Neural Biomimetic Materials for Investigating Cell Behaviors

in 3D

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

Shreyas S. Rao, B.E., M.S.

Graduate Program in Chemical and Biomolecular Engineering

The Ohio State University

2012

Dissertation Committee:

Dr. Jessica O. Winter, Advisor

Dr. Stuart L. Cooper

Dr. John J. Lannutti

Dr. Atom Sarkar

Copyright by

Shreyas S. Rao

2012

Abstract

Central nervous system (CNS) insult because of neurodegenerative disorders or cancer can be devastating. To restore lost neuronal function, prosthetic devices are commonly employed; however achieving robust tissue-electrode interfacing has been difficult because of mismatch in device-tissue properties. Likewise, CNS glial cancers are challenging to treat because of our limited understanding of cancer cell behaviors. To address these challenges, biomimetic materials that can serve as coatings to improve the tissue-electrode interface or as three dimensional (3D) physiologically relevant in vitro models to examine cancer behaviors are needed. This dissertation, therefore, has developed and evaluated 3D biomimetic materials composed of hydrogels, electrospun nanofibers, or a combination of the two.

To enhance the tissue-prostheses interface, synthetic poly(ethylene) glycol based- biomimetic hydrogel coatings endowed with adhesion molecules (polylysine) were developed. Polylysine modified hydrogels were shown to promote better cell adhesion over unmodified controls. Coatings were stable for at least four weeks in vitro suggesting that these biocompatible hydrogels hold potential to enhance the stability of chronic neural interfaces.

In addition to coatings, biomimetic hydrogels were also evaluated as in vitro tumor cell culture models to understand glioblastoma multiforme (GBM) glial cancer cell behaviors

ii in 3D that closely mimic in vivo behaviors as opposed to traditionally studied two dimensional behaviors. Using experimental and computational techniques, we demonstrated that edge effects or mechanical cues resulting from a rigid support-soft hydrogel interface significantly influence GBM morphology, spreading, elongation, migration, and actin organization in a 3D Matrigel biomaterial. These results have import for hydrogel-based, 3D cell culture and suggest that such inherent mechanical gradients should be considered while evaluating 3D cell behaviors.

Next, we developed composite hydrogels composed of collagen and hyaluronic acid

(HA), one of the major components of the brain tumor microenvironment. To mimic increased HA levels observed in GBM tumors, composite hydrogels with increasing HA content were synthesized and characterized. Patient derived GBM cell morphology, spreading, and migration were all strongly dependent on HA density in 3D with higher compositions promoting little or no migration. These findings suggest that the interplay of these hydrogel components guide cell behavior in 3D.

To mimic white matter topography, a major GBM migration highway, we developed an electrospun nanofiber platform and performed a comprehensive investigation of the influence of mechanics and chemistry on GBM behaviors utilizing core-shell electrospinning. Modulating nanofiber mechanics using different polymers (gelatin, polyethersulfone, polydimethylsiloxane) in the ‘core’ with a common poly (ε- caprolactone) (PCL) ‘shell’ revealed GBM elongation, migration, focal adhesion kinase and myosin light chain 2 expression sensitivity to nanofiber mechanics. Similarly, modulating nanofiber chemistry using HA, collagen, and Matrigel as a ‘shell’ on PCL

iii

‘core’ nanofibers revealed GBM sensitivity to HA, specifically, in which a negative effect on migration was observed.

To incorporate multiple cues and more importantly, mimic multiple key features of the in vivo microenvironment within a single 3D system, an integrated nanofiber-hydrogel model was also developed. Preliminary results indicate that GBMs align with nanofibers within 3D hydrogels allowing us to further investigate highly directed migration processes in 3D. Taken together, we have identified several factors influencing tumor cell behaviors in 3D by developing novel in vitro models as well as developed biomimetic coatings that can potentially enhance neural prosthesis biocompatibility. Thus, these findings should have far reaching implications in neural engineering and oncology with the potential for clinical translation.

iv

Dedication

To my late grandparents, my parents, and Almighty God.

v

Acknowledgments

I would like to acknowledge all my mentors, colleagues and friends who provided me with great advice and support throughout my graduate life. Firstly, I would like thank

Prof. Jessica O. Winter for her excellent mentorship as my thesis advisor. In the last five years, I have learnt a great deal from her. She has always encouraged us to think out of the box, think big and positive and come up with solutions that will potentially have a great impact on the scientific community. I cannot be more thankful and I must admit that it has been an honor to work with you.

I would also like to thank my other mentors, Dr. Atom Sarkar and Prof. John J. Lannutti.

Dr. Sarkar has been a great clinical mentor answering all my clinical questions and providing crucial clinical inputs to my projects. I also appreciate help and invaluable suggestions from Prof. Lannutti on the work with electrospun fibers who was virtually a

“second” advisor for small period during Prof. Winter’s absence during later half of my graduate life. Also, I would like thank Prof. Mariano Viapiano, for taking out time to answer some of my neurobiology related questions and also helping with molecular biology techniques described in this work. I look forward to interacting with all of you in the future. I would sincerely like to thank all my qualifier and candidacy committee members, Prof. Stuart L. Cooper, Prof. John J. Lannutti, Prof. Jeffery J. Chalmers, Dr.

vi

Atom Sarkar and Prof. Jessica O. Winter for providing critical inputs on my research. I would also like to thank Prof. David Wood for all his help and career advice.

I would also like to thank all members of the Winter lab who have helped me in my research endeavors. First and foremost, I would like to thank Dr. Ning Han and

Dhananjay Thakur who have been great colleagues providing research advice and help when I first started as a graduate student in the lab in 2008. Thanks Dhananjay for those stimulating research discussions in our office. I would like to thank my awesome undergraduate researchers Kunal Parikh, John Larison, Alex Hissong, and Caroline

Dahlem for providing experimental help as well as for taking on new research directions.

I would also like to thank Dr. Shuang Deng, Dr. Gang Ruan, Dr. Jianquan Xu, Joe

Grodecki, Aaron Short, Kalpesh Mahajan, Jenny Dorcena, Craig Buckley and Qirui Fan for making my stay in the lab enjoyable.

My thanks also goes to Sarah Bentil and Dr. Rebecca Dupaix for providing their FEM simulation expertise to my projects and helping me understand the mechanics of hydrogels using finite element models. I would like to thank the Lannutti lab members

Dr. Jed Johnson, Carol Lee, Tyler Nelson, Ruipeng Xue and Jason Drexler (Nanofiber

Solutions) for providing me with electrospun fibers for my research. I would also like to thank the Sarkar lab members, Jessica DeJesus, Lisa Denning and Kristin Dahl for all their help in confocal microscopy and related experiments. I would also like to thank Dr.

Sara Cole (Campus Microscopy and Imaging Facility (CMIF), OSU) for her help and assistance with confocal reflectance microscopy and Henk Colijn (Campus Electron

vii

Optics Facility (CEOF), OSU) for help and assistance with transmission electron microscopy.

I would like to thank all my Columbus friends for making me feel at home during my graduate study. More specifically, I would like to thank my roommates Kartik and

Ashutosh, colleagues Preshit and Shweta for spending and sharing 5 years of graduate life both as colleagues and friends. Thanks to Dr. Vikas Khanna and Dr. Somnath Sinha for making my initial years in Columbus enjoyable. I would also like to acknowledge my gym buddies Ganita, Varun, and Anshuman. Gymming at the Jesse Owens Recreation

Center could not have been more fun! Also, my special thanks to Dr. Khanna, Kartik,

Archana, and Harshit for being a positive sounding board and providing advice in times of need. Thanks Guys!

I would also like to acknowledge financial assistance from the National Science

Foundation (NSF), Women in Philanthropy at the Ohio State University, fellowship support via a Pelotonia Graduate Fellowship and a Joseph H. Koffolt Graduate

Fellowship. I am particularly indebted to the Pelotonia Fellowship program for making these last two years of graduate study enriching via symposiums enabling efficient exchange of highly inter-disciplinary research ideas in the cancer field.

Last but not the least; I would like thank my parents, Mr. Subhaschandra Rao and Mrs.

Subhashini Rao and my brother, Sandesh Rao for their unconditional love and support.

My heartfelt thanks and gratitude also goes to my late grandparents Mr. Srinivas Rao and

Mrs. Indira Rao for inculcating in me a scientific curiosity since my early days.

viii

Vita

September, 1985 ...... Born in Mumbai, India

June, 2001 ...... Little Angels High School, Mumbai, India

June, 2003 ...... Swami Vivekanand Junior College,

Mumbai, India

June, 2007 ...... B.E., Chemical Engineering, R. V. College

of Engineering, Visvesvaraya Technological

University, India

September, 2007-March 2008 ...... Graduate Research Associate

The Ohio State University

April, 2008- September, 2009 ...... Joseph H. Koffolt Graduate Fellow

The Ohio State University

September, 2009-Decemeber, 2010 ...... Graduate Research Associate

The Ohio State University

January, 2011- ...... Pelotonia Graduate Fellow

The Ohio State University

ix

Publications

1. K. S. Parikh, S. S. Rao, L. B. Zimmerman, H. Ansari, L. J. Lee, S. A. Akbar, J. O. Winter (2012). Ceramic nanopatterned surfaces to explore the effects of nanotopography on cell attachment. Materials Science and Engineering:C Materials for Biological Applictions, 32(8):2469-2475. 2. S. S. Rao, S. Bentil, J. DeJesus, J. Larison, A. Hissong, R. Dupaix, A. Sarkar, J. O. Winter (2012). Inherent interfacial mechanical gradients in 3D hydrogels influence tumor cell behaviors. PLoS ONE, 7(4):e35852. 3. N. Han, S. S. Rao, J. Johnson, K. S. Parikh, P. A. Bradley, J.J. Lannutti, J. O. Winter (2011). Hydrogel-electrospun fiber mat composite coatings for neural prostheses. Frontiers in Neuroengineering, 4(2):1-8. 4. S. S. Rao, N. Han, J. O. Winter (2011). Polylysine modified PEG-based hydrogels to enhance the neuro-electrode interface. Journal of Biomaterials Science: Polymer Edition, 22(4):611-625. 5. S. S. Rao, J. O. Winter (2009). Adhesion molecule-modified biomaterials for neural tissue engineering, Frontiers in Neuroengineering, 2(6): 1-14. (Invited Article) 6. G.M. Madhu, M. A. L. Antony Raj, K. V. K. Pai, S Rao (2007). Photo degradation of methylene blue dye using UV/BaTiO3, UV/H2O2 and UV/BaTiO3/H2O2 oxidation processes, Indian Journal of Chemical Technology, 14: 139-144. 7. G.M. Madhu, M.A. L. Antony Raj, K. V. K. Pai, S Rao (2006). Photo catalytic degradation of Orange III, Chemical Products Finder, The Journal of Materials & Equipment for the Process Industries, 25(2):19-24.

Fields of Study

Major Field: Chemical and Biomolecular Engineering

Studies in:

Tissue Engineering Bioengineering Biomechanics and Mechanobiology Biomaterials Oncology

x

Table of Contents

Abstract ...... ii

Dedication ...... v

Acknowledgments...... vi

Vita ...... ix

List of Tables ...... xxi

List of Figures ...... xxiii

Chapter 1: Introduction ...... 1

1.1 CNS disorders resulting from loss of neuronal functions ...... 3

1.2 CNS disorders resulting from cancer ...... 7

1.3 Dissertation overview ...... 14

Chapter 2: Adhesion Molecule-Modified Biomaterials for Neural Engineering ...... 20

2.1 Introduction ...... 20

2.2 Adhesion molecules (AMs) ...... 23

2.3 Approaches to biomaterial modification ...... 26

2.3.1 Surface deposition ...... 26

xi

2.3.2. Blending ...... 27

2.3.3. Electrostatic attachment ...... 30

2.3.3.1. Layer-by-Layer (LbL) technique (using polyelectrolytes) ...... 30

2.3.3.2. Electrochemical polymerization (using conducting polymers) ...... 33

2.3.4 Covalent attachment ...... 38

2.4 Patterning AM-modified biomaterials ...... 47

2.5 Conclusions ...... 51

Chapter 3: Toward 3D Biomimetic Models for Investigating Glioblastoma Multiforme

Tumor Cell Behaviors ...... 53

3.1 Introduction ...... 54

3.2 Normal brain versus cancer brain ...... 56

3.3 Modeling tissue architecture in 2D ...... 59

3.3.1 Monolayer wound healing assay (gap assay) ...... 59

3.3.2 Microliter scale migration assay ...... 60

3.3.3 Chamber assays ...... 61

3.4 Animal derived models ...... 62

3.4.1 Brain slice assays ...... 63

3.4.2 Confrontational tissue assays ...... 63

3.4.3 Tumor xenograft models ...... 64

xii

3.5 Making the transition to 3D: Bridging the gap between 2D and animal models .... 65

3.5.1 Synthetic biomaterials ...... 66

3.5.2 Natural biomaterials ...... 67

3.6 Potential benefits of 3D brain tissue models over conventional 2D models:

Concluding remarks ...... 73

Chapter 4: Polylysine modified PEG-hydrogels: Biomimetic Coatings to Enhance the

Neural Tissue-Electrode Interface ...... 77

4.1 Introduction ...... 78

4.2 Materials and methods ...... 80

4.2.1 Preparation of Acryl-PEG-Polylysine (Acryl-PEG-PL) ...... 80

4.2.2 Preparation and characterization of Poly (ethylene glycol)-poly (caprolactone)

(PEG-PCL) copolymers ...... 81

4.2.3 Hydrogel formation and quantification of polylysine attachment ...... 82

4.2.4 PC12 cell culture ...... 83

4.2.5 Quantification of cell adhesion using calcein-AM staining ...... 83

4.2.6 PEG-PCL electrode adhesion ...... 84

4.3 Results ...... 85

4.3.1 PL conjugation to PEG hydrogels ...... 85

4.3.2 PC12 cell adhesion on PEG hydrogels ...... 87

xiii

4.3.3 PEG-PCL hydrogel coatings ...... 88

4.3.3.1 Synthesis and characterization ...... 89

4.3.3.2 PEG-PCL electrode adhesion ...... 92

4.4 Discussion ...... 94

4.5 Conclusions ...... 97

Chapter 5: Inherent Interfacial Mechanical Gradients in 3D Hydrogels Influence

Glioblastoma Multiforme Tumor Cell Behaviors ...... 99

5.1 Introduction ...... 100

5.2 Materials and methods ...... 102

5.2.1 Ethics statement ...... 102

5.2.2 Modeling ...... 102

5.2.3 OSU-2 cell isolation and in vitro cell culture ...... 103

5.2.4 OSU-2 cell seeding in BD Matrigel ...... 105

5.2.5 OSU-2 morphology and cell spreading characterization in 3D Matrigel ...... 105

5.2.6 Immunostaining for actin in 3D Matrigel ...... 106

5.2.7 Real time cell tracking in 3D Matrigel ...... 106

5.2.8 Statistical analysis...... 107

5.3 Results ...... 107

5.3.1 Modeling the substrate/gel interface ...... 107

xiv

5.3.2 OSU-2 cell spreading in 3D Matrigel ...... 110

5.3.3 OSU-2 intracellular morphology in 3D Matrigel (actin organization) ...... 117

5.3.4 OSU-2 cell migration in 3D Matrigel ...... 118

5.4 Discussion ...... 120

5.5 Conclusions ...... 125

Chapter 6: Collagen-Hyaluronan Composite Hydrogels: 3D Extracellular Matrix Mimics of the Glioblastoma Multiforme Tumor Microenvironment ...... 126

6.1 Introduction ...... 127

6.2 Materials and methods ...... 130

6.2.1 Cell culture ...... 130

6.2.1.1 Patient tumor derived OSU-2 cell culture ...... 130

6.2.1.2 Normal (non-cancerous) astrocyte culture ...... 131

6.2.2 3D cell encapsulation in collagen-HA composite hydrogels ...... 131

6.2.3 Characterization of composite hydrogels ...... 134

6.2.3.1 Rheological characterization ...... 134

6.2.3.2 Confocal reflectance microscopy ...... 135

6.2.3.3 Scanning electron microscopy ...... 135

6.2.4 OSU-2 morphology analysis and cell spreading in collagen-HA composite

hydrogels ...... 136

xv

6.2.5 Real time OSU-2 cell tracking in 3D composite hydrogels ...... 136

6.2.6 Statistical analysis...... 137

6.3 Results and discussion ...... 137

6.3.1 Composite hydrogel modulus ...... 137

6.3.2 Composite hydrogel micro-architecture ...... 139

6.3.3 OSU-2 and normal astrocyte behaviors in 3D composite hydrogels ...... 141

6.3.4 OSU-2 migration in 3D composite hydrogels ...... 146

6.4 Conclusions ...... 151

Chapter 7: White Matter Tract Topography-Mimetic Biomaterial Platform for Examining

Glioblastoma Multiforme Tumor Cell Behaviors In Vitro ...... 153

7.1 Introduction ...... 154

7.2 Materials and methods ...... 158

7.2.1 Preparation of pure PCL and aligned core-shell nanofibers ...... 158

7.2.2 Morphological, surface and mechanical characterization of aligned PCL and

core-shell nanofibers ...... 161

7.2.2.1 Scanning electron microscopy (SEM) ...... 161

7.2.2.2 Fiber diameter, fiber density and fiber alignment ...... 161

7.2.2.3 Mechanical properties of nanofibers using tensile testing ...... 162

7.2.2.4 Surface properties of PCL and core-shell nanofibers ...... 162

xvi

7.2.2.5 Characterization of core-shell structure using transmission electron

microscopy (TEM) ...... 163

7.2.3 Patient derived OSU-2 cell culture ...... 163

7.2.4 Analysis of cell adhesion on PCL and core-shell nanofibers ...... 163

7.2.5 Morphological analysis of OSU-2 Cells on PCL and core-shell nanofibers. . 165

7.2.6 Analysis of single cell migration using time lapse confocal imaging on PCL

and core-shell nanofibers ...... 165

7.2.7 Analysis of FAK/MLC2 signaling expression using western blotting on PCL

and core-shell nanofibers ...... 166

7.2.8 Analysis of matrix metalloproteinase (MMP) expression using quantitative

polymerase chain reaction (PCR) ...... 167

7.2.9 Statistical analysis...... 167

7.3 Results ...... 167

7.3.1 Characterization of PCL nanofibers and core shell nanofibers ...... 167

7.3.2 OSU-2 cell adhesion on PCL nanofibers and core-shell nanofibers ...... 172

7.3.3 OSU-2 cellular morphology on PCL nanofibers and core-shell nanofibers ... 174

7.3.4 OSU-2 single cell migration on PCL nanofibers and core-shell nanofibers .. 177

7.3.5 FAK/MLC2 signaling and MMP gene expression on PCL nanofibers and core-

shell nanofibers ...... 180

7.4 Discussion ...... 184 xvii

7.5 Conclusions ...... 190

Chapter 8: Incorporating Aligned Electrospun Nanofibers in 3D Hydrogel Systems:

Mimicking White Matter Tracts in Neural Extracellular Matrix ...... 192

8.1 Introduction ...... 192

8.2 Materials and methods ...... 195

8.2.1 Cell culture ...... 195

8.2.1.1 OSU-2 cell culture ...... 195

8.2.1.2 NIH 3T3 cell culture ...... 195

8.2.2 Preparation and characterization of aligned electrospun PCL nanofibers ...... 195

8.2.3 Preparation of aligned electrospun nanofiber-hydrogel–cell composite 3D

cultures...... 195

8.2.4 Imaging of cell-aligned nanofiber-hydrogel composite constructs using epi-

fluorescence microscopy ...... 198

8.3 Results and discussion ...... 198

8.4 Conclusions ...... 203

Chapter 9: Conclusions and Future Work ...... 205

9.1 Conclusions ...... 205

9.2 Future directions ...... 208

9.2.1 Future directions on neural prosthesis/tissue interfaces ...... 208

xviii

9.2.1.1 Combining soluble factor release, tethered cues and topographical cues to

develop multifunctional neural biomaterials for preliminary in vivo examination.

...... 209

9.2.1.2 Developing and examining 3D tissue models of glial scarring ...... 210

9.2.2 Future directions on 3D tissue biomimetics for glioma migration ...... 210

9.2.2.1 Investigating influence of growth factors implicated in glioma invasion in

3D ...... 212

9.2.2.2 Investigating effect of fluid flow on GBM behaviors in 3D using hydrogel

based microfluidic platforms ...... 212

9.2.2.3 Investigating the effect of electrical stimulation on the behavior of

glioblastomas using aligned electrospun nanofiber platforms ...... 213

9.2.2.4 Identification of specific pathways/signaling cascades and potential

molecular targets for glioblastomas in 3D ...... 214

9.2.2.5 Evaluation of different anti-cancer drug sensitivities in 3D ...... 215

9.2.2.6 Detailed investigations on comparison of normal and tumor cell behaviors

in 3D settings...... 215

9.2.2.7 Clinical correlations of patient derived tumor cell migration using

physiologically relevant in vitro models to predict survival outcomes...... 216

9.2.2.8 Evaluation of neuronal/glia biology in 3D settings using hydrogel and/or

electrospun fiber based platform biomaterials with physiological relevance. .... 217

xix

9.2.2.9 Translating 3D biomimetic models to explore behaviors of other tumor

cell types ...... 217

Bibliography ...... 219

Appendix: Experimental Protocols ...... 255

xx

List of Tables

Table 1. Classification of diffusively infiltrative astrocytic tumors and their associated features...... 8

Table 2. Differences between Primary and Secondary GBMs...... 10

Table 3. Advantages and disadvantages of electrostatic attachment techniques...... 38

Table 4. Covalent coupling of AM with biomaterials. Abbreviations: (DRG) Dorsal Root

Ganglion, (SCG) Superior Cervical Ganglion, (SC) Schwann Cells, (RC) Rat cortical

Neurons, (MC) Mouse Cortical Neurons, (EHC) Embyronic HippoAMpal Neurons,

(ERGC) Embryonic Retinal Ganglion Cells, (GB) Glioblastoma, (SD) Sprague Dawley,

(F) Female, (M) Male ...... 46

Table 5. Composition of the brain ECM. Normal versus Cancer [159, 160] ...... 57

Table 6. GBM interactions with natural and synthetic biomaterials with specific observations. Species: Hu=Human, M= Mice, R= Rat; Tumor Cell Model: S= single cells, A= tumor aggregates/spheroids; Culture Type: H= Hydrogel, F= Electrospun Fiber;

Culture Dimensionality: 2D = Cells cultured on TCPS/glass, 2.5D = Cells cultured on top of hydrogels/electrospun fibers, 3D = Cells encapsulated in hydrogels; Assay Type: S =

Static, End point based D = Dynamic, Time Lapse Imaging; *Tumor initating cells implanted into mice to generate tumors and tumor explants cultured on a biomaterial. f indicates functional dependence...... 71

xxi

Table 7. Concentration of bound and entrapped PL (Reported as Average ± SD) ...... 87

Table 8. Composition of composite hydrogels. Superscript 1 indicates composition used for cell studies and superscript 2 indicates composition used for confocal reflectance microscopy for Collagen-IV based compositions. Collagen-IV used at manufacturer supplied concentration...... 133

Table 9. Summary of electrospinning conditions used for nanofiber samples...... 160

Table 10. Micro-structural and mechanical characterization of PCL and core-shell nanofibers with altered cores but an identical surface chemistry (PCL). All nanofibers examined displayed a contact angle of 0˚, indicative of complete wetting and uniform surface chemistry...... 169

Table 11. Micro-structural characterization of core-shell nanofibers with altered surface chemistries and an identical core (PCL). All nanofibers examined displayed a contact angle of 0˚, indicative of complete wetting...... 171

Table 12. Protein expression levels quantified via densitometry from western blots as a function of nanofiber modulus...... 181

Table 13. Protein expression levels quantified via densitometry from western blots for

PCL versus PCL-HA nanofibers ...... 183

xxii

List of Figures

Figure 1. Schematic of the CNS microenvironment with relevant cell types...... 2

Figure 2. CNS response to injury. Figure taken from [9]...... 4

Figure 3. Response to an electrode array post implantation. Figure taken from

(http://knol.google.com/k/neural-prosthesis-brain-interactions#)...... 6

Figure 4. Brain mimetic hydrogel based-coatings on electrode surfaces. AM denotes adhesion molecules incorporated into hydrogels...... 7

Figure 5. Histology Images of Astrocytomas. (A) Example of a WHO grade II

Astrocytoma (Hematoxylin and Eosin (H&E staining), × 400). (B) Example of a WHO grade III AA (H&E staining, × 400). (C) Example of a WHO grade IV GBM showing microvascular proliferation (H&E staining, × 200). (D) Example of a WHO grade IV showing nuclear pleomorphism, hyper chromaticity and necrotic regions (H&E staining,

× 200). Figure taken from [48]...... 9

Figure 6. Magnetic Resonance Imaging (MRI) of bilateral glioblastoma with connection via the corpus callosom. Image courtesy of Dr. Atom Sarkar’s lab...... 11

Figure 7. Frequent alterations observed in signaling pathways in GBM tumors (A)

Receptor Tyrosine Kinase (RTK), RAS, and phosphoinositol-3-kinase (P13K) signaling

(B) P53 tumor suppressor signaling (C) Retinoblastoma (RB) tumor suppressor signaling.

EGFR= Epidermal Growth Factor Receptor, PDGFRA= Platelet Derived Growth Factor

xxiii

Receptor-A, PTEN= Phosphatase and Tensin Homolog, MET= Mesenchymal-Epithelial

Transition Factor. Figure taken from [51]...... 13

Figure 8. Receptor-mediated cell binding. Figure adapted from http://www.mun.ca/biology...... 22

Figure 9. Non-receptor mediated cell binding...... 26

Figure 10. Schematic of LbL technique. Figure taken from [95]...... 31

Figure 11. Immunofluorecence of chick cortical neurons on Si wafers (5 days): (A) (PEI-

Gelatin)8, (B) (PEI-gelatin)-(Chitosan-gelatin)7, (C) (PEI-Laminin)8. Figure courtesy of

Dr. Ravi Bellamkonda, Georgia Institute of Technology and Dr. Wei He, University of

Tennessee...... 33

Figure 12. Electrochemical polymerization. Figure courtesy of Nathalie Guimard and Dr.

Christine E. Schmidt, The University of Texas at Austin...... 35

Figure 13. SEM of Ppy/CDPGYIGSR on a microelectrode site of a neural probe. Figure taken from [104]. Scale bar = 2 µm...... 36

Figure 14. Neuroblastoma cell response on a coated neural probe. Figure taken from [37].

...... 37

Figure 15. Chemical modification mediated by thiol group...... 39

Figure 16. Chemical modification mediated via SMCC crosslinker...... 40

Figure 17. Representative light microscope image of DRG neurons on poly (Dex-MA-co-

AEMA) modified with CGRGDS. Figure taken from [115]...... 41

Figure 18. Chemical modification mediated via CDI chemistry...... 42

Figure 19. Chemical modifications via oxidation with periodate...... 43

xxiv

Figure 20. Chemical modification mediated via EDC chemistry...... 44

Figure 21. Schematic of biomaterial modification techniques...... 48

Figure 22. Hippocampal neuron alignment on surface modified glass [CSIKVAV (200

µm)/PEG (50 µm)] at 20X magnification. Figure taken from [136]...... 49

Figure 23. Clinical presentation of GBM tumors. (A) GBM cells seen around blood vessel periphery (arrow). (B) GBM cells seen below surface of pia mater, indicated by arrow (sub-pial spread). (C) GBM cells migrating along white matter tracts (labeled blue via Luxol staining). Both (A) and (B) are taken from [50] and (C) from [49]...... 59

Figure 24. State of the art cell culture models. (A) 2D Gap assay. (B) 2D Microliter scale migration assay. (C) Transwell insert assay or chamber assay. Both (A) and (B) taken from [175]...... 62

Figure 25. Animal derived models. (A) Brain slice assay. Figure adapted from [171],

GFP = Green fluorescent protein. (B) Confrontational tissue assays, arrows indicate direction of tumor cell infiltration. (C) Rat/Mouse models of cancer (tumor xenografts).

...... 65

Figure 26. Three dimensional cell culture models for GBM migration studies (A)

Hydrogel models used in different configurations. (B) Aligned electrospun nanofiber models...... 70

Figure 27. Schematic of the conjugation procedure...... 83

Figure 28. Conjugation of PL with PEG-DA hydrogels. (N=4 for 1X, N=4 for 0.1X, N=5 for 0.01X, * indicates statistical significance compared to negative control...... 86

xxv

Figure 29. PC12 cell response on PL-modified PEGDA hydrogels. (A) Sample. (B)

Sham. (C) Negative control (No PL). (D) Positive Control (PL)...... 88

Figure 30. Normalized fluorescence for PC12 cell adhesion on experimental PEGDA hydrogels. (N=6) * indicates statistical significance compared to negative control...... 89

Figure 31. Analysis of PEG-PCL polymer. (A) FTIR Spectroscopy. (B) NMR

Spectroscopy. (C) Chemical Structure...... 91

Figure 32. Conjugation of PL to PEG-PCL hydrogels. (A) PL conjugated/entrapped

(Sample). (B) PL entrapped in PEGPCL hydrogel. (C) Negative Control (No PL)...... 92

Figure 33. Retinal Implant Electrode...... 92

Figure 34. Coating integrity (A) before and (B) after insertion into an agarose tissue phantom. Black arrows indicate the edges of the polymer coating...... 93

Figure 35. Static PEG-PCL coating integrity under agarose tissue phantom at time (A) t=

0. (B) 14 days. (C) 28 days. Black arrows indicate the boundary of the hydrogel coating.

(Scale bar = 500 µm)...... 93

Figure 36. Patterned brain mimetic PEG hydrogel coatings. (A) Schematic of patterning technique. (B) Patterned “O” s stained using Rhodamine observed via fluorescence microscopy (C) Patterned “O” s observed using phase contrast microscopy (D) Ohio

State “O” logo...... 97

Figure 37. Schematic of cell isolation procedure...... 104

Figure 38. OSU-2 cells in culture. (A) Hoechst stain labels the nucleus blue; rhodamine-

GFAP (e.g., glial fibrillary acidic protein) labels the cytoskeleton. GFAP is an intermediate protein expressed by astrocytes. GFAP staining was performed to confirm

xxvi astrocytic lineage. (B) Phase contrast image of OSU-2 cells in culture. Scale bar indicates

100 µm...... 104

Figure 39. Mechanics of the gel-glass interface modeled using FEM. (A) Stress contour plots of Matrigel with varying height. Axisymmetric elements used. Von Mises stress is an equivalent stress that includes both normal stress (tension/compression) and shear stress contributions. It is calculated from the stress components acting at each location and gives a convenient way of comparing the overall magnitude of stress in different regions. (B) Stress felt at the Matrigel-glass interface as a function of gel height...... 108

Figure 40. Reaction force vs. 5 μm displacement of the indenter. Insert illustrates a decrease in stiffness with increasing Matrigel height due to a 5 μm indenter displacement.

The stiffness insert is the slope extracted from the displacement vs. reaction force plot using the reaction force experienced by Matrigel when the indenter reaches a displacement of 5 µm...... 109

Figure 41. Still images taken from a Z-stack of fluorescently-labeled cells in a 40% v/v

Matrigel (0-50 µm, step size = 5 µm). (A) Brightfield/fluorescence Z-stack shown as a montage. (B) Rotated views of the Z-stack shown in A. White arrow indicates the same cell, at position 30 µm, which is clearly embedded within the hydrogel...... 111

Figure 42. Images from a brightfield/fluorescence Z-stack of fluorescently-labeled cells in a 40% v/v Matrigel (0-100 µm, step size = 5 µm). White arrow indicates a cell, at position 15 µm, whose edge is in contact with the rigid glass support while the cell body is embedded in the hydrogel. The asterisk indicates a cell, at position 90 µm, fully embedded in the hydrogel...... 112

xxvii

Figure 43. Box plot of individual cell aspect ratios comparing cells in the lowest observation plane (< ~50 µm) in 40% (v/v) Matrigel versus Bare Glass. * indicates statistical significance (p < 0.0001), n = 206 cells for glass, n = 20 for lowest observation plane in 40% (v/v) Matrigel...... 113

Figure 44. OSU-2 cell behaviors as a function of observation plane in 40% (v/v)

Matrigel. (A) Schematic of cell-hydrogel constructs showing morphology observation at different “z” planes. (B) OSU-2 cell area. Representative cell morphologies are shown in the insets. As a result of surface roughness, zero height was set to the first plane of observed cells, which may not necessarily correspond to the substrate surface. (C) OSU-2 morphology at different heights. Representative heights are shown in the chart. Scale bar

= 200 µm. (D) OSU-2 aspect ratio...... 114

Figure 45. OSU-2 cell morphology quantification. (A) OSU-2 cell area and (B) aspect ratio as a function of observation plane in 55% (v/v) Matrigel...... 115

Figure 46. OSU-2 cell morphology quantification. (A) OSU-2 cell area and (B) aspect ratio as a function of observation plane in 70% (v/v) Matrigel...... 116

Figure 47. OSU-2 cell morphology quantification. (A) OSU-2 cell area and (B) aspect ratio as a function of observation plane in 85% (v/v) Matrigel...... 116

Figure 48. OSU-2 cell morphology in 2D Matrigel for all formulations. Scale bar = 200

µm...... 117

Figure 49. OSU-2 actin organization in 3D hydrogels at a higher observation (> ~500

µm) and lower observation (< ~50 µm) plane. Scale bar = 100 µm...... 118

xxviii

Figure 50. OSU-2 cell migration in a representative 40% v/v Matrigel at (A) lower (< ~50

µm) and (B) higher (> ~500 µm) observation planes shown as stills from time lapse microscopy. Time stamp is reported in hours (h). Scale bar = 100 µm...... 119

Figure 51. Quantification of migration speeds (average) of OSU-2 cells at the lowest (<

~50 µm) and highest observation planes (> ~500 µm) investigated. * indicates statistical significance...... 120

Figure 52. OSU-2 and non-cancerous human astrocytes cells in culture. (A) GFAP staining of OSU-2 cells in culture. Hoechst stain labels the nucleus blue, whereas rhodamine-GFAP labels the cytoskeleton red. (B) Phase contrast image of OSU-2 cells in culture. (C) Phase contrast image of non-cancerous astrocytes in culture. Scale bar = 100

µm...... 131

Figure 53. Schematic of encapsulation procedure to create cell laden 3D hydrogels. ... 134

Figure 54. Mechanical characterization of Collagen (I/III) and Collagen (I/III) composite hydrogels. (A) Elastic modulus values reported for various tissues. Figure adapted from

[176]. (B) Elastic modulus of Collagen (I/III)-HA composite hydrogels (N ≥ 3). * indicates statistically significant from collagen controls (p < 0.0001, as reported from

ANOVA)...... 138

Figure 55. Confocal Reflectance Microscopy (CRM) imaging of Collagen (I/III) and

Collagen (I/III)-HA composite hydrogels...... 139

Figure 56. Confocal Reflectance Microscopy (CRM) imaging of Collagen-IV and

Collagen IV-HA composite hydrogels...... 140

xxix

Figure 57. Scanning Electron Microscopy (SEM) imaging of Collagen (I/III), Collagen

(I/III)-HA composite and pure HA hydrogels. Scale bar = 50 µm...... 140

Figure 58. OSU-2 morphologies in Collagen (I/III)-HA composite hydrogels as observed via confocal microscopy. Insets with each image show representative cell morphologies.

Scale bar = 100 µm...... 141

Figure 59. OSU-2 cell areas in 3D composite hydrogels. * indicates pairs that are statistically significant compared to Col (Control) (p < 0.0001, as reported from

ANOVA) ...... 142

Figure 60. OSU-2 cell circularity in 3D composite hydrogels. * indicates pairs that are statistically significant compared to Col (Control) (p < 0.0001, as reported from

ANOVA) ...... 143

Figure 61. Percentage of rounded cells in composite hydrogels. * indicates pairs that are statistically significant compared to Col (Control) (p < 0.0001, as reported from

ANOVA)...... 143

Figure 62. OSU-2 cell morphology in Col-IV and Col-IV-HA composite hydrogels. ... 144

Figure 63. OSU-2 cell adhesion and spreading on Col (I/III) and Col-IV coated surfaces

(~100 µg/ml). (A) OSU-2 cell adhesion was quantified by prelabelling cells with Cell

Tracker Green and seeding them on coated surfaces and reading the fluorescent intensity of adhered cells (after 1 h wash) using a fluorescent plate reader. (B) OSU-2 cell spreading on Col (I/III) and Col-IV coated surfaces (~ 100 µg/ml) was quantified using

Image J. * indicates statistical significance, tissue culture polystyrene (TCPS) served as control in both cases...... 145

xxx

Figure 64. Non-cancerous human astrocyte morphology in Collagen (I/III)-HA composite hydrogels. Arrows indicate small processes extending from the astrocyte cell body. ... 146

Figure 65. OSU-2 cell migration in an example Collagen (I/III)-HA composite hydrogel

(Col-0.2HA) as shown through stills from time lapse microscopy movies. Time stamp reported in hours (h). Scale bar = 100 µm...... 148

Figure 66. Quantification of OSU-2 single cell migration speeds in Collagen (I/III)-HA composite hydrogels. N ≥ 40 individual cells analysed for each condition. * (in blue) indicate statistical significance when compared to Col (Control) (p < 0.0001, as reported from ANOVA). Representative cell morphologies are presented as insets. Red lines within the box indicate mean and black lines indicate median values...... 149

Figure 67. Cell culture well design modified for use with electrospun nanofiber discs. (A)

Part of a 24 well plate with a drilled hole (shown via bidirectional white arrow). (B)

Electrospun nanofiber disc used with this design attached to the bottom of the well using medical adhesive...... 164

Figure 68. Comparison of white matter histology to aligned electrospun nanofibers. (A)

Histology of corpus callosum showing stained white matter tracts with Luxol fast blue.

Image taken from [49]. (B) Aligned electrospun PCL nanofibers as observed via SEM closely mimicking the in vivo topographical features seen in (A)...... 168

Figure 69. Micro-structural features of PCL and core-shell nanofibers observed via SEM.

(A) Gelatin-PCL. (B) PCL. (C) PDMS-PCL. (D) PES-PCL. (E) PCL-Collagen. (F) PCL-

HA. (G) PCL-Matrigel. Scale bar indicates 20 µm. Fast Fourier Transform (FFT) with associated images are shown as insets...... 168

xxxi

Figure 70. FFT analysis of representative SEM images via radial summation of pixels normalized for all nanofiber samples and plotted versus degree...... 170

Figure 71. Transmission electron microscopy imaging of a representative PDMS-PCL core-shell nanofiber. (A) Schematic depicting the structure of core-shell nanofiber. (B)

TEM image of PDMS-PCL core-shell nanofiber showing the PDMS ‘core’ and PCL

‘shell’ surrounding the core. Scale bar = 0.2 µm...... 170

Figure 72. Adhesion of OSU-2 cells on various nanofibers as a function of nanofiber mechanics. No significant differences between samples observed...... 172

Figure 73. Adhesion of OSU-2 cells on PCL nanofibers and nanofibers with identical cores and various biomimetic shells. * indicates statistically significant difference compared to PCL nanofiber...... 173

Figure 74. OSU-2 cell morphologies on various nanofibers examined. (A) Gelatin-PCL.

(B) PCL. (C) PDMS-PCL. (D) PES-PCL. (E) PCL-Collagen. (F) PCL-HA. (G) PCL-

Matrigel. Scale bar indicates in (A) 100 µm. Bidirectional arrow indicates direction of fiber alignment...... 174

Figure 75. Feret diameter analysis of OSU-2 cells as a function of nanofiber mechanics.

N ≥ 142 individual cells analysed for each nanofiber. * and ** indicates statistically significant difference compared to PCL nanofiber. Levels marked by identical number of

* are not significantly different from each other as determined by Tukey-HSD test. .... 175

Figure 76. Feret diameter analysis of OSU-2 cells on PCL and core-shell nanofibers with an identical PCL core and various biomimetic chemistries as shells. N ≥ 199 individual cells analysed for each nanofiber. * and ** indicates statistically significant difference

xxxii compared to PCL nanofiber. Levels marked by identical number of * are not significantly different from each other as determined by Tukey-HSD test...... 176

Figure 77. Cell area analysis of OSU-2 cells on PCL and core-shell nanofibers with an identical PCL core and various biomimetic chemistries as shells. N ≥ 199 individual cells analysed for each nanofiber. * indicates statistically significant difference compared to

PCL nanofiber...... 177

Figure 78. Analysis of single cell migration speed as a function of nanofiber mechanics shown as a box and whisker plot. N ≥ 95 individual cells analysed for each nanofiber.

Blue * and ** indicates statistically significant difference compared to PCL nanofiber.

Levels marked by identical number of * are not significantly different from each other as determined by Tukey-HSD test. Red lines within the box indicate mean and black lines indicate median...... 178

Figure 79. Analysis of single cell migration speed as a function of biomimetic chemistries shown as a box and whisker plot. N ≥ 142 individual cells analysed for each nanofiber.

Blue * indicates statistically significant difference compared to PCL nanofiber. Red lines within the box indicate mean and black lines indicate median...... 180

Figure 80. Analysis of FAK/MLC2 expression via western blotting. (A) pFAK, FAK, pMLC2, MLC2 expression as a function of mechanics. (B) pFAK, FAK, pMLC2, MLC2 expression for PCL versus PCL-HA combination. In both cases, tubulin served as a loading control...... 181

Figure 81. Normalized MMP-2, MMP-9, and MMP-13 gene expression levels for PCL versus PCL-HA nanofibers. * indicates statistical significance...... 184

xxxiii

Figure 82. Schematic of hydrogel and hydrogel-fiber composites examined. (A)

Hydrogels only (B) Aligned electrospun nanofibers (ENF) combined with hydrogels (C)

(left) Aligned ENF “sandwiched” between two hydrogels, ENF not shown, (right) ENF embedded in the gel shown. Note hydrogel placed atop the aligned ENF is reduced in scale compared to the base hydrogel layer to show the presence of nanofibers...... 197

Figure 83. SEM image of aligned PCL nanofibers. Scale bar indicates 50 µm...... 200

Figure 84. OSU-2 cell behaviors on HA based gel-nanofiber composites (A) 2 wt % HA hydrogel (See Figure 82 A) (B) 2 wt % HA hydrogel with PCL nanofiber (See Figure 82

B) (C) PCL nanofiber “sandwiched” between 2 wt % HA base hydrogel and 0.5 wt %

HA hydrogel on top (See Figure 82 C). Left insets show representative cell morphologies. Right insets in B & C, bidirectional arrows show direction of nanofiber alignment...... 200

Figure 85. OSU-2 cell behaviors on HA and Col-HA based gel-nanofiber composites (A)

PCL nanofiber “sandwiched” between 1 wt% HA hydrogel base and Col-0.5HA composite hydrogel on top as described in Chapter 6 (B) PCL nanofiber “sandwiched” between Col-0.5HA composite hydrogels. Left insets show representative cell morphologies. Right insets (bidirectional arrows) show direction of nanofiber alignment.

...... 202

Figure 86. NIH 3T3 cell behaviors on gel-nanofiber composites. (A) 1 wt% HA hydrogel.

(B) 1 wt % Agarose hydrogel. (C) 22 wt% PEG-DMA hydrogel. (D) 1 wt % HA hydrogel-PCL nanofiber composite (E) 1 wt% Agarose hydrogel-PCL nanofiber composite (F) 22 wt% PEG-DMA hydrogel-PCL nanofiber composite. Right inset shows

xxxiv representative cell morphologies in each case. Note that cells on hydrogels are typically rounded. Cells on gel-nanofiber composites mostly displayed elongated bipolar morphologies. Bidirectional arrows (left inset, bottom row) show the direction of alignment of PCL nanofibers...... 203

Figure 87. Schematic of the multifunctional neural biomimetic coating...... 210

Figure 88. Schematic of GBM invasion microenvironment including major highways.

Figure taken from [156]...... 211

Figure 89. Schematic of incorporating fluid flow in ex vivo 3D systems...... 213

Figure 90. Live/Dead staining of circulating tumor cells obtained using a negative depletion technology. (A) Calcein AM (Live, Green) staining and (B) EthD-1 (Dead,

Red). Note the sample may also contain blood cell populations and this is not differentiated in this figure...... 218

xxxv

Chapter 1: Introduction

The central nervous system (CNS) is one of the most fascinating systems in our body and is responsible for integrating and coordinating events of all other body parts. The CNS consists of nerve cells (Neurons that make ~ 8% of the tissue volume and ~ 10% of the cell volume) and glial cells (Glia that make ~ 72% of the tissue volume and ~ 90% of the cell volume) [1]. Clearly, glial cells are a majority population in the CNS outnumbering neurons by a factor of more than 10-50. These cells function as the supporting cells of the

CNS and perform a variety of functions including removal of cell debris during injury and separating and insulating synaptic terminals formed between neurons. Glial cells in the nervous system can be further subdivided into Macroglia and Microglia. Microglia belong to a class of phagocytes that get activated in response to CNS traumas and injury.

Macroglial cells in the CNS include oligodendrocytes and astrocytes (Schwann cells are macroglial cells specific to the peripheral nervous system) [2]. Oligodendrocytes are myelinating cells of the CNS forming myelin sheaths that serve as insulators for axons (a long tubular process from a neuron) facilitating efficient signal transfer for information processing [2, 3]. Astrocytes play important roles in maintaining concentration of potassium (K+) ions by acting as a “sink” for these ions released by neurons that would otherwise interfere with the inter-neuron signaling activity. Further, their long processes termed as “astrocytic feet” terminate on neuronal cells, enabling transport of nutrients, as

1 well as on brain blood vessels, forming a tight junction called the “blood-brain barrier” that serves to protect the brain from infection [2, 4]. Neurons are functional signaling units of the CNS playing crucial roles in signal generation and transfer via short

(dendrites, signal receiving end) and long processes (axons, signal transferring end) [2].

In addition to the cellular components, the remaining ~ 20% of the tissue volume consists of extracellular matrix (ECM) molecules [5, 6] comprised of glycosoaminoglycans

(GAGs), proteoglycans (PGs) and proteins that tightly regulate both neuronal and glial cell behaviors. [Detailed information on the extracellular environment of the CNS is presented in Chapter 3]. A schematic of the CNS microenvironment is shown in figure 1.

Figure 1. Schematic of the CNS microenvironment with relevant cell types.

Several types of insults can significantly impair the functioning of CNS cell types and as a result affect the quality of life. CNS disorders resulting from loss of neuronal functions

(neuron) and those from cancers (glia) are discussed below. 2

1.1 CNS disorders resulting from loss of neuronal functions

Millions of individuals suffer from neuronal disorders such as Parkinson’s disease,

Alzheimer’s diseases, stroke, epilepsy or other traumas (e.g., malignant tumor presentation in the CNS that could result in tissue removal) [7, 8]. However, CNS neurons do not regenerate appreciably [9]. Several factors contribute to this failure.

Numerous glycoproteins present in the CNS microenvironment present inhibitory cues for neural regeneration. In the CNS, macrophage recruitment is delayed by the blood- brain barrier, limiting clearance of myelin, a glycoprotein that impedes regeneration [10-

12]. In addition, CNS injury results in activation of astrocytes that proliferate, resulting in

“reactive astrocytes” that form an inhibitory glial scar (typically composed of cell debris, myelin and cellular components such as oligodendrocytes, astrocytes and microglia as well as fibroblasts, monocytes and macrophages) impeding regeneration [13, 14] (Figure

2).

3

Figure 2. CNS response to injury. Figure taken from [9].

Several strategies have been employed to aid in tissue regeneration following injury.

These range from using guidance therapies (e.g., nerve grafts), biomolecular therapies

(e.g., using neurotrophic factors that promote neuronal survival and growth), cellular therapies (e.g., stem cells), advanced therapies (e.g., using advanced channel guidance fabrication techniques) and combination of the aforementioned techniques [9]. However, despite the promise and encouraging results seen using these strategies, complete functional recovery has not been achieved. To address this issue, neural prostheses, which restore lost electrical signaling by converting an external signal into an electrical pattern transmitted to cells via microelectrodes, have become an essential tool [15] .

These devices typically act as a “bridge” bypassing signals from the damaged neuronal cells [16]. These electrodes could be recording electrodes that record activity from single

4 or groups of neurons or stimulating electrodes designed to stimulate neurons. Neural prostheses have had some success, most notably the cochlear implant [17]. However, two primary factors hampering their efficacy and thereby preventing the formation of a robust neuron-electrode interface are: (a) difficulty in accessing target neuronal tissue

[18] and (b) scarring resulting from the host immune response to implantation [19, 20]

(i.e., biocompatibility issues).

The mechanical trauma of electrode insertion ruptures surrounding cells and blood vessels as well as extracellular matrix molecules [13]. This initiates wound healing processes in the CNS (i.e., immune response), resulting in activated astrocytes and microglia that compromise the efficacy of electrode function. Following this acute response (lasting ~ 1-3 weeks), a chronic immune response in the form of an encapsulating sheath (referred to as “glial scar”) is observed that surrounds the electrode

6-8 weeks after insertion [21-23] (Figure 3). This scar further compromises efficacy by increasing the electrode impedance and the distance between nearby neurons and electrode sites. Also the reduction in neuronal densities around the electrode as a result of trauma, often referred to as the “kill zone” [23-26], contributes. Taken together, all these factors may result in inconsistent performance over time and in some cases electrode failure. Thus, to improve the biocompatibility of these devices and aid in neuronal regeneration, it is crucial to reduce the formation of scar tissue and associated events.

5

Figure 3. Response to an electrode array post implantation. Figure taken from (http://knol.google.com/k/neural-prosthesis-brain-interactions#).

To minimize the immune response to neural electrodes, several strategies have been explored (e.g., alteration in electrode design, such as size, shape, area) [27-31]. For the last several years, research groups, including ours [18, 32-36] have examined the possibility of creating biomimetic materials that mimic the native tissue environment

(i.e., combining tissue engineering approaches with neural prostheses to create a composite material that promotes integration with target neural tissue) to significantly enhance function. Several biomaterials have been developed as electrode coatings, including conducting polymers [37-40], layer-by-layer (LbL) assemblies [41], and polymeric hydrogels [18, 32]). Among these, hydrogel-based coatings have attracted considerable attention because their structural and mechanical properties mimic native brain tissue. Hydrogels are crosslinked, hydrophilic polymeric structures that swell in

6 solution [42, 43]. Whereas strategies have been developed to deliver a variety of biomolecules (i.e., anti-inflammatory drugs, neurotrophic factors) many of these drugs

(i.e., Neurotrophins) have a very short half-life in vivo [44]. Additionally, efficacy is compromised after the supply of drugs is exhausted. To provide multiple cues in addition to drug delivery strategies developed by our group and others (e.g., Figure 4), this dissertation discusses various strategies to incorporate adhesion cues into biomimetic hydrogel materials and presents one possible route using polylysine that could potentially aid in creating stable chronic neuronal interfaces.

Figure 4. Brain mimetic hydrogel based-coatings on electrode surfaces. AM denotes adhesion molecules incorporated into hydrogels.

1.2 CNS disorders resulting from cancer

In addition to loss of neuronal functions, the number of individuals’ succumbing to CNS cancers is also on the rise with over 40,000 cases reported annually [45]. In general, CNS cancers are classified according to the specific location of origin in the central nervous system, and include Gliomas, Meningiomas, Medulloblastomas, Gangliogliomas,

Schwannomas (Neurilemmonmas), Craniopharyngiomas, Chordomas, and Non-Hodgkin lymphomas [46]. Gliomas account for ~ 49% of all CNS cancers [47]. Among gliomas,

7 tumors derived from the astrocytes make up ~75%, with Glioblastoma Multiforme

(GBM) (incidence in the United States is ~2.36 cases/100,000 persons [47]) and

Anaplastic Astrocytomas (AA) accounting for > 78% [48] of these. As per the World

Health Organization (WHO) classification, three types of diffusely infiltrative astrocytic tumors have been identified. These types with their characteristic hallmarks are listed in

Table 1 and their representative histologies are shown in Figure 5.

Tumor Type Characteristic Features [48] Diffuse Astrocytomas (WHO grade II) Mild pleomorphism and minimal increased cellularity Anaplastic Astrocytomas (WHO grade III) Increased cellularity over normal brain, nuclear atypia and mitotic activity Glioblastoma Multiforme and its variants Nuclear pleomorphism, Necrosis and/or (WHO grade IV) endothelial proliferation Table 1. Classification of diffusively infiltrative astrocytic tumors and their associated features.

8

Figure 5. Histology Images of Astrocytomas. (A) Example of a WHO grade II Astrocytoma (Hematoxylin and Eosin (H&E staining), × 400). (B) Example of a WHO grade III AA (H&E staining, × 400). (C) Example of a WHO grade IV GBM showing microvascular proliferation (H&E staining, × 200). (D) Example of a WHO grade IV showing nuclear pleomorphism, hyper chromaticity and necrotic regions (H&E staining, × 200). Figure taken from [48].

