere-loaded channels, 88% and 79% respectively. This retention of bioactivity may not be applicable to all protein-based drugs however, as different proteins may be more susceptible to denaturation. Our system of PLGA microspheres in chitosan channels is best suited for factors that are stable in acidic conditions, which becomes increasingly present during PLGA degradation.

5.3 Fibrin Scaffolds for Cell Delivery

Guidance channels present an opportunity to include matrix-filled interiors. Lumen-filling matrices provide additional surface area for tissue in-growth or for containing cell transplants. These matrices can be aligned fiber networks or random porous structures. In SCI guidance

85 channel studies, interior scaffolds have shown benefits in promoting endogenous axonal regeneration into the bridge compared to empty channels [49,68]. Tsai et al. tested a several lumen-filling matrices and observed differences in the quality of axonal regeneration depending on the biomaterial used. For example, rostral-caudal orientation of regenerating fibers was observed with methylcellulose whereas with collagen it was more disorganized [49]. Fibrin performed very well in the Tsai study as it promoted a significantly higher number of regenerating motor axons that could be traced back to the brainstem compared to the other materials [49]. Fibrin is a natural protein-based polymer that is important in wound healing. As such, it is biodegradable and biocompatible. In guidance channel studies for peripheral nerve regeneration, internal fibrin matrices promote early-stage bridging by promoting Schwann cell migration, resulting in increased rate of regeneration and enhanced functional recovery in long- gap injuries [280,281]. In the spinal cord, fibrin matrices have been shown to inhibit glial scarring and promote axonal outgrowth [45,108,109].

In the present work, fibrin scaffolds were used primarily as the cell delivery vehicle. In our previous work, NSPCs were seeded by coating chitosan channels with laminin then allowing neurospheres to adhere to the inner lumen. This resulted in relatively weak adhesion and uneven distribution along the channels. This protocol was also not suitable for tissue culture studies, as the neurospheres tended to dissociate from the channel after 2 to 3 days due to proliferation. For these reasons, fibrin matrices were investigated as a means of entrapping NSPCs to improve feasibility of one-week culture studies, as well as improving consistency in cell number and distribution during or prior to transplantation.

Another advantage of fibrin is its ability to form hydrogel networks from solution without the use of toxic chemical crosslinkers or organic solvents. Distributed cell seeding within three- dimensional scaffolds is not a trivial matter, and this property of fibrin was a particularly appealing feature permitting polymerization of the fibrin scaffold in the presence of cells. We demonstrated that this process was not harmful, and that the fibrin scaffolds supported NSPC viability and proliferation over one week in culture (Appendix Figure B.4). Indeed, fibrin scaffolds have been previously demonstrated by others for cell encapsulation of neural progenitor cells [115,119] and to evenly distribute MSCs within peripheral nerve guidance channels [282].

Another rationale for the use of fibrin was for potential effects on transplant cell survival in vivo. Although this effect was never directly tested in this work, there is a prevailing notion that cell survival of exogenous transplants may be improved if attached to a substrate. The use of biomaterials has been shown to improve the survival of exogenous cells versus direct injection in CNS tissue [77,78,79,283,284,285]. This was also the case with fibrin, as Itosaka et al. demonstrated greater BMSC survival and migration after SCI when transplanted in fibrin scaffolds compared to direct injection [76]. However, this may not be universal for all cell types or conditions. For example, cell survival was neither impacted positively nor negatively when fibrin was used as a delivery vehicle for ES cell-derived neural progenitor cells transplanted as embryoid bodies into the injured spinal cord [119].

As stated earlier, it is difficult to conclude the effect of fibrin on NSPC survival, as all groups in the animal study contained fibrin. We found approximately 10% survival of NSPCs seeded in fibrin scaffolds in the absence of dbcAMP treatment. In contrast, the rate of survival of dissociated NSPCs injected directly into the injured cord was about 3% at 2 weeks [73], but other differences in study design (eg. presence of channel, injury model, cell density) preclude meaningful comparison. Likewise, our previous experience of seeding NSPCs into chitosan channels without the use of fibrin resulting in much higher viability, greater than 100% [5,75], but key study design differences also preclude direct comparison. Specifically, the prior studies transplanted 1.5 to 3 million NSPCs as neurospheres, whereas the present study transplanted less than 100 K NSPCs as dissociated cells. Cells have previously been shown to survive much better when transplanted as neurospheres compared to single cells [73], and high cell density may be a positive influence due to paracrine trophic factor support. We chose to use dissociated cells in the present study for consistency with in vitro studies where the use of dissociated cells was more logical for differentiation studies. This was also the reason for the comparatively low cell density, as high cell densities often result in the aggregation of cells and the reforming of neurospheres. Regardless of whether or not the presence of fibrin enhances cell survival in vivo, fibrin was shown to support cell viability and differentiation in vitro, and served as a practical material for cell delivery.

5.4 Dibutyryl Cyclic-AMP and NSPCs

Dibutyryl cyclic-AMP was shown to promote the differentiation of neurons in a dose-dependent manner. This effect was quite dramatic, up to 80% with 1 mM dbcAMP over 7 days, especially given the low percentage of neurons (<10%) typically generated from NSPCs in control differentiation conditions. This effect of dbcAMP on adult brain-derived NSPCs [18] and other similar populations [144,145,253] has been previously described. In a hybrid mouse neuroblastoma and rat glioma cell line, Tojima et al. showed that cAMP-mediated neuronal differentiation occurs via the activation of protein kinase A (PKA) [146]. This PKA-dependent pathway was also shown to be important for differentiation of human blastoma cells [145] and rat hippocampal progenitor cells [144]. These studies show that PKA ultimately results in the activation of CREB, an important transcription factor traditionally associated with neurogenesis and maturation [147,148]. Indeed, dbcAMP or other molecules intended to increase intracellular cAMP have been widely investigated to promote axonal outgrowth [140,141,143].

The present work aimed to investigate the use of dbcAMP as an agent for promoting neuronal differentiation of adult NSPCs to: 1) demonstrate the in vitro and in vivo capabilities of microsphere-loaded channels as an effective means of local and sustained drug delivery (discussed in section 5.2); and 2) examine the role of directed neuronal differentiation on NSPC survival and integration into the injured spinal cord. The first step however, was to examine the suitability of dbcAMP as a drug candidate.

In vitro cell culture investigations confirmed the neuron-promoting ability of dbcAMP on adult rat brain-derived NSPCs as assessed by Zahir et al. [18]. Importantly, it was shown in the present work that dbcAMP-mediated differentiation can overcome competing signals present in 1% FBS, whereas in the Zahir study, dbcAMP was the sole differentiation factor in the media. This is particularly important to demonstrate for translation into animal studies, where the local environment is less-controlled and the ability of a factor to influence differentiation must show some ability to dominate cell fate decisions. Another important finding was that dbcAMP- mediated differentiation required greater than overnight (16 h) exposure for effect. The requirement of prolonged exposure was advantageous as it served as a validation criterion for sustained release of the drug delivery system.

This work also provided evidence to suggest that dbcAMP directly promotes neuronal differentiation of NSPCs, as opposed to preferential expansion of a neuronally-committed precursors or inhibition/selective toxicity to glial precursors. This claim arose from population characterization of NSPCs undergoing dbcAMP-mediated differentiation over time. Specifically, it was shown that early-stage (day 3) treatment with dbcAMP resulted in increased betaIII-tubulin-positive neurons primarily at the expense of nestin-positive progenitors. The presence or absence of dbcAMP did not affect total numbers of oligodendrocytes, astrocytes, or total cell number at 3 days. Furthermore, total numbers of oligodendrocytes and astrocytes did not decrease between 3 and 7 days in the presence of dbcAMP, suggesting that dbcAMP is not selectively toxic to glial populations. The presence of dbcAMP did significantly affect proliferation, however, as total cell numbers for dbcAMP-treated NSPCs did not change between 3 and 7 days, whereas proliferation was apparent in 1% FBS controls. Cessation of cell proliferation at 3 days was confirmed by Ki67 staining for mitotically active cells.

Interestingly, dbcAMP-mediated neuronal differentiation varied greatly depending on culture conditions, specifically during plating. In the studies described in Chapter 4, the culture protocol involved plating cells in proliferation/passaging media (contains EGF, FGF, and heparin, with or without dbcAMP) overnight prior to switching to differentiation media (1% FBS, no mitogens, with or without dbcAMP). In this ‘priming’ protocol of overnight plating in proliferation media, the addition of dbcAMP to the media resulted in greatly enhanced neuronal differentiation over time and was used for the remainder of the studies. By contrast, if NSPCs were plated overnight in media containing 1% FBS instead of proliferation media, the effects of dbcAMP were diminished greatly. The comparison of NSPC differentiation using ‘primed’ versus ‘unprimed’ plating protocols is shown in Appendix Figure B.5. This varied response based on a subtle change in the protocol was unexpected, particularly given that the NSPCs are grown in passaging media prior to plating. Therefore, the local environment during dissociation and adhesion appears to greatly influence the differentiation potential of NSPCs, especially as it relates to responsiveness to dbcAMP. The mechanism for this differential response is yet unclear, but warrants future investigation. Nevertheless, it is an important reminder of the complexities involved in cellular signaling and phenotype determination.

Differentiation in vivo was also examined. Spinal cord transplanted NSPCs were either pre- treated with dbcAMP, exposed to in situ dbcAMP delivery, or left untreated. Pre-treatment with

89 dbcAMP resulted in approximately 40% betaIII-positive neurons prior to transplant and this percentage was maintained for 2 weeks post-transplantation, indicating that no further differentiation of NSPCs occurred in vivo. Microsphere-based delivery also increased the percentage of betaIII-positive neurons (27%) compared to controls (15%), although this was not statistically significant in our limited sample size.

The most striking and potentially important result to come from this work was the remarkable increase in cell survival observed when dbcAMP-treated NSPCs were transplanted into the injured spinal cord. Approximately 80% cell survival was associated with this treatment after two and six weeks post-transplant, a marked increase over the 10% survival rates observed in untreated NSPCs. Moreover, this effect on survival was not observed with in situ dbcAMP delivery, indicating that pre-treatment with dbcAMP greater than 24 h was a prerequisite for this effect. As discussed extensively in Chapter 4, enhanced survival could be attributed to a dbcAMP-specific priming effect on cell viability, or increased resiliency of neuronally- committed compared to undifferentiated NSPCs. The differentiation data suggests that enhanced transplant viability may primarily be a dbcAMP-mediated effect. The percentage of betaIII- positive neurons pre-transplant (40%) and 2 weeks post-transplant (37%) do not differ significantly, indicating that there is not preferential survival of this population. Indeed, upregulation of intracellular cAMP has been shown to increase cellular resistance to excitotoxicity [136]. The increased survival of NSPCs based on dbcAMP pre-treatment, quite unexpectedly, became the most compelling result from our in vivo studies and further examination of the mechanism is certainly warranted.

5.5 NSPC Transplant Optimization

The success of stem cell transplantation therapies depend on transplant viability, proper cell differentiation, and meaningful integration with the host tissue. Previous studies in our lab have shown cell viability of NSPCs is impacted by factors such as cell distribution (ie. dissociated versus neurospheres), transplant time after injury, and transplant location (ie. lesion epicenter versus rostral and/or caudal) [72,73]. Moreover, several studies have demonstrated the benefits of using biomaterial vehicles to improve transplant survival compared to direct injection with a media vehicle [76,283,284,285]. This was part of the rationale for using a fibrin matrix within

90 chitosan channels as a cell delivery vehicle, although this aspect of fibrin and chitosan was not explicitly characterized in this work.

Directed differentiation of NSPCs is also of interest, as typically very few neurons are generated after transplantation [27]. This is primarily due to the inhibitory nature of the adult spinal cord towards neurogenesis. The use of lineage-restricted precursors [248,250], and genetic manipulation [249] have been attempted to introduce new neurons into the spinal cord after injury without much success. Other protocols including neuronal pre-differentiation [267,268] and in situ drug delivery [119] have been more successful in generating and maintaining neuronal populations after transplantation into injured CNS tissue, and have been shown to positively impact transplant survival [119], migration [268], and functional behavior [267].

The present work aimed to further investigate neuronal differentiation of NSPCs in the spinal cord and its role in survival, integration with host tissue, and functional improvement. In particular, our study was designed to also elucidate the importance of timing of neuronal differentiation, ie. whether pre-differentiation or in situ differentiation via drug delivery would affect these outcome measures. We found that pre-differentiated NSPCs resulted in greatly enhanced survival. However, the mechanism for this pro-survival effect could not be directly attributed to the differentiation state of the NSPCs. Pre-treatment with dbcAMP, which was used in the pre-differentiation protocol, can itself promote survival by increasing resistance against excitotoxicity [136]. Cell migration did not seem to be affected by differentiation state or timing, as grafted cells were typically found at the center of the bridge (both radially and rostral- caudal) at 2 weeks, and migrated radially outwards by 6 weeks independent of treatment.

