University of Florida Thesis Or Dissertation Formatting

ENGINEERING AN IN SITU OXYGEN GENERATING MATERIAL FOR EXTRAHEPATIC ISLET TRANSPLANTATION

MARIA MARGARITA CORONEL SALAS

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2016

To my mother, for her love and support

ACKNOWLEDGMENTS

Time flies. I first joined the Stabler lab during my junior year as an undergrad. I had no idea of how to hold a pipette but was very excited to be given a chance to immerse myself in the beautiful, and most times frustrating, world of research. Fast forward 8 years and here I am still fascinated with the work I’ve chosen. First and foremost, I owe a deep sense of gratitude to my mentor Dr. Cherie Stabler. I would not have been able to complete this work without her support both personally, and professionally. Her dedication, professionalism, scientific approach, and above all her profound interests for her students are some of the values I have learned from her, and will take with me in my new scientific endeavors. It has been a great honor to work under your supervision.

I would like to express my deepest thanks and sincere appreciation to Eileen

Pedraza, Jessica Weaver and Irayme Labrada. You three have provided me with invaluable guidance, support, and encouragement through different stages of my career. For all this and for more, thank you. I would also like to thanks all members of the Stabler lab past and present. The time I spent working along you has definitely made me a better scientist.

I would also like to express my deepest thanks and sincere appreciation to my committee members, Dr. Benjamin Keselowsky, Clayton Matthews, and Peter

McFetridge for taking time out of their busy schedules to be part of my dissertation committee as well as their constructive criticism and direction.

To Enrico, my rock in this last year, thank you for your patience and unconditional love. I would not have made it without you. Last but not least, I can never be grateful enough to my mother who supported my dream as a teenager to leave her

side to achieve my professional dreams. I would not be here today without her constant encouragements and support.

TABLE OF CONTENTS page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 9

LIST OF FIGURES ...... 10

LIST OF ABBREVIATIONS ...... 16

ABSTRACT ...... 18

CHAPTER

1 RATIONALE ...... 20

Introductory Remarks...... 20 Specific Aims ...... 21 Contents of Dissertation ...... 23

2 ENGINEERING AN EXTRAHEPATIC TRANSPLANT SITE ...... 25

Background ...... 25 Selecting the Optimal Engraftment Site ...... 26 3D Scaffolding ...... 28 Oxygenation and Vascularization ...... 31 Conclusion ...... 34

3 ENGINEERING AN IN SITU OXYGEN GENERATOR ...... 37

Background ...... 37 Methods ...... 39 Materials ...... 39 Fabrication of Oxygen Generator ...... 39 Leaching Studies ...... 40 Macroscopic Analysis ...... 40 Microscopic Analysis ...... 40 Surface Characterization ...... 41 Hydrogen Peroxide Measurements ...... 42 Calcium Concentration ...... 43 Non-invasive Oxygen Measurements ...... 43 Mathematical Model ...... 44 Computational Modeling ...... 46 Islet Isolation and Culture ...... 47 In Vitro Cellular Assessments ...... 48 Statistics ...... 49 Results ...... 49

Conclusion ...... 63

4 MITIGATING HYPOXIC STRESS ON PANCREATIC ISLETS VIA IN SITU OXYGEN GENERATION ...... 91

Background ...... 91 Research Design and Methods ...... 92 Materials ...... 92 Islet Isolation and Culture ...... 92 Fabrication of Oxygen Generating Material, OxySite ...... 93 Oxygen Studies with Pancreatic Islets ...... 93 In Vitro Cellular Assessments ...... 94 Islet Transplant and Graft Assessment ...... 97 Histological Assessment...... 98 Statistical Analysis ...... 99 Results ...... 99 Conclusion ...... 109

5 MACROPORUS SCAFFOLDS WITH ENHANCED OXYGENATION FOR EXTRAHEPATIC ISLET TRANSPLANTATION ...... 127

Background ...... 127 Methods ...... 129 Materials ...... 129 Composite Scaffold-OxySite Fabrication ...... 129 Non-invasive Oxygen Measurements ...... 130 Scanning Electron Microscopy ...... 131 Islet Isolation and Culture ...... 131 Islet Transplant and Graft Assessment ...... 131 Histological Assessment...... 133 Statistical Analysis ...... 134 Results ...... 134 Conclusions ...... 139

6 IN SITU OXYGENATION OF IMMUNOISOLATING DEVICES ...... 149

Background ...... 149 Methods ...... 151 Materials ...... 151 OxySite Fabrication ...... 151 MIN6 Culture ...... 151 Islet Culture ...... 152 Immunoisolating Device Fabrication ...... 152 In vitro Assessments ...... 153 Transplant and Graft Assessment ...... 153 Histological Assessment...... 154 Ex Vivo Graft Assessment ...... 154

Statistics ...... 155 Results ...... 155 Conclusions ...... 164

7 OPTIMIZATION OF OXYGEN GENERATION BY ADDITION OF OSMOTIC AGENTS ...... 181

Background ...... 181 Methods ...... 182 Materials ...... 182 Fabrication of Oxygen Generator with Osmotic Agents ...... 182 Macroscopic Analysis ...... 182 Microscopic Analysis ...... 183 Swelling Measurements ...... 183 Non-invasive Oxygen Measurements ...... 184 Hydrogen Peroxide Measurements ...... 184 MIN6 Culture ...... 184 MIN6 In Vitro Cultures ...... 185 In Vitro Cellular Assessments ...... 186 Statistics ...... 186 Results ...... 187 Conclusion ...... 191

8 SUMMARY ...... 202

Conclusion ...... 202 Future Work ...... 207

APPENDIX

A LOADING DENSITY CALCULATION FOR COMSOL MODELING ...... 210

B BIOCOMPATIBILITY OF SCAFFOLD OXYSITE COMPOSITE MATERIAL ...... 213

C LOADING TITRATION OF AGAROSE XENOGRAFT TRANSPLANT MODEL .... 220

D LOADING TITRATION OF AGAROSE ALLOGRAFT TRANSPLANT MODEL ..... 224

E OXYGEN GENERATING BIOMATERIAL SUPPORTS SCHWANN CELL VIABILITY IN A MODEL OF SPINAL CORD INJURY ...... 228

F OPTIMIZATION OF AGAROSE CONCENTRATION AND CASTING PROCESS FOR THE FABRICATION OF IMMUNOISOLATING DEVICES ...... 235

REFERENCES ...... 241

BIOGRAPHICAL SKETCH ...... 262

LIST OF TABLES

Table page

4-1 List of oligonucleotide primers used in RT-PCR analysis ...... 115

5-1 List of oligonucleotide primers used for gene profiling ...... 143

6-1 Primer Sequence for PCR Array...... 169

LIST OF FIGURES

Figure page

2-1 Various novel approaches implemented within islet transplants to improve overall function and efficacy include: ...... 36

3-1 Schematic diagram of OxySite disks dimensions in cylindrical coordinates (z,r). ... 71

3-2 Macroscopic characterization of OxySite material. A) Cartoon representation of OxySite disks. Inset represents actual photograph of material post fabrication...... 72

3-3 Porosity assessment of OxySite material via mercury porosimeter. A) Changes in porosity of OxySite material compared to a blank PDMS disks as well as after different leaching times...... 73

3-4 Percent volume intruded vs pore diameter distribution for different OxySite disks. A) Blank PDMS after 1 day leaching, B-D) OxySite after 1-7-100 days post leaching respectively...... 74

3-5 Surface characterization of OxySite disks. A) FTIR spectrum of OxySite disks (magenta, green, and blue), blank PDMS (black), and CaO2 powder (red)...... 75

3-6 Surface characterization of OxySite disks via SEM. Representative images of the surface and a cross section of the OxySite material 15 days post leaching are shown...... 76

3-7 Surface characterization of OxySite disks via LSM. Representative images of the surface of the OxySite material 1, 10, and15 days post-leaching are shown...... 77

3-8 Illustration representing sample collected via nanoCT. Brighter voxels within the samples were observed at different positions within the disk...... 78

3-9 Table representation of the different layers of high density voxels found within the OxySite material for all time intervals evaluated...... 79

3-10 Concentration of calcium peroxide reaction byproducts in supernatant at different leaching time points...... 80

3-11 Comprehensive oxygen kinetic release from OxySite disks of 10 mm in diameter for 30 days. Chamber was refreshed every day. Samples were measured for 6 to 24 hours...... 81

3-12 Mathematical model of oxygen release from OxySite disks. A) Cumulative release of oxygen release from OxySite disks collected via noninvasive oxygen sensing chambers ...... 82

3-13 Oxygen release kinetic and mathematical modeling of release kinetic profiles of OxySite disks of different dimensions...... 83

3-14 Oxygen release kinetic and mathematical modeling of release kinetic profiles of OxySite disks of different dimensions...... 84

3-15 Experimental design and parameter implemented for COMSOL modeling...... 85

3-16 Representative FEM simulation models illustrating effects of loading densities and external oxygen tensions on oxygen availability to pancreatic islets after 8 h of culture either at: ...... 86

3-17 Representative FEM simulation models illustrating effects of loading densities and external oxygen tensions on oxygen availability to pancreatic islets after 24 h of culture either at: ...... 87

3-18 Bright field images of islets culture under standard oxygen and loading density (Free Islets); standard oxygen and high loading density with a PDMS blank control (0.2 mM Oxygen Control)...... 88

3-19 Viability and function of rat islets exposed to OxySite 15 d post leaching and incubated for 48 h at 0.01 mM Oxygen...... 89

3-20 mRNA expression and confocal images of islets exposed to OxySite 15 d post leaching and incubated for 48 h at 0.01 mM Oxygen...... 90

4-1 Immunohistochemistry of whole islets after exposure to low oxygen conditions for 24 h. Hypoxia of Po2 <10 Torr is indicated by the oxygen sensitive hypoxyprobe dye (green)...... 116

4-2 OxySite mitigates activation of second stage hypoxic and apoptotic markers, leading to preservation of metabolic activity and viability...... 117

4-3 Culture of pancreatic islets with OxySite enhances islet viability and reduces activation of pro-apoptotic markers in culture long term...... 118

4-4 OxySite mitigates activation of anaerobic glycolysis and preserves glucose stimulated insulin release for rat pancreatic islets cultured under hypoxic conditions...... 119

4-5 Co-culture of islets with OxySite maintains islets OCR under hypoxia. A) ATP concentration was reduced in islets exposed to chronic hypoxia (24 h) with a blank disk (white bars) compared to OxySite treated islets (black bars)...... 120

4-6 OxySite reduces activation of anaerobic glycolysis in nonhuman pancreatic islet cultures...... 121

4-7 OxySite alters expression of inflammatory regulators and effectors of rat pancreatic islets under hypoxic conditions...... 122

4-8 Oxysite prevents hypoxia mediated activation of pro-inflammatory cytokines in non human primate cultures. NHP Islet were cultured for short (8 h) and long term (24 h) under continuous hypoxia ...... 123

4-9 Impact of OxySite on angiogenic pathways and VEGF protein release for low oxygen cultures of rat islets...... 124

4-10 OxySite does not hinder angiogenic potential of non-human pancreatic islet cultures...... 125

4-11 Co-culture of rat islets with OxySite resulted in enhanced functional outcomes when transplanted into diabetic syngeneic recipients...... 126

5-1 Optimization of macroporous scaffold as a platform for high loading density islet transplants...... 143

5-2 Scanning elecetron micrographs of composite scaffold with OxySite disk. Left panel represents top surface of the scaffold. Middle panel is the bottom layer macroporous scaffold...... 144

5-3 Transplantation of original size macroporous scaffold in a syngeneic lewis rat model...... 145

5-4 Transplantation of optimized macroporous scaffold in a syngeneic Lewis rat model. A) Photograph of composite OxySite-scaffold used for implantation (6 mm in diameter, 5 mm in thickness)...... 146

5-5 mRNA expression from PCR array of SS-OxySite scaffolds transplanted in the OP pouch of diabetic male rats for 30 d...... 147

5-6 Immunohistochemistry evaluation of grafts explanted 30 days post-transplant, stained with anti-insulin (green), tomato lectin (red), and nuclear staining (blue)...... 148

6-1 Schematic representation of immunoisolating agarose devices implemented for in vitro assessments. The yellow region represents the OxySite material while the blue region represents the 2% w/v agarose cells mixture...... 169

6-2 Schematic representation of immunoisolating agarose devices implemented for transplantation. The yellow region represents the OxySite material while the blue region represents the 2% w/v agarose cells mixture...... 170

6-3 Effects of OxySite supplementation in immunoisolated devices containing pancreatic islets exposed to low oxygen conditions in vitro...... 171

6-4 mRNA expression of 4000 IEQ islets encapsulated within an agarose hydrogel with an OxySite disk or with a blank PDMS disk under low and standard oxygen conditions...... 172

6-5 In vivo data of syngeneic imunoisolating device transplants in female Lewis rats. 173

6-6 Histological analysis of immunoisolated devices transplanted into the omental pouch of diabetic female recipients...... 174

6-7 In vivo data of allograft imunoisolating device transplants in mice. A) non-fasting blood glucose of animals post-transplant shows stark difference in normoglycemic reversal times between ctrl and OxySite group...... 175

6-8 Live Dead images of agarose hydrogels sections 33 d post-transplant. Top row represent agarose hydrogels containing a blank PDMS disk (control group). .. 176

6-9 Histological assessments of samples 33 d post-transplant. Top row represent agarose hydrogels containing a blank PDMS disk (control group). Bottom row represents constructs containing an OxySite disk...... 177

6-10 Ex vivo assessments of islets transplanted in encapsulated immunoisolated devices in diabetic b/6 mice. Grafts were electively removed 5 d post- transplant. Islet function was assessed via static glucose stimulated insulin release test...... 178

6-11 Ex vivo assessments of islets transplanted in encapsulated immunoisolated devices in diabetic b/6 mice. Grafts were electively removed 15 d post- transplant...... 179

6-12 In vivo data of xenograft imunoisolating device transplants in mice. A) non- fasting blood glucose of animals post-transplant shows no difference in normoglycemic reversal times between ctrl and OxySite group...... 180

7-1 Addition of osmotic agents to OxySite material can enhance porosity. A) Cartoon representation of OxySite material with added osmotic agents. Inset represents pore formation after water dissolution of excipients from polymer. . 196

7-2 Optimization of OxySite material by addition of PEG 200 as a porogen. A) Cartoon representation of OxyO material after addition of PEG 200. Inset represents pore formation after water dissolution of excipients from polymer. . 197

7-3 Incorporation of PEG200 in oxygen generating material can improve matrix porosity. A) Table comparison of porosities of OxySite material compared to OxyO after 1 d post leaching...... 198

7-4 Comprehensive oxygen kinetic measurement for OxyO sample of 10 mm in diameter. Oxygen tension measured via a non-invasive sensing system in a temperature controlled environment...... 199

7-5 Long term co-culture of beta cells with OxyO and OxySite under low oxygen tension...... 200

7-6 Hydrogen peroxide concentration of Blank PDMS (open circles) disks, OxySite disks (black squares), and OxyO disks (Grey open triangles) at selected time points for 60 days ...... 201

A-1 Schematic representation of homogeneous distribution of islets within surface area of insert...... 212

B-1 SEM images of surface and cross section of composite scaffold material. No direct effects on pore structure was observed by the addition of an OxySite disk to the bottom of the scaffold...... 216

B-2 Histological analysis of composite scaffolds transplanted in the omental pouch and skin of normal Lewis female rats after 30 days post-transplant...... 217

B-3 In vivo assessment of scaffold-OxySite composite material. Scaffolds were loaded with 1800 IEQ and transplanted in the omental pouch of diabetic Lewis female recipients...... 218

B-4 Histological analysis of explanted graft 35 d post-transplant. Good host integration was observed on the scaffold side (top)...... 219

C-1 Summary of all islet loading implemented (yellow inset) and the corresponding vol/vol loading density in the graft (red inset)...... 222

C-2 Representative non fasting blood glucose (top panel) and body weight (bottom panel) of all the different loading densities tested in vivo...... 223

D-1 Summary of all MIN6 loading implemented (yellow inset) and the corresponding vol/vol loading density in the graft (red inset). MIN6 loading was normalized to an IEQ using the conversion 1EQ=2000 cells...... 226

D-2 Representative non fasting blood glucose (top panel) and body weight (bottom panel) of all the different loading densities tested in vivo...... 227

E-1 Luxol fast blue (LFB) staining shows the morphology of myelin in the graft site. (A) sham experiment no material, (B) collagen only, (C), collagen+PDMS and (D) Collagen+O2 ...... 231

E-2 Immunofluorescence of Schwann cell GFP positive cells, with nuclear staining and a neurofilament marker. (A) Collagen only, (B) Collagen +PDMS and (C) collagen + O2...... 232

E-3 Immunofluorescence of Schwann cell GFP positive cells, with nuclear staining and a neurofilament marker and a neurofilament protein Top (A) Collagen +PDMS and (B) collagen + O2...... 233

E-4 Metabolic activity of 3D scaffolds containing embedded schwann cells in the presence of the oxygen generator culture under normoxic or hypoxic conditions ...... 234

F-1 Summary of agarose polymers implemented in optimization of fabrication protocol with respective tensile strengths and gelling temperature...... 238

F-2 Non-fasting blood glucose levels for mice receiving agarose immunoisolated devices from a subset of agarose polymers tested...... 239

F-3 Images illustrating improved agarose immunoisolated device fabricated using optimized manufacturing protocol...... 240

LIST OF ABBREVIATIONS

BMP-2 bone morphogenetic protein-2

CIT clinical islet transplantation

CMRL Connaught Medical Research Laboratorie medium s

DMEM dulbecco's modified eagle medium

ECM extracellular matrix

ELISA enzyme-linked immunosorbent assay

FBS fetal bovine serum

FGF fibroblast growth factor

FTIR fourier transform infrared spectroscopy

H&E hematoxylin and eosin

HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid )

HIF-1 hypoxia-inducible factor 1

IBMIR instant blood-mediated inflammatory reaction

IEQ islet equivalents

IP intraperitoneal

IV intravenous

MCP-1 monocyte chemoattractant protein-1

MMP matrix metalloproteinase

MSC mesenchymal stem cells

NO nitric oxide

PBS phosphate buffered saline

PDGF platelet-derived growth factor

PDMS poly dimethyl siloxane

PEG polyethylene glycol

PLGA poly(lactic-co-glycolic acid)

SEM scanning electron microscope

STZ streptozotocin

T1DM type 1 diabetes mellitus

VEGF vascular endothelial growth factor A

XPS x-ray photoelectron spectroscopy

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

ENGINEERING AN IN SITU OXYGEN GENERATING MATERIAL FOR EXTRAHEPATIC ISLET TRANSPLANTATION

Maria M. Coronel Salas

December 2016

Chair: Cherie Stabler Major: Biomedical Engineering

A main limitation in the translation of cellular based therapies is the deleterious oxygen gradients generated within these mainly avascular grafts. Pancreatic islet therapy, a potential treatment for type I diabetes mellitus (TIDM), is more susceptible to this harmful effect, due to the need for increased loading densities for clinical relevance and their elevated susceptibility to hypoxia-induced death and dysfunction. The objective of this thesis was to engineer an in situ oxygen generating biomaterial capable of improving oxygen availability and mitigating the harmful effects of hypoxic microenvironments. Our hypothesis was that a controllable and predictable in situ oxygen generating biomaterial would enhance islet viability and function, resulting in improved engraftment of the final three-dimensional implants. Thus, the primary objective of this thesis was to develop an in situ oxygen generator to enhance islet viability and function in 3D matrices for extrahepatic islet transplantation. Detailed characterization of physical and chemical material properties of these oxygen generating materials was performed, while the kinetics of oxygen release from the material were evaluated. The effect of oxygenation on cell viability, function, and

activation of inflammatory pathways was assessed for beta cells and islets in vitro and

in vivo. Translation of the biomaterial to three-dimensional constructs was screened

using an islet-engineered 3D construct in a chemically induced model of rodent

diabetes. It is our belief that by improving oxygenation to the cellular constructs, via the

introduction of an in situ oxygen generator, we can mitigate islet apoptosis and the

progressive metabolic burnout of the cellular graft, thus substantially improving

engraftment and long-term success. Resulting implants should enhance the potential of

cellular based therapies as an alternative treatment for TIDM.

CHAPTER 1 RATIONALE

Introductory Remarks

The development of tissue engineered cell-based therapies has emerged as a promising alternative to treat or repair malfunctioning tissues. While there are many potential and exciting applications for tissue engineered cell constructs, the ability to transplant large, functional structures has commonly been hindered by problems associated with inadequate nourishment. Specifically, a lack of proper oxygenation in the cellular based scaffold, particularly in the early stages after transplantation, is a

substantial challenge.

Supporting proper oxygenation is particularly challenging in clinical islet

transplantation (CIT) due to the high sensitivity of beta cells to variations in oxygen

tension. Islets in their native environment are richly vascularized by arterial blood flow.

Their dense vasculature provides the necessary nutrient support to maintain the high

metabolic demand of these cells, while also allowing for an immediate physiological

response to glucose changes. During isolation, the vascular network is stripped away.

Even following infusion into the microvasculature, such as the hepatic portal vein site

used in CIT, islets are exposed to lower oxygen tensions than those in their native

pancreas and they fail to appropriately revascularize over time. As a result, a significant

portion of the transplanted beta cell mass is lost in the early post-transplant period.

Furthermore, exposure of islets to hypoxia not only leads to cellular death, it inhibits the insulin responsiveness of the residing beta cells, while also elevating the release of stress factors such as inflammatory cytokines and the accumulation of detrimental free radicals in the surrounding milieu. As such, despite successful reversal of

hyperglycemia in patients, long-term success of these grafts has been hindered, in part, by their inadequate oxygenation.

Although many approaches have focused on enhancing and hastening the

revascularization process, acute hypoxia exposure, even for short periods of time,

incurs a long term impact on the engraftment and function of islet transplants. Providing

supplemental oxygenation, even only in the period prior to revascularization, can be

highly beneficial in preventing graft metabolic burnout and activation of local immune

responses, while supporting the engraftment and function of the implants. In situ oxygen

delivery is a highly promising approach, as it allows for the local delivery and control of

oxygen at the graft site. Previous approaches for localized oxygen delivery have

included equipment and/or techniques that are not easily translated into the clinic. Some

even require external lines, which are cumbersome and prone to infections in immune

compromised patients. In our approach, we seek to engineer a biomaterial of a relevant

geometry and design capable of delivering oxygen in situ.

Specific Aims

The long term objective of this project is to engineer a bioactive biomaterial that can provide an elegant means to spatially and temporally control the localized delivery of oxygen to cellular based grafts. It is our central hypothesis that the introduction of a controllable oxygen-generating device can mitigate hypoxia induced cellular death and enhance the viability and function of islet grafts. We plan to address this hypothesis via the following specific aims:

Specific Aim 1. Engineer an in situ oxygen generating biomaterial with

controllable and predictable release kinetics capable of mitigating hypoxic development

in cellular constructs. To accomplish this aim, a bioactive material was fabricated by

encapsulating solid peroxide in a PDMS matrix with or without osmotic agents.

Extensive physical and chemical analysis of the material properties was performed to fully characterize the final biomaterial prototypes. Fabrication processes were optimized based on the release kinetics obtained via non-invasive oxygen measurements. Further, a mathematical model of kinetic release was developed and fitted to the experimental data collected. Finally, a multiphysics computational model was developed to characterize oxygen gradient and to optimize in vitro culture systems for further

analysis.

Specific Aim 2. Evaluate the capacity of the in situ oxygen generator to mitigate hypoxia-induced activation of apoptotic pathways and reduce inflammatory processes.

To accomplish this aim, a comprehensive mechanistic study of the ability of the biomaterial to mitigate hypoxia-induced-apoptosis or necrosis and pro-inflammatory process was assessed in a 2D model of pancreatic culture. Further, the ability of this material to support viability and maintain the glucose responsive function of pancreatic islets was tested in an in vivo diabetic rodent model.

Specific Aim 3. Assess the ability of an in situ oxygen generator to enhance the viability and function of 3D based cellular constructs in an animal transplant model. To accomplish this aim, the capacity of the oxygen generating biomaterial to support islet graft viability and function in an extrahepatic site was screened in both macroporous and completely immunoisolated devices. Moreover, ex vivo assessments of graft function was performed at early and late time points to assess the effect of in situ oxygenation in the graft microenvironment and highlight the impact of supplemental oxygenation on islet health in vivo.

Contents of Dissertation

The main objective of this dissertation was to engineer an in situ oxygen generating material for extrahepatic islet transplantation. Most of the chapters herein

are adaptions of work published or submitted for publication with appropriate copyright

permissions. Each chapter provides a rationale for the research performed, as well as

detailed information on methods employed and results obtained, with a discussion of implications for the specific aim.

The three main tasks of this project were: (1) to fully characterize the material properties of our oxygen generator to develop a predictive mathematical model of release kinetics, based on reactive material loading density and biomaterial geometry;

(2) perform a comprehensive in vitro characterization of the effects of in situ oxygenation on pancreatic islet culture exposed to low oxygen tensions, in terms of activation of HIF-1α and downstream regulators and its effect on glucose stimulated insulin release and general islet quality; and (3) incorporate our bioactive material into two distinct platforms for extrahepatic islet transplantation to elucidate the effects of oxygenation in the viability of avascular 3D devices or as a bridge to functional vascular network formation within 3D macroporous devices.

This thesis consists of six distinct topics presented in Chapters 2-7. Chapter 2

provides background information on the pathophysiology of type I diabetes, current

approaches for the treatment of type I diabetes, and the significance of engineering an

extrahepatic site for islet transplantation and approaches to overcome limitations of

such sites. Chapter 3 evaluates the engineering of a predictable and controllable

bioactive material for in situ oxygen generation, as well as describes a mathematical

model for predictive release kinetics and a computational model to optimize in vitro

culture systems based on islet loading density and external oxygen tensions. Chapter 4 provides a comprehensive evaluation of the effects on in situ oxygenation on islet cultures at high loading densities under detrimental external oxygen tensions. Chapter 5

aims at translating this bioactive material to a 3D macroporous device for extrahepatic

islet transplantation in a syngeneic rat model. Focus is centered on assessing the effect

of the oxygen generating material on graft efficacy in vivo, as well as on graft

revascularizaton. Chapter 6 continues with the translation of the oxygen generator in

vivo in a 3D macroencapsulating device. The main aim of this chapter is to assess the

impact of in situ oxygen delivery on graft viability and transplant outcomes without the

interference of graft revascularization in syngeneic, allograft, and xenograft models of

murine diabetes. Chapter 7 aims to improve the duration of oxygen delivery by

incorporating osmotic agents into the bioactive material formulations. The goal of these

studies is to provide characterization of biomaterial properties, as well as to assess its

effects in an in vitro culture system. Chapter 8 provides a summary of the work

discussed in this dissertation and suggestions for future research.

CHAPTER 2 ENGINEERING AN EXTRAHEPATIC TRANSPLANT SITE

Background

Type 1 Diabetes Mellitus (T1DM) results from the autoimmune destruction of β-

cells within pancreatic islets of Langerhans. While exogenous insulin, supplemental

agents, and dietary regulation can provide reasonable management of blood glucose

and mitigate secondary complications, only the replacement or regeneration of β-cells can offer physiological glycemic control. Clinical Islet Transplantation (CIT) represents a viable treatment option for T1DM. The current procedure involves the infusion of allogeneic cadaveric islets into the hepatic portal vein of the recipient. While most CIT recipients exhibit stable glycemia, decreased hemoglobin A1c, and elimination of hypoglycemia for multiple years, long–term (> 5 years) insulin independence has been achieved in only a subset of patients [1].

Graft failure in the long-term can be explained, in part, by the chosen transplant site. Infusion of the islets into the hepatic vasculature leads to an instant blood mediated inflammatory reaction (IBMIR), resulting in islet necrosis. Delay in graft revascularization and the ensuing hypoxia further exacerbate islet loss. Moreover, the higher concentrations of drugs and toxic by-products in the hepatic microenvironment can

dampen the function and engraftment of the transplanted islets [2].

While the initial success of islet transplantation highlights the potential of these

cells to treat T1DM, the substantial loss of islets post-transplantation directs current

research towards engineering superior transplant microenvironments. Ideally the

Coronel, Maria M., and Cherie L. Stabler. "Engineering a local microenvironment for pancreatic islet replacement." Current opinion in biotechnology 24.5 (2013): 900-908.

appropriate 3D niche should be tailored to the specific requirements of islets, customized to the natural, unique extracellular matrix architecture of the native pancreas. To meet the high oxygen demand of the islets, it can be supplemented with oxygen prior to competent vascularization, while simultaneously directing and accelerating the revascularization process. A defined engineered microenvironment can also provide a platform for local modulation of the inflammatory and immunomodulatory process, thereby reducing the systemic burden. Finally, a defined transplant site permits noninvasive assessment and the safe retrieval/replacement of the graft, if necessary.

This chapter provides a description of the current research in engineering an optimal transplant site (Figure 2-1) and how a combinatorial approach can lead to dramatic improvements in the long-term success of islet transplantation.

Selecting the Optimal Engraftment Site

Engineering the optimal transplant microenvironment should begin with selecting the appropriate implant location. As evident from the native pancreas, vascular density and portal drainage of the site should be important factors. The sensitivity of islets to nonspecific inflammation during the peri-transplant period, with inflammatory reactions and oxidative stress promoting β-cell apoptosis, calls for the selection of a site with minimal inflammatory reactivity [3, 4]. Logistical factors, such as ease of implantation and adequate space of housing device, should also be taken into consideration.