GBM, the specific tumor examined in this work, is a primary tumor of the astrocytes (i.e., glial cells in the CNS) [9]). Based on genetic analysis data, GBMs can be further classified into primary and secondary GBMs with a major difference attributed to alterations in unique cell signaling pathways. These differences are highlighted in Table

2.

9

Primary GBMs [48] Secondary GBMs [48] Over expression of epidermal growth factor Over expression of platelet derived (EGFR) receptors, G protein coupled receptor growth factor (PDGF) 26 (GPR26) Loss of heterozygosity (LOH) on chromosome Loss of heterozygosity (LOH) on 10 (10q) chromosome 19q PTEN mutation, deletions in INK4a gene with loss of p14 and p16, deletion of CDKN2A and P53 mutation MDM2 genes Affects individuals > 50 years in age Affects individuals < 45 years in age Table 2. Differences between Primary and Secondary GBMs.

A characteristic feature of this primary brain tumor is its high invasion potential. For instance, it has been shown that these tumors redevelop in the opposing hemisphere of the brain following surgical resection of the tumor-containing hemisphere [49] (e.g.,

Figure 6). Another feature of these tumors is that they do not generally metastasize outside the brain, thereby favorably adapting to the local CNS environment [49]. Current treatment methods such as surgery, radiation and chemotherapy serve to reduce the intensity of disease progression. However, these are not effective as a treatment for a number of reasons. First, surgical attempts fail to completely resect the tumor because of their diffusively infiltrative nature as well as variability associated with location of tumor presentation [50]. In some cases, tumor presentation in regions of brain absolutely necessary for patient survival makes it extremely challenging to resect the tumor by surgical interventions [50]. Second, these tumor cells offer high chemotherapy resistance

[51]. One of the reasons for this is the presence of cells with “cancer stem cell” like characteristics (GBMs were one of the first solid tumors where such features were identified [47]) that over express multidrug resistance proteins [47]. Further, commonly

10 employed chemotherapy drugs in GBM treatments, Temozolomide (marketed as

Temodar) and Carmustine (bischloroethylnitrosourea, BCNU) target cell proliferation pathways as opposed to migration. For instance, Temozolomide is metabolized and results in the formation of the toxin O-methyl guanine, forcing cells to switch to the apoptotic pathway [52], whereas, Carmustine, a nitrosourea-alkylating agent functions by alkylating and crosslinking DNA and RNA, eventually initiating cell cycle arrest [53].

Third, the inherent complexity associated with the tumor is yet another reason for treatment failures.

Figure 6. Magnetic Resonance Imaging (MRI) of bilateral glioblastoma with connection via the corpus callosom. Image courtesy of Dr. Atom Sarkar’s lab.

11

As neurosurgeon, Dr. Eric C. Holland describes in his commentary:

“…glioblastoma is multiforme. It is multiforme grossly, showing regions of

necrosis and hemorrhage. It is multiforme microscopically, with regions of

pseudopalisading necrosis, pleomorphic nuclei and cells, and microvascular

proliferation. And it is multiforme genetically, with various deletions,

amplifications, and point mutations leading to activation of signal transduction

pathways downstream of tyrosine kinase receptors such as EGFR and PDGFR, as

well as to disruption of cell-cycle arrest pathways by INK4a-ARF loss or by P53

mutations associated with CDK4 amplications or RB loss.”[50].

The associated signaling pathways involved in genetic alterations are shown in Figure 7.

Finally, brain tissue is easily prone to damage by chemotherapy and once damaged has a limited capacity to regenerate under normal conditions, as discussed in Section 1.1 [9].

Additionally, most drugs designed to target these tumors fail to cross the blood-brain barrier [54]. Despite recent advancements in these techniques, median survival rate for a patient diagnosed with GBM remains low (~12-15 months) [55]. This, in part, is a consequence of our poor understanding of molecular pathogenesis of GBMs. Therefore, to refine existing techniques and expand present therapeutic options, there is a need to develop newer methods and models to understand the complex behavior of GBMs.

Current methods to study these behaviors mostly employ two dimensional (2D) substrates that do not mimic the complexity of the in vivo brain tissue microenvironment

(more details on these models, their advantages and disadvantages are presented in

Chapter 3).

12

Figure 7. Frequent alterations observed in signaling pathways in GBM tumors (A) Receptor Tyrosine Kinase (RTK), RAS, and phosphoinositol-3-kinase (P13K) signaling (B) P53 tumor suppressor signaling (C) Retinoblastoma (RB) tumor suppressor signaling. EGFR= Epidermal Growth Factor Receptor, PDGFRA= Platelet Derived Growth Factor Receptor-A, PTEN= Phosphatase and Tensin Homolog, MET= Mesenchymal-Epithelial Transition Factor. Figure taken from [51].

To aid either as tissue engineered scaffolds, or as in vitro models to study disease, this dissertation therefore attempts to address these crucial issues by developing biomaterials that mimic several features of the native brain tissue environment. The development of biomimetic materials will not only enable recapitulation of in vivo like behaviors, but also aid in the development of novel therapeutic targets yet unknown because of the lack of 13 biomimetic models. This includes biomaterials to improve the tissue electrode interface and also models to study tumor cell behaviors. This work specifically aims to advance the study of CNS cancers by “borrowing” concepts in tissue engineering to develop improved three dimensional (3D) in vitro disease models, thereby uniting the themes of neural tissue engineering and cancer biology.

1.3 Dissertation overview

This dissertation is divided into 9 chapters. In chapter 2, adhesion molecule-modified biomaterials for neural engineering applications is described. Adhesion molecules are specific proteins, or peptides that encourage cell adhesion by providing cues typically present in the native extracellular matrix. These molecules are often combined with various biomaterial scaffolds using several techniques depending on the application to enhance cell-biomaterial interactions. Several strategies can be employed to modify biomaterial surface/bulk with adhesion molecules (e.g., layer by layer techniques, blending, physical adsorption, covalent techniques). In addition to these techniques, specific patterns of adhesion proteins can be created in biomaterials to guide neural cell behaviors, thereby increasing the complexity of scaffolds in mimicking the native in vivo microenvironment. This chapter reviews mechanisms of cell adhesion to biomaterial surfaces, strategies to incorporate adhesion molecules (with a focus on collagen, laminin and polylysine, and their peptide derivatives) into biomaterial scaffolds, their advantages and disadvantages as well as methodologies to incorporate patterned adhesive cues.

Following the chapter on adhesion molecules that can be used to enhance cell adhesion and study these behaviors, Chapter 3 presents an overview of current methodologies to

14 study cell migration with a particular focus on glioblastoma multiforme tumor cell migration. Current methods to study cell migration mostly utilize two dimensional substrates that fail to mimic the complex in vivo microenvironment. This chapter discusses current models with their advantages and disadvantages as well as lays a foundation to developing three dimensional biomimetic polymeric materials that can be used as in vitro disease models in addition to serving as biomaterial scaffolds for neural repair therapies. Recent efforts using biomimetic models to study GBM behaviors and how these models can serve to bridge the gap between current 2D models and animal studies is discussed. Further, the utility of these models for different applications (e.g., drug discovery, tailored patient bioassays) is highlighted.

In Chapter 4, polylysine (PL) modified poly(ethylene) glycol (PEG) – based hydrogels are presented as biomimetic coatings to tailor the neural tissue electrode interface. PL, an adhesion molecule was incorporated into PEG-based hydrogels using photopolymerization. The amount of PL incorporated was quantified using a fluorescence spectrophotometer. PC12 cells, a model neuronal cell line, were used to assess the adhesion potential of PL-modified PEG hydrogels. It was observed that incorporation of

PL significantly enhanced cell adhesion over unmodified PEG control hydrogels. Further, since PEG-diacrylate hydrogels do not adhere to electrode surfaces, the adhesion behavior of copolymer of PEG with poly(ε-caprolactone) (PCL) was examined.

Additionally, the ability to pattern these biomimetic coatings is demonstrated.

Chapter 5 describes biomimetic models presenting inherent interfacial mechanical gradients and further discusses how such gradients present in 3D hydrogels influence the

15 behavior of patient derived glioblastoma multiforme tumor cells. Interfacial gradients resulting from edge effects were examined using a biomimetic Matrigel hydrogel of varying compositions supported on a rigid underlying glass substrate. (Matrigel is a complex hydrophilic hydrogel obtained from Engelbreth-Holm-Swarm (EHS) mouse sarcoma [56] composed primarily of laminin, collagen IV, heparin sulfate proteoglycans and entactin/nidogen). The rigid glass-soft gel interface significantly influenced the interfacial properties as well a boundary zone (~ 50 µm) within the hydrogel as identified using computational finite element methodologies. To further examine this influence in an experimental setting, tumor cells were encapsulated in Matrigel and their morphological and migration signatures were examined at a lower (< ~50 µm) and higher gel (> ~ 500 µm) positions. Experimentally, tumor cells close to this interface (< ~50 µm) exhibited bipolar morphologies, organized actin filaments and faster migration with mesenchymal features. In contrast, cells away from this interface (> ~500 um) displayed mostly rounded morphologies, poor actin organization and slower migration with some amoeboid features. Thus, a combination of computational and experimental observations supported our hypothesis that inherent mechanical gradients resulting from edge effects play an important role in guiding 3D cell behaviors and should be considered in other 3D cell culture systems.

Chapter 6 presents collagen-hyaluronic acid (HA) composite hydrogels as 3D models of the in vivo brain tumor microenvironment. Whereas, the Matrigel system provided significant insight into the migration of patient derived GBMs, it is difficult to acknowledge the role of individual components because of its inherent complexity and

16 further, since it is obtained from a sarcoma tumor tissue does not adequately recapitulate the in vivo brain tumor microenvironment. To bridge this gap, and develop tissue microenvironment-mimetic systems, collagen-HA multi-component hydrogels were developed. HA was chosen because it makes up ~50% of the brain extracellular environment; whereas collagen is found in the brain microenvironment localized to the blood vessels. Biomimetic hydrogels were synthesized and their micro-architectural and mechanical features were examined. To mimic the increasing levels of HA typically seen in GBM tumors in vivo, gels with increasing HA content were prepared and patient- derived GBM single cell morphology, spreading, and migration were examined. GBM spreading and migration were inversely dependent on HA density in 3D. In particular, lower concentrations promoted spread/spindle shaped cell morphologies as well as cell migration, whereas higher concentrations promoted rounded cell morphologies with little or no migration. In contrast, non-cancerous astrocytes primarily displayed rounded morphologies regardless of HA concentrations. These results suggest that GBM behaviors are highly sensitive to ECM mimetic materials in 3D and that these composite systems could be used to further develop improved physiological 3D brain mimetic models for studying migration processes.

Chapter 7 describes white matter tract-mimicking polymeric biomaterials with tunable features fabricated using core-shell electro spinning techniques to examine GBM tumor cell behaviors. Chapters 5 and 6 examined hydrogel based 3D models for GBM migration. However, although these biomaterials are fibrous materials at the micro/nano scale, these fibers are not usually oriented and thus do not mimic highly aligned white

17 matter tracts, one of the major migration highways employed by GBM tumors in vivo. To develop biomaterials mimicking aligned white matter tract topography, electrospinning technology was employed to created aligned nanofibers. To further examine the role of mechanics and chemistry in a nanofiber setting, core-shell electrospinning was utilised.

To modulate the mechancial properties of these nanofibers while retaining identical surface chemistries, a variety of polymers were used as the core material (i.e., gelatin, polyethersulfone, polydimethylsiloxane)) with a common poly(ε-caprolactone) (PCL) shell, in addition to aligned PCL nanofibers. GBM cell behaviors were highly sensitive to nanofiber mechanics with single cell morphology (feret diameters), migration speed, focal adhesion kinase (FAK) and myosin light chain 2 (MLC2) expression all being a strong function of nanofiber mechanics. Similarly, altering chemical cues using biologically relevant chemistries ((HA), collagen, and Matrigel, as discussed in Chapters

5 and Chapter 6) as a shell on PCL core nanofibers revealed GBM sensitivity to chemistry, with HA, specifically, producing a strong negative effect on migratory potential, FAK and MLC2 expression. Collectively, these results provide strong evidence that tumor cell behaviors are tightly regulated by the biophysical and biochemical properties of their microenvironment in an in vivo topography mimetic setting.

Chapter 8 describes preliminary efforts in designing hydrogel-aligned electrospun fiber composite biomaterials mimicking white matter tracts (nanofibers) encapsulated in neural

ECM (hydrogels). Whereas both hydrogel and electrospun nanofiber based models have yielded crucial information regarding GBM behaviors, to further increase the complexity of in vitro models, multicomponent biomaterial systems providing multiple cues are

18 highly desirable. To this end, this chapter provides proof-of-concept demonstration in incorporating aligned cues within 3D hydrogel systems in different configurations (i.e., gel-aligned nanofiber and gel-aligned nanofiber-gel sandwich constructs) for use as an integrated platform for examining aligned 3D GBM migration and invasion. Further, using gel-aligned nanofiber constructs, improved adhesion, spreading and cell alignment versus gel controls was also observed thereby improving cell adhesion ability of synthetic biomaterials without the addition of chemical cues (i.e., adhesion molecules) for neural tissue engineering and regenerative medicine applications.

In Chapter 9, the results are summarized with conclusions. Relevant future directions, both as follow up studies to this work or newer research topics related to the theme of this work, are highlighted and described.

19

Chapter 2: Adhesion Molecule-Modified Biomaterials for

Neural Engineering1

Adhesion molecules (AMs) represent one class of biomolecules that promote central nervous system regeneration. These tethered molecules provide cues to regenerating neurons that recapitulate the native brain environment. Improving cell adhesive potential of non-adhesive biomaterials is therefore a common goal in neural tissue engineering.

This chapter discusses common AMs used in neural biomaterials and the mechanism of cell attachment to these AMs. Methods to modify materials with AMs are discussed and compared. Additionally, patterning of AMs for achieving specific neuronal responses is explored.

2.1 Introduction

Damage to the central nervous system, which affects at least 2 million people per year [7,

8], can be devastating to the patient. Unlike neurons in the peripheral nervous system

(PNS), those of the central nervous system (CNS) do not regenerate under normal conditions [9]. Several factors contribute to this failure. In the CNS, macrophage recruitment is delayed by the blood-brain barrier, limiting clearance of myelin, a glycoprotein that impedes regeneration. In addition, CNS injury results in activation of astrocytes, a type of glial cell, which proliferate to form an inhibitory glial scar as

1 This chapter with minor modifications has been published in the following reference: S. S. Rao, J.O. Winter (2009). “Adhesion molecule-modified biomaterials for neural tissue engineering.” Frontiers in Neuroengineering. 2(6):1-14. 20 described in Chapter 1. Several neural biomaterials have been developed as treatment options for CNS injury, and have been used as regenerative, “tissue-engineered”, scaffolds [57] or as components of implanted neural prosthetic devices. These biomaterials are composed of either synthetic or natural materials. Natural materials recapitulate the native environment very well, but manipulating their characteristics may be difficult due to their complex structure. On the other hand, synthetic materials selected for their tunable structural and chemical properties are generally not conducive for neural cell adhesion and hence can impede the process of tissue regeneration.

Researchers have adopted several strategies to improve tissue integration of neural biomaterials, which fall into two general categories: soluble factor addition (reviewed in

[58]) and modification of materials with tethered biomolecules (i.e., adhesion molecules,

AMs). Cell adhesion is an important phenomenon in tissue regeneration. The interaction of the adherent cell with its surroundings can ultimately determine cell fate. For example, it has been shown that cells require a minimal contact area on a substrate to survive [59], and that the nature of this contact area can control the formation of connections with the outside environment [60].

On an artificial substrate, such as a biomaterial, this proceeds via the following steps

[61]: (1) initial cell attachment, (2) cell spreading, (3) organization of the actin cytoskeleton and (4) formation of specific focal contacts. The initial attachment of cells onto modified substrates results in immobilization, preventing detachment in response to mild shear forces. Once attached to the surface, the cell membrane begins to spread along the available surface area. This is followed by the creation of a filamentous actin

21 cytoskeleton. Finally, in response to force applied by the cytoskeleton, integrins form clusters, known as focal adhesion sites (Figure 8), that trigger signaling pathways, which can influence cell function, viability, and proliferation [62].

Figure 8. Receptor-mediated cell binding. Figure adapted from http://www.mun.ca/biology.

It is well known that cell-cell and cell-extracellular matrix (ECM) interactions are vital for tissue regeneration [63, 64]. ECM molecules such as collagen and laminin promote axonal regeneration, differentiation, adhesion, and migration in the central nervous system [65]. To enhance tissue integration, biomaterials can be modified with short recognition motifs that mimic the ECM to promote cell binding (e.g., proteins, peptides).

Incorporating AMs has thus become one of the standard methods for increasing tissue integration of neural biomaterials. This chapter focuses on strategies to modify biomaterials with AMs for tissue engineering applications in the CNS, and in particular the use of polylysine, collagen, and laminin and their peptide derivatives. In addition, the

22 influence of AM patterning on neuronal behavior is discussed. It should be noted that many of the techniques discussed in this chapter are equally applicable to other tissue engineering domains (e.g., RGD peptide in bone [61] and cardiovascular tissue engineering [66]), which can serve as a guide for neural biomaterial modification.

2.2 Adhesion molecules (AMs)

AMs derived from ECM proteins, including laminin and collagen, initiate receptor- mediated cell binding, inducing the formation of focal adhesions (Figure 8). Focal adhesion sites allow for two way signal transfer (i.e., into and out of the cell) through an elaborate mechanotransduction system [67]. Specifically, tension in the cytoskeleton can increase integrin receptor affinity for ligand binding (outward signal transfer), and conversely, ligand binding to an integrin receptor can induce a cascade of events within the cell that alters cytoskeletal composition (inward signal transfer). Various focal adhesion proteins assist in this signal transduction cascade, including talin, vinculin, α- actinin, filamin and paxillin [68]. This chapter will focus primarily on integrin-binding

AMs (e.g., collagen and laminin). Proteoglycans (PGs), another type of ECM molecule, also mediate cell adhesion using the receptor-mediated cell binding mechanism.

However, PGs are rarely used for neural tissue engineering because they have been shown to have inhibitory effects on axonal regeneration in the CNS [64]. Additionally, the cell adhesion molecule class of cell surface proteins (e.g., neural cell adhesion molecule (NCAM)), which primarily mediate cell-cell interactions, are not commonly used to modify neural biomaterials, and hence are not discussed in this chapter.

23

Integrin binding to ECM proteins has been shown to rely on short peptide motifs within the larger protein [69, 70]. Given that large-scale isolation of ECM proteins can be challenging, the application of these specific peptide sequences has received much attention in neural biomaterial modification [71]. Peptides are more stable, are more easily synthesized, and are less likely to exhibit steric hindrance after biomaterial modification than whole proteins [71]. For example, the trimer RGD peptide sequence, found in collagen, laminin and fibronectin has been identified as a minimum cell recognition sequence that can mediate adhesion of many cell types, including neurons

[71]. (It should be noted that in the case of laminin this sequence is not available for binding to integrin receptors until the domains in its vicinity are proteolytically cleaved

[72]). Sequences specific to neural adhesion are found primarily in the ECM molecule laminin, and include YIGSR, IKVAV, RNIAEIIKDI and RYVVLPR. YIGSR, found on the β1 laminin chain [70] and IKVAV, found on the C-terminal end of the α1 laminin chain [70], bind 67kDa and 110kDa proteins on the cell membrane, respectively. The combination of these peptides [73] and extended peptide sequences that incorporate both

YIGSR and IKVAV [74] have been found to significantly increase neuronal adhesion.

The peptide sequence, RNIAEIIKDI is present on the γ laminin chain [75], whereas

RYVVLPR is derived from β1 laminin chain [76].

In addition to laminin, the ECM protein collagen, whose primary function is to provide structural stability to tissues [77], can facilitate adhesion of neural cells through integrins.

Most neural cells express integrins belonging to the β1 and αvβ family [65, 71].

Specifically, α1β1 integrins can bind collagen type IV and XIII, α2β1 and α11β1 can bind

24 fibril forming collagens, whereas α10β1 can bind type II collagen [78]. (For a detailed description of different types of collagen and their functions the reader is referred to

[77]). A known integrin-binding peptide sequence that promotes neural adhesion in collagen is DGEA, which is present in collagen type I and fibril forming collagens [79].

In addition to natural biomolecules and peptides, some non-native proteins/peptides have been shown to promote neural adhesion. For example, polylysine, a polypeptide comprised of lysine sequences, enhances neural adhesion, proliferation, and neurite extension [80]. Polylysine modulates cell adhesion via a non-receptor-mediated cell binding mechanism (Figure 9). Positive charges on polylysine attract the negatively charged cell membrane resulting in electrostatic bond formation [80]. The negative charge on the cell membrane results from the glycocalyx, which is composed of short oligosaccharide chains containing a large number of sialic acid residues [81]. It is believed that free polylysine amino groups [80], which produce a monopolar basic surface [82], are necessary for cell adhesion and that adhesion is energy dependent [80]

(i.e., adhesion is drastically affected when cultures are exposed to inhibitors of respiration such as cyanide binding reagents). Adhesion is also temperature dependent indicating an affiliation with endocytotic metabolic pathways. Upon binding, polylysine produces a charge-induced redistribution of molecules in the cell membrane, resulting in a ‘cell- polylysine interaction’ similar to the ligand-receptor-mediated interaction [83].

Polylysine may also enhance attachment indirectly, by promoting the adsorption of medium proteins [84]. After the initial, polylysine-induced binding; cells secrete ECM, which is used to initiate mechanotransduction processes described above. It should be

25 noted that cells unable to secrete ECM cannot sustain binding through this mechanism and undergo apoptosis [68].

Polylysine in both of its forms (i.e., d and l) mediates neural cell adhesion. Whereas cell responses do not differ greatly, the d-form may be preferred over the l-form because of its resistance to proteases released in culture [84]. Another interesting aspect of polylysine is that cell response is drastically altered by changes in the molecular weight (i.e., number of lysine residues). In a red blood cell model widely used in cell biology and equally applicable to neurons, low polylysine concentrations and low molecular weights were found to only weakly promote cell adhesion. At intermediate concentrations and molecular weights, cells spread uniformly. However, high concentrations and molecular weights produced cells lysis [85]. These results suggest the importance of selecting polylysine of the appropriate molecular weight and form for the desired application.

Figure 9. Non-receptor mediated cell binding.

2.3 Approaches to biomaterial modification

2.3.1 Surface deposition

One of the easiest methods used to modify neural biomaterials is to physically adsorb/coat AMs onto biomaterial surfaces. AMs are physically bound to the biomaterial

26 via weak forces (e.g., van der Waals, hydrogen bonding, electrostatic interaction). This method has been used to apply laminin and polylysine to poly(lactide-co-glycolide)

(PLGA) films [86]; laminin, polylysine and collagen to polysialic acid hydrogel surfaces

[87]; and laminin to plasma-treated PLGA films, chitosan films [88] and poly(L-lactic acid) (PLA) nanofibers [89]. Although modified biomaterials created using this technique can promote neural cell adhesion, they are limited by poor stability of the AM layer.

Additionally, if proteins or long peptide sequences are utilized, physical deposition can create steric hindrance at the active site, lowering adhesion and hence regeneration potential. However, as a first approach to study the effect of AM-modified biomaterials on neuronal response, this technique can provide useful initial data.

2.3.2. Blending

As an alternative to direct adsorption, AMs may be blended with biomaterials to create composites. Blending results in a near uniform distribution of AMs in the biomaterial matrix, is simple and less time consuming than covalent approaches, and provides a more stable material than that formed using adsorption methods. Blended composites can be created as thin films or 3D polymeric constructs. For example, chitosan, a biodegradable polysaccharide obtained from chitin, has been blended with collagen and polylysine to form 2D films [90]. These films showed improved neural adhesion over unmodified chitosan, with chitosan-polylysine blends showing the highest adhesion potential.

Another study found that films consisting of chitosan-3 wt% polylysine supported neural adhesion more effectively than collagen-only films, which have been widely used as adhesive substrate coatings [91]. These films could potentially be applied as coatings for

27 neural prosthetic implants. However, possible disadvantages include instability of AM attachment as a result of non-covalent incorporation, reduction of the number of AMs available per unit surface area with increasing film thickness, and their 2D structure.

As an alternative to thin films, blending can also be used to physically entrap AMs in 3D hydrogel matrices. Hydrogels, insoluble, hydrophilic, cross-linked, polymer networks

[42], have been widely studied as brain mimetics because of their structural similarity to glycosoaminoglycans (GAGs), which are the primary component of brain ECM. The addition of AMs to hydrogels enhances neural cell adhesion, while preserving a matrix with similar properties to those found in vivo. AM modification can be used to increase neural adhesion at the interface with tissue (i.e., on the material surface) or to increase adhesion of neurons encapsulated within the hydrogel. For example, laminin blended with a keratin-based hydrogel was used to increase affinity of neurosphere forming cells to the hydrogel surface (2D) [92]. Blending can also be used to increase encapsulated cell adhesion; Anseth et al. investigated the effect of collagen-blended with poly(ethylene glycol) – poly(lactic acid) (PEG-PLA) hydrogels on neural cells encapsulated within the

3D hydrogels [93]. AM-modified hydrogels created through blending could be applied to prosthetic devices or used as tissue engineering constructs. In particular, the ability of these materials to recapitulate the 3D environment found in vivo should yield detailed insights into tissue-level behaviors.

In addition to hydrogels, AMs have also been blended with electrospun fibers to create

3D scaffolds [89]. Nanostructured scaffolds provide very high surface areas, which may increase the efficacy of AMs as a larger number will be displayed on the surface for the

28 same volumetric loading. Koh et al. have shown that the total amount of AM (i.e., laminin) in blended scaffolds can be higher than that produced by either physical adsorption or covalent modification. Despite this success, results were not comparable with a positive control of polylysine-coated coverslips. It should be noted that these studies are still in nascent stages; different protein molecules at different concentrations were investigated and direct comparison is therefore not possible.

From a manufacturing perspective, blending may be advantageous when compared to covalent modification methods for nanostructured scaffolds. AMs can be incorporated directly during processing (e.g., laminin was present in the electrospinning solution to form laminin-blended electrospun fibers [89]), rather than requiring subsequent synthesis steps. Post-synthetic modification of 3D porous, nanostructured scaffolds may be challenging because of the difficulty in achieving uniform biomolecule access to the scaffold interior. This becomes more challenging as the scaffold pore sizes approaches that of the biomolecule, limiting diffusion. Blending is less desirable; however, for scaffolds with larger pore sizes. As long as the pore size of the scaffold is less than that of the adhesion molecule, AMs can be physically entrapped, preventing their escape, but as pore size increases, biomolecules may diffuse into the surrounding medium.

In general, blending is an attractive method for creating AM-modified materials because of its simplicity. It can be applied to films and 3D constructs, which can serve as prosthetic material coatings or tissue engineering scaffolds. One of the major limitations of blending is the tendency of AMs to escape from the polymer matrix over time. This is mainly caused by the absence of strong attractive forces between AMs and the

29 biomaterial. Several other techniques (e.g., covalent modification) have been explored for developing more stable composite systems.

2.3.3. Electrostatic attachment

Electrostatic attachment is similar to blending or physical adsorption, but relies specifically on the electrostatic interaction between AMs and the biomaterial as the driving force for biomaterial assembly. Electrostatic attachment methods can be divided into two broad classes: layer-by-layer (LbL) assembly and electrochemical polymerization.

2.3.3.1. Layer-by-Layer (LbL) technique (using polyelectrolytes)

The Layer-by-Layer (LbL) technique involves the deposition of alternating layers of polycationic and polyanionic materials, which can self-assemble through electrostatic attraction to produce nanoscale coatings (~100 Å). For example, positively charged biomaterials can be electrostatically coupled with negatively charged AMs to produce an alternating structure of (-biomaterial-AM-)N. The initial deposition of either polycation or polyanion depends on the substrate charge. Fortunately, many likely biomaterial targets for AM-modification are inherently charged. For example, silicon, which is a common material for neural electrodes, has an inherent negative charge and therefore can be coated with polycations. In the event that the target biomaterial does not possess charge, it can be induced by various surface treatment techniques [94].

The methodology of LbL treatment (Figure 10) is relatively simple [95, 96]. Initial layer formation proceeds by dipping the negatively or positively charged substrate in alternating polycationic and polyanionic solutions. Between each deposition stage, the

30 excess surface polyelectrolyte is removed by rinsing. This alternate dipping process is repeated until a desired number of bilayers with certain thickness are obtained. Factors that are critical to LbL film formation and stability include pH, polyelectrolyte loading, and ionic strength of the polyelectrolytes. The polyelectrolyte solutions can consist of the biomaterial, other polymers, drugs, or AMs. For optimal performance of AM-modified substrates the terminal layer should contain the AM of interest, permitting direct interaction with cells and tissue.

Figure 10. Schematic of LbL technique. Figure taken from [95].

LbL coatings can be applied to tissue engineering constructs and implanted neural prostheses. Wu et al. demonstrated that LbL films comprised of hyaluronic acid

(HA)/collagen promote cortical neuron adhesion on glass [97], which is normally a non- permissive substrate. Similarly, Bellamkonda et al. examined LbL films consisting of polyethylimine (PEI)/Laminin, PEI/Gelatin/Chitosan/Gelatin, and PEI/Gelatin as neural recording electrode coatings [41]. Gelatin is obtained from collagen and is known to promote cell adhesion [98]. These coatings were shown to promote cortical neuronal

31 adhesion and neurite extension in vitro [41] (Figure 11) and rapid reduction of early microglia activation (over a period of four weeks) [99] in vivo. This implies that LbL coatings can lower immune response over a stipulated time period.

LbL films have many advantages. They are versatile and can be applied to virtually any charged substrate. They can achieve near conformal coatings, with nm control of thickness. However, their long term stability in vivo is still questionable. Films are pH sensitive, with even minor changes in pH altering organization and producing instability.

Instability might possibly be addressed by cross-linking polyelectrolyte layers either chemically or via photo coupling. This could enhance their stability due to the presence of multiple linkages (i.e., electrostatic as well as cross linking), but complicates material processing.

Furthermore, since LbL assembly is mainly charge based, it is necessary for the AM to possess sufficient charge opposite that of the biomaterial for coupling to occur. For instance, gelatin is available mainly as Type A (obtained from porcine skin) and Type B

(obtained from bovine skin). The isoelectric points of gelatin range from 4.5-9.4 with

Type A having a higher isoelectric point (7-9.4) compared to Type B (4.5-5.3) [100]. The type of gelatin selected and the experimental conditions chosen should be adjusted to reflect the charge of the biomaterial to be modified. Similarly, cell response to different types of gelatin may also vary with charge presented at the physiological pH and should be considered.

32

Figure 11. Immunofluorecence of chick cortical neurons on Si wafers (5 days): (A) (PEI- Gelatin)8, (B) (PEI-gelatin)-(Chitosan-gelatin)7, (C) (PEI-Laminin)8. Figure courtesy of Dr. Ravi Bellamkonda, Georgia Institute of Technology and Dr. Wei He, University of Tennessee.

2.3.3.2. Electrochemical polymerization (using conducting polymers)

Another class of biomaterials that can incorporate AMs through electrostatic interactions is electrically conducting polymers. Because nerve cells are electrically active, there is great interest in using electrically conducting biomaterials to more closely mimic the native neural environment. One of the most extensively studied conducting polymers is polypyrrole (Ppy), a heterocyclic conducting polymer, which promotes neurite outgrowth under the influence of electrical stimulation [101]. Ppy has also been utilized for drug delivery [102], neural probe coatings [34, 37, 103-105] and bioactuation (i.e., the generation of a mechanical force as required in artificial muscle actuators) [106-108].

Electrically conducting polymers consist of charged crystalline to semi-crystalline polymer chains that are doped with ions of the opposite charge. Dopants serve to balance the charge of the conducting polymer to produce a neutral composite. Charged AMs can be incorporated into electrically conducting polymers as dopants using electrochemical polymerization (Figure 12) [109]. For example, a neutral polymer such as Ppy develops a

33 positive charge following oxidation and can be coupled with negatively charged AMs during electrochemical polymerization. In this process, a three electrode system is typically employed. The apparatus consists of a working electrode (where the films deposited, usually Si for neural probes, ITO for other applications), a counter electrode

(e.g., platinum) and a reference electrode (e.g., calomel electrode) in a liquid solution of monomer and dopant in a suitable solvent. Applying electric current to the system produces conducting polymer/AM film deposition on the working electrode surface.

Polymer monomers undergo oxidation at anodic sites forming cations that can bind negatively charged dopants (e.g., AMs). The resulting composite thus has a net charge of zero. Film thickness is controlled by the amount of charge that passes through the electrode system. Parameters that can influence film topography and conductivity include deposition time, temperature, electrode system, and choice of solvent. The technique is straightforward and attractive because doping of AM and polymerization proceed simultaneously. Also, extremely thin films (~ 20 nm) can be prepared.

34

Figure 12. Electrochemical polymerization. Figure courtesy of Nathalie Guimard and Dr. Christine E. Schmidt, The University of Texas at Austin.

This method has been used to dope polypyrrole with CDPGYIGSR (Figure 13), an extended peptide sequence from laminin, on gold recording sites of Si-neural recording probes [37]. In vitro, these materials demonstrated increased neuroblastoma cell adhesion compared to control films (Figure 14). In vivo, the coatings have been shown to be stable for at least 1 week and to promote neural adhesion [104]. In later work,

Ppy/RNIAEIIKDI (a sequence from laminin) coatings were shown to be superior to the original Ppy/CDPGYIGSR composites in promoting neural adhesion and axonal growth

[103], demonstrating the importance of AM selection when creating modified 35 biomaterials. This work has also been extended to other conducting polymers, for example poly(hydroxymethylated-3, 4-ethlenedioxythiophene) (PEDOT-MeOH) has been doped with the laminin fragment CDPGYIGSR [110].

Figure 13. SEM of Ppy/CDPGYIGSR on a microelectrode site of a neural probe. Figure taken from [104]. Scale bar = 2 µm.

As an alternative to using AMs directly as dopants, other entities can be employed to tether the AM to the conducting polymer through the dopant. One advantage of this method is that different AMs can be incorporated onto the same ‘base’ material (i.e., conducting polymer/dopant composite). Song et al. [111] used this method to tether polylysine and laminin to Ppy doped with poly(glutamic acid) (Ppy/Pglu). Specifically,

Ppy was doped with Pglu using electrochemical polymerization. Polylysine and laminin were then attached to the resulting Ppy/Pglu composite by a covalent bond formed between Pglu and AMs using EDC chemistry (see below, covalent binding).

36

Figure 14. Neuroblastoma cell response on a coated neural probe. Figure taken from [37].

AM electrochemical incorporation into conducting polymers has specific advantages for neural prosthetic systems. Because the composites are films, they provide high surface area, and therefore high AM contact area. Although not specific to AM-modified materials, the high surface areas of electrically conducting polymer films also provide ample sites for Faradaic charge transfer, the primary mode of electrical conduction in implanted prostheses. Faradaic charge transfer depends on the storage capacity of the material, and in the case of surface reactions, on the surface area of the material. In addition to electrical considerations, composite surface properties, namely morphology and conductivity, could potentially be precisely tuned to encourage neurons to form firm contacts with the electrode, while minimizing the activation of astrocytes, a sign of immune response. However, despite extensive study of these composites, long term stability continues to be a major impediment to the use of conducting polymer-AM composites for neural interfaces. For instance, it has been observed that Ppy undergoes

37 structural changes and is subject to degradation in vivo within short time periods [110].

Alternative conducting polymer materials are being explored to address this problem. The advantages and disadvantages of electrostatic attachment techniques are summarized in

Table 3.

Techniques Utilizing Electrostatic Attachment Advantages Disadvantages LbL assembly (using . Better control of film . Lack long term stability polyelectrolytes) thickness . Highly pH sensitive . Simple technique Electrochemical . Thin films can be . Limited to monomers polymerization (using obtained that oxidize under the conducting polymers) . Doping of AM and influence of applied polymerization occur at potential the same time . Modifying the bulk of . High surface area can be conducting polymer after achieved for neural electrochemical interfacing polymerization may be difficult Table 3. Advantages and disadvantages of electrostatic attachment techniques.

2.3.4 Covalent attachment

Techniques based on weak interactions (i.e., physical adsorption, blending) often fail to strongly bind AMs to biomaterial scaffolds, which can permit escape via diffusion. This is disadvantageous, especially for in vivo applications in which the implanted material may be intended for months or years of use. Covalent binding of AMs to biomaterials can produce much more stable composites. Several covalent attachment techniques have been designed specifically for biomaterial modification (see [112] for details). This chapter will concentrate on attachment methods mediated by thiol, amine, carboxylate and 38 hydroxyl linkages, as these have been most commonly employed (Table 4). It should be noted that some of these techniques require pretreatment of surfaces to produce the desired surface functionalization; however, these methods (e.g., plasma treatment, ionizing radiation graft copolymerization [113]) are not discussed in this chapter.

Chemical modification mediated by thiol (-SH) groups involves the reaction of a sulfhydryl compound with a maleimide derivative to form a thioether bond (Figure 15)

[112].

O O

R N R' SH N + R S R' O Sulfhydryl O compound Maleimide derivative Thioether bond

Figure 15. Chemical modification mediated by thiol group.

This route has been used by the Shoichet group [114] to conjugate cell adhesion peptides

CDPGYIGSR and GQASSIKVAV (both laminin derived) to thiolated methacrylamide chitosan scaffolds. Other approaches employing similar chemistry use heterofunctional crosslinkers to conjugate AMs to biomaterials. Intermediary crosslinkers are generally preferred to direct biomolecule conjugation because the crosslinker separates the biomolecule from the biomaterial, reducing steric hindrance and preserving the conformation of the protein or peptide. The choice of the cross linking agent is dependent on application. For example, for in vivo use, a cross linker that minimizes immune

39 response should be chosen. One commonly examined –SH active crosslinker is sulfo-

SMCC (sulfo-succinimidyl-4-(N-maleinidomethyl) cyclohexane-1-carboxylate (Figure

16), which has an amine reactive NHS ester as well as a thiol reactive maleimide.

R' O S O O

N O N R NH N O O

O O O Crosslinked product

R NH2 Amine containing compound R' SH O Sulfhydryl containing compound O

N OH R NH N O

O O NHS SMCC activated intermediate

Figure 16. Chemical modification mediated via SMCC crosslinker.

This cross linker has been used to conjugate CRGDS and CDPGYIGSR/CQAASIKVAV to poly(methacrylated dextran-co-amino ethyl methacrylate) (p(dex-MA-co-AEMA))

[115], CDPGYIGSR/CQAASIKVAV to poly(hydroxyl ethyl methacrylate-co-2- aminoethyl methacrylate) (p(HEMA-co-AEMA) [74], and

CYIGSR/CDPGYIGSR/CIKVAV/CQAASIKVAV to e-poly(tetrafluroethylene) [116].

40

Cell adhesion was improved for all AM-modified biomaterials when compared to unmodified control biomaterials (e.g., Figure 17).

Figure 17. Representative light microscope image of DRG neurons on poly (Dex-MA-co-AEMA) modified with CGRGDS. Figure taken from [115].

Many polymers used for neural engineering applications contain hydroxyl groups (-OH)

(e.g., agarose, hyaluronan). Carbonyldiimidazole (CDI) chemistry can be exploited to couple biomaterials with AMs via a hydroxyl mediated reaction (Figure 18).

CDI, a carbonylating agent with acylimidazdole groups can react with (–OH) containing biomaterials to form imidazole carbamate active intermediates. These intermediates further react with amine containing compounds releasing imidazole, thus forming stable carbamate linkages [112]. Bellamkonda et al. [117] used this chemistry to covalently couple agarose to laminin fragments (CDPGYIGSR, IKVAV,GRGDSP and their combination). This method has also been used to bind laminin to hyaluronan [118].

41

O O

R N R OH + + NH N N O N

N N

Hydroxyl N compound CDI Imidazole carbamate active intermediate OH O

N R R R' NH + R' NH2 + O N O NH

Amine containing Carbamate Linkage compound N Imidazolyl carbamate

Figure 18. Chemical modification mediated via CDI chemistry.

It is also possible to target hydroxyl groups via periodate oxidation [112] wherein compounds containing internal diol groups, terminal diol groups or terminal hydroxylamine groups can be oxidized to form aldehydes (Figure 19). Aldehydes can then react with amine compounds to form a Schiff base (imine). This approach has been used to covalently bind laminin fragments to dextran [119] and to bind laminin to methylcellulose [120].

42

OH O H NaIO4 R R' R + R' H O

OH Compound with Aldehyde an internal diol group O OH H H NaIO4 R + R OH H O Compound with Aldehyde terminal diol group

OH O H H NaIO4 R R + H NH2 O O O Aldehyde Compound with terminal hydroxylamine group

Figure 19. Chemical modifications via oxidation with periodate.

Perhaps the most popular method used for covalent biomaterial modification is EDC chemistry, which proceeds via reaction with carboxylate (-COOH) groups. EDC or

EDAC (1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride) is a water soluble, zero length cross linker. Generally, EDC is used to covalently link a carboxylate

(-COOH) or a phosphate (-PO4) compound with an amine (-NH2) (Figure 20).

43

CH 3 - Cl + N NH R O + C CH3 O Isourea H C N 3 EDC OH CH3 + R' NH3 H3C NH NH NH Carboxylic acid R Amine compound CH O 3 O O R' R CH3 NH

H3C NH N NH Amide bond formation

CH o-acylisourea active intermediate 3

Figure 20. Chemical modification mediated via EDC chemistry.

Carboxylates react with EDC to form o-acylisourea intermediates, which in turn react with amine compounds to form stable amide bonds and isourea as a byproduct. This byproduct can be easily separated from the product since it is water soluble [112]. It may be necessary to use sulfo-N-hydroxysuccinimide (sulfo-NHS) in conjunction with EDC if the intermediate product is highly unstable (e.g., hydrolyzed) or if the amine compound is present in low concentrations. Addition of sulfo-NHS forms sulfo-NHS ester intermediates that are more stable and can efficiently react with amine compounds, increasing conjugation yields. A disadvantage of using EDC is that some target biomolecules (e.g., peptides) contain both carboxylate and amine groups, which can produce self-polymerization rather than conjugation to the desired biomaterial [112].

44

EDC chemistry has been widely employed in neural biomaterial modification. For example, it has been used to covalently modify chitosan with laminin derived peptides

[121], poly-D-lysine with hyaluronan [122], alginate with YIGSR [123], PLGA and chitosan with laminin [88], aminated glass with various integrin-binding peptide sequences [124], and glassy carbon implant materials with laminin peptide sequences

[125]. Recently, Koh et al. [89] also showed that EDC chemistry can be used to modify electrospun poly(L-lactic acid) fibers with laminin.

In addition to these more common cross-linking methods, carboxyl (-COOH) and hydroxyl (-OH) groups have also been covalently modified with tresyl chloride, SMCC and O-(N-succinimidyl)-N,N,N’,N’-tetramethyluronium tetrafluoroborate. This method was used to couple laminin derived peptides to poly(tetrafluroethylene-co- hexafluoropropylene) (FEP) [126]. Specifically, FEP was functionalized with reactive groups (i.e., hydroxyl or carboxyl) by a sequence of chemical reactions and covalently coupled to peptides using one of the aforementioned coupling agents. As an alternative, introducing amine functionality to FEP surfaces and coupling laminin peptides [127] and their combination [73] was also investigated.

Apart from purely chemical methods, photo-initiated coupling has also been used to covalently immobilize AMs onto neural biomaterials. Benefits of this technique include better control over the coupling reaction and also rapid reaction completion (e.g.,

~minutes). Examples of photo-coupling include conjugation of benzophenone-derivatized

YIGSR to agarose [128] and azidoaniline photocoupling of poly-D-lysine to chitosan

[129]. As with chemical methods, heterobifunctional crosslinkers can also be employed,

45 as demonstrated by used of an agent with photoreactive and amine-reactive groups to conjugate collagen to agarose [130].

Another interesting alternative to chemical crosslinking is enzymatic coagulation. In this method, an enzyme regulates covalent crosslinking of the AM to the biomaterial matrix.

This was demonstrated using fibrin gels, which are formed from fibrinogen in conjunction with the enzyme transglutaminase. This property has been used to develop peptide-modified fibrin materials utilizing custom designed peptides with bioactive domains as well as providing a substrate for fibrin forming enzymes [131]. Examples from literature employing specific chemistries have been summarized in Table 4.

Reactive In Vitro In Vivo Biomaterial/AM groups Cell Line Location References Methacrylamide chitosan -SH SCG - [114] /YIGSR/IKVAV Poly(dex-MA-co AEMA) -SH DRG - [115] /RGD/YIGSR/IKVAV -NH2 Poly(HEMA-co-AEMA)/ -SH DRG - [74] YIGSR / IKVAV -NH2 Agarose/RGD/YIGSR/IKVA -OH DRG - [117] V PC12 Hyaluronan/Laminin -OH - SD F rat [118] Brain Dextran/RGD/IKVAV -OH PC12 - [119] -CHO GB Methylcellulose/Laminin -OH RC [120] -CHO Chitosan/YIGSR/IKVAV -OH - SD M rat [121] -COOH brain continued

Table 4. Covalent coupling of AM with biomaterials. Abbreviations: (DRG) Dorsal Root Ganglion, (SCG) Superior Cervical Ganglion, (SC) Schwann Cells, (RC) Rat cortical Neurons, (MC) Mouse Cortical Neurons, (EHC) Embyronic HippoAMpal Neurons, (ERGC) Embryonic Retinal Ganglion Cells, (GB) Glioblastoma, (SD) Sprague Dawley, (F) Female, (M) Male 46

Table 4 continued

Hyaluronan/Poly-D-lysine -OH RC SD rat [122] brain Chitosan/Poly-D-lysine -NH2 Fetal MC - [129] -OH photo Agarose/Collagen -NH2 RC - [130] photo Poly(tetrafluoroethylene-co- -OH EHC - [126] hexafluropropylene) -COOH /YIGSR/IKVAV/RGD Poly(tetrafluoroethylene-co- -NH2 EHC - [73, 127] hexafluropropylene) /YIGSR/IKVAV/RGD and combination of YIGSR/IKVAV Alginate/YIGSR -COOH NB2a - [123] Poly-L-lactic acid/Laminin -COOH PC12 [89] Poly(lactide-co- -COOH SC [88] glycolide)/Laminin Chitosan/Laminin -OH SC [88] -COOH Glassy carbon/Laminin/ -COOH ERGCs - [125] Laminin derived peptides (chick) e-poly(tetrafluroethylene)/ -SH DRG - [116] Laminin derived peptides -NH2

Each biomaterial modification technique discussed is outlined in Figure 21.

2.4 Patterning AM-modified biomaterials

Bulk biomaterial modification with AMs increases neuronal adhesion compared to

unmodified materials; however, this response is not targeted to specific portions of the

material. In addition, AM distribution is not tailored to achieve specific cell responses.

There is a substantial body of evidence that suggests neural cells respond to micron and

nanoscale features with altered adhesion, proliferation and survival tendencies [132, 133]. 47

Figure 21. Schematic of biomaterial modification techniques.

The focal adhesions characteristic of cell adhesion to the ECM are typically on the order of nanometers [134]. It is therefore logical that micro- and nano- scale surfaces can better duplicate the in vivo environment and provide more detailed insights into neural cell behavior than bulk-modified materials.

Initial efforts to pattern AMs on neural biomaterials focused on 2D, planar surfaces because of the simplicity of fabrication. In an early example, fluorinated ethylene propylene films were patterned with laminin derived peptides YIGSR and IKVAV through radio frequency glow discharge [135]. AM patterns can promote specific neuronal responses. For example, AMs have been patterned, along with non-cell adhesive polyethylene glycol (PEG) domains, on activated glass surfaces. The non-cell adhesive 48 regions were created using covalent attachment methods, whereas the cell adhesive domains were created using the shadow masking technique involving Ti and Au sputter coating followed by AM attachment. The resulting alternating ‘stripe’ pattern promoted hippocampal neuron alignment and neurite extension along the length of the patterned

AM (Figure 10) [136]. Similarly, poly(chlorotrifluorothylene) (PCTFE) was also used for creating well defined alternating ‘stripe’ patterns of laminin derived peptides YIGSR and

IKVAV [137].

Figure 22. Hippocampal neuron alignment on surface modified glass [CSIKVAV (200 µm)/PEG (50 µm)] at 20X magnification. Figure taken from [136].

Most of these early studies examined AM patterning on relatively smooth substrates.

More recently, researchers have focused on creating topological patterns on the biomaterial and combining these with AM patterns by physical adsorption or covalent binding. Topological patterns can be created through a variety of techniques. For 49 example, microfluidic patterning has been used to create channels on polymeric surfaces with O2 plasma treated polydimethylsiloxane (PDMS) molds. These channels can then be modified with AMs through standard bioconjugation methods. For example, avidin-biotin interactions have been exploited to create patterns of IKVAV on PEGPLA [138].

Specifically, plasma treated PDMS molds were employed to form patterns on PLA-PEG- biotin surfaces. Avidin solution was then placed on this modified surface resulting in

PLA-PEG-biotin-avidin. Finally, biotinylated ligands (biotin-AMs) were anchored to the

PLA-PEG-biotin via avidin forming PLA-PEG-biotin-avidin-biotin-AM.

Photolithography can be used to create “master” patterns, which can then be transferred to polymeric surfaces by compression molding or similar techniques. This method has been used to modify poly (D, L-lactic acid) with laminin and was used to examine

Schwann cell culture [139, 140]. Microcontact printing has been used to print AMs directly onto biomaterials. For instance, laminin was applied to acrylamide hydrogel surfaces to examine astroglioma and primary rat hippocampal neurons [141].

These methods have primarily been applied to 2D surfaces. Unfortunately, the in vivo environment is a highly complex 3D structure, and many target neural biomaterials have been designed to mimic this environment. Alternative methods are needed to create 3D

AM patterns in these materials. Hydrogels, one of the most commonly employed 3D materials, can be modified with AMs using photo patterning (especially if vinyl or similar chemistries are employed for the hydrogel backbone [142]). Hydrogel patterning has been used to create channels of RGD adhesive peptides in agarose hydrogels [143] using a combination of covalent and photochemical coupling techniques and to study the effect

50 of channel RGD concentration gradients in hyaluronic acid hydrogels [144]. The presence of RGD-modified channels in agarose permitted dorsal root ganglion cells growth and attachment, which was not supported outside the channels [143]. High fidelity AM patterns in hydrogels can provide complex topographical and biochemical signals to neurons more closely mimicking the native environment.

2.5 Conclusions

Biomaterials developed for neural tissue engineering applications should replicate the native brain environment for successful biomaterial-tissue integration. Application of bound/tethered factors or AMs has become one of the primary strategies to achieve this goal. The interaction between the biomaterial and cells is crucial in determining cell fate.

Understanding the phenomenon of cell adhesion on biomaterials and the mechanisms by which AMs promote neural adhesion is crucial for the development of materials that promote neural tissue regeneration.

Among the different techniques discussed for creating biomaterial-AM composites, blending and surface deposition are the simplest methods. However, composites created using these techniques are less stable. Electrostatic and covalent attachment techniques can produce much stronger and more stable composites because of stronger attractive forces between the biomaterial and AM. Selection of AM incorporation method should thus be made with the end goal in mind. For example, in vivo implants with an expected lifetime of months to years would require a more stable interface than materials designed for short-term in vitro tests. Additionally, the morphology of AM deposition must be

51 considered. Whereas bulk material modification is easier, patterned AMs have been shown to promote specific neural responses.

AM modification using these methods has already produced materials for drug delivery, tissue engineering, and implanted neural prostheses. These materials have shown promise in minimizing the long term immune response to implanted materials and in promoting nerve regeneration. However, the stability of AM coatings is still a significant issue.

Future enhancements in AM-modification will likely include combinations of the techniques discussed and increases in the number of biomolecules bound. With continued improvement, AM-modification will likely become a critical element in enhancing biomaterial-tissue integration with the nervous system.

52

Chapter 3: Toward 3D Biomimetic Models for Investigating

Glioblastoma Multiforme Tumor Cell Behaviors2

In addition to loss of neuronal function as described in Chapter 2, cancerous tissue growth is yet another devastating CNS insult. Among the several tumor types, glioblastoma multiforme (GBM) tumors are one of the most deadly forms of human cancers and despite improved treatments, median survival time for majority of the patients is a dismal 12-15 months. A hallmark of GBM tumors is their unique ability to diffusively infiltrate normal brain tissue. To understand these unique behaviors and successfully target the underlying migration mechanisms, it is crucial to develop robust experimental ex vivo disease models. This chapter discusses current two dimensional

(2D) experimental models as well as animal based models used to examine GBM migration behaviors with their advantages and disadvantages. Recent attempts in developing three dimensional (3D) tissue models and their utility in unraveling the role of microenvironment on tumor cell behaviors is highlighted. Further, how such models could be used to bridge the gap between 2D and animal models is explored. Finally, the broad utility of such models in the context of brain cancer research as well as in neural

2 Major portions of this chapter are being prepared for publication: S. S. Rao, J. J. Lannutti, A. Sarkar, J.O. Winter (2012). Toward 3D models for investigating brain tumor migration. In Preparation. 53 regeneration therapies is examined. Whereas Chapter 2 focused closely on cell adhesion, this chapter focuses on cell migration, yet another important biological phenomenon.