Optimization of transplant survival and migration is important, but ultimately the goal of cell transplantation studies is to promote proper integration with the host. The term host integration for NSPCs can be interpreted in many ways and is different for the various cell phenotypes. For example, in the case of oligodendrocytes it may be a measure of the quantity and quality of remyelination. In our study, integration with the host was assessed by synaptic marker staining and was observed occasionally only with neuronally pre-differentiated cells. The significantly poorer survival of NSPCs in the other treatment groups made it difficult to assess whether this was solely attributed to the pre-differentiation treatment. A more direct measure of neuronal integration could be electrophysiological measurements of signal propagation and continuity, but

91 great care would be needed to ensure that the transplanted cells are a part of the assayed pathway. Alternatively, transplanted cells may promote recovery indirectly by supporting endogenous regeneration and plasticity. In the present work, endogenous axonal growth into the tissue bridges was observed to be extensive in all NSPC treatment groups, although quantitative measurements were not performed.

To date, strategies to augment stem cell transplantation have generally been on focused on improving stem cell survival and differentiation, whether through biomaterials or drug delivery. However, the ultimate aim of functional recovery will also require optimization of transplant integration with host tissue. For SCI, this will likely require promoting plasticity and formation of proper signaling pathways. It is clear that multidisciplinary approaches will be necessary to realize the full potential of cell replacement therapies.

5.6 Translational Considerations

5.6.1 Drug Delivery

Drug treatment is conceptually one of the simplest methods to influence the local environment after spinal cord injury, whether by remedying the local toxic environment (eg. free radicals), attenuating the inflammation and immune responses, promoting axonal growth and plasticity, or neutralizing growth inhibitory substrates such as myelin or the glial scar. However, in many cases drug delivery requires local and prolonged release to be effective. Repeated local bolus injections are not practical in CNS tissue, and systemic delivery is often not a viable option due to the blood-brain barrier (BBB). Many relevant factors have multiple effects, and thus their systemic delivery would likely result in undesirable side effects in non-target tissue. Catheter and minipump systems have been used by some groups to stimulate endogenous or transplanted stem cells [286]; however, these systems are under scrutiny for use in spinal cord injury due to concerns regarding safety, tissue damage and risk of infection [212,213].

This work utilized drug release via PLGA microspheres. PLGA microsphere drug delivery systems are in use clinically for in cardiovascular, cancer, and immunology applications (reviewed in [287]), but have not yet been clinically applied to spinal cord injury. Animal model research suggests that PLGA microspheres applied adjacent to or directly into the CNS is well-

92 tolerated [128,129]. Indeed, this present work showed no overt detrimental effect of PLGA or the related material Vicryl in spinal cord tissue. Stem cell transplant viability, tissue bridge formation, and ingrowth of endogenous axons were all observed in the presence of PLGA microspheres, with or without encapsulated drug. This does not imply that PLGA is an ideal material for microsphere delivery however. PLGA microspheres can have detrimental interactions with proteins resulting in denaturation or aggregation [231], can harbor a local acidic environment during degradation [288], and are often difficult to tailor release profiles [289,290]. Other materials of potential interest for microsphere-based drug delivery include synthetic polyanhydrides, which are surface eroding and thus have more predictable release curves [291], and natural polymers such as chitosan [292,293], hyaluronan [294], or dextran [295] which may have less detrimental interaction with encapsulated proteins or surrounding tissue. The ideal choice for material is largely dependent on the specific application, with considerations including site of action, required time and shape of release, and properties of the drug.

The modular nature of microsphere delivery systems, whether based on PLGA or not, allows them to be incorporated not only into defined implantable structures such guidance channels, but also into injectable hydrogels [126] or directly into tissue via media vehicle [129]. Therefore, tailored microsphere formulation for a particular drug shown to be effective in one study can be relatively easily adapted to work in another study or injury model by ensuring dosing and release rates are equivalent. In this case, the microsphere delivery of dbcAMP could be used to promote neuronal differentiation of transplanted NSPCs in a contusion model of SCI or in brain injury, or could potentially target differentiation of endogenous neural stem cell population.

5.6.2 Stem Cell Source

Despite the potential of stem cell therapy, there are several issues that must be considered as the field moves towards clinical applications. The first issue concerns stem cell source. Notwithstanding the ethical issues, embryonic stem cells are pluripotent, offering numerous advantages yet require careful purification strategies for differentiated progeny [296] and have the risk of tumor formation [297]. Significantly, human ES-derived oligodendrocyte progenitor cells are currently in clinical trials in North America for treatment of SCI [11], and positive safety data from these trials could accelerate future stem cell therapies. Indeed, retinal pigment epithelial (RPE) cells derived from embryonic stem cells have been recently approved for Phase

I/II clinical testing for age-related macular degeneration in the eye, stemming from pre-clinical data showing effectiveness in rodent [298]. This could potentially provide an alternative source to fetal-derived [299] or autologous [300,301] RPE transplants.

Autologous sources such as adult brain or spinal cord stem cells are impractical to biopsy, while others such as human skin-derived precursors, umbilical cord blood stem cells, or bone-marrow derived MSCs may not have the differentiation potential to repopulate CNS tissue. Induced pluriopotent stem cells are a promising source of patient-specific stem cells particularly in trauma injuries such as SCI, TBI, and stroke which are not genetic diseases. Patient-specific cell sources are ideal as it would reduce the likelihood of transplant rejection and obviate the need for immune-suppression. However, research into iPSCs is still relatively early and many of the concerns over ES cells exist with iPSCs.

The present work made use of homologous adult NSPCs in rat. In human, adult neural stem cells can be derived from organ donors or biopsy patients and have the advantages of being restricted to CNS cell lineages. Human adult NSPCs can be expanded long-term and are amenable to cryopreservation [302]. However, it is still not established whether relevant cell numbers can be generated from adult brain or spinal cord for clinical practice. Fetal neural stem cells may have greater expansion potential, and indeed clinical trials using human fetal brain- derived NSCs are in progress for the treatment of two separate fatal brain disorders in children, Pelizaeus-Merzbacher Disease and Neuronal Ceroid Lipofuscinosis (ie. Batten disease). These cells have also been approved for Phase I/II trials for chronic spinal cord injury in Europe. In the UK, trials of immortalized fetal brain-derived NSCs are underway for treatment of stroke, where preclinical data showed effectiveness in improving motor deficits in rat [303]. These studies will provide some indication as to the suitability of somatically-derived NSPCs as an allograft source, and whether immuo-suppression or graft rejection will be major concerns going forward.

5.6.3 Injury Model

An issue that must be considered going forward is the use of relevant injury models. In spinal cord injury, the most clinically relevant model is the contusion/compression model, which typically results in cavity formation in the center of the cord surrounded by a ring of spared tissue. However, many scaffold and guidance channel implantation studies, the present study included, rely on hemisection or transection models of SCI [5,50,77,175], in part due to

94 practicality of the implantation surgery. One can view these as proof-of-concept studies that help elucidate mechanisms or modes of regeneration that will inform future study design.

This work utilized the full transection model, as this model conceptually fits with a guidance channel approach. Transection models are useful as they provide unambiguous information about regeneration, as opposed to the more clinically relevant contusion/compression models where it is difficult to distinguish between neuroprotection and regeneration. However, full transection injuries are relatively rare in humans so this guidance channel approach may have limited translational value as a clinically feasible strategy. However, this approach may hold value in certain cases where there is high meningeal disruption. Indeed, our channels can be cut longitudinally and sutured in a sleeve-type implantation. Ultimately, accessibility to the injury site is a major concern, and minimally invasive strategies such as injectable or in situ gelling hydrogels will be desirable from a clinical perspective.

5.7 The Entubulation Strategy for SCI

The role of guidance channels in SCI is to provide an isolated environment in which new tissue growth can occur. While the clinical translation of such a strategy may be limited due to the rarity of complete transection injuries in humans, guidance channel studies are important for informing the field about important factors relating to the regenerative process. The basic guidance channel strategy has been shown previously to limit scarring at the spinal cord transection interface, encourage new tissue growth, and promote host axonal regeneration to various degrees [4,5,49]. The advantage of guidance channel studies lies in their ability to influence the local environment, whether through the choice of the channel material and various physical designs such as internal matrices or fiber networks, cellular transplantation, or drug delivery. The motivation of the present work was to develop a system that would permit the study of the combined action of cellular and molecular therapy within the injured spinal cord environment.

To this end, a microsphere-based drug delivery system was incorporated into the chitosan channel walls to provide a means of local and sustained release of molecules. Also, fibrin scaffolds were utilized to seed and distribute adult brain-derived NSPCs throughout the channel

95 interior. As a model for demonstrating efficacy of this system, dbcAMP was chosen for its effect on promoting neuronal differentiation of NSPCs, providing a defined endpoint for evaluating the drug delivery system. Moreover, the directed differentiation of neurons was of particular interest in a complete transection injury model where all neuronal pathways are severed, prioritizing the need for neuronal replacement.

This combined effect of dbcAMP and NSPCs resulted in enhanced differentiation towards the neuronal phenotype and decreased percentages of oligodendrocytes and astrocytes when compared to control transplants untreated with dbcAMP. Moreover, NSPCs pre-treated with dbcAMP prior to transplantation resulted in much higher rates of cell survival. At six weeks, longitudinal sections showed mature morphology of NSPCs, which seemed to form a continuous network across the length of the tissue bridge. Synaptic interactions between transplanted NSPCs and host axons in the bridge were also observed, as was a measurable gain in functional recovery.

These cell and drug-delivery channels were shown to promote robust tissue bridges after rat spinal cord transection as early as 2 weeks post-transplant, independent of dbcAMP treatment. Previous experience with stem cell-seeded channels indicated that tissue bridges generally require 5 weeks before bridges are thick enough for practical histological assessment. This suggests that the internal fibrin matrix accelerated the bridge formation process, as is the case for fibrin scaffolds in peripheral nerve conduits [280]. These bridges at 2 weeks were also shown to contain numerous blood vessels, although the majority of bridge cells consisted of collagen- producing fibroblasts. These constituents of the tissue bridge are consistent with our previous studies [5]. By 6 weeks, all treatment groups also contained numerous endogenous long-tract axons within the bridge, originating from the rostral cord and terminating near the caudal bridge/cord interface (Appendix Figure B.6). This pattern of endogenous regeneration is observed after many growth-promoting treatments for SCI and is largely due to the glial scar or inhibitory myelin at the re-entry zone [60,304,305]. The extent of directed axonal regeneration found in this study was quite substantial. However, no qualitative differences were observed between treatments. Unfortunately, retrograde axonal labeling was not performed, which may have provided further insight into the mechanism of behavioural benefit. Anterograde tracing of these axons done in similar studies from our lab suggest that many of these regenerated axonal

96 tracts in the tissue bridge may originate from various brain stem nuclei important for motor function [4].

The entubulation strategy was initially developed as a method to treat transection injuries. Although promising results have been attained with basic guidance channels in this model, it has become evident that an injury of this severity is at present beyond our capacity as researchers to overcome. The present research aimed to improve our ability to control the local environment by developing the entubulation strategy from a basic hydrogel channel to a device capable of serving as a vehicle for stem cell therapy and local and sustained drug delivery. This system was demonstrated here to successfully influence stem cell fate (ie. survival and differentiation) through treatment with the small molecule drug dbcAMP.

The capability of guidance channels to serve as a vehicle for stem cell transplants and drug delivery allows for a multifaceted and versatile approach towards discovering effective combination therapies to optimize the regenerative response. While it may be naïve to suggest that our strategy will soon fully restore function after complete spinal cord transection, these studies can provide important insights that can contribute to the development of effective treatments in the future.

6 Conclusions 6.1 Achievement of Objectives

The original hypothesis for this work was:

Several objectives were set in order to assess this hypothesis and are revisited here with a summary of the work presented to meet these objectives.

1. Assess the long-term biocompatibility of chitosan channels. In the context of this study, biocompatibility refered to minimal long-term effects on chronic inflammation or immune system reaction. Chitosan channels of varying degrees of deacetylation were prepared and investigated in both non-injured and injured spinal cord models for up to one year. The biocompatibility of chitosan was compared to two commercially, clinically used materials: degradable polyester Vicryl; and inert ePTFE-based Gore-Tex. Chitosan demonstrated excellent biocompatibility comparable to Gore-Tex, resulting in minimal attraction of phagocytic cells and no chronic immune or inflammatory response in either model. Chitosan sheets showed no macroscopic signs of degradation after one year in vivo, as assessed by SEM and mass loss measurement. However, evidence of degradation was observed at the chemical level based on differential staining with hematoxylin and eosin. These data are presented in Chapter 2.

2. Design a drug delivery vehicle capable of sustained release within chitosan channels. A method for incorporating microspheres into channels was developed whereby they are embedded within a thin chitosan layer on the interior surface of the channel. This process resulted in minimal alteration in channel dimensions, most notably wall thickness, which is an important consideration for practical use. Sustained bioactive release of the model protein alkaline phosphatase was demonstrated from this system over a three month period. This work was described in Chapter 3.

3. Evaluate the effects of directed neuronal differentiation caused by the bioactive factor on transplanted neural/stem progenitor cell fate. The small molecule drug dbcAMP was chosen as a candidate factor based on its ability to promote neuronal differentiation of NSPCs. DbcAMP was encapsulated into PLGA microspheres and tailored for short term release within microsphere-loaded channels. Microsphere-loaded channels containing dbcAMP were able to enhance neuronal differentiation of NSPCs seeded within a fibrin matrix in vitro.