Numerous locations have been explored as alternate islet implantation sites, with variations in efficacy and reproducibility [5]. Clinically relevant sites include: intraperitoneal; intramuscular; subcutaneous; omentum; and venous sac. The intraperitoneal cavity is amendable to greater transplant volumes, making it a popular site for encapsulated islets; however, it is challenged by lower oxygenation, limited

capacity for revascularization, and delays in glucose responsiveness. Transplantation within striated muscles has shown promise as an alternate transplantation site, with recent clinical efficacy using autologous islets [6]. Its high oxygen tension, as well as

rapid revascularization, makes the intramuscular location a favorable environment for

islets; however, care must be taken in the implantation procedure to minimize clumping

and mechanical stress. The subcutaneous space is a highly desirable location due to

ease in accessibility and transplantation. Its potential has been limited by mechanical

stress and inflammation, with suboptimal vascularization and systemic circulation

drainage [7], although recent approaches using devices and/or biomaterials to develop

it as a more favorable site are promising [8, 9]. The omentum shows potential as a

transplant site, whereby the tissue can be folded or wrapped to create a transplant

pocket, permitting flexibility in transplant volumes [10, 11]. It is well vascularized with

portal drainage, and has demonstrated properties for enhancing the graft

microenvironment, such as controlling the spread of inflammation, and promoting

reconstruction and tissue regeneration [12]. A recent report by Kakabadze et al.

explored the potential of an isolated venous sac, nourished by its vasa vasorum, for

transplanted islets [13]. Results in rat isografts were highly promising, with improved

efficacy compared to the intraportal site. Alternatively, several locations have been

investigated for their immuno-privileged properties, with recent works of interest in the

lymph node [14] and the anterior chamber of the eye [15]. While likely not immune-

privileged per se, these sites may provide a less inflammatory microenvironment.

Overall, careful consideration of the properties of the graft site is a critical component for

optimal islet engraftment and function.

3D Scaffolding

In their native environment, islets are embedded within extracellular matrix

(ECM) proteins, typically composed of interstitial matrix and basement membrane (BM) proteins, predominated by collagen type IV, laminin, and fibronectin [16]. These dynamic 3D structures play an instructive role in islet survival, function, and proliferation. β-cell-ECM interactions also play a critical role in the activation of NF-κB signaling, a critical pro-inflammatory regulator [17]. Moreover, BM proteins influence endothelial cell attachment and migration [18], thus facilitating vascular network growth.

Isolation procedures, however, disrupt this microenvironment, resulting in islet apoptosis via an anoikis-like pathway [19].

The introduction of a 3D platform can provide a spatial surrogate to the isolated islets, with the benefits of islet distribution and mechanical protection. Architectural properties, such as porosity and pore size, play an important role in the design of these platforms. While microporous designs, which prevent cellular migration into or out of the implant, have the desirable property of dampening immune attack, the engineering of immunoisolating, micro-scale devices is greatly challenged by diffusion length scales.

Macroporous scaffolds, characterized by pores larger than 50 µm, can provide an open environment for housing islets. This open framework supports efficient nutrient delivery and waste removal, while permitting infiltration of host cells within the implant. This accessibility for cellular infiltration facilitates the deposition of host ECM and the formation of an intra-graft / intra-islet capillary network. Various biomaterials have been proposed for creating these scaffolds [20]. Synthetic polymers are commonly employed, due to their stability, reproducibility, and ease of functionalization. A popular choice for biodegradable macroporous scaffolds is poly(lactic-co-glycolic acid) (PLGA). The

biodegradability of the PLGA can promote host integration and ECM formation. Gibly et al. demonstrated the ability of a macroporous PLGA scaffold to improve islet survival and engraftment in both rodent and porcine models, while allowing for cellular infiltration and revascularization [21]. Additionally, a macroporous scaffold made from Ethisorb, a composite degradable polymer made from a blend of polyglycolic acid (e.g. VICRYL) and poly-p-dioxanone (PDS), was shown to be effective in both nonhuman primate and dog allografts [22, 23]. Alternatively, biostable scaffolds can illustrate similar outcomes, such as the one recently reported by our lab using polydimethyl-siloxane (PDMS) [24].

Islet viability within these PDMS macroporous scaffolds was enhanced under low oxygen culture conditions, when compared to microporous scaffolds of comparable dimensions. Furthermore, stabilization of glycemia and independence from exogenous insulin was observed in both rodent and nonhuman primate models [24, 25].

The introduction of proteins or ligands responsible for cell-ECM or cell-cell

interactions onto the base 3D material scaffold can further minimize islet anoikis by

recapitulating their native microenvironment. The significance of β-cell-ECM interactions

has been highlighted heavily in the field via culture of islets atop surfaces coated with

selected BM proteins [26, 27]. When incorporated within 3D matrices, such as the

poly(ethylene glycol) (PEG)-based hydrogels utilized by Weber et al., reduction of

apoptotic pathways and improved glucose-stimulated insulin-secretion, for both a β cell

line and murine islets, was observed [28, 29]. In translating this in vivo, Salvay et al.

showed a decreased mean time to euglycemia and improved engraftment, when

embedding islets in a PLGA macroporous scaffold functionalized with BM proteins,

specifically collagen IV [30]. Alternatively, the selected presentation of specific cell

adhesion motifs of importance to β-cells may provide a more elegant means to functionalize the scaffold framework, avoiding immunogenicity and sourcing issues associated with incorporation of full matrix proteins. For example, Lin et al. evaluated the use of a PEG-based hydrogel functionalized with EphA–ephrinA fusion proteins to

mimic native cell-cell binding [31]. Encapsulation of β-cells, either derived from a cell

line or dissociated from murine islets, within these biomimetic hydrogels resulted in

enhanced survival and function. Another approach explored peptide amphiphiles

containing arginine-glycine-aspartic acid (RGD), in addition to a matrix metalloproteinase-2 (MMP-2) sensitive sequence, for β-cell encapsulation [32]. Peptide

hydrogels resulted in improved retention of glucose responsiveness and islet

morphology in culture. While efficacy in vivo has yet to be evaluated, 3D matrices

functionalized with bioactive motifs designed to facilitate ECM-cell or cell-cell

interactions can provide a powerful platform to enhance transplant outcomes by

preserving differentiated function and minimizing apoptosis.

Of note, several innovative approaches have sought to utilize these 3D structures

as bioactive platforms to direct host responses, such as inflammation, either through

incorporation of supplemental cells or through the tethering of bioactive agents. For

example, ethylene carbodiimide (ECDI)-fixed splenocytes were co-localized with rodent

islets within macroporous PLGA scaffolds prior to transplantation in an allogeneic rodent

model, whereby reversal of diabetes without the need of immunosuppression was

observed [33]. Protection was evidenced by an increased accumulation of Treg cells

and a reduction of interferon-γ (INF-γ), a pro-inflammatory cytokine. Elsewhere, researchers functionalized hydrogels with interleukin-1 (IL-1), an anti-inflammatory

peptide, for the local protection of encapsulated islets from immunoreactive cells [34].

Results showed improved protection of encapsulated islets embedded within these

bioactive surfaces when exposed to pro-inflammatory cytokines. Thus, localized 3D

microenvironments provide a potential for bio-mimicry that can recreate the spatial

architecture and environmental signaling present in the islet’s native microenvironment.

Oxygenation and Vascularization

Insufficient oxygen tension is one of the major obstacles for the success of

cellular based constructs, particularly when composed of highly metabolic cells, such as

pancreatic islets. While islets comprise about 1% of the pancreas, they consume

approximately 15% of the blood flow going into the pancreas [35]. Not only is the

oxygen consumption rate for pancreatic islets elevated compared to many other cell

types, they are also susceptible to functional impairment at moderate oxygen tensions

[36]. Proper oxygenation during the initial post-transplantation period plays a critical role

in islet engraftment and viability, as islets, stripped from their natural microvasculature

during isolation, rely solely on diffusion to obtain adequate nutrients and oxygen.

Hypoxia negatively impacts islet survival, mainly through the stabilization of hypoxia

inducible factor (HIF)-1α and the subsequent activation of its target genes, which lead to

a cascade of events that terminate in islet apoptosis [37]. In addition, HIF-1α has been

shown to result in impairment of glucose responsiveness [38, 39], which may further

delay islet revascularization [40]. While prompt re-vascularization of islets grafts is

critical for transplant success, complete prevention of hypoxia exposure is more

desirable. This is a particular challenge for extrahepatic sites, where these initially

avascular implants are plagued with substantial nutrient delivery problems, leading to

central necrosis and sub-optimal performance. Two promising approaches seeking to address this challenge are: in situ oxygen generation and revascularization guidance.

In situ oxygen generation is an attractive approach to improve oxygen tensions in the transition period from implantation to the formation of a functional vascular bed. The ideal oxygen generator must ensure adequate and stable oxygenation for the duration of the revascularization process. It should not drastically affect transplant size, the transplant procedure, or require multiple surgeries. While various prototypes have been developed, more recent publications have employed oxygen chambers or hydrolytically

reactive solid peroxides. Ludwig et al. developed an immunoprotective, macrochamber

device containing an oxygen tank, which is refreshed daily via an external port [41].

Islet-loaded macrochambers, retrieved 13 days post-transplant from an allogeneic porcine model, demonstrated a 50% increase in islet oxygen consumption rate (OCR) over controls. Moreover, in a rat model, this device was found to improve both islet OCR and insulin secretion levels, when islets were pretreated with growth hormone releasing hormone [42]. Recently, our group reported on the development of a hydrolytically reactive, oxygen generating biomaterial, based on the encapsulation of solid calcium peroxide [43]. The material was capable of generating oxygen over 6 weeks in culture, while co-culture with β-cells or rat pancreatic islets mitigated hypoxic-induced cell death when cells were cultured under hypoxic conditions.

An alternative approach to alleviate hypoxia during islet engraftment is improving the efficiency and competency of implant vascularization. Given that the formation of a functional vascular network is an intricate process, engineering a microenvironment that efficiently coordinates and directs these pro-angiogenic cells and factors is a complex

task. Interventions that focus on the delivery of pro-angiogenic cues must address

issues associated with pharmacokinetics, while approaches that target pro-angiogenic

cells must consider cellular interactions, delivery, and nutritional burden. One recent

publication explored the use of ultrasound-targeted microbubble destruction to deliver

the gene for human vascular endothelial growth factor (VEGF) to the host liver prior to

intrahepatic infusion of human islets in a diabetic nude mouse. Enhanced VEGF

secretion in the liver was detected, resulting in significant improvement in islet

engraftment and efficacy [44]. Additional strategies include the use of pro-angiogenic

cells, such as mesenchymal stem cells (MSC), as modulators of the revascularization

process, which has proven to increase vascular sprout formation, intra-islet

vascularization, and graft efficacy [45, 46]. Moreover, the transplantation of MSCs with

islets may have the additional benefit of immunomodulation, as recently observed in

rodent and nonhuman primate models [47-49].

The delivery of pro-angiogenic factors to the local site is a common means used

to initiate and hasten the revascularization process. The use of pro-angiogenic

polymers, i.e. the combination of angiogenic growth factors (GFs) and biomaterials, is

an attractive approach that has shown enhanced therapeutic effect over delivery of GF

alone [50]. Of recent interest are hydrogels incorporated with both cell matrix and

growth factor binding sites. Martino et al. has published on the use of a system

comprised of: fibrin network; angiogenic growth factors (PDGF-BB and VEGF-A); and a

recombinant fragment of fibronectin (FN) containing binding sites for fibrin, integrins,

and GF [51]. This system has been shown to provide potent synergistic signaling and

morphogenesis between α5β1 integrin and GF receptors, when their binding sites are

proximally presented in the same polypeptide chain. This synergy translated to enhanced regenerative effects in the treatment of chronic wounds in a diabetic mouse model. A similar presentation was engineered using synthetic polymers, such as the

PEG-based gels developed by Phelps et al., where resulting bioactive hydrogels

demonstrated enhanced vascularization and functional outcomes in a hind-limb

ischemia model [52]. This approach has recently shown enhanced benefit for islet

transplants in a syngeneic murine model [53]. For islet applications, the group led by

Samuel Stupp has explored the use of heparin-binding peptide amphiphiles, doped with

VEGF and FGF, to house islets. Incorporation of these GFs resulted in an increase in islet endothelial cell sprouting, improved viability and glucose stimulation indices, and enhanced efficacy in islet transplants within a syngeneic mouse model [53, 54].

Conclusion

The engineering of a 3D microenvironment for islet transplantation can provide a means to recapitulate the natural dynamic matrix environment of the islet niche.

Moreover, the selection of an appropriate transplant site with desirable mechanical and biological properties can enhance the long-term success of the graft. However, the unique challenge of autoimmune responses to the allogeneic beta cells is a significant obstacle in ensuring long-term survival of these cells. This grand challenge of

engineering microenvironments for islet transplantation is met with many hurdles, such

as impaired islet revascularization and function, as well as the fact that current

immunotherapy options are unable to completely abrogate autoimmune attack [56].

It is critical that immunomodulatory strategies are converged with optimal

implantation microenvironments to ensure success. While multiple researchers are exploring the potential of immunoprotective biomaterials or devices for dampening

immune attack, all too often these innovative immunomodulatory approaches are employed using devices or implantation sites that limit islet viability and function. For example, the implantation of immunoisolating devices within the subcutaneous site, which has unfavorable inflammatory responses and nutrient profiles, can impair appropriate assessment of the device’s potential. Further, the confinement of highly metabolically active islets within devices at loading densities substantially higher than native tissues, likely results in undesirable nutrient profiles. It was the goal of this dissertation to develop a bioactive material capable of delivering predictable and controllable oxygen profiles at levels capable of sustaining islet metabolic demand at high loading densities within 3D macroenpasulation devices for extrahepatic islet transplantation. This bioactive material can serve as a versatile platform for addressing

the formation of detrimental oxygen gradients within transplanted grafts, one of the key challenges hindering the success of islet transplantation in the clinical setting.

Figure 2-1. Various novel approaches implemented within islet transplants to improve overall function and efficacy include: A) macroporous scaffold designed to provide mechanical protection and 3D distribution of islet clusters; B) co- encapsulation of islets with helper cells such as MSCs or ECs to promote revascularization and engraftment; C) incorporation or tethering of growth factors to biomaterials to hasten and direct the revascularization process; D) implementation of devices/materials to supplement oxygen; and E) incorporation of ECM proteins to recapitulate native islet niche (illustration courtesy of Jessica A Weaver).

CHAPTER 3 ENGINEERING AN IN SITU OXYGEN GENERATOR

Background

Cellular transplantation is an attractive approach for the replacement or repair of damage tissue. Nevertheless, the success of these engineered tissues has been hindered by poor oxygenation upon transplantation, due to the inevitable delay in the vascularization process. These detrimental oxygen gradients have been shown to induce apoptotic and necrotic process both in vitro and in vivo. Moreover, the inadequate oxygen delivery of cellular constructs leads to a dysfunctional and highly pro-inflammatory environment [55, 56].

Some of the approaches that have been implemented to enhance the mass transfer of oxygen to this 3D devices includes: (1) enhanced vascularization around the graft; (2) in situ oxygenation; (3) and external oxygen delivery. While all of the aforementioned approaches have been implemented with variable levels of success, in situ oxygen delivery remains, by far, the most attractive approach, as it allows for the immediate oxygenation of the graft without burdensome external devices. In situ

oxygenation has been accomplished via multiple approaches such as the electrolytic

decomposition of water in an electrolyzer [57] and photo-synthetic algae able to

produced oxygen when exposed to a light source [58]. Although proof of principle has

been demonstrated, the translatability to an in vivo platform will likely be invasive and

cumbersome.

In situ oxygen generation has also been achieved via encapsulation of oxygen

releasing agents such as peroxides (calcium peroxide, magnesium peroxide and

hydrogen peroxide), and oxygen carriers, such as perfluorocarbons (PFC) and

fluorinated compounds [59]. PFCs have been widely investigated due to their enhanced

solubility for physiologically relevant gases, such as oxygen and carbon dioxide,

resulting in a transport behavior similar to hemoglobin. Perfluorocarbon emulsions have

been employed as blood substitutes, as well as incorporated in bioreactor systems to

improve its oxygen permeability [60]. Likewise, thermoresponsive fluorinated materials

with enhanced oxygen storage ability have been developed and investigated for

increased oxygenation of cells [61].

The work in our laboratory has centered on the implementation of calcium

peroxide as an oxygen generating agent. While a potent oxygen generator, the abrupt

decomposition of calcium peroxide into oxygen and other byproducts upon water

contact results in a harmful hyperoxic environment resulting in the accumulation of

detrimental byproducts, which results in decreased cellular viability [62]. As such,

calcium peroxide must be encapsulated within a material to control its hydrolytic

reactivity. Others have explored the encapsulation of calcium peroxide in degradable

PLGA; however, oxygenation was short-lived (< 7 d) and translatability was hindered by

the accumulation of negative byproducts and the need to implement scavenging

catalase to mitigate cytotoxicity [63]. In our approach, we encapsulated calcium

peroxide within the biostable hydrophobic material polydimethyl-siloxane (PDMS), termed herein as OxySite [64]). We found that encapsulation within this material generated a unique and unexpected degree of controlled reactivity, resulting in the extended generation of oxygen without the accumulation of reactive oxygen species

(ROS) and the development of hyperoxia. Moreover, in vitro assessments found

improved viability and function of beta cell and pancreatic islet cultures under low

oxygen in the presence of this patented oxygen generating biomaterial [65].

Nevertheless, a complete understanding of the mechanism of action of oxygen

release, as well a comprehensive study of kinetic of oxygen generation, was lacking. As

such, the purpose of this study was to systemically improve each fabrication step, to

optimize the reproducibility of reactivity, and to carefully evaluate and characterize changes in biomaterial properties over an extended reactivity period. Furthermore, an extensive kinetic characterization was conducted, which was subsequently translated into a mathematical model of oxygen release, which was then implemented in a

multiphysics model of simulated in vitro cultures.

Methods

Materials

All chemical reagents were purchased from Sigma-Aldrich, unless otherwise

noted. All culture media was sourced from Mediatech. Poly(dimethyl)siloxane (PDMS)

(NUSIL 6215, clinical-grade) was sourced from NuSil Silicone, USA.

Fabrication of Oxygen Generator

The oxygen generating material was fabricated in a manner similar to previously

published, but with optimization in mixing and mold casting to improve the

reproducibility of the resulting OxySite disks [65]. Briefly, 25% w/w CaO2 was added to

PDMS in a 4:1 vol/vol PDMS monomer to platinum catalyst and mixed/defoamed using

a high viscosity centrifugal mixer (Thinky Mixer, USA). The resulting PDMS/CaO2 mixture was then poured into a cylindrical mold (29 mm) to 1 mm thickness and cured

overnight at 40 oC. Prior to use, smaller OxySite disks (10 mm) were punched out of the

larger cured disk. Resulting OxySite disks were sterilized with ethanol for 30 min and rinsed five times in PBS (Invitrogen) prior to performing experiments

Leaching Studies

In order to study the effects of water exposure on the OxySite disk’s chemical and surface characteristics, samples were hydrated for selected time periods. Shortly,

OxySite disks were punched out from mother slabs, sterilized in ethanol for 30 min and consequently rinsed in PBS for five times (five minute rinse each time). Samples were then placed in a 24 non-tissue culture treated plate. 1 mL of sterile PBS (w/o Ca, and

Mg) was added to each well. Plates were placed in a 37oC incubator throughout the

duration of the leaching study. Sample supernatant was refreshed daily. After selected

time points samples would be collected and used for specific assessments as described

below.

Macroscopic Analysis

Dimension analysis was performed on OxySite disks of 10 mm in diameter after

leaching them in 1 mL PBS buffer (replacing buffer every day) for 60 days. Samples

were checked for weight uniformity (Mettler Toledo, USA), diameter, and thickness

(Digital caliper, Fisher) at n = 3 per time point.

Microscopic Analysis

OxySite morphological changes were investigated using a scanning electron

microscope Hitachi VP-SEM S-3000 coupled using a secondary electron detector (SE).

Scans were performed at 3-10 kV.

The porosity and pore parameters of the OxySite after selected leaching times

was determined using a mercury intrusion technique conducted using a Quantachrome

Poremaster Mercury Porosimeter (Quantachrome Instruments, USA). Briefly, whole

OxySite constructs were weighed and placed in the cup of the penetrometer, which was closed by tightening the cap. The penetrometer, along with the sample, was placed into the pressure chamber of the porosimeter and pore properties measured. Intrusion and extrusion assessments were performed at both low (up to 50 Pa) and high (up to 60

KPa) pressures.

A GE Vtome xm240 CT scanner (GE, USA) was employed for high resolution CT images of the OxySite material after different leaching times. Projected x-ray images were acquired (Magnification 30.8, Voxel size 6.5 µm, FOD 26 mm, number of images acquired 2600 at 2004x2024 pixels), and reconstructed using a VGA studio software.

After reconstruction, images were exported as tiffs and uploaded to Matlab for volumetric analysis. A code was developed to segment the disk from the background and detect the high density peaks. A median filter was implemented to reduce noise (i.e. not related to calcium peroxide). After filtering, histogram of image voxel intensity was inspected and thresholded to isolate voxels related to high intensity peaks only. For volumetric calculation of loading density, thresholded high intensity voxels from stacks of x-y images (to give total segmented volume) were added and divided by the total voxels corresponding to the disk. Gradient calculation was performed on selected images at different penetration depths to confirm observed difference in calcium peroxide deposition.

Surface Characterization

Elemental chemical analysis was performed by X-ray photoelectron spectroscopy

(XPS) on a PHI Versa Probe III (ULVAC- PHI Electronics, MN) under high vacuum (5 <

10-7Torr) using Al Kα radiation with a pass energy of 200 eV for survey scans and 50 eV for elemental scans, a 200 μm spot size, and 50 ms dwell time. Corresponding peak

assignments were analyzed using a Smart soft Versa probe software. Elemental analysis scans (n =2) on three spots for each top and bottom surfaces per disk for disks that were leached at different time points.

Electron dispersive X-ray analysis (EDX) was implemented to perform chemical surface characterization of the OxySite material. The EDX-system was coupled to a FEI

Nova NanoSEM 430 (Nova nano, USA) with lateral secondary electron (LSE) signal detectors.

Quantitative transmission FT-IR measurements were recorded using a Frontier niIR (Perkin Elmer). Scan range was from 2500 to 1800 cm−1 at a 4 cm−1 resolution. Spectral processing of signatures obtained was performed by

Spectrum Software V.6.2.0. Spectral graphs presented correspond to one OxySite disk or a blank control.

Roughness analysis of OxySite samples leached at different time points were evaluated via a 3D Laser Scanning Microscopy (LSM) analysis (VK-9700K, Keyence

GmbH, Germany). For each sample, LSM images (n = 3) at 20-fold magnification were taken.

Hydrogen Peroxide Measurements

Hydrogen peroxide production from OxySite disks was assessed for samples (3 different disks for each individual time point, assay was run in triplicates for each sample) incubated in 24-well plates with 1 mL of PBS at 37 °C for 60 days. Samples were refreshed every day and collected at defined time points. Hydrogen peroxide concentration was measured using a colorimetric assay (Assay Designs).

Calcium Concentration

To assess changes in calcium concentration based on hydration exposure time, supernatant samples from the non-invasive oxygen recordings were collected after each

refreshing time point and stored at -80oC prior to testing (n=3 samples at each time

point collected). A calcium colorimetric kit (Sigma) was implemented to analyze samples. Briefly, an aliquot of 50 µL per sample (i.e 3 different samples run in triplicates for each time point assessed) were plated in a 96 well plate. 90 µL of a chromogenic reagent was added and mixed. For detection, 60 µl of a calcium assay reagent was added, samples were incubated for 10 min, and read at 575 nm in a Spectramax M5 spectrophotometer (Spectramax, USA). To calculate concentrations, OD was correlated to calcium standard curve diluted in saline buffer used for non-invasive oxygen collections.

Non-invasive Oxygen Measurements

Oxygen release profiles of OxySite disks of different geometries were obtained via a custom-built, non-invasive, sealed titanium chamber system (Instech Laboratories,

PA). Partial oxygen tensions in the solution medium were measured using non-invasive oxygen sensors (PreSens, Germany). Sensor spots consisted of a ruthenium coated via a sol-gel polymer. The optical signal was sent to the sensors via fiber cables from a custom transmitter box (PreSens, Germany). Ruthenium in spot sensors quenched oxygen in the medium shifting fluorescent signal, which can then be correlated to the oxygen concentration in the solution. Oxygen concentrations were recorded by Presens open software in units of mmHg (torr). Samples were refreshed between 16-28 h. For

mathematical modeling, these values were converted to mol/cm3 using Henry’s law

(Equation 1) and a Bunsen solubility coefficient (α) of 1.6x10-9 mol·mmHg·cm-3 for

oxygen at 25oC.

= (3-1)

Mathematical Model 𝑐𝑐 𝛼𝛼𝛼𝛼 The proposed mathematical model is based on the following assumptions:

1. Oxygen diffusion from matrix follows a Fickian diffusion

2. Sink conditions

3. The reaction coefficient of calcium peroxide with water was order of magnitude faster than oxygen diffusion out of the PDMS matrix

4. No matrix swelling or contraction upon water contact, with constant total volume of the system with time

Based on those assumptions, the oxygen diffusion out of the matrix can be described by Fick’s second law in cylindrical coordinates:

= + + (3-2) 2 2 𝜕𝜕𝜕𝜕 𝜕𝜕 𝑐𝑐 1 𝜕𝜕𝜕𝜕 𝜕𝜕 𝑐𝑐 2 2 𝜕𝜕𝜕𝜕 𝜕𝜕𝑟𝑟 𝑟𝑟 𝜕𝜕𝜕𝜕 𝜕𝜕𝑧𝑧 0𝐷𝐷 � , � (3-3) (See Figure 3-1)

Using the time conditions of ≤Cmatrix𝑟𝑟 ≤ 𝑎𝑎 (t− =0)𝑙𝑙 ≤ = 𝑧𝑧C≤o and𝑙𝑙 Cmatrix (t→∞) = 0 and

considering both axial and radial mass transport, the solution to the diffusion equation,

as given by Vergnaud [66], corresponds to:

( ) = 1 ( ) (3-4) 𝑀𝑀 𝑡𝑡 ( ) exp ( ) exp 𝑀𝑀 ∞ 2 ( ) 2 2− 32 ∞ 1 𝑞𝑞𝑛𝑛 ∞ 1 2𝑝𝑝+1 𝜋𝜋 2 2 2 2 2 𝜋𝜋 ∑𝑛𝑛=1 �𝑞𝑞𝑛𝑛� −𝐷𝐷 𝑅𝑅 𝑡𝑡 ∑𝑝𝑝=1 2𝑝𝑝+1 �−𝐷𝐷𝐷𝐷 𝐻𝐻 � Where Mt and M∞ represent the absolute cumulative amounts of drug released at

time t and infinity, respectively; n and p denotes the summation variables; qn are the

roots of the Bessel function for the first zero order derivative [J0(qn) = 0], and R and H symbolize the radius and height of the cylinder.

Assuming that the release profile from the surface is dominant over the axial release, the above equation can be reduced from a 3D domain to a 2D domain, as shown by [67]:

/ = (3-5) 𝑀𝑀𝑡𝑡 𝑆𝑆 𝐷𝐷𝑒𝑒𝑒𝑒𝑒𝑒𝑡𝑡 1 2 𝑀𝑀∞ 𝑉𝑉 � 𝜋𝜋 � Where Deff is an adjusted diffusion coefficient due the presence of porous and encapsulated material in the matrix causing the diffusional length path to change [68].

Dimensional analysis has shown that the diffusion coefficient can be scaled based on tortuosity, τ, [69] as follows:

= (3-6) 𝐷𝐷 𝑒𝑒𝑒𝑒𝑒𝑒 where D corresponds to the diffusion𝐷𝐷 coefficient𝜏𝜏 of oxygen in the matrix. For our model, D was estimated as 3.5 e-5 cm2/s based on literature values [70]. Furthermore, the tortuosity of a matrix, τ, is defined as the total length between the continuous paths of two throats in a matrix divided by their straight line length [71]. This value is a representation of the degree of difficulty of mass transport in the (connected) pore structure [72]. Hence, it is a paramount variable dictating the permeation and diffusion process in the matrix [73]. In the literature, several models of tortuosity are implemented based on pore shape, size and distribution. For the purpose of this thesis, the model proposed by Salmas and Androutsopoulos [74], based on pores of cylindrical shape and pore entrapped volume fraction illustrated below, appeared the most appropriate and was implemented.

4.996 = 4.6242 1 5.8032 (3-7) 1 𝜏𝜏 𝑙𝑙𝑙𝑙 � − � − − 𝛼𝛼𝑒𝑒𝑒𝑒 = (3-8) 𝑣𝑣𝑒𝑒𝑒𝑒 𝑒𝑒𝑒𝑒 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 ∝ 𝑖𝑖𝑖𝑖 These values of αen can be calculated𝑣𝑣 from parameters obtained from the porosimeter analysis.

Computational Modeling

COMSOL Multiphysics 5.2a was used to generate a finite element analysis model of oxygen gradients and in situ oxygen generation within 2D islet cultures at different loading densities and external oxygen tensions. The general diffusion equation

(3-9) with negligible convection was used to model oxygen diffusion.