3.1 Introduction

Glioblastoma multiforme (GBM) represents one of the most common brain tumors in adults, accounting for ~15% of intracranial tumors and affecting over 20,000 individuals annually in the United States [48, 55, 145, 146]. Unfortunately, coupled with its frequency, these are among the most malignant human cancers, and as such, prognosis associated with this lesion is bleak [48, 51, 55]. Nearly 60 years ago, a grading scale was devised by the pathologist Kernohan [147], which linked tumor grade to survival. Despite dramatic improvements in micro-neurosurgical techniques, neuro-anesthesia, neuro- imaging, chemotherapy, and radiation therapy, the outcomes for patients with aggressively managed tumors still remain dismal [148]. Further, it has been shown that migrating GBM cells at the leading front divide more slowly than the core making cytotoxic therapies based on proliferation futile [149, 150]. This is, in part, a consequence of the highly infiltrative nature of these tumors with recurrence of tumors both locally and distantly within the brain [49]. In fact, median survival for a patient with optimal care for a GBM is approximately a year, with many patients succumbing to their illnesses precipitously [48, 55, 151].

Most therapeutic strategies aimed at GBMs target rapidly proliferating cells and/or a combination of cytotoxic therapies [152-154]. There are have been very few attempts if any that target GBM migration as a therapeutic target and it is likely that techniques targeting migration will provide a significant benefit to patients over the long term given

54 their highly migratory potential [152]. Understanding the aggressive migratory behaviors of GBMs is therefore crucial to the development of new targeted therapeutics [155, 156].

A major limitation in understanding these behaviors and developing these newer treatments for patients with GBM is the lack of powerful experimental in vitro models that predict migration patterns in the brain. Current models in use, specifically two dimensional (2D) models, do not adequately model the complex in vivo tumor microenvironment and therefore continue to remain poor predictors of tumor cell behaviors. To gain detailed insight into GBM migration patterns in brain tissues, experimental in vitro models recapitulating the in vivo niche as well as providing a highly reproducible and tunable environment are urgently needed. These models would also allow investigation and identification of factors that play a pivotal role in disease progression, eventually leading to the development of novel therapeutic options with direct implications to cancer treatment.

This chapter examines changes occurring in the normal and cancer brain extracellular matrix (ECM) compositions. State of the art experimental models to study GBM migration in vitro and the limitations of these models in providing reproducible in vivo- like behaviors are discussed. Further, recent attempts to develop three dimensional (3D) models that mimic several aspects of the in vivo environment are highlighted. Finally, how creation of improved 3D tissue analogs will impact brain cancer research as well as other cancers is discussed.

55

3.2 Normal brain versus cancer brain

The ECM consists of the local tissue surrounding cells and is comprised of glycosoaminoglycans (GAG), proteoglycans (PG), and proteins that support and discourage cell adhesion. ECM influences have long been recognized as important contributors in tumorigenesis and tumor cell migration [157]. In fact, it has been suggested that the migratory potential of cancer results not only from genetic mutation within the cell, but also from external influences, including soluble factors and the insoluble matrix environment [158]. Brain exhibits substantially different ECM composition to other tissues in the body. Normal human brain is composed of ~80% cells

(of which 90% are glial) and ~20% extracellular matrix. This matrix is comprised predominantly of Hyaluronic acid or Hyaluronan (HA), a hydrophilic, anionic glycosoaminoglycan making up ~ 50% of the ECM. HA interacts non-covalently with other proteoglycan members (e.g., brevican, versican)[159]. The primary components of

ECM found in many other systems (e.g., collagen, laminin and fibronectin) are mainly found in basement membranes in the brain [160]. Additionally, the composition of brain

ECM changes dramatically in cancerous tissue. For example, glycosoaminoglycan

(GAG) production increases 3 fold [161], and the proteoglycan composition is altered, with brevican increasing and versican decreasing. Clearly, the chemical composition of tissue is altered with unique presentation of several molecules and up/down regulation of existing molecules. These alterations in chemical compositions are summarized in Table

5 along with their specific anatomical locations.

56

Adhesive Non-adhesive GAGs PGs Proteins Proteins Blood Fibronectin, Tenascin-C, Chondroitin Vessels Collagen, thrombospondin-1, sulfate (Tumor Laminin, Sparc proteoglycan associated) Osteopontin (CSPG), Versican Blood Fibronectin, Heparan sulfate Vessels Collagen, (Normal) Laminin, Vitronectin, Entactin Neuropil Vitronectin, Tenascin-C, Sparc Hyaluronan ↑ Brevican ↑ (Tumor Osteopontin Chondroitin Versican ↓ Associated) sulfate (CS) Heparan sulfate ↑ Keratin sulfate Dermatan Sulfate ↑ Neuropil No No Tenascin-C Hyaluronan Phosphocan (Normal) Vitronectin GAGs (3 times Neurocan lower than tumor Brevican ↓ associated Versican ↑ neuropil) Table 5. Composition of the brain ECM. Normal versus Cancer [159, 160]

In addition to alteration of chemical cues in the tumor microenvironment, the mechanical environment is also compromised. For example, clinical observations using magnetic resonance elastography reveal that cancerous brain tissue presents with altered mechanical properties compared to normal brain tissue [162-165]. However, it has been difficult to conclusively determine whether tissue tends to become stiffer or softer when compared to normal tissue given the observed heterogeneity of these tumors and a variety of other factors including interstitial pressure, edema, and secretion of several ECM

57 components. Nonetheless, these evidences suggest, that the mechanical properties of the microenvironment can also strongly influence migration capacity of GBM tumors.

Further, in contrast to alteration of chemical and mechanical characteristics of the microenvironment that might favor tumor dissemination, unique topographical architectures inherent in the in vivo brain environment (e.g., white matter tracts) have been noted to potentially contribute to migration. White matter tracts are formed of myelinated axons aligned and organized into bundles by oligodendrocytes. A single oligodendrocyte can myelinate up to 7-10 axons [166], with the number and length of lamella depending on axon diameter. Larger axons have thicker myelin and vice versa

[167]. Bundles are aligned, fibrillar structures with individual fiber diameters ranging from 0.5-3 µm and fiber densities of ~10,000-30,000 fibers/mm2 [168, 169]. Collections of these myelinated axons are organized into white matter tracts and the largest white matter tract, the corpus callosom, connects brain hemispheres [170] that could potentially act as a cellular highway for contralateral tumor migration. In fact, clinical observations suggest that GBMs migrate as single cells along the blood vessel periphery, subpial glial space, and most frequently through white matter tracks (Figure 23) independent of their origins [49, 50, 159, 171, 172]. Recent studies have shown that GBMs strongly respond to topographical cues [173, 174], matching closely with clinical observations of cells migrating along aforementioned topography presenting anatomical structures. Thus, it is possible that biochemical, biomechanical and unique architectural features presented at these locations drive cell migration by providing paths of least resistance. With altered environments that present favorable cues for cell migration in vivo, it is crucial that such

58 scenarios be recapitulated using in vitro models that mimic several aspects of this microenvironment to gain fundamental insights into single tumor cell migration patterns.

Figure 23. Clinical presentation of GBM tumors. (A) GBM cells seen around blood vessel periphery (arrow). (B) GBM cells seen below surface of pia mater, indicated by arrow (sub-pial spread). (C) GBM cells migrating along white matter tracts (labeled blue via Luxol staining). Both (A) and (B) are taken from [50] and (C) from [49].

3.3 Modeling tissue architecture in 2D

3.3.1 Monolayer wound healing assay (gap assay)

The Monolayer wound healing assay (Figure 24) was one of the earliest methods employed to investigate GBM cell migration. In this technique [175], GBM cells are cultured until they form a confluent monolayer. To stimulate creation of a wound, a scratch is made, and the time required for the cells to fill the gap is measured. The migration rate is calculated by dividing the distance travelled by GBMs (represented by the equation below) over the time period investigated.

59

It is important to note that the width of the wound decides the migration rate. For example, cells would take longer time to fill larger width wounds as compared to smaller width wounds. Although calculation of migration rates is straightforward, this assay examines GBM migration on 2D, rigid substrates (e.g., tissue culture polystyrene (TCPS)

,glass, Elastic modulus (E) > ~ 100,000 Pa [176]), culture conditions that do not replicate the in vivo mechanical environment (E ~ 100-1000 Pa [176]), and are therefore less physiologically relevant.

3.3.2 Microliter scale migration assay

Microliter-scale migration assays [175] examine migration of GBMs in the presence of an ECM molecule (Figure 24). Specifically, these molecules are deposited on a particular substrate (e.g., 10-well Teflon printed microscopic slides), following which, the cell solution is placed at the center of these wells. Radial migration of cells is then monitored by quantifying the increase in area of the circle that encompasses the cells over a stipulated time period. This assay has been used with physiologically relevant ECM molecules to study tumor cell migration, such as collagen type IV [177-179], laminin

[177, 179], vitronectin [177-180], fibronectin [177-179], merosin [178, 180], tenascin

[178, 181], hyaluronic acid [178] and myelin extracts [182, 183]. However, similar to the gap assay, the micro liter scale migration assay examines glioma cell migration on 2D rigid substrates that do not provide in vivo like mechanical environments. Also, quantification of areas may be difficult if there are several outlier cells, leading to biasing of results. Furthermore, ECM proteins on hard surfaces do not approximate the highly complex 3D in vivo environment to which GBMs are exposed.

60

3.3.3 Chamber assays

The Chamber assay (also referred to as the Boyden chamber assay) examines migration of GBMs in response to specific cues (e.g., attractants) through a transwell insert. Cells are seeded on the insert membrane and the number of stained cells crossing the membrane is counted. However, the pore size of the membrane (e.g., ~ 8 µm), is larger than those observed in vivo and hence this system does not present the adequate barriers to cell migration. Further, rigidity of the plastic substrate may dramatically influence migration capacity.

Matrigel-modified Chamber assays overcome some of the shortcomings. For example,

Matrigel provides a less rigid mechanical microenvironment [184]. In this assay [175,

185-189], the transwell insert is coated with Matrigel. The cell solution with appropriate media is placed in the top chamber. Cells that pass through the Matrigel layer and the pores of the insert are then fixed and stained for observation (Figure 24). The number of cells can be evaluated using optical microscopy. Typically, these assays provide a driving force for migration, for example culture in serum free medium above the membrane with serum containing medium below. Chamber-based assays have been employed to explore the role of physiologically relevant ECM molecules (e.g., unmodified HA [186, 187,

189]) by incorporating them into Matrigel as an additive or as a chemo attractant in the lower chamber (e.g., type I and IV collagen, laminin, fibronectin [188]). In both versions of the chamber assays, migrating cells contact the synthetic membrane (typically made of poly ethylene terephthalate (PET) that is mechanically and chemical distinct from brain tissue), which in turn can influence the results significantly. Furthermore, Matrigel, an

61

ECM derived from Engelbreth Holm mouse sarcoma (connective tissue tumor matrix) does not adequately recapitulate features of brain extracellular matrix. In addition, these assays are static in nature and thereby fail to provide dynamic information about migration patterns of tumor cells. More importantly, we have shown that the presence of the membrane could substantially diminish the benefits of dimensionality provided by a thin Matrigel layer. For example, we have shown in 3D Matrigel culture, patient derived

GBM cells within ~ 50 µm of a rigid support experience mechanical edge effects that significantly alter their adhesion and migration response [190]. Distinct effects are also seen at interfaces (i.e., gel-glass) when compared to standard 2D culture [More details of this work are provided in Chapter 5].

Figure 24. State of the art cell culture models. (A) 2D Gap assay. (B) 2D Microliter scale migration assay. (C) Transwell insert assay or chamber assay. Both (A) and (B) taken from [175].

3.4 Animal derived models

It is now widely recognized that 2D cell cultures can produce substantially different results in terms of tumor cell behaviors and signaling cascades than those of 3D cultures

[191, 192] that more closely mimic features of the in vivo environment. With this in

62 mind, several animal derived models for studying aspects of GBM migration ex vivo under physiologically relevant contexts have been developed.

3.4.1 Brain slice assays

This method [171, 175, 193-195] involves obtaining brain slices (~ 400 µm thick sections) from the whole brain using a vibratome. The brain slices are then placed on transwell inserts supplemented with media and stored until use. Glioma tumor cells (pre- labeled with a fluorescent molecule for observation) are then placed on these brain slices and supplemented with medium and their lateral movement is observed using fluorescence microscopy (Figure 25). Using the images, the extent of invasion can then be obtained (distance that yields half of the maximum invasive cell area) by quantifying the area of fluorescently labeled cells as a function of distance following formaldehyde fixation of brain slices. Despite the fact that these assays model tissue level behaviors closely, it becomes difficult to maintain cultures for long time periods because of the limited lifespan of tissues in vitro. For instance, brain slice consist of many cell types, and neuronal cells are extremely sensitive. As culture period increases, dying neuronal cells may affect the behavior of glioma cells. Brain slice models are also time consuming and may take 3-7 days for completion. Further, long term cultures must account for the influence of proliferation while examining tumor cell migration to avoid confounding effects.

3.4.2 Confrontational tissue assays

Confrontational assays [196-200] are employed to assess the migration of tumor cells by culturing normal and cancerous brain tissue in close proximity (e.g., one tissue on top of

63 the other or side by side, Figure 25). Other versions of this assay include employing tumor spheroids in confrontation with normal brain tissue or using normal brain cell aggregates in confrontation with tumor cells. The infiltration of the tumor cells into the normal tissue is then investigated using microscopy. Confrontational assays are limited in utility presenting shortcomings similar to the brain slice assays, the important ones being costly harvesting of normal tissue, failure to permit investigator control of the local environment, and animal-animal variation.

3.4.3 Tumor xenograft models

In addition to the brain slice and confrontation assays, examination of tumor cell dispersion in rat or mouse models is routinely performed. In these models (Figure 25), tumor cells are implanted into the rat brain and their progression over time is monitored

[201-205]. At different time points, the animal is perfused with a fixing agent

(paraformaldehyde) and tissue sections are analyzed using histology and immunofluorescence for cell-specific markers. Analyzing the tissue sections using microscopy and image analysis tools; the extent of tumor invasion can be quantified.

These models adequately mimic the in vivo tumor microenvironment and can be employed to study spatiotemporal tumor distribution (i.e., the different anatomical structures that favor tumor cell dispersion in vivo.) as well as identify dynamics of tumor cell migration (e.g., intracellular cytoskeletal organization [206]). Despite the fact animal models replicate the in vivo environments faithfully; several issues continue to arise in terms of costs associated with animal studies as well as ethical concerns. More specifically, results from these assays may be skewed depending on animal-animal

64 variability and hence are difficult to reproduce. Additionally, results from animal based assays do not translate well to human studies. These assays also fail to provide investigator control of local environments, thereby making studies of specific interactions

(e.g., tumor cell response to specific material cues such as soluble factors) difficult.

Figure 25. Animal derived models. (A) Brain slice assay. Figure adapted from [171], GFP = Green fluorescent protein. (B) Confrontational tissue assays, arrows indicate direction of tumor cell infiltration. (C) Rat/Mouse models of cancer (tumor xenografts).

3.5 Making the transition to 3D: Bridging the gap between 2D and animal models

Whereas most 2D models are relatively easy and fast to execute, they fail to recapitulate the highly complex, three dimensional in vivo niches. A number of scientific reports have demonstrated that cells behave differently in 3D versus traditional 2D microenvironments

[191, 207]. Yet, procedures involving animal models are costly and often present with ethical concerns. Lessons learnt from tissue engineering principles have now been used to 65 create well defined 3D niches for a variety of cell types [208-211] including GBMs.

These environments are usually recreated using either naturally derived biomaterials or synthetic biomaterials that recapitulate several aspects of their in vivo environment and provide important material cues to tumor cells.

3.5.1 Synthetic biomaterials

Synthetic biomaterial platforms, inspired from tissue engineering approaches are now being utilized as in vitro engineered disease models to elucidate mechanisms of tumor progression. Hydrogels and Electrospun Fibers (EFs), biomaterials commonly employed in tissue engineering [42, 212], are the two most successful in vitro disease models currently in use. Both models are attractive in terms of their mechanical and chemical tunability to incorporate a number of cell responsive cues (e.g., adhesion molecules, growth factors). Specifically, hydrogels, cross-linked polymeric materials, mimic structural and mechanical features of brain tissue because of their similarity to glycosoaminoglycans and proteoglycans, found in the native brain ECM. Electrospun fibers mimic fibrous structures (e.g., white matter, blood vessels) that act as highways for

GBM dispersion in vivo.

Several synthetic biomaterials have been used to study glioblastoma cell behaviors. For example, sheets of silicone rubber (poly(methylphenyl) siloxane) [213] and poly

(acrylamide)-based hydrogels [214] have been used to elucidate the role of mechanics on glioma cell migration. Both studies showed that migration is proportional to the rigidity of the underlying substrate. In particular, with poly (acrylamide) based hydrogels [214], it was observed that as rigidity (E ~ 0.8 kPa) approached that of brain tissue, migration was

66 drastically reduced compared to stiff substrates (E ~ 119 kPa) demonstrating a strong role played by the mechanical environment in guiding migration. Similarly, synthetic electrospun fibers derived from poly(ε-caprolactone) (PCL) have been used to investigate the role of topography on glioblastoma behaviors. In particular, glioma migration was observed to be a strong function of substrate topography with glioblastoma cells migrating much faster on aligned PCL fibers versus random PCL fibers. This system also recapitulated in vivo-like migratory morphologies with migration correlating to STAT-3

(Signal transducer and activator of transcription-3) signaling, a known driver of cell migration in vivo [173, 174, 215].

Synthetic materials provide user control over material properties, but unfortunately do not capture complex cell responses unless specific biological recognition sequences are added. Also, to date, studies with these materials do not provide a barrier to migrating tumor cells and represent an intermediate environment between 3D and traditional 2D (a

2.5D environment), making it difficult to examine the role of matrix metalloproteinases

(MMPs) on tumor cell migration. While there have been few studies with these materials, a dramatic increase in their use as 3D tissue analogs with the development of new biomaterial combinations is expected in the coming years primarily because of their simple preparation procedures, tunability and reproducibility.

3.5.2 Natural biomaterials

In addition to synthetic materials, naturally derived scaffold materials are also ideal candidates for developing 3D models for tumor cell migration investigations. Some standard models employed include Matrigel- [190, 216-218] and Collagen- [219-222]

67 based assays. These assays either investigate invasion into the gel by seeding cells on the gel surface or use tumor cell spheroids to examine radial migration of cells away from the tumor core. Whereas these assays are good starting points for studying cell migration in

3D environments and continue to be used, they usually access a limited range of physicochemical properties (e.g., stiffness, ligand density) making certain tumor cell characteristics difficult to capture.

To circumvent these issues, several studies have focused on developing techniques to isolate the role of these singular factors on cell behaviors in 3D by developing multicomponent, tunable natural biomaterial systems. For example, using hybrid gels of collagen and agarose, GBM migration was shown to be inversely related to matrix stiffness in 3D. Further, a change in migration pattern from mesenchymal to amoeboid was also observed [223]. Similarly, by exploiting the gelation dynamics of collagen gels,

GBM migration was also examined as a function of pore size in 3D networks [224].

With a goal of moving towards more physiologically relevant tissue models, investigators have also employed HA as the starting material to create hydrogels as well as incorporating it with other biomolecules to create multicomponent systems. As opposed to other biomaterial systems discussed above, HA is the primary ECM component present in both normal and cancerous brain ECM [160]. These biomaterials have been employed to investigate migration capacity of GBM cells in both 2.5D and 3D cultures

[225-230]. Specifically, using HA-RGD gels, the migration capacity of GBM cells was shown to be a strong function of stiffness and ligand density in 2.5D culture [227]. In 3D culture, migration was permitted only in less dense HA cultures and strongly inhibited by

68 highly dense HA-based hydrogel cultures [227, 228]. This behavior was also observed with CS addition to collagen based gels wherein migration of GBM spheroids was inhibited [231]. In addition to matrix parameters, several other factors such as elastin derived peptides (i.e., kappa elastin, кE) [226] and growth factors (i.e., stromal cell- derived factor-1α and basic fibroblast growth factor) [225] have also been shown to increase migratory capacities of GBMs in HA-based hydrogel systems. Studies have also pointed toward cell specific differences in invading HA-based hydrogels, indicating inherent differences in cell type [229]. Interactions of GBMs with several biomaterial systems and the associated observations are summarized in Table 6.

Studies with natural biomaterial based-systems offer several advantages. For example, with an increase in the development of systems that offer tunability, it is becoming easier to assess the role of competing cues (e.g., chemistry, stiffness) on tumor cell behavior.

Whereas this is also possible with synthetic biomaterials, the importance of translating these findings to the clinic using natural biomaterial systems representative of both, chemical and mechanical features of native tissue cannot be underscored. Also, naturally derived materials can elicit specific signaling responses that can influence cell behaviors in ways as similar to those observed in vivo. However, it is also important to realize that for natural systems, variations in composition can influence experimental findings.

Finally, detailed studies of intracellular signaling cascades and other migration regulatory pathways using existing as well as forthcoming physiological biomaterial platforms should enable development of new therapeutic targets.

69

Compared to animal tissue based models, results obtained using 3D in vitro biomaterial assays are more reproducible. Furthermore, these assays are inexpensive, quick and simple to execute compared to brain slice models or confrontational assays. Brain slice assays or confrontational assays usually require extreme care to avoid contamination

(usually bacterial) of slices or tissues. However, recognizing that glioblastoma behavior is extremely complex, in some cases, it might still be required to utilize a combination of in vitro assays (e.g., brain slice, hydrogel and/or fiber-based assays) to better understand tumor behaviors. For instance, 3D models could be used to identify specific effects in a high throughput manner, which can be further explored in detail using animal models. A schematic of three dimensional cell culture models is shown in Figure 26.

Figure 26. Three dimensional cell culture models for GBM migration studies (A) Hydrogel models used in different configurations. (B) Aligned electrospun nanofiber models.

70

Cell Cell Tumor Culture Assay Observations Refs Biomaterial type Sou Cell Type Type rce Model and Dimensi onality poly SNB- Hu S H, 2.5D D Migration = f [213] (methylphen 19 (stiffness) yl) siloxane) poly (acryl U-373 Hu S H, 2.5D D Migration = f [214] amide) U-87 Hu (stiffness) coated with U-251 Hu fibronectin SNB- Hu 19 R C6 poly (ε- X-12 Hu S, A F, 2.5D S, D Migration is [173, caprolactone U-251 Hu topography ) coated U-87 Hu dependent. 174] with G-8, M fibronectin G-9* Matrigel U-87 Hu A H, 3D D Tumor exerts [217] mechanical stress and traction on its surrounding Collagen U-87 Hu A H, 3D S Invasion = f [219] (collagen concentration at early time points) Collagen U-87 Hu A H, 3D S Invasion was [221] U-373 influenced by GBM 1 tissue cohesion GBM 2 and N-Cadherin GBM 3 expression GBM 4

continued

Table 6. GBM interactions with natural and synthetic biomaterials with specific observations. Species: Hu=Human, M= Mice, R= Rat; Tumor Cell Model: S= single cells, A= tumor aggregates/spheroids; Culture Type: H= Hydrogel, F= Electrospun Fiber; Culture Dimensionality: 2D = Cells cultured on TCPS/glass, 2.5D = Cells cultured on top of hydrogels/electrospun fibers, 3D = Cells encapsulated in hydrogels; Assay Type: S = Static, End point based D = Dynamic, Time Lapse Imaging; *Tumor initating cells implanted into mice to generate tumors and tumor explants cultured on a biomaterial. f indicates functional dependence. 71

Table 6 continued Collagen U-87 Hu S H, 3D D Migration = f [220] (Epidermal growth factor stimulation) Collagen C6 R S, A H, 3D S Migration = f [224] (pore size) Collagen- U-251 Hu S H, 3D S Presence of [222] Tenascin C U-178 Tenascin-C increases invasiveness Chitosan- U-87 Hu S H, 2.5D S Provided an [232] Alginate U-118 Hu environment C6 R leading to formation of solid tumor-like cells. VEGF and MMP-2 secretion ↑ for human cell lines. Collagen- U-373 Hu S, A H, 3D S, D With increasing [223] Agarose agarose, migration mechanism is altered and eventually abrogated. HA U-87 Hu S H, 2.5D S The extent of [229] U-251 invasion was U-343 cell type U-373 dependent. HA CB-191 Hu S H, 2.5D S Invasion [230] CB-193 correlated with hyaluronidase activity. continued

72

Table 6 continued

HA CB-191 Hu S H, 2.5D S Invasion was [225] influenced by adhesion molecules (collagens) and growth factors (stromal cell- derived factor- 1α and basic fibroblast growth factor) HA-кE CB-74 Hu S H, 2.5D S Invasion [226] CB-109 increased in the CB-191 presence of kappa-Elastin (кE), MMP-2 ↑ HA-RGD U-373 Hu S H, 2.5D S, D Migration = f [227] U-87 Hu A H, 3D (stiffness and C6 R ligand density) HA- OSU-2 Hu S H, 3D D Migration = f [228] Collagen (HA density). Collagen- C6 R A H, 3D S Migration = f [231] HA (CS Collagen-CS concentration). Effect of HA not significant

3.6 Potential benefits of 3D brain tissue models over conventional 2D models: Concluding remarks

One area where 3D models could make a substantial impact is in drug discovery and drug testing. 2D cell culture models (e.g., monolayer cultures) are currently used to test efficacy of anticancer drugs and evaluate their potential prior to clinical trials [233].

Despite rapid advances in high throughput drug screening procedures, 2D assays fail to 73 predict the efficacy of anticancer drugs and several studies have indicated that 2D assays have only provided marginal benefits in evaluating anticancer drug sensitivities [234,

235]. This is mainly because cells grown on 2D tissue culture polystyrene (TCPS) undergo a selection process, and, eventually over repeated passaging in vitro, result in cells that have a strong ability to proliferate on TCPS. For instance, some of the cell lines used to study glioblastoma behaviors such as U-87, U-118, and U-138 were established between 1966-69, nearly 45 years ago and over this time period have gained improved survival capacity. Further, a variety of complex environmental cues are lost in 2D TCPS, altering the properties of cells compared to the primary tumor. Therefore, it is of no surprise that current models do not adequately predict tumor in vivo response and that very few drugs, as low as ~5%, eventually achieve approval (with the rest failing in phase

II and III clinical trials)[236]. This is a huge challenge both scientifically and financially as several million dollars are wasted in the process on targets that have no future. Finally, most drug candidates (i.e., for chemotherapy) target proliferation of glioblastomas as opposed to migration.

It is anticipated that biomimetic 3D models would overcome these limitations and serve as powerful predictors for high throughput drug screening and drug discovery by uniquely bridging the gap between 2D and animal models in a cost effective manner

[207, 237]. Thus, it is possible that this would shorten the time for a specific therapeutic drug to reach the market. Further, 3D brain tissue models will also be of immense use to clinicians to investigate tumor migration rate in vitro, thereby guiding patient care and treatment decisions. For example, creation of improved 3D tissue analogs can serve as

74 tailored bioassays for patient derived tumor biopsies for testing different drug combinations. Whereas these models are mainly designed based on ECM of the tumor, the complexity of such models can be increased by adding a variety of cues found in the in vivo microenvironment. For example, cell-cell interactions are equally important and hence investigating interaction of tumor cells with other relevant cell types in the brain

(e.g., oligodendrocytes, neurons, [238, 239]) will improve the field’s understanding of how these interactions influence tumor progression. While increasing the model complexity is not without challenges, understanding the nutritional requirement and culture conditions for each specific cell type and their co-cultures will be crucial. Further, the role of several growth factors in combination with ECM cues on tumor cell behaviors can be also be explored. In addition to predicting tumor cell behaviors, brain biomimetic

3D models would also give us a powerful tool to understand and evaluate fundamental questions in neuroscience (e.g., migration of other cell types in the brain) that are yet unexplored.

However, there are several significant challenges that still need to be overcome in 3D cultures. For example, oxygen limitations are one of the concerns especially for long term cultures. In addition, downstream assays such as western blots, immunofluorescence assays are relatively difficult to perform in some 3D culture platforms. Some of these challenges can be partially overcome. For example, to increase oxygen transport in 3D cultures, bioreactors may be employed taking lessons from the tissue engineering community [240, 241]. Similarly, downstream assays can be performed on thin 3D slices taken from the samples. Finally, crucial information gained from these models could be

75 employed as guide in designing scaffolds for neural tissue engineering, creating stable nerve tissue-electrode interfaces and tissue engineering in general.

76

Chapter 4: Polylysine modified PEG-hydrogels: Biomimetic

Coatings to Enhance the Neural Tissue-Electrode Interface3

Neural prostheses are a promising technology in the treatment of lost neural function.

However, poor biocompatibility of these devices, as discussed in Chapter 1, inhibits the formation of a robust neuron-electrode interface. Several factors including mechanical mismatch between the device and tissue, inflammation at the implantation site, and possible electrical damage contribute to this response. Many researchers are investigating polymeric brain mimetic coatings as a means to improve integration with nervous tissue.

Specifically, hydrogels, constructs also employed in tissue engineering, have been explored because of their structural and mechanical similarity to native tissue. However, many hydrogel materials (e.g., poly(ethylene glycol), (PEG)) do not support cell adhesion. In this chapter, we report a technique to enhance the interface between polymeric brain mimetic coatings and neural tissue using adhesion molecules. In particular, polylysine-modified PEG-based hydrogels were synthesized, characterized and shown to promote neural adhesion using a PC12 cell line. In addition, we examined adhesion behavior of a PEG-copolymer and found that these materials adhere to

3 This chapter with minor modifications has been published in the following reference: S. S. Rao, N. Han, J. O. Winter (2011). Polylysine modified PEG-based hydrogels to enhance the neuro-electrode interface. Journal of Biomaterials Science, Polymer Edition. 22(4-6). 611-625. 77 electrodes for at least 4 weeks. These results suggest that polylysine-PEG hydrogel biomaterials are biocompatible and can enhance stability of chronic neural interfaces.

4.1 Introduction

Neurodegenerative disorders (e.g., Alzheimer’s disease, Parkinson’s disease) cause substantial loss of neural function, affecting over a million individuals every year [7].

Unfortunately, central nervous system (CNS) neurons do not regenerate following injury

[9]. One strategy for treatment has been the use of tissue engineering to enhance nerve regeneration at the injury site; however, these approaches have seen only limited success

[242]. Alternatively, researchers have developed neural prostheses, which restore lost electrical signaling by converting an external signal into an electrical pattern transmitted to cells via microelectrodes [15]. Neural prostheses have had some success, most notably the cochlear implant [17]. However, several challenges, including difficulty in accessing target neuronal tissue [18] and scarring resulting from the host immune response to implantation [19], have prevented the formation of a robust neuron-electrode interface.

For the last several years, our group [18, 32, 33] and others [34-36] have examined the possibility of combining tissue engineering approaches with neural prostheses to create a composite material that promotes integration with target neural tissue. Several biomaterials have been developed as electrode coatings, including conducting polymers

[37, 38], layer-by-layer (LbL) assemblies [41], and polymeric hydrogels [18, 32, 33]).

Among these, hydrogel-based coatings have attracted considerable attention because their structural and mechanical properties mimic native brain tissue. Hydrogels are crosslinked, hydrophilic polymeric structures that swell in solution [42]. Specifically,

78 hydrogels have been employed as drug eluting devices (e.g., for neurotrophin delivery)

[18, 243, 244] for improving electrode-tissue integration. Unfortunately, many of these drugs (i.e., neurotrophins) have a very short half-life in vivo [44]. Additionally, electrode efficacy is compromised after the supply of drugs is exhausted. Therefore, additional cues will be needed to sustain and improve electrode-tissue integration. In the PNS, regeneration occurs not only because of soluble factors but also because of adhesion molecules (i.e., proteins/peptides that promote neural adhesion) [9]. Further, adhesion molecule-modified hydrogels have been utilized as tissue engineering constructs [245].

Adhesion molecules thus are attractive candidates for improving tissue integration with neural biomaterials [64, 246]. These molecules have been previously utilized in combination with conducting polymers [104] as well as with LbL assemblies [41] as neuro-prosthetic coatings although not with hydrogel systems.

Here, we examined the ability of adhesion molecule-modified hydrogel coatings to promote adhesion to electrode surfaces, thereby enhancing the neuron-electrode interface.

Specifically, we investigated poly(ethylene glycol) (PEG)-based hydrogels modified with the non-native adhesion molecule, polylysine (PL). PEG was selected as a hydrogel material because of its non-immunogenic properties [247] whereas PL was selected because of its ability to support neuronal adhesion [80]. PL-modified poly(ethylene glycol) diacrylate (PEGDA) hydrogels were developed, characterized, and investigated for their ability to promote neuronal adhesion. Additionally, recognizing the fact that

PEGDA hydrogels do not normally adhere to hydrophobic electrode surfaces; we examined the adhesion of PEG-hydroxyacid copolymers (e.g., lactide, caprolactone) to

79 electrode surfaces. The addition of hydroxyacids also provides a strategy for biodegradation of the coating, so that it is removed from the electrode surface after the acute immune response to implantation. As proof of concept, poly(ethylene glycol)-poly

(caprolactone) diacrylate (PEG-PCL)) hydrogels were synthesized and their adhesion to model retinal prosthesis electrodes was characterized. Adhesion molecule-modified hydrogel coatings offer great promise to enhance electrode-host integration. Furthermore, this technique is broadly applicable to any implant material i.e., stimulating/recording electrodes. Hydrogels can be tuned to mechanically resemble target tissue and to release soluble factors. The addition of adhesion molecules to this system provides a means to combine chemical and mechanical similarities to native tissue.

4.2 Materials and methods

4.2.1 Preparation of Acryl-PEG-Polylysine (Acryl-PEG-PL)

Polylysine was conjugated to acryl-poly(ethylene glycol)-N-hydroxyl succinimide

(Acryl-PEG-NHS) (Laysan Bio, Inc.) using standard approaches [112]. Briefly, polylysine (MW 30,000-70,000, Sigma Aldrich) was dissolved in 50 mM sodium bicarbonate (Sigma Aldrich) and Acryl-PEG-NHS (MW 3400, Laysan Bio, Inc.) was dissolved in DMSO (Sigma Aldrich) in a 1:2 molar ratio. These solutions were mixed and reacted for ~ 2 hours in an ice bath. The resulting sample was then dialyzed (Slide-a- lyzer cassettes, Fisher) overnight and reconcentrated to the original volume in phosphate buffer saline (PBS) using a centrifugal concentrator (Fisher). For quantification, FITC labeled poly-L-lysine (PLL) (MW 30,000-70,000, Sigma Aldrich) and poly-D-lysine

(PDL) in a 1:10 (weight ratio) were used, whereas PDL (MW 30,000-70,000, Sigma

80

Aldrich) was utilized for all other experiments. Different concentrations of Acryl-PEG-

PL were prepared by diluting the above stock solution in PBS.

4.2.2 Preparation and characterization of Poly (ethylene glycol)-poly

(caprolactone) (PEG-PCL) copolymers

Poly(ethylene glycol)-poly(caprolactone) (PEG-PCL) copolymers were synthesized following Hubbell’s method [248]. Briefly, PEG (MW 950-1050, Sigma-Aldrich) was azeotropically distilled in anhydrous toluene (Sigma-Aldrich) and reacted with – caprolactone (Sigma-Aldrich) (1:2 molar ratio) for ~24h, under an argon blanket, at 130 º

C in the presence of stannous octanoate (Sigma- Aldrich) catalyst. PEG-PCL copolymer was then cooled to ambient temperature, dissolved in anhydrous dichloromethane

(Sigma-Aldrich), and slowly added to cold hexane to precipitate the product. PEG-PCL copolymer was collected using Buchner funnel filtration and vacuum dried overnight.

This intermediate was then subjected to azeotropic distillation in anhydrous toluene and dissolved in anhydrous dichloromethane in presence of triethylamine (Sigma-Aldrich)

(for HCl neutralization). Acrylation of PEG-PCL was performed by reaction with acryloyl chloride (Sigma-Aldrich) (1:2 molar ratio) in anhydrous dichloromethane (1: 4 v/v) added at the rate of ~0.20mL/min. The system temperature was maintained at -20º C for ~ 3h and at ambient temperature for another 60 h. Acrylated PEG-PCL was purified by removal of TEA-HCl salt via filtration, removal of the solvent by rotary evaporation, and subsequently dissolution in anhydrous tetrahydrofuran (Sigma-Aldrich). The product was collected via filtration, dissolved in anhydrous dichloromethane, precipitated using cold diethyl ether, filtered, and vacuum dried in the presence of P2O5 (Sigma-Aldrich).

81

Characterization of PEG-PCL and acrylated PEG-PCL copolymers was performed using

Fourier transform infrared spectroscopy (Nicolet 6700 FT-IR spectrometer, Thermo

Scientific) and nuclear magnetic resonance spectroscopy (NMR Bruker DPX400).

4.2.3 Hydrogel formation and quantification of polylysine attachment

Poly(ethylene glycol)-diacrylate (PEGDA) (MW 3400) was procured from Laysan Bio,

Inc. PEGDA and PEG-PCL hydrogels were prepared following Hubbell’s method [248].

Briefly, PEGDA or PEG-PCL was dissolved in PBS at 22 wt % and supplemented with

1% v/v initiator (0.33 g Irgacure 2959 (Ciba) in 1mL of N-vinyl-2-pyrrolidone (Sigma

Aldrich)). Sixty µl of this hydrogel solution was placed in a 96 well plate and subjected to UV illumination (peak power 11.21 mW/cm2, sample distance ~9mm) for ~10 min.

Acryl-PEG-PL (~30 µl) with initiator was then added to the surface of the PEG-based hydrogel, and the sample was subjected to UV illumination (~ 10 min) (Figure 27). For

PEGDA hydrogels, conjugation efficiency was determined by monitoring the diffusion of unreacted FITC-polylysine over a period of 7 days, against a sham (unconjugated polylysine) and a negative control (no polylysine) using a fluorescent microplate reader

(Excitation wavelength= 485 nm, Emission wavelength = 535 nm). [Sham consisted of

PEGDA hydrogels to which unconjugated PL (i.e., PL in PBS) supplemented with initiator was added and photo polymerized]. The conjugation efficiencies were evaluated using standard concentrations of polylysine. Data was analyzed using ANOVA and pair- wise comparisons were performed using Tukey Kramer’s HSD test (α = 0.05) via JMP statistical software (Version 7). For PEG-PCL hydrogels, conjugation was confirmed

82 using fluorescence microscopy and compared to PEG-PCL hydrogels with unconjugated

FITC-polylysine, as well as PEG-PCL hydrogels with no polylysine.

Figure 27. Schematic of the conjugation procedure.

4.2.4 PC12 cell culture

PC12 cells (ATCC, CRL-1721, Manassas, VA) were cultured using Powdered Ham’s

F12-K Media (Kaighn's modification) (Sigma Aldrich) supplemented with 1.5 g/L

NaHCO3, 12.5% horse serum, 2.5% fetal bovine serum (all Sigma Aldrich) and 1% penicillin-streptomycin (Invitrogen). Cells were cultured in an incubator at 37 C and 5%

CO2. Medium was exchanged every 2-3 days, and cells were sub-cultivated weekly before experimentation.

4.2.5 Quantification of cell adhesion using calcein-AM staining

Cell adhesion to modified hydrogel surfaces was quantified using the Live-Dead assay

(Invitrogen). Non-fluorescent calcein AM is converted by intracellular esterases to intensely fluorescent calcein in live cells, whereas EthD-1 permeates damaged cell membranes and enhances fluorescence by binding to nucleic acids in dead cells.

Polylysine-modified hydrogels were prepared as described previously using PDL. For cell experiments, all solutions were sterile filtered using a 13mm syringe filter, pore size 83

0.22 µm (Fisher) prior to gelation. Modified hydrogels were immersed in sterile PBS for one week and then subjected to UV illumination overnight (for sterilization). PC12 cells at ~2×104 cells/cm2 were seeded onto all hydrogel surfaces and the positive control (well plate coated with polylysine). After incubating for ~ 24 h, all surfaces were washed with sterile PBS thoroughly (3 times). Cells were then stained with EthD-1(1µM) and calcein-

AM (2µM) prepared in sterile Dulbecco’s PBS (D-PBS) (Sigma). Following incubation for ~45 minutes, excess dye was removed using sterile D-PBS (3 times) and fluorescence was observed. To quantify cell adhesion, calcein-AM fluorescence was used since fluorescence intensity is directly proportional to cell number. Background fluorescence from dye entrainment in the hydrogel was subtracted from all measurements by observing hydrogels stained at the same dye concentrations, but with no cells. Images were captured using an Olympus IX71 inverted optical microscope equipped with fluorescence filters at 10X magnification. The fluorescence intensities (a.u.) were normalized to the negative control and compared using ANOVA. Pair-wise comparison was performed using Tukey Kramer’s HSD test (α = 0.05) via JMP statistical software

(Version 7).

4.2.6 PEG-PCL electrode adhesion

Multielectrode arrays (3 x 5 electrodes) used to investigate adhesion of PEG-PCL hydrogels were generously provided by Dr. Stuart Cogan (EIC Laboratories, Norwood,

MA). The arrays consisted of gold electrodes patterned on polyimide substrates [18].

PEG-PCL hydrogel was formed by depositing ~ 1 µl of the precursor solution onto the electrode array surface and subjecting this bolus to UV photo-polymerization. To

84 evaluate the efficacy of the hydrogel coating, static adhesion of the coating as well as adhesion following electrode insertion was monitored. For electrode insertion, PEG-PCL hydrogel coated electrodes were inserted into a 1 wt% agarose brain tissue phantom and removed immediately. Static integrity of the hydrogel coating over time was evaluated by placing coated electrodes under a 1 wt% agar gel blanket, which physically constrained the hydrogel bolus on the electrode surface, mimicking the effect of tissue. Reflected DIC images for both the tests were collected using an Olympus BX41 reflected optical microscope. Images were converted to grayscale using Adobe Photoshop (Version 10.0).

4.3 Results

4.3.1 PL conjugation to PEG hydrogels

The extent of PL conjugation to PEG hydrogels was examined using fluorescence microscopy to evaluate FITC-PL-conjugated PEG hydrogels (sample), PEG hydrogels with FITC-PL added but not conjugated to the hydrogel backbone (sham), and PEG- based hydrogels with no PL (negative control). Fluorescence values were normalized to the negative control (no FITC-PL, fluorescence intensity = 1). PL conjugation at different concentrations of acryl-PEG-PL was quantified and in all cases was higher and statistically significant (p < 0.0001 for 1X and 0.01X, p = 0.0056 for 0.1X, ANOVA,

Tukey test) for PL-conjugated PEG hydrogels (sample) versus the negative control

(Figure 28). Unconjugated PL (sham) hydrogels were also evaluated for comparison.

Intuitively, unconjugated PL should be released from the hydrogel over time. However, we observed that some PL becomes physically entrapped within the gel matrix, and unconjugated PL (sham) hydrogels generally exhibit fluorescence intensity higher and

85 statistically significant (Tukey test) from the negative control. [In the case of 0.1X dilution, this value, although higher than negative control was not statistically significant

(power = 0.93 < 0.95).] Comparing PL-conjugated (sample) and unconjugated (sham) hydrogels, it was observed that fluorescent intensity of sample hydrogels was higher and statistically significant from that of sham hydrogels at all dilutions investigated. Exact concentration values (obtained from a standard curve of known concentrations) are summarized in Table 7.

Figure 28. Conjugation of PL with PEG-DA hydrogels. (N=4 for 1X, N=4 for 0.1X, N=5 for 0.01X, * indicates statistical significance compared to negative control. 86

Dilution of Acryl-PEG-PL 1X 0.1X 0.01X (N=4) (N=4) (N=5) Conjugated + Entrapped PL (µg/ml) 31.52 ± 5.32 6.21 ± 1.88 0.45 ± 0.11 Entrapped PL (µg/ml) 9.18 ± 6.77 3.44 ± 0.49 0.28 ± 0.05 Table 7. Concentration of bound and entrapped PL (Reported as Average ± SD)

4.3.2 PC12 cell adhesion on PEG hydrogels

Cell response to PL-modified PEG hydrogels was assessed using the PC12 cell line, a model for neuronal behavior [249-251]. PC12 cells convert to a neuronal phenotype after exposure to nerve growth factor (NGF) [251], including extension of neurites. Because the primary focus of this study was cell adhesion and because neurite extension complicates quantification of cell surface area, NGF was excluded from the cell culture medium. Additionally, because this study focused on proof-of-concept and demonstration of biocompatibility, additional cell types (e.g., astrocytes, microglia) were not investigated. PC12 cell adhesion was quantified using calcein AM staining with fluorescence intensity serving as an indirect measure of cell number.

PC12 cells adhered to PL-modified PEG hydrogels at the lower range of acryl-PEG-PL concentrations investigated. Both ‘sample’ and ‘sham’ hydrogels displayed statistically significant adhesion (p < 0.0001 ANOVA, Tukey test, power = 0.99) from the negative control (Figure 29, 30). At high concentrations of PL (i.e., 1X dilution), substantial cell death was seen (EthD-1 staining). This is not unusual, as PL is known to be cytotoxic at high concentrations [85]. Although PC12 cell adhesion to the well plate (positive control) was also examined, cell adhesion was not compared with ‘sample’, ‘sham’ and negative control hydrogels because it is known that differences in substrate stiffness (i.e., 87 soft hydrogel vs. rigid tissue culture plastic) strongly influence cell morphology and spreading [85].

Figure 29. PC12 cell response on PL-modified PEGDA hydrogels. (A) Sample. (B) Sham. (C) Negative control (No PL). (D) Positive Control (PL).

4.3.3 PEG-PCL hydrogel coatings

Given these promising initial results with PEG hydrogels, we next investigated response to PEG-hydroxyacid (PEG-HA) hydrogel composites (i.e., PEG-PCL). PEG-HA composites have been previously used as neuro-prosthesis coatings that provide delivery of soluble neurotrophins [18, 32, 33] to enhance neuronal survival, extension, and tissue integration. Additionally, in contrast to PEG hydrogels, PEG-HA coatings demonstrate much greater adhesion to hydrophobic electrode array surfaces, which are primarily 88 composed of parylene or polyimide. PEG-HA hydrogels are also degradable through hydrolysis of the ester linkage, thus providing a means of controllably removing the coating after the acute immune response. Given a target of ~ 6-8 weeks for the acute immune response, we investigated slowly-degrading PEG-poly(caprolactone) hydrogels for ability to support PL-conjugation and promote adhesion to neural prosthesis electrode arrays.

Figure 30. Normalized fluorescence for PC12 cell adhesion on experimental PEGDA hydrogels. (N=6) * indicates statistical significance compared to negative control.

4.3.3.1 Synthesis and characterization

PEG-PCL polymers were synthesized and characterized by FTIR and NMR spectroscopy. In the FTIR spectra for PEG-PCL (Figure 31A), notable peaks consisted of 89

-1 1735 cm = PCL, COO; 2870 = PEG, CH2; 3500 = PEG, OH. For the diacrylate PEG-

PCL, peaks included 1725 = acryl, COO; 1735 = PCL, COO; 2870 = PEG, CH2; 3550 =

PEG, OH [18, 252, 253]. The reduction of the OH peak (3550, PEG) confirms the creation of the ester bond between PEG-PCL and acryloyl chloride following the acrylation step. In the 1H NMR spectrum for diacrylate PEG-PCL (Figure 31B), notable peaks consisted of 3.64 = PEG, CH2, 4.08, 4.15, 4.23, 4.31 = PCL, CH2 (a, position as shown in chemical structure, Figure 31C), 1.67 = PCL, CH2 (b), 2.33 = PCL, CH2 [253]

(c), 5.81, 6.12, 6.41 = acryl, CH, CH2 [18, 254]. The average number of CLs per side

PEG is ~ 0.967 as calculated using yield of PEG-PCL. The yield for acrylated PEG-PCL copolymer as calculated from NMR spectra using peak integration was ~ 88.4%. PL- conjugated hydrogels were created using UV-photopolymerization of acryl-PEG-PL segments to the PEG-PCL hydrogel backbone. To confirm conjugation, FITC-PL was used and its fluorescence was observed. As expected, PEG-PCL hydrogels with FITC-PL modification exhibit fluorescence, whereas unmodified PEG-PCL hydrogels do not

(Figure 32).

90

Figure 31. Analysis of PEG-PCL polymer. (A) FTIR Spectroscopy. (B) NMR Spectroscopy. (C) Chemical Structure.

91

Figure 32. Conjugation of PL to PEG-PCL hydrogels. (A) PL conjugated/entrapped (Sample). (B) PL entrapped in PEGPCL hydrogel. (C) Negative Control (No PL).

4.3.3.2 PEG-PCL electrode adhesion

Adhesion of PEG-PCL hydrogels to an electrode array (Retinal prosthesis, Center for

Innovative Visual Rehabilitation, Figure 33) was confirmed using insertion-removal and agar blanket tests.

Figure 33. Retinal Implant Electrode.

The insertion-removal test showed that the hydrogel bolus remains intact and is minimally influenced by the shear force arising from the electrode insertion and removal 92 process (Figure 34, N=2). Further, the agar blanket test showed that PEG-PCL hydrogels can adhere to electrode surfaces for at least 4 weeks (Figure 35, N=3).

Figure 34. Coating integrity (A) before and (B) after insertion into an agarose tissue phantom. Black arrows indicate the edges of the polymer coating.

Figure 35. Static PEG-PCL coating integrity under agarose tissue phantom at time (A) t= 0. (B) 14 days. (C) 28 days. Black arrows indicate the boundary of the hydrogel coating. (Scale bar = 500 µm).

Degradation of the polymer was observed (evidenced by decreasing surface area occupied by the hydrogel bolus on the electrode surface); however, the rate of

93 degradation was slower when compared with other PEG copolymers e.g., poly(ethylene glycol)-poly(lactic acid) [18], permitting hydrogel adhesion over longer time periods.

4.4 Discussion

Here, we investigated the potential of adhesion molecule-modified PEG-hydrogels to promote neuronal adhesion and thereby improve integration of prosthetic devices with the body. In particular, we examined the effect of polylysine (PL), a non-native biomolecule known to enhance neuronal adhesion. At almost all concentrations of PL investigated, conjugated PL (sample) and sham hydrogels displayed higher and statistically significant PL incorporation from the negative control. This indicates that PL can be both bound to the PEG hydrogel (sample) and physically entrapped within the hydrogel matrix (sample and sham). In either case, only limited diffusion of PL from the hydrogel matrix is observed, indicating the stability of the adhesion molecule-modified neural interface. It should be noted that in the case of the 0.1 X Acryl-PEG-PL dilution, the PL concentration in the sham was higher, although not statistically significant from that of the negative control. The statistical power of this experiment was 0.93, slightly lower than 0.95 and this difference would most likely have been detected by increasing the sample size, N.

PC12 cells exhibited two behaviors on these surfaces. At low PL concentrations (i.e.,

0.01X dilution), PC12 cell adhesion was improved on PL-modified hydrogel surfaces when compared with the negative control. At high PL concentrations (i.e., 1X dilution), cell death was observed. This is not unexpected as PL at high concentrations is known to be cytotoxic [85], resulting in cell lysis. Thus, the potential for PL release to the body

94 will be important to quantify in future in vivo work. PL release from PEGDA gels is limited to desorption of surface bound and lightly entrapped molecules during an initial five day period. In this study, all hydrogels were immersed in solution for one week before cell testing to eliminate adsorbed PL. Release from PEGPCL hydrogels would occur as the polymer degraded and is expected to proceed on the order of weeks [248], a sufficiently slow time course to prevent high concentrations of PL from being achieved.

In the low PL concentration region, PC12 adhesion on the conjugated sample was higher than that to the unconjugated sham, although not statistically significant; indicating that physically entrapped PL was equally able to support neuronal adhesion. This could be explained through the mechanism of PL-mediated cell adhesion. PL promotes adhesion via the non-receptor mediated cell binding mechanism; i.e., positively charged PL attracts the negatively charged cell membrane resulting from the glycocalyx [246]. It also enhances deposition of serum proteins from medium, thereby indirectly improving cell adhesion. This is in contrast to other adhesion molecules (e.g., laminin, collagen) that promote cell adhesion via ligand-receptor (LR) [246] interactions. In those cases, conjugation of the adhesion molecule to the hydrogel may be necessary to generate traction as the cell pulls on the adhesion molecule. Molecules that are not bound, but only “entrapped” within the hydrogel, often do not permit the cell to generate the necessary traction force to support adhesion. Additionally, adhesion molecules that employ the LR mechanism primarily promote adhesion through specific sites (e.g., RGD in the case of collagen, IKVAV in the case of laminin [246]). Steric hindrance of an entrapped molecule can limit access to the adhesion site. Our results suggest that because

95

PL supports adhesion through an alternative mechanism, conjugation may not be required. However, when employing PL, it is important to consider molecular weight as well as PL concentration, as cell response may be altered to PL of varying molecular weight and concentration [85]. It should also be noted that these results may not apply only to neural adhesion, adhesion of other cell types (e.g., astrocytes, microglia) may also be encouraged by PL-modified hydrogels; which will be investigated more fully in future in vivo studies.