In vivo studies investigated the effect of directed neuronal differentiation of NSPCs, comparing untreated NSPCs, NSPCs with dbcAMP-microspheres (in situ differentiated), and NSPCs pre- differentiated with dbcAMP prior to transplantation. In situ delivery of dbcAMP via microsphere-loaded channels resulted in increased neuronal differentiation and decreased oligodendrocyte and astrocyte differentiation compared to untreated NSPCs. Although statistical significance was not achieved, this phenotypic profile followed the expected trend based on that observed for dbcAMP pre-differentiated cells. Pre-differentiation with dbcAMP resulted in greatly enhanced survival of transplanted NSPCs. These cells engrafted well throughout the regenerated tissue bridge and showed some evidence of synaptic integration with host axons. This treatment also demonstrated modest but significant behavioural improvement in spinal cord injured rats after 6 weeks. This work is described in Chapter 4.

6.2 Major Contributions

In this thesis, it was demonstrated that chitosan is indeed a biocompatible, minimally degrading material in the spinal cord. Chitosan can be formulated to create guidance channels which serve to promote directed tissue regeneration following spinal cord injury. These channels were modified with the capability of local and sustained drug release through the incorporation of PLGA microspheres. These microsphere-loaded channels demonstrated the ability for both short-term bioactive release of the small molecule dbcAMP and the long-term bioactive release of the protein alkaline phosphatase.

Enhanced neuronal populations of adult rat brain-derived NSPCs was able to be attained with dbcAMP exposure. This effect was quite robust, resulting in approximately 80% betaIII-positive neurons after one week of culture, compared to less than 10% in control conditions. Timecourse characterization of dbcAMP-mediated neuronal differentiation suggests that dbcAMP acts as a direct promoter of neuronal differentiation as opposed to an agent that selectively enhances a neuronally-committed progenitor cell population or selectively inhibits a glial cell population.

The potential benefits of dbcAMP-mediated directed differentiation was also shown. NSPCs primed with dbcAMP for four days prior to transplant display a remarkable increase in survival in the typically hostile environment of acute SCI. This survival effect is not seen with in situ dbcAMP delivery, suggesting timing of treatment is a major factor. NSPCs were shown to engraft throughout the tissue bridge, and in the case of dbcAMP-treated groups, demonstrated mature morphology and network formation, interaction with host axons, and a measurable increase in functional recovery of hindlimb movement after SCI.

100

7 Recommendations for Future Work 7.1 Further Investigation dbcAMP-Mediated Survival

Pre-treatment of NSPCs with dbcAMP in vitro prior to transplantation resulted in greatly enhanced viability in vivo. This pre-treatment occurred over 4 days in vitro to achieve neuronal differentiation of the transplant population prior to implantation into the animal model. Interestingly, the enhanced survival effect was not observed with in vivo delivery of dbcAMP, indicating that the pre-treatment was necessary for the effect. Moreover, we controlled for the extended incubation time required for pre-differentiation, and it was found that survival was not impacted by changes in cell distribution, fibrin scaffold integrity, etc. associated with holding implants for 4 div versus 1 div.

Therefore, the enhanced survival effect was attributed to two potential explanations. First, the pre-treatment with dbcAMP could directly protect NSPCs during transplantation into the injury site. Increases in intracellular cAMP concentration have been shown to be neuroprotective against excitotoxicity in culture [136], and have resulted in increased transplant survival of other cell types [147,263,264]. Cell death of transplanted NSPCs occurs mainly in the first day [73], so it is reasonable that pre-treatment or ‘priming’ with dbcAMP would be necessary for this effect, and that in vivo delivery, as observed in our studies, would not protect the cells in time.

The alternative explanation is that the main factor for increased survival was due to the difference of transplanting undifferentiated NSPCs versus differentiated neurons. There have been several studies investigating the transplantation of cells at differing stages of maturity, but no consensus is evident on how this affects cell viability or resiliency. Indeed, it is certainly possible that the stage of maturity (ie. immature or terminally differentiated) as well as cell type (neuron, oligodendrocyte, or astrocyte) and sub-type (ie. GABAergic, glutamatergic, cholinergic, dopaminergic, etc) would be a major factor. The observation that survival did not affect the percentage of neurons pre- and post-transplant in the dbcAMP treated group suggests that there was no differential survival effect based on cell phenotype of the transplanted population. Thus, pre-differentiation may not be the dominant reason for the pro-survival effect. However, increasing cell survival is of universal importance, so further investigation is warranted.

101

To investigate whether dbcAMP-mediated neuroprotection is the dominant effect, a couple additional different studies are required. First, as discussed in Section 5.4, the neuronal differentiation effect of dbcAMP can be greatly attenuated if the initial seeding is performed in the absence of mitogens. If a 4 day pre-incubation time is maintained, the survival effects of the mitogens, present only for the first day, would be minimal. A decreased percentage of betaIII- positive neurons would need to be confirmed following this adapted protocol, and if successful, direct comparison dbcAMP-treated neuronally-enriched or normal populations could be evaluated for survival.

Alternatively, dbcAMP could be administered simultaneously with inhibitors of the intracellular downstream pathways that affect differentiation. Targets that are further down the relevant pathways would be recommended so as not to affect other potential signaling aspects of intracellular cAMP. For example, the cAMP pathway ultimately converges on the phosphorylation of the transcription factor CREB, which is responsible for gene regulation related to neuronal development [148]. CREB activity can be attenuated by upregulating protein phosphatase-1 (PP1) [306], which dephosphorylates CREB, or through RNA interference [307]. If these treatments can nullify the directed differentiation effect of dbcAMP, then co-treatment with dbcAMP prior to transplantation could be used to separate the effects of dbcAMP versus neuronal pre-differentiation.

The role of neuronal pre-differentiation could also be investigated by using dbcAMP- independent protocols. For instance, the cytokine interferon-γ (IFNγ) has also shown to greatly enrich neuronal differentiation of adult brain-derived NSPCs [18] and can be compared to pre- differentiated NSPCs treated with dbcAMP.

Independent of mechanism, it is recommended to try to translate dbcAMP-mediated transplant survival into other SCI models. For example, it would be worthwhile to see if exposing neurospheres to dbcAMP-containing media prior to transplant is effective at increasing survival in a traditional transplantation via direct injection. This could inform us whether a substrate such as fibrin is important.

102

7.2 Utilizing the Versatility of the Drug Delivery System

Microsphere-loaded channels are fabricated by mixing drug-loaded microspheres into a dilute chitosan solution, which is then coated onto the interior walls of chitosan channels. This process can be easily adapted by mixing different drug-loaded microsphere populations prior to embedding into the channels. This would allow for simultaneously incorporation of multiple drug factors, where each drug could be tailored for its own optimal temporal release profile. For instance, neuroprotective agents such as erythropoietin [47] can be delivered during the acute phase while a growth promoting molecule such as NT-3 [155] can be delivered over the long- term.

Modifications to the PLGA microspheres can also be made to tailor release profiles. For example, microspheres can be coated with a polyanhydride shell. Polyanhydrides degrade by surface erosion, resulting in an intial delay or lag in the release profile [308]. This could be beneficial for factors like chondroitinase ABC which break down the glial scar. Initial glial scar formation is important in reducing spread of the secondary injury [309] but ultimately becomes a barrier for regeneration.

This flexibility to be able to combine multiple drug therapies with individually-optimized release profiles makes microsphere-loaded chitosan channels an appealing tool for studying combination therapies in the injured spinal cord.

7.3 Directed Differentiation of Other Cell Types

The directed differentiation of NSPCs into neurons was investigated in this work, but it is also of interest to be able to enrich the population of oligodendrocytes or astrocytes. It is not yet known what cell type, or relative mixture of cell types, is the best combination for inducing regeneration and functional connections after SCI. It is likely as well that the optimal population may change based on injury level, type, or severity. Therefore, the ability to control NSPC fate determination is of great interest.

Several factors have been identified for directed differentiation of NSPCs. Enriched oligodendrocyte populations can be achieved through treatment with PDGF-AA [251], whereas

103 enriched astrocyte populations have been achieved with BMP4 [310,311]. These factors could be tested in both pre-differentiation and in situ delivery models. Moreover, it may be possible that relative dosing of molecules that enhance neurons, oligodendrocytes, and astrocytes will allow for manipulation of the relative population of these phenotypes formed. It may even be possible through in situ delivery via microspheres to control the temporal enhancement of these populations, for instance to promote neuronal differentiation early followed by glial differentiation, which may better mimic how CNS tissue is formed during development [26].

7.4 Application in Other SCI Models

The present work was done in the context of an acute full transection model of SCI where the implantation of the device occurred directly following the injury. This treatment paradigm was chosen to provide better comparison to previous studies done in our lab, and due to its relative convenience of only needing a single surgery. However, it is well established that the elapsed time after injury plays a large role in determining the make-up of the local milieu, particularly as it relates to inflammation and immune responses [28]. The time of treatment can therefore influence such factors as cell transplant viability and endogenous axonal response. Moreover, in any clinical situation, it would be impossible to apply a treatment at the time of injury. Standard surgical interventions such as decompression occur between hours and days, and most cell-based treatments in clinical trials are not administered until a week to months following initial injury [312]. Therefore, one recommendation for future work would be to test our treatment in subacute (7 to 14 days post-injury) or chronic models. Additionally, the guidance channel treatment can be miniaturized and implanted into the spinal cord in more clinically-relevant compression/contusion injury models [98,99]. It has been previously shown that implantation of channels into a spinal cord cavity does not itself negatively affect function [98]. Microsphere- loaded channels can be adapted for this intramedullary model of implantation. These studies would provide more meaningful data on the translation potential of this strategy.

104

References

1. One Degree of Separation: Paralysis and Injury in the United States. Christopher and Dana Reeve Foundation. 2010.

2. (2010) Spinal Cord Injury - Facts and Figures at a Glance [Internet]. Birmingham, Alabama: National Spinal Cord Injury Statistical Center. Available from: www.uab.edu/nscisc [cited February 2010].

3. David S, Aguayo AJ (1981) Axonal elongation into peripheral nervous system "bridges" after central nervous system injury in adult rats. Science 214: 931-933.

4. Tsai EC, Dalton PD, Shoichet MS, Tator CH (2004) Synthetic hydrogel guidance channels facilitate regeneration of adult rat brainstem motor axons after complete spinal cord transection. J Neurotrauma 21: 789-804.

5. Nomura H, Zahir T, Kim H, Katayama Y, Kulbatski I, et al. (2008) Extramedullary chitosan channels promote survival of transplanted neural stem and progenitor cells and create a tissue bridge after complete spinal cord transection. Tissue Eng Part A 14: 649-665.

6. Seif GI, Nomura H, Tator CH (2007) Retrograde axonal degeneration "dieback" in the corticospinal tract after transection injury of the rat spinal cord: a confocal microscopy study. J Neurotrauma 24: 1513-1528.

7. Gensel JC, Donnelly DJ, Popovich PG (2011) Spinal cord injury therapies in humans: an overview of current clinical trials and their potential effects on intrinsic CNS macrophages. Expert Opin Ther Targets.

8. Hawryluk GW, Rowland J, Kwon BK, Fehlings MG (2008) Protection and repair of the injured spinal cord: a review of completed, ongoing, and planned clinical trials for acute spinal cord injury. Neurosurg Focus 25: E14.

9. Kwon BK, Sekhon LH, Fehlings MG (2010) Emerging repair, regeneration, and translational research advances for spinal cord injury. Spine (Phila Pa 1976) 35: S263-270.

10. Karimi-Abdolrezaee S, Eftekharpour E, Wang J, Morshead CM, Fehlings MG (2006) Delayed transplantation of adult neural precursor cells promotes remyelination and functional neurological recovery after spinal cord injury. J Neurosci 26: 3377-3389.

11. Alper J (2009) Geron gets green light for human trial of ES cell-derived product. Nat Biotechnol 27: 213-214.

12. Mayor S (2010) First patient enters trial to test safety of stem cells in spinal injury. Bmj 341: c5724.

13. Keirstead HS, Nistor G, Bernal G, Totoiu M, Cloutier F, et al. (2005) Human embryonic stem cell-derived oligodendrocyte progenitor cell transplants remyelinate and restore locomotion after spinal cord injury. J Neurosci 25: 4694-4705.

105

14. Huebner EA, Strittmatter SM (2009) Axon regeneration in the peripheral and central nervous systems. Results Probl Cell Differ 48: 339-351.

15. Reier PJ (2004) Cellular transplantation strategies for spinal cord injury and translational neurobiology. NeuroRx 1: 424-451.

16. Enzmann GU, Benton RL, Talbott JF, Cao Q, Whittemore SR (2006) Functional considerations of stem cell transplantation therapy for spinal cord repair. J Neurotrauma 23: 479-495.

17. Garbossa D, Fontanella M, Fronda C, Benevello C, Muraca G, et al. (2006) New strategies for repairing the injured spinal cord: the role of stem cells. Neurol Res 28: 500-504.

18. Zahir T, Chen YF, MacDonald JF, Leipzig N, Tator CH, et al. (2009) Neural stem/progenitor cells differentiate in vitro to neurons by the combined action of dibutyryl cAMP and interferon-gamma. Stem Cells Dev 18: 1423-1432.

19. Reynolds BA, Weiss S (1992) Generation of neurons and astrocytes from isolated cells of the adult mammalian central nervous system. Science 255: 1707-1710.

20. Morshead CM, Reynolds BA, Craig CG, McBurney MW, Staines WA, et al. (1994) Neural stem cells in the adult mammalian forebrain: a relatively quiescent subpopulation of subependymal cells. Neuron 13: 1071-1082.