+ ( ) = (3-9) 𝜕𝜕𝜕𝜕 The reaction𝜕𝜕𝜕𝜕 rate ∇term∙ − for𝐷𝐷 ∇oxygen𝑐𝑐 𝑅𝑅 consumption was modeled as a Michaelis-

Menten reaction (3-10) with the constant term corresponding to a critical oxygen level

3 where cellular oxygen consumption drops by half, with a CMM equal to 1e-3 mol/m .

Moreover, a dirac step was added to stop oxygen consumption in areas where oxygen

3 level fell below a critical oxygen tension, Ccr at 1e-4 mol/m , as described elsewhere

[75].

= , ( > ) (3-10) 𝐶𝐶𝑂𝑂2 , 𝑂𝑂2 𝑀𝑀𝑀𝑀𝑀𝑀 𝑂𝑂2 𝑂𝑂2 𝑂𝑂𝑐𝑐𝑐𝑐 𝑅𝑅 𝑅𝑅 𝐶𝐶𝑂𝑂2+𝐶𝐶𝑀𝑀𝑀𝑀 𝑂𝑂2 ∙ 𝛿𝛿 𝐶𝐶 𝐶𝐶 The model consisted of two subdomains: i) the suspension medium and ii) the mimic spherical islets of 100, 150 and 250 um in diameter. A maximum oxygen

-1 -3 consumption rate Rmax of 0.034 mol·s ·m (per islet volume) [75] was used for subdomain ii. The oxygen diffusion coefficient was assumed to be 3 e-9 m2/s in both

domains. An initial oxygen concentration of 0.2 mol/m3 was assumed for all cultures.

Moreover, a concentration boundary at the top of subdomain i was implemented to

model the external oxygen concentration of the culture system, i.e. 0.01 mol/m3 for low

oxygen and 0.2 mol/m3 for standard oxygen. Insulation, no flux, was assumed for the

walls of the insert. Finally, an inward flux boundary condition was implemented at the

bottom of subdomain i to represent the oxygen release rate from OxySite. This flux

boundary was calculated from average release kinetics of OxySite disks performed

using sealed noninvasive oxygen release chambers (Instech Laboratories, PA). For an

explanation on the method implemented to model a 2D distribution of a 3D loading

density see APPENDIX A.

Islet Isolation and Culture

All animal procedures were performed under protocols approved by the

University of Florida IACUC and in accordance with the National Institutes of Health

guidelines. Rat islet isolations were performed using methods described previously [22],

using male Lewis rats (Harlan, Indianapolis, IN). After isolation, islets were cultured

2 o overnight under standard culture conditions (500 IEQ/cm , 37 C, 0.20 mM O2, and 0.05 mM CO2) in supplemented CMRL 1066 medium prior to use. In order to validate in vitro

release kinetics after long term hydration, OxySite disks were prepared as described

above, sterilized, and placed in PBS buffered in a 24-well plate in a humidified incubator

at 37oC. The buffered solution was replaced daily to mimic release kinetics experienced

during non-invasive oxygen monitoring.

Disks were used for studies 15 d post leaching. For in vitro cultures, disks were

placed in a 24-well plate and a non-tissue culture treated insert (Millipore) was placed

on top of the disks. Rat islets at a loading density of 1326 IEQ/cm2 were resuspended in

300 µl of supplemented CMRL 1066 media and placed inside the insert. 700 µl of

supplemented CMRL was placed outside the insert. Samples were incubated either

under standard oxygen conditions (i.e. 0.2 mM (20%) O2 with 0.05 mM (5%) CO2) or

under low oxygen conditions (i.e. 0.01 mM (1%) O2 with 0.05 mM (5%) CO2 using a C-

Chamber Hypoxia Chamber, Biospherix, New York). Samples were collected after 24

and 48 h of culture.

In Vitro Cellular Assessments

The metabolic activity was assessed via MTT (Promega), as previously

published [65]. Briefly, at time points designated, samples were incubated in 250 μL of

complete CMRL media with 28 μL of MTT for 1 h under standard oxygen conditions.

Stop solution (185 µL) was then added under agitation and incubated for 48 h to permit

complete formazan crystals solubilization. Absorbance (570 nm) was then measured

from triplicates samples using a SpectraMax M5 microplate reader (Molecular Devices).

Lactate dehydrogenase (LDH) release was quantified using methods previously

described [65]. Briefly, 100 μL of supernatant was mixed with 100 μL of working reagent

(Cytotoxicity Detection Kit, Roche) and incubated for 30 min at RT. Stop solution (50 μL

of 1 N HCl) was added and the resulting absorbance in triplicate samples was

measured at 492 nm.

Live/Dead samples were visualized using the Viability/Cytotoxicity Kit (Invitrogen)

and a Zeiss LSM 510 inverted confocal microscope, as described previously [65]. After

incubation, samples were imaged for live cells (green) and dead cells (red). Z stacks

(2.29 µm thickness; 7 slices per image; 1024x1024; 20x objective) were merged into a

3D projection using Zeiss Zen Software.

Statistics

Macroscopic analysis corresponds to an average of 3 random samples represented as mean ±SD. Microscopic and surface characterization analysis corresponds to one disk per sample time point with multiple readings made per disk (n >

3). Kinetic assessments for comprehensive oxygen release profiling for all OxySite

samples corresponds to an average of 3 independent samples (3 samples from 3 different OxySite mother slabs). Hydrogen Peroxide and calcium assays correspond to supernatant collected from three independent disks at each collection time point. Data is presented as mean ±SD of group replicates (n=3 for all time points assessed). In vitro data was collected from one islet isolation with data represented as the mean ±SEM of

group replicates (n = 2 for 48 h of each test and group).

Results

A biomaterial platform for oxygen delivery was previously developed in our

laboratory based on the encapsulation of calcium peroxide within a PDMS matrix

(Figure 3-2 Panel A). Those studies revealed the unique ability of PDMS to mitigate the

burst release of oxygen from calcium peroxide decomposition and to extend the

duration of its reactivity. Nevertheless, a full characterization of the material properties

and the mechanism of oxygen release had not been explored. Thus, it was the aim of

this project to examine its biomaterial properties, as well as a comprehensive evaluation

of the oxygen generating capabilities of the material. Finally, a mathematical model of

oxygen release was developed and evaluated, both experimentally and in a

computational model.

One of the first challenges encountered in gaining an understanding of our

oxygen generating platforms was the lack of reproducibility from our devices. Visual

inspection of polymer mixture with reactive agent demonstrated spots of higher peroxide concentration at different sites within the casting frame. Moreover, casting within a wider cylindrical shaped mold resulted in disparate heights within the mother slab, and thus, higher sedimentation rates at certain spots. Hence, it became obvious that the previous

casting method did not result in the complete homogenous distribution of the reactive

agent into the system. The first order of experiments set to improve the consistency in

the fabrication method of our devices. In order to obtain a homogeneous distribution of

calcium peroxide within the polymer network prior to crosslinking, a planetary mixer was

employed, which utilizes two simultaneous centrifugal forces to allow for complete

mixing of the powder agents. Moreover, the mixture was deaerated within the mixing

step prior to the degassing of the system; decreasing air entrapment within the cure

polymeric matrix. These enhanced devices are herein referred to as OxySite.

To begin our analysis, macroscopic changes of the biomaterial were assessed in

a time course manner. For this purpose, OxySite disks were fabricated, sterilized, and

incubated in saline buffer at 37oC for up to 60 d. At selected time points, disks were

collected dried and weighed and dimension changes were recorded. As seen in Figure

3-2 Panel C-D, no significant changes in the weight or radial/axial dimensions of the

OxySite disks were observed at any of the time points assessed. This suggests that the

hydrophobicity of the material hindered water diffusion and accumulation within the

matrix, during the 60 d incubation time. Further, while not quantifiable, visual differences

in the opacity of the material were observed throughout the leaching period; hinting at

reactivity of the calcium peroxide within the matrix.

More in-depth analysis of the initial material porosity and the effects of leaching

were performed via a mercury porosimeter. Samples were tested at both low (50 pa)

and high (60 Kpa) pressure to obtain a comprehensive overview of nano and macro

pores within the material. The entrapped volume collected after mercury intrusion was

measured to estimate material porosity according to the following equation:

= + (3-11)

𝑏𝑏 𝑔𝑔 𝑝𝑝 Vp was assumed to be 𝑣𝑣the volume𝑣𝑣 𝑣𝑣intruded after mercury saturation, while Vb was determined by weighing the sample post mercury intrusion. The porosity was calculated as:

+ (3-12) = 𝑣𝑣𝑏𝑏 𝑣𝑣𝑔𝑔 𝜑𝜑 Figure 3-3 panel A summarizes𝑣𝑣 porosity𝑏𝑏 measurements obtained using the

above equations for OxySite disks after incubation in buffered solution for varying time

lengths. Preliminary comparison of OxySite to control blank PDMS disks found a

decreased porosity in the former (ϕ = 0.007 for control PDMS versus 0.01 for OxySite

samples); suggesting that the addition of calcium peroxide crystals to the polymer

matrix might aid in the formation of pores. Nevertheless, a more robust characterization

of changes in the PDMS matrix alone should be conducted in order to validate this

hypothesis. While a limited change in total porosity was observed for OxySite samples

within the 1 to 7 d time period (ϕ = 0.01 at day 1 compared to 0.009 at day 7),

differences in porosity were apparent at the 30 d time point, with an increase in porosity

of 80% after 30 days post initial hydration. Moreover, enhanced average pore size

dimensions were seen in OxySite samples following increased leaching time at both

assessed scales (Figure 3-3 Panel B low pressure black line, high pressure red line),

with pores ranging from 60-70 nm under high pressure, and 200-230 µm after low

pressure intrusion for samples exposed to elution medium for 30 to 100 days;

conversely, pore sizes leached for 1-7 d ranged in size from 50-60 nm and 130-150 µm.

Likewise, a histogram of the percent volume intrusion as a function of pore size depicts

a strengthening in the distribution at higher pore diameters for OxySite samples leached

for prolonged periods of time, with 9% of the pores ranging from 250-300 µm at day 100

vs 1.5% at day 1, and 24% between 100-250 µm vs 11% (Figure 3-4 Panel C-D). In

stark contrast, blank samples exhibited a lower range pore size with pores ranging from

4 nm to 10 µm. Evaluation of hydrophobic matrixes within the drug delivery field has

demonstrated that interconnected aqueous pores can be formed by increasing water

migration into the matrix [76, 77]. Thus, it is our working hypothesis that calcium

peroxide dissolution within throats of the material (generated either by polymer cracking

or air entrapment) was responsible for the formation of pores within the PDMS matrix.

Within our theory, blank PDMS matrix pore formation is only dependent on polymer

cracking via water filled pores; hence, porosity and pore distribution remain in the lower

range (compared to OxySite) due to the poor water wettability of the polymer matrix.

Conversely, the changes herein observed suggest that the oxygen generating agent

within OxySite, at this fraction loading density, its able to increase the osmotic pressure

within the matrix leading to the formation of growing cavities within the polymer; and

thus, the observed changes in porosity and pore size distribution overtime. Further, this

bulk changes in the OxySite material porosity as a function of exposure time can be

valuable in characterizing the resulting oxygen kinetics, as highlighted below.

While porosity was an important factor in elucidating diffusional paths within the material, an understating of the surface chemistry could provide insight as to calcium peroxide availability and thus a sense of the reactivity of the material. To measure

surface peroxide within the material, as well as track changes in peroxide presence as a

function of time, surface characterization techniques XPS, EDX, Raman spectroscopy,

and EDX were implemented on multiple samples. Limitations on either the technique

itself (penetration depth, wavelength requirements out of the scope of the equipment

capability) or the sample (melting during acquisition, need to coat sample which would

mask surface), however, impaired our ability to obtain relevant information regarding the

chemical composition of the OxySite surfaces. As can be seen in the FTIR data (Figure

3-5 Panel A), no characteristic fingerprint for calcium peroxide was observed on the

surface of our OxySite disk (green). Moreover, samples that had not been exposed to

buffer or sterilized prior to testing (magenta) lack the characteristic fingerprint observed

for the powder alone (red). Likewise, no discernible calcium peroxide peaks were

evident after cutting the OxySite disk to expose the inner surface. Further, XPS data

displays similar results, with no characteristic peaks for calcium found in OxySite

samples without leaching (Figure 3-5 Panel B) or after leaching for varying periods of

time (data not shown). Lastly, no difference in the oxygen concentration on the surface

was found via XPS among the samples tested (data not shown).

Visual characterization of the surface via SEM demonstrates particle

sedimentation, believe to be calcium peroxide, on the surface of the OxySite disks.

Moreover, pores ranging from 100 to 250 µm were found to be present in the cross

section of the material (Figure 3-6); however, no visual changes were observed in the

surface of the material versus leaching time (data not shown). Furthermore, characterization of the surface via LSM demonstrated visual changes in the surface roughness that appeared to be proportional to the leaching time on the selected sections imaged (Figure 3-7).

Characterization of the calcium peroxide particles within the PDMS material was performed using a nanoCT system (GE). Scan of the OxySite disk could discriminate only two different attenuation energies, one for the background and one for the combination of the calcium peroxide silicone mixture. Considering the relatively low quantity of CaO2 compared to Si, the peak of CaO2 was obfuscated by the right end tail of the Si distribution. In fact, a comparison between the histogram of a blank PDMS disk and an OxySite disk revealed a wider distribution towards the right tail end for the latter.

Hence, knowing the energy distribution of the Si-only, we estimated an intensity to

differentiate between the Si and CaO2. As all images herein presented were taken

during a single scan, the same energy distribution applies. Further, the same

thresholding estimate was implemented for all samples. Hence, segmentation of the two

materials allows us to perform semi-quantitative volumetric analysis of our OxySite

disks to validate the homogeneity of our samples, as well as to track changes in calcium

peroxide deposition over time.

Extracted regions of interests (ROI) from the scans were obtained from the

nanoCT multiscan. Each ROI represented a sample OxySite disk leached for selected

time points to explore the change of overall calcium peroxide within the system. ROI

segmentation was implemented in Matlab, to threshold the energy peaks between

background, silicone, and calcium peroxide particle. As depicted in Figure 3-8 Panel B,

the total concentration of calcium peroxide within the OxySite disks prior to leaching (i.e. sum of all high density energy peaks divided by the sum of high and low density peaks) match the theoretical loading densities calculated for fabrication (i.e. 9.8 experimental loading vs 9.9% vol/vol calculated theoretically, which corresponds to a 25% w/v loading

density). These results support the ability of our optimized fabrication process to

manufacture homogeneous OxySite disks. Unexpectedly, no changes in total loading

density was observed between OxySite disks leached for short (5-15 d) or long (30 or

60 d) time intervals, as the loading density ranged from 9.2 at day 5 to 9.9 at day 30.

This was in contrast to the visual changes in opacity observed from the OxySite disks at

these same time points (see Figure 3-2). This disparity could be a result of noise within

the system affecting the accurate estimation of total density for small changes in total

loading density. In the future, samples should be run at a higher resolution (i.e. higher

energy and smaller voxel size) to explore the capacity of the system to distinguish two

true peaks within the energy histogram to better differentiate between the samples.

Further, decreasing the voxel size could also help in providing distinct edges between

the calcium peroxide material and the PDMS, thus, allowing for an area fragmentation

based both on energy attenuation as well as location.

Of note, 3D nanoCT images provided visual inspection of calcium peroxide

distribution within the OxySite. In these images, a vertical gradient of calcium peroxide

within the OxySite disks was revealed, with a greater accumulation of the particles at

the bottom of the construct. This pattern indicates potential settling of the calcium

peroxide-PDMS during the curing process. Investigation of the subsequent layers of

peroxide that form within the material could provide a means to determine the presence

of a “reactive zone” responsible for total oxygen capacity of the system. Our theory is

that only a selective outer layer within the material can be penetrated by water and

react, leading to calcium dissolution and oxygen release. This is based on the same

mass transport implemented for the mathematical model presented herein.

Understanding the location, thickness, and volume of calcium peroxide within each layer

could aid in the determination of total duration of oxygen release and assist in the

optimization of an OxySite geometry (i.e. reducing the volume from the total disk to only the reactive layer). A 0.2% vol/vol difference in the first layer (i.e. from the top to the bottom of the disk) was observed between OxySite disks leached for 5 d vs 15 d (Figure

3-9 Panel A); however, no discernible change in this layer was observed at later time points (15 d versus 30, 45, 60 d).These preliminary results should be reassessed after validation of thresholding methods. Nevertheless, these results demonstrate the implementation of this noninvasive technique, for the first time to our knowledge, to track biomaterial changes over time. Herein, we proposed the application of this technique as a means to explore changes in calcium peroxide that can affect OxySite performance and lifetime, and thus can aid in customizing the material for individual applications.

Accumulation of reactive byproducts in the collecting supernatant due to calcium peroxide decomposition was an important parameter to address due to its implication on cell viability. Hydrogen peroxide, an intermediate step in the reaction and a potent free radical generator, was quantified in the supernatant of OxySite disks after continuous leaching for an extended time period (60 d, samples were refreshed daily). As observed in Figure 3-10 Panel A, the concentration of hydrogen peroxide in solution averaged

59.3 µM ± 10.1 at all-time point tested. This was substantially below toxic

concentrations reported in the literature (e.g. 200 to 500 µM, represented by the orange

inset in the graph) [78, 79]. Further, accumulation of calcium in the OxySite milieu were

measured, as summarized on Figure 3-10 Panel B. Released calcium into the medium

fluctuated for the first 13 d post release, with a maximum peak of 0.87 mmol/L ± 0.014

achieved at day 7. From day 15 to 30, calcium levels in the bathing media averaged

0.056 ± 0.015 mmol/L, considerably lower than reported cytotoxic concentrations [80].

A comprehensive evaluation of oxygen kinetics was performed for our original

OxySite prototype (10 mm in diameter, 1 mm in height). OxySite disks were punched

from a mother slab, sterilized, and placed in a temperature controlled sealed chamber.

Partial oxygen tensions in solution were measured via a noninvasive oxygen sensors,

with daily sample refreshing. Figure 3-11 summarizes the extensive oxygen release

profile of OxySite disks (n = 3) procured from three separate mother slabs. Partial

oxygen released in solution peaked after 24 h of release to 219.6 ± 49.5 mmHg.

Subsequent release averaged 77.4 ± 15 mmHg during the first six days post exposure.

At longer incubation times (i.e. 6-30 days), the pO2 in solution averaged 44.8 ± 11.07

mmHg. These results demonstrate the ability of our OxySite material to not only

generate robust levels of oxygen, but to do so for extended periods of time (> 20 days);

a quality that has not been reported, to date. Significantly, current measurements

demonstrate the ability of our material to achieve pO2 of 39 ± 3.2 mmHg in vitro even at

40 days leaching (data not shown). These oxygen tensions are well within physiological

levels and in line with the oxygen tensions reported of native islets in the pancreas [81].

As such, these results highlight the ability of our material to generate oxygen in a manner that is relevant for pancreatic islet culture.

For the next phase of study, the oxygen release kinetic profile depicted in Figure

3-11 was converted to a cumulative release plot, to compare kinetic data to the mathematical model of release described in the methods section. Cumulative release was calculated from plotting total oxygen concentration released in moles (from

Equation 3-1) divided by the total theoretical number of oxygen moles that can be

produced in the system, which was based on the total calcium peroxide loaded

(obtained by the stoichiometry reaction of calcium peroxide decomposition). A

mathematical model (see methods section, Equation 3-2 to 3-10) was implemented and

compared to our experimental data. In short, the model represents an approximation of diffusion controlled released from a cylindrical shaped monolithic reservoir. This system

assumes that the drug was homogeneously dispersed within the matrix. Moreover, this

model does not take into account matrix swelling. On the other hand, the diffusion

velocity of oxygen out of the matrix was considered to be dependent on the length of the

diffusion pathway (tortuosity) and thus required the calculation of an effective diffusivity

(Equation 3-6). Finally, release kinetics were assumed to be primarily dependent on

time and matrix geometry (Equation 3-5). This assumption could be justified by

considering that the scale of the driving force for calcium peroxide decomposition was

much higher than that of oxygen diffusion out of the system. Thus, based on the mass

transport theory of Noyes and Whiteney, when two driving forces are occurring

sequentially, the faster one can be neglected when calculating the overall mass

transferred from the system [82].

Fitting of our experimental data to results obtained from the proposed model are depicted in Figure 3-12 Panel A. As it can be seen, a strong agreement between our

experimental data and the proposed model was obtained (R2 = 0.97, p<0.0001). Fitness of the model was assessed via a magnitude analysis that varied the value of effective diffusion coefficient by a factor of 10. As it can be seen in Figure 3-12 Panel B, variation of the effective diffusivity greatly deviates the predicted release curves compared to the experimental data collected.

To examine the impact of scale on oxygen generation from OxySite disks, as well as to validate our mathematical model, oxygen traces were obtained from OxySite disks

of varying diameters (8, 6, 5, and 3 mm) (Figure 3-13 and Figure 3-14). These

geometries were chosen as they were of appropriate scale for implementation in murine

animal models, as is elucidated in later chapters. In comparison to the original 10 mm

disks, a similar burst release (223 ± 4.5 mmHg) was observed for the 8 mm disks. After

a week of leaching, oxygen tensions decreased to 58.7 ± 14.8 mmHg; a decrease of

approximately 20 mmHg when compared to original size disks. After 30 d, oxygen

tensions of 16 ± 9.6 mmHg were still observed (Figure 3-13 panel A). Decreasing the

OxySite size by 40% (6 mm) lead to a reduction in burst release (131 ± 69 mmHg; p =

0.02 ). Oxygen tensions remained at about 58 ± 10.3 mmHg for the next 6 d, while

oxygen profiles of 35.4 ± 7.7 mmHg were still observed 17 d post initial hydration

(Figure 3-13 panel C). Reducing the size of the original disk by half to 5 mm reduced

the burst release and generated similar profiles as those for the 6 mm disk (Figure 3-14

panel A). Remarkably, decreasing the size by 70%, i.e. 10 mm to 3 mm, still resulted in

a significant increase in oxygen tension, with a burst release of 86.5 ± 0.24 mmHg and

oxygen profiles of 23.7 ± 2.85 mmHg after 7 d of leaching (Figure 3-14 panel C). Of

note, complete expiration of oxygen generation was not observed in any of the

geometries at the time points evaluated. These results demonstrate our ability to control

release kinetics based on geometrical changes and highlights the ability to tune the

material depending on the metabolic demand of the system.

In order to determine the robustness of the mathematical model developed,

cumulative release profiles from the different OxySite dimensions were generated and

compared to theoretical results adjusted for calcium peroxide loading (Figure 3-13 and

Figure 3-14 panels B-D). For all dimensions tested (5 in total), a robust fit between the

predicted and the experimental values was found (R2 0.89, 0.92, 0.99, 0.98 and p <

0.0001 for 8, 6, 5, and 3 mm disks respectively). Overall, these results demonstrate our

ability to predict oxygen release from our OxySite material without geometrical

restrictions.

Furthermore, a final element analysis model was developed to simulate in vitro

pancreatic islet culture systems in the presence of our OxySite material. Models can

provide guidance for future studies using OxySite and islets by finding the ideal

parameters for islet loading and oxygen tensions that will highlight the impact of in situ

oxygenation. Figure 3-15 summarizes the experimental design of the culture system, as

well as the parameters implemented. Models were performed in 2D for ease of

computation. This model was subsequently translated to experimental studies in

Chapter 4, permitting direct comparison between computational models and

experimental results. On note, a complete explanation of the loading density calculation

from 3D to 2D is presented in APPENDIX A. Figure 3-16 represents 2D models of

oxygen gradients within our systems after 8 h post incubation in different culture settings. No negative oxygen gradients were seen for low or high loading density groups under standard oxygen conditions (Figure 3-16 Panel A-B). Contrariwise, the culture of

islets at high loading density under low oxygen conditions resulted in regions with

oxygen levels below the Km of islets (labeled as hypoxic areas and represented in

black), in particular for the large diameter islets (250 µm). Conversely, the introduction

of OxySite to this culture system resulted in more beneficial oxygen gradients, with no

hypoxic areas present (Figure 3-16 Panel C-D). After 24 h, the detrimental oxygen gradients become more pronounced in low oxygen control groups with almost all islet regions exhibiting hypoxia. While some hypoxic areas begin to emerge in the OxySite group, this was an 80 % decrease compared to the low oxygen control group (Figure 3-

17 Panel C-D). Further, while a decrease in the oxygen tension was observed in the high density standard culture conditions, no hypoxic areas were present. These results are in stark contrast to those obtained experimentally, where cellular fragmentation and necrotic cores were observed for this group (Figure 3-18 Panel A). The disparity of

these results can be reconciled by the difference in oxygen tension in the culture

system. Specifically, our models implemented a theoretical oxygen tension present in

standard incubators (148 mmHg). Yet, reports have demonstrated that the actual

oxygen tensions experienced by cellular constructs under this conditions is

approximately 100 mmHg lower when high cellular loading densities are employed [83].

Further, differences in actual pO2 would be heightened by the presence of highly

metabolic cells such as pancreatic islets. Finally, the model was also implemented to

detect not only hypoxic areas, but the effect of oxygen gradients on projected islet function, as seen by Figure 3-18 Panel B.

Based on the enhanced released kinetics observed for the original prototype

disk, we set to evaluate the effect of OxySite on pancreatic culture systems after two

weeks post initial hydration. For this purpose, OxySite disks were leached as described

for 15 d prior to experiments. For in vitro cultures, rat islets (1500 IEQ) were cultured in

a manner identical to that in our previously published report (and as outlined in the

Methods section) [65]. After 48 h, an increased LDH was measured in the culture

supernatant for control samples, when compared to OxySite (Figure 3-19 panel A).

Moreover, low oxygen exposure resulted in complete loss of glucose responsiveness for

the cultured islets, as observed by a lack of insulin response following glucose

stimulation, with a mean index of 0.98 ± 0.01. Contrarily, islets cultured in the presence

of OxySite under low oxygen tensions exhibited a normal insulin response to glucose

stimulation, with a mean index of 2.39 ± 0.03 (Figure 3-19 panel B).

Gene expression of OxySite treated islets demonstrated a decrease of glycolytic

enzymes Glut1 and Pgk1, when compared to low oxygen controls. Likewise, down-

regulation of the inflammatory marker Nos2 was observed in OxySite containing

samples, when compared to controls. Of interest, islets cultured in the presence of

OxySite resulted in an up-regulation of Vegfa, when compared to low oxygen controls

(Figure 3-20 panel A). Moreover, live/dead imaging of samples cultured under low

oxygen demonstrated an increase in cellular debris and dead cells within the center of

islet clusters. Conversely, islets cultured with OxySite exhibited intact morphology with a

high presence of live cells (Figure 3-20 panel B).

Conclusion

The main objective of this thesis was to optimize our in situ oxygen generating material, OxySite to support islet viability and function when transplanted within extrahepatic sites. OxySite is based on the encapsulation of calcium peroxide within the

hydrophobic polymer PDMS. Peroxides have been applied in the past for the

bioremediation of estuaries due to their oxygen generating capabilities [84, 85].

Furthermore, calcium peroxide is the most widely applied among oxygen generating

peroxides, as it provides slower release kinetics of decomposition, thereby extending

the duration of oxygen release [86]. Previous research conducted in our laboratory had

demonstrated the capability of calcium peroxide to generate oxygen when hydrolytically

activated. Moreover, our ability to modulate kinetics of oxygen release was explored by

encapsulating the reactive agent (i.e. calcium peroxide) within different polymers (i.e.

agarose, PLGA, PDMS) [87]. Findings from these studies demonstrated the unique

ability of PDMS to modulate the release of oxygen and prolong the sustainability of

release into the milieu for > 2 weeks in an open system [65].

PDMS is a non-degradable polymer that has been extensively implemented in the drug

delivery field [88, 89]. The biostability and biocompatibility of PDMS have made it an

attractive choice for the development of controlled drug delivery systems [89].

Moreover, its hydrophobic nature permit ease in the solubility of many agents,

permitting it to serve as a loading depot for hydrophobic drugs to achieve sustain

release, such as in the delivery of contraceptives for periods > 1 month [88, 90]. Further,

a coated rod formulation fabricated out of PDMS and loaded with ivermectin provided

prolonged release with nearly zero order kinetics [91]. Despite these notable findings, a

comprehensive characterization of the biomaterial properties and their effect on oxygen

release modulation was absent. Thus, the purpose of this chapter was to investigate the

properties of our oxygen generator, in particular after hydration for selected periods of

times, in order to gain an understanding of the mechanism of oxygen release from the

device that could then be translated into a mathematical model.

Initial experiments on macroscopic changes post hydration demonstrated the limited water uptake of our system, with no noticeable changes in dimension or biomaterial weight after extensive exposure to a hydrating medium. These results demonstrated that polymeric swelling, or material degradation had no effect on driving oxygen release from our polymeric matrix. Moreover, studies examining the material

penetrability observed enhanced porosity in the system when calcium peroxide crystals

were incorporated into the polymer prior to casting. Exposure of the biomaterial system

to hydration for different periods of times lead to an increase in material porosity and

enhanced pore size distribution. Pore size and total material porosity proved to be an

invaluable tool in determining other significant properties that could influence release

kinetics; in particular the diffusional path inside the system, tortuosity, and how this

affected the diffusion coefficient of oxygen into our system. Understanding how the

presence of solid particles within the OxySite device would affect the diffusional path

was a critical parameter for our theorized model of oxygen release; in which the

availability of reactive agent on the surface of these pore throats upon water contact

would dissolve and form pathways for oxygen diffusion. Equations relating all three

parameters have been developed for systems of different pore sizes and shapes [73,

74]. Based on the SEM images obtained for our devices, at different time points post

leaching, it was determined that a cylindrical pore size model would provide the best fit

to our system. Thus, a previously developed model [74] was implemented to calculate how the formation of 3D cylindrical pore throats within the OxySite platform would affect diffusion out of the matrix.