We also synthesized degradable PEG-PCL hydrogels and demonstrated PL attachment to these surfaces. The mechanism of PL attachment to this hydrogel is similar to that of

PEGDA hydrogels because both hydrogels are formed via photo-polymerization. The acryl terminus of acryl-PEG-PL serves as an ‘anchor’ for PL attachment in both cases.

We demonstrated similar PL attachment for both PEGDA and PEG-PCL hydrogels.

Additionally, we examined the stability of PEG-PCL hydrogels on neural prostheses surfaces. We showed that PEG-PCL hydrogels can adhere to electrodes for at least 4 weeks, at which point degradation processes become significant. This is an enhancement over previous PEG-PLA hydrogel coatings that adhered for up to 11 days [18]. Long term adhesion of the polymeric hydrogel is vital for providing sustained biochemical cues that permit neurons to form firm contacts with the electrode array. In addition to adhesion, the hydrogel coating should withstand shear forces arising from the electrode insertion process. This was examined via the insertion test, and the coating evidenced very little difference before and after insertion into an agarose tissue phantom. It is possible that hydrogel adhesion and degradation may be altered by electrical stimulation.

96

Detailed electrical stimulation experiments should yield further insight into this behavior.

Additionally, we have also demonstrated the ability to pattern these coatings to minimize bulk coating resistance and selectively coat desired electrode zones using masks of specific shapes (Figure 36 shows an “O” pattern, mimicking the Ohio State logo).

Figure 36. Patterned brain mimetic PEG hydrogel coatings. (A) Schematic of patterning technique. (B) Patterned “O” s stained using Rhodamine observed via fluorescence microscopy (C) Patterned “O” s observed using phase contrast microscopy (D) Ohio State “O” logo.

4.5 Conclusions

PEG-based hydrogels are non-immunogenic, biocompatible and are composed of FDA approved materials. As such, they hold tremendous promise as coatings to enhance biocompatibility of implanted devices. Here, we show that adhesion molecule-modified

PEG hydrogels could enhance integration of neural prostheses with target tissue.

However, the PL-modified PEG hydrogel system described here is a passive coating that 97 promotes neuron attachment by direct contact and is limited by anatomical considerations and proximity to target tissue. In future work, we hope to combine adhesion molecule presenting hydrogels with soluble factor elution to further enhance tissue-device integration. We have already shown that soluble growth factors can influence neurons distant from the electrode surface [18] and hope to combine this approach with PL- modified hydrogels. Adhesion molecule-modified hydrogels, especially in combination with factor elution, could enhance the neuro-electrode interface and increase efficacy of current neural prostheses.

98

Chapter 5: Inherent Interfacial Mechanical Gradients in 3D

Hydrogels Influence Glioblastoma Multiforme Tumor Cell

Behaviors4

Cells sense and respond to the rigidity of their microenvironment by altering their morphology and migration behavior. To examine this response, hydrogels with a range of moduli or mechanical gradients have been developed. Here, we show that edge effects inherent in hydrogels supported on rigid substrates also influence cell behavior. A biomimetic Matrigel hydrogel was supported on a rigid glass substrate, an interface which computational techniques revealed to yield relative stiffening close to the rigid substrate support. To explore the influence of these gradients in 3D, hydrogels of varying

Matrigel content were synthesized and the morphology, spreading, actin organization, and migration of glioblastoma multiforme (GBM) tumor cells were examined at the lowest (< 50 µm) and highest (> 500 µm) gel positions. GBMs adopted bipolar morphologies, displayed actin stress fiber formation, and evidenced fast, mesenchymal migration close to the substrate, whereas away from the interface, they adopted more rounded or ellipsoid morphologies, displayed poor actin architecture, and evidenced slow

4 This chapter with minor modifications has been published in the following reference: S. S. Rao, S. Bentil, J. DeJesus, J. Larison, A. Hissong, R. Dupaix, A. Sarkar, J. O. Winter (2012). Inherent interfacial mechanical gradients in 3D hydrogels influence tumor cell behaviors. PLoS ONE, 7(4): e35852. 99 migration with some amoeboid characteristics. Mechanical gradients produced via edge effects could be observed with other hydrogels and substrates and permit observation of responses to multiple mechanical environments in a single hydrogel. Thus, hydrogel- support edge effects could be used to explore mechanosensitivity in a single 3D hydrogel system and should be considered in 3D hydrogel cell culture systems.

5.1 Introduction

Cell migration is a complex, broad-ranging phenomenon strongly influenced by cues from the external environment such as its chemical nature, topographical architecture, and rigidity [255]. It is now widely appreciated that cells can sense the stiffness of their environment and accordingly alter their response [256]. This was first established in a landmark publication by Pelham and Wang [257], who showed that fibroblasts as well as kidney epithelial cells alter their spreading behavior and motility when plated on substrates with different moduli. Since then, several studies in both two dimensional (2D) and three dimensional (3D) environments have corroborated this finding with other cell types in vitro (e.g., neurons [258], endothelial cells [259], myoblasts [260], cancer cells

[261]).

Both artificial (e.g., poly(acrylamide) and poly(ethylene glycol)-based systems [262]) and natural (e.g., collagen [263]) polymer hydrogels have been extensively employed to study the effect of cell response to changing substrate rigidity. However, if a single modulus hydrogel is used; several hydrogels are needed to explore effects across a range of mechanical properties. Recently, investigators have begun to incorporate stiffness gradients into hydrogel systems [264-272]. These gradients have been shown to better

100 mimic in vivo cell response compared to culture in a single modulus mechanical environment [273]. Additionally, they can induce directed cell migration (referred to as

“durotaxis” or “mechanotaxis”), in contrast to the random migration that occurs in uniformly rigid microenvironments. However, most of these approaches limit cell culture to 2D, which has been shown to differ from 3D in vivo conditions. There are very few studies in 3D exploring the effects of mechanical gradients on cell behaviors [268, 274].

Here, we explore mechanical gradients produced by edge effects at the interface of a rigid support with a soft gel on tumor cell, specifically GBM cell behavior in 3D. Edge effects are a specific engineering phenomenon in which properties of a material alter as a result of interactions with the surrounding medium. This occurs both at the material interface and also in an interfacial boundary region adjacent to the interface. As such, we have examined cell behaviors at a hydrogel-support interface, in a ~ 200 µm region surrounding this interface, and as a control, in the bulk of the gel (> 500 µm from the rigid support). We speculated the existence of these gradients in hydrogels, as softer gels

(e.g., Matrigel, modulus ~ 450 Pa [275]) are usually supported on rigid plastic/glass tissue culture plates (modulus of glass, plastic > 100,000 Pa [176]), providing a sharp interface between mechanical moduli. We performed finite-element analyses to support our expectation of the presence of these gradients. We then explored the response of glioblastoma multiforme (GBM) cells, a type of brain cancer previously shown to be sensitive to stiffness in 2D [214] and 3D [223], in this system, characterizing cell spreading and morphology, intracellular actin organization, and migration capacity. As a model, we used Matrigel hydrogels, which have previously been employed as a substrate

101 to study GBMs in the traditional transwell insert assay and as individual gels [216, 218], supported on glass. These inherent stiffness gradients do not require external devices or alteration of ligand density or chemical composition and could easily be expanded to other hydrogels and substrates. Further, to our knowledge, this is one of the first studies to show that inherent interfacial mechanical environments in a single, 3D hydrogel influence tumor cell response.

5.2 Materials and methods

5.2.1 Ethics statement

The Ohio State University's Institutional Review Board approved this study under IRB protocol 2005C0075 (dated November 7, 2008). Written consent from all participants in the study was obtained in accordance with the protocol.

5.2.2 Modeling

A Finite Element Model (FEM) was created using ABAQUS CAE 6.8-1 software

(Dassault Systèmes Simulia Corporation, Providence, RI, 2008). This model focused on the mechanical environment exhibited by 100% v/v Matrigel in the vicinity of the bottom of a cell culture well plate having a diameter of ~ 7 mm. The sides of the gel were free and without curvature at the bottom thus forming a cylinder. The model simulated an indentation test where a steel indenter tool was used to compress the top of Matrigel with heights of 12.5 μm, 25 μm, 50 μm, 100 μm, and 200 μm. The tip of the indentation tool was spherical and had a diameter of 10 μm, which is approximately the same diameter as the cells being studied. Following the Matrigel indentation, the mechanical properties were obtained. The effective stiffness was determined by fitting a straight line to the

102 force-deflection curve. An axisymmetric model captured the cylindrical geometry of the

Matrigel and spherical indenter. The bottom of the Matrigel was fixed mechanically to simulate the hard substrate at the well bottom, by preventing the nodes of the elements from horizontal and vertical translation. Finally, modeling results for heights > 200 µm are not included, as additional changes in height did not influence results.

The following assumptions were made for the model: a) The Matrigel was assumed to be elastic and isotropic with an elastic modulus of 450

Pa [275]. A linear elastic model was chosen for simplicity. It is recognized that

Matrigel may in fact exhibit viscoelastic properties; however, because the model

examines effective stiffness (i.e., instantaneous elastic response), the material will

appear stiffer near the rigid support regardless of the model chosen. b) The contact surface between the indenter and Matrigel was assumed to be frictionless.

Cells interact with gels via both compression and tension forces. Compression was

chosen for this model as this is the least sensitive to the adhesion between the

Matrigel and glass. However, results using a tension or shear model would be

qualitatively similar.

5.2.3 OSU-2 cell isolation and in vitro cell culture

OSU-2 cells were isolated from a GBM patient at the Ohio State University under human

IRB protocol 2005C0075. Briefly, tumors were washed with media containing 200 units penicillin, 200 µg streptomycin, and 0.5 µg/ml amphotericin B (all from Invitrogen).

Tumor samples were then subjected to 200 units/ml type 1A collagenase (Sigma) for ~ 4 hours, triturated, centrifuged at 250g (~5 min), and resuspended in cell culture media

103

(DMEM/F12, Invitrogen) containing 10% fetal bovine serum (Invitrogen), 100 units penicillin, 100 µg streptomycin, 0.25 µg/ml amphotericin B. Cells were cultured in a

37ºC, 5% CO2 environment, fed 2-3 times weekly, and passaged on reaching confluency.

A schematic of the isolation procedure is shown in Figure 37.

Figure 37. Schematic of cell isolation procedure.

Histopathology at the time of operation confirmed tumor type and grade and to further confirm their astrocyte lineage, OSU-2 cells were stained with the glial fibrillary acidic protein (GFAP) marker (Figure 38).

Figure 38. OSU-2 cells in culture. (A) Hoechst stain labels the nucleus blue; rhodamine-GFAP (e.g., glial fibrillary acidic protein) labels the cytoskeleton. GFAP is an intermediate protein expressed by astrocytes. GFAP staining was performed to confirm astrocytic lineage. (B) Phase contrast image of OSU-2 cells in culture. Scale bar indicates 100 µm. 104

5.2.4 OSU-2 cell seeding in BD Matrigel

To encapsulate cells in BD Matrigel (BD Biosciences), OSU-2 cells were pre-labeled with Cell Tracker Green CMFDA (Invitrogen), suspended in cell culture medium, and mixed at ~3000 cells/80 µL hydrogel with ice-cold Matrigel at varying concentrations

(40, 55, 70, 85 v/v %) in an ice bath. Constructs were incubated at 37ºC, 5% CO2 for ~

0.5 hours prior to addition of additional OSU-2 cell culture media to encapsulate the cells. Cells were also seeded on BD Matrigel in 2D after the initial gelation of Matrigel constructs at similar concentrations. All Matrigel constructs were prepared in 16-well

Lab-Tek chamber slides (Thermo scientific).

5.2.5 OSU-2 morphology and cell spreading characterization in 3D Matrigel

OSU-2 laden Matrigel constructs were prepared as described above. After ~16 hours, still images were captured from each gel at different gel heights using an inverted microscope

(Olympus IX71) (N=3 hydrogels for every formulation) equipped with a spinning disk confocal attachment and a Photometrics Evolve EMCCD camera. Data were subjected to image analysis using NIH ImageJ image analysis software. Discrete cells that were in focus in each image were analyzed to obtain cell area and aspect ratio at different gel heights. Cell areas and aspect ratios versus height are reported as average ± S.D. for total cells found at a particular gel height. Cell areas at the lowest and highest gel positions

(images obtained using a confocal microscope (LSM 510; Zeiss, Minneapolis, MN)) were compared and analyzed for statistical differences. Because of variations in surface roughness, cell height was measured from the first plane of observed cells. Thus, the zero point of each chart is equivalent to the lowest plane in which cells were observed and not

105 necessarily the bottom of the substrate. To further quantify cell position, confocal Z- stacks were collected and prepared using the ImageJ Volume Viewer Plugin (metadata available on request).

5.2.6 Immunostaining for actin in 3D Matrigel

OSU-2 cells were seeded in Matrigel constructs as described above. After ~16 hours, cell-gel constructs were fixed in 4 wt/v% paraformaldehyde (Sigma) for 20 min, washed with phosphate buffer saline (PBS), extracted with Triton X-100 (Sigma) solution for 15 min, and blocked with bovine serum albumin (BSA) (Jackson ImmunoResearch) solution overnight at 4ºC. Constructs were then incubated with Alexa Fluor® 633 phalloidin

(Invitrogen) overnight at 4ºC and imaged using fluorescence microscopy to observe actin distribution in 3D Matrigel constructs.

5.2.7 Real time cell tracking in 3D Matrigel

OSU-2 cells (~5000 cells/well) pre-labeled with Cell Tracker Green CMFDA

(Invitrogen) were encapsulated in varying Matrigel concentrations as described above.

OSU-2 cell migration experiments were performed using a confocal microscope (LSM

510; Zeiss, Minneapolis, MN) equipped with a weather station to maintain a 37ºC, 5%

CO2 environment. After ~16 hours, a series of still images of cells in the lowest and highest planes (difference in lowest and highest plane heights ≥ ~ 900 µm) inside the gels were captured every 20 minutes for 12 hours. These images were then concatenated and converted to movies using NIH Image J and were subsequently tracked using the M-

Track J plug-in. At least 15 individual cells were tracked at the lowest and highest gel positions (N=3 hydrogels per condition). In most cases, considerable gel movement (i.e.,

106 swelling) was observed as the experiment progressed. This was corrected using the

StackReg plugin (available at http://bigwww.epfl.ch/thevenaz/stackreg/) that permitted stack alignment at different time points. Migration speeds were then computed for individual cells by dividing the total length of movement by the observation time and are reported as average ± SD for the lowest and highest gel planes examined per condition.

5.2.8 Statistical analysis

Statistical analysis was performed using JMP software (Version 9). All measurements at the highest and lowest gel positions were compared using ANOVA and the Student’s t- test.

5.3 Results

5.3.1 Modeling the substrate/gel interface

Recognizing that there are often edge effects at the interface of two materials with dramatically different moduli, we hypothesized that hydrogels supported by rigid substrates (i.e., glass or tissue culture plates) would display edge effects in their mechanical moduli near the substrate interface. We further hypothesized that these inherent mechanical gradients might influence cell behavior in a single, 3D hydrogel construct. To determine the potential extent of edge effects, we modeled the substrate- hydrogel interface for Matrigel, a common hydrogel biomaterial, on glass using a finite element model. Because it would be difficult to measure forces directly within 3D gels,

FEM was employed on gels of different heights to imitate different z positions within a thick 3D gel. However, the cell-laden gels used for experimentation were of the same height (~ 2 mm) with cells occupying different positions within the gel.

107

In each simulation, the indenter was displaced by 5 μm and the resulting force and stress contour were obtained (Figure 39).

Figure 39. Mechanics of the gel-glass interface modeled using FEM. (A) Stress contour plots of Matrigel with varying height. Axisymmetric elements used. Von Mises stress is an equivalent stress that includes both normal stress (tension/compression) and shear stress contributions. It is calculated from the stress components acting at each location and gives a convenient way of comparing the overall magnitude of stress in different regions. (B) Stress felt at the Matrigel-glass interface as a function of gel height. 108

As Matrigel thickness approached the size of the indenter tip, which represents the size of the cell, the maximum stress in the gel increased. For Matrigel samples with heights greater than 50 μm, the stress field around the indenter did not interact with the rigid well bottom, whereas for heights less than 50 μm the stress field did interact with the well bottom. Consequently, decreasing the gel thickness led to a stiffer response, since the indenter (i.e., cell) started to feel the effects of the rigid substrate beneath the Matrigel.

The influence of gel height on stiffness was also examined using the slope generated by plotting the reaction force experienced by the Matrigel due to the indenter displacement

(Figure 40).

Figure 40. Reaction force vs. 5 μm displacement of the indenter. Insert illustrates a decrease in stiffness with increasing Matrigel height due to a 5 μm indenter displacement. The stiffness insert is the slope extracted from the displacement vs. reaction force plot using the reaction force experienced by Matrigel when the indenter reaches a displacement of 5 µm.

109

The stiffness of the Matrigel is its resistance to deformation due to an applied force, represented by the slope of the curves in Figure 40. The stiffness of the Matrigel decreased as the thickness of the Matrigel increased (Figure 40), as evidenced by the decrease in slope of the reaction force versus displacement curve. Changes in stiffness were more dramatic for smaller thicknesses of Matrigel (i.e., < 50 m), whereas at heights > 50 m changes in stiffness with increasing gel height were negligible.

5.3.2 OSU-2 cell spreading in 3D Matrigel

To examine the influence of inherent interfacial mechanical gradients on cell behavior,

OSU-2 cells encapsulated in Matrigel were analyzed for cell spreading area and aspect ratio. Because of variations in surface roughness, zero height was normalized to the first plane in which cells were observed. To further verify cell position, confocal imaging was performed. Z-stack 3D views of cells in interfacial regions (e.g., ~ 0-50 µm and 0-100

µm) are shown in Figures 41 and 42 taken from supplementary stacks S1 (available at http://youtu.be/r0loQ1LkSAk) and stack S2 (available at http://youtu.be/q_P7ohN1J24) respectively. These stacks and images indicate that cells at positions lower than ~ 15 µm most likely make some contact with the rigid support, whereas cells at positions above

~30 µm are most probably fully embedded in hydrogel. However, cells at the lowest observation plane (i.e., position 0) demonstrated a distinct morphology from those cultured on 2D rigid supports. These cells displayed mostly spindle-shaped morphologies with large processes versus those on bare 2D glass surfaces, which evidenced mostly fan- or tear drop-shaped morphologies (Video 5.1, available at http://youtu.be/ADLaZtKtf8U vs. Video 5.9, available at http://youtu.be/A3QTxwOKDWc).

110

Figure 41. Still images taken from a Z-stack of fluorescently-labeled cells in a 40% v/v Matrigel (0-50 µm, step size = 5 µm). (A) Brightfield/fluorescence Z-stack shown as a montage. (B) Rotated views of the Z-stack shown in A. White arrow indicates the same cell, at position 30 µm, which is clearly embedded within the hydrogel.

111

Figure 42. Images from a brightfield/fluorescence Z-stack of fluorescently-labeled cells in a 40% v/v Matrigel (0-100 µm, step size = 5 µm). White arrow indicates a cell, at position 15 µm, whose edge is in contact with the rigid glass support while the cell body is embedded in the hydrogel. The asterisk indicates a cell, at position 90 µm, fully embedded in the hydrogel.

Also, calculation of individual cell aspect ratios (ratio of major to minor axis of a single cell by fitting an ellipse) showed that cells at the lowest position had statistically higher aspect ratios compared to those plated on glass (Figure 43, 40% (v/v) Matrigel, data for

55%, 70% and 85% (v/v) compositions are also significantly higher compared to glass).

112

Figure 43. Box plot of individual cell aspect ratios comparing cells in the lowest observation plane (< ~50 µm) in 40% (v/v) Matrigel versus Bare Glass. * indicates statistical significance (p < 0.0001), n = 206 cells for glass, n = 20 for lowest observation plane in 40% (v/v) Matrigel.

Morphology in 3D gels varied with changing gel height. For example, cells near the lowest observation plane (i.e., < ~50 µm) showed more elongated, highly bipolar morphologies, whereas cells at higher observation planes (i.e., > ~500 µm) showed rounded morphologies with short processes in some cases (Figure 44).

113

Figure 44. OSU-2 cell behaviors as a function of observation plane in 40% (v/v) Matrigel. (A) Schematic of cell-hydrogel constructs showing morphology observation at different “z” planes. (B) OSU-2 cell area. Representative cell morphologies are shown in the insets. As a result of surface roughness, zero height was set to the first plane of observed cells, which may not necessarily correspond to the substrate surface. (C) OSU-2 morphology at different heights. Representative heights are shown in the chart. Scale bar = 200 µm. (D) OSU-2 aspect ratio. 114

[Still images, with their observations planes, from a typical experiment of cells encapsulated in 40% v/v Matrigel are shown in Figure 44C]. This behavior was quantified as a function of observation plane (Figure 44, 45, 46, and 47) with OSU-2 cells displaying drastically reduced cell area as well as aspect ratio as distance from the lowest observation plane increased for all gel formulations investigated.

Figure 45. OSU-2 cell morphology quantification. (A) OSU-2 cell area and (B) aspect ratio as a function of observation plane in 55% (v/v) Matrigel.

For instance, OSU-2 cells encapsulated in 40% v/v Matrigel at the lowest observation plane displayed an average cell area of ~ 1340 ± 470 µm2 (aspect ratio ~ 10.4 ± 7.3) versus cells at the highest position investigated displaying an average area of ~ 400 ± 270

µm2 (aspect ratio ~ 1.6 ± 0.8). In comparison, the area of cells cultured on bare glass was

2309 ± 1232 µm2 (aspect ratio ~ 2.2 ± 1.5), distinct from observations in both higher and lower positions. Also, deviations in cell area measurements reduced with increasing distance from the lowest observation plane. 115

Figure 46. OSU-2 cell morphology quantification. (A) OSU-2 cell area and (B) aspect ratio as a function of observation plane in 70% (v/v) Matrigel.

Figure 47. OSU-2 cell morphology quantification. (A) OSU-2 cell area and (B) aspect ratio as a function of observation plane in 85% (v/v) Matrigel.

The large deviations observed at low observation planes are a result of the presence of two populations of cells (Figure 44): cells displaying spindle-shaped spread morphology and rounded cells, whereas at more distant observation planes, cells were primarily rounded. Statistical analysis (Student’s t-test as well as non-parametric data comparison

116 using Wilcoxon method) confirmed that cell areas for highest and lowest observation planes were statistically significant for all formulations (p = 0.0003 for 40% v/v and 55% v/v, p = 0.0005 for 70% v/v, p = 0.0001 for 85% v/v reported for the Wilcoxon method).

For comparison, we also examined cells plated on 2D Matrigel (i.e., cultured on top of

Matrigel surfaces), which behaved similar to those at the highest 3D gel positions for all gel formulations (Figure 48).

Figure 48. OSU-2 cell morphology in 2D Matrigel for all formulations. Scale bar = 200 µm.

5.3.3 OSU-2 intracellular morphology in 3D Matrigel (actin organization)

To further evaluate the influence of inherent mechanical gradients on cell behavior, we also examined actin organization in cells as a function of gel position. Consistent with

OSU-2 cell morphology observations, actin filaments in OSU-2 cells at the lowest observation plane were highly organized and resulted in the formation of mature stress fibers. In contrast, cells at higher observation planes, displayed actin fibers that were not organized, poorly developed and did not display stress fiber formation (Figure 49). This behavior was also maintained for cells plated on 2D Matrigel.

117

Figure 49. OSU-2 actin organization in 3D hydrogels at a higher observation (> ~500 µm) and lower observation (< ~50 µm) plane. Scale bar = 100 µm.

5.3.4 OSU-2 cell migration in 3D Matrigel

OSU-2 cells encapsulated in Matrigel were tracked in real time to gain insight into their migration patterns as well as to quantify migration speeds. There were striking differences in migration speeds and patterns of OSU-2 cells at the lowest observation plane, where edge effects were expected to dominate, compared to those near the gel surface. For example, in the case of 40% v/v Matrigel, cells at the lowest observation plane migrated at ~ 29.5 ± 11.3 µm/hr, ~4X faster than cells near the gel surface that migrated at ~ 7.6 ± 3.1 µm/hr. This trend was maintained at all concentrations of Matrigel investigated. OSU-2 cells at lower observation planes migrated by displaying highly bipolar cell bodies and long processes (see video 5.1 [http://youtu.be/ADLaZtKtf8U], video 5.3 [http://youtu.be/jksg1jnK50Y], video 5.5 [http://youtu.be/7iXqQgun5_I], video

5.7 [http://youtu.be/QM9RvHlXU_I]), whereas cells at higher gel positions migrated by displaying short processes and mostly rounded or ellipsoid cell bodies (see video 5.2 [

118 http://youtu.be/Q6YpG3H-rEQ], video 5.4 [http://youtu.be/juodjYK-Z7s], video 5.6

[http://youtu.be/7M-T1y_UwlM], video 5.8 [http://youtu.be/9UMWoelSmdM]).

Migration stills from a typical time lapse experiment for cells at lower and higher observation planes in the 40% v/v gel are shown in Figure 50.

Figure 50. OSU-2 cell migration in a representative 40% v/v Matrigel at (A) lower (< ~50 µm) and (B) higher (> ~500 µm) observation planes shown as stills from time lapse microscopy. Time stamp is reported in hours (h). Scale bar = 100 µm.

Migration speeds for each gel formulation were computed for the lowest and highest observation planes investigated and were compared with cells plated on the bottom of glass surface controls using ANOVA and the Student’s t-test. In all gel formulations, statistically significant differences (p < 0.001 in all cases) were detected for cell migration speeds at lower vs. higher observation planes (Figure 51). Migration speeds of cells at lower observation planes were not statistically significant from those of cells on glass substrates (video 5.9, available at http://youtu.be/A3QTxwOKDWc; OSU-2 cell speed = 28.2 ± 9.7 µm /hr). It is likely that cells at the “zero” observation plane are in contact with the glass surface; however, sequential data collected from cells throughout 119 the 0 – 200 µm interfacial region (e.g., Figure 41, 42, 44B, Stack S1, Stack S2) clearly demonstrates that cell area decreases with gel height, supporting the hypothesis that interfacial gradients are influencing cell behavior.

Figure 51. Quantification of migration speeds (average) of OSU-2 cells at the lowest (< ~50 µm) and highest observation planes (> ~500 µm) investigated. * indicates statistical significance.

5.4 Discussion

Mechanical gradients that exist in vivo (e.g., in the brain) have been shown to modulate cell migration, differentiation, proliferation, and cytoskeletal organization [273]. We hypothesized the existence of mechanical gradients at the interface of soft hydrogel

120 materials and rigid substrate supports that could influence cell behavior. To support our expectation of the presence of these gradients, FEM was used to simulate an indentation test on 100% v/v Matrigel supported on glass. Other computational models that examine similar interfaces focus on the cell and gel simultaneously [276, 277]. Despite the fact that our simplified FEM does not consider either a 2D [277] or 3D matrix [276] in contact with cells, our results corroborate those of more complex models. FEM shows that the stiffness exhibited by Matrigel changes as the Matrigel depth approaches the size of the cells. Specifically, the model shows an increase in stiffness for Matrigel heights ≤

50 μm from the hard substrate. This increased stiffness yields an increase in Matrigel stresses near the well bottom. In addition to FEM, active microrheology could be further employed to experimentally support simulation observations. In particular, this could be achieved by embedding micro beads in 3D hydrogels and tracking their response to external fields (e.g., using magnetic fields, or optical tweezers) at different gel heights.

This could then be translated to appropriate stress-strain relationships and mechanical properties.

Experimentally, GBM tumor cells responded to this increased stiffness by exhibiting a spread or bipolar morphology. Morphological changes exhibited an exponential response

(Figure 44), decaying with increasing distance from the rigid support. Cells also displayed an increased migration capacity, in stark contrast to cells distant from the interface. This outcome is in agreement with recent studies, which show that the stiffness of the gel and substrate are crucial factors affecting cell morphology during migration

[261, 264, 278]. This was also evident from individual cell area and aspect ratio analyses

121 in which cells at lower observation planes exhibited higher and statistically significant cell areas and aspect ratios compared to cells more distant from the interface. Thus, these data show that inherent gradients in 3D culture systems can dramatically influence the ability of cells to attach and spread, and demonstrate that tumor cells encapsulated in 3D hydrogels can “sense” the stiffness of an underlying rigid support (in this case, glass).

This phenomenon has been previously observed by Discher and colleagues in 2D gels for mesenchymal stem cells [176, 279] and others [280-282]. A similar result has also been obtained computationally by van Dommelen et al. [283], who showed that glass plates used to support brain tissue samples played a significant role on the force level in indentation. Our simulations and experimental findings complement these studies by examining tumor cell behaviors in a 3D setting. Our findings also suggest caution in interpretation of cellular outcomes in 3D hydrogels depending upon the location of these cells in hydrogels, as edge effects can significantly alter findings. Further, individual cells can influence the behavior of neighboring cells several microns distant [278], and hence the influence of edge effects could extend beyond the interfacial region. Our findings are consistent with a recent study that demonstrated that an underlying rigid support dramatically influences human mesenchymal stem fate when cultured on 2D collagen gels of different heights (i.e., 130 µm versus 1440 µm) [284]. Several experimental and computational studies have attempted to identify a “critical” height that defines how far cells can sense microenvironments in 2D [176, 277, 280-282]. Our computational findings suggest a possible threshold of ~50 µm for a 3D Matrigel model. This is comparable to values obtained in previous studies, which yield “critical” height values

122 that range from on the order of focal adhesions (i.e., a few microns) [176, 277, 281] to ~

2-3 times the cell length (e.g., ~ 60 microns) [280, 282].

In addition to changes in cell spreading and morphology, we also observed differences in cell migration at the lowest and the highest observation planes. Individual cancer cells can migrate in a mesenchymal or amoeboid fashion in 3D matrices [285]. In mesenchymal migration, cells attach to the matrix via formation of focal contacts that are dissolved during migration [285]. Cells migrating in amoeboid mode squeeze their cell body through matrix pores with minimal matrix contact or no attachment [285]. OSU-2 cells at lower observation planes migrated faster (~4X) and in a mesenchymal fashion with filopodia (finger like protrusions) at the leading cell edges (e.g., Video 5.1). In contrast, cells more distant from the rigid support showed continuous short process extension and retraction (e.g., Video 5.2). Some cells seem to migrate in an amoeboid fashion (e.g., Video 5.6); however, this was not consistently maintained for all cells.

Differences in migration modes were further supported by data on actin organization.

Cells at lower observation planes displayed highly organized stress fiber formation, which enables cells to generate traction forces for migration, in contrast to cells more distant from this interface. These results demonstrate that tumor cells may adopt different migration mechanisms in a single 3D hydrogel in response to these inherent interfacial mechanical gradients.

Interestingly, no specific trends in cell behaviors (i.e., cell spreading, migration, actin organization) were observed with increasing Matrigel concentration (e.g., Figure 51).

Thus, the influence of edge effects observed here was more significant than that of the

123 increased modulus generated by increasing Matrigel concentration. Across the concentration ranges investigated, Matrigel demonstrates an ~ 5 fold difference in stiffness [261], whereas interfacial gradients may access as much as a 10 fold increase in stiffness (e.g., Figure 39). Furthermore, cells cultured in single gels far from the substrate will experience a uniform mechanical environment, as opposed to gradients, such as those found near the interface with a support. Gradients can produce different cell responses than uniform mechanical environments [264-274]. Also, it should be noted that additional variables, such as matrix porosity, which would differ in gradient and single gel systems, may have influenced results. Thus, it is not surprising that the responses seen with increasing gel concentration differ from those produced by interfacial gradients.

There are several factors that may influence these results, which could be minimized by experimental modifications. The addition of fluorescent beads as markers of gel position or the use of computational algorithms to reduce gel movement would potentially permit tracking throughout the gel over longer time frames. Additionally, dynamic investigations of cell behaviors immediately after cell encapsulation should provide insight into the metabolic rates of tumor cells that further relate to differential cell spreading and migration behaviors. It should also be noted that a possible limitation of this approach is that different conditions in the bulk vs. at the gel-support interface (e.g., nutrient supply, ligand or crosslinker density) could lead to differences in properties that influence cell behaviors. The correlation between interfacial stiffness and pore size at different gel heights should be experimentally explored in detail. This will enable examination of

124 complex relationships of matrix parameters (i.e., pore size) on tumor cell migration in 3D microenvironments.

5.5 Conclusions

Here, we demonstrate that inherent interfacial mechanical gradients produced by edge effects between soft hydrogels and rigid substrate supports can modulate cell behavior.

To our knowledge, this study is the first to examine the role of inherent interfacial mechanical gradients on the behavior of tumor cells in a single 3D hydrogel. These findings are broadly applicable to virtually any hydrogel and adhesive rigid support combination, and could have import for hydrogel-based, 3D cell culture. Thus, inherent mechanical gradients can influence cell behavior in single, 3D hydrogels.

125

Chapter 6: Collagen-Hyaluronan Composite Hydrogels: 3D

Extracellular Matrix Mimics of the Glioblastoma Multiforme

Tumor Microenvironment5

Glioblastoma multiforme (GBM) tumors are one of the most deadly forms of human cancer with a median survival time of ~1 year. Their high infiltrative capacity makes them extremely difficult to treat, and even with aggressive multimodal clinical therapies, outcomes are dismal. To improve understanding of cell migration in these tumors, three dimensional (3D) multi-component composite hydrogels consisting of collagen and hyaluronic acid, or hyaluronan (HA), were developed. Collagen is a component of blood vessels known to be associated with GBM migration, whereas HA is one of the major components of the native brain extracellular matrix (ECM). We characterized hydrogel micro structural features and utilized these materials to investigate patient tumor-derived, single cell morphology, spreading, and migration in 3D culture. GBM morphology was influenced by collagen type with cells adopting a rounded morphology in collagen-IV versus a spindle-shaped morphology in collagen-I/III. GBM spreading and migration were inversely dependent on HA concentration; with higher concentrations promoting

5 This chapter with minor modifications has been submitted for publication. S.S. Rao, J. DeJesus, A. Sarkar, J. O. Winter (2012). Glioblastoma behaviors in 3D collagen-hyaluronan hydrogels. Submitted 126 little or no migration. Further, non-cancerous astrocytes primarily displayed rounded morphologies at lower concentrations of HA; in contrast to the spindle-shaped (spread) morphologies of GBMs. These results suggest that GBM behaviors are sensitive to ECM mimetic materials in 3D and that these composite hydrogels could be used to develop 3D brain mimetic models for studying migration processes.

6.1 Introduction

Glioblastoma multiforme (GBM), a primary tumor of the astrocytes and one of the most lethal forms of human cancer, affects ~22,500 individuals in the United States annually

[48, 55, 145, 146]. GBMs are characterized by their extremely high invasion potential

[51]. For example, tumors can redevelop in the opposing brain hemisphere following surgical resection of the afflicted hemisphere [49]. Current treatment methods (e.g., surgery, radiation, and chemotherapy) have been largely unsuccessful, mainly because of the highly infiltrative nature of these tumors [148]. Despite advances in these techniques, median survival time remains low (~12-15 months) [48, 55, 146, 151]. This, in part, is a consequence of our poor understanding of the molecular and mechanical pathogenesis of

GBMs. Thus, there is a need to develop newer methods and models to understand the complex behavior of GBM tumors.

Many existing models to investigate tumor cell migration (e.g., scratch assay, micro-liter migration assay) utilize two dimensional (2D) substrates (e.g., plastic, glass) that do not recapitulate the complex in vivo tumor microenvironment [175]. It is now widely accepted that cell behavior is drastically altered when exposed to three dimensional (3D) microenvironments [191, 192, 207] and that extracellular matrix (ECM) cues play a

127 significant role in tumor progression [286]. In recent years, there has been increasing interest in using hydrogels, 3D biomaterials commonly employed as tissue engineered scaffolds, to understand tumor cell biology. For example, seminal work by Bissell and coworkers using 3D Matrigel biomaterials to explore breast cancer has unraveled several tumor cell characteristics observed in vivo under in vitro conditions [287, 288]. For GBM studies, both naturally-derived (e.g., Matrigel [216-218], collagen [219, 220, 224]) and synthetic (e.g., poly(acrylamide) [214]) hydrogels have been utilized. Naturally-derived materials present a rich in vitro GBM invasion platform, but are limited in the tunability of their physicochemical properties. For example, ligand density, stiffness, and porosity cannot be varied beyond a certain range, which may prevent certain tumor cell characteristics from being captured. In addition, most of these assays do not employ hyaluronic acid/hyaluronan (HA), a significant component of the brain ECM [160, 289,

290]. Synthetic hydrogels can overcome many of these limitations, providing highly tunable properties and user control over several parameters, but they often lack the complexity of naturally-derived materials and therefore may not fully capture in vivo response. Additionally, regardless of the materials used, most of these studies have investigated behavior of well-established tumor cell lines isolated >25 ago, which is important because phenotypic and genotypic alterations have been reported after repeated culture in cell lines [291].

To increase the complexity of the 3D tumor microenvironment beyond that provided by single component natural gels [217, 219, 220, 292], we investigated GBM behaviors in collagen-HA, multi-component composite hydrogels. Similar constructs have been used

128 in neural tissue engineering [9, 293-295]. Collagen was chosen because it is found in the cancer brain microenvironment. Specifically, collagen types I, III, and IV have been observed in the glial limitans externa and vascular basement membranes, with types I and

III also found in the tumor ECM [150]. Additionally, clinical observations suggest that

GBMs migrate as single cells along these structures [49, 50, 159, 171], and previous animal studies have shown the formation of a thicker collagen ECM around blood vessels in gliomas compared to normal tissue [296]. HA, a high molecular weight, non-sulfated anionic, glycosoaminoglycan (GAG) [297], was chosen because in both normal and cancerous neuropil it is the primary ECM component [160] and is present in high levels in many gliomas when compared to normal tissues [152, 227, 298]. HA in its unmodified form has previously been used as a transwell insert coating in a glioma cell motility assay

[187, 189] and as an additive to Matrigel [186, 189] and fibrin [299] in the traditional invasion assay.

Here, we combined these two ECM components (i.e., collagen types I/III and IV and HA) to yield GAG-protein composite hydrogels, characterized their architecture, and examined 3D GBM response to altered HA composition, mimicking the increasing levels of HA typically observed in GBM tumors in vivo. Very few studies have examined GBM behavior in 3D [217, 219, 220, 223, 224, 227, 231], and even fewer have utilized collagen-HA composite hydrogels [231]. Further, this work is the first to examine single cell morphology, spreading, and migration of GBM cells in 3D collagen-HA composites, and in contrast to prior 3D hydrogel studies [217, 219, 220, 223, 224, 227, 231] is the

129 first to explore behaviors of patient-derived GBM cells rather than well-established culture lines.

6.2 Materials and methods

6.2.1 Cell culture

6.2.1.1 Patient tumor derived OSU-2 cell culture

Glioblastoma cells were directly procured from primary, patient brain tumors (OSU,

Neurosurgery) in accordance with OSU approved IRB protocol 2005C0075 (dated

11/08/08). Written consent was obtained from participants involved in the study. These

Ohio State University (OSU)-2 cells were sub-cultured for experimental use as described previously [190]. Briefly, patient-derived tumors were prepared from discarded “tissue” by washing thoroughly with cell culture media (DMEM/F12 (Invitrogen)) containing 200 unit penicillin (Invitrogen), 200 µg streptomycin (Invitrogen), and 0.5 µg/ml amphotericin B (Invitrogen). Following this, samples were digested by treatment with

200U/ml type 1A collagenase (Sigma) for ~4 h, triturated to eliminate cell aggregates, centrifuged at 250 g (~5 min), and resuspended in cell culture media (DMEM/F12

(Invitrogen)) containing 10% fetal bovine serum (Invitrogen), 100 units penicillin, 100

µg streptomycin, and 0.25 µg/ml amphotericin B. The well dispersed cell solution was then transferred into a petri-dish and incubated at 37 ºC in a 5% CO2 environment. Cells were fed 2-3 times per week and passaged on reaching confluency. Histopathology at the time of operation confirmed the type of tumor and grade. Further, to confirm astrocyte lineage indicative of GBM tumors, cells were stained for glial fibrillary acidic protein

(GFAP), an astrocyte marker (Figure 52 A, 52 B).

130

6.2.1.2 Normal (non-cancerous) astrocyte culture

Human astrocytes (Figure 52 C) were obtained from Invitrogen (Gibco Human

Astrocytes) and sub-cultured for one passage on Geltrex (Invitrogen) coated tissue culture plates (1:100 dilution in DMEM, ~200 µL/cm2). Cells were cultured and fed 2-3 times per week with complete astrocyte medium containing 88% DMEM, 1% N-2 supplement, 10% fetal bovine serum, and 1% penicillin-streptomycin (Invitrogen). For passaging, cells were washed with phosphate buffer saline (PBS), detached using Stem

Pro Accutase (Invitrogen), centrifuged at 200 g for 4 min, and then transferred to new

Geltrex coated plates or used for hydrogel experiments.

Figure 52. OSU-2 and non-cancerous human astrocytes cells in culture. (A) GFAP staining of OSU-2 cells in culture. Hoechst stain labels the nucleus blue, whereas rhodamine-GFAP labels the cytoskeleton red. (B) Phase contrast image of OSU-2 cells in culture. (C) Phase contrast image of non-cancerous astrocytes in culture. Scale bar = 100 µm.

6.2.2 3D cell encapsulation in collagen-HA composite hydrogels

Composite hydrogels were created using collagen (PureCol, pepsin solubilized bovine collagen composed of ~97% collagen type-I and ~3% type-III (Advanced BioMatrix

Inc.)) and thiolated hyaluronic acid (HA) (Glycosan Biosystems Inc.). Collagen and

131 thiolated HA both independently form hydrogels in situ at 37 ºC providing permissible conditions for cell encapsulation. Sterile collagen (I/III) solution (1.5 mg/ml, pH ~ 7.4) was prepared with DMEM/F12 (Invitrogen) in a cold environment. Thiolated HA was sterilized using UV illumination (peak power 11.2 mW/cm2) for ~30 min and placed in a

96 well plate. OSU-2 cells pre-labeled with Cell Tracker Green CMFDA (Invitrogen) at

~175,000 cells/ml in cell culture medium were then mixed with the diluted collagen solution and directly added to thiolated HA. Thus, cell-laden hydrogel constructs with a constant collagen-I/III concentration of 1 mg/ml and HA concentrations ranging from 0 mg/ml-20 mg/ml (0 wt/v% - 2 wt/v%) (N = 3, See Table 8 for all compositions) were created. Cell-laden composite hydrogels were incubated at 37 ºC, 5% CO2 for ~1 hour prior to the supplementation with additional OSU-2 cell culture media. In addition to

OSU-2 cells, human derived, normal (non-cancerous) astrocytes at a cell density of

~175,000 cells/ml in cell culture medium were also fluorescently labeled and encapsulated within these hydrogels for morphology observations.

OSU-2 cells pre-labeled with Cell Tracker Green CMFDA (Invitrogen) were also encapsulated in human collagen-IV (Col-IV) (BD Biosciences) and Col-IV-HA composite hydrogels (Table 8). In both cases, a base gel layer was formed first to prevent cell settling through the loose hydrogel. Approximately 30 µL of 0.45 mg/ml

(concentration as supplied by manufacturer) sterile Col–IV under neutral conditions was used independently or added to 0.27 mg pre-sterilized thiolated HA and solidified at 37

ºC, 5% CO2 in a 96 well plate for ~2 h to form Col-IV and Col-IV-HA base layers, respectively. Then, a cell laden solution was created using pre-labeled OSU-2 cells at a

132 density of ~350,000 cells/ml in Col-IV or Col-IV-HA solution. Approximately 30 µL of these solutions were added to the pre-gelled layer to yield cell-laden Col-IV (N = 2) and

Col-IV-HA (N = 2) hydrogels with a final concentration of ~175,000 cells/ml, respectively. For Col-IV-HA hydrogels, the weight ratio of Col-IV to HA was held constant at ~1:20 (i.e., HA = 0.9 wt %), similar to Col-2HA hydrogels. Collagen IV is one of the primary components of blood vessels; however forms very weak hydrogels.

Therefore, multiple compositions of Col-IV-HA were not investigated. Cell-laden hydrogels were incubated for 1 h at 37 ºC, 5% CO2 to permit gelation of the upper layer before supplementation with 60 µl additional cell culture media. A schematic of encapsulation process is shown in Figure 53.

Collagen-I/III based Collagen-IV based Col- Col- Col- Col- Col- Col-IV Col-IV- Sample Col 0.1HA 0.2HA 0.5HA 1HA 2HA HA HA (mg/ml) 0 1 2 5 10 20 0 91/52 HA (wt %) 0 0.1 0.2 0.5 1 2 0 0.91/0.52 Collagen (mg/ml) 1 1 1 1 1 1 0.451/0.32 0.451/0.32 Table 8. Composition of composite hydrogels. Superscript 1 indicates composition used for cell studies and superscript 2 indicates composition used for confocal reflectance microscopy for Collagen-IV based compositions. Collagen-IV used at manufacturer supplied concentration.

133

Figure 53. Schematic of encapsulation procedure to create cell laden 3D hydrogels.

6.2.3 Characterization of composite hydrogels

6.2.3.1 Rheological characterization

Hydrogel mechanical properties were characterized using unconfined compression testing

(RSAIII, TA Instruments). Acellular hydrogels (hydrogels without cells comprised of

Col-I/III and Col-I/III-HA), N ≥ 3, were prepared as described in section 6.2.2. After gelation, hydrogels were subjected to compression testing at a strain rate of 0.5 mm/min for ~20 s and then held for another 20 s in a multiple extension mode test. Stress-strain

134 curves generated from the compression tests were used to obtain elastic moduli of hydrogels. All measurements were performed at room temperature (~25 ºC).

6.2.3.2 Confocal reflectance microscopy

Acellular composite hydrogels were prepared as described in section 6.2.2. (i.e., Col,

Col-0.5HA and Col-2HA, N = 3). Gels were formed in a cover well perfusion chamber gasket (8 chambers, 9 mm diameter, 2 mm depth, Invitrogen) glued to a glass slide. Gels were overlaid with cell culture media for imaging purposes. For Col-IV gels (N = 3), 80

µL at a concentration of ~ 0.3 mg/ml (concentration as supplied by manufacturer) was used and for Col-IV-HA gels (N = 2), ~0.4 mg thiolated HA was added to keep the weight ratio approximately equivalent to that used for cell experiments (i.e., 1:20 collagen: HA). Images were acquired at random gel positions using a laser scanning confocal microscope (Fluoview MPE) in reflected mode with a 25X objective, 3X zoom and NA = 1.05. The excitation laser source was Alexa Fluor 488 nm and the reflected light was detected using a photo-multiplier tube detector (PMT).

6.2.3.3 Scanning electron microscopy

Acellular gels Col, Col-0.5HA, Col-2HA and pure HA gels (N = 3) were prepared as described in section 6.2.2. Gels were incubated with deionized water overnight and then flash frozen in liquid nitrogen to preserve the morphology and structure as described elsewhere [294, 300]. Samples were then lyophilized overnight and cut using a razor blade to observe the interior gel surface. Samples were mounted on aluminum stubs (Ted

Pella Inc.), coated with gold for 30 s (Model 3 Sputter Coater 91000, Pelco, Reading,

135

CA), and imaged using a scanning electron microscope (SEM, FEI XL-30 Sirion SEM,

FEI Company, Hillsboro, OR) at an accelerating voltage of 2 kV.

6.2.4 OSU-2 morphology analysis and cell spreading in collagen-HA composite hydrogels

OSU-2 cell laden hydrogels were prepared as described in section 6.2.2 and imaged after

~24 h. Three still images per hydrogel (N = 3 hydrogels for every formulation) were randomly collected using a confocal microscope (LSM 510; Zeiss, Minneapolis, MN) and subjected to image analysis. OSU-2 morphology was quantitatively analyzed by examining the discrete area and circularity of individual cells using NIH Image J software (available at http://rsbweb.nih.gov/ij/). Cell area and circularity were determined and reported as average ± SD (for three hydrogel replicates). In addition, the percent rounded cells in each hydrogel formulation was also examined (a cell was considered to be round if circularity ≥ 0.95).

6.2.5 Real time OSU-2 cell tracking in 3D composite hydrogels

OSU-2 cell laden hydrogels were prepared as described in section 6.2.2 and cell migration experiments were performed using confocal microscopy (LSM 510; Zeiss,

Minneapolis, MN). After an initial 12 h incubation, a series of images were collected every 10 min for a total of 8 h using a confocal microscope equipped with a motorized stage and an incubation chamber. Some samples experienced considerable movement

(i.e., swelling). This was corrected by applying the StackReg Plugin (available at http://bigwww.epfl.ch/thevenaz/stackreg/). Images were converted to movies using NIH

Image J and migration speeds were calculated using MTrack J by dividing the entire

136 length traveled (µm) by the total time (h) of cell tracking. Also, only spread cells were motile and hence cell speeds in each gel formulation was calculated only for spread or

“spindle” shaped cells. Migration speeds were computed from individual cells (n ≥ 40 cells per condition) for each gel formulation (N = 3 hydrogels) and are reported as box and whisker plots, showing mean, median, and outliers for each condition.

6.2.6 Statistical analysis

Statistical Analysis was performed using the JMP statistical software package. All samples were analyzed using ANOVA and observations in collagen-HA composite hydrogel samples were compared to control collagen samples using Dunnett’s Method

(comparison to a control).

6.3 Results and discussion

6.3.1 Composite hydrogel modulus

The elastic moduli of composite hydrogels were obtained from stress-strain curves generated by unconfined compression testing. The modulus of a composite hydrogel has been shown to strongly influence cell migration [261] and can alter with composition.

Pure collagen (Col-I/III) hydrogels had an elastic modulus of ~ 300.48 ± 39.5 Pa, and the addition of HA increased this modulus to > 1000 Pa with Col-1HA and Col-2HA samples having values statistically different from those of pure collagen (p < 0.001) (Figure 54).

Comparing these values with physiologically reported values for brain tissue (~200-1000

Pa [176], non-cancerous, values for cancerous brain have not been conclusively determined, however evidence suggests that tumor tissue is mechanically different from normal brain tissue [164]) (Figure 54), the last two samples exceed this modulus range,

137 whereas the other samples are within values reported for brain tissue. Thus, by changing the composition of HA in composites, mechanical stiffness could be controllably altered from 300-2065 Pa, with a maximum increase in modulus of 7X (e.g., Col-2HA) over pure collagen controls, comparable to the physiological range reported for brain tissue [176].

This demonstrates that the collagen-HA composite hydrogel system is mechanically tunable, and its stiffness can be modulated to span the physiological range for brain tissue while simultaneously permitting incorporation of GAGs (i.e., HA).

Figure 54. Mechanical characterization of Collagen (I/III) and Collagen (I/III) composite hydrogels. (A) Elastic modulus values reported for various tissues. Figure adapted from [176]. (B) Elastic modulus of Collagen (I/III)-HA composite hydrogels (N ≥ 3). * indicates statistically significant from collagen controls (p < 0.0001, as reported from ANOVA). 138

6.3.2 Composite hydrogel micro-architecture

Composite hydrogel micro-architectures were assessed using confocal reflectance microscopy (CRM) and scanning electron microscopy (SEM). CRM images showed that pure collagen (I/III) hydrogels demonstrated a strong fibrillar character, which was also present in composite hydrogels formed with HA, but declined with increasing HA composition (Figure 55).

Figure 55. Confocal Reflectance Microscopy (CRM) imaging of Collagen (I/III) and Collagen (I/III)-HA composite hydrogels.

In contrast, Col-IV gels, as well as composites of Col-IV with HA, had less fibrillar character (Figure 56). The structure of Col I/III hydrogels and composites was further confirmed using SEM (Figure 57). The fibrillar structure of pure collagen hydrogels is evidenced. In comparison, pure HA hydrogels had a more flat, smooth, sheet-like, dense architecture (Figure 57) as observed previously [227, 301], and this structure was more evident as HA composition increased in collagen-HA composite hydrogels. Thus, composite hydrogels demonstrated characteristics of both material components, similar to

139 collagen-HA interpenetrating networks (IPNs) described previously [294, 301], and presented a combination of unique architectures representative of blood vessels (i.e., fibrillar collagen) and brain ECM (i.e., flat sheet-like HA). Thus, collagen-HA hydrogels exhibit many brain mimetic features, including tunable control of collagen and HA composition, distinct microarchitectures, and physiologically relevant mechanical properties, which make them suitable 3D biomaterial scaffolds for the investigation of neural cells.

Figure 56. Confocal Reflectance Microscopy (CRM) imaging of Collagen-IV and Collagen IV- HA composite hydrogels.

Figure 57. Scanning Electron Microscopy (SEM) imaging of Collagen (I/III), Collagen (I/III)-HA composite and pure HA hydrogels. Scale bar = 50 µm.