21. Palmer TD, Takahashi J, Gage FH (1997) The adult rat hippocampus contains primordial neural stem cells. Mol Cell Neurosci 8: 389-404.

22. Roy NS, Wang S, Jiang L, Kang J, Benraiss A, et al. (2000) In vitro neurogenesis by progenitor cells isolated from the adult human hippocampus. Nat Med 6: 271-277.

23. Weiss S, Dunne C, Hewson J, Wohl C, Wheatley M, et al. (1996) Multipotent CNS stem cells are present in the adult mammalian spinal cord and ventricular neuroaxis. J Neurosci 16: 7599-7609.

24. Tropepe V, Hitoshi S, Sirard C, Mak TW, Rossant J, et al. (2001) Direct neural fate specification from embryonic stem cells: a primitive mammalian neural stem cell stage acquired through a default mechanism. Neuron 30: 65-78.

25. Hu BY, Weick JP, Yu J, Ma LX, Zhang XQ, et al. (2010) Neural differentiation of human induced pluripotent stem cells follows developmental principles but with variable potency. Proc Natl Acad Sci U S A 107: 4335-4340.

26. Okano H, Ogawa Y, Nakamura M, Kaneko S, Iwanami A, et al. (2003) Transplantation of neural stem cells into the spinal cord after injury. Semin Cell Dev Biol 14: 191-198.

27. Tetzlaff W, Okon EB, Karimi-Abdolrezaee S, Hill CE, Sparling JS, et al. (2010) A Systematic Review of Cellular Transplantation Therapies for Spinal Cord Injury. J Neurotrauma.

106

28. Ruff CA, Fehlings MG (2010) Neural stem cells in regenerative medicine: bridging the gap. Panminerva Med 52: 125-147.

29. Cao QL, Zhang YP, Howard RM, Walters WM, Tsoulfas P, et al. (2001) Pluripotent stem cells engrafted into the normal or lesioned adult rat spinal cord are restricted to a glial lineage. Exp Neurol 167: 48-58.

30. Hofstetter CP, Holmstrom NA, Lilja JA, Schweinhardt P, Hao J, et al. (2005) Allodynia limits the usefulness of intraspinal neural stem cell grafts; directed differentiation improves outcome. Nat Neurosci 8: 346-353.

31. Novak U, Kaye AH (2000) Extracellular matrix and the brain: components and function. J Clin Neurosci 7: 280-290.

32. Archibald SJ, Shefner J, Krarup C, Madison RD (1995) Monkey median nerve repaired by nerve graft or collagen nerve guide tube. J Neurosci 15: 4109-4123.

33. Liu S, Bodjarian N, Langlois O, Bonnard AS, Boisset N, et al. (1998) Axonal regrowth through a collagen guidance channel bridging spinal cord to the avulsed C6 roots: functional recovery in primates with brachial plexus injury. J Neurosci Res 51: 723-734.

34. Freier T, Koh HS, Kazazian K, Shoichet MS (2005) Controlling cell adhesion and degradation of chitosan films by N-acetylation. Biomaterials 26: 5872-5878.

35. Dalton PD, Flynn L, Shoichet MS (2002) Manufacture of poly(2-hydroxyethyl methacrylate- co-methyl methacrylate) hydrogel tubes for use as nerve guidance channels. Biomaterials 23: 3843-3851.

36. Nomura H, Tator CH, Shoichet MS (2006) Bioengineered strategies for spinal cord repair. J Neurotrauma 23: 496-507.

37. Straley KS, Foo CW, Heilshorn SC (2010) Biomaterial design strategies for the treatment of spinal cord injuries. J Neurotrauma 27: 1-19.

38. Madigan NN, McMahon S, O'Brien T, Yaszemski MJ, Windebank AJ (2009) Current tissue engineering and novel therapeutic approaches to axonal regeneration following spinal cord injury using polymer scaffolds. Respir Physiol Neurobiol 169: 183-199.

39. Kang CE, Kim H, Talpag V, Shoichet MS (2011) Approaches Using Biomaterials for Tissue Engineering. In: Fehlings MG, Vaccaro AR, Boayke M, Rossignol S, Burns A et al., editors. Essentials of Spinal Cord Injury. New York, NY: Thieme Publishers.

40. Jain A, Kim YT, McKeon RJ, Bellamkonda RV (2006) In situ gelling hydrogels for conformal repair of spinal cord defects, and local delivery of BDNF after spinal cord injury. Biomaterials 27: 497-504.

41. Prang P, Muller R, Eljaouhari A, Heckmann K, Kunz W, et al. (2006) The promotion of oriented axonal regrowth in the injured spinal cord by alginate-based anisotropic capillary hydrogels. Biomaterials 27: 3560-3569.

107

42. Mitsui T, Shumsky JS, Lepore AC, Murray M, Fischer I (2005) Transplantation of neuronal and glial restricted precursors into contused spinal cord improves bladder and motor functions, decreases thermal hypersensitivity, and modifies intraspinal circuitry. J Neurosci 25: 9624-9636.

43. Yoshii S, Ito S, Shima M, Taniguchi A, Akagi M (2009) Functional restoration of rabbit spinal cord using collagen-filament scaffold. J Tissue Eng Regen Med 3: 19-25.

44. Crompton KE, Goud JD, Bellamkonda RV, Gengenbach TR, Finkelstein DI, et al. (2007) Polylysine-functionalised thermoresponsive chitosan hydrogel for neural tissue engineering. Biomaterials 28: 441-449.

45. Taylor SJ, Rosenzweig ES, McDonald JW, 3rd, Sakiyama-Elbert SE (2006) Delivery of neurotrophin-3 from fibrin enhances neuronal fiber sprouting after spinal cord injury. J Control Release 113: 226-235.

46. Gupta D, Tator CH, Shoichet MS (2006) Fast-gelling injectable blend of hyaluronan and methylcellulose for intrathecal, localized delivery to the injured spinal cord. Biomaterials 27: 2370-2379.

47. Kang CE, Poon PC, Tator CH, Shoichet MS (2009) A new paradigm for local and sustained release of therapeutic molecules to the injured spinal cord for neuroprotection and tissue repair. Tissue Eng Part A 15: 595-604.

48. Baumann MD, Kang CE, Stanwick JC, Wang Y, Kim H, et al. (2009) An injectable drug delivery platform for sustained combination therapy. J Control Release 138: 205-213.

49. Tsai EC, Dalton PD, Shoichet MS, Tator CH (2006) Matrix inclusion within synthetic hydrogel guidance channels improves specific supraspinal and local axonal regeneration after complete spinal cord transection. Biomaterials 27: 519-533.

50. Oudega M, Gautier SE, Chapon P, Fragoso M, Bates ML, et al. (2001) Axonal regeneration into Schwann cell grafts within resorbable poly(alpha-hydroxyacid) guidance channels in the adult rat spinal cord. Biomaterials 22: 1125-1136.

51. Moore MJ, Friedman JA, Lewellyn EB, Mantila SM, Krych AJ, et al. (2006) Multiple- channel scaffolds to promote spinal cord axon regeneration. Biomaterials 27: 419-429.

52. Bini TB, Gao S, Wang S, Ramakrishna S (2005) Development of fibrous biodegradable polymer conduits for guided nerve regeneration. J Mater Sci Mater Med 16: 367-375.

53. Wong DY, Leveque JC, Brumblay H, Krebsbach PH, Hollister SJ, et al. (2008) Macro- architectures in spinal cord scaffold implants influence regeneration. J Neurotrauma 25: 1027-1037.

54. Nisbet DR, Yu LM, Zahir T, Forsythe JS, Shoichet MS (2008) Characterization of neural stem cells on electrospun poly(epsilon-caprolactone) submicron scaffolds: evaluating their potential in neural tissue engineering. J Biomater Sci Polym Ed 19: 623-634.

108

55. Novikov LN, Novikova LN, Mosahebi A, Wiberg M, Terenghi G, et al. (2002) A novel biodegradable implant for neuronal rescue and regeneration after spinal cord injury. Biomaterials 23: 3369-3376.

56. Tysseling-Mattiace VM, Sahni V, Niece KL, Birch D, Czeisler C, et al. (2008) Self- assembling nanofibers inhibit glial scar formation and promote axon elongation after spinal cord injury. J Neurosci 28: 3814-3823.

57. Guo J, Su H, Zeng Y, Liang YX, Wong WM, et al. (2007) Reknitting the injured spinal cord by self-assembling peptide nanofiber scaffold. Nanomedicine 3: 311-321.

58. Hejcl A, Lesny P, Pradny M, Sedy J, Zamecnik J, et al. (2009) Macroporous hydrogels based on 2-hydroxyethyl methacrylate. Part 6: 3D hydrogels with positive and negative surface charges and polyelectrolyte complexes in spinal cord injury repair. J Mater Sci Mater Med 20: 1571-1577.

59. Woerly S, Doan VD, Evans-Martin F, Paramore CG, Peduzzi JD (2001) Spinal cord reconstruction using NeuroGel implants and functional recovery after chronic injury. J Neurosci Res 66: 1187-1197.

60. Guest JD, Rao A, Olson L, Bunge MB, Bunge RP (1997) The ability of human Schwann cell grafts to promote regeneration in the transected nude rat spinal cord. Exp Neurol 148: 502- 522.

61. Belkas JS, Munro CA, Shoichet MS, Midha R (2005) Peripheral nerve regeneration through a synthetic hydrogel nerve tube. Restor Neurol Neurosci 23: 19-29.

62. Johansson F, Carlberg, P., Danielsen, N., Montelius, L., Kanje, M. (2006) Axonal Outgrowth on Nano-imprinted Patterns. Biomaterials 27: 1251-1258.

63. Kim IA, Park, S.A., Kim, Y.J., Kim, S., Shin, H.J., Lee, Y.J., Kang, S.G., Shin, J. (2006) Effects of Mechanical Stimuli and Microfiber-Based Substrate on Neurite Outgrowth and Guidance. Journal of Bioscience and Bioengineering 101: 120-126.

64. Shaw D, Shoichet, M.S. (2003) Toward Spinal Cord Injury Repair Strategies: Peptide Surface Modification of Expanded Poly(tetrafluoroethylene) Fibers for Guided Neurite Outgrowth in vitro. Journal of Craniofacial Surgery 14: 308-316.

65. Bellamkonda RV (2006) Peripheral Nerve Regeneration: An Opinion on Channels, Scaffolds and Anisotropy. Biomaterials 27: 3515-3518.

66. Yu TT, Shoichet, M.S. (2006) Guided Cell Adhesion and Outgrowth in Peptide-modified Channels for Neural Tissue Engineering. Biomaterials 26: 1507-1514.

67. Ichihara S, Inada Y, Nakamura T (2008) Artificial nerve tubes and their application for repair of peripheral nerve injury: an update of current concepts. Injury 39 Suppl 4: 29-39.

68. Li X, Yang Z, Zhang A, Wang T, Chen W (2009) Repair of thoracic spinal cord injury by chitosan tube implantation in adult rats. Biomaterials 30: 1121-1132.

109

69. Li GN, Hoffman-Kim D (2008) Tissue-engineered platforms of axon guidance. Tissue Eng Part B Rev 14: 33-51.

70. De Laporte L, Huang A, Ducommun MM, Zelivyanska ML, Aviles MO, et al. (2010) Patterned transgene expression in multiple-channel bridges after spinal cord injury. Acta Biomater 6: 2889-2897.

71. Himes BT, Neuhuber B, Coleman C, Kushner R, Swanger SA, et al. (2006) Recovery of function following grafting of human bone marrow-derived stromal cells into the injured spinal cord. Neurorehabil Neural Repair 20: 278-296.

72. Parr AM, Kulbatski I, Tator CH (2007) Transplantation of adult rat spinal cord stem/progenitor cells for spinal cord injury. J Neurotrauma 24: 835-845.

73. Mothe AJ, Kulbatski I, Parr A, Mohareb M, Tator CH (2008) Adult spinal cord stem/progenitor cells transplanted as neurospheres preferentially differentiate into oligodendrocytes in the adult rat spinal cord. Cell Transplant 17: 735-751.

74. Zvibel I, Smets F, Soriano H (2002) Anoikis: roadblock to cell transplantation? Cell Transplant 11: 621-630.

75. Zahir T, Nomura H, Guo XD, Kim H, Tator C, et al. (2008) Bioengineering neural stem/progenitor cell-coated tubes for spinal cord injury repair. Cell Transplant 17: 245- 254.

76. Itosaka H, Kuroda S, Shichinohe H, Yasuda H, Yano S, et al. (2009) Fibrin matrix provides a suitable scaffold for bone marrow stromal cells transplanted into injured spinal cord: a novel material for CNS tissue engineering. Neuropathology 29: 248-257.

77. Teng YD, Lavik EB, Qu X, Park KI, Ourednik J, et al. (2002) Functional recovery following traumatic spinal cord injury mediated by a unique polymer scaffold seeded with neural stem cells. Proc Natl Acad Sci U S A 99: 3024-3029.

78. Kim BG, Kang YM, Phi JH, Kim YH, Hwang DH, et al. (2010) Implantation of polymer scaffolds seeded with neural stem cells in a canine spinal cord injury model. Cytotherapy 12: 841-845.