Tracking of calcium peroxide changes within the material was attempted via nanoCT imaging. Reconstruction of x-ray scans demonstrated the presence of two energy attenuations, background and a broader peak corresponding to the OxySite material. Thresholding the OxySite energy distribution from the collected scans permitted the separation of the PDMS material from the entrapped calcium peroxide to perform semi-quantitative volumetric analysis of peroxide loading. Loading densities calculated from pixel quantification demonstrated packing densities within the same range as those calculated experimentally for material fabrication; providing corroboration on the loading accuracy of the OxySite prototypes. Further, upon inspection of reconstructed images, a distinct gradient of high energy clusters, theorized to be calcium peroxide particles, was observed. Quantification of gradient concentration was performed on disks leached for different time periods in order to track calcium peroxide changes based on water dissolution and diffusion out of the matrix; however, no discernible changes in concentration was detected in the evaluated samples, despite the distinct visual differences observed within the devices. The low signal-to-noise ratio

(SNR) and consequent lack of three discernible energy peaks within the histogram, hindered the ability to fully segment the materials. Thus, the accuracy of the quantification and tracking of changes in agent concentration was limited. In the future, a higher energy level should be employed during imaging to improve overall SNR; allowing for distinct voxel segmentation to provide a more accurate quantitative

assessment. Moreover, it cannot be discounted that the lack of changes in particle concentration calculated from this techniques was a result of calcium hydroxide

(secondary byproduct from calcium peroxide decomposition) remaining entrapped

within the PDMS disk. In the future, pure calcium hydroxide and calcium peroxide

powders should be evaluated to determine if attenuation energies of both chemicals fall

within an overlapping range. As this is a new technology employed for biomaterial

characterization, optimization of the technique is needed; however, the data collected

herein establishes the potential of this method in the noninvasive characterization of

biomaterials.

Moving forward, a comprehensive analysis of the release kinetics of oxygen from the system was performed for our primary OxySite prototype geometry (10 mm diameter; 1 mm height). Kinetic assessment of oxygen from these prototypes revealed the ability of our improved devices to generate oxygen at physiological relevant values for > 40 d. This was the first time that a comprehensive measurement of oxygen release from our OxySite disks was conducted. Further, there was robust reproducibility within samples (i.e. disks cut from different mother slabs). As one of our goals was to

personalize these OxySite devices to different cell culture platforms and tailor the release to different oxygen demands, the effects of geometry variations on oxygen release from our system was investigated. Examination of the impact of prototype

diameter on oxygen kinetics demonstrated the ability of these improved OxySite

platforms to generate oxygen at substantial levels, even after a 50% size reduction.

These results validate the pliability of our devices and provide a useful screening tool for

validating mass transport mechanisms involved in the oxygen release from our OxySite devices.

Characterization of the presence of potentially toxic byproducts following the decomposition of our reactive agent in our original prototypes was also conducted, with extensive assessment of ambient hydrogen peroxide for over > 40 d. Results demonstrated the presence of hydrogen peroxide, but at nontoxic concentrations [78,

79] within our OxySite systems. Likewise, ambient calcium levels were measured, as calcium could impact regulation of islet function [80]. Measurements at different time points demonstrated spikes in calcium concentrations at particular intervals; however, levels were well below that shown to impart cytotoxic effects [80]. In the future, it would be of interest to investigate if there is any relation between these calcium peaks and spikes in the oxygen kinetics from the OxySite devices.

An equally significant aspect of this thesis was the development of a mathematical model that could predict the release kinetic of oxygen from our OxySite platforms. Based on the extensive characterization collected from our devices, it was hypothesized that a Fickian diffusion model of solid transport from a polymeric matrix could fit the oxygen release kinetics observed from our OxySite prototypes. In the literature, there is a plethora of well-established models describing controlled release of various agents from polymeric matrices [66, 82]. Based on the properties of our system, a model describing diffusion out of a cylindrical monolithic matrix was chosen [67].

Further, this model was simplified from a 3D scale to a 2D scale, as it was theorized that the observed oxygen release originated predominantly from the surface of the

OxySite material. Moreover, the system was simplified from a multiple driving force

mechanism (i.e. water diffusion, calcium peroxide decomposition and oxygen diffusion), to a single transport process (i.e. oxygen diffusion). Despite, the low water uptake seen in our systems, it was hypothesize that the driving force of water migration and activation of calcium peroxide would be orders of magnitude higher than that of oxygen diffusion out of the system; making the limiting step oxygen migration out of the matrix.

Following the completion of our mathematical model development, theoretical curves were compared to experimentally obtained curves. Plotting oxygen generation as cumulative release, a significant correlation in oxygen profiles for the two systems for our original OxySite prototype was found (R2>0.9). This model was further validated by

comparing theoretical to experimental curves for other dimensions of OxySite.

Comparison to four other geometries revealed the sensitivity of the model, validating the results obtained from our porosity analysis, as well as the implemented tortuosity model.

Thus, we can say that we have developed a mathematical model that is able to

accurately predict the kinetics and duration of oxygen release from our OxySite devices.

Future experiments will test the validity of this model against OxySite devices with

different loading densities and different shapes to authenticate the robustness of this

system.

With the possibility to predict the effect of OxySite parameters (i.e. geometry, and

calcium peroxide loading) on the resulting oxygen kinetics using our proposed model,

the development and tailoring of different prototypes based on personalized

requirements is now feasible. Our ability to do so was tested with the development of a

computational model, specifically finite element analysis, which seeks to predict oxygen

gradients within in vitro pancreatic cultures. Results demonstrated enhanced

oxygenation in culture systems containing our OxySite device (implementing the flux derived from the kinetics of oxygen release) based on controls. These results highlight our ability to implement computational models based on predicted release kinetics, and to evaluate the potential effects they could have on cellular culture systems. This is of valuable importance in planning and developing experiments, as it can significantly

reduce the time and money spent on evaluating potential outcomes of a plethora of

cellular combinations and oxygen delivery profiles.

Finally, we wanted to validate in vitro the robust oxygen release kinetics

observed from the primary OxySite prototype at intermediate time points (> 10 d).

Previous experiments looking at the effects of our oxygen generating materials on

pancreatic islet cultures were performed immediately after sterilization of the material

(see Chapter 4 and [65]). Herein, we wanted to experimentally evaluate if at the

observed tapered oxygen levels (219 mmHg post initial hydration compared to 45.7

mmHg at 15 d), was our material able to support viability of islet cultures under low

oxygen conditions. In vitro assessment of this system revealed an enhance insulin

responsiveness to glucose stimulation, as well as a reduction of cytotoxic byproducts of

islets cultured with OxySite compared to low oxygen controls. This results hint at the

potential of our system to maintain viability and functionality of pancreatic cultures for

extended time periods - a quality that is of exceptional importance when translating

these platforms in the in vivo setting.

Overall, herein we have demonstrated the development of an enhanced in situ

oxygen generating device capable of supplying oxygen at significant levels for >40 d.

Moreover, we have presented extensive characterization on the material properties over

time. Of note, we developed a mathematical model capable of predicting, with great accuracy, oxygen kinetic profiles from different OxySite prototypes. Finally, we demonstrated our ability to implement this in a computational models to predict cell viability outcomes.

Figure 3-1. Schematic diagram of OxySite disks dimensions in cylindrical coordinates (z,r).

Figure 3-2. Macroscopic characterization of OxySite material. A) Cartoon representation of OxySite disks. Inset represents actual photograph of material post fabrication. B) Pictogram representation of OxySite material after various leaching times. Strong differences in optical denisties can be observed in material after 15 d post leaching. C) Dimensional changes in thickness (open circle) and diameter (black squares) of OxySite after hydration for 60 d. D) Weight changes in OxySite material after hydration for 60 d.

Figure 3-3. Porosity assessment of OxySite material via mercury porosimeter. A) Changes in porosity of OxySite material compared to a blank PDMS disks as well as after different leaching times. B) Average pore size after low and high pressure mercury intrusion for OxySite disks leaching at 0, 7, 30 and 100 days.

Figure 3-4. Percent volume intruded vs pore diameter distribution for different OxySite disks. A) Blank PDMS after 1 day leaching, B-D) OxySite after 1-7-100 days post leaching respectively.

Figure 3-5. Surface characterization of OxySite disks. A) FTIR spectrum of OxySite disks (magenta, green, and blue), blank PDMS (black), and CaO2 powder (red). B) XPS binding energy spectrogram of OxySite disk without leaching. Representative peaks for silicone, carbon, and oxygen were found on surface.

Figure 3-6. Surface characterization of OxySite disks via SEM. Representative images of the surface and a cross section of the OxySite material 15 days post leaching are shown. Pores of 100 to 250 µm were found within the material. Scale 100 µm.

Figure 3-7. Surface characterization of OxySite disks via LSM. Representative images of the surface of the OxySite material 1, 10, and15 days post-leaching are shown. Surface roughness was more pronounced on samples after increased leaching times.

Figure 3-8. A) Illustration representing sample collected via nanoCT. Brighter voxels within the samples were observed at different positions within the disk. B) Quantification of these voxels vs total voxel number revealed loading density within the sample. Loading density was calculated for all OxySite samples evaluated.

Figure 3-9. A) Table representation of the different layers of high density voxels found within the OxySite material for all time intervals evaluated. B) Selected images chosen to calculate gradient for OxySite sample leaching in buffer for 5 d. Samples were at different depths within the material. Slices at the same location were chosen for OxySite disks at all time points evaluated.

Figure 3-10. Concentration of calcium peroxide reaction byproducts in supernatant at different leaching time points. A) Hydrogen peroxide concentration of both Blank PDMS (control) disks and OxySite disks at selected time points for 60 days. Orange box represents levels known to be toxic to cells. B) Concentration of calcium released from OxySite disks after select leaching times.

Figure 3-11. Comprehensive oxygen kinetic release from OxySite disks of 10 mm in diameter for 30 days. Chamber was refreshed every day. Samples were measured for 6 to 24 hours.

Figure 3-12. Mathematical model of oxygen release from OxySite disks. A) Cumulative release of oxygen release from OxySite disks collected via noninvasive oxygen sensing chambers (black), theoretical oxygen release profile from model (red). B) Order of magnitude analysis of mathematical model fit.

Figure 3-13. Oxygen release kinetic and mathematical modeling of release kinetic profiles of OxySite disks of different dimensions. Cumulative release of oxygen release from OxySite disks collected via noninvasive oxygen sensing chambers (black), theoretical oxygen release profile from model (red). A-B) OxySite disks of 8 mm in diameter. C-D) OxySite disks of 6 mm in diameter.

Figure 3-14. Oxygen release kinetic and mathematical modeling of release kinetic profiles of OxySite disks of different dimensions. Cumulative release of oxygen release from OxySite disks collected via noninvasive oxygen sensing chambers (black), theoretical oxygen release profile from model (red). A-B) OxySite disks of 5 mm in diameter. C-D) OxySite disks of 3 mm in diameter.

Figure 3-15. Experimental design and parameter implemented for COMSOL modeling.

Figure 3-16. Representative FEM simulation models illustrating effects of loading densities and external oxygen tensions on oxygen availability to pancreatic islets after 8 h of culture either at: A) standard oxygen and loading density (Free Islets); B) standard oxygen and high loading density with a PDMS blank control (0.2 mM Oxygen Control); C) low oxygen and high density with a PDMS blank control (0.01 mM Oxygen Control); or D) low oxygen and high density with an OxySite disk (0.01 mM OxySite). Hypoxic areas were defined as zones with an oxygen concentration below 1×10-4 mol/m3 and labeled black. Scale = mM oxygen.

Figure 3-17. Representative FEM simulation models illustrating effects of loading densities and external oxygen tensions on oxygen availability to pancreatic islets after 24 h of culture either at: A) standard oxygen and loading density (Free Islets); B) standard oxygen and high loading density with a PDMS blank control (0.2 mM Oxygen Control); C) low oxygen and high density with a PDMS blank control (0.01 mM Oxygen Control); or D) low oxygen and high density with an OxySite disk (0.01 mM OxySite). Hypoxic areas were defined as zones with an oxygen concentration below 1×10-4 mol/m3 and labeled black. Scale = mM oxygen.

Figure 3-18. A) Bright field images of islets culture under standard oxygen and loading density (Free Islets); standard oxygen and high loading density with a PDMS blank control (0.2 mM Oxygen Control); and low oxygen and high density with an OxySite disk (0.01 mM OxySite). B) Summary of Oxygen Impacts on Islet. % area of total islet area (cross-section) where the oxygen tension was below necrotic (CCR-VIABILITY) or non-functional (CCR-INSULIN) levels, as predicted from multiphysics modeling.

Figure 3-19. Viability and function of rat islets exposed to OxySite 15 d post leaching and incubated for 48 h at 0.01 mM Oxygen. A) LDH release in culture milieu from islets exposed to low oxygen conditions with or without OxySite. B) Static glucose stimulated insulin release from islets post culture. Data Presented as mean ± SEM.

Figure 3-20. mRNA expression and confocal images of islets exposed to OxySite 15 d post leaching and incubated for 48 h at 0.01 mM Oxygen. A) mRNA expression of glycolytic enzymes, angiogenic factors and inflammatory markers for islets exposed to OxySite under low oxygen conditions. Results are presented as fold regulation from low oxygen controls B) Live/Dead confocal imaging of islets exposed to low oxygen with OxySite demonstrates intact morphology and reduced cellular debris compared to low oxygen controls. Scale 100 µm. Data Presented as mean ± SEM.

CHAPTER 4 MITIGATING HYPOXIC STRESS ON PANCREATIC ISLETS VIA IN SITU OXYGEN GENERATION

Background

Clinical islet transplantation provides an attractive approach to restore endogenous insulin control for Type 1 Diabetics [92, 93]. Poor islet engraftment following transplantation, however, significantly hampers islet function, leading to the need for multiple islet infusions to achieve desired glycemic control and the steady decline of islet function over time [92, 94, 95]. While multiple factors contribute to poor islet engraftment, hypoxia, both in culture and during early and late-stage engraftment in vivo, has been suggested as a major contributor [38, 39, 96, 97]. Isolated islets, stripped of their native dense pancreatic vasculature, suffer from fluctuations in ambient oxygen tension during processing, as well as unfavorable oxygen gradients during culture [98].

Immediately following embolization of the islets in the liver, grafts are subjected to hypoxic conditions due to inadequate oxygen tension of the portal vein and lack of intra- islet vascularization [81, 99]. The significant delay in islet revascularization, as well as the inadequate competency of the resulting vascular network, further exacerbates these unfavorable oxygen gradients [100, 101].

Exposing islets to a hypoxic environment has been shown to lead to deleterious responses. Islets cultured under low oxygen tensions have been found to exhibit loss of glucose stimulated insulin secretion, elevated apoptosis, activation of caspase cascades, release of inflammatory cytokines, and defragmentation of islet structure

Coronel, Maria M., et al. "Mitigating Hypoxic Stress on Pancreatic Islets via In situ Oxygen Generating Biomaterial”. Under review.

[102-106]. A key modulator of activation of these hypoxic pathways is the stabilization of hypoxia-inducible factor-1α (HIF-1α); a tightly regulated protein, modulated by prolyl hydroxylase domain containing enzymes (PHDs) [107, 108]. With the importance of mitigating HIF-1α activation and subsequent deleterious downstream pathways, interventions focused on improving islet oxygenation during culture and in the early engraftment period could be highly beneficial.

We have previously reported the development of a biomaterial capable of in situ oxygen generation following aqueous hydration, herein termed OxySite [65]. The degree of oxygen supplementation of OxySite can be modulated via biomaterial design parameters (e.g. reactive product loading and geometry) to tailor oxygen tensions to the needs of the cultured islets. In this study, we sought to evaluate the impact of oxygen supplementation using OxySite on hypoxic pathways activated within rodent and nonhuman primate islets, with a particular emphasis on cell viability, glycolysis, inflammatory cytokines, and angiogenesis. Based on these positive results, we translated our findings in vivo to quantify the impact of OxySite culture on islet engraftment and function in a diabetic rodent model.

Research Design and Methods

Materials

All chemical reagents were purchased from Sigma-Aldrich, unless otherwise noted. All culture media was sourced from Cellgro. Poly(dimethyl)siloxane (PDMS)

(NUSIL 6215, clinical-grade) was sourced from NuSil Silicone, USA.

Islet Isolation and Culture

All animal procedures were performed under protocols approved by the

University of Miami or University of Florida IACUC and in accordance with National

Institutes of Health guidelines. Rat islet isolations were performed using methods described previously [22], using male Lewis rats (Harlan, Indianapolis, IN). Nonhuman primate (NHP) islets, isolated from Cynomolgus monkeys, were donated from Dr.

Norma Kenyon’s laboratory at the Diabetes Research Institute at the University of

Miami, using methods as previously described [109]. Rat and NHP islets were cultured

2 o overnight under standard culture conditions (1326 IEQ/cm , 37 C, 0.20 mM O2, and

0.05 mM CO2) in supplemented CMRL 1066 medium prior to use.

Fabrication of Oxygen Generating Material, OxySite

OxySite was fabricated as described in Chapter 3. Briefly, 25% w/w CaO2 was

added to PDMS in a 4:1 vol/vol PDMS monomer to platinum catalyst and

mixed/defoamed via centrifugal mixer (Thinky Mixer, USA). The resulting PDMS/CaO2

mixture was then poured into a cylindrical mold (100 mm) to 1 mm thickness and cured

overnight at 40 oC. Prior to use, smaller OxySite disks (10 mm) were punched out of the

larger cured disk. Resulting OxySite disks were sterilized with ethanol for 30 min prior to

performing experiments.

Oxygen Studies with Pancreatic Islets

Based on modeling predictions performed in Chapter 3, 1500 islet equivalents

(IEQ) were seeded within Millicell inserts in 24-well plates supplemented with 700 µL of

CMRL 1066 (loading density = 1326 IEQ/cm2). Treated groups received a single

OxySite disk, placed underneath the insert. Control groups received a blank PDMS disk

of equivalent dimension at the same location. For standard oxygen tension controls,

samples were incubated in a standard incubator set at 0.2 mM (20%) O2 with 0.05 mM

(5%) CO2. For low oxygen conditions studies, islets were incubated in a C-Chamber

Hypoxia Chamber (Biospherix, New York) set to 0.01 mM (1%) O2 with 0.05 mM (5%)

CO2 using controllers (Biospherix). Oxygen levels were validated using oxygen sensor

spots (Presense, Germany). Samples were incubated for 4, 8, or 24 h, depending on

the study. Groups tested and nomenclature include: islets cultured at standard oxygen

conditions with a blank PDMS disk (0.2 mM Control), islets cultured at low oxygen

conditions with a blank PDMS disk (0.01 mM Control), islets cultured at low oxygen

conditions with an OxySite disk (0.01 mM OxySite), and islet cultured under standard

oxygen conditions at low IEQ/cm2 density (Free Islets). Assessments included:

metabolic activity (MTT), LDH, gene arrays, qPCR, dynamic stimulated insulin release,

Live/Dead, cytokine production, oxygen consumption rate, and respirometry

assessments (details below).

For time course studies with non-human primates (NHP), identical culture

conditions were employed. Assessments included: metabolic activity, LDH release,

qPCR, Live/Dead, and cytokine production.

In Vitro Cellular Assessments

The metabolic activity was assessed via MTT (Promega), as previously

published [65]. Briefly, at time points designated, samples were incubated in 250 μL of

complete CMRL media with 28 μL of MTT for 1 h under standard oxygen conditions.

Stop solution (185 µL) was then added under agitation and incubated for 48 h to permit

complete formazan crystals solubilization. Absorbance (570 nm) was then measured

from triplicates samples using a SpectraMax M5 microplate reader (Molecular Devices).

Islet function was assessed via a dynamic glucose stimulated insulin release

(GSIR), using a 12 channel perifusion machine (Biorep, Miami), as described previously

[110]. Islets (75 handpicked and loaded in perifusion columns) were incubated for 1 h in

basal glucose buffer prior to assessment. Islets were then stimulated according to the following series: 10 min basal glucose (3 mM), 20 min high glucose (11 mM), 15 min low glucose (3 mM), 5 min KCL (25 mM), and 15 min basal (3 mM). Insulin concentration was measured using a rat insulin ELISA (Mercodia, Inc., Winston Salem,

NC).

Lactate dehydrogenase (LDH) release was quantified using methods previously

described [65]. Briefly, 100 μL of supernatant was mixed with 100 μL of working reagent

(Cytotoxicity Detection Kit, Roche) and incubated for 30 min at RT. Stop solution (50 μL

of 1 N HCl) was added and the resulting absorbance in triplicate samples was

measured at 492 nm.

Gene expression profiling was performed using PCR arrays. Total RNA was

extracted (RNase easy kit, Qiagen) and reverse transcribed (RT2 First Strand Kit,

Qiagen), according to the manufacturer's instructions. The resulting cDNA was added to

96-well-plate PCR arrays for qPCR (RT2 profiler PCR arrays for hypoxia signaling and

inflammatory cytokines and receptors, Qiagen) assessment using a QuantiStudio 6 flex

real-time PCR system (Life Technologies, California, USA). Relative expression of

analyzed genes was calculated by the ΔΔCt method. PCR array data were represented

in heat maps as fold change from standard oxygen (0.2 mM) controls.

Total RNA for qRT-PCR was isolated (mirVana miRNA Isolation kit, Ambion,

Invitrogen). Reverse transcription of total RNA to single stranded cDNA was performed

with a High Capacity cDNA Reverse Transcription Kit (Invitrogen). Primers used are

listed in Table 4-1. Relative gene expression was calculated using Taqman assays in a

StepOne plus cycler (Applied Biosystems, Life Technologies). Gene expression was

normalized against β-actin for the rat islet experiments and 18 S for the NHP islets. The

ΔΔCt method for relative quantification was used and results were expressed as fold regulation over standard oxygen (0.2 mM) controls.

Inflammatory cytokines and chemokines secreted from islets under experimental conditions were quantified via collection of culture supernatants at designated time points (8 and 24 h) and measurement using milliplex multiplex kits (EMD Millipore) on a

Luminex xMAP (Luminex 200), following manufacturer instructions. Supernatants were incubated with the capture antibodies (MIP1a, MCP1, IL8, VEGFa, and IL-6) for 24 h at

4 oC with shaking prior to measurement. Data was analyzed using Milliplex Analysit

(EMD Millipore).

Live/Dead samples were visualized using the Viability/Cytotoxicity Kit (Invitrogen) and a Leica SP5 inverted confocal microscope, as described previously [65]. After incubation, samples were imaged for live cells (green) and dead cells (red). Z stacks (1

µm thickness; 10 slices per image; 1024x1024; 20x objective) were merged into a 3D projection using LAS lite Leica Software.

The oxygen consumption rate (OCR) of islets was assessed using a custom 4- chamber oxygen sensing chambers (Instech, Plymouth Meeting, PA). All samples were measured concomitantly. OCR measurements were made using methods similar to that described elsewhere [111]. In short, after 24 h culture, 250 IEQ were rinsed in PBS, resuspended in 1 mL of pre warmed CMRL media, loaded into the chamber, and sealed. Oxygen readings were recorded for at least 180 min. Final OCR was normalized to DNA concentration per sample (Quant-iT PicoGreen dsDNA assay kit, Invitrogen).

Islet respirometry assessments after 24 h incubation were performed using an

XF24 (Seahorse Bioscience, Billerica, MA), using methods described previously [112].

Briefly, islets were transferred to XF media (supplemented with 3 mM glucose, 1% FBS without HEPES; Seahorse Bioscience, Billerica, MA), loaded into a XF24 plate (70 islets/well; n = 3/group), and incubated for 1 h in a 0% CO2 incubator at 37 °C before placing in the XF24 respirometry machine (Seahorse Bioscience). OCR was measured at basal (3 mM) and high (11 mM) glucose levels. Final OCR was normalized to DNA concentration per sample. The shift in OCR from basal to high glucose (ΔOCRglc) was calculated as the difference in OCR between high and basal glucose, averaged through the duration of high glucose exposure. Respirometry assessment was performed in a single set of experiments, with triplicates per each sample group. Data is expressed as mean ± SEM.

Islet Transplant and Graft Assessment

Diabetes was induced in Lewis female rats recipients via 2 intraperitoneal injections of 60mg/kg streptozotocin and were only used as recipients after 3 consecutive readings of non-fasting blood glucose levels > 350 mg/dL, as previously described [113]. Islets groups used for transplant were 1800 IEQ cultured for 24 h at 0.2 mM Control (standard oxygen conditions with a blank PDMS disk) or at 0.01 mM

OxySite (low oxygen conditions with an OxySite disk). For transplant, the omental pouch (OP) of the anesthetized rat was exposed and spread. Islets were collected into a

Hamilton syringe, distributed onto the exposed OP, and sealed in place using a fibrin hydrogel, similar to that previously described [114]. Briefly, 15 µL of a thrombin (4

U/mL), 9 mM CaCl2, 150 mM NaCl, (on ice) was quickly mixed with 15 µL of fibrinogen

(8 mg/mL), aprotinin (85 ug/mL) and 20 mM HEPES solution at RT, and placed on top of the islets. The omentum was then folded over the graft and the seam was sealed using the same fibrin hydrogel. Note, fibrin gel was used to seal the omentum in lieu of

sutures for ease in closure and mitigation of surgical issues associated with suturing

[113, 115]. The OP was placed back in the peritoneal cavity, the incision was sutured,

and the skin was closed using surgical staples.

Blood glucose (BG) and body weight (BW) of recipients were monitored.

Normoglycemia was defined as stable nonfasting glycemic levels < 200 mg/dL for at

least two consecutive BG readings. For metabolic assessment of grafts, an intravenous glucose tolerance test (IVGTT) was performed on 30 d post-transplant on selected animals from both groups (OxySite n = 5; Control n = 3), as previously described [114].

Briefly, after overnight fasting, rats received a bolus of glucose (IV; 2 ul 50% dextrose / g BW) and BG levels were monitored until reading < 200 mg/dL or after 100 mins. The area under the curve (AUC) of glucose was calculated, as previously described [8]. To ensure observed normoglycemia was due to the islet graft and not residual native pancreas function, the islet-loaded omentum was removed in a survival surgery and the animal BG was monitored for a minimum of 3 d.

Histological Assessment

Explanted OPs were fixed in 10% formalin buffer and embedded in paraffin blocks. Resulting slides were stained for insulin (Dako A0564, 1:100), SMA (Abcam ab5694, 1:50), glucagon (Abcam ab10988, 1:50), pimonidazole (hypoxyprobe, 1:50) and DAPI (Invitrogen D1306, 1:500), as described elsewhere [116]. Histological images were collected using a Zeiss LSM 510 confocal microscope.

Statistical Analysis

A minimum of three independent experiments (e.g. three separate islet

isolations) with a minimum of triplicates per sample/group/assay performed for each

assessment, with the exception of gene microarrays, OCR, and respiratory

measurements, which was collected from one islet isolation. Comparisons between

groups were made using the same islet preparation, with graphs summarizing results

from a representative experiment. Unless otherwise stated, results are expressed as the

mean ± SD. Statistical analysis was performed using the unpaired Student’s t test for

pairwise comparisons and two-way analysis of variance (ANOVA) for multiple

comparisons. Post-hoc analysis for multiple comparisons was done using a Bonferroni-

Holm test. Statistical significance was considered at P < 0.05.

Results

To evaluate the impact of OxySite on alleviating hypoxia-induced islet death and

dysfunction, a time-course culture study was performed. Cell culture conditions of

environmental oxygen tension and islet loading density (IEQ/cm2) were altered to

highlight the impact of oxygen supplementation and more closely resemble loading

densities used for transplantation [117, 118]. Multiphysics modeling was employed to

characterize predicted oxygen tensions within culture conditions for selection of final

parameters (see Chapter 3 for details). In summary, pancreatic islets were cultured up

to 24 h at a loading density of 1,326 IEQ/cm2 under standard incubator (0.2 mM) or low

(0.01 mM) oxygen culture conditions. To provide oxygen supplementation for low

oxygen cultures, a single OxySite disk (1 mm thick; 10 mm diameter) was placed within

the culture dish (labeled OxySite). For control groups, an inert material (PDMS-only) of

identical dimensions was used (labeled 0.01 mM Control and 0.2 mM Control). Rat

pancreatic islets were the primary species studied; however, to validate observations in a species more closely resembling human islets, nonhuman primate pancreatic islets were used [119, 120].