140

6.3.3 OSU-2 and normal astrocyte behaviors in 3D composite hydrogels

The morphology of patient derived OSU-2 cells encapsulated in composite hydrogels was characterized, including cell spreading, circularity, and percentage of rounded cells

(Figure 58, 59, 60, 61). OSU-2 cells adopted a spread or “spindle” morphology in collagen hydrogels (control) (Figure 58).

Figure 58. OSU-2 morphologies in Collagen (I/III)-HA composite hydrogels as observed via confocal microscopy. Insets with each image show representative cell morphologies. Scale bar = 100 µm.

This morphology was maintained at lower HA concentrations (≤ 0.5 wt% HA) and is representative of the morphology observed in vivo [173, 201, 206, 302]. As the concentration of HA increased, OSU-2 cells transitioned to a rounded morphology. For example, cells in 2 wt% HA were mostly rounded (~92.77 ± 6.73%) compared to those in 141

1 wt% HA (~54 ± 9.6 %). This was further confirmed by quantification of cell area and circularity (Figures 58, 59, 60, 61) with higher HA concentration reducing cell spreading and increasing cell circularity, indicating minimal interaction of OSU-2 cells with higher

HA wt% composite hydrogels.

Figure 59. OSU-2 cell areas in 3D composite hydrogels. * indicates pairs that are statistically significant compared to Col (Control) (p < 0.0001, as reported from ANOVA)

142

Figure 60. OSU-2 cell circularity in 3D composite hydrogels. * indicates pairs that are statistically significant compared to Col (Control) (p < 0.0001, as reported from ANOVA)

Figure 61. Percentage of rounded cells in composite hydrogels. * indicates pairs that are statistically significant compared to Col (Control) (p < 0.0001, as reported from ANOVA). 143

Cell morphology in Col IV and Col IV-HA hydrogels was also examined; however, gel integrity was not sufficient to prevent cell settling over the time period investigated (e.g.,

24 h). Thus, cells contacted the bottom of the dish and displayed typical 2D culture behaviors. However, at shorter time points (i.e., 6 h), OSU-2 cells in Col-IV and Col-IV-

HA maintained a rounded shape (Figure 62), suggesting that cells preferred Col-I/III environments versus Col-IV environments.

Figure 62. OSU-2 cell morphology in Col-IV and Col-IV-HA composite hydrogels.

The altered response may have been a consequence of the unique architecture of these hydrogels. Collagen-I monomers self-assemble at physiological temperature and pH to form hydrogels, first forming aggregates and then filaments that eventually form fibrils by lateral crosslinking. Three dimensional hydrogels are created when these fibrils entangle in a non-covalent fashion [223, 303]. In contrast, collagen-IV is less fibrillar (as confirmed using confocal reflectance microscopy, Figure 56) and is weaker than 144 collagen-I fibrillar networks [304]. The inherent, weak nature of collagen-IV hydrogels might hinder cell attachment and spreading, which in turn could result in extremely weak traction forces. This is further supported by results of 2D culture on Col-I/III and Col-IV coated surfaces. Whereas, initial cell adhesion does not differ, 2D spreading is significantly higher for Col-I/III versus Col-IV (p < 0.0001) demonstrating that the differences in cell spreading do not result from differences in adhesion (Figure 63).

Figure 63. OSU-2 cell adhesion and spreading on Col (I/III) and Col-IV coated surfaces (~100 µg/ml). (A) OSU-2 cell adhesion was quantified by prelabelling cells with Cell Tracker Green and seeding them on coated surfaces and reading the fluorescent intensity of adhered cells (after 1 h wash) using a fluorescent plate reader. (B) OSU-2 cell spreading on Col (I/III) and Col-IV coated surfaces (~ 100 µg/ml) was quantified using Image J. * indicates statistical significance, tissue culture polystyrene (TCPS) served as control in both cases.

The behavior of normal human astrocytes was also examined in these materials. In almost all compositions, normal astrocytes displayed rounded morphologies occasionally with

145 short processes (Figure 64). This behavior is in contrast with that of tumor cells, which showed a spindle-shaped morphology at lower HA concentration.

Figure 64. Non-cancerous human astrocyte morphology in Collagen (I/III)-HA composite hydrogels. Arrows indicate small processes extending from the astrocyte cell body.

This could result from altered integrin expression in normal versus cancer cells. For example, β1 integrins, which mediate attachment to collagen I, are rare in normal astrocytes, but commonly found in glioblastomas [305] These contrasting behaviors including normal astrocyte migration and their adhesion-dependent 3D biology will be more fully investigated in future studies.

6.3.4 OSU-2 migration in 3D composite hydrogels

The migration capacity of OSU-2 cells encapsulated in 3D composite hydrogels was characterized using time-lapse confocal imaging. Since cells did not spread in Col-IV or

146

Col-IV-HA gels, migration in these hydrogels was not investigated. Cells in pure collagen (I/III) exhibited the fastest migration speeds at 9.4 ± 3.4 µm/h (Video 6.1, available at http://youtu.be/UAVtUcyE4dE). OSU-2 cells in composite hydrogels with lower concentrations of HA migrated in a similar fashion to pure collagen (I/III) controls.

Individual tumor cells are known to migrate via mesenchymal or amoeboid migration modes in 3D matrices [285]. In mesenchymal mode, cells attach to the ECM via formation of focal contacts that are eventually dissolved upon migration to an adjacent site [285]; whereas in amoeboid mode, cells squeeze through the matrix pores with minimal attachment to the matrix [285]. At low HA concentration, migration appeared mesenchymal in nature for both composites and collagen I/III controls. Further, OSU-2 cells in Col-0.1HA hydrogels migrated at 7.7 ± 3.9 µm/h, speeds that were statistically insignificant compared to pure collagen (Video 6.2, available at http://youtu.be/2ZPgKX2kHlc). However, as HA concentration increased, cell migration speeds showed a decreasing trend and eventually cells failed to migrate (Still images from a typical time lapse experiment for cells in Col-0.2HA hydrogels are shown in

Figure 65, Video 6.1 [http://youtu.be/UAVtUcyE4dE], Video 6.2

[http://youtu.be/2ZPgKX2kHlc], Video 6.3 [http://youtu.be/rVWylpClsIE], Video 6.4

[http://youtu.be/BTWoZ2fXYiQ], Video 6.5.1 [http://youtu.be/vRnyXxFike4], Video

6.5.2 [http://youtu.be/dq_mnTo0NEY], Video 6.6 [http://youtu.be/_cZaNyovOSQ] and

Figure 66). This behavior can be seen clearly in video 6.5.2 that shows some rounded cells trying to interact with the matrix, but failing to migrate. These results obtained from our 3D single patient cell imaging are consistent with a recent report that investigated

147 cell-line derived GBM spheroid invasion in HA gels modified with the RGD peptide

[227].

Figure 65. OSU-2 cell migration in an example Collagen (I/III)-HA composite hydrogel (Col- 0.2HA) as shown through stills from time lapse microscopy movies. Time stamp reported in hours (h). Scale bar = 100 µm.

148

Figure 66. Quantification of OSU-2 single cell migration speeds in Collagen (I/III)-HA composite hydrogels. N ≥ 40 individual cells analysed for each condition. * (in blue) indicate statistical significance when compared to Col (Control) (p < 0.0001, as reported from ANOVA). Representative cell morphologies are presented as insets. Red lines within the box indicate mean and black lines indicate median values.

Both collagen [150] and HA [160] are important components of the tumor microenvironment, with increased HA expression evidenced over levels in normal tissues

[152, 227, 298]. OSU-2 cells cultured in these 3D materials exhibited greater spreading and migration at lower HA wt%, demonstrating patient derived glioma sensitivity to HA concentration. As the concentration of HA was increased, OSU-2 cells transitioned to a rounded morphology and lost the ability to migrate through the gel structure. These

149 findings further corroborate the view that tumor cell behaviors are regulated by the structural and mechanical properties of 3D ECMs [227, 231]. Consistent with our observations, David et al., showed collagen type I and III to be strong stimulators of invasion [225]. However, these studies investigated 2D invasion using static methods, rather than the dynamic, single cell tracking methods reported here. In addition, cell morphology in these constructs was primarily rounded, whereas our constructs display both spindle-shaped and rounded morphologies, consistent with the morphologies observed in migratory versus non migratory gliomas in vivo [173, 201, 206, 302]. These results are also consistent with spheroid migration observations in other HA-based systems, [227] but contradict reports using HA alone [189] or in a composite with

Matrigel [187, 189] or collagen [231]. This may result from the difference in HA used; here, cross-linked thiolated-HA was used versus the free polymer or may result because of the tumor cell type or source. Nonetheless, these results demonstrate that material presentation is crucial in dictating tumor cell spreading and migration, further enhancing our understanding of the role of specific physiological ECM components (i.e., HA) on tumor progression in 3D contexts.

The observed migration response could be caused by a number of factors: (1) cell response to increasing mechanical stiffness, (2) reduced porosity and steric barriers resulting from increased HA density (as observed using SEM) [It is recognized that (1) mechanics and (2) matrix pore size are inextricably linked.], (3) inability of tumor cells to secrete HA degrading enzymes, or (4) repellent chemical interactions between HA and cell surface adhesion proteins. Detailed interrogation of HA receptors (i.e., CD44 and

150

RHAMM (receptor for hyaluronan-mediated motility)) [152] with studies of intracellular signaling cascades as well as evaluation of HA degrading enzymes in physiologically relevant contexts should provide additional insight into this mechanism.

6.4 Conclusions

The potential of collagen-HA hydrogels as 3D biomimetic systems with tunable mechanical and chemical properties to explore the role of microenvironment on the migration of patient derived brain tumor cells was examined. To our knowledge, this is one of the first studies to examine the morphology and migration behaviors of human, patient-derived tumor cells in a 3D, protein-GAG composite hydrogel in real time. Tumor cells adopted in vivo-like spindle-shaped morphologies at lower HA concentrations, in contrast to normal human cells that maintained rounded morphologies at all concentrations investigated. GBM migration was an inverse function of HA concentration, with HA impeding and eventually stopping cell movement. Three dimensional materials that combine relevant ECM molecules, such as those described here, could greatly enhance our understanding of GBM migration, which is crucial to the development of improved therapeutic options. Also, these composite hydrogels offer great potential to investigate migration capacity of other cancers, as HA and collagen are widely found in the extracellular environment of many tissues, demonstrating the broad applicability of these biomaterials. Further, in contrast to other collagen-HA interpenetrating network hydrogels reported previously [294, 295, 301], the HA employed was chemically cross-linked, adding to the existing library of multi-component

151

3D hydrogels that can be employed for soft tissue engineering and regenerative medicine applications.

152

Chapter 7: White Matter Tract Topography-Mimetic

Biomaterial Platform for Examining Glioblastoma Multiforme

Tumor Cell Behaviors In Vitro6

Glioblastoma multiforme (GBM), one of the deadliest forms of human cancer is characterized by its high infiltration capacity, partially regulated by the neural extracellular matrix (ECM). A major limitation in developing effective treatments is the surprising lack of tunable in vitro models that mimic features of GBM migration highways. Ideally, these models should allow independent control of mechanics and chemistry to allow the role of these individual components in tumor behaviors to be examined. To address this need, we developed aligned nanofiber biomaterials via core- shell electrospinning that permit systematic study of mechanics and chemistry. These models consist of aligned electrospun fibers that mimic the topography of the white matter tract, a major GBM migration ‘highway’. To independently investigate the influence of chemistry and mechanics on GBM behaviors, nanofiber mechanics were modulated by using different polymers (i.e., gelatin, polyethersulfone, polydimethylsiloxane) in the ‘core’ while employing a common poly (ε-caprolactone)

(PCL) ‘shell’. These materials revealed GBM sensitivity to nanofiber mechanics, with

6 This chapter with minor modification is under preparation for publication: S. S. Rao, T. Nelson, R. Xue, J. DeJesus, M. S. Viapiano, J.J. Lannutti, A. Sarkar, J. O. Winter (2012). In Preparation. 153 single cell morphology (Feret diameter), migration speed, focal adhesion kinase (FAK) and myosin light chain 2 (MLC2) expression all showing a strong dependence on nanofiber modulus. Similarly, modulating nanofiber chemistry using potentially biologically relevant chemistries (i.e., hyaluronic acid (HA), collagen, and Matrigel) in the ‘shell’ material with a common PCL ‘core’ nanofibers revealed GBM sensitivity to

HA; specifically, in which a negative effect on migration was observed. This biomimetic system, which mimics the topographical features of white matter tracts, should allow further examination of the complex interplay of mechanics, chemistry, and topography in regulating brain tumor behaviors.

7.1 Introduction

Glioblastoma multiforme (GBM) accounts for nearly 50% of reported malignant brain tumors [55]. GBM is a primary tumor of astrocytes that, despite decades of research, remains resistant to chemotherapy [306]. Median survival for a patient diagnosed with

GBM remains dismal (i.e., ~1 year) [55], with tumor recurrence and progression being inevitable in almost all cases. Recurrent tumors can appear either at the site of the original tumor and also at noncontiguous or even contralateral sites [307]. Clinical observations suggest that these tumors migrate as single cells along the blood vessel periphery, glial limitans externa, and most frequently through white matter tracts [48, 49,

156, 171]. For reasons not yet fully understood, migration occurs despite the fact that white matter is an inhibitory substrate for neurite outgrowth and astrocyte migration

[308]. Thus, it is clear that additional strategies and technologies are needed to understand the complex and unexpected migration behaviors of GBM tumor cells.

154

Traditionally, cancer cell migration has been assessed using a number of two dimensional

(2D) assays, such as the micro liter migration assay [175, 178] or the wound healing assay [175] (reviewed in detail in Chapter 3). However, these assays employ tissue culture dish substrates that are far from the aligned, fibrous topography characteristic of white matter tracts. Further, these 2D culture models can produce substantially different results than three dimensional (3D) cultures [191] that more closely mimic in vivo conditions [184]. Alternatively, chamber assays, which examine migration through a thin,

3D substrate on a supporting membrane, have been employed [175]. However, the presence of the membrane substantially diminishes the benefits of dimensionality. For example, we have shown in 3D culture that cancer cells within ~ 50 µm of a rigid support experience mechanical edge effects that alter their adhesion and migration response

[190]. The most common 3D, non-chambered migration assays are based on hydrogels, networks of chemically or physically crosslinked, natural or synthetic polymers. The most popular natural material for evaluating tumor cell migration is Matrigel [216-218,

309-311]. However, Matrigel is a mouse tumor extract that bears little resemblance to the composition of brain, which is substantially different from normal tissues [171]. Specific to GBM migration, several brain mimetic hydrogels, including hyaluronic acid, have been employed by us [228] and others [219, 223-227, 231]. However, although hydrogels are fibrous materials at the micro/nano scale, these fibers are not usually oriented

(Chapters 5 and 6). Further, decoupling the physical (e.g., porosity, mechanics) and chemical properties of 3D hydrogels is challenging [312] thereby limiting the ability to

155 separate the effects of these variables. Therefore, hydrogels alone do not adequately mimic the physical and chemical features of white matter tracts.

There are few in vitro models that attempt to mimic white matter specifically. The most common approaches use oligodendrocytes in primary culture [313] or organ cultures such as brain slices [314] or whole regions of the brain [315, 316]. Whereas these are valuable in vitro mimics, these models exhibit several limitations. Simple oligodendrocyte primary culture does not recapitulate the formation of the fiber tracts that characterize white matter. Many of these limitations can be addressed through the use of organotypic cultures; however, animal-to-animal variation and the difficulty of organ culture can make these models challenging to use. In addition, models based solely on organ/primary cultures generally do not permit investigator control of selected parameters (e.g., chemistry, mechanics, or topography) necessary for systematic study.

Polymeric electrospun nanofibers (ENFs) are alternative neural tissue engineering materials [317-321] that have been used as guides for neural repair and regeneration [317,

322-324] and substrates for Schwann cell maturation [325] and neural stem cell differentiation [326]. Aligned ENFs are particularly interesting as neural guides because of their topographical similarity to white matter [327]. Aligned ENFs can induce specific neural behaviors including neuron elongation and proliferation correlated to the direction of fiber alignment [323, 325, 328-330]. Additionally, aligned ENFs (i.e., poly (ε- caprolactone), PCL) reproduce the morphological and molecular signatures of glioma migration ex vivo [173, 174]. However, to the best of our knowledge, these tunable materials that mimic highly aligned white matter tract-topography have not been 156 employed previously to examine the role of microenvironment specifically, mechanics and chemistry, on GBM behaviors.

In this study, aligned ENFs fabricated using electrospinning were used to mimic the topography of the native in vivo environment, as well as some features of white matter tracts. Specifically, coaxial, core-shell ENFs with tunable mechanics and chemistry were fabricated and characterized to further examine biomimetic material properties that can influence tumor cell behaviors. To examine the influence of mechanics, various core materials were used (i.e., gelatin, polydimethylsiloxane (PDMS), and polyethersulfone

(PES)) with an identical shell (i.e., poly (ε-caprolactone) (PCL)). To study the role of brain tissue chemical composition, the glycosoaminoglycan, hyaluronic acid or hyaluronan (HA), also found in white matter [150] was spun as a shell on PCL core nanofibers. Finally, although not specific to white matter, collagen and Matrigel were also spun as shells on PCL core fibers for comparison because of their wide applicability in 3D cell culture models [303, 331]) and examining their topographical effect. These core-shell ENFs displayed identical micro-architectural features allowing independent assessment of the role of nanofiber mechanics and chemistry on tumor cell behaviors.

Using this ENF system, the role of mechanics and biologically relevant chemistries on

GBM cell adhesion, morphology, feret diameters, migration speed, focal adhesion kinase

(FAK) and myosin light chain 2 (MLC2) expression were examined. Additionally, for a specific chemistry (i.e., HA), matrix metalloproteinase (MMP) gene expression levels for

MMP-2, MMP-9, and MMP-13 genes was also examined because of their involvement in

GBM migration processes in vivo [332-336].

157

7.2 Materials and methods

7.2.1 Preparation of pure PCL and aligned core-shell nanofibers

Aligned core-shell nanofibers were prepared by coaxial electrospinning. For preparation of gelatin core-PCL shell samples, 5-wt% PCL (Mn 70,000-90,000, Sigma-Aldrich, St.

Louis, MO) and 6.7-wt% porcine gelatin type A (300 Bloom, Sigma-Aldrich, St. Louis,

MO) solutions were prepared separately in 1,1,1,3,3,3-hexafluoro-2-propanol (HFP) (>

99% purity; Oakwood Products, Inc., Columbia, SC) by stirring at room temperature (~

25 °C) for 24 hours. PCL and gelatin solutions were then poured individually into separate 20 cc syringes. One syringe containing the core solution (e.g., 6.7 wt% gelatin) was fitted with a stainless steel 22 gauge blunt tipped needle feed through a hollow, stainless steel T-joint core-shell nozzle (Small Parts Inc., Miramar, FL), whereas the syringe containing the shell solution (i.e., 5 wt% PCL) was connected to nozzle. The syringes were then placed into individual syringe pumps, set to a flow rate of 2 mL/hr and 4 mL/hr for the gelatin and PCL solutions respectively, and electrospun using a DC high voltage power supply (Glassman High Voltage, Inc., High Bridge, NJ) at positive 20 kV and a 20 cm needle-to-collector distance [337], for ~ 45 min at an average relative humidity of ~ 30%. Using these conditions, electrospun fiber was deposited on a mandrel coated with tissue culture polystyrene (TCPS) substrates rotating with a linear velocity of

15 m/s to produce aligned core-shell gelatin-PCL scaffolds. Aligned PCL scaffolds were produced using a single syringe fitted with a 20 gauge blunt tipped needle, at a flow rate of 5 ml/hr without the T-joint core-shell nozzle and the same spinning conditions stated above for ~25 min. For preparation of polyethersulfone (PES)-PCL core shell nanofibers,

158

5 wt% PES was dissolved in HFP at 55 °C for the core and a 5 wt % PCL solution for the shell. After cooling to room temperature, the polymer solution was placed in 60 cc syringes and a 20-guage blunt tip needle and coaxially electrospun using a high DC power supply set to 25 kV and a 20 cm tip-to substrate distance and a flow rate of 4mL/h for both core and the shell for ~ 20 min. For preparation of poly (dimethylsiloxane)

(PDMS)-PCL core shell nanofibers, 1 g PDMS base (Dow Corning Corporation,

Midland, MI) was dissolved in a mixture of dichloromethane (DCM) and toluene (weight ratio 3:2) together with 0.1 g curing agent. Following this, the mixture was electrospun as the core with 5 wt% PCL as the shell using a DC power supply set to 25kV, 20 cm tip-to substrate distance and a flow rate of 1 mL/h for the core and 4mL/h for the shell for ~ 30 min. Then, the samples were cured for ~ 24 hours at room temperature.

Nanofibers with identical cores but altered biomimetic surface chemistries as shells were prepared likewise. PCL (5 % w/w), 10 % w/v acid soluble collagen type I from bovine hide (Kensey Nash; Exton), and 1 % w/v Matrigel (BD Matrigel, BD Biosciences) solutions were prepared by stirring with HFP at room temperature (25 °C). A 1.1 % w/v hyaluronic acid (HA) (from Streptococcus Equi, Sigma Aldrich) solution was prepared by stirring with a 1:1 ratio of deionized water (DI water) and N, N-dimethylformide

(DMF) (>99% purity, Oakwood Products, Inc., Columbia, SC) at room temperature.

After complete dissolution of the polymer in the solvent, the dissolved PCL, collagen,

Matrigel and HA solutions were poured individually into separate 20 cc syringes. Flow rates were set to 4 ml/hr for the core and 2 ml/hr for the shell solution for PCL-collagen,

PCL-HA, and PCL-Matrigel, respectively. Electrospun aligned core-shell samples were

159 deposited on a mandrel coated with TCPS substrates rotated with an average linear velocity of 15 m/s, using a DC high voltage power supply, a negative 5 kV was applied to the mandrel and a positive 20 kV was applied to the coaxial nozzle while maintaining the same 20 cm needle-to-collector distance. Core-shell PCL-collagen samples were spun for

30 min, while PCL-HA and PCL-Matrigel samples were spun for 35 min. The as-spun

ENF mats were placed in a vacuum oven (< 30 mm Hg) at 25 °C for 24 hours to ensure removal of residual solvents [338]. The electrospinning conditions for each fiber type are summarized in Table 9.

Scaffold Flow rate Voltage Needle-to- Electrospinning (mL/h) (kV) collector Time (min) distance (cm) PCL 5 20 20 25 Core Shell Gelatin-PCL 2 4 20 20 45 PES-PCL 4 4 25 20 20 PDMS-PCL 1 4 25 20 30 PCL-Collagen 4 2 20 20 30 PCL-HA 4 2 20 20 35 PCL-Matrigel 4 2 20 20 35 Table 9. Summary of electrospinning conditions used for nanofiber samples.

As spun-PCL and core-shell ENFs were air plasma treated by placing the samples inside a Harrick plasma cleaner chamber (Harrick Plasma, Ithaca, NY, USA) under vacuum at

1000 mTorr and exposing the samples to a plasma radio frequency of 8-12 MHz for 2.5 minutes. After 2.5 minutes, samples were removed from the chamber and kept in a sealed container until use. Following this, the mats were then cut into ~ 16 mm diameter cylinders using a metal punch (Arch Punch; C.S. Osborne & Co, Harrison, N.J.) and used 160 for cell culture experiments. For analysis of signaling pathways via western blotting, mats were cut to fit snugly in a 60 mm petri dish (Fisher Scientific).

7.2.2 Morphological, surface and mechanical characterization of aligned PCL and core-shell nanofibers

7.2.2.1 Scanning electron microscopy (SEM)

For examination of fiber micro-architecture, as spun nanofiber discs (N=3) were placed on aluminum stubs using a carbon tape (Ted Pella, Inc.), sputter coated with gold for 30 s

(Model 3 Sputter Coater 91000, Pelco, Reading, CA) and imaged using a scanning electron microscope (Quanta 200 SEM or XL30F ESEM, FEI Company, Hillsboro, OR).

7.2.2.2 Fiber diameter, fiber density and fiber alignment

Individual fiber diameters and fiber density were measured using Image J software

(available at http://rsbweb.nih.gov/ij/) from the SEM images. For measuring fiber diameter, a line tool was used to measure the edge-to-edge distance perpendicular to individual nanofibers (n ≥ 75 fibers). For measuring fiber density (also known as line density), a line of known length was drawn perpendicular to the alignment of fibers and the number of fibers crossing the line was manually counted as described previously

[339]. For examining fiber alignment, a Fast Fourier Transform (FFT) was used [339-

341], which indicates the extent of fiber alignment. A radial summation of pixel intensities on the FFT output image was performed using the oval profile plug in (Image

J) and plotted as normalized intensity or gray value versus degree for each nanofiber examined. The degree with highest intensity was set to 90º/270º (as the plot is symmetric) to compare different nanofiber samples.

161

7.2.2.3 Mechanical properties of nanofibers using tensile testing

The mechanical properties of the aligned electrospun core-shell gelatin-PCL, PES-PCL,

PDMS-PCL and PCL nanofibers were determined using a uniaxial bench-top testing machine (TEST RESOURCES- Type R, TestResources Inc, Shakopee, MN). Samples were cut to a gage-length of 20 mm and gage-width of 2.4 mm by placing the fiber mats between two, 2 mm thick, stainless steel ‘dog-bone’ shaped templates and carefully cut as described previously [342, 343]. A stainless-steel surgical blade was employed to make the straight cuts of the template, and a 6-mm dermal biopsy punch used to cut the radii.

Great care was taken to ensure clean-cut samples, reducing sample flaws that could result in inaccurate testing. Placing the gage-length of the ‘dog bone’ shaped samples between two glass-slides and measuring the thickness of the sample using a digital micrometer determined sample thickness. The as-spun PCL, PES-PCL and PCL-PDMS samples were placed into aluminum grips and pulled at a cross-head speed of 50 mm/min by a 50-lb load cell to failure. Core-shell gelatin-PCL samples were soaked in phosphate buffered saline (PBS) solution for 1 hour at 37 ºC, to simulate the rapid hydration and softening of the gelatin in aqueous solution, and pulled to failure immediately following the soak using the methods described above. Force as a function of displacement was recorded and converted to engineering-stress vs. percent elongation for analysis.

7.2.2.4 Surface properties of PCL and core-shell nanofibers

Surface wetting properties were quantified using contact angle goniometry. PCL and core-shell nanofiber scaffolds plasma treated with air gas, as described previously, were cut into 5 x 1 cm segments and water contact angle was measured using a Krüss

162

Easydrop (Krüss, Hamburg, Germany) water contact system. A 300-µL drop of deionized water was placed on a dry area of the fiber, and using the Easydrop software, water contact angle was measured using a sessile drop contact to surface measurement.

Five measurements were made and the average ± standard deviation (SD) was recorded.

7.2.2.5 Characterization of core-shell structure using transmission electron microscopy (TEM)

To examine the core-shell structure, polymer solutions were directly spun as described above onto TEM 200 mesh copper grids (Ernest F. Fullam, Inc., NY) for ~ 2 seconds to obtain representative PDMS-PCL core-shell nanofibers. These were then imaged directly using a transmission electron microscope (Tecnai 20, FEI Company, Hillsboro, OR).

7.2.3 Patient derived OSU-2 cell culture

Primary tumor cells were obtained from a patient with GBM under OSU approved IRB protocol 2005C0075 (dated 11/08/08). Written consent was obtained from all participants involved in this study. Patient derived GBM (OSU-2) cells obtained from tumor tissue were cultured routinely as described previously [190]. Briefly, cells were cultured in

(DMEM/F12 (Invitrogen)) containing 10% fetal bovine serum (Invitrogen) with 1% Pen-

Strep (Invitrogen), fed 2-3 times a week and passaged on reaching confluency.

7.2.4 Analysis of cell adhesion on PCL and core-shell nanofibers

Scaffolds (ENFs deposited on ~ 16 mm discs) were fixed, using a medical adhesive (Dow

Corning Silastic Brand, Medical Adhesive, Silicone Type A), to the bottom of a 24 well

163 plate where the bottom was drilled (hole size ~ 11.2 mm dia) to reduce working distance during imaging (Figure 67).

Figure 67. Cell culture well design modified for use with electrospun nanofiber discs. (A) Part of a 24 well plate with a drilled hole (shown via bidirectional white arrow). (B) Electrospun nanofiber disc used with this design attached to the bottom of the well using medical adhesive.

For sterilization, scaffolds were then incubated with 70% ethanol combined with UV in the tissue culture hood (0.5 h), after which they were washed with PBS and allow to dry overnight to remove any residual ethanol. Before cell seeding, scaffolds were washed with OSU-2 cell culture media (3X) and incubated for ~ 1 h. To examine initial cell adhesion, ~ 20,000 cells pre-labeled with cell tracker (CMFDA Invitrogen) were seeded on all scaffolds (N=3, for each scaffold). After 0.5 h, scaffolds were washed with cell culture media (3X) and imaged using an inverted fluorescent microscope (Olympus IX

71) with a 10X objective. Images were acquired at random locations (n ≥ 7, for each well) and were thresholded to allow rapid quantification of cell adhesion using the particle analysis function in Image J. Cell adhesion to all scaffolds is reported as average number of cells adhered per unit area (mm2) ± SD.

164

7.2.5 Morphological analysis of OSU-2 Cells on PCL and core-shell nanofibers.

Nanofiber scaffolds were sterilized as described previously (section 7.2.4). Following this, OSU-2 cells, pre-labeled with a cell tracker dye (CMFDA Green, Invitrogen) were seeded on all scaffolds. After ~ 24 h, images of cells on nanofiber scaffolds (N = 3, for each scaffold) were acquired at random positions and subjected to image analysis. To examine the interaction of cells with nanofiber scaffolds, the Feret diameter (i.e., maximum distance between any two points in a region of interest, i.e., in this case, a single cell) was quantified using Image J. This parameter provides the extent of elongation of a cell along the nanofiber. At least 140 individual cells were analysed on each nanofiber scaffold and are reported as the average ± SD for all cells analysed.

7.2.6 Analysis of single cell migration using time lapse confocal imaging on

PCL and core-shell nanofibers

All scaffolds were sterilized as described in section 7.2.4. OSU-2 cells, pre-labeled with cell tracker dye (CMFDA, Green, Invitrogen) were seeded on scaffolds (N=3) at ~ 20,000 cells/well in a 24 well plate. After ~ 6 h, scaffolds were washed with OSU-2 cell culture media (3X) to remove non-adherent cells and then placed on the microscope stage equipped with a Weather Station (Precision Control LLC) to maintain incubator conditions and adjust to this setting for ~ 3 h prior to imaging. Following this, a 50 µm z- stack (step size = 10 µm, no. of steps = 6) was captured at multiple, random locations for each scaffold every 20 min for a total of 12 hours using an inverted microscope

(Olympus IX 71) equipped with a spinning disk confocal attachment. The z-stacks were

165 then projected (maximum) and concatenated to create migration movies. Movies were analyzed using the Image J MTrack J plug in. Cells that underwent proliferation during the time course of analysis were not considered for migration analysis (e.g., Video 7.1, available at http://youtu.be/sWGg-cOb_Lc). Atleast 95 individual cells for each substrate type were tracked using MTrack J. Migration speeds for all nanofiber scaffolds are reported as box and whisker plots showing median, mean and associated outliers.

7.2.7 Analysis of FAK/MLC2 signaling expression using western blotting on

PCL and core-shell nanofibers

OSU-2 cells were cultured as described in section 7.2.3. Approximately 5 × 105 cells were seeded on pre-sterilized electrospun core-shell and PCL scaffolds glued to 60 mm culture dishes and cultured for ~ 48 h. After 48 h, OSU-2 cell-nanofiber constructs were frozen at -80 ºC. Following this, cells were lysed in 20 mM Tris-HCl buffer, pH 7.6, containing 150 mM NaCl, 1% v/v NP-40, 0.5% v/v sodium deoxycholate, and protease plus phosphatase inhibitors (Complete and PhosSTOP cocktails; Roche Applied Science,

Indianapolis, IN). Proteins were processed for Western blot analysis with antibodies against focal adhesion kinase (FAK) and phospho-Tyr925 FAK (pFAK), myosin light chain 2 (MLC2) and phospho-Ser19 MLC2, and β-tubulin (all from Cell Signaling,

Danvers, MA). Protein expression was quantified by relative densitometry. All values were normalized to the tubulin loading control and expressed as ratio to the band intensities obtained on aligned PCL.

166

7.2.8 Analysis of matrix metalloproteinase (MMP) expression using quantitative polymerase chain reaction (PCR)

RNA from OSU-2 cells cultured on nanofibers (specifically, PCL versus PCL-HA) was extracted using Trizol as described (detailed procedure for extraction is available at http://tools.invitrogen.com/content/sfs/manuals/trizol_reagent.pdf). Two µg of RNA from each sample was used to make cDNA via a 20 µL reverse transcription reaction. One µL of cDNA was used in each 20 µL qPCR reaction. Triplicate reactions were run for each gene of interest. Forward and reverse universal primers detecting glyceraldehyde-3- phosphate dehydrogenase (GAPDH, a house keeping gene, normalization control),

MMP-2, MMP-9, and MMP-13 were used.

7.2.9 Statistical analysis

All data was analyzed using statistical analysis software (JMP Pro 9) by Oneway

ANOVA. Statistical differences between nanofiber scaffolds post ANOVA were analyzed using the Tukey-Kramer HSD test. In all cases unless otherwise noted, p < 0.05 was considered to be statistically significant.

7.3 Results

7.3.1 Characterization of PCL nanofibers and core shell nanofibers

Aligned electrospun PCL nanofibers exhibited submicron fiber diameters (~ 0.9 µm) with cylindrical morphologies as observed via SEM, and closely mimicked the topographical features of white matter tracts (Figure 68, 69, Table 10).

167

Figure 68. Comparison of white matter histology to aligned electrospun nanofibers. (A) Histology of corpus callosum showing stained white matter tracts with Luxol fast blue. Image taken from [49]. (B) Aligned electrospun PCL nanofibers as observed via SEM closely mimicking the in vivo topographical features seen in (A).

Figure 69. Micro-structural features of PCL and core-shell nanofibers observed via SEM. (A) Gelatin-PCL. (B) PCL. (C) PDMS-PCL. (D) PES-PCL. (E) PCL-Collagen. (F) PCL-HA. (G) PCL-Matrigel. Scale bar indicates 20 µm. Fast Fourier Transform (FFT) with associated images are shown as insets.

168

Scaffold Fiber diameter Fiber density Elastic Ultimate Tensile (nm) (mm-1) Modulus Strength (UTS) (MPa) (MPa) Gelatin-PCL 880 ± 295 577 ± 151 2.4 ± 0.6 0.8 ± 0.2 PCL 925 ± 183 594 ± 100 7.9 ± 1 4.6 ± 1.3 PES-PCL 854 ± 302 648 ± 159 28.6 ± 6.6 9.2 ± 1.7 PDMS-PCL 829 ± 232 638 ± 94 33.3 ± 6.9 13.1 ± 2.2 Table 10. Micro-structural and mechanical characterization of PCL and core-shell nanofibers with altered cores but an identical surface chemistry (PCL). All nanofibers examined displayed a contact angle of 0˚, indicative of complete wetting and uniform surface chemistry.

Nanofibers produced via core-shell electrospinning with an identical shell (PCL) but varying core polymeric materials exhibited micro structural characteristics very similar to

PCL-only ENFs with diameters ranging from ~ 0.8-0.9 µm (Table 10) and fiber densities of ~ 570-650 fibers/mm (Table 10). These diameters are within the range observed for white matter tracts in vivo (0.5-3 µm) [168]. FFT analysis of these images showed nearly aligned nanofibers for all substrates investigated, with the resultant FFT image showing symmetric pixels and narrow areas in the center (Figure 69, FFT images shown as insets).

Radial summation of pixel intensities, followed by normalization, further confirmed nearly uniform alignment for all samples examined, with two narrow peaks seen for all samples at 90º and 270º (because of symmetry) with a significant overlap of normalized intensity profiles (Figure 70).

169

Figure 70. FFT analysis of representative SEM images via radial summation of pixels normalized for all nanofiber samples and plotted versus degree.

In a representative PDMS-PCL nanofiber, TEM revealed the presence of the expected core-shell structure i.e., PDMS ‘core’ surrounded by a PCL ‘shell’ (Figure 71).

Figure 71. Transmission electron microscopy imaging of a representative PDMS-PCL core-shell nanofiber. (A) Schematic depicting the structure of core-shell nanofiber. (B) TEM image of PDMS-PCL core-shell nanofiber showing the PDMS ‘core’ and PCL ‘shell’ surrounding the core. Scale bar = 0.2 µm.

170

Despite the great similarities in micro structure, PCL, and core-shell gelatin-PCL, PES-

PCL and PDMS-PCL exhibited dramatically different moduli. The modulus of gelatin-

PCL was the lowest among the core-shell fibers examined (2.4 MPa) followed by PCL

(7.9 MPa, ~ 3X that of gelatin-PCL), PES-PCL (28.6 MPa) and PDMS-PCL (33.3 MPa, both ~ 4X compared to PCL, the moduli of PES-PCL and PDMS-PCL not being dramatically different from each other). Ultimate tensile strength (UTS) for these nanofibers followed trends observed for the modulus (Table 10). Nearly identical surface chemistries for these nanofibers were confirmed via contact angle goniometry. A contact of 0˚ indicated complete surface wetting following plasma treatment [344]. Thus, these materials demonstrate similar surface properties, but varying mechanical moduli.

Nanofibers with identical core materials (i.e., PCL) but varying shell materials also exhibited similar microstructural features to pure aligned PCL nanofibers: fiber diameters of ~ 0.8-0.85 µm and densities of ~ 600-700 fibers/mm (Table 11).

Scaffold Fiber diameter (nm) Fiber density (mm-1) PCL-Collagen 841 ± 329 622 ± 145 PCL-HA 813 ± 246 597 ± 106 PCL-Matrigel 815 ± 300 684 ± 101 Table 11. Micro-structural characterization of core-shell nanofibers with altered surface chemistries and an identical core (PCL). All nanofibers examined displayed a contact angle of 0˚, indicative of complete wetting.

Further, complete water wetting was observed for these nanofibers. Plasma treatment was not necessary for nanofibers with varying biomimetic chemistries (as all of them are hydrophilic chemistries), however was performed on all nanofibers to compare to the otherwise hydrophobic PCL nanofiber that was also plasma treated to maintain identical 171 experimental conditions to the control, PCL. Fiber alignment of both types of ENFs was confirmed via FFT analysis (Figure 69, 70).

7.3.2 OSU-2 cell adhesion on PCL nanofibers and core-shell nanofibers

To examine the interaction of OSU-2 cells with various nanofiber scaffolds, the initial adhesion of OSU-2 cells to these scaffolds was examined. Shorter times were investigated to avoid confounding influences from cell proliferation. In aligned core-shell nanofiber scaffolds presenting altered mechanical properties, no significant differences in initial cell adhesion were observed (p = 0.09, ANOVA) although this value was the highest for the PCL nanofiber scaffold (Figure 72).

Figure 72. Adhesion of OSU-2 cells on various nanofibers as a function of nanofiber mechanics. No significant differences between samples observed. 172

In contrast, OSU-2 cells on nanofibers having identical PCL cores but varied biomimetic chemistries as shells showed differences in cell adhesion (p < 0.0001, ANOVA). For example, HA surface chemistry reduced cell adhesion compared to PCL. Also, surprisingly, this behavior was also observed for PCL-collagen compared to PCL.

However, the Matrigel chemistry did not significantly alter adhesion versus PCL (Figure

73).

Figure 73. Adhesion of OSU-2 cells on PCL nanofibers and nanofibers with identical cores and various biomimetic shells. * indicates statistically significant difference compared to PCL nanofiber.

173

7.3.3 OSU-2 cellular morphology on PCL nanofibers and core-shell nanofibers

To examine the morphological signatures of OSU-2 cells on various nanofiber scaffolds, the Feret diameter of single OSU-2 cells was quantified. In almost all scaffolds examined, cells elongated along the nanofibers, exhibiting a bipolar, elongated, spindle shaped morphology, reminiscent of the morphology observed in vivo [173, 174, 201, 206,

302] (Figure 74).

Figure 74. OSU-2 cell morphologies on various nanofibers examined. (A) Gelatin-PCL. (B) PCL. (C) PDMS-PCL. (D) PES-PCL. (E) PCL-Collagen. (F) PCL-HA. (G) PCL-Matrigel. Scale bar indicates in (A) 100 µm. Bidirectional arrow indicates direction of fiber alignment.

However, the extent of elongation was significantly influenced by the mechanical properties of the underlying nanofiber scaffold. For example, Feret diameters were influenced by nanofiber mechanics with the highest diameters observed on PCL (~ 236

µm) and lower values on lower moduli (i.e., gelatin-PCL, ~ 2 MPa), and higher moduli 174 nanofibers (i.e., PES-PCL and PDMS-PCL, ~ 30 MPa) when compared to PCL (~ 8

MPa). Further, the lowest feret diameters (~ 169 µm) were seen on the softest nanofiber examined (i.e., gelatin-PCL, ~ 2 MPa) revealing morphological sensitivity to nanofiber mechanics (Figure 75).

Figure 75. Feret diameter analysis of OSU-2 cells as a function of nanofiber mechanics. N ≥ 142 individual cells analysed for each nanofiber. * and ** indicates statistically significant difference compared to PCL nanofiber. Levels marked by identical number of * are not significantly different from each other as determined by Tukey-HSD test.

Similarly, this extent of elongation was also significantly influenced by biomimetic chemistries of aligned PCL core nanofibers. For example, for all chemistries examined, feret diameters were significantly lower compared to PCL (p < 0.0001) (Figure 76).

175

However, although OSU-2 cells on PCL-collagen nanofibers were less elongated than on

PCL, individual measured cell areas were equivalent to that observed on PCL (Figure

77). Thus, cells on PCL-collagen were more likely to also spread transversely possibly engaging multiple nanofibers. Similarly, with reduced feret diameters observed on PCL-

Matrigel, OSU-2 cells on these nanofibers were also more likely to possess rounded morphologies extending some processes along the nanofibers. In sum, biomimetic chemistries on an aligned PCL core significantly influenced the morphological behavior of OSU-2 cells.

Figure 76. Feret diameter analysis of OSU-2 cells on PCL and core-shell nanofibers with an identical PCL core and various biomimetic chemistries as shells. N ≥ 199 individual cells analysed for each nanofiber. * and ** indicates statistically significant difference compared to PCL nanofiber. Levels marked by identical number of * are not significantly different from each other as determined by Tukey-HSD test. 176

Figure 77. Cell area analysis of OSU-2 cells on PCL and core-shell nanofibers with an identical PCL core and various biomimetic chemistries as shells. N ≥ 199 individual cells analysed for each nanofiber. * indicates statistically significant difference compared to PCL nanofiber.

7.3.4 OSU-2 single cell migration on PCL nanofibers and core-shell nanofibers

OSU-2 single cell migration was examined on PCL and other core-shell nanofibers using time lapse confocal microscopy. Migration was sensitive to nanofibers with different mechanical properties with the fastest migration speeds observed at ~ 8 MPa moduli for

PCL nanofibers (~ 11 µm/h, Video 7.2 [http://youtu.be/atBl-vGAvjE]) and slower migration on both lower (i.e., gelatin-PCL, ~ 2 MPa, Video 7.3 [http://youtu.be/io0Yc-

Fzabw]) and higher (i.e., PDMS-PCL, Video 7.4 [http://youtu.be/92_LQYBLRIY] and

PES-PCL, Video 7.5 [http://youtu.be/AmAm9hg6ndQ], ~30 MPa) moduli nanofibers

(Figure 78). Interestingly, this range of mechanical environments demonstrated for the

177 first time that glioma migration, in addition to being a function of modulus, can exhibit a peak value at a particular substrate modulus. Further increases in the modulus (i.e., PES-

PCL or PDMS-PCL (~ 30 MPa) versus PCL (~ 8 MPa)) reduced the migration speeds although these speeds were still significantly higher (p < 0.005, Tukey-Kramer HSD) than those observed with the softest core-shell nanofiber (gelatin-PCL, ~ 2 MPa). These trends are in agreement with the observed morphological responses. In no case was migration inhibited; further demonstrating that aligned topographical features can guide migration. However, migration potential was a strong function of nanofiber mechanics.

Figure 78. Analysis of single cell migration speed as a function of nanofiber mechanics shown as a box and whisker plot. N ≥ 95 individual cells analysed for each nanofiber. Blue * and ** indicates statistically significant difference compared to PCL nanofiber. Levels marked by identical number of * are not significantly different from each other as determined by Tukey- HSD test. Red lines within the box indicate mean and black lines indicate median. 178

In addition to mechanics, core-shell nanofibers with various chemistries (i.e., HA,

Collagen-I and Matrigel) with a PCL core were fabricated to examine the role of chemical cues on tumor cell migration ex vivo. Interestingly, OSU-2 cell migration was sensitive only to PCL-HA core-shell nanofibers (Video 7.6, available at http://youtu.be/Wh0aZafUZvs) with significantly decreased cell migration speeds observed when compared to PCL nanofibers (p < 0.0001, Tukey-Kramer HSD) (Figure

79). Thus, HA acts as a negative chemical cue that hinders migration in this setting.

Interestingly, the addition of collagen-I (Video 7.7, available at http://youtu.be/Vw-

Flqrn6C4) and Matrigel (Video 7.8, available at http://youtu.be/WDXfgR3gGEQ) as shells to PCL core nanofibers had no significant effect (p > 0.2, Tukey-Kramer HSD) compared to bare PCL nanofibers, demonstrating that OSU-2 migration potentials were insensitive to these chemistries when examined in an aligned nanofiber setting. In contrast to observations with nanofibers with altered mechanics but identical PCL shell, migration results did not correlate strongly with morphological observations except for the case of PCL-HA. This is because, although cells elongated less on PCL- collagen/PCL-Matrigel versus PCL, their migration potential was not significantly compromised indicating possible involvement of multiple alternative signaling pathways.

179

Figure 79. Analysis of single cell migration speed as a function of biomimetic chemistries shown as a box and whisker plot. N ≥ 142 individual cells analysed for each nanofiber. Blue * indicates statistically significant difference compared to PCL nanofiber. Red lines within the box indicate mean and black lines indicate median.

7.3.5 FAK/MLC2 signaling and MMP gene expression on PCL nanofibers and core-shell nanofibers

To further examine the underlying mechanism of the observed migration results, we investigated signaling expression of an adhesion (FAK) and migration (MLC2) pathway implicated in high grade gliomas, as both FAK [345] and MLC2 (Isoform A) [346] are overexpressed in GBM tumor tissues. Consistent with live, single cell imaging migration observations, FAK, MLC2 and their phosphorylated forms were highly overexpressed on aligned PCL nanofibers. The lowest expression was observed on the softest substrate,

180 gelatin-PCL (~ 2 MPa, Figure 80 A). For example, normalized pFAK and FAK expression levels on gelatin-PCL were 0.34 and 0.28 versus 1 for PCL (Table 12).

Similarly, normalized pMLC2 and MLC2 expression levels on gelatin-PCL were 0.34 and 0.1 versus 1 for PCL.

Figure 80. Analysis of FAK/MLC2 expression via western blotting. (A) pFAK, FAK, pMLC2, MLC2 expression as a function of mechanics. (B) pFAK, FAK, pMLC2, MLC2 expression for PCL versus PCL-HA combination. In both cases, tubulin served as a loading control.

Levels expressed as proportion of control (PCL = 1) Gelatin-PCL PCL PDMS-PCL PES-PCL pFAK 0.34 1 0.84 0.67 FAK 0.28 1 0.74 0.73 pMLC2 0.34 1 0.7 0.67 MLC2 0.1 1 0.74 0.88 Table 12. Protein expression levels quantified via densitometry from western blots as a function of nanofiber modulus.

181

Although expression levels for nanofibers with moduli greater than PCL (i.e., PES-PCL and PDMS-PCL, ~ 30 MPa) were less than PCL (~ 8 MPa) and correlated well with migration data, the differences in expression levels were less dramatic when compared to gelatin-PCL/PCL combinations. For example, normalized expression levels of FAK,

MLC2 and their phosphorylated versions for both PES-PCL and PDMS-PCL ranged from 0.67-0.88 versus 1 for PCL (Table 12). Nonetheless, these expression levels were much higher compared to gelatin-PCL. In sum, FAK/MLC2 signaling events were positively correlated with GBM cell migration observed on varying moduli nanofibers further suggesting involvement of such pathways triggered purely by nanofiber mechanics.

The influence of chemistry on expression levels of FAK, MLC2 and their phosphorylated versions was also examined. Specifically PCL and PCL-HA materials were studied, since the HA ‘shell’ chemistry produced a significant reduction in migration potential compared to PCL. Whereas the total FAK and MLC2 expression levels were not very different for both PCL and PCL-HA, expression of the phosphorylated version was reduced compared to PCL with differences in pFAK expression being more dramatic than those of pMLC2 (Figure 80 B). For example, pFAK expression level was 0.3 versus 1 for

PCL whereas pMLC2 expression level was 0.74 versus 1 for PCL (Table 13). [Note that the densitometry values reported are representative intensities of the western blots

(Figure 80) for qualitative comparison only in addition to visual comparison of blots].

Thus, specific chemistries (i.e., HA) influenced adhesion and migration pathways, demonstrating that in addition to mechanics, the chemical nature of the underlying

182 scaffold also plays a crucial role in migration behaviors (modulus of PCL-HA nanofibers

= 7.6 ± 1.2 MPa, not different from modulus of PCL nanofibers = 7.9 ± 1 MPa).

Levels expressed as proportion of control (PCL = 1) PCL PCL-HA pFAK 1 0.3 FAK 1 1.3 pMLC2 1 0.74 MLC2 1 0.94 Table 13. Protein expression levels quantified via densitometry from western blots for PCL versus PCL-HA nanofibers

Because of the physiological relevance of these materials and the potential of clinical translation, MMP (i.e., MMP-2, MMP-9 and MMP-13) gene expression was also investigated for cells grown on PCL and PCL-HA nanofiber scaffolds. Several studies have demonstrated a relationship between brain tumor cell malignancy and MMP levels

[172, 347-349]. In particular, MMP-2 and MMP-9 are highly expressed in vivo, enhancing invasion potential and have also been shown to activate growth factors implicated in GBM migration (e.g., transforming growth factor-β or TGF-β) [332-334].

GBM cells with induced MMP-13 expression have been demonstrated to possess higher migratory potential in vitro [335] and further, GBM cancer stem-like cells with high invasion potential have been shown to express MMP-13 [336]. Consistent with a reduction in migratory potential on PCL-HA nanofiber scaffolds, a significant decrease in

MMP-2 levels was evidenced (Figure 81), indicating the potential involvement of MMP-

2 (p = 0.013, ANOVA) in these behaviors. However, the expression levels of two other

183

MMPs (i.e., MMP-9 and MMP-13) were not significantly different for PCL versus PCL-

HA scaffolds (p = 0.262 and p = 0.6 as reported from ANOVA for MMP-9 and MMP-13 respectively).

Figure 81. Normalized MMP-2, MMP-9, and MMP-13 gene expression levels for PCL versus PCL-HA nanofibers. * indicates statistical significance.

7.4 Discussion

In this study, aligned electrospun nanofibers were fabricated via core-shell electrospinning to mimic the topographical architectures found in white matter. Using this approach, nanofibers with different mechanical properties, yet nearly identical micro- architectures and surface chemistries, were prepared. Similarly, core-shell nanofibers were synthesized that displayed different, physiologically-relevant biological surface

184 chemistries on an identical core (i.e., PCL) nanofiber that provided a consistent mechanical modulus while altering surface chemistry. This work also addresses one of the most crucial issues in brain cancer biology by providing tunable and well-defined 3D platforms that mimic the major highways (i.e., white matter tracts) for GBM migration while simultaneously allowing examination of biophysical and biochemical cues regulating tumor cell behaviors.

Core-shell spinning thus provides a unique method to isolate the effects of individual factors (e.g., mechanics, chemistry) while simultaneously retaining the in vivo-like morphological and molecular signatures of glioma migration as observed here and as demonstrated previously [173, 174]. In contrast, decoupling the physical (e.g., porosity, mechanics) and chemical properties of 3D hydrogels is challenging [79], limiting the ability to independently investigate the effects of these variables on tumor cell behaviors.

For example, in most mechanical studies using 3D hydrogel systems, modulus is adjusted by altering the polymer density, which introduces changes in chemistry as well as alteration in adhesion site density. Similarly, in other electrospun systems, modulus is altered by varying polymer composition, which may be associated with changes in microstructure and chemistry [350, 351].

Using core-shell electrospinning, nanofiber modulus was altered to examine the effect of mechanics on GBM cell behaviors. Nanofibers fabricated displayed moduli ranging from

~ 2-30 MPa, which spans the observed physiological range and also sufficiently captured biologically relevant migration behavior better than TCPS/glass. For the first time, a peak in migration speed was observed (e.g., PCL, ~ 8 MPa). Nanofibers exhibiting moduli less

185 than that of PCL (i.e., ~ 2 MPa) and greater than PCL (~ 30 MPa) allowed slower migration, with the softest nanofiber examined promoting the slowest migration.