79. Pritchard CD, Slotkin JR, Yu D, Dai H, Lawrence MS, et al. (2010) Establishing a model spinal cord injury in the African green monkey for the preclinical evaluation of biodegradable polymer scaffolds seeded with human neural stem cells. J Neurosci Methods 188: 258-269.

80. Gustafson SB, Fulkerson P, Bildfell R, Aguilera L, Hazzard TM (2007) Chitosan dressing provides hemostasis in swine femoral arterial injury model. Prehosp Emerg Care 11: 172- 178.

81. Fernandes JC, Tavaria FK, Fonseca SC, Ramos OS, Pintado ME, et al. (2010) In vitro screening for anti-microbial activity of chitosans and chitooligosaccharides, aiming at potential uses in functional textiles. J Microbiol Biotechnol 20: 311-318.

110

82. Fernandes JC, Tavaria FK, Soares JC, Ramos OS, Joao Monteiro M, et al. (2008) Antimicrobial effects of chitosans and chitooligosaccharides, upon Staphylococcus aureus and Escherichia coli, in food model systems. Food Microbiol 25: 922-928.

83. Synowiecki J, Al-Khateeb NA (2003) Production, properties, and some new applications of chitin and its derivatives. Crit Rev Food Sci Nutr 43: 145-171.

84. Howard MB, Ekborg NA, Weiner RM, Hutcheson SW (2003) Detection and characterization of chitinases and other chitin-modifying enzymes. J Ind Microbiol Biotechnol 30: 627-635.

85. Varum KM, Holme HK, Izume M, Stokke BT, Smidsrod O (1996) Determination of enzymatic hydrolysis specificity of partially N-acetylated chitosans. Biochim Biophys Acta 1291: 5-15.

86. Tomihata K, Ikada Y (1997) In vitro and in vivo degradation of films of chitin and its deacetylated derivatives. Biomaterials 18: 567-575.

87. Nagpal K, Singh SK, Mishra DN (2010) Chitosan nanoparticles: a promising system in novel drug delivery. Chem Pharm Bull (Tokyo) 58: 1423-1430.

88. VandeVord PJ, Matthew HW, DeSilva SP, Mayton L, Wu B, et al. (2002) Evaluation of the biocompatibility of a chitosan scaffold in mice. J Biomed Mater Res 59: 585-590.

89. Schiffman JD, Schauer CL (2007) One-step electrospinning of cross-linked chitosan fibers. Biomacromolecules 8: 2665-2667.

90. Freier T, Montenegro R, Shan Koh H, Shoichet MS (2005) Chitin-based tubes for tissue engineering in the nervous system. Biomaterials 26: 4624-4632.

91. Mi FL, Tan YC, Liang HF, Sung HW (2002) In vivo biocompatibility and degradability of a novel injectable-chitosan-based implant. Biomaterials 23: 181-191.

92. Wen P, Gao J, Zhang Y, Li X, Long Y, et al. (2010) Fabrication of Chitosan Scaffolds with Tunable Porous Orientation Structure for Tissue Engineering. J Biomater Sci Polym Ed.

93. Ahmadi R, de Bruijn JD (2008) Biocompatibility and gelation of chitosan-glycerol phosphate hydrogels. J Biomed Mater Res A 86: 824-832.

94. Lavertu M, Filion D, Buschmann MD (2008) Heat-induced transfer of protons from chitosan to glycerol phosphate produces chitosan precipitation and gelation. Biomacromolecules 9: 640-650.

95. Lapitsky Y, Zahir T, Shoichet MS (2008) Modular biodegradable biomaterials from surfactant and polyelectrolyte mixtures. Biomacromolecules 9: 166-174.

96. Leipzig ND, Wylie RG, Kim H, Shoichet MS (2010) Differentiation of neural stem cells in three-dimensional growth factor-immobilized chitosan hydrogel scaffolds. Biomaterials 32: 57-64.

111

97. Cho Y, Shi R, Borgens RB (2010) Chitosan produces potent neuroprotection and physiological recovery following traumatic spinal cord injury. J Exp Biol 213: 1513-1520.

98. Nomura H, Baladie B, Katayama Y, Morshead CM, Shoichet MS, et al. (2008) Delayed implantation of intramedullary chitosan channels containing nerve grafts promotes extensive axonal regeneration after spinal cord injury. Neurosurgery 63: 127-141; discussion 141-123.

99. Bozkurt G, Mothe AJ, Zahir T, Kim H, Shoichet MS, et al. (2010) Chitosan channels containing spinal cord-derived stem/progenitor cells for repair of subacute spinal cord injury in the rat. Neurosurgery 67: 1733-1744.

100. Brass EP, Forman WB, Edwards RV, Lindan O (1978) Fibrin formation: effect of calcium ions. Blood 52: 654-658.

101. Zhao H, Ma L, Zhou J, Mao Z, Gao C, et al. (2008) Fabrication and physical and biological properties of fibrin gel derived from human plasma. Biomed Mater 3: 015001.

102. Ahmed TA, Dare EV, Hincke M (2008) Fibrin: a versatile scaffold for tissue engineering applications. Tissue Eng Part B Rev 14: 199-215.

103. Galla TJ, Vedecnik SV, Halbgewachs J, Steinmann S, Friedrich C, et al. (2004) Fibrin/Schwann cell matrix in poly-epsilon-caprolactone conduits enhances guided nerve regeneration. Int J Artif Organs 27: 127-136.

104. Wood MD, MacEwan MR, French AR, Moore AM, Hunter DA, et al. (2010) Fibrin matrices with affinity-based delivery systems and neurotrophic factors promote functional nerve regeneration. Biotechnol Bioeng 106: 970-979.

105. Kalbermatten DF, Pettersson J, Kingham PJ, Pierer G, Wiberg M, et al. (2009) New fibrin conduit for peripheral nerve repair. J Reconstr Microsurg 25: 27-33.

106. Pettersson J, Kalbermatten D, McGrath A, Novikova LN (2010) Biodegradable fibrin conduit promotes long-term regeneration after peripheral nerve injury in adult rats. J Plast Reconstr Aesthet Surg 63: 1893-1899.

107. Williams LR, Longo FM, Powell HC, Lundborg G, Varon S (1983) Spatial-temporal progress of peripheral nerve regeneration within a silicone chamber: parameters for a bioassay. J Comp Neurol 218: 460-470.

108. Johnson PJ, Parker SR, Sakiyama-Elbert SE (2010) Fibrin-based tissue engineering scaffolds enhance neural fiber sprouting and delay the accumulation of reactive astrocytes at the lesion in a subacute model of spinal cord injury. J Biomed Mater Res A 92: 152-163.

109. King VR, Alovskaya A, Wei DY, Brown RA, Priestley JV (2010) The use of injectable forms of fibrin and fibronectin to support axonal ingrowth after spinal cord injury. Biomaterials 31: 4447-4456.

112

110. Taylor SJ, McDonald JW, 3rd, Sakiyama-Elbert SE (2004) Controlled release of neurotrophin-3 from fibrin gels for spinal cord injury. J Control Release 98: 281-294.

111. Thomopoulos S, Das R, Sakiyama-Elbert S, Silva MJ, Charlton N, et al. (2010) bFGF and PDGF-BB for tendon repair: controlled release and biologic activity by tendon fibroblasts in vitro. Ann Biomed Eng 38: 225-234.

112. Wood MD, Borschel GH, Sakiyama-Elbert SE (2009) Controlled release of glial-derived neurotrophic factor from fibrin matrices containing an affinity-based delivery system. J Biomed Mater Res A 89: 909-918.

113. Wood MD, Sakiyama-Elbert SE (2008) Release rate controls biological activity of nerve growth factor released from fibrin matrices containing affinity-based delivery systems. J Biomed Mater Res A 84: 300-312.

114. Lord-Fontaine S, Yang F, Diep Q, Dergham P, Munzer S, et al. (2008) Local inhibition of Rho signaling by cell-permeable recombinant protein BA-210 prevents secondary damage and promotes functional recovery following acute spinal cord injury. J Neurotrauma 25: 1309-1322.

115. Johnson PJ, Tatara A, McCreedy DA, Shiu A, Sakiyama-Elbert SE (2010) Tissue- engineered fibrin scaffolds containing neural progenitors enhance functional recovery in a subacute model of SCI. Soft Matter 6: 5127-5137.

116. Willerth SM, Arendas KJ, Gottlieb DI, Sakiyama-Elbert SE (2006) Optimization of fibrin scaffolds for differentiation of murine embryonic stem cells into neural lineage cells. Biomaterials 27: 5990-6003.

117. Liu J, Zhang Z, Gong A, Cao X, Qian L, et al. (2010) Neuronal progenitor cells seeded in fibrin gel differentiate into ChAT-positive neuron. In Vitro Cell Dev Biol Anim 46: 738- 745.

118. Willerth SM, Faxel TE, Gottlieb DI, Sakiyama-Elbert SE (2007) The Effects of Soluble Growth Factors on Embryonic Stem Cell Differentiation Inside of Fibrin Scaffolds. Stem Cells.

119. Johnson PJ, Tatara A, Shiu A, Sakiyama-Elbert SE (2010) Controlled release of neurotrophin-3 and platelet-derived growth factor from fibrin scaffolds containing neural progenitor cells enhances survival and differentiation into neurons in a subacute model of SCI. Cell Transplant 19: 89-101.

120. Asano M, Fukuzaki H, Yoshida M, Kumakura M, Mashimo T, et al. (1989) In vivo characteristics of low molecular weight copoly (D,L-lactic acid) formulations with controlled release of LH-RH agonist. Biomaterials 10: 569-573.

121. Cheng Y, Deng S, Chen P, Ruan R (2009) Polylactic acid (PLA) synthesis and modifications: a review. Front Chem China 4: 259-264.

113

122. Albertsson AC, Varma IK (2003) Recent developments in ring opening polymerization of lactones for biomedical applications. Biomacromolecules 4: 1466-1486.

123. Seyednejad H, Ghassemi AH, van Nostrum CF, Vermonden T, Hennink WE (2011) Functional aliphatic polyesters for biomedical and pharmaceutical applications. J Control Release.

124. Chen BK, Knight AM, de Ruiter GC, Spinner RJ, Yaszemski MJ, et al. (2009) Axon regeneration through scaffold into distal spinal cord after transection. J Neurotrauma 26: 1759-1771.

125. Krych AJ, Rooney GE, Chen B, Schermerhorn TC, Ameenuddin S, et al. (2009) Relationship between scaffold channel diameter and number of regenerating axons in the transected rat spinal cord. Acta Biomater 5: 2551-2559.

126. Baumann MD, Kang CE, Tator CH, Shoichet MS (2010) Intrathecal delivery of a polymeric nanocomposite hydrogel after spinal cord injury. Biomaterials 31: 7631-7639.

127. Chvatal SA, Kim YT, Bratt-Leal AM, Lee H, Bellamkonda RV (2008) Spatial distribution and acute anti-inflammatory effects of Methylprednisolone after sustained local delivery to the contused spinal cord. Biomaterials 29: 1967-1975.

128. Lagarce F, Faisant N, Desfontis JC, Marescaux L, Gautier F, et al. (2005) Biopharmaceutics of intrathecal baclofen-loaded microparticles in a goat model. Int J Pharm 298: 68-79.

129. Wang YC, Wu YT, Huang HY, Lin HI, Lo LW, et al. (2008) Sustained intraspinal delivery of neurotrophic factor encapsulated in biodegradable nanoparticles following contusive spinal cord injury. Biomaterials 29: 4546-4553.

130. Baptiste DC, Fehlings MG (2006) Pharmacological approaches to repair the injured spinal cord. J Neurotrauma 23: 318-334.

131. Schmidt CE, Leach JB (2003) Neural tissue engineering: strategies for repair and regeneration. Annu Rev Biomed Eng 5: 293-347.

132. Mocchetti I, Wrathall JR (1995) Neurotrophic factors in central nervous system trauma. J Neurotrauma 12: 853-870.

133. Dergham P, Ellezam B, Essagian C, Avedissian H, Lubell WD, et al. (2002) Rho signaling pathway targeted to promote spinal cord repair. J Neurosci 22: 6570-6577.

134. Karimi-Abdolrezaee S, Eftekharpour E, Wang J, Schut D, Fehlings MG (2010) Synergistic effects of transplanted adult neural stem/progenitor cells, chondroitinase, and growth factors promote functional repair and plasticity of the chronically injured spinal cord. J Neurosci 30: 1657-1676.

135. Tojima T, Ito E (2001) A cyclic AMP-regulated negative feedforward system for neuritogenesis revealed in a neuroblastomaxglioma hybrid cell line. Neuroscience 104: 583-591.

114

136. Nakao N (1998) An increase in intracellular levels of cyclic AMP produces trophic effects on striatal neurons developing in culture. Neuroscience 82: 1009-1020.

137. Cai D, Deng K, Mellado W, Lee J, Ratan RR, et al. (2002) Arginase I and polyamines act downstream from cyclic AMP in overcoming inhibition of axonal growth MAG and myelin in vitro. Neuron 35: 711-719.

138. Gao Y, Nikulina E, Mellado W, Filbin MT (2003) Neurotrophins elevate cAMP to reach a threshold required to overcome inhibition by MAG through extracellular signal-regulated kinase-dependent inhibition of phosphodiesterase. J Neurosci 23: 11770-11777.