Initial tests focused on validation of hypoxic conditions and evaluation of global islet loss. While oxygen spot sensors confirmed environmental conditions (data not shown), immunohistochemical analysis of whole islets exposed to low oxygen conditions depicted increase accumulation of hypoxia sensible dyes in the control group compared to those treated with OxySite (Figure 4-1). Further, gene expression profiling characterized activation of hypoxic pathways, specifically early and late stage hypoxia- inducible factors (HIFs) and selected downstream genes (Figure 4-2 panel A). As expected, exposure of islets to low oxygen (0.01 mM) lead to strong up-regulation of early hypoxic markers, such as HIF-1α, and downstream regulators (e.g. Arnt, Per1,

Hnf4a and Cops5) after both acute (8 h) and chronic (24 h) exposure, when compared to 0.2 mM controls. The second-phase HIF gene, HIF-3α, and its downstream targets

(Nrdg1 and Ruvbl2) were also prominently upregulated. OxySite treated cultures still exhibited upregulation of markers associated with HIF-1α stabilization; however, second phase HIF genes were downregulated.

A major consequence of accumulation of HIF1a is the activation of deleterious apoptotic pathways [102]. Supplementation of low oxygen cultured islets with OxySite resulted in gene switching from up- to down-regulation of pro-apoptotic proteins, such as Bnip3, Bnip3l, Ddit4, Adm and Btg1, with comparable up-regulation of anti-apoptotic genes (Ler3 and Pim1). Similarly, nitric oxide, which plays a prominent role in the

100

regulation of beta cell suicide, was strongly down-regulated for OxySite treated islets, even when compared to islets cultured under 0.2 mM oxygen.

The impact of apoptotic gene activation on overall islet metabolic activity and viability was characterized. Results found a significant decline in metabolic activity for the low oxygen tension group after 8 and 24 h culture (29.8 and 59.6% decrease, respectively) (Figure 4-2 panel B). Metabolic activity for rat islets cultured under 0.01 mM oxygen with OxySite, however, was significantly higher than even the 0.2 mM oxygen control group at all time points. Observations were validated using Live/Dead imaging (Figure 4-2 panel C), where evidence of rat islet fragmentation and increased cell death in the low oxygen control group emerged by 8 h and was amplified by 24 h.

Conversely, robust, intact, and viable islets were found when OxySite was added to the cultures.

Similar trends in preservation of islet viability using OxySite were also observed for nonhuman primate islets, as demonstrated in Figure 4-3 panel A, with preservation of islet morphology when OxySite was added to low oxygen cultures. Activation of apoptotic pathways was quantified via the expression of BCL-2 family pro-apoptotic members, Bax and Bak. Intriguingly, we observed a significant up-regulation of Bax, in the OxySite treated group compared to the low oxygen and standard oxygen controls at

8 h. This trend was reversed after chronic exposure (24 h) with a significant up- regulation of these pro-apoptotic proteins in the low oxygen control, compared to

OxySite treated islets. Of note, recent evidence suggests that BCL-2 proteins might play dual roles in balancing stimulation of apoptosis with islet metabolism and insulin secretion coupling [121]. Moreover, Bax has recently been linked to mitochondrial

101

energy production, with deficiency of this protein being linked to blunted ATP biosynthesis in colorectal cancer cells [122].

With the capacity of OxySite to support islet viability under hypoxia established, its impact on activation of pathways associated with glucose processing and insulin responsiveness was explored. It has been well established that hypoxic stabilization of

HIF-1α results in the transcription of glucose transporters such as, Glut1 and Glut2, and glycolytic enzymes, such as Pgk1 [123]. Gene expression profiling focused on glycolysis pathways demonstrated an expected up-regulation of all enzymes involved in the breakdown of glucose, as well glucose transporters that facilitate glucose diffusion to the cytosol (Figure 4-4 panel A). This enhanced expression, together with the accumulation of downstream by-products of glycolysis such as Ldh and Pdk1 that block access of glycolytic end products to the mitochondria [123], suggests a prominent shift to anaerobic glycolytic pathways for pancreatic islets exposed to low oxygen tensions.

In contrast, supplementation of low oxygen cultures with OxySite reversed this phenotype, with decreased expression of glycolytic enzymes and transporters when compared to even standard oxygen controls. These results indicate the ability of

OxySite to maintain electron transport respiration. Trends were validated using qPCR of glucose transport (Glut1), glycolytic enzymes (Pgk1), and by-products of anaerobic glycolysis (Ldh) (Figure 4-4 panel B-E), whereby significant down-regulation of all genes were observed for the OxySite treated group, even when compared to islets cultured under standard incubator conditions (P < 0.001). Further, elevated Ldh gene expression translated to increased LDH protein in the culture media, although no significant

102

difference in LDH protein levels was measured between OxySite and 0.2 mM oxygen cultured islets.

Examination of pancreatic islet intracellular ATP concentrations and oxygen respiration served to confirm glycolysis gene pathway analysis (Figure 4-5 panel A).

Exposure of islets to low oxygen tensions lead to a reduction in ATP production, while the addition of OxySite not only restored but increased ATP levels to above standard oxygen controls. Islet oxygen consumption rates (OCR) also indicated decreased islet metabolic activity at low (0.01 mM) and standard (0.2 mM) oxygen culture conditions, with depressed OCR (430 and 170 nmol/min-ng DNA, respectively), when compared to control islets (Free Islets, e.g. islet cultured under standard culture and density conditions; 973 nmol/min-mg DNA) (Figure 4-5 panel B). The addition of OxySite to low oxygen cultures increased OCR to levels above the 0.2 mM control (673 nmol/min-mg

DNA). Delineation of the impact on beta cells within the pancreatic islet was investigated by measuring OCR under glucose challenge (Figure 4-5 panel C). Up- regulation of OCR relative to basal after glucose simulation was stronger in islets co- cultured with OxySite (0.71 ± 0.23 pMol/min·ng/mL), when compared to both 0.01 mM and 0.2 mM control islets (0.26 ± 0.11 and 0.25 ± 0.12 pMol/min·ng/mL, respectively).

Quantification of this trend via ΔOCRglc found comparable results for free islets and islets co-cultured with OxySite, at 0.24 ± 0.10 and 0.23 ± 0.07 nmol/min per 100 islets, respectively (Figure 4-5 panel D). In contrast, the stimulated OCR for low oxygen and standard oxygen controls was 0.06 ± 0.03 and 0.08 ± 0.04 nmol/min per 100 islets, accordingly. Overall, the preserved capacity of mitochondrial glucose oxidation further supports the capacity of OxySite to suppress anaerobic glycolysis.

103

A critical impact of hypoxia on pancreatic islets is the loss of insulin transcription and insulin responsiveness to glucose stimulation due to mitochondrial dysfunction

[102]. For this purpose, a dynamic glucose stimulated insulin release test was conducted after 8 and 24 h culture (Figure 4-4 panel F-G). Culture in low oxygen conditions for 8 h resulted in limited response to high glucose stimulation, with a single, late, and unsustained peak near the end of the stimulation period. Intracellular insulin was not diminished, as validated by islet depolarization via high KCl stimulation (SI =

11.8 ± 10.7). By 24 h of low oxygen culture, islets exhibited a complete loss of glucose responsiveness and minimal insulin content, with impaired KCl stimulation insulin release (SI = 3.29 ± 1.98; P = 0.031 compared to OxySite treated group). While the first phase peak for the OxySite group was dampened after 24 h, this did not impact overall responsiveness, as the AUC was statistically comparable to 8 h (P = 0.06; Figure 4-4 panel H). Total insulin content was impacted by 24 h (P = 0.013), but was significantly higher than low oxygen controls (P = 0.005).Total insulin content was impacted by 24 h

(P = 0.013), but was significantly higher than low oxygen controls (P = 0.005).

Examination of the impact of low oxygen tension and the presence of OxySite on glucose processing was also conducted for nonhuman primate islets. Similar trends were observed with decreased activation of Glut1, Pgk1, and Ldh genes and LDH protein release for NHP islets co-cultured with OxySite under low oxygen conditions.

(Figure 4-6).

The response to hypoxic stress is tightly coupled to the immune response via

HIF-1α activated NF-κb signaling [124]. Liberation of NF-κb from the cytoplasm results in the transcription of downstream target genes, including inflammatory cytokines, such

104

as Ccl2 and iNos [106]. Moreover, activation of metabolic stress in pancreatic islet cultures has been shown to be responsible for 60-80% of the enhanced Il-1b, Il-6 and

TNF-a production by islets [125]. As such, examination of the effect of hypoxia on the mRNA expression of inflammatory markers, as well as their subsequent cytokine production, was of interest.

Gene expression profiling of known islet inflammatory regulators found a general up-regulation of most pro-inflammatory cytokines and subsequent islet-derived chemokines after 8 and 24 h exposure to low oxygen (0.1 mM). OxySite mitigated this response at 8 h, resulting in down-regulation of several of these factors, including Ifng,

Il17a, Cxcl2, Ccr2, and Cxcr2. This suppression became more pronounced after 24 h, with 8 of the 15 genes studied down-regulated, even when compared to 0.2 mM control islets (Figure 4-7 panel A).

A more in-depth analysis of inflammatory effectors, iNos and Ccl2, was performed via qPCR. While all groups exhibited up-regulation of iNos at 8 h, significant down-regulation was observed in the OxySite treated group, when compared to both low oxygen (P < 0.001) and standard oxygen controls (P < 0.0005) after 24 h.

Interestingly, iNos expression was also down-regulated in low oxygen controls when compared to those cultured under standard oxygen concentration (P < 0.0005). In accordance with gene microarray data, mRNA expression of Ccl2 was highest for islets cultured with OxySite at 8 h; however, after 24 h, the OxySite group was comparable to

0.2 mM controls, while low oxygen controls expressed elevated Ccl2 expression (P <

0.0005; Figure 4-7 panel C).

105

To evaluate if changes in gene expression translated to modified secretion of

inflammatory factors, IL6, IL-8, CCL2, and MIP-1α levels in the incubation media were

measured. A significant elevation in MIP-1α was observed after 24 h culture in low

oxygen, when compared to OxySite supplemented (P < 0.001) or standard oxygen

cultures (P < 0.05; Figure 4-7 panel D). Conversely, OxySite treated cultures exhibited a

significant elevation in CCL2 release, compared to both low and standard oxygen

controls throughout duration of the experiment (P <0.05, < 0.0001, and < 0.0001 at 4, 8,

and 24 h, respectively) (Figure 4-6 panel E). No change in IL-6 or Il-8 release was observed between groups (Figure 4-7 panel F-G).

Characterization of inflammatory pathways of nonhuman primate islets validated aforementioned observations (Figure 4-8), whereby elevated secretion of CCL2 was

observed for islets co-cultured with OxySite under hypoxic conditions at 8 h, with

reversal of this trend after 24 h; however, elevated IL-8 and IL-6 release for low oxygen cultures of nonhuman primate islets was detected after 24 h, when compared to

OxySite (P < 0.0005, < 0.001) and standard oxygen controls (P < 0.0005, < 0.001).

These results indicate dampening of inflammatory pathways in higher order species, although further investigation is needed.

Of particular interest in the development of strategies to alleviate HIF-1α downstream pathways is its potential impact on neovascularization. It is well documented that stabilization of HIF-1α results in activation of pro-angiogenic factors, such as vascular endothelial cell growth factor (VEGF), which can promote islet revascularization. A screening of pro-angiogenic genes (Figure 4-9 panel A) demonstrated up-regulation of major HIF induced pro-angiogenic factors (Vegfa, Pgf,

106

Angptl4, and anxa2), as well as some of their respective downstream regulators (Lox,

Hmox1, Jmjd6, Serpine1, and Mmp9), after 8 h low oxygen exposure, when compared to standard oxygen controls. This trend was sustained and even elevated for selected factors after chronic exposure (24 h). The addition of OxySite to the culture lead to down-regulation of most angiogenic regulators and their downstream factors. The quantification of Vegfa mRNA expression and subsequent VEGF protein secretion

(Figure 4-9 panel B-C), however, revealed contrary trends. While significant down- regulation of Vegfa was quantified for both the 0.01 mM control and OxySite treated groups after 8 h, this trend was not translated at the protein level, as VEGFa protein levels were significantly higher for the OxySite treated group than low and even standard control groups at both 4 and 8 h. As protein studies examine cumulative protein release during the designed time period, it is reasonable to suspect that early up-regulation of Vegfa in the OxySite group (e.g. < 8 h) could lead to these results.

Further, gene expression of Vegfa for low oxygen control groups were unexpectedly down-regulated for all time points studied. This result is inconsistent with microarray studies; however, the robust nature of the qPCR analysis (3 isolations vs 1 isolation) favors the conclusions of the qPCR data. After 24 h culture, protein levels begin to reflect gene expression levels, whereby low oxygen cultures exhibit the lowest VEGFa protein levels, followed by the 0.01 mM oxygen OxySite treated and 0.2 mM oxygen control groups (P < 0.001 and < 0.0005, respectively).

Validation of results for nonhuman primate islets lead to similar conclusions, although gene expression data indicate greater sensitivity to culture conditions (Figure

4-10). Specifically, Vegfa was up-regulated for both low and standard oxygen control

107

cultures at both 8 and 24 h time points, while OxySite treated islets exhibited significant down-regulation (P < 0.0001 at both time points). As with the rat islets, gene expression profiles did not mirror cumulative protein release, as the levels of VEGFa protein in the culture media was inverted, with the highest VEGFa concentration observed for OxySite treated islets. Nevertheless, results indicate that treatment of hypoxic islets with OxySite did not impair VEGFa protein release.

The impact of oxygen culture conditions on the resulting efficacy of islets was tested using a syngeneic diabetic rat model. In this experiment, two groups were tested:

1) islets cultured under low (0.01 mM) oxygen with OxySite; and 2) islets cultured under standard (0.2 mM) oxygen with an inert PDMS disk. Following 24 h in vitro culture, islets were retrieved and transplanted into the extrahepatic omental pouch site, using a previously published procedure and similar to the approach currently being used in clinical trials (P.I. Rodolfo Alejandro, clinicaltrials.gov: NCT02213003) [114, 126]. Within the first 10 days post-transplantation, stabilization of diabetic rats to euglycemia was observed in 63% (5 out of 8) of the recipients of OxySite treated islets (Figure 4-11 panel A), with grafts remaining stable until elective omentum removal 40 days post- transplant. Restoration to the diabetic state was observed following elective graft explantation, verifying that normoglycemia was due to the transplanted islets. The remaining 3 OxySite treated recipients exhibited blood glucose levels that fluctuated between 100 and 300 mg/dL for most of the 40 d study. This was in stark comparison to recipients of 0.2 mM oxygen control islets, whereby 0% (0 out the 8) reverted to normoglycemia and all exhibited blood glucose levels exceeding 250 mg/dL.

108

To further assess engraftment potency, an intravenous glucose tolerance test

(IVGTT) was performed on euglycemic recipients of OxySite treated islets (n = 5), as well as recipients of 0.2 mM oxygen control islets (n = 3), with a nondiabetic animal used for metabolic reference. Animals transplanted with OxySite cultured islets cleared glucose in a manner comparable with that of naïve animals (AUC = 8677 ± 505 vs

AUC= 7505, respectively) (Figure 4-11 panel B). As expected, nonstable recipients of

0.2 mM oxygen cultured islets failed to clear glucose in an efficient or effective manner

(AUC = 25496 ± 7584, P = 0.0186 vs OxySite cultured islets).

Immunohistochemical analysis for insulin, glucagon, and SMA on explanted grafts for OxySite treated islets revealed preserved cytoarchitecture with expected insulin-positive beta cells surrounded by glucagon-positive alpha cells and intra-islet

SMA staining, indicating intra-vascularized islets within the omentum tissue (Figure 4-11 panel C). In contrast, control grafts exhibited smaller and/or fragmented islets that lacked the architecture and organization of the treated group, although insulin and glucagon positive cells and SMA-positive tubules were observed. These data support graft efficacy observations with preserved islet viability and function for OxySite treated islets.

Conclusion

Shortage of oxygen for transplanted islets after liver embolization has been recognized as a major factor contributing to significant initial graft loss and long-term graft failure [127]. Oxygen deprivation is paradoxically linked to the inappropriate accumulation of free radicals, which causes stress on proteins and DNA in the cell

[128]. Free radical formation is a critical concern in the field of islet transplantation, as islets are highly susceptible to injury mediated by reactive oxygen species and oxidative

109

cellular damage [129]. HIF proteins are major regulators of this cellular response [130];

they are tightly controlled, and their translocation and stabilization are dependent on

oxygen gradients [130, 131], thus making them a general marker of hypoxia. Both low

oxygen and OxySite treated islets demonstrated activation of HIF-1α and its dimer

ARNT (also known as HIF-1β). Activation of the HIF-1 heterodimer in the OxySite

culture may read contradictory; however, activation of HIF-1 under normoxia (i.e. not

due to oxygen insufficiency) as a result of glucose stimulated oxygen consumption has

been demonstrated in β-cells [131]. Further, the significant reduction in the glucose-

stimulated ATP generation of HIF-1 α null mice and subsequent impaired insulin

secretion hints at the role of HIF-1 complex in β-cell metabolism and insulin

responsiveness [38, 130]. Thus, HIF-1 complex in β-cell cultures may be modestly

activated during insulin secretion, but elevated accumulation can lead to negative

pathways associated with cell death. In contrast, evidence of downregulation of HIF-3 α and its target pro-apoptotic pathways was observed in our OxySite cultures compared to both low oxygen and standard oxygen controls. These results, together with the subsequent preservation of metabolic activity and viability, indicate that OxySite can significantly mitigate key markers associated with hypoxia-induced islet loss, while not hindering β-cell homeostasis.

Reprogramming of cellular metabolism is a critical adaptive response to oxygen and nutrient deprivation in models of ischemic insults in diabetic patients [132]. HIF-1α plays a prominent role in glucose catabolism and cellular oxidative response in hypoxic settings [128] by promoting the expression of glucose transporters, glycolytic enzymes, and lactate dehydrogenase [103]. Glycolytic and enzyme markers that direct a shift

110

toward anaerobic glycolysis were identified in islets cultured at low oxygen conditions.

This shift also lead to dampened cellular respiration. Supplementation with OxySite,

however, demonstrated the ability of our oxygen generator to support aerobic

respiration and preserve OCR, even under chronic low oxygen conditions (24 h).

Preventing anaerobic glycolysis and preserving islet and beta cell respiration are likely

contributing factors to the observed improved graft outcomes for the OxySite treated

islets, as anaerobic pathways negatively impact β-cell function and insulin transcription,

while dampened OCR leads to poor transplant efficacy.

Early graft loss has also been linked to non-specific inflammation around the

graft, which is unrelated to specific host immune responses [133]. Previous studies

have shown that concomitant activation of HIF-1α and NFκB regulates activation of pro-

inflammatory genes under hypoxia [134]. Microarray data found OxySite to suppress hypoxia-induced activation of several potent inflammatory regulators and effectors, notably Infg, Il-17, and Il-6, which play prominent roles in the activation of beta cell’s

“extrinsic” apoptotic pathway [135, 136]. As such, downregulation of these factors can

lead to a significant positive impact on islet heath and clinical efficacy. Concurrently,

OxySite resulted in significant downregulation of iNos; a potent player in the generation

of oxidative stress, and cytokine induced β-cell death [137]. Suppression of iNos activation and expression would be beneficial in protecting the islet graft from inflammatory stimulus in the surrounding microenvironment. Of interest, iNos was not upregulated in the low oxygen control group after 24 h of culture. Activation of iNos under strict hypoxia has been found to be minimal, and to require complimentary cytokines, such as Il-1b, to elevate expression [138]. The observed downregulation of Il-

111

1b for both OxySite and control islets after 24 h of culture, thus supports iNos observations. Further, gene and protein expression of CCL2 was upregulated in our

OxySite co-culture system. While CCL2 expression has been associated with prominent cellular infiltrates in pancreatic islets [139], there is recent evidence that suggests a protective role via promotion of tolerogenic dendritic cell migration to the graft [140].

Moreover, CCL2 has been correlated with enhanced glucose responsiveness and in the regeneration of the peri-islet capsule post-isolation [141]. Thus, the role of CCL2 under our oxygenated cultures is an interesting aspect that requires further investigation.

The role of HIF-1α in the vascular response has been well documented in the literature, in particular its effect on angiogenic genes such as VEGF [142]. Limiting the ability of HIF-1α to activate these pathways via an oxygen generator might be seen as counterproductive. Nevertheless, our results found no benefit of severe hypoxia on

VEGF protein release, for the time points tested. While seemingly contrary to other published reports [143], the degree of hypoxia-induced damage observed in low oxygen controls in this study suggests that, in the face of severe oxygen deprivation, there is a shift towards mechanisms of conservation and reduced energy that hinders stable translation of pro-angiogenic pathways. This trend is supported by literature demonstrating a general suppression of mRNA transcription, including VEGF [144, 145], and a reduction of the angiogenic capacity of isolated islets [146] after chronic hypoxia.

Thus, the incorporation of OxySite to prevent islet hypoxia should not impair the efficiency or competency of the resulting vascular network.

Condensed in vitro assessments conducted using NHP islets demonstrated trends comparable to that observed for rat islets, with preservation of islet architecture,

112

viability, mitochondrial glucose oxidation, and angiogenic potential for the OxySite treated group. Interestingly, Bax levels were initially upregulated for OxySite and standard oxygen control groups (8 h), although not sustained. Recent evidence suggests that BCL-2 proteins may play dual roles in balancing stimulation of apoptosis, with islet metabolism and insulin secretion coupling [121]. In terms of cytokine activation, NHPs demonstrated elevated sensitivity to hypoxia, with a significant increase in the release of pro-inflammatory cytokines Il-6 and Il-8 in low oxygen controls. The capacity of OxySite to inhibit the expression of these cytokines, which contribute to delayed glucose responsiveness, induce direct β-cell toxicity, and lead to the recruitment and activation of host immune cells [147], should result in beneficial effects on islet engraftment and long-term acceptance.

The favorable islet health profile promoted by the culture of islets with OxySite resulted in a significant impact in overall graft efficacy, whereby OxySite treated islets outperformed islets cultured under standard oxygen tensions. Further, strong intra-islet vascularization observed for OxySite treated islets corroborates that healthy islets result in strong graft outcomes. Altogether, these results highlight the importance of oxygen in the preservation of islets, particularly when islet yields are unpredictable and have a strong correlation to overall clinical efficacy [148, 149].

In conclusion, while supporting oxygen in vitro is desirable, of further interest is increasing oxygenation of islet grafts [81]. While many efforts have concentrated on identifying appropriate transplant sites, optimizing the graft microenvironment, and enhancing the revascularization process [150], few approaches are focused on providing in situ oxygenation. Recent publications highlight the promise of oxygen

113

supplementation for improved graft viability [151, 152], although indefinite oxygen supplementation is required for these macroencapsulation approaches. Alternatively, oxygen supplementation during the early engraftment period, when islets are most vulnerable to detrimental low oxygen tensions, could provide the necessary boost to greatly improve not only early, but late stage, efficacy. The incorporation of temporary in situ oxygenation into biomaterial scaffolds engineered for housing islets within extrahepatic sites is the focus of future studies [153].

114

Table 4-1. List of oligonucleotide primers used in RT-PCR analysis Gene Organism Gene ID Amplicon Context Seq. (bp) Glut1 Rattus norvegicus Rn01417099_m1 73 Pgk1 Rattus norvegicus Rn01474008_gH 142 Ldha Rattus norvegicus Rn00820751_g1 85 Actinb Rattus norvegicus Rn00667869_m1 91 iNos Rattus norvegicus Rn00561646_m1 77 Ccl2 Rattus norvegicus Rn00580555_m1 95 Vegfa Rattus norvegicus Rn01511601_m1 69 18S Homo sapiens Hs99999901_s1 187 Bax Macaca mulatta Rh01016552_m1 92 Bak Macaca mulatta Hs00940249_m1 64 Glut1 Homo sapiens Hs00892681_m1 76 Pgk1 Macaca mulatta Rh02621707_g1 107 Ldha Macaca mulatta Rh02822039_gH 173 Vegfa Macaca mulatta Rh02621759_m1 142

115

Figure 4-1. Immunohistochemistry of whole islets after exposure to low oxygen conditions for 24 h. Hypoxia of Po2 <10 Torr is indicated by the oxygen sensitive hypoxyprobe dye (green). Spatial distribution of Insulin and glucagon is represented in yellow and magenta respectively. Scale bar = 50 µm.

116

Figure 4-2. OxySite mitigates activation of second stage hypoxic and apoptotic markers, leading to preservation of metabolic activity and viability. A) Heatmap summary of differential gene expression of early (above dashed line) and second stage (below dashed line) stage hypoxic and apoptotic (below solid line) markers from islets following low oxygen (0.01 mM) culture without (c) or with OxySite (o) after 8 and 24 h. Results are expressed as fold regulation over standard oxygen (0.2 mM) controls. B) Metabolic activity (measured via MTT) for islets following 4, 8, and 24 h culture under low oxygen (0.01 mM) culture without (white bars) or with OxySite (black bars), compared to standard oxygen (0.02 mM) controls (grey bars). C) Visualization of islet viability via live/dead staining (green = live; red = dead) after 4, 8, and 24 hr culture under 0.01 mM oxygen without (Control; left panel) or with Oxysite (middle panel), compared to 0.2mM Oxygen control cultures (right panel). Scale bar = 100 µm. * p < 0.05, ** p < 0.01, **** p< 0.0001.

117

Figure 4-3. Culture of pancreatic islets with OxySite enhances islet viability and reduces activation of pro-apoptotic markers in culture long term. A) NHP Islet viability of co-culture systems by calcein-AM and ethidium bromide live/dead stains. Co-culture of islets with OxySite under hypoxia resulted in increased viability and morphological intact islets at all culture time points comparable to normoxic conditions (0.2mM). In contrast increased presence of dead cells was observed by 8 h after hypoxic exposure (0.01mM). After 24 h of hypoxic exposure NHP islets were fragmented and only cellular debris remained. Live (green fluorescence) and dead (red fluorescence). Scale 100 um. B-C) qRT- PCR expression of BLC-2 pro-apoptotic proteins BAX and BAK in NHP islets cultured under continuous hypoxia for short (8 h) and long-term (24 h) with (black) and without an OxySite disk. For panel B-C * p < 0.05, ** p < 0.01, **** p<0.0001.

118

Figure 4-4. OxySite mitigates activation of anaerobic glycolysis and preserves glucose stimulated insulin release for rat pancreatic islets cultured under hypoxic conditions. A) Heatmap summary of differential gene expression of glycolysis (above dashed line: glycolytic pathway; below dashed line: membrane glucose transporters) from islets following low oxygen (0.01 mM) culture without (c) or with OxySite (o) after 8 and 24 h. Results are expressed as fold regulation over standard oxygen (0.2 mM) controls. B-D) qRT-PCR of selected glycolytic genes Glut1, Pgk1, and Ldh quantify the extent of anaerobic glycolytic gene activation in low oxygen control (white bars) islets after 8 and 24 h culture, compared to 0.01 mM oxygen OxySite treated (black bars) and 0.02 mM oxygen control (grey bars) islets. E) Protein levels of LDH following hypoxic exposure for all culture groups. F-G) Dynamic glucose perifusion and resulting insulin release from islets cultured under low oxygen without (orange) or with OxySite (blue) after acute (8 h; F) and chronic (24 h; G) culture. Grey region = high glucose (11 mM); Striped region = KCl H) Resulting area under the curve (AUC) during high glucose stimulation for low oxygen cultures without (white bar) or with OxySite (black bar), compared to islet cultured under standard oxygen and density conditions (free islets; grey bar). Inset: KCL Stimulation Index (SI) for low oxygen cultures without (white bar) or with OxySite (black bar). ** p < 0.01, *** p< 0.0005, **** p< 0.0001.

119

Figure 4-5. Co-culture of islets with OxySite maintains islets OCR under hypoxia. A) ATP concentration was reduced in islets exposed to chronic hypoxia (24 h) with a blank disk (white bars) compared to OxySite treated islets (black bars). B) Representative OCR from islets cultured under hypoxia with (black bar) or without an OxySite disk (white bar), as well as normoxic control (grey bars) and islets cultured at decreased densities (checkered bar). Culture of islets under high loading densities at both normoxia and hypoxia results in a decrease in OCR. Introduction of OxySite into culture system can restore OCR to levels comparable to islets culture at low densities. C) Representative results of respirometry assessments of islets exposed to a glucose challenge. OCR rate was enhanced in islets cultured under hypoxia with OxySite (orange line) compared to hypoxic (blue line) and normoxic (grey line) controls. OCR for OxySite group was comparable to islets cultured at low densities (black line). D) Change in OCR after glucose stimulation was calculated relative to basal for all samples assessed (Data represented as mean +/- SEM). For panel A and D *p<0.05.

120

Figure 4-6. OxySite reduces activation of anaerobic glycolysis in nonhuman pancreatic islet cultures. A-C) mRNA expression of glycolytic enzymes GLUT1, PGK1 and LDH of NHP islets cultured in hypoxic culture (0.01mM) with (black bars) or without (white bars) OxySite as well as a normoxic (0.2mM) control (grey bars). Co-culture of NHP islets with OxySite resulted in decrease mRNA expression of all glycolytic markers at all-time points. D) LDH release was used as a marker of cytotoxicity following hypoxic (0.01mM) exposure with (black bars) or without (white bars) OxySite. Elevation of LDH activation can be observed as early as 8 h after hypoxic exposure. Nevertheless, co-culture of NHP islets with OxySite hinders activation to levels even lower than normal culture conditions (grey bars). For panel A-D * p < 0.05, ** p < 0.01, *** p< 0.0005, **** p< 0.0001.