Although few reports have examined the mechanical properties of myelinated central nervous system axons/nerve fibers, atomic force microscopy (AFM) studies have reported modulus values of ~ 0.1-1.5 MPa for mouse-derived peripheral axons [352]. In our nanofiber-based white matter mimicking system, we observed strong migration sensitivity in the range of ~2-30 MPa. Additionally, this sensitivity to nanofiber modulus correlates well to a known cellular sensing process referred to as the “catch-bond formation” mechanism [353] in which cell interaction with the underlying substrate is significantly enhanced at a particular rigidity level - in our case, ~ 8 MPa.

This observed mechanical sensitivity in migration was not directly correlated to initial, short-term cell adhesion, as no significant alterations as a function of mechanical properties were observed. Thus, initial cell adhesion is most likely guided by the surface properties of the underlying nanofiber scaffold and not mechanical features. This is not surprising given the nearly identical surface properties observed using contact angle goniometry and SEM. However, migration results were further validated by studies of

FAK/MLC2 signaling of cells cultured (~ 48 h) on nanofiber scaffolds. The lowest expression levels of these proteins were seen on gelatin-PCL, correlating to adhesion and migration data. This suggests that gelatin-PCL (the softest ENF, ~ 2 MPa) provided the least resistance to OSU-2 cell-generated traction forces, thereby significantly slowing movement compared to other core-PCL shell nanofibers. Whereas involvement of these pathways in glioma migration has been implicated previously [345, 346, 354, 355], this is

186 the first study to show a strong dependence of these pathways on nanofiber modulus.

These results suggest that GBM cell behaviors including feret diameters, migration speed, FAK and MLC2 expression, are all tightly regulated by the mechanical properties of the underlying nanofiber scaffold.

To compare the influence of mechanics to that of surface chemistry, biologically-relevant molecules (i.e., collagen, HA and Matrigel) were then incorporated as the ‘shell’ on PCL

‘core’ nanofibers permitting examination of these chemistries in conjugation with aligned topography and consistent mechanical modulus. Whereas each chemistry elicited unique responses either in terms of morphology or adhesion, HA was the only chemistry that significantly influenced (i.e., impeded) cell motion in an aligned topography-mimetic setting. This correlated well with observations of reduced initial cell adhesion, reduced feret diameter, reduced expression levels of pFAK and pMLC2, and reduced MMP-2 gene expression levels observed for PCL-HA versus the PCL control. These observations were not surprising given the role of HA as a chemo-repellant in vivo [152] and our previous in vitro observations of GBM (OSU-2) migration behaviors in HA-based hydrogel systems [228]. However, our findings in a topography-mimetic setting contradict previous studies suggesting that HA promotes tumor cell migration and invasion, conclusions drawn primarily using HA either as a soluble factor or incorporated with other ECM components [186, 187, 189]. In agreement with observations using HA- hydrogel systems [227], our findings strongly suggest that the physical presentation of

HA plays a critical role in migration. Increased HA levels are often observed in GBM tumors in vivo, along with over expression of HA receptors (i.e., CD44, RHAMM)

187 contributing to tumor progression. Thus, this biomimetic materials platform, mimicking in vivo topography and HA chemistry, should enable further investigation of CD44 signaling and associated events, in addition to those of the FAK/MLC2 pathways investigated here.

In contrast, other surface chemistries explored (i.e., PCL-Matrigel and PCL-collagen nanofibers) altered the morphology of OSU-2 cells, but not migration behaviors.

Interestingly, initial cell adhesion, cell area, and ferret diameter (i.e., alignment to the scaffold) were equivalent to or less than that of the PCL control for both PCL-Matrigel and PCL-collagen nanofibers. Similar results for aligned PCL versus PCL blended with collagen nanofibers have been reported using the U-373 astrocytoma cell line [319]. In no case were cell attachment or spreading enhanced by altered surface chemistry, despite the fact that these chemistries are known to provide enhanced adhesive cues compared to the bare PCL nanofiber control (e.g., [356, 357]). It is important to note that short-term adhesion (0.5 h) was examined to minimize confounding influence from proliferation versus longer time points (i.e., 24 h) that might yield different results. This suggests that alternative pathways/mechanisms could be at play in supporting cell movement vs. adhesion, and that topographical cues alone could be sufficient to support migration.

Detailed interrogation into the adhesion and migration mechanisms should provide insight into such behaviors. Collectively, these results demonstrate that core-shell ENFs permit examination of cell behaviors in response to specific, individual, chemical, mechanical, or topographical cues in a near-3D setting.

188

ENFs have been classified as 2D (i.e., randomly-aligned ENFs), 3D (i.e., ENF scaffolds), and even 1D (i.e., aligned ENFs) systems [358-360]. In general, porosity is insufficient to permit cell migration through ENF structures without modification [361], and thus cell culture is limited to their surfaces (e.g., 2D). However, individual tumor cells migrate differently on ENFs versus a flat surface (e.g., 2D TCPS), with the scaffold being sufficiently irregular for cells to exhibit 3D migratory patterns (e.g., elongation along fiber length and formation of protrusions along the fiber that closely mimic cell protrusions through a 3D porous hydrogel matrix). The ability of aligned ENFs to mimic the morphological, and in part, molecular signatures, of 3D migration (e.g., migration driven by myosin-II) comparable to those seen in brain slices has been demonstrated previously [173, 174]. This work further extends the application of this biomimetic system to unravel the role of mechanics and chemistry on tumor cell behaviors and associated signaling pathways, which should enable identification and discovery of mechanobiologically or chemically inspired therapeutic candidates. Inarguably, such a system offers tremendous advantages in terms of ease of preparation, scalability, tunability, and compatibility with other downstream cell assays, as well as in studying the role of competing cues in a physiologically relevant setting.

Because these materials provide 3D topography that mimics the in vivo microenvironment ex vivo, this work also advances the field of tissue engineering and regenerative medicine. Aligned core-shell biomaterials with tunable mechanics and chemistry can be employed as mimetics of the white matter microenvironment. These ex vivo biomimetic models could also serve as powerful tools to examine diseases of the

189 white matter, and have potential implications for stem cell therapy, studies of neural regeneration and development, and myelination. Whereas aligned, as well as core-shell, nanofibers have been fabricated previously [339, 362-370], few studies have utilized such nanofiber systems to examine and elucidate the influence of mechanical/chemical properties on cell behaviors in vitro in general (e.g., [343]) and none have examined such influences on tumor cell migratory potential. Further, in comparison to core-shell nanofibers spun previously in random orientations (i.e., gelatin-PCL [364, 366, 371],

PES-PCL [343], PCL-collagen [356, 372]), this work employed aligned core-shell nanofibers and also introduced new biomaterial combinations (i.e., PDMS-PCL, PCL-HA and PCL-Matrigel core-shell nanofibers). Thus, this work complements and adds to the library of existing aligned core-shell nanofiber scaffold materials that can be fabricated and used as well defined ex vivo biomimetic cell environments.

7.5 Conclusions

Here, we developed and characterized a white matter tract, topography-mimetic, aligned

ENF platform fabricated via core-shell electrospinning. This platform displayed independently tunable mechanical and chemical properties, thus providing the potential to further explore the complex interplay of mechanics, chemistry and topography on malignant tumor cell behaviors. We have further demonstrated its utility in studying patient-derived GBM behaviors, including cell adhesion, morphology, migration and

FAK/MLC2 signaling expression, revealing GBM sensitivity to nanofiber mechanics as well as biologically relevant chemistries. This biomimetic material platform should have broad applicability for examining and further evaluating fundamental questions in the

190 biology of brain tumors as well as in neuroscience ex vivo. Finally, these mechanical and chemically tunable biomaterials could also be employed to examine behaviors of other types of tumor cells (e.g., prostrate, head and neck tumors) that metastasize via perineural structures [373] and cell behaviors of other tissues having similar topographical architectures (e.g., the aligned myocardium in the heart [374]) thereby demonstrating its broad utility.

191

Chapter 8: Incorporating Aligned Electrospun Nanofibers in

3D Hydrogel Systems: Mimicking White Matter Tracts in

Neural Extracellular Matrix

Recapitulating complex tissue architectures and mimicking multiple features of such architectures is highly desirable for creating advanced in vitro 3D cell culture models. To this end, we have integrated aligned electrospun nanofibers within 3D hydrogels to mimic white matter tracts in neural ECM. Preliminary observations using such models indicate that cells align and elongate along nanofibers in hydrogel systems, including hydrogels that do not typically support cell adhesion. An integrated hydrogel-aligned electrospun nanofiber platform would be tremendously useful in examining processes of highly directed invasion/migration in a 3D setting and is also amenable to other biological investigations.

8.1 Introduction

To develop complex tissue architectural features in vitro (e.g., those found in brain), a variety of factors including biomaterial properties, is critical. Chapter 6 examined the use of hydrogels as a model of the in vitro brain tumor microenvironment. Hydrogels can function as tissue mimics, however they do not provide aligned features (such as those found in white matter tracts) that act as highways for tumor dissemination. Further, as discussed in Chapter 7, electrospun nanofibers (ENF) capture in vivo like morphologies

192 and in part, the in vivo molecular signatures of tumor cells by mimicking the aligned white matter topographical features. However, the motion of tumors cells across the roughened surface shares features of that in three dimensional (3D) substrates, a majority of cell motion is restricted to nanofiber surfaces. Most importantly, components of neural and tumor ECM, mimicking the enveloping 3D matrix that surrounds white matter tracts, is not incorporated. Further, although MMP driven cell migration evidenced in Chapter 7 in the case of PCL-HA nanofiber scaffolds, MMP expression levels in response to a 3D hydrogel-like barrier would be different and in fact, enhanced as it is necessary to degrade the hydrogel for efficient invasion.

In the context of tissue engineering and regenerative medicine, controlling the orientation of individual cells is also important, aiding in development of complex tissue architectures via contact guidance [375-377]. Strategies to control cell orientation in hydrogels have been reported (e.g., patterning cues to direct cell behavior (micro patterning)), however most of these do not have good control over individual cytoskeletal cell orientation. To develop complex tissue mimics, controlling cell orientation at the single cell level is crucial, as adhesion and morphology are strongly influenced by nanotopography [378].

To address these issues, aligned electrospun nanofibers were integrated with 3D hydrogel systems for use in tissue engineering and as tumor cell culture models, combining multiple features in a single 3D system. Hydrogel-electrospun nanofiber composite materials in different configurations have been previously utilized by our research group for controlled release of soluble factors [379], as neural prosthetic coatings [380] and by

193 others as biomimetics of the extracellular matrix or for tissue engineering applications

[381-385]. However, in most of these studies, the fibers were randomly organized and did not provide aligned topographical features. Very few studies have examined aligned nanofiber-hydrogels constructs [386]. For example, a recent study developed collagen hydrogel-poly (lactic) acid (PLA) nanofiber assemblies [386] for regenerative medicine applications. However, to the best of our knowledge, these composites have not been used as 3D tumor cell culture models. Moreover, aligned electrospun nanofibers have not been combined with non-adhesive hydrogels commonly used in tissue engineering and regenerative medicine to improve cell adhesion, spreading and most importantly to control cell orientation. To demonstrate this utility, hydrogel biomaterials commonly used in tissue engineering (e.g., hyaluronic acid, agarose, and poly (ethylene) glycol

(PEG)) were combined with aligned electrospun fibers. The addition of nanofiber components not only enhanced cell adhesion and spreading compared to non-adhesive hydrogel controls, but also induced directed cell alignment, thereby adding an important component of topography to 3D hydrogels. Also, these materials, in the context of tumor cell models, have the potential to reveal details of highly directed invasion and migration patterns; and in particular, permit further examination of matrix metalloproteinase

(MMP)-driven mechanisms of GBM dispersion in the brain.

194

8.2 Materials and methods

8.2.1 Cell culture

8.2.1.1 OSU-2 cell culture

OSU-2 cells procured from patient with GBM were cultured as described in detail in previous chapters (Chapter 5, 6, 7).

8.2.1.2 NIH 3T3 cell culture

NIH 3T3 cells (a standard fibroblast cell line) were procured from American Type

Culture Association (ATCC). Cells were cultured in Dulbecco’s Modified Eagle’s

Medium (DMEM) supplemented with 10% calf bovine serum (both from ATCC). Cells were passaged every 3 days and were used for experimentation following passaging.

8.2.2 Preparation and characterization of aligned electrospun PCL nanofibers

Aligned PCL electrospun fibers were prepared and characterized via scanning electron microscopy (SEM) as described previously in Chapter 7. The aligned fibers were deposited on an aluminum foil and sterilized before use in cell culture experiments as described in Chapter 7. All nanofibers were plasma treated as described in Chapter 7 before use.

8.2.3 Preparation of aligned electrospun nanofiber-hydrogel–cell composite

3D cultures

To prepare aligned electrospun nanofiber-hydrogel-tumor cell composite cultures, HA hydrogels and Col (I/III)-HA composite hydrogels were employed as described in

Chapter 6. Briefly, for HA-PCL gel-nanofiber composites, ~ 20 mg of thiolated HA (~ 2 wt%) was mixed with ~ 100 µL of OSU-2 cell culture medium and allowed to incubate 195 for ~ 15 min (Figure 82 A) following which an aligned PCL nanofiber mat was placed atop the gel and allowed to incubate for another ~ 45 min to form HA-PCL gel-nanofiber composites (See Figure 82 B). Further, to create composite sandwich constructs (i.e., gel- nanofiber-tumor cell-gel, See Figure 82 C), another 0.5 mg of thiolated HA (~ 0.5 wt%) was laid atop the gel-nanofiber-tumor cells constructs and allowed to incubate for another

~ 1h at 37 ºC, 5% CO2. For Col (I/III)-HA composite cultures, two types of sandwich configurations were examined. In the first configuration, 1 wt% base layer of HA hydrogel, followed by aligned PCL and Col-0.5HA composite and in the second configuration, aligned PCL was sandwiched between two Col-0.5HA composite hydrogels. In all cases, ~ 9000 OSU-2 cells were utilized to examine their interactions with composite biomaterials.

196

Figure 82. Schematic of hydrogel and hydrogel-fiber composites examined. (A) Hydrogels only (B) Aligned electrospun nanofibers (ENF) combined with hydrogels (C) (left) Aligned ENF “sandwiched” between two hydrogels, ENF not shown, (right) ENF embedded in the gel shown. Note hydrogel placed atop the aligned ENF is reduced in scale compared to the base hydrogel layer to show the presence of nanofibers.

To examine if attachment and alignment in typically non-adhesive gels (such as HA, agarose and PEG) used for a variety of tissue engineering applications was influenced by the presence of fibers, gel-nanofiber-cell composites (Figure 82 B) were formed. For HA-

PCL gel-nanofiber composites, ~ 10 mg of thiolated HA (~ 1 wt%) was mixed with ~

100 µL of NIH 3T3 cell culture medium and allowed to incubate for ~ 15 min, following which an aligned PCL nanofiber mat was placed atop the gel and allowed to incubate for another ~ 45 min to form HA-PCL gel-nanofiber composites. For agarose-PCL gel- nanofiber composites, ~1 wt% solution of SeaPrep- Agarose (Lonza, Rockland, ME) was

197 heated to dissolve the contents in a water bath and the agarose solution was poured into

96 well plates. The plates were then incubated at 4 ºC for ~ 15 min, following which an aligned PCL nanofiber mat was placed atop the gel and allowed to incubate for another ~

1 h to form agarose-PCL gel-nanofiber composites. For PEG-PCL gel-nanofiber composites, ~22 wt% poly (ethylene glycol)-dimethacrylate (PEG-DMA) (Polysciences,

Inc.) was polymerized partially under UV for ~ 15 min, following which an aligned PCL nanofiber mat was placed atop the gel and exposed to UV for another 15 min to form

PEG-PCL gel-nanofiber composites. All gel-nanofiber composites were washed with media and in all cases, ~ 9000 NIH 3T3 cells were seeded on each gel-nanofiber composite examined.

8.2.4 Imaging of cell-aligned nanofiber-hydrogel composite constructs using epi-fluorescence microscopy

Following incubation for ~ 24h, cell-nanofiber-hydrogel composites constructs were washed with appropriate cell culture media. They were then imaged using an epifluorescence microscope (Olympus IX 71) with appropriate fluorescent filters. Three- five images were acquired at random positions for each construct examined.

8.3 Results and discussion

Hydrogel-aligned electrospun nanofiber composite biomaterials were successfully created using several different combinations of 3D hydrogel constructs, demonstrating the wide applicability of this approach. These results are, however, preliminary, as well as qualitative and only serve to provide proof-of-concept demonstration of this methodology and the resulting cell behaviors observed.

198

To examine the utility of such systems as 3D brain tumor cell culture models, their interactions with patient derived OSU-2 cells were examined. Aligned nanofibers were used as mimetics of white matter tracts (Chapter 7) and hydrogel systems were used as mimics of the neural ECM (Chapter 6). This approach combined these biomaterials into a single construct. As reported in Chapter 6, OSU-2 cells adopted a rounded morphology in bare HA hydrogels (control). However, the addition of an aligned nanofiber component

(Figure 83), not only improved cell-material interaction, but also enhanced cell spreading, alignment and elongation with highly bipolar and elongated morphologies seen in the presence of aligned nanofiber constructs. This behavior was also retained when another

HA hydrogel layer was added atop the fiber layer (Figure 84).

199

Figure 83. SEM image of aligned PCL nanofibers. Scale bar indicates 50 µm.

Figure 84. OSU-2 cell behaviors on HA based gel-nanofiber composites (A) 2 wt % HA hydrogel (See Figure 82 A) (B) 2 wt % HA hydrogel with PCL nanofiber (See Figure 82 B) (C) PCL nanofiber “sandwiched” between 2 wt % HA base hydrogel and 0.5 wt % HA hydrogel on top (See Figure 82 C). Left insets show representative cell morphologies. Right insets in B & C, bidirectional arrows show direction of nanofiber alignment.

200

This preliminarily demonstrates that cell behaviors in 3D hydrogel systems can be enhanced by inclusion of nanofibers without the need for adding specific recognition sequences (i.e., RGD, YIGSR).

Similarly, “sandwich” like structures were also created using a combination of GAG- protein composite hydrogels (i.e., Col (I/III) - HA). In both sandwich combinations examined, cells underwent spreading and exhibited bipolar morphologies (Figure 85).

Future experiments will investigate migration behaviors of OSU-2 cells in such environments, thereby allowing examination of aligned migration in a 3D hydrogel system with the potential to investigate the effects of MMP secretion implicated in GBM migration. These results are in agreement with the work of Yang et al., wherein the incorporation of aligned PLA nanofibers into 3D collagen hydrogel constructs dictated the orientation of individual cell cytoskeleton (i.e., corneal fibroblasts and bovine nucleus pulposus (NP) cell types) [386]. Further, they also complement and extend the applicability of this approach to other synthetic and naturally derived biomaterial scaffolds as well as for applications in cancer biology.

201

Figure 85. OSU-2 cell behaviors on HA and Col-HA based gel-nanofiber composites (A) PCL nanofiber “sandwiched” between 1 wt% HA hydrogel base and Col-0.5HA composite hydrogel on top as described in Chapter 6 (B) PCL nanofiber “sandwiched” between Col-0.5HA composite hydrogels. Left insets show representative cell morphologies. Right insets (bidirectional arrows) show direction of nanofiber alignment.

To further investigate applicability in tissue engineering and regenerative medicine; some of the most commonly employed non-adhesive hydrogels (i.e., HA, PEG and Agarose) were combined with aligned nanofibers constructs. Similar to the general qualitative observations reported for tumor cells, aligned nanofiber-hydrogel composite constructs enhanced NIH 3T3 fibroblast cell adhesion compared to bare hydrogel controls (Figure

86). Further, composite constructs also promoted better spreading, elongation and alignment as opposed to the rounded cell morphologies seen in unmodified hydrogel controls, thereby demonstrating the ability of aligned cues in 3D hydrogel systems to direct cell behaviors. In general, several strategies are employed to improve the cell adhesion potential of non-adhesive biomaterials (reviewed in chapter 2). The addition of nanofibers to non-adhesive hydrogels provides a new strategy in tissue engineering to

202 improve cell adhesion with the further ability to control alignment through aligned nanofibers. Additionally, such an approach is extremely simple compared to other 3D patterning techniques for hydrogels (e.g., by using materials with specific chemistries).

Figure 86. NIH 3T3 cell behaviors on gel-nanofiber composites. (A) 1 wt% HA hydrogel. (B) 1 wt % Agarose hydrogel. (C) 22 wt% PEG-DMA hydrogel. (D) 1 wt % HA hydrogel-PCL nanofiber composite (E) 1 wt% Agarose hydrogel-PCL nanofiber composite (F) 22 wt% PEG- DMA hydrogel-PCL nanofiber composite. Right inset shows representative cell morphologies in each case. Note that cells on hydrogels are typically rounded. Cells on gel-nanofiber composites mostly displayed elongated bipolar morphologies. Bidirectional arrows (left inset, bottom row) show the direction of alignment of PCL nanofibers.

8.4 Conclusions

Hydrogel-aligned ENF composite biomaterial platforms were developed and shown to guide 3D cell behaviors. Detailed migration studies are needed to examine the dynamic interaction of tumor cells with such 3D materials and would further enable identification of crucial cues strongly favoring cell migration. In general, combination of such 203 biomaterial systems (i.e., hydrogels and electrospun nanofibers) can be highly advantageous. For example, different types of therapeutics could also be incorporated in these two systems individually thereby providing dual release in addition to the physical guidance cues in 3D systems. Also, this combination strategy lays the foundation to develop further complicated architectures with different features found in other tissues.

204

Chapter 9: Conclusions and Future Work

9.1 Conclusions

Whereas a majority of the literature has focused on studying cell behaviors in traditional two dimensional surfaces such as plastic and glass, it is becoming increasingly clear that such behaviors are far from those observed within the body. Three dimensional materials that faithfully capture features of the tissue microenvironment are therefore needed to enhance our understanding of individual cell behaviors. Such biomaterials are being developed, however are yet in their very early stages. This dissertation presents a variety of such models for applications in neural engineering and tumor cell migration studies enabling exploration of the complex relationships between micro-environmental cues and the resulting cell behaviors.

For applications in neural engineering, particularly enhancing the neuro-electrode interface, hydrogel based biomimetic coatings endowed with adhesion molecules were developed using synthetic PEG-based hydrogel systems. Polylysine was successfully conjugated to PEG-based hydrogels using photopolymerization reactions. These materials better promoted PC-12 cell adhesion over unmodified control hydrogels.

Additionally, adhesion behavior of such hydrogels to electrode surfaces showed that these coatings were stable for at least a month. It was also demonstrated that these

205 coatings could be patterned to minimize coating resistance and increase cell specific interactions with neural electrode systems.

In addition to brain mimetic coatings for neural electrodes, hydrogel-based, 3D biomimetic cell culture models were developed and employed to examine patient derived tumor cell behaviors. Using a 3D Matrigel model, we discovered, for the first time, that the underlying rigid support used to support a hydrogel also plays a significant role in influencing cell behaviors in a 3D hydrogel system and that this biomimetic system could be used as a model to study the role of interfacial gradients on cell behaviors. Utilizing both experimental and computational techniques, we showed that tumor cell morphology, spreading, elongation, migration potential, and cytoskeletal organization were all influenced in 3D hydrogels at regions close to the glass substrate-gel interface. Moreover, these cell behaviors were different from those seen farther away from the interfacial region. Most importantly, the mechanobiological response of tumor cells was demonstrated ex vivo in a single 3D hydrogel for the first time by exploiting an inherent cue in the system (i.e., edge effects).

Next, physiologically-relevant, 3D, biomimetic models mimicking features of the brain tumor microenvironment, consisting of collagen-HA composite (or interpenetrating network (IPN)) hydrogels were developed. Composite gels with increasing HA content were synthesized that mimicked the increasing levels of HA typically observed in GBM tumors. These composite hydrogels, in addition to simulating some chemical features of the tumor microenvironment, also displayed similar mechanical features to native tissue

(e.g., composite hydrogels displayed modulus ~ 300-2065 Pa, which extends across the

206 physiological reported values of 200-1000 Pa for normal brain tissue, values for cancer brain have not been determined conclusively). Using this system, tumor cell morphology, spreading, circularity, and migration were shown to be a strong function of HA concentration with in vivo spindle-like morphologies observed in lower HA concentration gels. Higher concentrations of HA (i.e., 2 wt%) completely abrogated tumor cell migration, demonstrating that migration could be tightly controlled by external cues and in this case, was strongly HA-density dependent in 3D. This work thus showed that the interplay of ECM components in 3D is crucial in dictating tumor cell behaviors.

In addition to hydrogel based models, aligned electrospun nanofiber models, closely mimicking brain white matter tracts (one of the most common routes for GBM tumor dissemination) were also developed and examined. In particular, by utilizing a core-shell spinning methodology, aligned ENF models with tunable mechanical (gelatin-PCL, PCL,

PDMS-PCL, PES-PCL) and biologically relevant-chemical features (PCL-Collagen,

PCL-HA, PCL-Matrigel) were developed. Tumor cells were extremely sensitive to nanofiber mechanics with cell morphology, elongation (feret diameter), migration, FAK, and MLC2 expression all being a strong function of nanofiber mechanics. Similarly, tumor cells were sensitive to chemistry, with HA chemistry significantly reducing migration potential. In contrast to most 3D models, where it is extremely difficult to decouple polymer properties to provide insights into mechanisms governing cell behaviors (e.g., porosity versus mechanics in 3D gels), aligned core-shell ENF model allowed modulation as well as near independent assessment of key variables (e.g., mechanics, chemistry) on tumor cell behaviors without significant confounding factors.

207

Finally, an integrated nanofiber-hydrogel model was developed to combine multiple functionalities into a single 3D system. This platform methodology could further enable evaluation of complex tumor cell behaviors (e.g., highly directed migration along white matter tracts, combined with high MMP activity needed for hydrogel degradation, in 3D contexts). Overall, this dissertation contributed a variety of novel, 3D, cell culture models possessing unique features for studying cell behaviors. In addition, tumor cell studies were performed using patient derived cells as opposed to cell lines used in majority of the studies. Changes in cell properties over repeated culture (i.e., for cell lines) have been reported and hence it is important to examine patient derived cells with minimal passages to accurately mirror the in vivo tumor cell response. Finally, dynamic evaluation techniques to monitor cell migration were used that clearly delineate migration and cell proliferation processes, which are a common issue in static assays or with spheroid models. Although this distinction is a fine one, it is extremely vital in the context of GBM tumors, as these tumors are diffusively migratory, infiltrating the normal brain tissue, in contrast to most others that grow via uncontrolled proliferation.

9.2 Future directions

9.2.1 Future directions on neural prosthesis/tissue interfaces

As mentioned in Chapter 4, adhesion molecules provide an attractive strategy for enhancing the tissue electrode-interfaces and promote chronic interfacing. Recapitulating the complex 3D neuronal microenvironment requires a range of different chemical/mechanical cues and hence it is necessary to combine multiple cues in a single

208 system to develop a robust interface (an example of such a system was demonstrated in

Chapter 8). The following section discusses this aspect in more detail.

9.2.1.1 Combining soluble factor release, tethered cues and topographical cues to develop multifunctional neural biomaterials for preliminary in vivo examination.

Our laboratory has developed techniques to release soluble factors from hydrogel as well as hydrogel-electrospun fiber biomaterials. Based on our preliminary results, it is anticipated that the soluble factors released from this coating can influence neurons several microns from the interface. The nerve cells then extend neurites in response to the released soluble cue. As these cells reach the interface, tethered cues present on the surface should enable favorable cell-material interactions to interact with electrode resulting in better interfacing. A schematic of the multifunctional coating is shown in

Figure 87. Several in vitro (e.g., using primary brain derived cells) and in vivo (e.g., implanting into rat brains) tests will be required to examine the feasibility of this multifunctional electrode coating. In vivo electrode performance could then be monitored using impedance spectroscopy (EIS), cyclic voltammetry (CV), and electrophysiological data over long term.

209

Figure 87. Schematic of the multifunctional neural biomimetic coating.

9.2.1.2 Developing and examining 3D tissue models of glial scarring

Current state of the art techniques for studying glial scarring processes in response to electrode implantation employ 2D substrates [387-389]. Astrocytes plated in 2D are typically “reactive” and hence it is difficult to differentiate the effects of a poorly compatible material from those of simple 2D culture. Our preliminary results using primary human (Chapter 6) as well as rat cortical astrocytes encapsulated in 3D gels show that these cells are primarily rounded (a morphology typical of the non-reactive phenotype), and hence these materials may serve as supports for investigations of the scarring response (e.g., transformation of astrocytes to reactive phenotype) in a straightforward and physiological 3D model.

9.2.2 Future directions on 3D tissue biomimetics for glioma migration

The models presented herein, could be used to further understand and answer several fundamental questions regarding glioblastoma multiformes and/or other cancers.

Refinement of these 3D models might also be required to further approximate the 210 complex 3D in vivo environment including major GBM migration routes (schematic representation shown in Figure 88).

Figure 88. Schematic of GBM invasion microenvironment including major highways. Figure taken from [156].

It is hoped that these 3D models mimicking several features of the native brain tissue microenvironment, will better predict in vivo outcomes when compared to standard 2D assays. Some of the future directions for this research have been summarized below:

211

9.2.2.1 Investigating influence of growth factors implicated in glioma invasion in 3D

Development of 3D models in this dissertation focused on components of the extracellular matrix or tethered cues. However, it is also known that soluble signal gradients (e.g., growth factors [390]) can influence migration pattern of gliomas.

Specifically, growth factors such as Vascular Endothelial Growth Factor (VEGF), Nerve

Growth Factor (NGF) and Interleukins (IL) have been reported to play important roles in glioblastoma migration [390]. Growth factors specifically bind to tumor cell receptors and activate associated signaling cascades that directly influence tumor cell adhesion, proliferation and migration. In future work, the influence of growth factors in 3D hydrogel systems should be investigated to examine their effects on tumor cell migration in 3D. An interesting aspect of this study is that growth factors will diffuse through a barrier (i.e., 3D hydrogel) to interact with tumor cells, in contrast to existing 2D studies where these factors have the least diffusion resistance.

9.2.2.2 Investigating effect of fluid flow on GBM behaviors in 3D using hydrogel based microfluidic platforms

It is believed that migration of glioblastomas is also influenced by flow of fluids (i.e., cerebrospinal fluid) in vivo [150]. For instance, it is possible that these cells are transported to different anatomical locations in the central nervous system by the flux of cerebrospinal fluid. To explore this possibility ex vivo, a micro fluidic platform could be developed that allows monitoring of the behavior of these cells to flow in 3D contexts.

The cell-encapsulated gels can be formed inside the fluidic chambers and fluid (e.g., similar to the cerebrospinal fluid (CSF)) can be passed through the chamber through

212 inlets (Figure 89). The entire set-up can be placed on a microscope stage to observe cell migration in situ in response to flow and compared against a control (without flow). A similar model incorporating aligned topographical features could also be constructed.

Figure 89. Schematic of incorporating fluid flow in ex vivo 3D systems.

9.2.2.3 Investigating the effect of electrical stimulation on the behavior of glioblastomas using aligned electrospun nanofiber platforms

Given that the unique microenvironment in the brain also presents strong electrical cues caused by neuronal firing, it is possible that migrating tumor cells contact nerve fibers with electrical conductivity. To mimic these features, aligned electrospun nanofibers could be either doped with conducting polymers (i.e., poly(pyrrole) similar to the core- shell nanofibers employed in this work (Chapter 7)) or prepared entirely using conducting polymers. Examples of such biomaterials for tissue engineering exist in the literature

(e.g., poly(pyrrole) coated poly(lactic-co-glycolic) acid (PLGA) electrospun nanofibers

[320, 391]), and should be easy to adapt to tumor cell migration studies.

213

9.2.2.4 Identification of specific pathways/signaling cascades and potential molecular targets for glioblastomas in 3D

We have identified and examined the influence of the key factors (e.g., stiffness, chemistry and topography) on the migration pattern of glioblastomas using 3D hydrogel and electrospun nanofiber based models. We have also examined the influence of FAK and MLC2 pathways in our electrospun nanofiber-based models. However, the picture is incomplete without a detailed fundamental understanding of several specific signaling cascades/pathways that are activated in response to these factors in 3D. Clearly, if we can further pin point the mechanisms that dictate tumor cell behavior, molecular targets that block these pathways (i.e., pathways that favor migration of tumors) could be identified.

Further, investigating cell response in 3D when these pathways are blocked using certain inhibitors should yield critical information. These will eventually enable design of better therapeutic agents as well as effective treatment methods against glioblastomas.

To further explore 3D hydrogel based models, gene expression profiling (similar to those performed by Bissell and colleagues [392]) should be performed. Detailed studies could reveal genes responsible for a particular cell behavior (e.g., morphology) in 3D (e.g., round vs. spread), different from those identified from 2D gene expression profiles.

These expression profiles can be used as a further guide to predict the performance of chemotherapeutic agents [393]. Most importantly, gene expression profiling of patient- derived cells plated on tissue mimetic models before being plated on standard 2D tissue culture polystyrene (TCPS) dishes should provide additional insight to novel genes expressed in response to tissue mimetic microenvironments. Routine culture of cells on

214

2D rigid TCPS can over time alter cell behaviors and lead to irreversible changes in cell properties.

9.2.2.5 Evaluation of different anti-cancer drug sensitivities in 3D

At present, efficacy of most drugs is determined using standard 2D cell culture assays

[394]. These methods fail to accurately predict the required dosage for in vivo conditions, thereby resulting in variable clinical outcomes. Development of robust 3D cell culture methods will dramatically improve our understanding of cancer cell sensitivities to these drugs and their associated toxicities, as well as reduce the costs involved in drug discovery and clinical trials. Potential new drugs or a combination of drugs can be tested in these 3D models by examining tumor cell response to different concentration levels of these drugs. Improvement of existing drug delivery methods (e.g., Carmustine- impregnated wafers (Gliadel wafers [395]) can also be achieved by utilizing our 3D models to better predict their release profiles and evaluate tumor cell response under in vitro conditions.

9.2.2.6 Detailed investigations on comparison of normal and tumor cell behaviors in

3D settings.

Whereas there have been studies that have focused on examining tumor cell behaviors, almost none have focused on a direct comparison of tumor and normal cells in 3D settings. Evaluation of dynamics of cell-material interactions utilizing both normal and tumor cells (e.g., investigating adhesion, spreading, migration, expression of integrins, cytoskeletal proteins) should yield crucial information on which specific properties are altered in tumor cells compared to normal cells. This could potentially enable

215 identification of several factors that favor tumor cell behaviors compared to normal cell behaviors. Additionally, valuable information could be obtained if both the normal and cancer cell was derived from a single source.

9.2.2.7 Clinical correlations of patient derived tumor cell migration using physiologically relevant in vitro models to predict survival outcomes.

Another important aspect of 3D tissue mimetic models includes their applicability in predicting survival outcomes and as tailored patient bio-assays to predict desired drug combinations. To investigate the efficacy of model prediction of survival outcomes, a large number of patient samples will be required to generate a graph of in vitro migration rates versus in vivo clinical data. The migration data can then be extrapolated with reasonable assumptions to predict survival time or tumor spread, thereby enabling neurosurgeons to tailor their treatment course.

Additionally, the most invasive tumor cells and stem cells share many features, a fact that has led to the hypothesis of “cancer stem cells” or “glioma stem cells” that are believed to closely mirror primary tumors [291, 396]. In addition to patient-derived cells, it is necessary to study the behaviors of glioma stem cells in such models and compare their behaviors to patient cells and traditional cell lines. Such detailed investigations will not only provide further confirmation for the cancer stem cell hypothesis, but also open up new avenues for targeted GBM therapies and their use as ex vivo tumor models.

216

9.2.2.8 Evaluation of neuronal/glia biology in 3D settings using hydrogel and/or electrospun fiber based platform biomaterials with physiological relevance.

In addition to the aforementioned applications, 3D models can also be employed to answer several fundamental, unexplored questions in neuronal and glia biology. For example, how do key factors such as stiffness, chemistry and topography influence oligodendrocyte myelination processes? Examining which of these attributes is crucial will dramatically improve our understanding of myelination as well as diseases of the

CNS white matter (e.g., Multiple Sclerosis, MS) thereby paving the way for development of targeted therapies.

9.2.2.9 Translating 3D biomimetic models to explore behaviors of other tumor cell types

3D biomimetic models discussed in this dissertation can also be employed to explore cancer cell behaviors of other tissues that share similar extracellular matrix molecules

(e.g., hyaluronic acid (HA) in cartilage [397]). Work in progress is exploring the possibility of using such models for expanding and studying properties of circulating tumor cells (CTCs) from several patient cancer tissue samples obtained using a negative depletion technology developed by Prof. Jeffery J. Chalmers (Chemical and

Biomolecular Engineering, OSU) [398-400]. Preliminary results show circulating cells enriched by this technology survive this process as evidenced using a standard Live-Dead

Assay (Figure 90).

217

Figure 90. Live/Dead staining of circulating tumor cells obtained using a negative depletion technology. (A) Calcein AM (Live, Green) staining and (B) EthD-1 (Dead, Red). Note the sample may also contain blood cell populations and this is not differentiated in this figure.

In sum, biomimetic 3D models offer an exciting opportunity to study a variety of biological processes in physiologically-relevant contexts with applications ranging from

“organ-on-a-chip” platforms [401] to drug discovery [402]. As leading biologist, Prof.

Mina Bissell points out in her review, “half of the secret of the cell lies outside the cell”

[403], 3D models will help unravel such secrets enabling a huge leap forward not only in regenerative medicine but also in battling cancer. Inarguably, these models are poised to make a dramatic impact in the coming years, offering hope to individuals affected by devastating diseases and enhancing their quality of life.

218

Bibliography

1. C. Zimmer. http://discovermagazine.com/2009/sep/19-dark-matter-of-the-human- brain. 2009.

2. E. Kandel, J. Schwartz, and T. Jessell, Principles of Neural Science. Fourth ed2000: McGraw Hill Companies, Inc.

3. N. Baumann and D. Pham-Dinh. Biology of oligodendrocyte and myelin in the mammalian central nervous system. Physiol Rev, 2001. 81(2): 871-927.

4. N.J. Abbott, L. Ronnback, and E. Hansson. Astrocyte-endothelial interactions at the blood-brain barrier. Nature Reviews Neuroscience, 2006. 7(1): 41-53.

5. E. Sykova. Diffusion properties of the brain in health and disease. Neurochem Int, 2004. 45(4): 453-66.

6. D.R. Zimmermann and M.T. Dours-Zimmermann. Extracellular matrix of the central nervous system: from neglect to challenge. Histochem Cell Biol, 2008. 130(4): 635-53.

7. M. Bani-Yaghoub, R.G. Tremblay, A. Ajji, M. Nzau, S. Gangaraju, D. Chitty, B. Zurakowski, and M. Sikorska. Neuroregenerative strategies in the brain: emerging significance of bone morphogenetic protein 7 (BMP7). Biochem Cell Biol, 2008. 86(5): 361-9.

8. R.L. Ruff, L. McKerracher, and M.E. Selzer. Repair and Neurorehabilitation Strategies for Spinal Cord Injury. Annals of the New York Academy of Sciences, 2008. 1142(The Year in Neurology 2008): 1-20.

9. C.E. Schmidt and J.B. Leach. Neural tissue engineering: strategies for repair and regeneration. Annu Rev Biomed Eng, 2003. 5: 293-347.

10. G. Mukhopadhyay, P. Doherty, F.S. Walsh, P.R. Crocker, and M.T. Filbin. A novel role for myelin-associated glycoprotein as an inhibitor of axonal regeneration. Neuron, 1994. 13(3): 757-67.

11. M.S. Chen, A.B. Huber, M.E. van der Haar, M. Frank, L. Schnell, A.A. Spillmann, F. Christ, and M.E. Schwab. Nogo-A is a myelin-associated neurite outgrowth inhibitor and an antigen for monoclonal antibody IN-1. Nature, 2000. 403(6768): 434-9. 219

12. D.A. Morgenstern, R.A. Asher, and J.W. Fawcett. Chondroitin sulphate proteoglycans in the CNS injury response. Prog Brain Res, 2002. 137: 313-32.

13. V.S. Polikov, P.A. Tresco, and W.M. Reichert. Response of brain tissue to chronically implanted neural electrodes. J Neurosci Methods, 2005. 148(1): 1-18.

14. R.J. McKeon, R.C. Schreiber, J.S. Rudge, and J. Silver. Reduction of neurite outgrowth in a model of glial scarring following CNS injury is correlated with the expression of inhibitory molecules on reactive astrocytes. J Neurosci, 1991. 11(11): 3398-411.

15. E.J. Tehovnik. Electrical stimulation of neural tissue to evoke behavioral responses. J Neurosci Methods, 1996. 65(1): 1-17.

16. G. Santhanam, S.I. Ryu, B.M. Yu, A. Afshar, and K.V. Shenoy. A high- performance brain-computer interface. Nature, 2006. 442(7099): 195-8.

17. J.C. Middlebrooks, J.A. Bierer, and R.L. Snyder. Cochlear implants: the view from the brain. Curr Opin Neurobiol, 2005. 15(4): 488-93.

18. J.O. Winter, S.F. Cogan, and J.F. Rizzo, 3rd. Neurotrophin-eluting hydrogel coatings for neural stimulating electrodes. J Biomed Mater Res B Appl Biomater, 2007. 81(2): 551-63.

19. W.M. Grill and J.T. Mortimer. Electrical properties of implant encapsulation tissue. Ann Biomed Eng, 1994. 22(1): 23-33.

20. N.A. Kotov, J.O. Winter, I.P. Clements, E. Jan, B.P. Timko, S. Campidelli, S. Pathak, A. Mazzatenta, C.M. Lieber, M. Prato, R.V. Bellamkonda, G.A. Silva, N.W.S. Kam, F. Patolsky, and L. Ballerini. Nanomaterials for Neural Interfaces. Advanced Materials, 2009. 21(40): 3970-4004.

21. D.H. Szarowski, M.D. Andersen, S. Retterer, A.J. Spence, M. Isaacson, H.G. Craighead, J.N. Turner, and W. Shain. Brain responses to micro-machined silicon devices. Brain Res, 2003. 983(1-2): 23-35.

22. D.J. Edell, V.V. Toi, V.M. McNeil, and L.D. Clark. Factors influencing the biocompatibility of insertable silicon microshafts in cerebral cortex. IEEE Trans Biomed Eng, 1992. 39(6): 635-43.

23. R. Biran, D.C. Martin, and P.A. Tresco. Neuronal cell loss accompanies the brain tissue response to chronically implanted silicon microelectrode arrays. Exp Neurol, 2005. 195(1): 115-26.

24. K. Katayama, K. Wada, A. Nakajima, Y. Kamisaki, and T. Mayumi. Nuclear receptors as targets for drug development: The role of nuclear receptors during 220

neural stem cell proliferation and differentiation. Journal of Pharmacological Sciences, 2005. 97(2): 171-176.

25. D. Giulian, M. Corpuz, S. Chapman, M. Mansouri, and C. Robertson. Reactive mononuclear phagocytes release neurotoxins after ischemic and traumatic injury to the central nervous system. J Neurosci Res, 1993. 36(6): 681-93.

26. D. Giulian, K. Vaca, and M. Corpuz. Brain Glia Release Factors with Opposing Actions Upon Neuronal Survival. Journal of Neuroscience, 1993. 13(1): 29-37.

27. M.A. Nicolelis. Brain-machine interfaces to restore motor function and probe neural circuits. Nat Rev Neurosci, 2003. 4(5): 417-22.

28. J.M. Carmena, M.A. Lebedev, R.E. Crist, J.E. O'Doherty, D.M. Santucci, D.F. Dimitrov, P.G. Patil, C.S. Henriquez, and M.A. Nicolelis. Learning to control a brain-machine interface for reaching and grasping by primates. PLoS Biol, 2003. 1(2): E42.

29. R.C. Kelly, M.A. Smith, J.M. Samonds, A. Kohn, A.B. Bonds, J.A. Movshon, and T.S. Lee. Comparison of recordings from microelectrode arrays and single electrodes in the visual cortex. J Neurosci, 2007. 27(2): 261-4.

30. P.J. Rousche and R.A. Normann. Chronic recording capability of the Utah Intracortical Electrode Array in cat sensory cortex. J Neurosci Methods, 1998. 82(1): 1-15.

31. K.A. Ludwig, J.D. Uram, J.Y. Yang, D.C. Martin, and D.R. Kipke. Chronic neural recordings using silicon microelectrode arrays electrochemically deposited with a poly(3,4-ethylenedioxythiophene) (PEDOT) film. J Neural Eng, 2006. 3(1): 59-70.

32. J.O. Winter, M. Gokhale, R.J. Jensen, S.F. Cogan, and J.F. Rizzo, 3rd. Tissue Engineering Applied to the Retinal Prosthesis: Neurotrophin-Eluting Polymeric Hydrogel Coatings. Mater Sci Eng C Mater Biol Appl, 2008. 28(3): 448-453.

33. J.O. Winter, N. Han, R. Jensen, S.F. Cogan, and J.F. Rizzo. Adhesion molecules promote chronic neural interfaces following neurotrophin withdrawal. Conf Proc IEEE Eng Med Biol Soc, 2009. 2009: 7151-4.

34. D.H. Kim, M. Abidian, and D.C. Martin. Conducting polymers grown in hydrogel scaffolds coated on neural prosthetic devices. J Biomed Mater Res A, 2004. 71(4): 577-85.

35. Y. Zhong, X. Yu, R. Gilbert, and R.V. Bellamkonda. Stabilizing electrode-host interfaces: a tissue engineering approach. J Rehabil Res Dev, 2001. 38(6): 627-32.

221

36. M.R. Abidian and D.C. Martin. Multifunctional Nanobiomaterials for Neural Interfaces. Adv Funct Mater, 2009. 19(4): 573-585.

37. X. Cui, V.A. Lee, Y. Raphael, J.A. Wiler, J.F. Hetke, D.J. Anderson, and D.C. Martin. Surface modification of neural recording electrodes with conducting polymer/biomolecule blends. J Biomed Mater Res, 2001. 56(2): 261-72.

38. X. Cui and D.C. Martin. Electrochemical deposition and characterisation of poly(3,4-ethylenedioxythiophene) on neural microelectrode arrays. Sensors and Actuators B:Chemical, 2003. 89(1-2): 92-102.

39. L. Poole-Warren, N. Lovell, S. Baek, and R. Green. Development of bioactive conducting polymers for neural interfaces. Expert Rev Med Devices, 2010. 7(1): 35-49.

40. D.H. Kim, S. Richardson-Burns, L. Povlich, M.R. Abidian, S. Spanninga, J.L. Hendricks, and D.C. Martin, Soft, Fuzzy, and Bioactive Conducting Polymers for Improving the Chronic Performance of Neural Prosthetic Devices, in Indwelling Neural Implants: Strategies for Contending with the In Vivo Environment, W.M. Reichert, Editor 2008: Boca Raton (FL).

41. W. He and R.V. Bellamkonda. Nanoscale neuro-integrative coatings for neural implants. Biomaterials, 2005. 26(16): 2983-90.

42. N.A. Peppas, J.Z. Hilt, A. Khademhosseini, and R. Langer. Hydrogels in Biology and Medicine: From Molecular Principles to Bionanotechnology. Advanced Materials, 2006. 18: 1345-1360.

43. A.S. Hoffman. Hydrogels for biomedical applications. Adv Drug Deliv Rev, 2002. 54(1): 3-12.

44. A. Kishino, N. Katayama, Y. Ishige, Y. Yamamoto, H. Ogo, T. Tatsuno, T. Mine, H. Noguchi, and C. Nakayama. Analysis of effects and pharmacokinetics of subcutaneously administered BDNF. Neuroreport, 2001. 12(5): 1067-72.

45. J.C. Buckner, P.D. Brown, B.P. O'Neill, F.B. Meyer, C.J. Wetmore, and J.H. Uhm. Central nervous system tumors. Mayo Clin Proc, 2007. 82(10): 1271-86.

46. http://www.cancer.org/Cancer/BrainCNSTumorsinAdults/DetailedGuide/brain- and-spinal-cord-tumors-in-adults-what-are-brain-spinal-tumors.

47. M.A. Hayat, Tumors of the Central Nervous System, Gliomas: Glioblastoma (Part 1). 1st ed. Vol. 1. 2011.

222

48. A. Sarkar and E.A. Chiocca, Glioblastoma and malignant astrocytoma, in Brain tumors, 3rd Edition: An encyclopedic approach., K.a. Laws, Editor 2011, Churchill Livingstone: Edinburgh, New York, USA.

49. A.C. Bellail, S.B. Hunter, D.J. Brat, C. Tan, and E.G. Van Meir. Microregional extracellular matrix heterogeneity in brain modulates glioma cell invasion. Int J Biochem Cell Biol, 2004. 36(6): 1046-69.

50. E.C. Holland. Glioblastoma multiforme: the terminator. Proc Natl Acad Sci U S A, 2000. 97(12): 6242-4.

51. E.G. Van Meir, C.G. Hadjipanayis, A.D. Norden, H.K. Shu, P.Y. Wen, and J.J. Olson. Exciting new advances in neuro-oncology: the avenue to a cure for malignant glioma. CA Cancer J Clin, 2010. 60(3): 166-93.

52. G. Dresemann. Temozolomide in malignant glioma. Onco Targets Ther, 2010. 3: 139-46.

53. S.H. Lin and L.R. Kleinberg. Carmustine wafers: localized delivery of chemotherapeutic agents in CNS malignancies. Expert Rev Anticancer Ther, 2008. 8(3): 343-59.

54. H.C. Lawson, P. Sampath, E. Bohan, M.C. Park, N. Hussain, A. Olivi, J. Weingart, L. Kleinberg, and H. Brem. Interstitial chemotherapy for malignant gliomas: the Johns Hopkins experience. J Neurooncol, 2007. 83(1): 61-70.

55. P.Y. Wen and S. Kesari. Malignant gliomas in adults. N Engl J Med, 2008. 359(5): 492-507.

56. S. Vukicevic, H.K. Kleinman, F.P. Luyten, A.B. Roberts, N.S. Roche, and A.H. Reddi. Identification of multiple active growth factors in basement membrane Matrigel suggests caution in interpretation of cellular activity related to extracellular matrix components. Exp Cell Res, 1992. 202(1): 1-8.

57. R. Langer and J.P. Vacanti. Tissue engineering. Science, 1993. 260(5110): 920- 926.

58. S.M. Willerth and S.E. Sakiyama-Elbert. Approaches to Neural Tissue Engineering Using Scaffolds for Drug Delivery. Advanced Drug Delivery Reviews, 2007. 59(4-5): 325-338.

59. C.S. Chen, M. Mrksich, S. Huang, G.M. Whitesides, and D.E. Ingber. Geometric control of cell life and death. Science, 1997. 276(5317): 1425-8.

223

60. C.S. Chen, J.L. Alonso, E. Ostuni, G.M. Whitesides, and D.E. Ingber. Cell shape provides global control of focal adhesion assembly. Biochem Biophys Res Commun, 2003. 307(2): 355-61.

61. R.G. LeBaron and K.A. Athanasiou. Extracellular matrix cell adhesion peptides: functional applications in orthopedic materials. Tissue Eng, 2000. 6(2): 85-103.

62. B. Alberts, D. Bray, B. Lewis, M. Raff, K. Roberts, and J.D. Watson, Molecular Biology of the Cell. Third Edition ed1996: Garland Publishing, Inc

63. C.S. Chen, J. Tan, and J. Tien. Mechanotransduction at cell-matrix and cell-cell contacts. Annu Rev Biomed Eng, 2004. 6: 275-302.

64. L.M.Y. Yu, N.D. Leipzig, and M.S. Shoichet. Promoting neuron adhesion and growth. Materials Today, 2008. 11(5): 36-43.

65. K.A. Venstrom and L.F. Reichardt. Extracellular matrix. 2: Role of extracellular matrix molecules and their receptors in the nervous system. Faseb J, 1993. 7(11): 996-1003.

66. N. Alobaid, H.J. Salacinski, K.M. Sales, G. Hamilton, and A.M. Seifalian. Single stage cell seeding of small diameter prosthetic cardiovascular grafts. Clin Hemorheol Microcirc, 2005. 33(3): 209-26.

67. R.O. Hynes. Cell adhesion: old and new questions. Trends Cell Biol, 1999. 9(12): M33-7.

68. L. Bacakova, E. Filova, F. Rypacek, V. Svorcik, and V. Stary. Cell adhesion on artificial materials for tissue engineering. Physiol Res, 2004. 53 Suppl 1: S35-45.

69. M.D. Pierschbacher and E. Ruoslahti. Cell attachment activity of fibronectin can be duplicated by small synthetic fragments of the molecule. Nature, 1984. 309(5963): 30-3.

70. J. Graf, R.C. Ogle, F.A. Robey, M. Sasaki, G.R. Martin, Y. Yamada, and H.K. Kleinman. A pentapeptide from the laminin B1 chain mediates cell adhesion and binds the 67,000 laminin receptor. Biochemistry, 1987. 26(22): 6896-900.