139. Cai D, Qiu J, Cao Z, McAtee M, Bregman BS, et al. (2001) Neuronal cyclic AMP controls the developmental loss in ability of axons to regenerate. J Neurosci 21: 4731-4739.

140. Hannila SS, Filbin MT (2008) The role of cyclic AMP signaling in promoting axonal regeneration after spinal cord injury. Exp Neurol 209: 321-332.

141. Qiu J, Cai D, Dai H, McAtee M, Hoffman PN, et al. (2002) Spinal axon regeneration induced by elevation of cyclic AMP. Neuron 34: 895-903.

142. Spencer T, Filbin MT (2004) A role for cAMP in regeneration of the adult mammalian CNS. J Anat 204: 49-55.

143. Pearse DD, Pereira FC, Marcillo AE, Bates ML, Berrocal YA, et al. (2004) cAMP and Schwann cells promote axonal growth and functional recovery after spinal cord injury. Nat Med 10: 610-616.

144. Kim G, Choe Y, Park J, Cho S, Kim K (2002) Activation of protein kinase A induces neuronal differentiation of HiB5 hippocampal progenitor cells. Brain Res Mol Brain Res 109: 134-145.

145. Kume T, Kawato Y, Osakada F, Izumi Y, Katsuki H, et al. (2008) Dibutyryl cyclic AMP induces differentiation of human neuroblastoma SH-SY5Y cells into a noradrenergic phenotype. Neurosci Lett 443: 199-203.

146. Tojima T, Kobayashi S, Ito E (2003) Dual role of cyclic AMP-dependent protein kinase in neuritogenesis and synaptogenesis during neuronal differentiation. J Neurosci Res 74: 829- 837.

147. Jagasia R, Steib K, Englberger E, Herold S, Faus-Kessler T, et al. (2009) GABA-cAMP response element-binding protein signaling regulates maturation and survival of newly generated neurons in the adult hippocampus. J Neurosci 29: 7966-7977.

148. Dworkin S, Mantamadiotis T (2010) Targeting CREB signalling in neurogenesis. Expert Opin Ther Targets 14: 869-879.

149. Monaghan TK, Mackenzie CJ, Plevin R, Lutz EM (2008) PACAP-38 induces neuronal differentiation of human SH-SY5Y neuroblastoma cells via cAMP-mediated activation of ERK and p38 MAP kinases. J Neurochem 104: 74-88.

115

150. Gerdin MJ, Eiden LE (2007) Regulation of PC12 cell differentiation by cAMP signaling to ERK independent of PKA: do all the connections add up? Sci STKE 2007: pe15.

151. Jones LL, Tuszynski MH (2001) Chronic intrathecal infusions after spinal cord injury cause scarring and compression. Microsc Res Tech 54: 317-324.

152. Jimenez Hamann MC, Tator CH, Shoichet MS (2005) Injectable intrathecal delivery system for localized administration of EGF and FGF-2 to the injured rat spinal cord. Exp Neurol 194: 106-119.

153. Johnson PJ, Parker SR, Sakiyama-Elbert SE (2009) Controlled release of neurotrophin-3 from fibrin-based tissue engineering scaffolds enhances neural fiber sprouting following subacute spinal cord injury. Biotechnol Bioeng 104: 1207-1214.

154. Lee H, McKeon RJ, Bellamkonda RV (2010) Sustained delivery of thermostabilized chABC enhances axonal sprouting and functional recovery after spinal cord injury. Proc Natl Acad Sci U S A 107: 3340-3345.

155. Fan J, Xiao Z, Zhang H, Chen B, Tang G, et al. (2010) Linear ordered collagen scaffolds loaded with collagen-binding neurotrophin-3 promote axonal regeneration and partial functional recovery after complete spinal cord transection. J Neurotrauma 27: 1671-1683.

156. Willerth SM, Sakiyama-Elbert SE (2007) Approaches to neural tissue engineering using scaffolds for drug delivery. Adv Drug Deliv Rev 59: 325-338.

157. Benoit JP, Faisant N, Venier-Julienne MC, Menei P (2000) Development of microspheres for neurological disorders: from basics to clinical applications. J Control Release 65: 285- 296.

158. Sinha VR, Bansal K, Kaushik R, Kumria R, Trehan A (2004) Poly-epsilon-caprolactone microspheres and nanospheres: an overview. Int J Pharm 278: 1-23.

159. Sinha VR, Trehan A (2003) Biodegradable microspheres for protein delivery. J Control Release 90: 261-280.

160. Whittlesey KJ, Shea LD (2004) Delivery systems for small molecule drugs, proteins, and DNA: the neuroscience/biomaterial interface. Exp Neurol 190: 1-16.

161. Eroglu H, Nemutlu E, Turkoglu OF, Nacar O, Bodur E, et al. (2010) A quadruped study on chitosan microspheres containing atorvastatin calcium: preparation, characterization, quantification and in-vivo application. Chem Pharm Bull (Tokyo) 58: 1161-1167.

162. Klose D, Laprais M, Leroux V, Siepmann F, Deprez B, et al. (2009) Fenofibrate-loaded PLGA microparticles: effects on ischemic stroke. Eur J Pharm Sci 37: 43-52.

163. Piotrowicz A, Shoichet MS (2006) Nerve guidance channels as drug delivery vehicles. Biomaterials 27: 2018-2027.

116

164. Goraltchouk A, Scanga V, Morshead CM, Shoichet MS (2006) Incorporation of protein- eluting microspheres into biodegradable nerve guidance channels for controlled release. J Control Release 110: 400-407.

165. Johnson EO, Soucacos PN (2008) Nerve repair: experimental and clinical evaluation of biodegradable artificial nerve guides. Injury 39 Suppl 3: S30-36.

166. Nomura H, Katayama Y, Shoichet MS, Tator CH (2006) Complete spinal cord transection treated by implantation of a reinforced synthetic hydrogel channel results in syringomyelia and caudal migration of the rostral stump. Neurosurgery 59: 183-192; discussion 183-192.

167. Ramon-Cueto A, Plant GW, Avila J, Bunge MB (1998) Long-distance axonal regeneration in the transected adult rat spinal cord is promoted by olfactory ensheathing glia transplants. J Neurosci 18: 3803-3815.

168. Cheng H, Huang YC, Chang PT, Huang YY (2007) Laminin-incorporated nerve conduits made by plasma treatment for repairing spinal cord injury. Biochem Biophys Res Commun 357: 938-944.

169. Borgens RB (1999) Electrically mediated regeneration and guidance of adult mammalian spinal axons into polymeric channels. Neuroscience 91: 251-264.

170. Xu XM, Guenard V, Kleitman N, Bunge MB (1995) Axonal regeneration into Schwann cell-seeded guidance channels grafted into transected adult rat spinal cord. J Comp Neurol 351: 145-160.

171. Maia J, Santos T, Aday S, Agasse F, Cortes L, et al. (2011) Controlling the neuronal differentiation of stem cells by the intracellular delivery of retinoic acid-loaded nanoparticles. ACS Nano 5: 97-106.

172. Kim H, Tator C, Shoichet M (2011) Chitosan implants in the rat spinal cord: Biocompatibility and biodegradation. J Biomed Mater Res A 97: 395-404.

173. Samadikuchaksaraei A (2007) An overview of tissue engineering approaches for management of spinal cord injuries. J Neuroengineering Rehabil 4: 15.

174. Novikova LN, Novikov LN, Kellerth JO (2003) Biopolymers and biodegradable smart implants for tissue regeneration after spinal cord injury. Curr Opin Neurol 16: 711-715.

175. Woerly S, Pinet E, de Robertis L, Van Diep D, Bousmina M (2001) Spinal cord repair with PHPMA hydrogel containing RGD peptides (NeuroGel). Biomaterials 22: 1095-1111.

176. Gazzeri R, Neroni M, Alfieri A, Galarza M, Faiola A, et al. (2009) Transparent equine collagen biomatrix as dural repair. A prospective clinical study. Acta Neurochir (Wien) 151: 537-543.

177. Williams DF (2008) On the mechanisms of biocompatibility. Biomaterials 29: 2941-2953.

117

178. Anderson JM, Rodriguez A, Chang DT (2008) Foreign body reaction to biomaterials. Semin Immunol 20: 86-100.

179. Di Vita G, Patti R, Sparacello M, Balistreri CR, Candore G, et al. (2008) Impact of different texture of polypropylene mesh on the inflammatory response. Int J Immunopathol Pharmacol 21: 207-214.

180. Kohane DS, Tse JY, Yeo Y, Padera R, Shubina M, et al. (2006) Biodegradable polymeric microspheres and nanospheres for drug delivery in the peritoneum. J Biomed Mater Res A 77: 351-361.

181. Ghanaati S, Barbeck M, Orth C, Willershausen I, Thimm BW, et al. Influence of beta- tricalcium phosphate granule size and morphology on tissue reaction in vivo. Acta Biomater.

182. Kofuji K, Ito T, Murata Y, Kawashima S (2001) Biodegradation and drug release of chitosan gel beads in subcutaneous air pouches of mice. Biol Pharm Bull 24: 205-208.

183. Ren D, Yi H, Wang W, Ma X (2005) The enzymatic degradation and swelling properties of chitosan matrices with different degrees of N-acetylation. Carbohydr Res 340: 2403-2410.

184. Kim H, Tator CH, Shoichet MS (2008) Design of protein-releasing chitosan channels. Biotechnol Prog 24: 932-937.

185. Hirai A, Odani H, Nakajima A (1991) Determination of degree of deacetylation of chitosan by 1H NMR spectroscopy. Polymer Bulletin 26: 87-94.

186. Azab AK, Doviner V, Orkin B, Kleinstern J, Srebnik M, et al. (2007) Biocompatibility evaluation of crosslinked chitosan hydrogels after subcutaneous and intraperitoneal implantation in the rat. J Biomed Mater Res A 83: 414-422.

187. Popovich PG, Wei P, Stokes BT (1997) Cellular inflammatory response after spinal cord injury in Sprague-Dawley and Lewis rats. J Comp Neurol 377: 443-464.

188. Barbolt TA, Odin M, Leger M, Kangas L, Hoiste J, et al. (2001) Biocompatibility evaluation of dura mater substitutes in an animal model. Neurol Res 23: 813-820.

189. Keller JT, Weil SM, Ongkiko CM, Jr., Tew JM, Jr., Mayfield FH, et al. (1989) Repair of spinal dural defects with vicryl (polyglactin 910) mesh. J Spinal Disord 2: 87-92.

190. Park YK, Tator CH (1998) Prevention of arachnoiditis and postoperative tethering of the spinal cord with Gore-Tex surgical membrane: an experimental study with rats. Neurosurgery 42: 813-823; discussion 823-814.

191. Shigemasa Y, Saito K, Sashiwa H, Saimoto H (1994) Enzymatic degradation of chitins and partially deacetylated chitins. Int J Biol Macromol 16: 43-49.

192. Temel A, Kazokoglu H, Taga Y (1991) Tear lysozyme levels in contact lens wearers. Ann Ophthalmol 23: 191-194.

118

193. Currie GA, Eccles SA (1976) Serum lysozyme as a marker of host resistance. I. Production by macrophages resident in rat sarcomata. Br J Cancer 33: 51-59.

194. Mishra OP, Batra P, Ali Z, Anupurba S, Das BK (2003) Cerebrospinal fluid lysozyme level for the diagnosis of tuberculous meningitis in children. J Trop Pediatr 49: 13-16.

195. Zhang KH, Xiao HS, Lu PH, Shi J, Li GD, et al. (2004) Differential gene expression after complete spinal cord transection in adult rats: an analysis focused on a subchronic post- injury stage. Neuroscience 128: 375-388.

196. Peluso G, Petillo O, Ranieri M, Santin M, Ambrosio L, et al. (1994) Chitosan-mediated stimulation of macrophage function. Biomaterials 15: 1215-1220.

197. Benesch J, Tengvall P (2002) Blood protein adsorption onto chitosan. Biomaterials 23: 2561-2568.

198. Bettinger C, Zimmermann HW (1991) New investigations on hematoxylin, hematein, and hematein-aluminium complexes. II. Hematein-aluminium complexes and hemalum staining. Histochemistry 96: 215-228.

199. Bunel S, Ibarra C, Moraga E, Blasko A, Bunton CA (1993) Structures of Complexes of Cobalt(Iii) and Glucosamine - Applications of Molecular Mechanics and Nmr- Spectroscopy. Carbohydrate Research 244: 1-14.

200. Sogias IA, Khutoryanskiy VV, Williams AC (2010) Exploring the Factors Affecting the Solubility of Chitosan in Water. Macromolecular Chemistry and Physics 211: 426-433.

201. Cunha-Reis C, TuzlaKoglu K, Baas E, Yang Y, El Haj A, et al. (2007) Influence of porosity and fibre diameter on the degradation of chitosan fibre-mesh scaffolds and cell adhesion. J Mater Sci Mater Med 18: 195-200.

202. Tator CH (2006) Review of treatment trials in human spinal cord injury: issues, difficulties, and recommendations. Neurosurgery 59: 957-982; discussion 982-957.

203. Bertleff MJ, Meek MF, Nicolai JP (2005) A prospective clinical evaluation of biodegradable neurolac nerve guides for sensory nerve repair in the hand. J Hand Surg [Am] 30: 513-518.