121

Figure 4-7. OxySite alters expression of inflammatory regulators and effectors of rat pancreatic islets under hypoxic conditions. A) Heatmap summary of differential gene expression of inflammatory regulators (above dashed line) and effectors (below dashed line) from islets following low oxygen (0.01 mM) culture without (c) or with OxySite (o) after 8 and 24 h. Results are expressed as fold regulation over standard oxygen (0.2 mM) controls.. B-C) qRT-PCR of selected inflammatory genes (iNos and Ccl2) quantify the extent of cytokine gene expression activation in low oxygen (0.01 mM) control (white bars) islets after 8 and 24 h culture, compared to 0.01 mM oxygen OxySite treated (black bars) and 0.02 mM oxygen control (grey bars) islets. D-G) Protein levels of cytokines MIP1a, CCL2, IL-6, and IL-8 in the culture supernatant after 8 and 24 h incubation under low oxygen conditions without (white bars) or with OxySite (black bar), compared to 0.2 mM control cultures (grey bars). * p < 0.05, ** p < 0.01, *** p< 0.0005, **** p< 0.0001.

122

Figure 4-8. Oxysite prevents hypoxia mediated activation of pro-inflammatory cytokines in non human primate cultures. NHP Islet were cultured for short (8 h) and long term (24 h) under continuous hypoxia with (black) or without (white) OxySite. A-C) Cytokine expression levels were assessed for IL-8, IL-6, and CCL-2/MCP-1 supernatant proteins. Exposure of NHP islets to hypoxia resulted in elevation of pro-inflammatory cytokines IL-8 and IL-6, which was stalled by the introduction of OxySite to the culture system. Interestingly, up regulation of CCL2 chemokine was also seen at the protein level. Assays are from multiplex protein measurements. For panels A-C ** p < 0.01, *** p< 0.0005.

123

Figure 4-9. Impact of OxySite on angiogenic pathways and VEGF protein release for low oxygen cultures of rat islets. A) Heatmap summary of differential gene expression of angiogenic gene regulators (above dashed line) and downstream factors (below dashed line) from islets following low oxygen (0.01 mM) culture without (c) or with OxySite (o) after 8 and 24 h. Results are expressed as fold regulation over standard oxygen (0.2 mM) controls. B) qRT-PCR of Vegfa quantifies the extent of gene expression activation in low oxygen (0.01 mM) control (white bars) islets after 8 and 24 h culture, compared to 0.01 mM oxygen OxySite treated (black bars) and 0.02 mM oxygen control (grey bars) islets. C) Protein levels of VEGFa in culture media after 8 and 24 h culture under low oxygen conditions without (white bars) or with OxySite (black bar), compared to 0.2 mM control cultures (grey bars). * p < 0.05, ** p < 0.01, *** p< 0.0005, **** p< 0.0001.

124

Figure 4-10. OxySite does not hinder angiogenic potential of non-human pancreatic islet cultures. A-B) Angiogenic activity of NHP Islets cultured under continuous hypoxia with (black) or without (white) OxySite was evaluated via mRNA expression profiles and supernatant protein concentration. While co-culture of NHP islets with an OxySite disk seemed to reduce the transcription of VEGFa at the mRNA level, it did not hinder the translation of the protein as seen by increased protein concentration in the culture supernatant. For panel A-B ** p < 0.01, *** p < 0.0005, **** p< 0.0001.

125

Figure 4-11. Co-culture of rat islets with OxySite resulted in enhanced functional outcomes when transplanted into diabetic syngeneic recipients. A) Average nonfasting blood glucose levels of rats receiving islets cultured under low oxygen (0.01 mM) conditions with OxySite for 24 h prior to transplant (black squares; n = 8). Control islets were cultured under standard oxygen (0.2 mM) for 24 h prior to transplant (grey circles; n = 8) or with a blank PDMS (grey line; n=8). B) Tracking of glucose clearance following intravenous glucose tolerance test (IVTT) performed in a subset of animals at 30 d post-transplant. Metabolic clearance was compared to a nondiabetic rat. (OxySite n = 5; Control, n = 3; Nondiabetic control, n = 1). D) Immunohistochemistry evaluation of grafts explanted 40 days post-transplant, stained with anti- insulin (green), anti-smooth muscle actin (SMA, red), and nuclear staining (blue) (top row) or anti-insulin (green), anti-glucagon (pink), and nuclear staining (blue) (top row). Scale bar = 50 µm.

126

CHAPTER 5 MACROPORUS SCAFFOLDS WITH ENHANCED OXYGENATION FOR EXTRAHEPATIC ISLET TRANSPLANTATION

Background

The potential of islet transplantation as a cellular based treatment for Type 1

Diabetes has been elucidated by the nearly complete halt of exogenous insulin (for at least one year) and the excellent metabolic control observed in recent islet transplant patients [154]. Long-term insulin independence and graft stability, however, has been

achieved in only a marginal subset of islet transplant recipients. The main hindrances in

successful translation of this therapy partially stems from the chosen transplantation

site. Infusion of islets directly into the vasculature of the hepatic portal vein, has proven

to lead to an instant inflammatory blood mediated reaction (IBMIR) and the subsequent

activation of a cascade of contributing factors that lead to early graft failure [155, 156].

Thus, islet transplantation would greatly benefit from the development of a more

supportive transplant location.

Research on the design of an alternative transplant site has focused on the

subcutaneous space [157, 158], intravascular [159] or intramuscular [160, 161]

compartment, as well as the intraperitoneal space [162, 163]; however, limited success

has been observed due to issues with the invasiveness of the procedure, detrimental

oxygen gradients, mechanical stress, clumping of the islet clusters, and the lack of

proximity to the vascular network [164]. In recent years, the omental pouch has

emerged as a promising extrahepatic site for islet transplantation. It has the advantage

of increased transplant volumes, portal drainage, and improved proximity to a rich

vascular network [150, 164]. Our laboratory and others have established the potential

127

of the omental pouch as a transplant site in both rat and non-human primate models of diabetes [10, 113, 116, 165].

Our approach for islet transplantation into this novel site incorporated the use of a non-degradable macroporous scaffold that favors host infiltration and graft revascularization, which can enhance graft insulin responsiveness to glucose challenges [113]. Current devices prototypes range in size from 10 mm for rats, to 30 mm for nonhuman primates, to 60 and 90 mm for humans. This scale is dictated by the maximum islet loading density within the devices, which is largely regulated by oxygen availability. As these are initially avascular devices, this loading density is targeted to be within the range of 2 to 6 vol/vol. As such, for a smaller human patients (< 60 kg), a

10,000 IEQ/kg islet transplant would require one 90 mm scaffold and one 60 mm scaffold to maintain the targeted islet loading density. While this is still a feasible approach, increasing the packing density of these macroporous scaffolds would permit the implantation of smaller devices. Decreasing the size of the implant could also have the beneficial effect of decreasing the space requirement in the omentum, the invasiveness of the surgical procedure, and complications from retrieval, while heightening feasibility of future implants.

In order to increase the islet loading density within our macroporous scaffolds, the availability and transport of oxygen would need to be dramatically increased. While it has been demonstrated that the prevascularization of a transplant site can increase the survival of islets [8, 166], this enhanced oxygenation is restricted to only the periphery of the implantation site, limiting the scale of the implant. In order to significantly elevate the loading density of a device, in situ oxygenation is required.

128

Further, it would be most desirable to incorporate this in situ oxygenation at the location of the device where it is most needed, centrally.

It is the purpose of this chapter to introduce and test an optimized version of our macroporous scaffold that is based on a composite fabrication of our scaffold platform and the OxySite bioactive material introduced in Chapter 3. These scaffold plus OxySite composites are termed herein SS-OxySite, while comparable PDMS-only controls are termed SS-PDMS. Through these studies, we seek to demonstrate that delivering oxygen in situ to high loading density grafts can improve achievement of normoglycemia in vivo.

Methods

Materials

The poly-dimethyl-siloxane polymer was purchased from Nusil (MED 6215, medical grade). Human plasma fibronectin (FN) was purchased from invitrogen. All culture media was purchased from Mediatech. Insulin ELISA was purchased from

Mercodia (Winston Salem, NC). All other materials were purchased from sigma Aldrich unless otherwise specified.

Composite Scaffold-OxySite Fabrication

Macroporous PDMS scaffolds were fabricated by (Silicone Specialty Fabricators) in a cGMP facility to our explicit specifications. Briefly, scaffolds were fabricated using a solvent casting and particulate leaching technique (SCPL) with control pore size ranging from 250 µm to 425 µm in diameter as described previously [113]. OxySite was fabricated as described in Chapter 3. Briefly PDMS monomer was mixed with platinum catalyst, 4:1 v/v via a thinky mixer. Mixture was poured into a mold of 2.9 cm in diameter, and placed under vacuum for 1 h. Samples were then cured overnight at 40

129

°C. After curing, the top and bottom scaffolds were attached to both top and bottom

surfaces of OxySite material using a mixture of the PDMS/curing agent at 4:1 v/v ratio.

Control implants were made using PDMS-only disks (no calcium peroxide). The

composite scaffolds were then cured again overnight at 40 °C. After curing, scaffolds

were punched out at different sizes (10 to 6 mm in diameter, depending on the study)

and sterilized 30 min in ethanol. After sterilization scaffolds were rinsed five times with

PBS (Invitrogen).To facilitate islet loading within the hydrophobic scaffold, human

plasma fibronectin (FN) was absorbed to the PDMS by incubating it at 250 µg/mL FN

o solution in water and placed in an incubator (37 C, 0.20 mM O2, and 0.05 mM CO2) for

24 h. Resulting composite grafts were termed SS-OxySite (for scaffolds containing a single OxySite disk) or SS-PDMS (for scaffolds containing a single PDMS-only control disk).

Non-invasive Oxygen Measurements

Oxygen release profiles of the SS-OxySite was obtained via a non-invasive, sealed titanium chamber system (Instech Laboratories, PA). Partial oxygen tensions in the solution medium were measured using non-invasive oxygen sensors (PreSens,

Germany). Sensor spots consists of a ruthenium coated via a sol-gel polymer. Optical signal is sent to the sensors via optical fiber cables from a custom transmitter box

(PreSens, Germany). Ruthenium in spot sensors quenches oxygen in the medium shifting fluorescent signal which can then be correlated to oxygen concentration in the solution. Oxygen concentrations were recorded by Presens open software in units of mmHg (torr).

130

Scanning Electron Microscopy

Scanning electron microscopy (SEM) (JEOL, JSM-5600LV, 29 Pa, 20 kV,

Peabody, MA, USA) was employed to visualize the resulting composite scaffolds.

Images were representative from multiple samples.

Islet Isolation and Culture

All animal procedures were performed under protocols approved by the

University of Florida IACUC and in accordance with National Institutes of Health

guidelines. Rat islet isolations were performed following methods described previously

[22], using male Lewis rats (Harlan, Indianapolis, IN). After isolation islets were cultured

2 o overnight under standard culture conditions (500 IEQ/cm , 37 C, 0.20 mM O2, and 0.05 mM CO2) in supplemented CMRL 1066 medium prior to use.

Islet Transplant and Graft Assessment

For transplants, diabetes was induced in Lewis female and male rat recipients

(Envigo) via 2 intraperitoneal injections of 60mg/kg streptozotocin and were only used

as recipients after 3 consecutive readings of non-fasting blood glucose levels > 350

mg/dL, as previously described [113]. The first transplant trial sought to compare the

impact of the OxySite on engraftment. As such, SS-OxySite and SS-PDMS implants (10

mm diameter) were compared to the standard PDMS scaffold (single scaffold disk with

no solid PDMS disk or OxySite) used in previous studies, at the same published loading

density, scale, and recipient weight (female rates 160-175 g) [113]. Devices were placed onto the spread omentum of anesthetized rats, islets (isolated 24 h earlier) were then collected into a Hamilton syringe and distributed onto only the top side of the scaffold. The islets were sealed within the scaffold using a fibrin hydrogel, similar to that previously described [114]. Briefly, 15 µL of a thrombin (4 U/mL), 9 mM CaCl2, 150

131

mM NaCl, (on ice) was quickly mixed with 15 µL of fibrinogen (8 mg/mL), aprotinin (85 ug/mL) and 20 mM HEPES solution at RT, and placed on top of scaffold. The scaffold was then flipped, the omentum was folded over the graft, and the seam was sealed using the same fibrin hydrogel. Note, as explained in Chapter 4, the fibrin gel was used to seal the omentum in lieu of sutures for ease in closure. The OP was placed back in the peritoneal cavity, the incision was sutured, and the skin was closed using surgical staples.

Additional efficacy transplants seeking to examine the impact of elevated loading densities on graft outcomes were conducted using larger male diabetic Lewis rats (280 -

300 g). In this study, the size of the SS-OxySite and SS-PDMS composite implants were decreased to 6 mm in diameter to further exacerbate loading densities. The transplant was then conducted in a manner identical to that described above, except for an elevated islet loading of 3,000 IEQ.

The blood glucose (BG) and body weight (BW) of transplant recipients were monitored daily until normoglycemia was achieved. Normoglycemia was classified as stable nonfasting glycemic levels < 200 mg/dL for at least two consecutive BG readings

[115]. Metabolic assessment was conducted via an intravenous glucose tolerance test

(IVGTT) 30 d post-transplant on selected animals from both groups (OxySite n =5;

Control n =4), as previously described [114]. Briefly, after overnight fasting, rats received a bolus of glucose (IV; 2 µ 50% dextrose / g BW) and BG levels were monitored until reading < 200 mg/dL or after 100 mins. To ensure observed normoglycemia was due to the islet graft and not residual native pancreas function, the omentum wrapped scaffold was removed in a survival surgery in selected

132

normoglycemic grafts and the animal BG was monitored until hyperglycemia was restored (typically < 3 d).

Gene expression profiling of grafts explanted at acute and chronic time point post-transplantation was performed using a custom made PCR array (for a table of the primers used refer to Table 5-1). Total RNA was extracted (RNase easy kit, Qiagen) and reverse transcribed (RT2 First Strand Kit, Qiagen), according to the manufacturer's instructions. The resulting cDNA was added to 96-well-plate PCR arrays for qPCR (RT2 profiler PCR arrays custom, Qiagen) assessment using a QuantiStudio 6 flex real-time

PCR system (Life Technologies, California, USA). Relative expression of analyzed genes was calculated by the ΔΔCt method. PCR array data collected from SS-OxySite implants were represented in heat maps as fold change from SS-PDMS grafts.

Histological Assessment

Explanted OPs were fixed in 10% formalin buffer and embedded in paraffin blocks and 10 μm sections were stained with hematoxylin and eosin (H&E) or Masson's trichrome (Richard Allan Scientific) for global graft/host interaction visualization.

Immunohistochemistry was performed for insulin (Dako A0564, 1:100), SMA (Abcam ab5694, 1:50), glucagon (Abcam ab10988, 1:50), and DAPI (Invitrogen D1306, 1:500), pimonidazole (hypoxryprobe, 1:50) as described elsewhere [116]. For lectin staining a tomato lectin (vectorlabs, DL1174) was injected IV at 1 µl/kg concentration. After 30 min animals were perfused with saline and the graft extracted and fixed in 10% formalin.

Post processing grafts were stained for insulin and DAPI as described above. IHC images were collected using a Zeiss LSM 510 confocal microscope.

133

Statistical Analysis

Two independent experiments (e.g. three separate islet isolations) were performed for efficacy transplants with control and OxySite groups compared for each transplant. Nonfasting blood glucose levels are represented as the mean ± SD with the number of animals for each group represented in the figure legend. Survival curves demonstrating % of normoglycemic animals remaining were implemented to evaluate the difference among the experimental groups.

Results

The aim of this chapter was to develop a composite scaffold + OxySite implant that could be used to examine the effects of in situ oxygenation on the engraftment and long term function of islets in a diabetic animal model. For the scaffold platform, we utilized a previously described macroporous silicone scaffold. This original prototype scaffold was optimized to minimize detrimental oxygen gradients by maximizing islet distribution within the device and has been shown to be effective in rodent and nonhuman primate transplant models [113]. Initial prototypes involved the incorporation of the reactive agent in the PDMS solution of the macroporous scaffold prior to fabrication. The idea of this prototype was to have a single system that would serve as both a 3D housing and an oxygen reservoir. While fabrication was plausible, the incorporation of a reactive agent within this highly porous scaffold lead to highly reactive system, resulting in a cytotoxic environment. Additional platforms involved the fabrication of a donut shaped macroporous scaffold that would allow for the placement and retrieval of a fully cured OxySite disks. Fabrication proved to be challenging and resulting scaffolds lacked the stability of the initial prototypes. Based on these

134

unsuccessful approaches, it was concluded that the scaffold and OxySite should be combined after the individual fabrication of each platform.

The final composite SS-OxySite scaffold implemented was fabricated by incorporating an OxySite disk into the device by adhering it to the base of a PDMS porous scaffold. In addition, a thinner scaffold was added to the bottom of the OxySite disk, as it was found necessary to mitigate the foreign body response to the solid PDMS disk interface (which was observed for both OxySite and pure PDMS solid disks, as described in APPENDIX B).

In order to examine the impact of in situ oxygenation on reducing the overall size of the implant needed, the volume of the macroporous scaffold was reduced by decreasing the diameter from 10 mm (original prototype) to 6 mm (Figure 5-1 panel A), a 64% reduction of volume. The kinetics of oxygen generation from the resulting SS-

OxySite composite was evaluated using a sealed, temperature controlled chamber system. Similar early stage (0-3 d post initial hydration) kinetics to a single OxySite disk of the same dimensions were observed (significant correlation with p<0.0001, see

Chapter 3), although later release (4-6 d post initial hydration) kinetics exhibited a decrease from 52 ± 3.8 to 23 ± 4.7 mmHg in the SS-OxySite, when compared to the single OxySite disk (p<0.001). Results demonstrate that the addition of scaffold and the adhesive PDMS did not significantly hinder the reactivity of the OxySite material during the first three days post fabrication. While, a decrease in oxygen generation was observed at later time points, oxygen production was still robust and within physiological levels.

135

Scanning electron microscopy was used to examine the impact of the addition of

the OxySite disk on the porosity of the macroporous scaffold. As shown, no changes in

the overall porosity was observed for the top islet-loaded scaffold, even at the scaffold-

OxySite interface (Figure 5-2). For the bottom layer (which served only as a buffer from

the host and not for housing islets), some compression was observed due to the

fabrication technique; however, this did not impact the function of this layer in mitigating

foreign body responses.

As the impact of the OxySite material on islet engraftment has not been previously investigated, initial experiments sought to compare these SS-OxySite

composites with our original single macroporous scaffolds, using standard loading

densities (i.e. not elevated) and comparable scales (i.e. 10 mm in diameter). Diabetic

female rats received 1800 IEQ (or 1,100 IEQ/cm2 loading density) loaded within a

traditional single macroporous scaffold (n = 1), a SS-OxySite (n = 2), or a SS-PDMS (n

= 1). All implants exhibited euglycemia within 48 h post-transplantation and throughout

the duration of the graft (Figure 5-3 panel A). Upon graft removal (36 d post-transplant), animals quickly reverted to a hyperglycemic state, demonstrating that diabetes reversal was due to the transplanted graft and not due to residual native beta cells in the pancreas. No variance in body weight was observed among all animals transplanted

(Figure 5-3 panel B). Moreover, histological assessment via H&E staining revealed robust islets and vascular formation throughout the graft. The scaffold appeared well integrated with the surrounding tissue and no evidence of a fibrotic capsule formation was observed for either side of the composite scaffold. Immunohistochemical staining of insulin and α-SMA confirm the aforementioned results (Figure 5-3 panel C).

136

The next series of transplants focused on validating the potential of supplemental in situ oxygenation to improve islet transplant efficacy. In order to highlight this impact, an ideal balance of islet transplant number (i.e. transplantation of a minimum number of islets) and loading density needed to be achieved, whereby the islets transplanted within the composite scaffold containing the OxySite disk would be fully supported and functional, while islets within the control composite scaffolds containing the blank PDMS disk would experience hypoxia-induced dysfunction at a level that would result in graft inefficacy. As noted above, the diameter of the scaffold was reduced significantly from

10 to 6 mm to elevate the loading density. This reduction of scale resulted in a 2.36-fold increase in loading density (IEQ/ cm2) from 1,100 IEQ/cm2 to 2,600 IEQ/cm2 when transplanting 1800 IEQ into a female rat recipient. This elevated density, however, was likely not elevated enough to fully highlight the potential of the OxySite, and still below loading densities used for other oxygen supplementation approaches [41, 167]. While a further reduction in the device diameter would elevate the loading density, decreasing the scale of the OxySite further would impair the duration of oxygen generation. In order to increase the loading density without altering the composite device geometry, male rats were employed. While male rats are not only inherently larger than their female counterparts, their juvenile age results in an exponential-like increase in growth during the transplant period. As such, transplantation into male rats would not only increase the loading density, but also permit examination of the stability of the system as the islet mass is pushed to maintain stable blood glucose during the substantial weight gain of the recipient . Maintaining the same dosage of 10,000 IEQ/kg, the islet mass required to treat these male recipients was 3,000 IEQ, resulting in an increase in loading density

137

from 1,100 to 3,200 IEQ/cm2 or a 2.9-fold increase. Transplantation of composite SS-

OxySite scaffolds into the omentum of diabetic male recipients lead to 100% reversal (5

out of 5 animals) (Figure 5-4). In contrast, only 1 out of 4 recipients of SS-PDMS control

composites reverted to a normoglycemic state (25%). During the ~20% weight gain experienced by the functional recipients of SS-OxySite implants, 3 of the 5 grafts began

to destabilize as the functional mass decreased from 10,0000 IEQ/kg to 8,700 IEQ/kg;

however, the remaining 2 grafts were able to maintain stability during this significant and

rapid weight gain, highlighting the capacity of these OxySite supplemented grafts to

support robust islet function. Moreover, no significant differences in body weight was

observed up to 20 d post-transplant between animals receiving OxySite composite graft,

when compared to composite scaffolds transplanted with PDMS-only disks. Gene

expression of graft post-explantation demonstrated downregulation of hypoxia

regulators Hif3a and Nfkb2 in SS-OxySite compared to SS-Controls (Figure 5-5).

Downregulation of apoptotic marker Bak, angiogenic factors Gpx1, and Epx, as well as immune regulator Tgfβ, were also observed. Contrariwise, upregulation of the immunoglobulin superfamily protein Icam1 was detected in OxySite composites compared to controls, perhaps suggesting an increase in the recruitment of endothelial cells to the graft. While very preliminary, these results hint at the ability of our OxySite disk to mitigate activation of detrimental factors. A more in-depth characterization of these profiles should be performed to validate the herein observed results. Lastly, qualitative analysis of islets and competent vascular networks, via insulin and lectin injections, respectively, exhibited more robust, highly vascularized islets in the OxySite containing scaffold compared to the blank PDMS controls (Figure 5-6).

138

While the results for the efficacy transplant were overall very encouraging, a

higher statistical power must be achieved to strengthen the conclusions herein

presented. A repeat of this challenging transplant was performed to increase the sample

size of both treatment groups; however, unforeseen complications were encountered

while rendering the animals diabetic (i.e. streptozotocin toxicity was observed in animals

requiring euthanization of a significant number of recipients); albeit, transplants were

performed on remaining animals. Results from this transplant showed minimal efficacy

for both groups, with euglycemia achieved in only 1 out of 5 and 0 out of 4 recipients of

SS-OxySite and SS-PDMS implants, respectively, although another recipient of an SS-

OxySite implant became euglycemic after 13 d post-transplant. Streptozotocin toxicity of these animals could have imparted a negative impact on device engraftment, which may explain the divergent results observed in this set of transplants when compared to results from the two other transplants herein summarized. In the future, a titration of the streptozotocin dosage will be performed to optimize the diabetic induction protocol for male rats of this size. Once this is completed, a repeat of the experiment will be performed.

Conclusions

The success of clinical islet transplantation has been limited by the substantial graft loss observed after islet infusion into the hepatic portal vein [168]. This has led to the exploration of different extrahepatic sites as alternative locations for islet placement

[150, 164, 169]. The kidney capsule has been the gold standard in rodents. Placing a marginal mass of islets under the subrenal capsule has been shown to correct hyperglycemia [170]. Nevertheless, scalability to a human model remains irrelevant due to the invasiveness of the procedure and the limited space for the required transplant

139

mass [169]. Likewise, the subcutaneous space has long been considered an attractive site due to the minimal invasiveness associated with the transplantation procedure.

Recently, some success in deviceless prevascularization under the skin for islet transplantation has been reported in small animal models [166]. Translation of this into larger size animals remains in question, as clinical relevance of the subcutaneous space has been hindered by graft loss due to poor revascularization and the associated mechanical stress of the location. Further, graft protection from the immune response has been sought by transplanting islets into immunoprivileged sites such as the testes

[171], anterior chamber of the eye [15], and the thymus [172]; and while feasibility has been demonstrated in animal models, clinical application is challenged by the limited space of these site, as well as their delicate nature. Recently, the omentum has been explored as a more suitable extrahepatic environment due to its enhanced vascularization, portal drainage, ease of accessibility, and capability to hold large transplant volumes [150, 169]. Moreover, the superiority of the omental pouch as an extrahepatic transplant site for islet engraftment has been elucidated by us [25, 116,

153] and others [126, 173], and more recently in a human clinical trial (P.I. Rodolfo

Alejandro, clinicaltrials.gov: NCT02213003).

Transplantation of islets into the omental pouch has been achieved by housing islets in biological or synthetic materials. Grafts consisting of islets resuspended in a fibrin plug and placed in the omentum have shown success in animal models [165], and are currently under clinical trial; however, the uncontrolled placement of islets, leading to variability, and the degradation of the fibrin can lead to islet clumping. Moreover, this approach prevents localization of the graft and thus limits the retrievability of the

140

implant. Our group has explored transplantation into the omentum by loading islets into a macroporous PDMS scaffold. Islets loaded into these non-adsorbable devices have demonstrated successful diabetes reversal and sustained long-term normoglycemia.

Moreover, grafts transplanted within these devices exhibited robust host integration and capillary formation upon explantation [153]. Herein, we set to combine our scaffold platform with our in situ oxygen generating material, OxySite, introduced in Chapter 3.

Initial efforts were concentrated on combining these two approaches in a manner that would not hinder the qualities of each system individually when merged (i.e. porosity of macroporous scaffold, oxygen kinetics from OxySite disk). The final prototype presented herein for our SS-OxySite composite involved the fusion of a scaffold to our fully cured OxySite disk. Fabricating our composite material in this manner resulted in stable constructs, with no visual differences in porosity. Surprisingly, biocompatibility, at acute and chronic time points, of composite devices transplanted into the subcutaneous space of control animals demonstrated unforeseen foreign body responses. Moreover, when this platform was implemented in a diabetic model, grafts failed to restore normoglycemia at loading densities proven to work in our original platform. Histological assessment of the graft demonstrated a robust foreign body response that seem to be directed to the solid PDMS disk, which was irrespective of the presence of the reactive agent (i.e. the response was observed when implanting either a blank control PDMS disk or OxySite disk). These results are in line with previous publications demonstrating the effect of material surface topography on biocompatibility and host responses [174].

141

To avoid these detrimental responses, a material was needed to camouflage the bottom of the solid PDMS disk from the host. Given the extensive biocompatibility data on the macroporous platform collected in our laboratory [116, 153] it was decided that a

PDMS scaffold should be implemented to help mitigate these unfavorable responses.

As demonstrated herein, the addition of the OxySite to the scaffolds imparted no significant impact on the scaffold porosity or on the oxygen kinetics of the OxySite material. Implementation of these devices in a baseline diabetic model demonstrated no detrimental effects from the incorporation of a solid disk into our system. Significantly, when these platforms were implemented in a model that would enhance the detrimental oxygen gradients of the grafts, SS-OxySite devices outperformed comparable implants containing a control PDMS disk only, as observed by a significant improvement in engraftment efficacy for SS-OxySite treated animals, when compared to controls.

Histological examination of explanted grafts demonstrated robust islets and vascularization for SS-OxySite implants, further demonstrating that in situ oxygenation did not impair device vascularization. In the future, these experiments will be repeated to increase the statistical power of our observations.

Overall, studies herein demonstrated our ability to fabricate composite materials, involving two platforms developed in our laboratory, for extrahepatic islet transplantation. Data post fabrication demonstrated our ability to preserve material properties (i.e. porosity and oxygen generation). Further, we revealed the capacity of these novel platform to maintain graft viability early post transplantation. Thus, these devices can prove highly beneficial as a bridge to revascularization and in the delicate period before the formation of a fully functional vascular network.

142

Table 5-1. List of oligonucleotide primers used for gene profiling Gene Organism Gene ID Epx Rattus norvegicus 303414 Gpx1 Rattus norvegicus 030826 Icam1 Rattus norvegicus 25464 Hif3a Rattus norvegicus 64345 Tgfβ Rattus norvegicus 25008 Nfkb2 Rattus norvegicus 309452 Bax Rattus norvegicus 24887 18SrRNA Rattus norvegicus PPR57734E

Figure 5-1. Optimization of macroporous scaffold as a platform for high loading density islet transplants. A) Picture depicting dimensions of original scaffold as well as optimized composite OxySite-scaffold. B) Oxygen release kinetics from OxySite-scaffold composite measured in a sealed chamber via a non-invasive oxygen sensors. Inset represents top and side view of bioactive composite material.

143

Figure 5-2. Scanning elecetron micrographs of composite scaffold with OxySite disk. Left panel represents top surface of the scaffold. Middle panel is the bottom layer macroporous scaffold. Right panel represents cross section of composite material. No effects in porosity were visible after incorporation of OxySite disk to scaffolds.