71. U. Hersel, C. Dahmen, and H. Kessler. RGD modified polymers: biomaterials for stimulated cell adhesion and beyond. Biomaterials, 2003. 24(24): 4385-415.

72. M. Aumailley and B. Gayraud. Structure and biological activity of the extracellular matrix. J Mol Med, 1998. 76(3-4): 253-65.

224

73. Y.W. Tong and M.S. Shoichet. Enhancing the neuronal interaction on fluoropolymer surfaces with mixed peptides or spacer group linkers. Biomaterials, 2001. 22(10): 1029-34.

74. T.T. Yu and M.S. Shoichet. Guided cell adhesion and outgrowth in peptide- modified channels for neural tissue engineering. Biomaterials, 2005. 26(13): 1507-14.

75. P. Liesi, A. Narvanen, J. Soos, H. Sariola, and G. Snounou. Identification of a neurite outgrowth-promoting domain of laminin using synthetic peptides. FEBS Lett, 1989. 244(1): 141-8.

76. A.P. Skubitz, J.B. McCarthy, Q. Zhao, X.Y. Yi, and L.T. Furcht. Definition of a sequence, RYVVLPR, within laminin peptide F-9 that mediates metastatic fibrosarcoma cell adhesion and spreading. Cancer Res, 1990. 50(23): 7612-22.

77. K. Gelse, E. Poschl, and T. Aigner. Collagens--structure, function, and biosynthesis. Adv Drug Deliv Rev, 2003. 55(12): 1531-46.

78. W.F. Vogel. Collagen-receptor signaling in health and disease. Eur J Dermatol, 2001. 11(6): 506-14.

79. M.J. Blewitt and R.K. Willits. The effect of soluble peptide sequences on neurite extension on 2D collagen substrates and within 3D collagen gels. Ann Biomed Eng, 2007. 35(12): 2159-67.

80. E. Yavin and Z. Yavin. Attachment and culture of dissociated cells from rat embryo cerebral hemispheres on polylysine-coated surface. J Cell Biol, 1974. 62(2): 540-6.

81. D.A. Lauffenburger and J.L. Linderman, Receptor-Mediated Cell Behavioral Responses, in Receptors: Models for Binding, Trafficking, and Signaling., D.A. Lauffenburger and J.L. Linderman, Editors. 1996, Oxford University Press, Inc. p. 236-344

82. E.M. Harnett, J. Alderman, and T. Wood. The surface energy of various biomaterials coated with adhesion molecules used in cell culture. Colloids Surf B Biointerfaces, 2007. 55(1): 90-7.

83. G. Rainaldi, A. Calcabrini, and M.T. Santini. Positively charged polymer polylysine-induced cell adhesion molecule redistribution in K562 cells. J Mater Sci Mater Med, 1998. 9(12): 755-60.

225

84. D. Heggins and G. Banker, Primary Dissociated Cell Cultures, in Culturing Nerve Cells, G. Banker and K. Goslin, Editors. 1998, MIT Press. p. 37-78.

85. A. Hategan, K. Sengupta, S. Kahn, E. Sackmann, and D.E. Discher. Topographical pattern dynamics in passive adhesion of cell membranes. Biophys J, 2004. 87(5): 3547-60.

86. Park K.S., Kim S.M., Kim M.S., Lee I., Rhee J.M., Lee H.B., and Khang G. Effect of Cell-Adhesive-Molecule-Coated Poly(lactide-co-glycolide) Film on the Cellular Behaviors of Olfactory Ensheathing Cells and Schwann Cells. Journal of Applied Polymer Science, 2008. 107: 1243-1251.

87. Y. Haile, S. Berski, G. Drager, A. Nobre, K. Stummeyer, R. Gerardy-Schahn, and C. Grothe. The effect of modified polysialic acid based hydrogels on the adhesion and viability of primary neurons and glial cells. Biomaterials, 2008. 29(12): 1880- 91.

88. Y.C. Huang, C.C. Huang, Y.Y. Huang, and K.S. Chen. Surface modification and characterization of chitosan or PLGA membrane with laminin by chemical and oxygen plasma treatment for neural regeneration. J Biomed Mater Res A, 2007. 82(4): 842-51.

89. H.S. Koh, T. Yong, C.K. Chan, and S. Ramakrishna. Enhancement of neurite outgrowth using nano-structured scaffolds coupled with laminin. Biomaterials, 2008. 29(26): 3574-82.

90. M. Cheng, W. Cao, Y. Gao, Y. Gong, N. Zhao, and X. Zhang. Studies on nerve cell affinity of biodegradable modified chitosan films. J Biomater Sci Polym Ed, 2003. 14(10): 1155-67.

91. C. Mingyu, G. Kai, L. Jiamou, G. Yandao, Z. Nanming, and Z. Xiufang. Surface modification and characterization of chitosan film blended with poly-L-lysine. J Biomater Appl, 2004. 19(1): 59-75.

92. T. Nakaji-Hirabayashi, K. Kato, and H. Iwata. Self-assembling chimeric protein for the construction of biodegradable hydrogels capable of interaction with integrins expressed on neural stem/progenitor cells. Biomacromolecules, 2008. 9(5): 1411-6.

93. M.J. Mahoney and K.S. Anseth. Contrasting effects of collagen and bFGF-2 on neural cell function in degradable synthetic PEG hydrogels. Journal of Biomedical Materials Research Part A, 2007. 81(2): 269-278.

94. C.S. Peyratout and L. Dahne. Tailor-made polyelectrolyte microcapsules: from multilayers to smart containers. Angew Chem Int Ed Engl, 2004. 43(29): 3762-83.

226

95. J.L. Lutkenhaus and P.T. Hammond. Electrochemically enabled polyelectrolyte multilayer devices: from fuel cells to sensors. Soft Matter, 2007. 3: 804-816.

96. Z. Tang, Y. Wang, P. Podsiadlo, and N.A. Kotov. Biomedical Applications of Layer-by-Layer Assembly:From Biomimetics to Tissue Engineering. Advanced Materials, 2006: 3203-3224.

97. Z.R. Wu, J. Ma, B.F. Liu, Q.Y. Xu, and F.Z. Cui. Layer-by-Layer assembly of polyelectrolyte films improving cytocompatibility to neural cells. Journal of Biomedical Materials Research Part A, 2007. 81A(2): 355-362.

98. S. Young, M. Wong, Y. Tabata, and A.G. Mikos. Gelatin as a delivery vehicle for the controlled release of bioactive molecules. J Control Release, 2005. 109(1-3): 256-74.

99. W. He, G.C. McConnell, and R.V. Bellamkonda. Nanoscale laminin coating modulates cortical scarring response around implanted silicon microelectrode arrays. J Neural Eng, 2006. 3(4): 316-26.

100. J. Poppe, Gelatin, in Thickening and Gelling Agents for Food, A. Imeson, Editor 1997, Springer. p. 144-168.

101. C.E. Schmidt, V.R. Shastri, J.P. Vacanti, and R. Langer. Stimulation of neurite outgrowth using an electrically conducting polymer. Proc Natl Acad Sci U S A, 1997. 94(17): 8948-53.

102. Y. Li, K.G. Neoh, and E.T. Kang. Controlled release of heparin from polypyrrole- poly(vinyl alcohol) assembly by electrical stimulation. J Biomed Mater Res A, 2005. 73(2): 171-81.

103. W.R. Stauffer and X.T. Cui. Polypyrrole doped with 2 peptide sequences from laminin. Biomaterials, 2006. 27(11): 2405-13.

104. X. Cui, J. Wiler, M. Dzaman, R.A. Altschuler, and D.C. Martin. In vivo studies of polypyrrole/peptide coated neural probes. Biomaterials, 2003. 24(5): 777-87.

105. P.M. George, A.W. Lyckman, D.A. LaVan, A. Hegde, Y. Leung, R. Avasare, C. Testa, P.M. Alexander, R. Langer, and M. Sur. Fabrication and biocompatibility of polypyrrole implants suitable for neural prosthetics. Biomaterials, 2005. 26(17): 3511-9.

106. G.M. Spinks, B. Xi, V.-T. Troung, and G.G. Wallace. Actuation behaviour of layered composites of polyaniline,carbon nanotubes and polypyrrole. Synthetic Metals, 2005. 151: 85-91.

227

107. G.M. Spinks, T.E. Campbell, and G.G. Wallace. Force generation from polypyrrole actuators. Smart Materials Structure, 2005. 14: 406-412.

108. M.R. Gandhi, P. Murray, G.M. Spinks, and G.G. Wallace. Mechanism of electrochemical actuation in polypyrrole. Synthetic Metals, 1995. 73: 247-256.

109. N.K. Guimard, N. Gomez, and C.E. Schmidt. Conducting polymers in biomedical engineering. Progress in Polymer Science, 2007. 32: 876-921.

110. Y. Xiao, D.C. Martin, X. Cui, and M. Shenai. Surface modification of neural probes with conducting polymer poly(hydroxymethylated-3,4- ethylenedioxythiophene) and its biocompatibility. Appl Biochem Biotechnol, 2006. 128(2): 117-30.

111. H.K. Song, B. Toste, K. Ahmann, D. Hoffman-Kim, and G.T. Palmore. Micropatterns of positive guidance cues anchored to polypyrrole doped with polyglutamic acid: a new platform for characterizing neurite extension in complex environments. Biomaterials, 2006. 27(3): 473-84.

112. G.T. Hermanson, Bioconjugate Techniques1996: Academic Press.

113. B.D. Ratner, A.S. Hoffman, F.J. Schoen, and J.E. Lemons, Biomaterials Science: An Introduction to Materials in Medicine. Second ed2004: Academic Press. 851.

114. L.M. Yu, K. Kazazian, and M.S. Shoichet. Peptide surface modification of methacrylamide chitosan for neural tissue engineering applications. J Biomed Mater Res A, 2007. 82(1): 243-55.

115. S.G. Levesque and M.S. Shoichet. Synthesis of cell-adhesive dextran hydrogels and macroporous scaffolds. Biomaterials, 2006. 27(30): 5277-85.

116. D. Shaw and M.S. Shoichet. Toward spinal cord injury repair strategies: peptide surface modification of expanded poly(tetrafluoroethylene) fibers for guided neurite outgrowth in vitro. J Craniofac Surg, 2003. 14(3): 308-16.

117. R. Bellamkonda, J.P. Ranieri, and P. Aebischer. Laminin oligopeptide derivatized agarose gels allow three-dimensional neurite extension in vitro. J Neurosci Res, 1995. 41(4): 501-9.

118. S. Hou, Q. Xu, W. Tian, F. Cui, Q. Cai, J. Ma, and I.S. Lee. The repair of brain lesion by implantation of hyaluronic acid hydrogels modified with laminin. J Neurosci Methods, 2005. 148(1): 60-70.

119. S.P. Massia, M.M. Holecko, and G.R. Ehteshami. In vitro assessment of bioactive coatings for neural implant applications. J Biomed Mater Res A, 2004. 68(1): 177- 86. 228

120. S.E. Stabenfeldt, A.J. Garcia, and M.C. LaPlaca. Thermoreversible laminin- functionalized hydrogel for neural tissue engineering. J Biomed Mater Res A, 2006. 77(4): 718-25.

121. M. Suzuki, S. Itoh, I. Yamaguchi, K. Takakuda, H. Kobayashi, K. Shinomiya, and J. Tanaka. Tendon chitosan tubes covalently coupled with synthesized laminin peptides facilitate nerve regeneration in vivo. J Neurosci Res, 2003. 72(5): 646- 59.

122. W.M. Tian, S.P. Hou, J. Ma, C.L. Zhang, Q.Y. Xu, I.S. Lee, H.D. Li, M. Spector, and F.Z. Cui. Hyaluronic acid-poly-D-lysine-based three-dimensional hydrogel for traumatic brain injury. Tissue Eng, 2005. 11(3-4): 513-25.

123. N.O. Dhoot, C.A. Tobias, I. Fischer, and M.A. Wheatley. Peptide-modified alginate surfaces as a growth permissive substrate for neurite outgrowth. J Biomed Mater Res A, 2004. 71(2): 191-200.

124. L. Kam, W. Shain, J.N. Turner, and R. Bizios. Selective adhesion of astrocytes to surfaces modified with immobilized peptides. Biomaterials, 2002. 23(2): 511-5.

125. M. Huber, P. Heiduschka, S. Kienle, C. Pavlidis, J. Mack, T. Walk, G. Jung, and S. Thanos. Modification of glassy carbon surfaces with synthetic laminin-derived peptides for nerve cell attachment and neurite growth. J Biomed Mater Res, 1998. 41(2): 278-88.

126. Y.W. Tong and M.S. Shoichet. Peptide surface modification of poly(tetrafluoroethylene-co-hexafluoropropylene) enhances its interaction with central nervous system neurons. J Biomed Mater Res, 1998. 42(1): 85-95.

127. Y.W. Tong and M.S. Shoichet. Enhancing the interaction of central nervous system neurons with poly(tetrafluoroethylene-co-hexafluoropropylene) via a novel surface amine-functionalization reaction followed by peptide modification. J Biomater Sci Polym Ed, 1998. 9(7): 713-29.

128. M. Borkenhagen, J.F. Clemence, H. Sigrist, and P. Aebischer. Three-dimensional extracellular matrix engineering in the nervous system. J Biomed Mater Res, 1998. 40(3): 392-400.

129. K.E. Crompton, J.D. Goud, R.V. Bellamkonda, T.R. Gengenbach, D.I. Finkelstein, M.K. Horne, and J.S. Forsythe. Polylysine-functionalised thermoresponsive chitosan hydrogel for neural tissue engineering. Biomaterials, 2007. 28(3): 441-449.

229

130. D.K. Cullen, M.C. Lessing, and M.C. LaPlaca. Collagen-dependent neurite outgrowth and response to dynamic deformation in three-dimensional neuronal cultures. Ann Biomed Eng, 2007. 35(5): 835-46.

131. J.C. Schense, J. Bloch, P. Aebischer, and J.A. Hubbell. Enzymatic incorporation of bioactive peptides into fibrin matrices enhances neurite extension. Nat Biotechnol, 2000. 18(4): 415-9.

132. G.M. Whitesides, E. Ostuni, S. Takayama, X. Jiang, and D.E. Ingber. Soft lithography in biology and biochemistry. Annu Rev Biomed Eng, 2001. 3: 335-73.

133. J.O. Winter, Nanoparticles and Nanowires for Cellular Engineering, in Tissue, Cell and Organ Engineering, C.S. Kumar, Editor 2006, Wiley VHC: Weinhem. p. 388-460

134. R. Zaidel-Bar, M. Cohen, L. Addadi, and B. Geiger. Hierarchical assembly of cell-matrix adhesion complexes. Biochem Soc Trans, 2004. 32(Pt3): 416-20.

135. J.P. Ranieri, R. Bellamkonda, E.J. Bekos, J.A. Gardella, Jr., H.J. Mathieu, L. Ruiz, and P. Aebischer. Spatial control of neuronal cell attachment and differentiation on covalently patterned laminin oligopeptide substrates. Int J Dev Neurosci, 1994. 12(8): 725-35.

136. S. Saneinejad and M.S. Shoichet. Patterned glass surfaces direct cell adhesion and process outgrowth of primary neurons of the central nervous system. J Biomed Mater Res, 1998. 42(1): 13-9.

137. S. Saneinejad and M.S. Shoichet. Patterned poly(chlorotrifluoroethylene) guides primary nerve cell adhesion and neurite outgrowth. J Biomed Mater Res, 2000. 50(4): 465-74.

138. N. Patel, R. Padera, G.H. Sanders, S.M. Cannizzaro, M.C. Davies, R. Langer, C.J. Roberts, S.J. Tendler, P.M. Williams, and K.M. Shakesheff. Spatially controlled cell engineering on biodegradable polymer surfaces. Faseb J, 1998. 12(14): 1447- 54.

139. C. Miller, H. Shanks, A. Witt, G. Rutkowski, and S. Mallapragada. Oriented Schwann cell growth on micropatterned biodegradable polymer substrates. Biomaterials, 2001. 22(11): 1263-9.

140. C. Miller, S. Jeftinija, and S. Mallapragada. Micropatterned Schwann cell-seeded biodegradable polymer substrates significantly enhance neurite alignment and outgrowth. Tissue Eng, 2001. 7(6): 705-15.

230

141. M.R. Hynd, J.P. Frampton, N. Dowell-Mesfin, J.N. Turner, and W. Shain. Directed cell growth on protein-functionalized hydrogel surfaces. J Neurosci Methods, 2007. 162(1-2): 255-63.

142. M.S. Hahn, J.S. Miller, and J.L. West. Three-dimensional biochemical and biomechanical patterning of hydrogels for guiding cell behavior. Advanced Materials, 2006. 18(20): 2679-2684

143. Y. Luo and M.S. Shoichet. A photolabile hydrogel for guided three-dimensional cell growth and migration. Nat Mater, 2004. 3(4): 249-53.

144. P. Musoke-Zawedde and M.S. Shoichet. Anisotropic three-dimensional peptide channels guide neurite outgrowth within a biodegradable hydrogel matrix. Biomed Mater, 2006. 1(3): 162-9.

145. M. Nakada, S. Nakada, T. Demuth, N.L. Tran, D.B. Hoelzinger, and M.E. Berens. Molecular targets of glioma invasion. Cell Mol Life Sci, 2007. 64(4): 458-78.

146. W.J. Thoman, M. Ammirati, L.P. Caragine, Jr., J.M. McGregor, A. Sarkar, and E.A. Chiocca. Brain tumor imaging and surgical management: the neurosurgeon's perspective. Top Magn Reson Imaging, 2006. 17(2): 121-6.

147. J.W. Kernohan, R.F. Mabon, and et al. A simplified classification of the gliomas. Mayo Clin Proc, 1949. 24(3): 71-5.

148. R.H. Goldbrunner, J.J. Bernstein, and J.C. Tonn. Cell-extracellular matrix interaction in glioma invasion. Acta Neurochir (Wien), 1999. 141(3): 295-305; discussion 304-5.

149. T. Demuth and M.E. Berens. Molecular mechanisms of glioma cell migration and invasion. J Neurooncol, 2004. 70(2): 217-28.

150. A. Giese and M. Westphal. Glioma invasion in the central nervous system. Neurosurgery, 1996. 39(2): 235-50; discussion 250-2.

151. A. Sarkar and E.A. Chiocca. Multiple craniotomies. J Neurosurg, 2011. 114(3): 574-5; discussion 575.

152. M.S. Viapiano and S.E. Lawler, Glioma Invasion: Mechanisms and Therapeutic Challenges, in CNS Cancer: Models, Markers, Prognostic Factors, Targets, and Therapeutic Approaches, E.G. Van Meir, Editor 2009, Humana Press (Springer): New York. p. 1219-1252.

231

153. E. Anderson, R. Grant, S.C. Lewis, and I.R. Whittle. Randomized Phase III controlled trials of therapy in malignant glioma: where are we after 40 years? Br J Neurosurg, 2008. 22(3): 339-49.

154. M.L. Salgaller and L.M. Liau. Current status of clinical trials for glioblastoma. Rev Recent Clin Trials, 2006. 1(3): 265-81.

155. A. Claes, A.J. Idema, and P. Wesseling. Diffuse glioma growth: a guerilla war. Acta Neuropathol, 2007. 114(5): 443-58.

156. D.N. Louis. Molecular pathology of malignant gliomas. Annu Rev Pathol, 2006. 1: 97-117.

157. M.L. Rosenblum, M.E. Berens, and J.T. Rutka. Recent perspectives in brain tumor biology and treatment. Clin Neurosurg, 1989. 35: 314-35.

158. S. Huang and D.E. Ingber. A non-genetic basis for cancer progression and metastasis: self-organizing attractors in cell regulatory networks. Breast Dis, 2006. 26: 27-54.

159. C.L. Gladson. The extracellular matrix of gliomas: modulation of cell function. J Neuropathol Exp Neurol, 1999. 58(10): 1029-40.

160. H.B. Newton. Molecular neuro-oncology and the development of targeted therapeutic strategies for brain tumors. Part 3: brain tumor invasiveness. Expert Rev Anticancer Ther, 2004. 4(5): 803-21.

161. A. Bertolotto, M.L. Magrassi, L. Orsi, C. Sitia, and D. Schiffer. Glycosaminoglycan changes in human gliomas. A biochemical study. J Neurooncol, 1986. 4(1): 43-8.

162. M. Scholz, V. Noack, I. Pechlivanis, M. Engelhardt, B. Fricke, U. Linstedt, B. Brendel, D. Ing, K. Schmieder, H. Ermert, and A. Harders. Vibrography during tumor neurosurgery. J Ultrasound Med, 2005. 24(7): 985-92.

163. L. Xu, Y. Lin, J.C. Han, Z.N. Xi, H. Shen, and P.Y. Gao. Magnetic resonance elastography of brain tumors: preliminary results. Acta Radiol, 2007. 48(3): 327- 30.

164. G. Unsgaard, O.M. Rygh, T. Selbekk, T.B. Muller, F. Kolstad, F. Lindseth, and T.A. Hernes. Intra-operative 3D ultrasound in neurosurgery. Acta Neurochir (Wien), 2006. 148(3): 235-53; discussion 253.

165. T. Selbekk, J. Bang, and G. Unsgaard. Strain processing of intraoperative ultrasound images of brain tumours: initial results. Ultrasound Med Biol, 2005. 31(1): 45-51. 232

166. Y. Bakiri, R. Karadottir, L. Cossell, and D. Attwell. Morphological and electrical properties of oligodendrocytes in the white matter of the corpus callosum and cerebellum. J Physiol, 2011. 589(Pt 3): 559-73.

167. D.L. Sherman and P.J. Brophy. Mechanisms of axon ensheathment and myelin growth. Nat Rev Neurosci, 2005. 6(9): 683-90.

168. Y. Benninger, H. Colognato, T. Thurnherr, R.J. Franklin, D.P. Leone, S. Atanasoski, K.A. Nave, C. Ffrench-Constant, U. Suter, and J.B. Relvas. Beta1- integrin signaling mediates premyelinating oligodendrocyte survival but is not required for CNS myelination and remyelination. J Neurosci, 2006. 26(29): 7665- 73.

169. M. Makino, K. Mimatsu, H. Saito, N. Konishi, and Y. Hashizume. Morphometric study of myelinated fibers in human cervical spinal cord white matter. Spine (Phila Pa 1976), 1996. 21(9): 1010-6.

170. F. Aboitiz, A.B. Scheibel, R.S. Fisher, and E. Zaidel. Fiber composition of the human corpus callosum. Brain Res, 1992. 598(1-2): 143-53.

171. S. Jung, C. Ackerley, S. Ivanchuk, S. Mondal, L.E. Becker, and J.T. Rutka. Tracking the invasiveness of human astrocytoma cells by using green fluorescent protein in an organotypical brain slice model. J Neurosurg, 2001. 94(1): 80-9.

172. J.S. Rao. Molecular mechanisms of glioma invasiveness: the role of proteases. Nat Rev Cancer, 2003. 3(7): 489-501.

173. J. Johnson, M.O. Nowicki, C.H. Lee, E.A. Chiocca, M.S. Viapiano, S.E. Lawler, and J.J. Lannutti. Quantitative analysis of complex glioma cell migration on electrospun polycaprolactone using time-lapse microscopy. Tissue Eng Part C Methods, 2009. 15(4): 531-40.

174. P.A. Agudelo-Garcia, J.K. De Jesus, S.P. Williams, M.O. Nowicki, E.A. Chiocca, S. Liyanarachchi, P.K. Li, J.J. Lannutti, J.K. Johnson, S.E. Lawler, and M.S. Viapiano. Glioma cell migration on three-dimensional nanofiber scaffolds is regulated by substrate topography and abolished by inhibition of STAT3 signaling. Neoplasia, 2011. 13(9): 831-40.

175. A. Valster, N.L. Tran, M. Nakada, M.E. Berens, A.Y. Chan, and M. Symons. Cell migration and invasion assays. Methods, 2005. 37(2): 208-15.

176. A. Buxboim, K. Rajagopal, A.E. Brown, and D.E. Discher. How deeply cells feel: methods for thin gels. J Phys Condens Matter, 2010. 22(19): 194116.

233

177. M.E. Berens, M.D. Rief, M.A. Loo, and A. Giese. The role of extracellular matrix in human astrocytoma migration and proliferation studied in a microliter scale assay. Clin Exp Metastasis, 1994. 12(6): 405-15.

178. A. Giese, M.A. Loo, M.D. Rief, N. Tran, and M.E. Berens. Substrates for astrocytoma invasion. Neurosurgery, 1995. 37(2): 294-301; discussion 301-2.

179. A. Giese, M.D. Rief, M.A. Loo, and M.E. Berens. Determinants of human astrocytoma migration. Cancer Res, 1994. 54(14): 3897-904.

180. A. Giese, M.A. Loo, N. Tran, D. Haskett, S.W. Coons, and M.E. Berens. Dichotomy of astrocytoma migration and proliferation. Int J Cancer, 1996. 67(2): 275-82.

181. A. Giese, M.A. Loo, S.A. Norman, S. Treasurywala, and M.E. Berens. Contrasting migratory response of astrocytoma cells to tenascin mediated by different integrins. J Cell Sci, 1996. 109 ( Pt 8): 2161-8.

182. A. Giese, L. Kluwe, B. Laube, H. Meissner, M.E. Berens, and M. Westphal. Migration of Human Glioma Cells on Myelin. Neurosurgery, 1996. 38(4): 755- 764.

183. V.R. Amberger, T. Hensel, N. Ogata, and M.E. Schwab. Spreading and migration of human glioma and rat C6 cells on central nervous system myelin in vitro is correlated with tumor malignancy and involves a metalloproteolytic activity. Cancer Res, 1998. 58(1): 149-58.

184. V.M. Weaver, A.R. Howlett, B. Langton-Webster, O.W. Petersen, and M.J. Bissell. The development of a functionally relevant cell culture model of progressive human breast cancer. Semin Cancer Biol, 1995. 6(3): 175-84.

185. A.P. Amar, S.J. DeArmond, D.R. Spencer, P.F. Coopersmith, D.M. Ramos, and M.L. Rosenblum. Development of an in vitro extracellular matrix assay for studies of brain tumor cell invasion. J Neurooncol, 1994. 20(1): 1-15.

186. B. Radotra and D. McCormick. Glioma invasion in vitro is mediated by CD44- hyaluronan interactions. J Pathol, 1997. 181(4): 434-8.

187. T. Nakagawa, T. Kubota, M. Kabuto, and T. Kodera. Hyaluronic acid facilitates glioma cell invasion in vitro. Anticancer Res, 1996. 16(5A): 2917-22.

188. S.K. Chintala, Z.L. Gokaslan, Y. Go, R. Sawaya, G.L. Nicolson, and J.S. Rao. Role of extracellular matrix proteins in regulation of human glioma cell invasion in vitro. Clin Exp Metastasis, 1996. 14(4): 358-66.

234

189. S. Koochekpour, G.J. Pilkington, and A. Merzak. Hyaluronic acid/CD44H interaction induces cell detachment and stimulates migration and invasion of human glioma cells in vitro. Int J Cancer, 1995. 63(3): 450-4.

190. S.S. Rao, S. Bentil, J. Dejesus, J. Larison, A. Hissong, R. Dupaix, A. Sarkar, and J.O. Winter. Inherent Interfacial Mechanical Gradients in 3D Hydrogels Influence Tumor Cell Behaviors. PLoS One, 2012. 7(4): e35852.

191. E. Cukierman, R. Pankov, D.R. Stevens, and K.M. Yamada. Taking cell-matrix adhesions to the third dimension. Science, 2001. 294(5547): 1708-12.

192. D.J. Webb and A.F. Horwitz. New dimensions in cell migration. Nat Cell Biol, 2003. 5(8): 690-2.

193. T. Ohnishi, H. Matsumura, S. Izumoto, S. Hiraga, and T. Hayakawa. A novel model of glioma cell invasion using organotypic brain slice culture. Cancer Res, 1998. 58(14): 2935-40.

194. M. Nakada, J.A. Niska, H. Miyamori, W.S. McDonough, J. Wu, H. Sato, and M.E. Berens. The phosphorylation of EphB2 receptor regulates migration and invasion of human glioma cells. Cancer Res, 2004. 64(9): 3179-85.

195. D. Yoshida, K. Watanabe, M. Noha, H. Takahashi, A. Teramoto, and Y. Sugisaki. Anti-invasive effect of an anti-matrix metalloproteinase agent in a murine brain slice model using the serial monitoring of green fluorescent protein-labeled glioma cells. Neurosurgery, 2003. 52(1): 187-96; discussion 196-7.

196. G.C. Easty and D.M. Easty. An Organ Culture System for the Examination of Tumor Invasion. Nature, 1963. 199: 1104-5.

197. R. Bjerkvig, O.D. Laerum, and O. Mella. Glioma cell interactions with fetal rat brain aggregates in vitro and with brain tissue in vivo. Cancer Res, 1986. 46(8): 4071-9.

198. S.K. Steinsvag, O.D. Laerum, and R. Bjerkvig. Interaction between Rat Glioma- Cells and Normal Rat-Brain Tissue in Organ-Culture. Journal of the National Cancer Institute, 1985. 74(5): 1095-1104.

199. S.K. Steinsvag and O.D. Laerum. Transmission Electron-Microscopy of Cocultures between Normal Rat-Brain Tissue and Rat Glioma-Cells. Anticancer Research, 1985. 5(2): 137-146.

200. O. Engebraaten, R. Bjerkvig, M. Lund-Johansen, K. Wester, P.-H. Pedersen, S. Mork, E.-O. Backlund, and O.D. Laerum. Interaction between human brain

235

turnout biopsies and fetal rat brain tissue in vitro. Acta Neuropathol, 1990. 81: 130-140.

201. J.S. Guillamo, F. Lisovoski, C. Christov, C. Le Gu´erinel, G.L. Defer, M. Peschanski, and T. Lefranc¸ois. Migration pathways of human glioblastoma cells xenografted into the immunosuppressed rat brain. Journal of Neuro-Oncology, 2001. 52: 205-215.

202. E.R. Laws, Jr., W.J. Goldberg, and J.J. Bernstein. Migration of human malignant astrocytoma cells in the mammalian brain: Scherer revisited. Int J Dev Neurosci, 1993. 11(5): 691-7.

203. W.J. Goldberg, E.R. Laws, Jr., and J.J. Bernstein. Individual C6 glioma cells migrate in adult rat brain after neural homografting. Int J Dev Neurosci, 1991. 9(4): 427-37.

204. J.J. Bernstein, W.J. Goldberg, and E.R. Laws, Jr. Human malignant astrocytoma xenografts migrate in rat brain: a model for central nervous system cancer research. J Neurosci Res, 1989. 22(2): 134-43.

205. J.J. Bernstein, W.J. Goldberg, and E.R. Laws, Jr. Immunohistochemistry of human malignant astrocytoma cells xenografted to rat brain: apolipoprotein E. Neurosurgery, 1989. 24(4): 541-6.

206. E.M. Caspani, D. Echevarria, K. Rottner, and J.V. Small. Live imaging of glioblastoma cells in brain tissue shows requirement of actin bundles for migration. Neuron Glia Biol, 2006. 2(2): 105-14.

207. K.M. Yamada and E. Cukierman. Modeling tissue morphogenesis and cancer in 3D. Cell, 2007. 130(4): 601-10.

208. D.W. Hutmacher, D. Loessner, S. Rizzi, D.L. Kaplan, D.J. Mooney, and J.A. Clements. Can tissue engineering concepts advance tumor biology research? Trends Biotechnol, 2010. 28(3): 125-33.

209. E. Burdett, F.K. Kasper, A.G. Mikos, and J.A. Ludwig. Engineering tumors: a tissue engineering perspective in cancer biology. Tissue Eng Part B Rev, 2010. 16(3): 351-9.

210. M.W. Tibbitt and K.S. Anseth. Hydrogels as Extracellular Matrix Mimics for 3D Cell Culture. Biotechnology and Bioengineering, 2009. 103(4): 655-663.

211. L.G. Griffith and M.A. Swartz. Capturing complex 3D tissue physiology in vitro. Nat Rev Mol Cell Biol, 2006. 7(3): 211-24.

236

212. Q.P. Pham, U. Sharma, and A.G. Mikos. Electrospinning of polymeric nanofibers for tissue engineering applications: a review. Tissue Eng, 2006. 12(5): 1197-211.

213. T.W. Thomas and P.A. DiMilla. Spreading and motility of human glioblastoma cells on sheets of silicone rubber depend on substratum compliance. Med Biol Eng Comput, 2000. 38(3): 360-70.

214. T.A. Ulrich, E.M. de Juan Pardo, and S. Kumar. The mechanical rigidity of the extracellular matrix regulates the structure, motility, and proliferation of glioma cells. Cancer Res, 2009. 69(10): 4167-74.

215. E. Devarajan and S. Huang. STAT3 as a central regulator of tumor metastases. Curr Mol Med, 2009. 9(5): 626-33.

216. J.J. Bernstein, E.R. Laws, Jr., K.V. Levine, L.R. Wood, G. Tadvalkar, and W.J. Goldberg. C6 glioma-astrocytoma cell and fetal astrocyte migration into artificial basement membrane: a permissive substrate for neural tumors but not fetal astrocytes. Neurosurgery, 1991. 28(5): 652-8.

217. V.D. Gordon, M.T. Valentine, M.L. Gardel, D. Andor-Ardo, S. Dennison, A.A. Bogdanov, D.A. Weitz, and T.S. Deisboeck. Measuring the mechanical stress induced by an expanding multicellular tumor system: a case study. Exp Cell Res, 2003. 289(1): 58-66.

218. J.J. Bernstein, W.J. Goldberg, and E.R. Laws, Jr. Migration of fresh human malignant astrocytoma cells into hydrated gel wafers in vitro. J Neurooncol, 1994. 18(2): 151-61.

219. L.J. Kaufman, C.P. Brangwynne, K.E. Kasza, E. Filippidi, V.D. Gordon, T.S. Deisboeck, and D.A. Weitz. Glioma expansion in collagen I matrices: analyzing collagen concentration-dependent growth and motility patterns. Biophys J, 2005. 89(1): 635-50.

220. H.D. Kim, T.W. Guo, A.P. Wu, A. Wells, F.B. Gertler, and D.A. Lauffenburger. Epidermal growth factor-induced enhancement of glioblastoma cell migration in 3D arises from an intrinsic increase in speed but an extrinsic matrix- and proteolysis-dependent increase in persistence. Mol Biol Cell, 2008. 19(10): 4249- 59.

221. B. Hegedus, F. Marga, K. Jakab, K.L. Sharpe-Timms, and G. Forgacs. The interplay of cell-cell and cell-matrix interactions in the invasive properties of brain tumors. Biophys J, 2006. 91(7): 2708-16.

237

222. S. Sarkar, R.K. Nuttall, S.H. Liu, D.R. Edwards, and V.W. Yong. Tenascin-C stimulates glioma cell invasion through matrix metalloproteinase-12. Cancer Research, 2006. 66(24): 11771-11780.

223. T.A. Ulrich, A. Jain, K. Tanner, J.L. MacKay, and S. Kumar. Probing cellular mechanobiology in three-dimensional culture with collagen-agarose matrices. Biomaterials, 2010. 31(7): 1875-84.

224. Y.L. Yang, S. Motte, and L.J. Kaufman. Pore size variable type I collagen gels and their interaction with glioma cells. Biomaterials, 2010. 31(21): 5678-5688.

225. L. David, V. Dulong, B. Coquerel, D. Le Cerf, L. Cazin, M. Lamacz, and J.P. Vannier. Collagens, stromal cell-derived factor-1alpha and basic fibroblast growth factor increase cancer cell invasiveness in a hyaluronan hydrogel. Cell Prolif, 2008. 41(2): 348-64.

226. B. Coquerel, F. Poyer, F. Torossian, V. Dulong, G. Bellon, I. Dubus, A. Reber, and J.P. Vannier. Elastin-derived peptides: matrikines critical for glioblastoma cell aggressiveness in a 3-D system. Glia, 2009. 57(16): 1716-26.

227. B. Ananthanarayanan, Y. Kim, and S. Kumar. Elucidating the mechanobiology of malignant brain tumors using a brain matrix-mimetic hyaluronic acid hydrogel platform. Biomaterials, 2011. 32(31): 7913-7923.

228. S.S. Rao, J. Dejesus, A. Sarkar, and J.O. Winter. Glioblastoma Behaviors in 3D Collagen-Hyaluronan Composite Hydrogels. Integrative Biology, Submitted, 2012.

229. S.G. Jin, Y.I. Jeong, S. Jung, H.H. Ryu, Y.H. Jin, and I.Y. Kim. The effect of hyaluronic Acid on the invasiveness of malignant glioma cells : comparison of invasion potential at hyaluronic Acid hydrogel and matrigel. J Korean Neurosurg Soc, 2009. 46(5): 472-8.

230. L. David, V. Dulong, D. Le Cerf, C. Chauzy, V. Norris, B. Delpech, M. Lamacz, and J.P. Vannier. Reticulated hyaluronan hydrogels: a model for examining cancer cell invasion in 3D. Matrix Biol, 2004. 23(3): 183-93.

231. Y.L. Yang, C. Sun, M.E. Wilhelm, L.J. Fox, J.L. Zhu, and L.J. Kaufman. Influence of chondroitin sulfate and hyaluronic acid on structure, mechanical properties, and glioma invasion of collagen I gels. Biomaterials, 2011. 32(31): 7932-7940.

232. F.M. Kievit, S.J. Florczyk, M.C. Leung, O. Veiseh, J.O. Park, M.L. Disis, and M. Zhang. Chitosan-alginate 3D scaffolds as a mimic of the glioma tumor microenvironment. Biomaterials, 2010. 31(22): 5903-10.

238

233. R.H. Shoemaker. The NCI60 human tumour cell line anticancer drug screen. Nat Rev Cancer, 2006. 6(10): 813-23.

234. T. Voskoglou-Nomikos, J.L. Pater, and L. Seymour. Clinical predictive value of the in vitro cell line, human xenograft, and mouse allograft preclinical cancer models. Clinical Cancer Research, 2003. 9(11): 4227-4239.

235. J.I. Johnson, S. Decker, D. Zaharevitz, L.V. Rubinstein, J.M. Venditti, S. Schepartz, S. Kalyandrug, M. Christian, S. Arbuck, M. Hollingshead, and E.A. Sausville. Relationships between drug activity in NCI preclinical in vitro and in vivo models and early clinical trials. Br J Cancer, 2001. 84(10): 1424-31.

236. I. Kola and J. Landis. Can the pharmaceutical industry reduce attrition rates? Nat Rev Drug Discov, 2004. 3(8): 711-5.

237. F. Pampaloni, E.G. Reynaud, and E.H. Stelzer. The third dimension bridges the gap between cell culture and live tissue. Nat Rev Mol Cell Biol, 2007. 8(10): 839- 45.

238. P. Oellers, M. Schallenberg, T. Stupp, P. Charalambous, V. Senner, W. Paulus, and S. Thanos. A coculture assay to visualize and monitor interactions between migrating glioma cells and nerve fibers. Nat Protoc, 2009. 4(6): 923-7.

239. P. Oellers, U. Schroer, V. Senner, W. Paulus, and S. Thanos. ROCKs are expressed in brain tumors and are required for glioma-cell migration on myelinated axons. Glia, 2009. 57(5): 499-509.

240. I. Martin, D. Wendt, and M. Heberer. The role of bioreactors in tissue engineering. Trends Biotechnol, 2004. 22(2): 80-6.

241. A. Ratcliffe and L.E. Niklason. Bioreactors and bioprocessing for tissue engineering. Ann N Y Acad Sci, 2002. 961: 210-5.

242. V.M. Tysseling-Mattiace, V. Sahni, K.L. Niece, D. Birch, C. Czeisler, M.G. Fehlings, S.I. Stupp, and J.A. Kessler. Self-assembling nanofibers inhibit glial scar formation and promote axon elongation after spinal cord injury. J Neurosci, 2008. 28(14): 3814-23.

243. S.J. Jhaveri, M.R. Hynd, N. Dowell-Mesfin, J.N. Turner, W. Shain, and C.K. Ober. Release of nerve growth factor from HEMA hydrogel-coated substrates and its effect on the differentiation of neural cells. Biomacromolecules, 2009. 10(1): 174-83.

239

244. Y. Lu, D. Wang, T. Li, X. Zhao, Y. Cao, H. Yang, and Y.Y. Duan. Poly(vinyl alcohol)/poly(acrylic acid) hydrogel coatings for improving electrode-neural tissue interface. Biomaterials, 2009. 30(25): 4143-51.

245. R.H. Schmedlen, K.S. Masters, and J.L. West. Photocrosslinkable polyvinyl alcohol hydrogels that can be modified with cell adhesion peptides for use in tissue engineering. Biomaterials, 2002. 23(22): 4325-32.

246. S.S. Rao and J.O. Winter. Adhesion molecule-modified biomaterials for neural tissue engineering. Front Neuroeng, 2009. 2: 1-14.

247. M. Eugene. Polyethyleneglycols and immunocamouflage of the cells tissues and organs for transplantation. Cell Mol Biol (Noisy-le-grand), 2004. 50(3): 209-15.

248. A.S. Sawhney, C.P. Pathak, and J.A. Hubell. Bioerodible hydrogels based on photopolymerized poly(ethylene glycol)-co-poly(.alpha.-hydroxy acid) diacrylate macromers. Macromolecules, 1993. 26: 581-587.

249. A. Ravni, S. Bourgault, A. Lebon, P. Chan, L. Galas, A. Fournier, H. Vaudry, B. Gonzalez, L.E. Eiden, and D. Vaudry. The neurotrophic effects of PACAP in PC12 cells: control by multiple transduction pathways. J Neurochem, 2006. 98(2): 321-9.

250. D. Vaudry, P.J. Stork, P. Lazarovici, and L.E. Eiden. Signaling pathways for PC12 cell differentiation: making the right connections. Science, 2002. 296(5573): 1648-9.

251. K. Fujita, P. Lazarovici, and G. Guroff. Regulation of the differentiation of PC12 pheochromocytoma cells. Environ Health Perspect, 1989. 80: 127-42.

252. S. Wang, L. Lu, J.A. Gruetzmacher, B.L. Currier, and M.J. Yaszemski. Synthesis and characterizations of biodegradable and crosslinkable poly(epsilon- caprolactone fumarate), poly(ethylene glycol fumarate), and their amphiphilic copolymer. Biomaterials, 2006. 27(6): 832-41.

253. S. Zhou, X. Deng, and H. Yang. Biodegradable poly(epsilon-caprolactone)- poly(ethylene glycol) block copolymers: characterization and their use as drug carriers for a controlled delivery system. Biomaterials, 2003. 24(20): 3563-70.

254. K. Hatada, T. Kitayama, T. Nishiura, and W. Shibuya. Relation between the reactivities of vinyl monomers in ionic polymerizations and their 1H NMR spectra. Journal of Polymer Science Part A: Polymer Chemistry, 2002. 40(13): 2134-2147.

240

255. S. Even-Ram and K.M. Yamada. Cell migration in 3D matrix. Curr Opin Cell Biol, 2005. 17(5): 524-32.

256. D.E. Discher, P. Janmey, and Y.L. Wang. Tissue cells feel and respond to the stiffness of their substrate. Science, 2005. 310(5751): 1139-43.

257. R.J. Pelham, Jr. and Y. Wang. Cell locomotion and focal adhesions are regulated by substrate flexibility. Proc Natl Acad Sci U S A, 1997. 94(25): 13661-5.

258. L.A. Flanagan, Y.E. Ju, B. Marg, M. Osterfield, and P.A. Janmey. Neurite branching on deformable substrates. Neuroreport, 2002. 13(18): 2411-5.

259. T. Yeung, P.C. Georges, L.A. Flanagan, B. Marg, M. Ortiz, M. Funaki, N. Zahir, W. Ming, V. Weaver, and P.A. Janmey. Effects of substrate stiffness on cell morphology, cytoskeletal structure, and adhesion. Cell Motil Cytoskeleton, 2005. 60(1): 24-34.

260. A.J. Engler, M.A. Griffin, S. Sen, C.G. Bonnemann, H.L. Sweeney, and D.E. Discher. Myotubes differentiate optimally on substrates with tissue-like stiffness: pathological implications for soft or stiff microenvironments. Journal of Cell Biology, 2004. 166(6): 877-87.

261. M.H. Zaman, L.M. Trapani, A. Siemeski, D. MacKellar, H.Y. Gong, R.D. Kamm, A. Wells, D.A. Lauffenburger, and P. Matsudaira. Migration of tumor cells in 3D matrices is governed by matrix stiffness along with cell-matrix adhesion and proteolysis. Proc Natl Acad Sci U S A, 2006. 103(29): 10889-10894.

262. S. Nemir and J.L. West. Synthetic materials in the study of cell response to substrate rigidity. Ann Biomed Eng, 2010. 38(1): 2-20.

263. R.K. Willits and S.L. Skornia. Effect of collagen gel stiffness on neurite extension. J Biomater Sci Polym Ed, 2004. 15(12): 1521-31.

264. C.M. Lo, H.B. Wang, M. Dembo, and Y.L. Wang. Cell movement is guided by the rigidity of the substrate. Biophysical Journal, 2000. 79(1): 144-152.

265. N. Zaari, P. Rajagopalan, S.K. Kim, A.J. Engler, and J.Y. Wong. Photopolymerization in microfluidic gradient generators: Microscale control of substrate compliance to manipulate cell response. Advanced Materials, 2004. 16(23-24): 2133-2137.

266. B.C. Isenberg, P.A. Dimilla, M. Walker, S. Kim, and J.Y. Wong. Vascular smooth muscle cell durotaxis depends on substrate stiffness gradient strength. Biophys J, 2009. 97(5): 1313-22.

241

267. Y.K. Cheung, E.U. Azeloglu, D.A. Shiovitz, K.D. Costa, D. Seliktar, and S.K. Sia. Microscale Control of Stiffness in a Cell-Adhesive Substrate Using Microfluidics-Based Lithography. Angewandte Chemie-International Edition, 2009. 48(39): 7188-7192.

268. H.G. Sundararaghavan, G.A. Monteiro, B.L. Firestein, and D.I. Shreiber. Neurite growth in 3D collagen gels with gradients of mechanical properties. Biotechnol Bioeng, 2009. 102(2): 632-43.

269. S. Nemir, H.N. Hayenga, and J.L. West. PEGDA hydrogels with patterned elasticity: Novel tools for the study of cell response to substrate rigidity. Biotechnol Bioeng, 2010. 105(3): 636-44.

270. S. Kidoaki and T. Matsuda. Microelastic gradient gelatinous gels to induce cellular mechanotaxis. J Biotechnol, 2008. 133(2): 225-30.

271. R.A. Marklein and J.A. Burdick. Spatially controlled hydrogel mechanics to modulate stem cell interactions. Soft Matter, 2010. 6(1): 136-143.

272. J.Y. Wong, A. Velasco, P. Rajagopalan, and Q. Pham. Directed movement of vascular smooth muscle cells on gradient-compliant hydrogels. Langmuir, 2003. 19(5): 1908-1913.

273. S. Sant, M.J. Hancock, J.P. Donnelly, D. Iyer, and A. Khademhosseini. Biomimetic gradient hydrogels for tissue engineering. The Canadian Journal of Chemical Engineering, 2010. 88(6): 899-911.

274. M.A. Kotlarchyk, S.G. Shreim, M.B. Alvarez-Elizondo, L.C. Estrada, R. Singh, L. Valdevit, E. Kniazeva, E. Gratton, A.J. Putnam, and E.L. Botvinick. Concentration independent modulation of local micromechanics in a fibrin gel. PLoS One, 2011. 6(5): e20201.

275. S.S. Soofi, J.A. Last, S.J. Liliensiek, P.F. Nealey, and C.J. Murphy. The elastic modulus of Matrigel (TM) as determined by atomic force microscopy. Journal of Structural Biology, 2009. 167(3): 216-219.

276. M.H. Zaman, R.D. Kamm, P. Matsudaira, and D.A. Lauffenburger. Computational model for cell migration in three-dimensional matrices. Biophysical Journal, 2005. 89(2): 1389-1397.

277. S. Sen, A.J. Engler, and D.E. Discher. Matrix Strains Induced by Cells: Computing How Far Cells Can Feel. Cellular and Molecular Bioengineering, 2009. 2(1): 39-48.

242

278. J.P. Winer, S. Oake, and P.A. Janmey. Non-linear elasticity of extracellular matrices enables contractile cells to communicate local position and orientation. PLoS One, 2009. 4(7): e6382.

279. A. Buxboim, I.L. Ivanovska, and D.E. Discher. Matrix elasticity, cytoskeletal forces and physics of the nucleus: how deeply do cells 'feel' outside and in? J Cell Sci, 2010. 123(Pt 3): 297-308.

280. R. Merkel, N. Kirchgessner, C.M. Cesa, and B. Hoffmann. Cell force microscopy on elastic layers of finite thickness. Biophys J, 2007. 93(9): 3314-23.

281. J.M. Maloney, E.B. Walton, C.M. Bruce, and K.J. Van Vliet. Influence of finite thickness and stiffness on cellular adhesion-induced deformation of compliant substrata. Physical Review E, 2008. 78(4): 041923.

282. Y.C. Lin, D.T. Tambe, C.Y. Park, M.R. Wasserman, X. Trepat, R. Krishnan, G. Lenormand, J.J. Fredberg, and J.P. Butler. Mechanosensing of substrate thickness. Phys Rev E Stat Nonlin Soft Matter Phys, 2010. 82(4 Pt 1): 041918.

283. J.A.W. van Dommelen, T.P.J. van der Sande, M. Hrapko, and G.W.M. Peters. Mechanical properties of brain tissue by indentation: Interregional variation. Journal of the Mechanical Behavior of Biomedical Materials, 2010. 3(2): 158- 166.

284. W.S. Leong, C.Y. Tay, H. Yu, A. Li, S.C. Wu, D.H. Duc, C.T. Lim, and L.P. Tan. Thickness sensing of hMSCs on collagen gel directs stem cell fate. Biochem Biophys Res Commun, 2010. 401(2): 287-92.

285. P. Friedl and K. Wolf. Tumour-cell invasion and migration: diversity and escape mechanisms. Nat Rev Cancer, 2003. 3(5): 362-74.

286. M.J. Bissell and D. Radisky. Putting tumours in context. Nat Rev Cancer, 2001. 1(1): 46-54.

287. O.W. Petersen, L. Ronnov-Jessen, A.R. Howlett, and M.J. Bissell. Interaction with basement membrane serves to rapidly distinguish growth and differentiation pattern of normal and malignant human breast epithelial cells. Proc Natl Acad Sci U S A, 1992. 89(19): 9064-8.

288. G.Y. Lee, P.A. Kenny, E.H. Lee, and M.J. Bissell. Three-dimensional culture models of normal and malignant breast epithelial cells. Nat Methods, 2007. 4(4): 359-65.

289. T.C. Laurent and J.R. Fraser. Hyaluronan. FASEB J, 1992. 6(7): 2397-404.

243

290. B.P. Toole. Hyaluronan: from extracellular glue to pericellular cue. Nat Rev Cancer, 2004. 4(7): 528-39.

291. J. Lee, S. Kotliarova, Y. Kotliarov, A. Li, Q. Su, N.M. Donin, S. Pastorino, B.W. Purow, N. Christopher, W. Zhang, J.K. Park, and H.A. Fine. Tumor stem cells derived from glioblastomas cultured in bFGF and EGF more closely mirror the phenotype and genotype of primary tumors than do serum-cultured cell lines. Cancer Cell, 2006. 9(5): 391-403.

292. R.D. Maestro, R. Shivers, W. McDonald, and A.D. Maestro. Dynamics of C6 astrocytoma invasion into three-dimensional collagen gels. J Neurooncol, 2001. 53(2): 87-98.

293. T.W. Wang and M. Spector. Development of hyaluronic acid-based scaffolds for brain tissue engineering. Acta Biomater, 2009.

294. S. Suri and C.E. Schmidt. Photopatterned collagen-hyaluronic acid interpenetrating polymer network hydrogels. Acta Biomater, 2009. 5(7): 2385-97.

295. S. Suri and C.E. Schmidt. Cell-Laden Hydrogel Constructs of Hyaluronic Acid, Collagen, and Laminin for Neural Tissue Engineering. Tissue Eng Part A 2010. 16(5): 1703-1716.

296. E. di Tomaso, N. London, D. Fuja, J. Logie, J.A. Tyrrell, W. Kamoun, L.L. Munn, and R.K. Jain. PDGF-C induces maturation of blood vessels in a model of glioblastoma and attenuates the response to anti-VEGF treatment. PLoS One, 2009. 4(4): e5123.

297. X.Z. Shu, S. Ahmad, Y. Liu, and G.D. Prestwich. Synthesis and evaluation of injectable, in situ crosslinkable synthetic extracellular matrices for tissue engineering. J Biomed Mater Res A, 2006. 79(4): 902-12.