204. Navissano M, Malan F, Carnino R, Battiston B (2005) Neurotube for facial nerve repair. Microsurgery 25: 268-271.

205. Kwon BK, Liu J, Lam C, Plunet W, Oschipok LW, et al. (2007) Brain-derived neurotrophic factor gene transfer with adeno-associated viral and lentiviral vectors prevents rubrospinal neuronal atrophy and stimulates regeneration-associated gene expression after acute cervical spinal cord injury. Spine 32: 1164-1173.

206. Zhou L, Baumgartner BJ, Hill-Felberg SJ, McGowen LR, Shine HD (2003) Neurotrophin-3 expressed in situ induces axonal plasticity in the adult injured spinal cord. J Neurosci 23: 1424-1431.

119

207. Namiki J, Kojima A, Tator CH (2000) Effect of brain-derived neurotrophic factor, nerve growth factor, and neurotrophin-3 on functional recovery and regeneration after spinal cord injury in adult rats. J Neurotrauma 17: 1219-1231.

208. Liebscher T, Schnell L, Schnell D, Scholl J, Schneider R, et al. (2005) Nogo-A antibody improves regeneration and locomotion of spinal cord-injured rats. Ann Neurol 58: 706- 719.

209. Wang X, Baughman KW, Basso DM, Strittmatter SM (2006) Delayed Nogo receptor therapy improves recovery from spinal cord contusion. Ann Neurol 60: 540-549.

210. Monnier PP, Sierra A, Schwab JM, Henke-Fahle S, Mueller BK (2003) The Rho/ROCK pathway mediates neurite growth-inhibitory activity associated with the chondroitin sulfate proteoglycans of the CNS glial scar. Mol Cell Neurosci 22: 319-330.

211. Barritt AW, Davies M, Marchand F, Hartley R, Grist J, et al. (2006) Chondroitinase ABC promotes sprouting of intact and injured spinal systems after spinal cord injury. J Neurosci 26: 10856-10867.

212. Motta F, Buonaguro V, Stignani C (2007) The use of intrathecal baclofen pump implants in children and adolescents: safety and complications in 200 consecutive cases. J Neurosurg 107: 32-35.

213. Protopapas MG, Bundock E, Westmoreland S, Nero C, Graham WA, et al. (2007) The complications of scar formation associated with intrathecal pump placement. Arch Phys Med Rehabil 88: 389-390.

214. Midha R, Munro CA, Dalton PD, Tator CH, Shoichet MS (2003) Growth factor enhancement of peripheral nerve regeneration through a novel synthetic hydrogel tube. J Neurosurg 99: 555-565.

215. Hadlock T, Sundback C, Koka R, Hunter D, Cheney M, et al. (1999) A novel, biodegradable polymer conduit delivers neurotrophins and promotes nerve regeneration. Laryngoscope 109: 1412-1416.

216. Aebischer P, Salessiotis AN, Winn SR (1989) Basic fibroblast growth factor released from synthetic guidance channels facilitates peripheral nerve regeneration across long nerve gaps. J Neurosci Res 23: 282-289.

217. Bloch J, Fine EG, Bouche N, Zurn AD, Aebischer P (2001) Nerve growth factor- and neurotrophin-3-releasing guidance channels promote regeneration of the transected rat dorsal root. Exp Neurol 172: 425-432.

218. Barras FM, Pasche P, Bouche N, Aebischer P, Zurn AD (2002) Glial cell line-derived neurotrophic factor released by synthetic guidance channels promotes facial nerve regeneration in the rat. J Neurosci Res 70: 746-755.

219. Yang Y, De Laporte L, Rives CB, Jang JH, Lin WC, et al. (2005) Neurotrophin releasing single and multiple lumen nerve conduits. J Control Release 104: 433-446.

120

220. Lavertu M, Xia Z, Serreqi AN, Berrada M, Rodrigues A, et al. (2003) A validated 1H NMR method for the determination of the degree of deacetylation of chitosan. J Pharm Biomed Anal 32: 1149-1158.

221. Aubert-Pouessel A, Bibby DC, Venier-Julienne MC, Hindre F, Benoit JP (2002) A novel in vitro delivery system for assessing the biological integrity of protein upon release from PLGA microspheres. Pharm Res 19: 1046-1051.

222. Dorati R, Genta I, Montanari L, Cilurzo F, Buttafava A, et al. (2005) The effect of gamma- irradiation on PLGA/PEG microspheres containing ovalbumin. J Control Release 107: 78- 90.

223. Ruhe PQ, Boerman OC, Russel FG, Spauwen PH, Mikos AG, et al. (2005) Controlled release of rhBMP-2 loaded poly(dl-lactic-co-glycolic acid)/calcium phosphate cement composites in vivo. J Control Release 106: 162-171.

224. Srinivasan C, Katare YK, Muthukumaran T, Panda AK (2005) Effect of additives on encapsulation efficiency, stability and bioactivity of entrapped lysozyme from biodegradable polymer particles. J Microencapsul 22: 127-138.

225. Pean JM, Venier-Julienne MC, Boury F, Menei P, Denizot B, et al. (1998) NGF release from poly(D,L-lactide-co-glycolide) microspheres. Effect of some formulation parameters on encapsulated NGF stability. J Control Release 56: 175-187.

226. Jollivet C, Aubert-Pouessel A, Clavreul A, Venier-Julienne MC, Remy S, et al. (2004) Striatal implantation of GDNF releasing biodegradable microspheres promotes recovery of motor function in a partial model of Parkinson's disease. Biomaterials 25: 933-942.

227. Kang F, Singh J (2001) Effect of additives on the release of a model protein from PLGA microspheres. AAPS PharmSciTech 2: 30.

228. Rizkalla N, Range C, Lacasse FX, Hildgen P (2006) Effect of various formulation parameters on the properties of polymeric nanoparticles prepared by multiple emulsion method. J Microencapsul 23: 39-57.

229. Sandor M, Riechel A, Kaplan I, Mathiowitz E (2002) Effect of lecithin and MgCO3 as additives on the enzymatic activity of carbonic anhydrase encapsulated in poly(lactide-co- glycolide) (PLGA) microspheres. Biochim Biophys Acta 1570: 63-74.

230. Jiang G, Thanoo BC, DeLuca PP (2002) Effect of osmotic pressure in the solvent extraction phase on BSA release profile from PLGA microspheres. Pharm Dev Technol 7: 391-399.

231. Estey T, Kang J, Schwendeman SP, Carpenter JF (2006) BSA degradation under acidic conditions: A model for protein instability during release from PLGA delivery systems. J Pharm Sci 95: 1626-1639.

232. Bhatti AR, Alvi A, Walia S, Chaudhry GR (2002) pH-dependent modulation of alkaline phosphatase activity in Serratia marcescens. Curr Microbiol 45: 245-249.

121

233. Faisant N, Akiki J, Siepmann F, Benoit JP, Siepmann J (2006) Effects of the type of release medium on drug release from PLGA-based microparticles: experiment and theory. Int J Pharm 314: 189-197.

234. Zolnik BS, Burgess DJ (2007) Effect of acidic pH on PLGA microsphere degradation and release. J Control Release 122: 338-344.

235. Husmann M, Schenderlein S, Luck M, Lindner H, Kleinebudde P (2002) Polymer erosion in PLGA microparticles produced by phase separation method. Int J Pharm 242: 277-280.

236. Chung TW, Tsai YL, Hsieh JH, Tsai WJ (2006) Different ratios of lactide and glycolide in PLGA affect the surface property and protein delivery characteristics of the PLGA microspheres with hydrophobic additives. J Microencapsul 23: 15-27.

237. Mittal G, Sahana DK, Bhardwaj V, Ravi Kumar MN (2007) Estradiol loaded PLGA nanoparticles for oral administration: effect of polymer molecular weight and copolymer composition on release behavior in vitro and in vivo. J Control Release 119: 77-85.

238. Cao X, Schoichet MS (1999) Delivering neuroactive molecules from biodegradable microspheres for application in central nervous system disorders. Biomaterials 20: 329- 339.

239. Berchane NS, Carson KH, Rice-Ficht AC, Andrews MJ (2007) Effect of mean diameter and polydispersity of PLG microspheres on drug release: experiment and theory. Int J Pharm 337: 118-126.

240. Yeh MK, Tung SM, Lu DW, Chen JL, Chiang CH (2001) Formulation factors for preparing ocular biodegradable delivery system of 5-fluorouracil microparticles. J Microencapsul 18: 507-519.

241. Sanchez A, Tobio M, Gonzalez L, Fabra A, Alonso MJ (2003) Biodegradable micro- and nanoparticles as long-term delivery vehicles for interferon-alpha. Eur J Pharm Sci 18: 221- 229.

242. Kang SW, Cho ER, Jeon O, Kim BS (2007) The effect of microsphere degradation rate on the efficacy of polymeric microspheres as bulking agents: an 18-month follow-up study. J Biomed Mater Res B Appl Biomater 80: 253-259.

243. Zhu G, Schwendeman SP (2000) Stabilization of proteins encapsulated in cylindrical poly(lactide-co-glycolide) implants: mechanism of stabilization by basic additives. Pharm Res 17: 351-357.

244. Kim H, Zahir T, Tator C, SHoichet MS (2011) Effects of dibutyryl cyclic-AMP on survival and neuronal differentiation of neural stem/progenitor cells transplanted into spinal cord injured rats. PLoS One submitted.

245. Martens DJ, Seaberg RM, van der Kooy D (2002) In vivo infusions of exogenous growth factors into the fourth ventricle of the adult mouse brain increase the proliferation of neural

122

progenitors around the fourth ventricle and the central canal of the spinal cord. Eur J Neurosci 16: 1045-1057.

246. Mothe AJ, Tator CH (2005) Proliferation, migration, and differentiation of endogenous ependymal region stem/progenitor cells following minimal spinal cord injury in the adult rat. Neuroscience 131: 177-187.

247. Aubert-Pouessel A, Venier-Julienne MC, Clavreul A, Sergent M, Jollivet C, et al. (2004) In vitro study of GDNF release from biodegradable PLGA microspheres. J Control Release 95: 463-475.

248. Cao QL, Howard RM, Dennison JB, Whittemore SR (2002) Differentiation of engrafted neuronal-restricted precursor cells is inhibited in the traumatically injured spinal cord. Exp Neurol 177: 349-359.

249. Tang BL, Low CB (2007) Genetic manipulation of neural stem cells for transplantation into the injured spinal cord. Cell Mol Neurobiol 27: 75-85.

250. Rossi SL, Nistor G, Wyatt T, Yin HZ, Poole AJ, et al. (2010) Histological and functional benefit following transplantation of motor neuron progenitors to the injured rat spinal cord. PLoS One 5: e11852.

251. Aizawa Y, Leipzig N, Zahir T, Shoichet M (2008) The effect of immobilized platelet derived growth factor AA on neural stem/progenitor cell differentiation on cell-adhesive hydrogels. Biomaterials 29: 4676-4683.

252. Nakashima K, Takizawa T, Ochiai W, Yanagisawa M, Hisatsune T, et al. (2001) BMP2- mediated alteration in the developmental pathway of fetal mouse brain cells from neurogenesis to astrocytogenesis. Proc Natl Acad Sci U S A 98: 5868-5873.

253. Mena MA, Casarejos MJ, Bonin A, Ramos JA, Garcia Yebenes J (1995) Effects of dibutyryl cyclic AMP and retinoic acid on the differentiation of dopamine neurons: prevention of cell death by dibutyryl cyclic AMP. J Neurochem 65: 2612-2620.

254. Kim H, Tator C, Shoichet M (2011) Chitosan implants in the rat spinal cord: biocompatibility and biodegradation. J Biomed Mater Res A Accepted.

255. Johnson PJ, Tatara A, Shiu A, Sakiyama-Elbert SE (2009) Controlled release of neurotrophin-3 and platelet derived growth factor from fibrin scaffolds containing neural progenitor cells enhances survival and differentiation into neurons in a subacute model of SCI. Cell Transplant.

256. Mothe AJ, Kulbatski I, van Bendegem RL, Lee L, Kobayashi E, et al. (2005) Analysis of green fluorescent protein expression in transgenic rats for tracking transplanted neural stem/progenitor cells. J Histochem Cytochem 53: 1215-1226.

257. Hakamata Y, Tahara K, Uchida H, Sakuma Y, Nakamura M, et al. (2001) Green fluorescent protein-transgenic rat: a tool for organ transplantation research. Biochem Biophys Res Commun 286: 779-785.

123

258. Leipzig ND, Shoichet MS (2009) The effect of substrate stiffness on adult neural stem cell behavior. Biomaterials 30: 6867-6878.

259. Basso DM, Beattie MS, Bresnahan JC (1995) A sensitive and reliable locomotor rating scale for open field testing in rats. J Neurotrauma 12: 1-21.

260. Cooke MJ, Zahir T, Phillips SR, Shah DS, Athey D, et al. Neural differentiation regulated by biomimetic surfaces presenting motifs of extracellular matrix proteins. J Biomed Mater Res A 93: 824-832.

261. Saha K, Keung AJ, Irwin EF, Li Y, Little L, et al. (2008) Substrate modulus directs neural stem cell behavior. Biophys J 95: 4426-4438.

262. Dyke HJ, Montana JG (2002) Update on the therapeutic potential of PDE4 inhibitors. Expert Opin Investig Drugs 11: 1-13.