144

Figure 5-3. Transplantation of original size macroporous scaffold in a syngeneic lewis rat model. A) Average nonfasting blood glucose of recipients following transplantation of 2500 IEQ into the omental pouch either control scaffold (gray line; n=1), within a composite PDMS scaffold (black line open circles; n=1), or within composite OxySite-scaffold (black line squares; n=2). Red line points to graft removal. B) Average body weight of recipients following transplantation of both control and composite scaffolds. Legend is similar to panel A. C) Representative histopathological images of hematoxylin/eosin- stained scaffolds explanted from the omental pouch, top panel represents control-PDMS scaffold while bottom panel is OxySite-Scaffold composite. Dashed lines encircle islets. D) Immunohistochemistry evaluation of grafts explanted 30 d post-transplant, stained with anti-insulin (green), anti-smooth muscle actin (SMA, red), and nuclear staining (blue). Sacle bar 50 µm.

145

Figure 5-4. Transplantation of optimized macroporous scaffold in a syngeneic Lewis rat model. A) Photograph of composite OxySite-scaffold used for implantation (6 mm in diameter, 5 mm in thickness). B) Survival curve depicting percent normoglycemic animals after composite OxySite-scaffold transplant. Black solid lines represents animals transplanted with a composite scaffold with a blank PDMS disk (n=4), dash red lines represents OxySite-scaffold receiving animals (n=5). C) Average body weight of recipients following transplantation of composite scaffolds loaded with 3000 IEQ on the omental pouch. Black lines represents control material, red line represents composite OxySite- scaffold transplanted animals.

146

Figure 5-5. mRNA expression from PCR array of SS-OxySite scaffolds transplanted in the OP pouch of diabetic male rats for 30 d. Data is expressed as fold regulation over SS-Ctrl-PDMS Scaffolds. Boxed numbers represent actual fold regulation of each sample.

147

Figure 5-6. Immunohistochemistry evaluation of grafts explanted 30 days post- transplant, stained with anti-insulin (green), tomato lectin (red), and nuclear staining (blue). Top row represent SS-OxySite scaffolds, bottom row SS-Ctrl- PDMS. Scale bar = 20 µm.

148

CHAPTER 6 IN SITU OXYGENATION OF IMMUNOISOLATING DEVICES

Background

Clinical Islet Transplantation (CIT), the infusion of allogeneic islets into the intrahepatic site, has emerged as a promising therapeutic option for patients with T1DM, providing a physiological metabolic control superior to exogenous insulin [129, 175-

177]. Successful islet transplants, in a cohort of patients with challenging glucose

dysregulation, glycemic liability and severe hypoglycemia unawareness, has highlighted

the potential of this treatment to reduce the incidence of hypoglycemia, deliver a tighter

glycemic control, and overall provide a better quality of life [176]. Nevertheless, long

term success and translation of this therapy to a wider range of the diabetic population

has been hindered by: (1) poor revascularization, (2) ensuing hypoxia, (3) remaining

autoimmunity; and (4) allogeneic responses to the foreign graft [164].

Encapsulation of islets in semipermeable immunoisolating devices has been an

attractive approach to avert host immune destruction, and potentially mitigate the need

for immunosuppression [164]. A single device that is capable of protecting the

transplant from immune cell attack, while also providing a confined, retrievable implant,

could address two of the major challenges in islet transplantation: beta cell mass

sourcing and immunosuppression. The translation of macroencapsulation devices,

however, has been greatly hindered by the limited oxygen availability within the

implanted devices [178]. As opposed to open macroporous scaffolds, as those

presented in Chapter 5, immunoisolating devices avert host infiltration, hindering the

formation of a vascular network within the material. Thus, these devices must rely solely

on passive oxygen diffusion for nutrient delivery. Moreover, the mass transfer of oxygen

149

is impacted by the device size, as increasing radial distance increases the inadequate oxygen tensions within the core of the device [65].This becomes particularly challenging in grafts that contain pancreatic islets, as their function is highly sensitive to inadequate oxygen tensions, whereby modest decreases in oxygen can result in substantial (up to

50%) declines in insulin responsiveness [36].

Various approaches have been investigated to improve mass transfer of oxygen to immunoisolated devices containing pancreatic islets. These methods range from hastening the revascularization process surrounding the device [179], delivery of growth factors [180], incorporation of oxygen carriers [181], or in situ oxygen delivery either via external purging mechanisms [41], photosynthetic decomposition of algae [58], or the electrochemical decomposition of water [57]. Nevertheless, most methods still result in inadequate oxygen tensions or cumbersome approaches to sustain cell viability, particularly at the core of the device.

In this chapter, we sought to evaluate the ability of our in situ oxygen generator,

OxySite, to sustain islet and beta cell viability, while decreasing hypoxia mediated activation of detrimental factors within a macroencapsulation device. Early studies focused on engineering the appropriate scale and structure of the device to withstand the mechanical stress experienced in the implant environment (see APPENDIX F).

Following the design of an appropriate macroencapsulation construct, in vitro validation studies examined the benefits of in situ oxygenation on islet survival and function.

Furthermore, the in vivo efficacy of the system was examined in syngeneic, allogeneic, and xenograft transplant models.

150

Methods

Materials

The poly-dimethyl-siloxane polymer was purchased from Nusil (MED 6215, medical grade). Seaplaque Agarose low melting temperature was obtained from Lonza.

All other materials were purchased from sigma Aldrich unless otherwise specified.

OxySite Fabrication

OxySite was fabricated as described in Chapter 3. Briefly, 25% w/w CaO2 was added to PDMS in a 4:1 vol/vol PDMS monomer to platinum catalyst and mixed/defoamed via centrifugal mixer (Thinky Mixer, USA). The resulting PDMS/CaO2 mixture was then poured into a cylindrical mold (29 mm) to 1 mm thickness and cured overnight at 40 oC. Prior to use, smaller OxySite disks (5 mm) were punched out of the larger cured disk. Resulting OxySite disks were sterilized in ethanol for 30 min and rinsed 5 times in PBS (Invitrogen) prior to incorporation with macroencapsulation devices.

MIN6 Culture

MIN6 cells were cultured as monolayers in tissue culture treated T-flasks with the medium change every 2-3 days. Media consisted of Dulbecco’s Modified Eagle’s

Medium (DMEM) (invitrogen) with 5.5 mM glucose and supplemented with 10% fetal bovine serum, 1% penicillin-streptomycin, 1% L-glutamine, and 0.001 % (v/v) β- mercaptoethanol. Upon reaching confluency, MIN6 cells were trypsinized using 0.25%

Trypsin with EDTA (Invitrogen) and either plated at a lower density for propagation or used in construct fabrication (passage numbers 28-45 were used for transplants).

151

Islet Culture

All animal procedures were performed under protocols approved by the

University of Florida IACUC and in accordance with National Institutes of Health

guidelines. Rat islet isolations were performed using methods described previously [75],

using male Lewis rats (Harlan, Indianapolis, IN). Rat islets were cultured overnight

2 o under standard culture conditions (500 IEQ/cm , 37 C, 0.20 mM O2, and 0.05 mM CO2)

in supplemented CMRL 1066 medium prior to use.

Immunoisolating Device Fabrication

The dimensions of the OxySite-Agarose device used for in vitro assessments is

illustrated in Figure 6-1. Devices were fabricated in custom silicone molds, placing the

OxySite disk in the geometrical center (both radially and axially) of the mold using pins.

Lewis rat islets (1500 IEQ) were mixed within 458 µL of 2% w/v seaplaque agarose

warmed to 37ºC and subsequently added to each well, whereby the agarose filled the

void spaces of the mold, encasing the centrally located OxySite disk. Constructs were

then placed at 4ºC for 8 min to complete gelation. Control constructs were generated

using PDMS-only disks in lieu of the OxySite disk (termed Controls). Gel constructs

were removed from molds, placed in 24-well non tissue culture plate (Falcon), and

cultured in supplemented CMRL1066 media (2 mL) at 37oC, and standard incubator

conditions (0.20 mM O2, and 0.05 mM CO2) for 1 h. Constructs were then transferred to

a C-Chamber Hypoxia Chamber (Biospherix, New York) set to low oxygen (0.01 mM

(1%) O2 with 0.05 mM (5%) CO2) using controllers (Biospherix). Islet constructs were

then incubated for 24, 48 and/or 72 h.

For efficacy transplants, the loading density was scaled up for a rat model (4000

IEQ). Moreover, the scale of the OxySite construct was decreased to scale

152

appropriately to the mouse model (see Figure 6-2). Constructs were fabricated in an identical manner to that described above, at the desired loading density of MIN6 cells or islets. Following fabrication, constructs were cultured under standard incubator conditions for 1 h prior to transplantation.

In vitro Assessments

The metabolic activity of the constructs was assessed via MTT (Promega), as previously published [65]. Briefly, at time points designated, one-quarter piece of the construct was cut with a scalpel and incubated in 350 μL of complete CMRL media with

39 μL of MTT for 3 h under standard oxygen conditions. Stop solution (259 µL) was then added under agitation and incubated for 48 h to permit complete formazan crystals solubilization. Absorbance (570 nm) was then measured from triplicates samples using a SpectraMax M5 microplate reader (Molecular Devices).

Total RNA was extracted using a combination of DNA gel extraction kit (QIAquick

Gel Extraction Kit, QIagen) and a total RNA plant extraction kit (RNeasy Plant Min Kit,

Qiagen) following manufactures instructions. Primers used are listed in Table 6-1.

Relative gene expression was calculated using Taqman assays in a SetpOne cylce

PCR system (Life Technologies, California, USA). Gene expression was normalized against β-actin. The ΔΔCt method for relative quantification was used and results were expressed as fold regulation over control constructs containing a blank PDMS disk.

Transplant and Graft Assessment

Diabetes was induced in male B/6 mice via 1 intraperitoneal or intravenous injection of 200 mg/kg streptozotocin and were only used as recipients after 3 consecutive readings of non-fasting blood glucose levels > 350 mg/dL. For implantation, a 2 cm incision was made in the abdomen and the immunoisolating device was placed

153

in the peritoneal cavity of the diabetic mouse. The incision was sutured, and the skin

was closed using surgical staples. Blood glucose (BG) and body weight (BW) of

recipients were monitored in the morning daily. As for all transplants, normoglycemia

was delineated as stable nonfasting glycemic levels < 200 mg/dL for at least two

consecutive BG readings.

Histological Assessment

Explanted OPs were fixed in 10% formalin buffer and embedded in paraffin

blocks and 5 μm sections were stained with hematoxylin and eosin (H&E) or Masson's

trichrome (Richard Allan Scientific) for global graft/host interaction visualization.

Ex Vivo Graft Assessment

A static glucose stimulated insulin release (GSIR) assay was used to assess the insulin secretion response of electively explanted devices. For each transplant, 3

controls (blank PDMS disk) or 3 treated (OxySite disk) devices were retrieved and

transferred into a 24-well non tissue culture treated plate (Falcon). Basal insulin

secretion was measured after a 2 h incubation at 37°C in 3 mM krebs buffer. The

devices were transferred into hyperglycemic culture medium (11 mM krebs buffer for

islet containing devices, and 25 mM for MIN6 containing devices) for an additional 2 h.

A second exposure to basal conditions was performed for an additional 2 h. Aliquots of

the medium were stored at -80°C. Further, constructs were placed in 1.7 mL Eppendorf

tube with 1 mL total insulin extraction buffer (0.18 M HCL, 70-75% Ethanol, prepared in

house) and stored at -80°C. Rat and mouse insulin ELISA quantified total insulin and

GSIR samples for islets and MIN6, respectively (Mercodia, Inc., Winston Salem, NC).

The total RNA of grafts explanted at elective time points was extracted, as

described previously. Primers used are the same as those listed in Table 5-1. Relative

154

gene expression was calculated using Taqman assays in a QuantiStudio 6 flex real-time

Statistics

Islets in vitro assessments corresponds to one single islet isolation with a n=3 for all groups. Data is represented as mean ± SD. Islet graft ex vivo assessments corresponded to three separate experiments (e.g. three islet isolations) with triplicate samples assess for each group. Data is represented as the mean of group replicates ±

SD. One independent experiments was used for MIN6 transplants. Non fasting blood glucose levels are represented as the mean ± SD, with the number of animals for each group shown in the figure legend.

Results

In Chapter 6, we demonstrated the potential of our in situ oxygen generating material to maintain islet viability and function in the early engraftment period in a syngeneic diabetic transplant model. The transplant platform implemented, however, screened the potential of in situ oxygenation to serve as a bridge to vascularization, with the goal to preserve islet viability until engraftment. In this chapter, the objective was to characterize the impact of OxySite using a completely avascular encapsulating device.

This approach provides a more rigorous model for examination of the impact of in situ oxygenation, while also exploring the potential of OxySite to be used for macroencapsulation approaches. For this purpose, islets were encapsulated within an agarose hydrogel with a single OxySite disk centrally located within the construct to provide optimal oxygen supplementation. For controls, agarose constructs were loaded

155

with a blank PDMS disk. The first line of experiments involved developing a fabrication platform that would combine the polymer with our material, which proved to be a significant challenge. The mechanical mismatch between the hydrogel and the PDMS disk, as well as the two step fabrication procedure, caused fractures and the consequent separation of the device into two separate pieces. Agarose of different tensile strengths, different gelation temperatures, and concentrations were screened to overcome this hurdle. After multiple tests, a prototype using a low gelling temperature agarose at a 2% w/w concentration was found to be robust enough to prevent fracture, while providing a mild gelation procedure to minimize impacts on the encapsulating cells. Moreover, modifications in the fabrication procedure were implemented (e.g. the central suspension of the OxySite within a custom made mold during pouring of the agarose liquid), which resulted in a single gelled construct.

A schematic of the early construct prototype is illustrated in Figure 6-1. In vitro screening of macrodevices with control or OxySite disks under low oxygen conditions over 24 or 48 h sought to examine the impact of supplemental oxygen on islet viability, phenotype, and function. Results demonstrated the ability of our OxySite disk to maintain islet viability under low oxygen conditions, as observed by improved metabolic activity over controls for both time points evaluated (Figure 6-3 Panel A, p< 0.05 at 24 and 48 h). Moreover, a decrease in cytotoxic LDH was also detected in culture medium from OxySite containing devices compared to controls (Figure 6-3 Panel B, p <0.05, and p< 0.001 at 24 and 48 h respectively). Further, islets encapsulated in OxySite treated constructs were responsive to a glucose stimulation after 24 h, with a mean index of 1.1

± 0.12. Conversely, complete dysfunction was observed for the low oxygen controls,

156

with no insulin release after high glucose stimulation observed and a mean index of 0.4

± 0.05 (Figure 6-3 Panel C). Gene expression analysis demonstrated the superiority of

OxySite containing devices to mitigate activation of glycolytic enzymes, Glut1, Pgk1 and

Ldh; as well as diminish activation of DNA damage proteins such as Didit4 24 h post initial low oxygen exposure. Conversely, Vegfa was upregulated in low oxygen controls compared to devices with an OxySite disk present (Figure 6-3 Panel D, p < 0.05).

Moreover, no difference in proapoptotic proteins Bax and Bak was observed at the time point evaluated.

Loading density in macrodevices was scaled up from 1500 IEQ to 4000 IEQ for in vivo efficacy studies in a rat model. Prior to in vivo transplants, gene expression analysis of these high loading density devices was performed to elucidate the impact of in situ oxygen delivery on glucose metabolism, apoptosis, and angiogenesis at this significantly higher packing density. Results found a down-regulation in the activation of glucose transporters, Glut1, and glycolytic enzymes, Ldha and Pgk1, in the OxySite containing devices, when compared to controls at both time points evaluated (Figure 6-

4 red dotted inset). Furthermore, while no difference was observed in apoptotic markers after 24 h, a substantial down-regulation of the BCL regulated pro-apoptotic marker Bax and the DNA damage regulator Didit4 was observed in the OxySite group after 72 h culture (Figure 6-4 blue dotted inset). Lastly, no difference in Vegf-a was observed at any of the time points tested, while a sharp downregulation of Tgfa was observed in the

OxySite group after 24 h post encapsulation.

Macroencapsulation devices were subsequently tested in vivo using a syngeneic female diabetic rat model. Devices were placed in pockets created from the omental

157

pouch of recipients. Figure 6-5 panel A exhibits non-fasting blood glucose levels for transplanted devices. Euglycemia was observed for both control and OxySite devices for the first 7 d post-transplant. Glucose levels below 200 mg/dL were measured for

OxySite containing devices up to 10 d post-transplant, with blood glucose levels fluctuating from 200 to 300 mg/dL at subsequent time points. Conversely, blood glucose levels for blank PDMS containing devices ranged between 300 to 500 mg/dL as early as

8 d post-implantation. Moreover, differences in metabolic control was observed in the

OxySite group 30 d post-transplant, when compared to PDMS-only controls.

Nevertheless, no statistical significance could be calculated as IVGTT was performed on only one control animal (2 out of the 4 control animals had to be euthanized due to poor glycemic control and health deterioration and technical challenges were experienced during the IVGTT on the third animal). Histological analysis of the explanted devices demonstrated a robust foreign body response to the graft, irrespective of the presence of an OxySite or a blank control disk (Figure 6-6). This fibrotic encapsulation of the graft created an additional barrier for oxygen and insulin diffusion, likely contributing to the observed graft destabilization at these time points.

The unfavorable fibrotic responses observed for the rat transplants at this site required the transition to a different model and transplant location. The intraperitoneal space (IP) has been widely investigated as an alternative site, particularly due to its ability to hold large transplant volumes [150, 164]. Thus, this site was selected for future transplants. Of concern, however, was the poor glucose sensing of this environment compared to vascularized sites, which can lead to delays in the systemic distribution of the released insulin for larger animal models [182-184]. As such, implants within this

158

site may experience a more exacerbated oxygen gradient, as well as variable glucose control. In addition to the transplant site, we transitioned these studies from rats to mice, as this smaller animal model would permit more efficient screenings. The dimensions of device was subsequently scaled down to accommodate for the smaller size of the new animal model. After numerous testing of oxygen generation and prototype fabrication, the construct size was reduced by 17%, while the OxySite disks dimensions were cut in half (Figure 6-2), again to ensure the mechanical stability of the graft within the IP space.

Efficacy experiments for the IP transplant site were first performed using MIN6 cells to efficiently examine the impact of in situ oxygenation on cell viability and efficacy in vivo. For this purpose MIN6 beta cells, at a loading of 2.5x106 total cells, were encapsulated within the devices and transplanted in the IP space of diabetic B/6 mice.

Recipients of devices supplemented with OxySite exhibited significant improvement in graft efficacy, with a mean reversal time of 5 d ± 3.8, compared to 16 d ± 6 for control groups (P = 0.008, Mantel-Cox test, Figure 6-7 panel A). While the blood glucose levels for control group recipients were not statistically different from pre-transplant levels for the first 10 d post-transplant, recipients of OxySite constructs demonstrated significantly lower blood glucose levels after only the first day post-transplant (P < 0.001, t-test of day 0 vs all time points). Further, the OxySite treated group exhibited more stable glucose control, as determined by a Levine’s test for homogeneity of variance (p =

0.2684); in comparison, a non homogeneity of variance was found for control samples

(p < 0.0001). Moreover, normoglycemia curves demonstrated significant differences between macroencapsulation devices containing OxySite, when compared to blank

159

controls (p < 0.008), with complete achievement of euglycemia in all animals receiving a

macroencapsulation device with an OxySite disk by day 11 post-transplant. Conversely, only 60% of animals with blank controls returned to normal blood glucose levels (Figure

6-7 Panel B). No variance in body weight was observed at all time points evaluated, but this was likely due to careful management of the animals to minimize the impact of hyperglycemia on weight loss (data not shown). These promising results demonstrate, for the first time, the ability of an in situ oxygen biomaterial to significantly improve the efficacy of a beta cell transplant in a diabetic animal model.

Following explantation of grafts 33 d post-transplantation, live/dead whole mount imaging revealed a more homogenous distribution of viable cells within the OxySite containing device; with live cells found both at the outer edge of the device as well within the center of the device (i.e. the edge next to the in situ oxygen generator).

Conversely, samples from control groups had a greater disparity of viable cells, with a preference to the outer periphery of the implant, as it is common within macroencapsulation devices, with minimal viable cells found within the central regions of the device (Figure 6-8). Likewise, H&E assessments of grafts revealed a lower concentration of cells within the edge exposed to the biomaterial in the control group, when compared to the OxySite containing group. Further, a higher number of cell clusters was observed in the OxySite containing device compared to controls (Figure 6-

9 right panel); perhaps, suggesting that a more rich oxygen environment was present within these devices compared to controls. In regards to implant biocompatibility, no difference was observed in the host response to devices containing OxySite and those containing PDMS-only (Figure 6-9 left panel).

160

To screen for the impact of OxySite on enhancing pancreatic islet viability and function in vivo, islet-containing agarose constructs were transplanted into mice, electively removed at selected time points (5 and 15 d post-transplantation), and examined via ex vivo functional assays and gene expression analysis. After 5 d in the peritoneal space of a mouse, the functional activity of islets, assessed via static low and high glucose stimulation, and the total insulin content was comparable for both OxySite and control implants (P = 0.8630 and 0.5007 respectively; Figure 6-10); demonstrating robust function for islets encapsulated within these immunoisolating devices. Yet, mRNA expression from immunoisolated devices upon explantation demonstrated a down regulation in the activation of known hypoxia activated factors such as HIF1a, and

HIF3a; as well as its downstream effectors Adm, Bnip3, and Pim1. Lower expression of apoptotic regulators Bax, Nfkb2, and Faslg was also observed in the OxySite group compared to controls. A more supportive metabolic profile was also observed in OxySite containing devices, as demonstrated by the down-regulation of glycolytic enzymes Ldha and Pgk1, as well as Glut1; hinting at the ability of our OxySite disk to support aerobic respiration within these devices in vivo. Interestingly, a down-regulation of Ins1 expression in the OxySite treated group compared to controls was also observed, which might seem contradictory to the results observed from the GSIR and total insulin assays. Nevertheless, insulin has been demonstrated to play a significant role under hypoxia, regulating the presence not only of HIF1a but also glycolytic enzymes and transporters to increase anaerobic respiration [185]. Thus, an increase in the expression of Ins1 in the control group might reflect a more hypoxic environment within the control compared to the OxySite containing devices. Further, a down-regulation in the

161

expression of immune mediators’ il6, il1b, ifng, and Tgf-β was observed in OxySite containing devices. Conversely, an up-regulation of Ccl2 (MCP-1) was shown in the

OxySite group versus control implant. This result correlates with those observed in vitro cultures with the presence of OxySite (see Chapter 4) and should be explored further to elucidate its effect on islet culture. Lastly, results demonstrated a decreased expression of pro-angiogenic factors Angptl4, Icam1, and Vegfa in OxySite macrodevices, when compared to blank controls. As expression of these factors in the blood plasma was not evaluated, translation of this gene expression trend to protein release cannot be validated. As observed in Chapter 4, mRNA expression patterns of these factors do not always lead to protein translation [186]. This is particularly true in stressful environments, such as inadequate oxygen tensions [187]. Thus, while these devices will remain avascular in vivo, future studies should investigate their potential to promote the secretion of pro-angiogenic proteins that can support vascularization around the device.

Gene expression analysis of explants retrieved at a later time point (15 d) found a significant shift in functional activity, with islets encapsulated in the presence of

OxySite material exhibiting response to high glucose challenge and islets encapsulated with control disks demonstrating dysfunction (Figure 6-11). The mean stimulation index

(insulin released during high glucose normalized to that released during low glucose) for the OxySite group was significantly higher than that observed for control samples (1.6 vs 0.5; p = 0.025). Furthermore, significant differences were observed in the total insulin content of OxySite samples compared to controls (p = 0.003). Finally, in line with results observed for day 5 explants, gene expression profiles for islets within the different

162

constructs were notably divergent, with down-regulation of HIF-1a, Glut1, Vegfa, and

Nos2 genes for OxySite samples, when compared to controls.

With promising results observed for MIN6 cells and in explant studies, the focus turned to characterize the impact of OxySite on improving engraftment in a diabetic mouse model using xenogeneic rat islets. Our ability to obtain higher islet yields from rat isolations compared to mice, motivated us to implement rat islets as our primary cell source. Therefore, while it was never the objective of this chapter to assess the immunoisolating properties of the encapsulating polymer, the superior yields and consequently higher number of grafts that can be tested lead us to move to a xenograft model. Since this construct scale, transplant site, islet source, and animal model combination had never been optimized before, significant efforts in this area were concentrated on determining a minimum islet transplant mass needed to achieve normoglycemia, and then decrease this to identify a dosage whereby oxygen effects on graft could be evident shortly after transplantation (See APPENDIX C). Figure 6-12 depicts the results from one of three different xenograft transplants performed, specifically the loading of 700 IEQ per construct (loading density of 0.5% v/v). No difference in diabetes reversal was observed between the control and OxySite group for the first 8 d post-transplant, indicating this mass was not a minimum amount at this oxygen threshold; however, decreasing the dosage to 600 IEQ resulted in loss of graft efficacy for both groups (see APPENDIX C); indicating an inadequate dosage. Adding to the complexity of this study, destabilization of the graft was observed around day 10 for both groups. This destabilization was observed for all of the transplants performed, irrespective of loading density. The timing of this destabilization, as well as the absence

163

of this responses in allograft-like models (as shown in MIN6 studies), indicate that this

could be instigated or augmented by xenograft responses.

Conclusions

The concept of immunoisolating barriers as a means to dampen the immune

attack of islet transplants is a highly desirable one that has been tested for almost five

decades [164]. During this time, considerable efforts have been expended on

developing and testing materials that are biocompatible and provide adequate oxygen

and nutrient permeability, while still delivering an effective barrier from the host immune

attack [164, 188]. Agarose, a thermoselective polymer extracted from seaweed [164],

was implemented as the immunobarrier polymer for the studies herein presented. The

disparity of its gelation and melting temperature (26-30 oC and 65 oC, respectively)

ensure gelation at cell compatible temperatures while retaining stability in the body,

thereby making it an attractive hydrogel for immunoisolated macrodevices [189, 190].

Moreover, agarose encapsulation of beta cells and islets does not hinder cell viability,

as observed from others, as well as our in vitro and in vivo data [65, 191, 192].

Despite numerous improvements in immunobarrier material selection, the implementation of this approach has only achieved marginal long term success, primarily due to the nonsupportive oxygen gradients that form post-transplantation. The

current clinical approach of islet transplantation, whereby islets are implanted into the

liver, leads to exposure of the graft to lower pO2 than those in their native environment

(i.e. 5 mmHg compared to 40 mmHg in the pancreas), resulting in significant islet cell

loss due to inadequate oxygenation [81]. Further, these grafts eventually regain a

vascular network that can support the residual graft. In contrast, islets encapsulated

within immunoisolating barriers never revascularize and rely solely on passive diffusion

164

for nutrient and oxygen exchange. Moreover, transplant site and thus accessibility to a vascular network can have a significant impact on graft oxygenation. Sites that can accommodate implantation of these grafts in vivo are typically plagued with poor vascular networks [150, 164], exacerbating the detrimental oxygen gradients observed within these devices. Oxygen transport through this macroencapsulated devices relies on diffusion from the vasculature, which, depending on the transplant site, can be several micrometers away from the device. During this process, oxygen is consumed by the cells entrapped on the outer portion of the device, prior to reaching the core. As diffusion is limited proportionally to the radial dimension of the device [65], approaches at reducing the size of these implants from macro to micro scale sizes have also been explored. Itawa et al. have reported successful diabetes reversal in mice receiving microencapsulated allograft islets for over 80 d without immunosuppression [193].

Microencapsulation has also been shown effective in xenograft models of porcine islets transplanted to diabetic cynomologus monkeys without systemic immunosuppression

[194]. Moreover, techniques that further decrease the diameter of the capsules and the overall membrane thickness, such as conformal coating, have shown some success in small animal models [195]. Nevertheless, scaling these approaches to larger size animals has been hindered by the required volume needed to attain a clinically relevant transplant mass [196], the inadequate oxygen tensions observed from aggregation of these capsules in vivo, when placed in biped animals [164]; and the challenging retrievability of these implants.

Increasing oxygen availability within macroencapsulation devices has been demonstrated by enhancing the vascularization around the graft, either by preimplanting

165

devices to induce vascularization [197] or by implementing angiogenic motifs that would hasten the revascularization process [180], as well as by techniques that increase oxygen delivery through external oxygen tanks [167]. Neovascularization of the subcutaneous space, either in the presence or absence of encapsulating devices, have demonstrated success in small animal models [8, 166]. Yet, the double injury required for this approach, during pre-vascularization and transplant, the required transplant mass, and the mechanical stress associated with the site can prove challenging when translating into higher animal models. Implementation of angiogenic motifs tethered to polymers has also been employed to hasten the revascularization process [198].

Nonetheless, translation can be hindered by issues associated with vascular recession after depletion of signaling molecule, as well as leaky networks due to the nonphysiological mechanism of vascular formation. Lastly, oxygenation of islets encapsulated within an alginate hydrogel, encased by an immunoisolated barrier in the presence of oxygen supplementation, has proven successful at restoring normal blood glucose levels in allograft models [41]. Still, the higher risk for infections, the need for patient compliance, and the requirement for an external oxygen tank make this technique highly cumbersome.