298. B. Delpech, C. Maingonnat, N. Girard, C. Chauzy, R. Maunoury, A. Olivier, J. Tayot, and P. Creissard. Hyaluronan and hyaluronectin in the extracellular matrix of human brain tumour stroma. Eur J Cancer, 1993. 29A(7): 1012-7.

299. W. Hayen, M. Goebeler, S. Kumar, R. Riessen, and V. Nehls. Hyaluronan stimulates tumor cell migration by modulating the fibrin fiber architecture. J Cell Sci, 1999. 112: 2241-51.

300. J. Yang, C. Xu, C. Wang, and J. Kopecek. Refolding hydrogels self-assembled from N-(2-hydroxypropyl)methacrylamide graft copolymers by antiparallel coiled-coil formation. Biomacromolecules, 2006. 7(4): 1187-95.

244

301. M.D. Brigham, A. Bick, E. Lo, A. Bendali, J.A. Burdick, and A. Khademhosseini. Mechanically robust and bioadhesive collagen and photocrosslinkable hyaluronic acid semi-interpenetrating networks. Tissue Eng Part A, 2009. 15(7): 1645-53.

302. S.O. Suzuki and T. Iwaki. Dynamic analysis of glioma cells: looking into "movement phenotypes". Neuropathology, 2005. 25(3): 254-62.

303. J.A. Pedersen and M.A. Swartz. Mechanobiology in the third dimension. Ann Biomed Eng, 2005. 33(11): 1469-90.

304. M. van der Rest and R. Garrone. Collagen family of proteins. Faseb J, 1991. 5(13): 2814-23.

305. W. Paulus, I. Baur, D. Schuppan, and W. Roggendorf. Characterization of integrin receptors in normal and neoplastic human brain. Am J Pathol, 1993. 143(1): 154-63.

306. S.D. Bao, Q.L. Wu, R.E. McLendon, Y.L. Hao, Q. Shi, A.B. Hjelmeland, M.W. Dewhirst, D.D. Bigner, and J.N. Rich. Glioma stem cells promote radioresistance by preferential activation of the DNA damage response. Nature, 2006. 444(7120): 756-760.

307. D.A. Lim, S. Cha, M.C. Mayo, M.H. Chen, E. Keles, S. VandenBerg, and M.S. Berger. Relationship of glioblastoma multiforme to neural stem cell regions predicts invasive and multifocal tumor phenotype. Neuro-Oncology, 2007. 9(4): 424-429.

308. A.T.J. Belien, P.A. Paganetti, and M.E. Schwab. Membrane-type 1 matrix metalloprotease (MT1-MMP) enables invasive migration of glioma cells in central nervous system white matter. Journal of Cell Biology, 1999. 144(2): 373- 384.

309. C. Hebner, V.M. Weaver, and J. Debnath. Modeling morphogenesis and oncogenesis in three-dimensional breast epithelial cultures. Annu Rev Pathol, 2008. 3: 313-39.

310. K.R. Johnson, J.L. Leight, and V.M. Weaver. Demystifying the effects of a three- dimensional microenvironment in tissue morphogenesis. Methods Cell Biol, 2007. 83: 547-83.

311. V.M. Weaver, O.W. Petersen, F. Wang, C.A. Larabell, P. Briand, C. Damsky, and M.J. Bissell. Reversion of the malignant phenotype of human breast cells in three- dimensional culture and in vivo by integrin blocking antibodies. J Cell Biol, 1997. 137(1): 231-45.

245

312. M.B. Browning, T. Wilems, M. Hahn, and E. Cosgriff-Hernandez. Compositional control of poly(ethylene glycol) hydrogel modulus independent of mesh size. J Biomed Mater Res A, 2011. 98(2): 268-73.

313. J.E. Merrill. In vitro and in vivo pharmacological models to assess demyelination and remyelination. Neuropsychopharmacology, 2009. 34(1): 55-73.

314. G. Kronenberg, K. Gertz, G. Cheung, A. Buffo, H. Kettenmann, M. Gotz, and M. Endres. Modulation of fate determinants Olig2 and Pax6 in resident glia evokes spiking neuroblasts in a model of mild brain ischemia. Stroke, 2010. 41(12): 2944-9.

315. I. Burd, J. Chai, J. Gonzalez, E. Ofori, H. Monnerie, P.D. Le Roux, and M.A. Elovitz. Beyond white matter damage: fetal neuronal injury in a mouse model of preterm birth. Am J Obstet Gynecol, 2009. 201(3): 279 e1-8.

316. V. Tabar, G. Panagiotakos, E.D. Greenberg, B.K. Chan, M. Sadelain, P.H. Gutin, and L. Studer. Migration and differentiation of neural precursors derived from human embryonic stem cells in the rat brain. Nat Biotechnol, 2005. 23(5): 601-6.

317. J.M. Corey, C.C. Gertz, B.S. Wang, L.K. Birrell, S.L. Johnson, D.C. Martin, and E.L. Feldman. The design of electrospun PLLA nanofiber scaffolds compatible with serum-free growth of primary motor and sensory neurons. Acta Biomater, 2008. 4(4): 863-75.

318. B. Carlberg, M.Z. Axell, U. Nannmark, J. Liu, and H.G. Kuhn. Electrospun polyurethane scaffolds for proliferation and neuronal differentiation of human embryonic stem cells. Biomed Mater, 2009. 4(4): 045004.

319. J. Gerardo-Nava, T. Fuhrmann, K. Klinkhammer, N. Seiler, J. Mey, D. Klee, M. Moller, P.D. Dalton, and G.A. Brook. Human neural cell interactions with orientated electrospun nanofibers in vitro. Nanomedicine (Lond), 2009. 4(1): 11- 30.

320. J.Y. Lee, C.A. Bashur, A.S. Goldstein, and C.E. Schmidt. Polypyrrole-coated electrospun PLGA nanofibers for neural tissue applications. Biomaterials, 2009. 30(26): 4325-35.

321. M.P. Prabhakaran, J.R. Venugopal, and S. Ramakrishna. Mesenchymal stem cell differentiation to neuronal cells on electrospun nanofibrous substrates for nerve tissue engineering. Biomaterials, 2009. 30(28): 4996-5003.

322. V.J. Mukhatyar, M. Salmeron-Sanchez, S. Rudra, S. Mukhopadaya, T.H. Barker, A.J. Garcia, and R.V. Bellamkonda. Role of fibronectin in topographical guidance of neurite extension on electrospun fibers. Biomaterials, 2011. 32(16): 3958-68.

246

323. E. Schnell, K. Klinkhammer, S. Balzer, G. Brook, D. Klee, P. Dalton, and J. Mey. Guidance of glial cell migration and axonal growth on electrospun nanofibers of poly-epsilon-caprolactone and a collagen/poly-epsilon-caprolactone blend. Biomaterials, 2007. 28(19): 3012-25.

324. H.B. Wang, M.E. Mullins, J.M. Cregg, C.W. McCarthy, and R.J. Gilbert. Varying the diameter of aligned electrospun fibers alters neurite outgrowth and Schwann cell migration. Acta Biomater, 2010. 6(8): 2970-8.

325. S.Y. Chew, R. Mi, A. Hoke, and K.W. Leong. The effect of the alignment of electrospun fibrous scaffolds on Schwann cell maturation. Biomaterials, 2008. 29(6): 653-61.

326. G.T. Christopherson, H. Song, and H.Q. Mao. The influence of fiber diameter of electrospun substrates on neural stem cell differentiation and proliferation. Biomaterials, 2009. 30(4): 556-64.

327. N. Jovanov-Milosevic, V. Benjak, and I. Kostovic. Transient cellular structures in developing corpus callosum of the human brain. Coll Antropol, 2006. 30(2): 375- 81.

328. F. Yang, R. Murugan, S. Wang, and S. Ramakrishna. Electrospinning of nano/micro scale poly(L-lactic acid) aligned fibers and their potential in neural tissue engineering. Biomaterials, 2005. 26(15): 2603-10.

329. S.Y. Chew, R. Mi, A. Hoke, and K.W. Leong. Aligned Protein-Polymer Composite Fibers Enhance Nerve Regeneration: A Potential Tissue-Engineering Platform. Adv Funct Mater, 2007. 17(8): 1288-1296.

330. A. Martins, J.V. Araujo, R.L. Reis, and N.M. Neves. Electrospun nanostructured scaffolds for tissue engineering applications. Nanomedicine (Lond), 2007. 2(6): 929-42.

331. G. Benton, J. George, H.K. Kleinman, and I.P. Arnaoutova. Advancing science and technology via 3D culture on basement membrane matrix. J Cell Physiol, 2009. 221(1): 18-25.

332. A. Nakano, E. Tani, K. Miyazaki, Y. Yamamoto, and J. Furuyama. Matrix metalloproteinases and tissue inhibitors of metalloproteinases in human gliomas. J Neurosurg, 1995. 83(2): 298-307.

333. K. Lampert, U. Machein, M.R. Machein, W. Conca, H.H. Peter, and B. Volk. Expression of matrix metalloproteinases and their tissue inhibitors in human brain tumors. American Journal of Pathology, 1998. 153(2): 429-437.

247

334. M. Platten, W. Wick, and M. Weller. Malignant glioma biology: role for TGF- beta in growth, motility, angiogenesis, and immune escape. Microsc Res Tech, 2001. 52(4): 401-10.

335. W.L. Yeh, D.Y. Lu, M.J. Lee, and W.M. Fu. Leptin induces migration and invasion of glioma cells through MMP-13 production. Glia, 2009. 57(4): 454-64.

336. A. Inoue, H. Takahashi, H. Harada, S. Kohno, S. Ohue, K. Kobayashi, H. Yano, J. Tanaka, and T. Ohnishi. Cancer stem-like cells of glioblastoma characteristically express MMP-13 and display highly invasive activity. Int J Oncol, 2010. 37(5): 1121-31.

337. J. Gaumer, A. Prasad, D. Lee, and J. Lannutti. Structure-function relationships and source-to-ground distance in electrospun polycaprolactone. Acta Biomater, 2009. 5(5): 1552-61.

338. J. Nam, Y. Huang, S. Agarwal, and J. Lannutti. Materials selection and residual solvent retention in biodegradable electrospun fibers. Journal of Applied Polymer Science, 2008. 107(3): 1547-1554.

339. H.B. Wang, M.E. Mullins, J.M. Cregg, A. Hurtado, M. Oudega, M.T. Trombley, and R.J. Gilbert. Creation of highly aligned electrospun poly-L-lactic acid fibers for nerve regeneration applications. J Neural Eng, 2009. 6(1): 016001.

340. J. Lee and Y.L. Deng. Increased mechanical properties of aligned and isotropic electrospun PVA nanofiber webs by cellulose nanowhisker reinforcement. Macromolecular Research, 2012. 20(1): 76-83.

341. C.E. Ayres, B.S. Jha, H. Meredith, J.R. Bowman, G.L. Bowlin, S.C. Henderson, and D.G. Simpson. Measuring fiber alignment in electrospun scaffolds: a user's guide to the 2D fast Fourier transform approach. J Biomater Sci Polym Ed, 2008. 19(5): 603-21.

342. J. Lannutti, J. Johnson, and A. Ghosh. Microstructure-property relationships in a tissue-engineering scaffold. Journal of Applied Polymer Science, 2007. 104(5): 2919-2927.

343. J. Nam, J. Johnson, J.J. Lannutti, and S. Agarwal. Modulation of embryonic mesenchymal progenitor cell differentiation via control over pure mechanical modulus in electrospun nanofibers. Acta Biomater, 2011. 7(4): 1516-24.

344. B.N. Blackstone, J.J. Willard, C.H. Lee, T.M. Nelson, R.T. Hart, J.J. Lannutti, and H.M. Powell. Plasma Surface Modification of Electrospun Fibers for Adhesion-Based Cancer Cell Sorting. Integrative Biology, 2012, DOI: 10.1039/C2IB20025B

248

345. B.D. Cox, M. Natarajan, M.R. Stettner, and C.L. Gladson. New concepts regarding focal promotion of cell migration adhesion kinase and proliferation. Journal of Cellular Biochemistry, 2006. 99(1): 36-52.

346. C. Beadle, M.C. Assanah, P. Monzo, R. Vallee, S.S. Rosenfeld, and P. Canoll. The role of myosin II in glioma invasion of the brain. Mol Biol Cell, 2008. 19(8): 3357-68.

347. M. Yamamoto, S. Mohanam, R. Sawaya, G.N. Fuller, M. Seiki, H. Sato, Z.L. Gokaslan, L.A. Liotta, G.L. Nicolson, and J.S. Rao. Differential expression of membrane-type matrix metalloproteinase and its correlation with gelatinase A activation in human malignant brain tumors in vivo and in vitro. Cancer Res, 1996. 56(2): 384-92.

348. T.E. VanMeter, H.K. Rooprai, M.M. Kibble, H.L. Fillmore, W.C. Broaddus, and G.J. Pilkington. The role of matrix metalloproteinase genes in glioma invasion: co-dependent and interactive proteolysis. J Neurooncol, 2001. 53(2): 213-35.

349. D.K. Binder and M.S. Berger. Proteases and the biology of glioma invasion. J Neurooncol, 2002. 56(2): 149-58.

350. M.Y. Li, M.J. Mondrinos, M.R. Gandhi, F.K. Ko, A.S. Weiss, and P.I. Lelkes. Electrospun protein fibers as matrices for tissue engineering. Biomaterials, 2005. 26(30): 5999-6008.

351. W.J. Li, J.A. Cooper, Jr., R.L. Mauck, and R.S. Tuan. Fabrication and characterization of six electrospun poly(alpha-hydroxy ester)-based fibrous scaffolds for tissue engineering applications. Acta Biomater, 2006. 2(4): 377-85.

352. A. Heredia, C.C. Bui, U. Suter, P. Young, and T.E. Schaffer. AFM combines functional and morphological analysis of peripheral myelinated and demyelinated nerve fibers. Neuroimage, 2007. 37(4): 1218-26.

353. S.W. Moore and M.P. Sheetz. Biophysics of substrate interaction: influence on neural motility, differentiation, and repair. Dev Neurobiol, 2011. 71(11): 1090- 101.

354. M. Natarajan, T.P. Hecker, and C.L. Gladson. FAK signaling in anaplastic astrocytoma and glioblastoma tumors. Cancer J, 2003. 9(2): 126-33.

355. S. Ivkovic, C. Beadle, S. Noticewala, S.C. Massey, K.R. Swanson, L.N. Toro, A.R. Bresnick, P. Canoll, and S.S. Rosenfeld. Direct inhibition of myosin II effectively blocks glioma invasion in the presence of multiple motogens. Mol Biol Cell, 2012. 23(4): 533-42.

249

356. Y.Z. Zhang, J. Venugopal, Z.M. Huang, C.T. Lim, and S. Ramakrishna. Characterization of the surface biocompatibility of the electrospun PCL-collagen nanofibers using fibroblasts. Biomacromolecules, 2005. 6(5): 2583-2589.

357. W. He, Z. Ma, T. Yong, W.E. Teo, and S. Ramakrishna. Fabrication of collagen- coated biodegradable polymer nanofiber mesh and its potential for endothelial cells growth. Biomaterials, 2005. 26(36): 7606-15.

358. A. Martins, R.L. Reis, and N.M. Neves. Electrospinning: processing technique for tissue engineering scaffolding. International Materials Reviews, 2008. 53(5): 257- 274.

359. H. Fernandes, L. Moroni, C. van Blitterswijk, and J. de Boer. Extracellular matrix and tissue engineering applications. Journal of Materials Chemistry, 2009. 19(31): 5474-5484.

360. J. Fang, H.T. Niu, T. Lin, and X.G. Wang. Applications of electrospun nanofibers. Chinese Science Bulletin, 2008. 53(15): 2265-2286.

361. H.W. Choi, J.K. Johnson, J. Nam, D.F. Farson, and J. Lannutti. Structuring electrospun polycaprolactone nanofiber tissue scaffolds by femtosecond laser ablation. Journal of Laser Applications, 2007. 19(4): 225-231.

362. Z.C. Sun, E. Zussman, A.L. Yarin, J.H. Wendorff, and A. Greiner. Compound core-shell polymer nanofibers by co-electrospinning. Advanced Materials, 2003. 15(22): 1929-1932.

363. D. Li, A. Babel, S.A. Jenekhe, and Y.N. Xia. Nanofibers of conjugated polymers prepared by electrospinning with a two-capillary spinneret. Advanced Materials, 2004. 16(22): 2062-2066.

364. Y.Z. Zhang, Z.M. Huang, X.J. Xu, C.T. Lim, and S. Ramakrishna. Preparation of core-shell structured PCL-r-gelatin Bi-component nanofibers by coaxial electrospinning. Chemistry of Materials, 2004. 16(18): 3406-3409.

365. H. Jiang, Y. Hu, Y. Li, P. Zhao, K. Zhu, and W. Chen. A facile technique to prepare biodegradable coaxial electrospun nanofibers for controlled release of bioactive agents. J Control Release, 2005. 108(2-3): 237-43.

366. Z.M. Huang, Y.Z. Zhang, and S. Ramakrishna. Double-layered composite nanofibers and their mechanical performance. Journal of Polymer Science Part B- Polymer Physics, 2005. 43(20): 2852-2861.

250

367. M. Wang, J.H. Yu, D.L. Kaplan, and G.C. Rutledge. Production of submicron diameter silk fibers under benign processing conditions by two-fluid electrospinning. Macromolecules, 2006. 39(3): 1102-1107.

368. P. Zhao, H. Jiang, H. Pan, K. Zhu, and W. Chen. Biodegradable fibrous scaffolds composed of gelatin coated poly(epsilon-caprolactone) prepared by coaxial electrospinning. J Biomed Mater Res A, 2007. 83(2): 372-82.

369. I.C. Liao, S.Y. Chew, and K.W. Leong. Aligned core-shell nanofibers delivering bioactive proteins. Nanomedicine (Lond), 2006. 1(4): 465-71.

370. C.Y. Wang, J.J. Liu, C.Y. Fan, X.M. Mo, H.J. Ruan, and F.F. Li. The effect of aligned core-shell nanofibres delivering NGF on the promotion of sciatic nerve regeneration. J Biomater Sci Polym Ed, 2012. 23(1-4): 167-84.

371. M. Gulfam, J.M. Lee, J.E. Kim, D.W. Lim, E.K. Lee, and B.G. Chung. Highly porous core-shell polymeric fiber network. Langmuir, 2011. 27(17): 10993-9.

372. C.M. Haslauer, A.K. Moghe, J.A. Osborne, B.S. Gupta, and E.G. Loboa. Collagen-PCL sheath-core bicomponent electrospun scaffolds increase osteogenic differentiation and calcium accretion of human adipose-derived stem cells. J Biomater Sci Polym Ed, 2011. 22(13): 1695-712.

373. F. Marchesi, L. Piemonti, A. Mantovani, and P. Allavena. Molecular mechanisms of perineural invasion, a forgotten pathway of dissemination and metastasis. Cytokine Growth Factor Rev, 2010. 21(1): 77-82.

374. D.H. Kim, P.P. Provenzano, C.L. Smith, and A. Levchenko. Matrix nanotopography as a regulator of cell function. Journal of Cell Biology, 2012. 197(3): 351-360.

375. M.D. Mccartney and R.C. Buck. Comparison of the Degree of Contact Guidance between Tumor-Cells and Normal-Cells Invitro. Cancer Research, 1981. 41(8): 3046-3051.

376. A. Curtis and C. Wilkinson. Topographical control of cells. Biomaterials, 1997. 18(24): 1573-83.

377. D. Hoffman-Kim, J.A. Mitchel, and R.V. Bellamkonda. Topography, cell response, and nerve regeneration. Annu Rev Biomed Eng, 2010. 12: 203-31.

378. C.J. Bettinger, R. Langer, and J.T. Borenstein. Engineering Substrate Topography at the Micro- and Nanoscale to Control Cell Function. Angewandte Chemie- International Edition, 2009. 48(30): 5406-5415.

251

379. N. Han, J. Johnson, J.J. Lannutti, and J.O. Winter. Hydrogel-electrospun fiber composite materials for hydrophilic protein release. J Control Release, 2012. 158(1): 165-70.

380. N. Han, S.S. Rao, J. Johnson, K.S. Parikh, P.A. Bradley, J.J. Lannutti, and J.O. Winter. Hydrogel-electrospun fiber mat composite coatings for neural prostheses. Front Neuroeng, 2011. 4: 2.

381. Y. Hong, A. Huber, K. Takanari, N.J. Amoroso, R. Hashizume, S.F. Badylak, and W.R. Wagner. Mechanical properties and in vivo behavior of a biodegradable synthetic polymer microfiber-extracellular matrix hydrogel biohybrid scaffold. Biomaterials, 2011. 32(13): 3387-94.

382. J. Coburn, M. Gibson, P.A. Bandalini, C. Laird, H.Q. Mao, L. Moroni, D. Seliktar, and J. Elisseeff. Biomimetics of the Extracellular Matrix: An Integrated Three-Dimensional Fiber-Hydrogel Composite for Cartilage Tissue Engineering. Smart Struct Syst, 2011. 7(3): 213-222.

383. A. Hsieh, T. Zahir, Y. Lapitsky, B. Amsden, W. Wan, and M.S. Shoichet. Hydrogel/electrospun fiber composites influence neural stem/progenitor cell fate. Soft Matter, 2010. 6: 2227-2237.

384. D. Kai, M.P. Prabhakaran, B. Stahl, M. Eblenkamp, E. Wintermantel, and S. Ramakrishna. Mechanical properties and in vitro behavior of nanofiber-hydrogel composites for tissue engineering applications. Nanotechnology, 2012. 23(9): 095705.

385. W. Schuurman, V. Khristov, M.W. Pot, P.R. van Weeren, W.J. Dhert, and J. Malda. Bioprinting of hybrid tissue constructs with tailorable mechanical properties. Biofabrication, 2011. 3(2): 021001.

386. Y. Yang, I. Wimpenny, and M. Ahearne. Portable nanofiber meshes dictate cell orientation throughout three-dimensional hydrogels. Nanomedicine, 2011. 7(2): 131-6.

387. A.C. Yu, Y.L. Lee, and L.F. Eng. Astrogliosis in culture: I. The model and the effect of antisense oligonucleotides on glial fibrillary acidic protein synthesis. J Neurosci Res, 1993. 34(3): 295-303.

388. E.F. Ellis, J.S. McKinney, K.A. Willoughby, S. Liang, and J.T. Povlishock. A new model for rapid stretch-induced injury of cells in culture: characterization of the model using astrocytes. J Neurotrauma, 1995. 12(3): 325-39.

389. V.W. Wu and J.P. Schwartz. Cell culture models for reactive gliosis: new perspectives. J Neurosci Res, 1998. 51(6): 675-81.

252

390. M.M. Mueller, T. Werbowetski, and R.F. Del Maestro. Soluble factors involved in glioma invasion. Acta Neurochir (Wien), 2003. 145(11): 999-1008.

391. J.Y. Lee, C.A. Bashur, C.A. Milroy, L. Forciniti, A.S. Goldstein, and C.E. Schmidt. Nerve Growth Factor-Immobilized Electrically Conducting Fibrous Scaffolds for Potential Use in Neural Engineering Applications. IEEE Transactions on Nanobioscience, 2012. 11(1): 15-21.

392. P.A. Kenny, G.Y. Lee, C.A. Myers, R.M. Neve, J.R. Semeiks, P.T. Spellman, K. Lorenz, E.H. Lee, M.H. Barcellos-Hoff, O.W. Petersen, J.W. Gray, and M.J. Bissell. The morphologies of breast cancer cell lines in three-dimensional assays correlate with their profiles of gene expression. Mol Oncol, 2007. 1(1): 84-96.

393. A. Potti, H.K. Dressman, A. Bild, R.F. Riedel, G. Chan, R. Sayer, J. Cragun, H. Cottrill, M.J. Kelley, R. Petersen, D. Harpole, J. Marks, A. Berchuck, G.S. Ginsburg, P. Febbo, J. Lancaster, and J.R. Nevins. Genomic signatures to guide the use of chemotherapeutics. Nat Med, 2006. 12(11): 1294-300.

394. L.A. Gurski, N.J. Petrelli, X. Jia, and M.C. Farach-Carson. 3D Matrices for Anti- Cancer Drug Testing and Development. Oncology Issues, 2010(25): 20-25.

395. J. Clarke, N. Butowski, and S. Chang. Recent advances in therapy for glioblastoma. Arch Neurol, 2010. 67(3): 279-83.

396. J. Chen, R.M. McKay, and L.F. Parada. Malignant glioma: lessons from genomics, mouse models, and stem cells. Cell, 2012. 149(1): 36-47.

397. W. Knudson and R.F. Loeser. CD44 and integrin matrix receptors participate in cartilage homeostasis. Cell Mol Life Sci, 2002. 59(1): 36-44.

398. L. Yang, J.C. Lang, P. Balasubramanian, K.R. Jatana, D. Schuller, A. Agrawal, M. Zborowski, and J.J. Chalmers. Optimization of an enrichment process for circulating tumor cells from the blood of head and neck cancer patients through depletion of normal cells. Biotechnol Bioeng, 2009. 102(2): 521-34.

399. M. Lustberg, K.R. Jatana, M. Zborowski, and J.J. Chalmers. Emerging Technologies for CTC Detection Based on Depletion of Normal Cells. Recent Results Cancer Res, 2012. 195: 97-110.

400. P. Balasubramanian, L.Y. Yang, J.C. Lang, K.R. Jatana, D. Schuller, A. Agrawal, M. Zborowski, and J.J. Chalmers. Confocal Images of Circulating Tumor Cells Obtained Using a Methodology and Technology That Removes Normal Cells. Molecular Pharmaceutics, 2009. 6(5): 1402-1408.

253

401. D. Huh, Y.S. Torisawa, G.A. Hamilton, H.J. Kim, and D.E. Ingber. Microengineered physiological biomimicry: Organs-on-Chips. Lab Chip, 2012. 12(12): 2156-64.

402. N.T. Elliott and F. Yuan. A review of three-dimensional in vitro tissue models for drug discovery and transport studies. J Pharm Sci, 2011. 100(1): 59-74.

403. M.J. Bissell. Modelling molecular mechanisms of breast cancer and invasion: lessons from the normal gland. Biochem Soc Trans, 2007. 35(Pt 1): 18-22.

254

Appendix: Experimental Protocols

Hydrogel Sample Preparation for Scanning Electron Microscopy (SEM)

Purpose: This protocol outlines the steps for preparing hydrogel samples for SEM to examine their interior micro-architectural features.

Materials: Hydrogel samples prepared in a microcentrifuge tube DI water Liquid Nitrogen Razor blade Tweezers Carbon Tape Aluminium Stubs Cryogenic Gloves

Safety Concerns and PPE: Liquid nitrogen and all other cryogenic materials can cause significant burns.  Gloves and goggles (not safety glasses) should be worn at all times.  Full length aprons can be worn when handling large quantities to protect against splashes.  Remove metallic jewelry on hands and wrists. PPE: When performing flash freezing, all operators should wear latex or nitrile gloves and cryogenic gloves, lab coat, and safety glasses.

Procedure: 1. Prepare the desired hydrogel samples in microcentrifuge (~ 0.6 mL) tubes. Since most gels are prepared in PBS or cell culture media, add DI water to the tubes to remove as many salts as possible and incubating at room temperature on a rocker over night (if needed). Change DI water and replace with fresh water at least twice. 2. Following this, remove all the water and immerse these tubes in a glass beaker containing liquid nitrogen inside the chemical fume hood (flash freezing). 3. Using a needle carefully poke one hole in the tube and place in a lyophiliser overnight with the tube closed. The hole will allow for moisture evaporation. I have used the lyophiliser in Dr. Guan’s lab in the Department of Materials Science and Engineering.

255

4. Using a razor blade, cut the gel with the tubes to expose the gel interior and carefully using tweezers place them on an aluminium stub with double sided carbon tape. 5. Coat samples with gold for ~ 30 s using a gold coater and view on a scanning electron microscope.

256

Hydrogel Sample Preparation for Confocal Reflectance Microscopy (CRM)

Purpose: This protocol outlines the steps for preparing hydrogel samples that can be observed using confocal reflectance microscopy (e.g., collagen hydrogels).

Materials: Hydrogel samples Culture Media (if hydrogels are prepared using culture media) Glass Slide Cover Glass Super Glue Chamber Gaskets (CoverWell™ Perfusion Chamber Gasket, Eight Chambers, 9 mm Diameter, 2.0 mm Deep, Invitrogen C18141)

Safety Concerns and PPE: PPE: All operators should wear latex or nitrile gloves, lab coat, and safety glasses.

Procedure: 1. Carefully remove the plastic on the chamber gasket and glue them to a clean glass slide using super glue and let it dry overnight. 2. Prepare hydrogel samples on the chamber gasket (~ 100 µL gel samples can be prepared) as per protocols. 3. Use cell culture media to cover the gel surface and carefully place a cover glass on top. 4. Image using a confocal reflectance microscope in Campus Microscopy and Imaging Facility (CMIF) located in the Biomedical Research Tower at OSU.

257

Cell Culture Media Preparation for OSU2 Cells

Purpose: This protocol describes methods to make media for the culture of OSU2 cells. OSU2 cells have been directly procured from primary brain tumor of patients at Ohio State University as per IRB protocols. Cell Source: Dr. Atom Sarkar, Neurosurgery, OSU, now at Geisinger Health System, Danville, PA.

Materials: Medium (250 mL) 222.5 mL D-MEM/F-12 (1X) [Invitrogen 11330057] 89% 25 mL FBS (Fetal Bovine Serum) [Sigma F6178] 10% 2.5 mL Pen/Strep7 [Invitrogen/ Gibco 15140-148] 1%

Safety Concerns and PPE: Cell culture media must be synthesized in a sterile environment. Note that a sterile environment DOES NOT provide protection against chemical hazards. If chemical hazards are present work should be performed with a respirator or similar protective device.  Operators must perform indicated steps in the sterile, laminar flow tissue culture hood.  Wipe all surfaces with 70% ethanol.  Work as far back in the hood as possible.  Avoid reaching over all sterile surfaces. Note that OSU2 should be cultured in BSL2 containment, since it a patient derived cell line. All researchers working with OSU2 should be registered with occupational health and safety services and must have completed health checkups. PPE: When making media all operators should wear latex or nitrile gloves, lab coat, and safety glasses.

Procedure Media Preparation (250 ml) 1. Add 222.5 ml of media stock to a sterile 250 ml pre-sterilized container in the tissue culture hood. 2. Add 25 ml of FBS to container. Sera should be thawed in advance, but cannot last in the refrigerator for more than a few weeks. If necessary, sera can be thawed in the water bath on the day of use, but it is better to thaw the vial in the refrigerator a few days in advance of use. If opening a new vial, triturate (mix using the pipette) several times to ensure that proteins at the bottom of the vial are evenly distributed. 3. Add 2.5 ml of pen-strep (penicillin-streptomycin). Media is now ready for use.

7 There are two concentrations 5000 and 10000 units. Usually these are diluted to 100 U/100 mL media (1%). Higher concentrations (2%) may be used for the culture of primary cells. 258

Cell Pre-Staining for Long Term Cell Tracking Studies

Purpose: This protocol describes cell staining using Cell Tracker Green (CMFDA) prior to tracking cell behaviors over long term using time lapse confocal imaging. As an example, this can be used for tracking tumor cell migration in 3D hydrogels, on electrospun fibers, or other 2D surfaces.

Materials: Cell Tracker™ Green CMFDA (5-Chloromethylfluorescein Diacetate) [Invitrogen C7025] D-MEM/F-12 (1X) [Invitrogen 11330057] OSU-2 Cells OSU-2 Cell Culture Media [Refer to OSU-2 Media Preparation Protocol]

Safety Concerns and PPE All operations must be performed in a sterile environment. Note that a sterile environment DOES NOT provide protection against chemical hazards. If chemical hazards are present work should be performed with a respirator or similar protective device.  Operators must perform indicated steps in the sterile, laminar flow tissue culture hood.  Wipe all surfaces with 70% ethanol.  Work as far back in the hood as possible.  Avoid reaching over all sterile surfaces. PPE: When making media all operators should wear latex or nitrile gloves, lab coat, and safety glasses.

Procedure 1. Passage cells as per standard protocols. 2. After passaging, resuspend the cells in 2 mL of serum free DMEM for respective cells (e.g., D-MEM/F-12 (1X) for OSU-2 cells). 3. Dissolve one vial of dye in ~ 10.8 µL high quality sterile DMSO to result in a concentration of ~ 10 mM. 4. Add ~ 0.4 µL of this CMFDA green dye to ~ 2 mL cell solution and place in incubator for ~ 45 min. This will give a concentration of ~ 2 µM of the dye bright enough for cell tracking. 5. After 45 min, centrifuge the cell suspension, remove the supernatant and resuspend in respective cell culture media for cell seeding experiments. Note if substantial background fluorescence arises while imaging, this could be due to presence of dye not taken up by cells. If so, try to remove unbound dye by repeated centrifugation and removal of media followed by fresh media addition.

259

Laser Scanning Confocal Microscopy for Time Lapse Experiments

Purpose: This procedure outlines steps for performing time lapse experiments for tracking cells on substrates on a Laser Scanning Confocal Microscope in the Department of Neurosurgery, Wiseman Hall, The Ohio State University.

Procedure: 1. Turn on the computer; then the scope. When the “log on” window pops up on the screen, just hit “enter.” 2. Assemble chamber a. Plate or slide holder w/ blank b. Clear plastic chamber—tape back side to plate holder, connect inflow/outflow tubes to heat/CO2 control box. c. Check CO2 tanks. Left controls chamber, ensure green tubing is connected. Right controls table top. Open both, following specifications listed on the tanks. d. Turn on heat CO2 controller. Monitor levels on the front of the machine. 3. Open LSM 510 and choose “scan new images” then “start expert mode” 4. Open the following windows under the “aquire” menu: laser, micro, config, scan, stage. a. Laser window—control laser settings b. Micro—controls scope (e.g. objective in use, switching from BF to fluorescent filters when looking into eyepiece) c. Config—access to previous configurations, save new configurations d. Scan—controls what you see on the computer monitor. Frequently use continuous scanning, can change detection levels of fluorescence/BF, Z- stack controls, scan speed/resolution. e. Stage—mark locations, etc. 5. TIP: usually I’ll open the database for a previous experiment and reuse the configuration—to do this, just hit the “reuse” button in the database window. It’ll prompt you as to what lasers need to be turned on to use that given configuration. 6. To look into the scope, select the “aquire” menu, then choose the “vis” button (eyeball). 7. When you find a location you’d like to scan, hit “continuous” on the scan window. Use the coarse/fine focus knobs on the scope to get the image into focus on the screen. Adjust levels of detection as needed. 8. Once satisfied with image, you can mark the location in the stage window. Or, if you are taking stills only, hit “stop” on the scan menu. Then press the “single” button and save that image. 9. For time-lapses, go back over your locations and tweak the configurations as needed, saving each different one with a unique name. This is done in the “config” window. Make note of which locations require each of your configurations.

260

10. Make new database for temp images. FileNew Databaseselect folder and name. 11. To start time-lapse. Select the “macro” menu from the LSM toolbar. From there, select “time-lapse.” 12. In the “time-lapse” window: a. Click “select image database” button at the bottom of the window. Select the folder you created for the temp images. b. Click the “options” button (on the right vertical panel) and select that same database) c. Click “update” (smiley face) to load the locations you’ve set. d. Go through each location and set the time delay (calculated based on scan time and amount of time between group repeats), configuration, and z- stack (if applicable). e. Click “start time lapse.” Calculations for time delay: (Minutes between scans of 1 location x 60) - (Scan time x number of locations) Number of locations

261

Imaging Single Cell Migration using Time Lapse Spinning Disk Confocal Microscopy

Purpose: This protocol describes imaging of single cells stained with Cell Tracker Green (CMFDA) using time lapse spinning disk confocal microscopy in the Winter lab. For cell staining, refer to the appropriate protocols.

Safety Concerns and PPE: All operations must be performed in a sterile environment. Note that a sterile environment DOES NOT provide protection against chemical hazards. If chemical hazards are present work should be performed with a respirator or similar protective device.  Operators must perform indicated steps in the sterile, laminar flow tissue culture hood.  Wipe all surfaces with 70% ethanol.  Work as far back in the hood as possible.  Avoid reaching over all sterile surfaces. PPE: All operators should wear latex or nitrile gloves, lab coat, and safety glasses.

Procedure 1. Seed pre-stained cells on appropriate culture surfaces (i.e., hydrogels and/or electrospun fibers) 2. Switch on the heating unit of the microscopy well ahead to allow the microscope to reach 37 ºC. For example, if running an overnight experiment (~ 9 PM), it is better to switch on the microscope heating system late morning (~ 11 AM). This will give enough time for heating. Also, microscope parts will slightly expand with heat. Hence, if enough time is not provided for heating the chamber, it might interfere with imaging by drifting the focus. 3. After ~ 6 hours of seeding, exchange the media to remove any floating cells. 4. Place the plate on microscope plate holder and switch on the 5% CO2 tank. Check the water level in the conical flask to ensure enough water is present to provide humidity. Allow the plate to sit for ~ 2-3 hours before commencing the automated time lapse to adjust to the new conditions. These will significantly minimize drift in the Z-direction enabling better image acquisition. 5. Start the time lapse experiment using the Metamorph Multidimensional Acquisition Platform.

262

Sterilizing Electrospun Fiber Mats

Purpose: Electrospun fiber mats may be used as physiological models for the study of tumor cell adhesion and migration. Before use, they must be sterilized with 70% ethanol.

Materials: Fiber mats Well plate 1000-µL Micropipette and tips Fine tweezers 70% Ethanol Sterile PBS Dow Corning Silastic Brand, Medical Adhesive, Silicone Type A Cell culture medium

Safety Concerns and PPE:  70% Ethanol is a flammable solvent, and when combined with heat sources (i.e., alcohol lamp for polishing pipet tips or hot lightbulbs) presents a combustion hazard. PPE: When sterilizing fiber mats all operators should wear latex or nitrile gloves, lab coat, and safety glasses.

Procedure: 1. In a sterile tissue culture hood, use tweezers to add fiber mats to the wells of a well plate, one mat per well. Take caution to add the mats fiber-side up. The side with fibers should look duller than the side without. Alternatively, drill appropriate size holes and glue the fiber mat facing up for cell culture using a medical adhesive. 2. Fill each well with enough ethanol for the mat to be fully submerged (~ 200- 300 µL for a 24 well plate). 3. Place the lid on well plate. 4. Allow the mats to soak for ~ 0.5 hours with UV on. 5. Aspirate any remaining ethanol with either a vacuum pump and sterile Pasteur pipettes or a micropipette. 6. Rinse each mat twice with PBS. 7. Replace the lid on top of the well plate, leaving a vent for air flow. 8. Close the sash of the hood and let the mat dry overnight without UV. 9. Rinse the mats once with appropriate cell culture medium for whichever cell line will be cultured on the mats.

263

8,9 Actin Staining Purpose: To stain actin filaments in the cytoskeleton of cells. This procedure can be used to study the cytoskeletal development and morphology of glioblastoma multiformes in and on various 3D hydrogels and other substrates10. Materials: PBS (Sigma P4417) Rocker Sucrose (Sigma S0389) Hot/Stir plate 1M NaOH (Sigma-Aldrich 480878) Stir bar Paraformaldehyde (Sigma-Aldrich 158127) Analytical balance Triton X-100 (Sigma T8787) Microspatula BSA (Jackson ImmunoResearch 001-000-161) Weighing paper Alexa Fluor® 633 phalloidin stain (Invitrogen A22284) 10-mL beaker Chemical fume hood Micropipettes Sterile tissue culture hood Micropipette tips Fluorescence microscope Microfuge tubes Incubator Large beaker Refrigerator 70% Ethanol

Safety Concerns and PPE: Paraformaldehyde is a toxin, teratogen, and suspected carcinogen. It is also an irritant to the skin, respiratory system, and sensory systems. Use extreme caution while handling paraformaldehyde. The powder has the potential to become airborne.  Wear safety goggles and a face shield to prevent contact with the eyes.  Wear a respirator to avoid inhalation.  Wear gloves at all times. Wearing two gloves on each hand is recommended.  Only work with paraformaldehyde in a chemical fume hood.  It may be helpful to line the bench with paper towels that can be easily cleaned up in the event of a spill.  Clean all contacting surfaces with 70% ethanol, including gloves, sleeves.  Wear safety glasses and exercise caution when handling D-PBS to avoid contact with eyes. (DPBS is a saline solution and therefore an eye irritant).

8 This SOP requires the use of a chemical fume hood, a sterile tissue culture hood, and a fluorescence microscope. Working knowledge of all three is assumed for the purposes of these instructions. Refer to the respective SOPs in the Winter Lab for further help or ask for assistance. 9 Measurements in this SOP are based on a 9-well basis (10% extra) for a 96-well plate. Scale measurements accordingly for smaller or larger experiments (50 µL of each solution should go in each well and no solution is viscous enough that more than 10 to 20% extra needs to be prepared of each). 10 Staining on less porous surfaces may require smaller volumes, lower concentrations, and shorter times. Please consult the supplier for information relevant to your purpose. 264

The stain must only be opened in a sterile environment. Note that a sterile environment DOES NOT provide protection against chemical hazards. If chemical hazards are present work should be performed with a respirator or similar protective device.  Operators must perform indicated steps in the sterile, laminar flow tissue culture hood.  Wipe all surfaces with 70% ethanol.  Work as far back in the hood as possible.  Avoid reaching over all sterile surfaces.  Wear UV-resistant safety glasses while imaging the cells (UV can be harmful to eyes).

PPE: When actin staining all operators should wear latex or nitrile gloves, lab coat, and safety glasses. Face shield and respirator should be worn when handling paraformaldehyde. Butyl rubber gloves may also be worn when handling paraformaldehyde for extra protection.

When imaging cells all operators should wear latex or nitrile gloves, lab coat, and UV- protective safety glasses.

Procedure (Preparing Fixing Solution): 1. Heat 2.2 mL11 of PBS12 in a 10 mL beaker to 60 °C on a hot plate. It is not important that the solution is exactly 60 °C, only that it is heated. Stir with a very small stir bar. 2. While the solution is heating, add 88 mg sucrose (40 mg/mL) and 22 µL of 1M NaOH (10 µL/mL). 3. Once the sucrose has dissolved and the solution has been warmed to approximately 60°C, add 88 mg of paraformaldehyde (40 mg/mL). 4. Before fixing the cells with this solution, allow the solution to cool to room temperature after the paraformaldehyde has dissolved.

Procedure (Preparing Triton Solution): 1. Add 500 µL of PBS to a microfuge tube. 2. Add 1 µL of Triton (2 µL/mL) to the PBS. Mix the two liquids by pumping with the micropipette several times.

Procedure (Preparing Blocking Solution): 1. Add 500 µL of PBS to a microfuge tube. 2. Add 15 mg of BSA (30 mg/mL) to the PBS.

11 It may be difficult to make any less than 2 mL of this solution. 12 The staining procedure, and specifically the fixing, will kill the cells, so it is not necessary to use sterile PBS. The staining may and should be done in a chemical fume hood. 265

Procedure (Preparing Stain)13: 1. Add 461.5 µL of PBS to a microfuge tube. 2. Add 15 mg of BSA (30 mg/mL) to the PBS. 3. Add 38.5 µL of stain14.

Procedure (Staining): 1. Wash the gels twice with PBS by adding 50 µL to each well and swirling. Aspirate the spent PBS off of each gel by adjusting the micropipette to 40 µL, inserting the tip into the well along the wall of the well, and being careful not to aspirate the gel. 2. Fix the cells with 50 µL of the paraformaldehyde solution for 20 minutes. After 20 minutes, aspirate the spent fixing solution as with the PBS wash in step 1. The paraformaldehyde presents a strong biological hazard. It, along with all other solutions that subsequently contact the gels, must be disposed of as biohazardous waste. While completing the stain procedure, dispose of all hazardous solutions in a beaker of ethanol to be kept in the hood. When the procedure is over, all laboratory supplies that contacted paraformaldehyde must be doused with ethanol, and then cleaned as usual. 3. Wash another two times with PBS. 4. Extract with 50 µL of the Triton solution for 15 minutes. 5. Wash another two times with PBS. 6. Add 50 µL of the blocking solution to each well. Store the 96-well plate in a 4 °C refrigerator overnight. 7. Wash another two times with PBS. 8. Add 50 µL of the prepared stain to each well. Incubate the 96-well plate on a rocker in an incubator at 37°C and 5% CO2 for 20 minutes. After 20 minutes, move the well plate to the fridge and leave it overnight. 9. Wash another two times with PBS15. 10. Image the cells using fluorescent (UV) microscopy.

13 This solution should not be prepared at the same time as the first three, but rather the day after (see the staining procedure). 14 This step must be done in the sterile tissue culture hood to protect the integrity of the stain. 15 It may be necessary to add an agent such as Slow-fade to maintain the effects of the stain. This has not been necessary in past experience with the stain. 266

Fixation for SEM Imaging for Electrospun fibers

Purpose: SEM imaging is useful for generating high-resolution images that can show useful information about a sample’s topography. Before a sample can be imaged, it must be dehydrated (if used for cell culture) and sputter-coated with an electrically conductive element such as osmium or gold. This SOP outlines the procedure for dehydrating samples. Alternatively, flash freezing can also be performed for SEM.

Materials: Phosphate Buffer Solution (PBS) [Sigma P4417] Sucrose [Sigma S0389] NaOH [Sigma-Aldrich 480878] para-Formaldehyde [Sigma-Aldrich 158127] 100% Ethanol (EtOH) Hexamethyldisilazane (HMDS) Assorted micropipettes and tips

Safety Concerns and PPE: Para-formaldehyde is a toxin, teratogen, and suspected carcinogen. It is also an irritant to the skin, respiratory system, and sensory systems. Use extreme caution while handling paraformaldehyde. The powder has the potential to become airborne.  Wear safety goggles and a face shield to prevent contact with the eyes.  Wear a respirator to avoid inhalation.  Wear gloves at all times. Wearing two gloves on each hand is recommended.  Only work with para-formaldehyde in a ventilated chemical fume hood.  It may be helpful to line the bench with paper towels that can be easily cleaned up in the event of a spill.  Clean all contacting surfaces with 70% ethanol, including gloves and sleeves.  Wear safety goggles to prevent contact with eyes.  Wear gloves at all times.  Wear safety glasses and exercise caution when handling D-PBS to avoid contact with eyes.  Only work with hexamethyldisilazane in a ventilated chemical fume hood. Hexamethyldisilazane is toxic, flammable, reactive, and corrosive. It should always be handled with care.

PPE: When dehydrating all operators should wear latex or nitrile gloves, lab coat, and safety glasses. Face shield and respirator should be worn when handling para- formaldehyde. Butyl rubber gloves may also be worn when handling paraformaldehyde for extra protection.

267

Procedure: 1. Rinse any residual protein off of the surface with PBS. Wash all samples 3 times, allowing them to soak in the PBS for 5 minutes each time. 2. Fix the cells in 4% para-formaldehyde16 for 1 hour at room temperature in a chemical fume hood. 3. Rinse the samples in 0.1 M phosphate buffer with 0.1 M sucrose17. Wash all samples 3 times, allowing them to soak in the buffer solution for 5 minutes each time. 4. Dehydrate the samples with a wash of 50% EtOH and distilled water for 5 minutes. 5. Dehydrate the samples with a wash of 70% EtOH and distilled water for 5 minutes. 6. Dehydrate the samples with a wash of 80% EtOH and distilled water for 10 minutes. 7. Dehydrate the samples with a wash of 950% EtOH and distilled water for 10 minutes. 8. Dehydrate the samples with a wash of 100% EtOH twice for 10 minutes each. 9. Dry the samples with a wash of 3:1 EtOH:HMDS for 15 minutes in a chemical fume hood. 10. Dry the samples with a wash of 1:1 EtOH:HMDS for 15 minutes in a chemical fume hood. 11. Dry the samples with a wash of 1:3 EtOH:HMDS for 15 minutes in a chemical fume hood. 12. Dry the samples with a wash of 100% HMDS twice for 15 minutes each in a chemical fume hood. 13. Soak the samples in 100% HMDS overnight in a chemical fume hood18.

16 The procedure for making 4% para-formaldehyde may be found in many staining SOPs, such as the Actin Staining SOP. 17 This is PBS with sucrose and NaOH. It is made exactly like fixing solution, but without para- formaldehyde. 18 For plastic samples on tissue culture plates, insert a piece of filter paper beneath the sample to prevent it from sticking to the bottom of the well. 268

Unconfined Compression Test via RSA

Purpose: These tests can be used to extract the mechanical properties (e.g., elastic modulus) of substrates (e.g., hydrogels). For these tests, the information that are needed for input is the compressive rate and how long you want to operate at this rate until you reach your desired displacement.

Safety Concerns and PPE: PPE: When performing compression tests, all operators should wear latex or nitrile gloves, lab coat, and safety glasses.

Procedure: 1. Prepare the desired hydrogel samples for testing. 2. Login to the computer. 3. Turn on RSA3 (Power switch is the left back corner). 4. Open the software: TA Orchestrator 5. Control  Instrument Control Panel a. Turn Motor Power On 6. Attach plates using torque wrench. 7. Control  Set End of Test Conditions a. Make sure “no” is selected for all 8. Control  Gap Instrument Control a. Offset Force to Zero b. Zero Fixture 9. Utilities  Units Format a. Make sure metric units are selected 10. Control  Edit/Start Instrument Test a. Sample Geometry  Stored Geometries  Cylindrical Geometry i. Edit Geometry  Type in the diameter b. Test Setup  Predefined Test Setups  Multiple Extension Mode Test i. Test Type  Strain-Controlled ii. Measurement Type  Transient iii. Edit Test  Type in your desired strain input 11. Select the “Begin Test” option in the bottom left corner 12. To view different plots, right click on the plot  Plot Setup  Layout tab  double click the variables to add and remove for the x-axis and y-axis 13. To save the data for export into Excel  File  Export  select Excel ASCII (.txt). Identify the linear portion of the graph. The slope of this linear region will provide the elastic modulus of the material. 14. Remove plates 15. Turn motor off via Control  Instrument Control Panel and Close TA Orchestrator 16. Turn of RSA3.

269

Nanofiber Characterization using Image J

Purpose: This procedure outlines steps for characterization of electrospun nanofibers using image analysis tools (Image J).

Procedure: Fiber diameter measurements 1. Open scanning electron microscopy (SEM) images of electrospun nanofibers in Image J. 2. Using the line tool, measure the edge-to-edge distance of single nanofiber in the image. To be accurate, zoom in the image to clearly mark the edges of the nanofiber. 3. The “Set Scale” option will provide the distance in pixel which can then be converted to micron using the scale bar in the image. 4. Repeat these for multiple fibers and multiple images from nanofiber samples.

Fiber density measurements 1. Open scanning electron microscopy (SEM) images of electrospun nanofibers in Image J. 2. Using the line tool, draw a line perpendicular to the direction of alignment of nanofibers. 3. Manually count the number of nanofibers crossing the line and repeat for at least 3 randomly chosen locations in the image. Also, repeat for multiple images from nanofiber samples. 4. Divide the number of nanofibers by the length of the line and convert to number of fibers/mm.

270

Fast Fourier Transform (FFT) Analysis of Aligned Electrospun Nanofibers

Purpose: This outlines steps for performing FFT analysis of aligned electrospun nanofibers to compare alignment of different samples. This function is used to convert the images representing data (e.g., SEM image of an electrospun nanofiber scaffold) to a mathematical domain that provides the distribution of pixel intensities indicating scaffold anisotropy.

Procedure: 1. Open scanning electron microscopy (SEM) images in Image J. Perform FFT using “Process” option in Image J toolbar. 2. Use the FFT image to quantify the pixel distribution/gray values using Oval profile Plugin (available at http://rsbweb.nih.gov/ij/plugins/oval-profile.html) 3. Select the FFT generated pixels using an oval tool and perform radial sums over 100 points to generate a graph of gray value versus degree. 4. For aligned nanofiber samples, this would have 2 peaks because of symmetry. Since the default degree is already set in Image J, the position of the peaks can be changed and can be set arbitrarily at 90º and 270º as reported in literature for these types of scaffolds. 5. Normalize all gray values to have a maximum value of 1 and a minimum of 0 and plot these profiles for all types of aligned nanofibers studied for comparing the degree of scaffold alignment.

271

Correcting Drift in Migration Movies Using Stack Correction Plug-in in Image J

Purpose: This outlines steps for correcting drift in movies by aligning stacks as a function of time.

Procedure: 1. This plug-in can be downloaded from http://bigwww.epfl.ch/thevenaz/stackreg/. Acknowledge the use of this plug-in in journal articles/reports. 2. There are four types of transformation listed (i.e., translation, rigid body, scaled rotation, affine) with their associated description. 3. Apply the necessary transformation by opening the movies in Image J. The transformation may vary depending on the type of drift in movies. For example, if the drift is only in one direction, a translation transformation could be useful to correct for this drift. Always recheck to see if the stacks have been aligned well after applying the transformation. 4. Save the corrected stack for further analysis of movies.

272