263. Whitaker CM, Beaumont E, Wells MJ, Magnuson DS, Hetman M, et al. (2008) Rolipram attenuates acute oligodendrocyte death in the adult rat ventrolateral funiculus following contusive cervical spinal cord injury. Neurosci Lett 438: 200-204.

264. Bretzner F, Plemel JR, Liu J, Richter M, Roskams AJ, et al. (2010) Combination of olfactory ensheathing cells with local versus systemic cAMP treatment after a cervical rubrospinal tract injury. J Neurosci Res 88: 2833-2846.

265. Nout YS, Culp E, Schmidt MH, Tovar CA, Proschel C, et al. (2011) Glial restricted precursor cell transplant with cyclic adenosine monophosphate improved some autonomic functions but resulted in a reduced graft size after spinal cord contusion injury in rats. Exp Neurol 227: 159-171.

266. Fouad K, Ghosh M, Vavrek R, Tse AD, Pearse DD (2009) Dose and chemical modification considerations for continuous cyclic AMP analog delivery to the injured CNS. J Neurotrauma 26: 733-740.

267. Tarasenko YI, Gao J, Nie L, Johnson KM, Grady JJ, et al. (2007) Human fetal neural stem cells grafted into contusion-injured rat spinal cords improve behavior. J Neurosci Res 85: 47-57.

268. Olstorn H, Varghese M, Murrell W, Moe MC, Langmoen IA (2011) Predifferentiated brain- derived adult human progenitor cells migrate toward ischemia after transplantation to the adult rat brain. Neurosurgery 68: 213-222; discussion 222.

269. Davies JE, Huang C, Proschel C, Noble M, Mayer-Proschel M, et al. (2006) Astrocytes derived from glial-restricted precursors promote spinal cord repair. J Biol 5: 7.

270. Davies JE, Proschel C, Zhang N, Noble M, Mayer-Proschel M, et al. (2008) Transplanted astrocytes derived from BMP- or CNTF-treated glial-restricted precursors have opposite effects on recovery and allodynia after spinal cord injury. J Biol 7: 24.

124

271. Smith PM, Blakemore WF (2000) Porcine neural progenitors require commitment to the oligodendrocyte lineage prior to transplantation in order to achieve significant remyelination of demyelinated lesions in the adult CNS. Eur J Neurosci 12: 2414-2424.

272. Le Belle JE, Caldwell MA, Svendsen CN (2004) Improving the survival of human CNS precursor-derived neurons after transplantation. J Neurosci Res 76: 174-183.

273. Yang M, Donaldson AE, Marshall CE, Shen J, Iacovitti L (2004) Studies on the differentiation of dopaminergic traits in human neural progenitor cells in vitro and in vivo. Cell Transplant 13: 535-547.

274. de Boer R, Knight AM, Spinner RJ, Malessy MJ, Yaszemski MJ, et al. (2010) In vitro and in vivo release of nerve growth factor from biodegradable poly-lactic-co-glycolic-acid microspheres. J Biomed Mater Res A 95: 1067-1073.

275. Abematsu M, Tsujimura K, Yamano M, Saito M, Kohno K, et al. (2010) Neurons derived from transplanted neural stem cells restore disrupted neuronal circuitry in a mouse model of spinal cord injury. J Clin Invest 120: 3255-3266.

276. Menei P, Montero-Menei C, Venier MC, Benoit JP (2005) Drug delivery into the brain using poly(lactide-co-glycolide) microspheres. Expert Opin Drug Deliv 2: 363-376.

277. Prabaharan M, Mano JF (2005) Chitosan-based particles as controlled drug delivery systems. Drug Deliv 12: 41-57.

278. Chang MY, Juang RS (2006) Stability and reactivity of acid phosphatase immobilized on composite beads of chitosan and ZrO(2) powders. Int J Biol Macromol.

279. Lavertu M, Methot S, Tran-Khanh N, Buschmann MD (2006) High efficiency gene transfer using chitosan/DNA nanoparticles with specific combinations of molecular weight and degree of deacetylation. Biomaterials 27: 4815-4824.

280. Williams LR (1987) Exogenous fibrin matrix precursors stimulate the temporal progress of nerve regeneration within a silicone chamber. Neurochem Res 12: 851-860.

281. Williams LR, Varon S (1985) Modification of fibrin matrix formation in situ enhances nerve regeneration in silicone chambers. J Comp Neurol 231: 209-220.

282. Kalbermatten DF, Kingham PJ, Mahay D, Mantovani C, Pettersson J, et al. (2008) Fibrin matrix for suspension of regenerative cells in an artificial nerve conduit. J Plast Reconstr Aesthet Surg 61: 669-675.

283. Tate CC, Shear DA, Tate MC, Archer DR, Stein DG, et al. (2009) Laminin and fibronectin scaffolds enhance neural stem cell transplantation into the injured brain. J Tissue Eng Regen Med 3: 208-217.

284. Zhong J, Chan A, Morad L, Kornblum HI, Fan G, et al. (2010) Hydrogel matrix to support stem cell survival after brain transplantation in stroke. Neurorehabil Neural Repair 24: 636- 644.

125

285. Ballios BG, Cooke MJ, van der Kooy D, Shoichet MS (2010) A hydrogel-based stem cell delivery system to treat retinal degenerative diseases. Biomaterials 31: 2555-2564.

286. Parr AM, Tator CH (2007) Intrathecal epidermal growth factor and fibroblast growth factor- 2 exacerbate meningeal proliferative lesions associated with intrathecal catheters. Neurosurgery 60: 926-933; discussion 926-933.

287. Lu JM, Wang X, Marin-Muller C, Wang H, Lin PH, et al. (2009) Current advances in research and clinical applications of PLGA-based nanotechnology. Expert Rev Mol Diagn 9: 325-341.

288. Fu K, Pack DW, Klibanov AM, Langer R (2000) Visual evidence of acidic environment within degrading poly(lactic-co-glycolic acid) (PLGA) microspheres. Pharm Res 17: 100- 106.

289. Messaritaki A, Black SJ, van der Walle CF, Rigby SP (2005) NMR and confocal microscopy studies of the mechanisms of burst drug release from PLGA microspheres. J Control Release 108: 271-281.

290. Zhang JX, Chen D, Wang SJ, Zhu KJ (2005) Optimizing double emulsion process to decrease the burst release of protein from biodegradable polymer microspheres. J Microencapsul 22: 413-422.

291. Jain JP, Modi S, Domb AJ, Kumar N (2005) Role of polyanhydrides as localized drug carriers. J Control Release 103: 541-563.

292. Jaganathan KS, Rao YU, Singh P, Prabakaran D, Gupta S, et al. (2005) Development of a single dose tetanus toxoid formulation based on polymeric microspheres: a comparative study of poly(D,L-lactic-co-glycolic acid) versus chitosan microspheres. Int J Pharm 294: 23-32.

293. Sinha VR, Singla AK, Wadhawan S, Kaushik R, Kumria R, et al. (2004) Chitosan microspheres as a potential carrier for drugs. Int J Pharm 274: 1-33.

294. Esposito E, Menegatti E, Cortesi R (2005) Hyaluronan-based microspheres as tools for drug delivery: a comparative study. Int J Pharm 288: 35-49.

295. Chen FM, Wu ZF, Sun HH, Wu H, Xin SN, et al. (2006) Release of bioactive BMP from dextran-derived microspheres: a novel delivery concept. Int J Pharm 307: 23-32.

296. Schriebl K, Lim S, Choo A, Tscheliessnig A, Jungbauer A (2009) Stem cell separation: a bottleneck in stem cell therapy. Biotechnol J 5: 50-61.

297. Cooke MJ, Stojkovic M, Przyborski SA (2006) Growth of teratomas derived from human pluripotent stem cells is influenced by the graft site. Stem Cells Dev 15: 254-259.

298. Lu B, Malcuit C, Wang S, Girman S, Francis P, et al. (2009) Long-term safety and function of RPE from human embryonic stem cells in preclinical models of macular degeneration. Stem Cells 27: 2126-2135.

126

299. Radtke ND, Aramant RB, Petry HM, Green PT, Pidwell DJ, et al. (2008) Vision improvement in retinal degeneration patients by implantation of retina together with retinal pigment epithelium. Am J Ophthalmol 146: 172-182.

300. Falkner-Radler CI, Krebs I, Glittenberg C, Povazay B, Drexler W, et al. (2011) Human retinal pigment epithelium (RPE) transplantation: outcome after autologous RPE-choroid sheet and RPE cell-suspension in a randomised clinical study. Br J Ophthalmol 95: 370- 375.

301. da Cruz L, Chen FK, Ahmado A, Greenwood J, Coffey P (2007) RPE transplantation and its role in retinal disease. Prog Retin Eye Res 26: 598-635.

302. Ayuso-Sacido A, Roy NS, Schwartz TH, Greenfield JP, Boockvar JA (2008) Long-term expansion of adult human brain subventricular zone precursors. Neurosurgery 62: 223-229; discussion 229-231.

303. Pollock K, Stroemer P, Patel S, Stevanato L, Hope A, et al. (2006) A conditionally immortal clonal stem cell line from human cortical neuroepithelium for the treatment of ischemic stroke. Exp Neurol 199: 143-155.

304. Davies SJ, Goucher DR, Doller C, Silver J (1999) Robust regeneration of adult sensory axons in degenerating white matter of the adult rat spinal cord. J Neurosci 19: 5810-5822.

305. Ramon-Cueto A, Cordero MI, Santos-Benito FF, Avila J (2000) Functional recovery of paraplegic rats and motor axon regeneration in their spinal cords by olfactory ensheathing glia. Neuron 25: 425-435.

306. Mauna JC, Miyamae T, Pulli B, Thiels E (2010) Protein phosphatases 1 and 2A are both required for long-term depression and associated dephosphorylation of cAMP response element binding protein in hippocampal area CA1 in vivo. Hippocampus.

307. Peters M, Bletsch M, Catapano R, Zhang X, Tully T, et al. (2009) RNA interference in hippocampus demonstrates opposing roles for CREB and PP1alpha in contextual and temporal long-term memory. Genes Brain Behav 8: 320-329.

308. Pollauf EJ, Berkland C, Kim KK, Pack DW (2005) In vitro degradation of polyanhydride/polyester core-shell double-wall microspheres. Int J Pharm 301: 294-303.

309. Rolls A, Shechter R, Schwartz M (2009) The bright side of the glial scar in CNS repair. Nat Rev Neurosci 10: 235-241.

310. Wislet-Gendebien S, Bruyere F, Hans G, Leprince P, Moonen G, et al. (2004) Nestin- positive mesenchymal stem cells favour the astroglial lineage in neural progenitors and stem cells by releasing active BMP4. BMC Neurosci 5: 33.

311. Liu SY, Zhang ZY, Song YC, Qiu KJ, Zhang KC, et al. (2004) SVZa neural stem cells differentiate into distinct lineages in response to BMP4. Exp Neurol 190: 109-121.

127

312. Sahni V, Kessler JA (2010) Stem cell therapies for spinal cord injury. Nat Rev Neurol 6: 363-372.

313. Kim H, Cooke MJ, Shoichet MS (2011) Recreating a permissive microenvironment for stem cell transplantation in spinal cord and brain repair. Trends in Biotechnology: In preparation.

314. Kim H, Wosnick JH, Yu LM, Shoichet MS (2007) Neural Tissue Engineering: Fundamental and Translational Approaches. In: Mao J, Vunjak-Novakovic G, Mikos A, Atala A, editors. Translational Approaches in Tissue Engineering and Regenerative Medicine. Norwood, MA: Artech House.

128

Appendix A: Abbreviations

BBB: Basso, Beattie, & Bresnahan (Locomotor Scale)

BBB: Blood-Brain Barrier

CNS: Central Nervous System

CSF: Cerebrospinal Fluid

CREB: Cyclic-AMP Responsive Element-Binding dbcAMP: Dibutyryl cyclic adenosine monophosphate

DD: Degree of Deacetylation

EE: Encapsulation Efficiency

EGF: Epidermal Growth Factor

FBS: Fetal Bovine Serum

FGF2: Fibroblast Growth Factor-2

MS: Microsphere

Mw: Molecular Weight

NSPC: Neural Stem/Progenitor Cell

PKA: Protein Kinase A

PLGA: Poly(lactic-co-glycolic acid)

SCI: Spinal Cord Injury

SEM: Scanning Electron Microscopy

129

Appendix B: Additional Figures

Figure B.1: In vitro degradation of different preparation of chitosan in 1mg/ml lysozyme solution at 37oC.

130

131

132

Figure B.4: In vitro measurement of cell viability as assessed by double stranded DNA content. Fibrin supports the proliferation and survival of NSPCs over one week in culture.

133

134

Figure B.6: Longitudinal sections showing endogenous axonal regeneration (betaIII- tubulin, red) from the rostral spinal cord stump through the bridge and aborting at the caudal bridge/spinal cord interface. This pattern was consistent in all treatment groups.

135

Sections of the introduction were adapted from review articles co-authored by Howard Kim [39,313,314], with copyright permissions granted by the respective publishers. Only sections written exclusively by Howard Kim were used in this thesis.

Chapters 2, 3, and 4 were adapted from published or accepted original research articles. Proper copyright permissions were granted by respective publishers. These works were primarily conducted and written by Howard Kim.