Herein, we explored the impact of an in situ oxygen generating material within an immunoisolated device for beta cell and islet transplantation. Our in vitro data highlighted the ability of our oxygen generator to support islet aerobic glycolysis under low oxygen conditions, as seen by decreased activation of glycolytic enzymes in

OxySite containing group compared to controls. In vivo testing in a syngeneic animal model demonstrated the ability of islets encapsulated with OxySite to improve plasma

166

glucose levels compared to blank controls. Moreover, translation into an allogeneic animal model with beta cells enabled us to demonstrate the ability of our in situ oxygen generator to promote beta cell growth and maintain function in vivo. Further, translation into a xenograft model confirmed the ability of our OxySite disk to maintain islet function, reduce mRNA expression of hypoxic and inflammatory markers early and at 15 d post-transplant.

These results demonstrate, for the first time, the ability of our material to maintain viability in vivo; however, its impact on islet function could not be robustly examined in a diabetic transplant model. As this was the first time these devices were translated in vivo, a significant amount of time was expended on assessing the dosage required for reversal in both the rat and mouse models. Even though promising results were obtained at the rat scale, an unforeseen robust foreign body reaction limited our ability to maintain stable function over time. Addressing this issue was beyond the scope of this thesis. The transition to a mouse model brought about different challenges, in particular the issue of dosage and plausible xenograft mediated responses. Significant efforts were implemented in determining a minimal mass, whereby oxygen supplementation could have an impact early on after transplantation. Tuning islet loading proved to be a fine balance in which 100 IEQ difference could switch a functional graft to highly unstable. From our results, it can be inferred that a dosage between 600-700 IEQ could be the tipping point in our system; however, due to time constraints, this will be tested in future studies. Moreover, destabilization of grafts was observed for all of our xenograft transplants, with the timing consistent with a xenograft response; further hindering our ability to maintain differences in graft efficacy in vivo.

167

Addressing this issue in the future would require implementation of allo or syngeneic mouse models, or application of more impermeable encapsulating polymers.

Overall, our data demonstrates our ability to fabricate mechanically stable immunoisolating devices containing OxySite disks for the encapsulation of beta cells and islets. Further, these devices can be modified and designed for specific animal models. Significantly, for the first time, we demonstrated the ability of our OxySite material to sustain islet function and promote beta cell function in vivo.

168

Table 6-1. Primer Sequence for PCR Array. Gene Organism Gene ID Il6 Rattus norvegicus 24498 Ins1 Rattus norvegicus 24505 Hif1a Rattus norvegicus 29560 Pim1 Rattus norvegicus 24649 Pgk1 Rattus norvegicus 24644 Il1b Rattus norvegicus 24494 Icam1 Rattus norvegicus 25464 Hif3a Rattus norvegicus 64345 Vegfa Rattus norvegicus 83785 Tgfb Rattus norvegicus 25008 Bax Rattus norvegicus 24887 Nfkb2 Rattus norvegicus 309452 Angptl4 Rattus norvegicus 362850 Ifng Rattus norvegicus 25712 Faslg Rattus norvegicus 25385 Ccl2 Rattus norvegicus 24770 18SrRNA Rattus norvegicus PPR57734E

Figure 6-1. Schematic representation of immunoisolating agarose devices implemented for n vitro assessments. The yellow region represents the OxySite material while the blue region represents the 2% w/v agarose cells mixture.

169

Figure 6-2. Schematic representation of immunoisolating agarose devices implemented for transplantation. The yellow region represents the OxySite material while the blue region represents the 2% w/v agarose cells mixture.

170

Figure 6-3. Effects of OxySite supplementation in immunoisolated devices containing pancreatic islets exposed to low oxygen conditions in vitro. A) Enhanced metabolic activity was observed for the group containing an OxySite disk compared to low oxygen controls after both time points evaluated. B) Increased release of cytotoxic byproduct LDH was observed in low oxygen controls compared to OxySite. C) Low oxygen exposure lead to dysfunctional islets after glucose stimulation. Viability could be restored by implementation of OxySite in the device. Inset represents mean index. D) Gene expression represented as fold over 0 h controls for controls and OxySite containing devices after 24 h of low oxygen exposure. P <0.05, 0.005, 0.0001 for *, **, **** respectively.

171

Figure 6-4. mRNA expression of 4000 IEQ islets encapsulated within an agarose hydrogel with an OxySite disk or with a blank PDMS disk under low and standard oxygen conditions. A-B) mRNA expression is expressed as fold regulation from standard oxygen controls. Red dotted square highlights gene involved in glucose metabolism. Blue dotted squares highlights represent genes involved in apoptosis after 24 h and 72 h of encapsulation respectively.

172

Figure 6-5. In vivo data of syngeneic imunoisolating device transplants in female Lewis rats. A) non-fasting blood glucose on female graft recipients highlights the potential of our in situ oxygen generator to support graft viability in vivo. B) Metabolic clearance measured via an IVGTT 30 d post-transplant.

173

Figure 6-6. Histological analysis of immunoisolated devices transplanted into the omental pouch of diabetic female recipients. Left panel represents devices transplanted with a Blank PDMS disk as controls; right panel represents devices transplanted with OxySite. Scale 600 µm.

174

Figure 6-7. In vivo data of allograft imunoisolating device transplants in mice. A) non- fasting blood glucose of animals post-transplant shows stark difference in normoglycemic reversal times between ctrl and OxySite group. B) Survival curve demonstrating time to reversal for both OxySite and Blank control containing devices.

175

Figure 6-8. Live Dead images of agarose hydrogels sections 33 d post-transplant. Top row represent agarose hydrogels containing a blank PDMS disk (control group). Bottom row represents constructs containing an OxySite disk. Left column represents the edge exposed to the host environment. Right column represents edge near the PDMS disk. Scale 100 µm.

176

Figure 6-9. Histological assessments of samples 33 d post-transplant. Top row represent agarose hydrogels containing a blank PDMS disk (control group). Bottom row represents constructs containing an OxySite disk. Left column represents the edge exposed to the host environment. Right column represents edge near the PDMS disk. Scale zoom in images100 µm, whole device 500 µm.

177

Figure 6-10. Ex vivo assessments of islets transplanted in encapsulated immunoisolated devices in diabetic b/6 mice. Grafts were electively removed 5 d post-transplant. Islet function was assessed via static glucose stimulated insulin release test. A) Insulin release after low and high glucose stimulation demonstrates robust function for islets encapsulated within immunoisolating devices. Insets represent mean index of islets post glucose stimulation. B) No difference in total insulin content was observed between grafts at 5 d post- transplant. C) mRNA expression from PCR array of immunoisolated devices containing an OxySite disk. Data is expressed as fold regulation over control devices.

178

Figure 6-11. Ex vivo assessments of islets transplanted in encapsulated immunoisolated devices in diabetic b/6 mice. Grafts were electively removed 15 d post-transplant. A) Insulin release after low and high glucose stimulation demonstrates strong function for islets encapsulated within immunoisolating device in the presence of OxySite compared to controls. Insets represent mean index of islets post glucose stimulation. B) Significant differences in total insulin content was observed between grafts at 15 d post-transplant. C) qRT-PCR expression of hypoxia regulated genes for immunoisolated devices transplanted with controls and OxySite disks at 15 d post-transplant.p<0.05.

179

Figure 6-12. In vivo data of xenograft imunoisolating device transplants in mice. A) non- fasting blood glucose of animals post-transplant shows no difference in normoglycemic reversal times between ctrl and OxySite group. Destabilization of the grafts was observed at around 10 d for all animals. B) Animals body weight was monitored daily post-transplant with no difference observed between both groups.

180

CHAPTER 7 OPTIMIZATION OF OXYGEN GENERATION BY ADDITION OF OSMOTIC AGENTS

Background

While encapsulation can provide an immunobarrier for protecting the islets from the immune system, the creation of a barrier between the islet and the host microenvironment carries the disadvantage of limited oxygen transport. Hence, successful outcomes for islet encapsulation, as presented in Chapter 6, will require implementation of oxygen delivery to preserve islet survival not only after implantation but for the duration of the graft. Our previous results highlighted the ability of OxySite to improve graft survival in the early period and up to 30 d post-transplant; however, the

duration of oxygen delivery needs to be prolonged to enhance clinical appeal. Thus, studies in this chapter sought to improve the duration of oxygen release for the OxySite bioactive material introduced in Chapter 3.

The incorporation of osmotic agents has been widely investigated as a method to overcome the highly hydrophobic nature of PDMS [76, 199, 200]. The addition of excipients such as inorganic salts can result in polymer cracking due to osmotically induced water migration, hence their designation as “water carriers” [76]. This polymer

cracking leads to an increase in the porosity of the system and an effective promotion of

hydration and swelling of the polymer [200]. Since oxygen delivery from our bioactive

material is essentially contingent on porosity and thus tortuosity of the material (see

Chapter 3), the manipulation of porosity could have the advantageous effect of

extending oxygen release kinetics.

This study tested the incorporation of different water carriers during the fabrication of our oxygen generating bioactive material. As a critical parameter of this

181

study was to assess the enhanced hydration of the matrix, changes in material

dimensions, as well as its effect in the surrounding milieu, were evaluated. Lastly, our

most promising prototype from this study was tested in the presence of a beta cell line

in vitro.

Methods

Materials

All chemical reagents were purchased from Sigma-Aldrich, unless otherwise noted. All culture media was sourced from Mediatech. Poly(dimethyl)siloxane (PDMS)

(NUSIL 6215, clinical-grade) was sourced from NuSil Silicone, USA.

Fabrication of Oxygen Generator with Osmotic Agents

Oxygen generating material with osmotic agents was fabricated using similar methods as those described in Chapter 3. Briefly, 25% w/w CaO2 was added to PDMS

in a 4:1 vol/vol PDMS monomer to platinum catalyst and osmotic agents, such as

sodium chloride (sifted between 250 to 425 µm), glucose, and PEG200, were

implemented at different concentrations. The mixture was then homogeneously mixed

and defoamed via centrifugal mixer (Thinky Mixer, USA), poured into a cylindrical mold

(29 mm diameter to 1 mm thickness), and cured overnight at 40 oC. Prior to use, smaller

disks (10 mm) were punched out of the larger cured disk. Resulting disks were sterilized

with ethanol for 30 min and rinsed five times in PBS (Invitrogen) prior to performing

experiments. OxySite disks doped with glucose, salt, or PEG were termed OxyG, OxyS, and OxyO, respectively.

Macroscopic Analysis

Dimensional measurements were made on test disks after leaching them in 1 mL

PBS buffer (replacing buffer every day) for 60 days. Samples were checked for weight

182

uniformity (Miller Toledo), diameter, and thickness (micrometer) with n = 3 disks per time point per prototype.

When glucose or salt was used as an osmotic agent, the amount of solute within the leaching solution of the disk prototypes were quantified using either a glucometer and glucose strips (one touch, ultra mini) or weight measurements of the lyophilized solution, respectively. All measurements were compared to initial buffer salt concentration.

Microscopic Analysis

For the OxyO prototype, morphological changes were investigated using a scanning electron microscope Hitachi VP-SEM S-3000 coupled using a secondary electron detector (SE). Scans were performed at 3-10 kV.

Porosity and pore parameters of OxyO biomaterial was determined by mercury intrusion technique using a Quantachrome Poremaster Mercury Porosimeter

(Quantachrome Instruments, USA). Whole OxyO constructs were weighed and placed in the cup of the penetrometers, which was closed by tightening the cap. The penetrometer, along with the sample, was placed into the pressure chamber of the porosimeter for measurement of pore properties. Intrusion and extrusion assessments were performed at low pressures (50 Pa) and high pressures (up to 60 KPa).

Swelling Measurements

Percent weight change and swelling ratio were determined at all designated time points for OxySite materials with different osmotic agents. Samples were measured and weighed prior to addition of buffer (PBS; Sigma-Aldrich), and subsequently after retrieval at different hydration time points. Water content was determined from change

183

in weight %. Finally, the swelling ratio was simply the wet weight normalized by the dry

weight.

Non-invasive Oxygen Measurements

Oxygen release profiles of OxyO disks (initial dimensions 10 mm in diameter 1

mm in height) were obtained via a custom-built, non-invasive, sealed titanium chamber

system (Instech Laboratories, PA). Partial oxygen tensions in the solution medium were measured using non-invasive oxygen sensors (PreSens, Germany). Sensor spots

consisted of a ruthenium coated via a sol-gel polymer. The optical signal was sent to the sensors via fiber cables from a custom transmitter box (PreSens, Germany). Ruthenium

in spot sensors quenched oxygen in the medium shifting fluorescent signal, which can

then be correlated to the oxygen concentration in the solution. Oxygen concentrations

were recorded by Presens open software in units of mmHg (torr). 1 independent disk

was assessed at each interval time point (0-4 d, 15-22 d, 60-62 d). Samples for the first

time interval were assessed for oxygen kinetics right after sterilization. Samples for the

two other time intervals were leached in a plate for 15 and 60 d, respectively, prior to

kinetic assessment.

Hydrogen Peroxide Measurements

Hydrogen peroxide production from OxyO disks was assessed for samples

incubated in 24-well plates with 1 mL of PBS at 37 °C for 60 days. Samples were

refreshed every day and collection taken at defined time points. Hydrogen peroxide

concentration was measured using a colorimetric assay (Assay Designs).

MIN6 Culture

MIN6 cells were cultured as monolayers in tissue culture treated T-flasks with the

medium change every 2-3 days. Media consisted of Dulbecco’s Modified Eagle’s

184

Medium (DMEM) (invitrogen) with 5.5 mM glucose and supplemented with 10% fetal

bovine serum, 1% penicillin-streptomycin, 1% L-glutamine and 0.001 % (v/v) β-

mercaptoethanol. Upon reaching confluency, MIN6 cells were trypsinized using 0.25%

Trypsin with EDTA (Invitrogen) and either plated at a lower density for propagation or

used in construct fabrication (passage numbers 29-31).

MIN6 In Vitro Cultures

In vitro cultures of MIN6, a beta cell line, were implemented to compare oxygen

generating capabilities of OxyO material versus OxySite after long term hydration. Both

OxyO and OxySite disks were prepared as described above and in Chapter 3,

respectively. Disks were sterilized and placed in PBS buffered in a 24-well plate in a

humidified incubator at 37oC. The incubating solution was replaced daily to simulate

release kinetics experienced during non-invasive oxygen monitoring. Disks were used for experimental cell studies 15 d post leaching. For in vitro cultures, disks were placed

in a 24-well plate and a non-tissue culture treated insert (Millipore) was placed on top of

the disks. MIN6 cells (350x103) were resuspended in 300 µL of DMEM media and

placed inside the insert, while 700 µL of DMEM was placed outside the insert. Samples

were incubated under standard oxygen conditions set at 0.2 mM (20%) O2 with 0.05

mM (5%) CO2. For low oxygen conditions studies, cell were incubated in a C-Chamber

Hypoxia Chamber (Biospherix, New York) set to 0.01 mM (1%) O2 with 0.05 mM (5%)

CO2 using controllers (Biospherix). Media was changed every 2 - 3 days. Every 5 days,

cells and supernatant was collected and assessed for cell viability by MTT metabolic

assay (Promega, WS), LDH production, hydrogen peroxide, and live/dead staining

(Invitrogen, CA).

185

In Vitro Cellular Assessments

The metabolic activity was assessed via MTT (Promega), as previously published [65]. Briefly, at time points designated, samples were incubated in 250 μL of complete CMRL media with 28 μL of MTT for 1 h under standard oxygen conditions.

Stop solution (185 µL) was then added under agitation and incubated for 48 h to permit complete formazan crystals solubilization. Absorbance (570 nm) was then measured from triplicates samples using a SpectraMax M5 microplate reader (Molecular Devices).

Lactate dehydrogenase (LDH) release was quantified using methods previously described [65]. Briefly, 100 μL of supernatant was mixed with 100 μL of working reagent

(Cytotoxicity Detection Kit, Roche) and incubated for 30 min at RT. Stop solution (50 μL of 1 N HCl) was added and the resulting absorbance in triplicate samples was measured at 492 nm.

Hydrogen peroxide assessments follow the instructions provided above on the supernatant samples collected after each time point or media change time point.

Live/Dead samples were visualized using the Viability/Cytotoxicity Kit (Invitrogen) and a Zeiss LSM 510 inverted confocal microscope, as described previously [65]. After incubation, samples were imaged for live cells (green) and dead cells (red). Z stacks

(2.29 µm thickness; 7 slices per image; 1024x1024; 20x objective) were merged into a

3D projection using Zeiss Zen Software.

Statistics

Macroscopic analysis corresponded to an average of 3 random samples represented as mean ± SD. Microscopic and surface characterization analysis corresponds to one disk per sample time point. In vitro data corresponds to two

186

independent experiments with data summarizing the mean ± SD of group replicates (n =

3 for treatment or prototype for each time point).

Results

The primary objective of this study was to examine the impact of osmotic agents on improving the duration of oxygen release for our OxySite material. As schematically illustrated in Figure 7-1 panel A, it was hypothesized that the addition of water carriers into the oxygen bioactive material would improve water migration by dissolution of excipients in the buffer medium during hydration (inset). To test this theory, we screened three different osmotic agents: sodium chloride, glucose, and PEG200, termed OxyS, OxyG, and OxyO, respectively. Figure 7-1 panel B illustrates changes observed in the biomaterial after excipient additions. As expected, changes in the excipient’s density at the surface of the biomaterial was increased, depending on the w/w concentration of the osmotic agent versus the biomaterial polymer.

While the inorganic salts and sugars are the most commonly employed water carriers, a major concern in their use in this system was the potential impact of the dissolution of these agents into the surrounding milieu. For OxyG, the concern was that glucose release into the surrounding media would lead to inadvertent stimulation of glucose sensitive beta cells. Alternatively, changes in salt concentration in the supernatant exposed to OxyS disks could result in variations in solution osmolarity, which could dramatically affect cell homeostasis. For this purpose, quantification of the levels of glucose and salt present in the buffer solutions during incubation of the OxyG and OxyS system was conducted. As it can be seen in Figure 7-1 panel C, the addition of sugar to the bioactive material lead to a concentration dependent increase in the amount of glucose released into the buffer system, from 300 mg/dL to above 600 mg/dL

187

on day 1 post leaching. Furthermore, an increase in the total concentration of salt in the

medium, ranging from 2.7% to 8%, was observed in the OxyS disks. Of interest, no

swelling, as measured by increased in weight and radial dimension, was observed in

these materials, despite the increase of excipient concentration in buffer system (data

not shown). Overall, both glucose and salt were eliminated as potential osmotic agents for our OxySite system due to the potential detrimental impacts on beta cell function and viability.

Based on the results presented above, the focus turned on utilization of a more inert material as a water carrier, specifically PEG. Hydrophilic polymers, such as PEG,

have been previously evaluated in drug delivery systems within silicone based matrices

[77]. The addition of PEG has been demonstrated to lead to considerable matrix

swelling following contact with water. Further, the extent of this swelling was impacted

by the PEG content, as well as by the size of the PEG polymer. PEG200 was chosen

for our system due to its low viscosity, which would aid in the homogenous mixing with

the PDMS matrix during fabrication; as well as its complete solubility in water, which

was theorized to help increase the osmotic pressure and enhance matrix swelling. The

impact of different w/w concentrations of the PEG200 material was assessed during

fabrication, in order to optimize pore formation. It was found that the addition of ≥ 20%

w/w PEG200 into the PDMS/CaO2 mixture inhibited proper curing of PDMS polymer,

even after exposure to heat for prolonged periods of time (data not shown).

Furthermore, concentrations below 10% w/w lead to minor polymer swelling and no

visual differences with the original OxySite prototype after prolonged hydration. An

intermediate concentration of 16% w/w was found optimal, as PDMS curing was

188

unaffected and visual swelling was observed after water exposure. Herein and through this chapter, this material is labeled as OxyO. Figure 7-2 panel B demonstrates the ability of the OxyO material to heighten polymer hydration, as observed by the enhanced swelling 30 d post hydration. Moreover, changes in the optical pigmentation of the material after prolonged buffer exposure were observed. Finally, as depicted in

Figure 7-2 panel C-D, hydration of the OxyO material for 60 d lead to significant increases in biomaterial weight and diameter (p<0.0001), but no significant changes in thickness. Testing of biomaterial swelling post-leaching found an increase in the swelling ratio of 82% ± 0.22 after 1 d of leaching and a maximum swelling of 128% ±

0.13 after 15 d post leaching.

Microscopic characterization of pore formation were performed using a mercury porosimeter. The intruded mercury volumes were recorded and converted to porosity, as presented in Figure 7-3 panel A for OxO material post fabrication. Over a 1700% increase in the material porosity was measured for the OxyO material, when compared to OxySite. Moreover, a histogram of pore diameter size intruded after mercury assessments highlights the expanded distribution of pore size, ranging from 4 to 254

µm, with larger pore sizes present when compared to OxySite (see in Chapter 3).

Likewise, SEM micrographs of OxyO disks post fabrication illustrated a greater presence of pores, both at the surface and in the cross section of the material.

Moreover, oxygen kinetics demonstrated the ability of the OxyO material to generate oxygen at levels comparable to the OxySite prototype, 130 mmHg ± 71 vs 117 mmHg ±

69 for the first 4 d post leaching (Figure 7-4, see Chapter 3). At later time points (i.e. 15 d post-leaching), the amount of oxygen generated from the OxyO material exceeded

189

that of the OxySite prototype with 132 mmHg for OxyO devices compared to 45.7

mmHg for OxySite. Lastly, robust oxygen tensions of 12 mmHg were still observed for

OxyO devices 60 d post initial hydration. These results indicate that the OxyO prototype

was able to support longer oxygen generation than the original OxySite prototype.

With material parameters indicating increased water migration and porosity, the

next experiments focused on the impact of these new OxyO materials on beta cell

viability and function in vitro. To test our hypothesis that improved porosity would lead to

enhanced temporal oxygen release, OxyO and OxySite materials were pre-leached for

15 d prior to assessments. Metabolic activity, as well as viability as observed by

live/dead confocal images, were evaluated every 5 d. As shown by Figure 7-5 panel A,

exposure of beta cells to standard oxygen conditions (0.2 mM O2) lead to an initial

increase in metabolic activity that stabilized for the remainder of the experiment.

Conversely, the culture of cells under low oxygen conditions (0.01 mM O2) lead to a hasty decrease in metabolic activity, as seen by decreased MTT absorbance levels, 5 d

post exposure. The introduction of our standard OxySite disk into the system preserved metabolic activity to levels comparable to standard oxygen controls for the first 10 d of culture; with decrease around day 15, but still greatly improved from the low oxygen

controls (p<0.001). Surprisingly, the addition of OxyO material into the culture system

lead to a swift decrease in metabolic activity at all time points collected.

To understand this unexpected and unfavorable impact of OxyO in the culture

system, the culture medium was analyzed for detrimental byproducts of calcium

peroxide decomposition. As seen in Figure 7-5 panel B, no differences in hydrogen

peroxide concentration were observed in standard and low oxygen cultures with or

190

without an OxySite disk; however, the presence of OxyO lead to a sharp increase in hydrogen peroxide concentrations to values above toxic range (>200 µM) in the first 7 days of culture. Results from MTT assessments were validated with live/dead confocal imaging of samples, with robust beta cell clusters apparent in both standard and

OxySite culture group; whereas more dead cells and cellular debris is present in low oxygen controls and OxyO groups (Figure 7-5 panel C).

Based on the aforementioned results, extensive studies on hydrogen peroxide accumulation from OxyO prototypes was performed in an acellular system, where the disk was leached continuously for 60 d. As observed in the cellular studies, a significant increase in hydrogen peroxide concentration was observed in the OxyO group for the first 7 d post-leaching, when compared to OxySite and control samples (p<0.0001 for

Oxysite vs OxyO and control vs OxyO), with a decrease to 50 µM from 10-30 d, and an upsurge in accumulation between 30 and 40 d (Figure 7-6), in accordance with the data collected from the in vitro cellular studies.

Conclusion

In previous chapters, we had demonstrated the ability of our OxySite material to generate oxygen for up to 30 days. Further, we have found this material capable of supporting cell viability and function under hypoxic conditions in both 2D and 3D culture systems. Nevertheless, continuous oxygenation over the current capabilities of our

OxySite system is necessary for the long-term implantation of immunoisolation devices, like those presented in Chapter 6. Formation of detrimental oxygen gradients at the core of 3D cell encapsulation devices have been extensively reported [201, 202]. While, these devices have been implemented successfully in vivo in both small [203, 204] and large animal [205] model of diabetes; inadequate oxygenation has hindered its long

191

term success and translation [178]. In our laboratory, we have demonstrated mitigation of the formation of hypoxic cores in 3D immunoisolating devices in vitro by the introduction of an oxygen generating material at the center of the device [65].Yet, oxygen delivery must be continuous through the life of the graft, as revascularization will not occur to alleviate these detrimental oxygen gradients. Thus, it was the aim of this chapter to optimize the fabrication procedure of our OxySite material to increase the availability of reactive peroxide to increase temporal oxygen release to time frames that would support macroencapsulation devices over extended time periods (i.e. > 6 months).

The innovation of our material relies on our ability to control the release kinetics of oxygen production, rendering its implementation free of any scavenging agents, such as catalase. Tunability of the kinetics relies mostly on the hydrophobicity of the PDMS encapsulating polymer. This property, in addition to its ability to incite minimal foreign body responses, has made PDMS an attractive polymer in the drug delivery field [206].

While, numerous polymers have been designed for drug release systems, the high permeability and solubility of PDMS to gases and hydrophobic compounds, as well as its availability in medical grade levels, low toxicity, and malleability to be casted in different shapes and sizes makes it a widely implemented polymer in the field of controlled release [88].PDMS has been implemented as a drug delivery depot for a variety of applications ranging from contraceptive delivery [207] to an intraocular drug delivery device [208]. Further, previous reports have demonstrated the enhancement matrix porosity of silicone by the addition of liquid or solid water carriers into the matrix during fabrication [76]. Porogens dissolve when in contact with the buffer medium,

192

diffusing out of the matrix and leaving behind pores within the structure. Moreover, the addition of crystalline solid water carriers can generate PDMS membranes with controllable porous structures [199]. In this chapter, we investigated three different porogens, a liquid (PEG 200) and two solid carriers (glucose and sodium chloride). The solid water carriers were chosen due to their wide application in particle leaching techniques, as well as their ability to form pores of controlled sizes [209, 210]. PEG was selected due to its hydrophilicity, as well as extensively characterized biocompatibility

[211, 212].

The addition of the two solid carriers, glucose and salt, did not hinder polymer curing for all concentrations tested. Moreover, no difference in water content was observed post leaching, as seen by consistent dry and wet weights post buffer exposure; indicating no swelling of the matrix. Hydration of these OxyG and OxyS materials, however, resulted in undesirable levels of solutes in the supernatant during the first days post leaching; to concentrations that can have detrimental effects on cell culture. Thus, further testing on these platforms were halted. Moving forward, the liquid water carrier, PEG 200, was implemented in our OxySite prototypes (OxyO). Screening of different loading densities of PEG supported the concentration of 16 w/w%, which did not impact curing and exhibited enhanced water migration, as seen by a significant increase in matrix swelling, as well as porosity and pore distribution. Further, evaluation of oxygen release kinetics from these devices demonstrated similar short term kinetics for OxyO disks compared to original OxySite disks, with no significant burst effects observed from enhanced porosity. Yet, a substantial increase in oxygen generation from

OxyO matrices was observed at later time points, with meaningful oxygen tensions

193

detected even after 2 months post initial hydration. Delivery of sustained oxygen concentrations for extended time periods, is a quality highly sought for in vivo oxygen delivery systems, as it enhances the potential of avascular systems, as those presented in Chapter 6, to maintain viability and function long-term. These results demonstrate our ability to manipulate material properties to enhance the oxygen generating profiles of our oxygen generating devices. Moreover, it corroborates our hypothesis that enhanced pore formation, and thus water migration into the system, can have the positive effect of extending temporal oxygen release.

The impact of OxyO on supporting beta cell viability was assessed in vitro. To highlight the potential of OxyO to provide elevated oxygenation for longer time periods, the OxySite and OxyO samples were leached for 15 d prior to testing. It was established in Chapter 3 that continuous exposure of the oxygen generating material to buffer lead to a decrease in the kinetics of oxygen release. Thus, the leaching of disks prior to testing allowed the examination of the capacity of the new OxyO prototype to sustain cell viability for at least 15 d post initial hydration. Unexpectedly, results found OxyO decreased the metabolic activity of beta cells, compared to both low and standard oxygen controls, and in stark contrast to the group containing the OxySite disk. Analysis of reactive byproducts in the supernatant found a significant increase in hydrogen peroxide accumulation in OxyO samples, compared to all other groups tested. Thus, it can be inferred that the enhanced water uptake and reactivity of the matrix, while promoting the decomposition of the reactive agent, lead to the elevated accumulation of toxic byproducts in the surrounding milieu.

194

Overall, this chapter demonstrates a facile method for the fabrication of porous

OxySite membranes from the addition of liquid water carriers into the polymer matrix. By controlling the percentage of liquid carrier content, the porosity, swelling and mechanical properties of the material can be drastically altered. Further, the oxygen kinetics of the system can be enhanced. Unfortunately, cell culture testing of these new prototypes resulted in unexpected negative profiles. The detrimental effects of OxyO in the culture system was attributed to the augmented concentration of hydrogen peroxide observed in the culture milieu. Nevertheless, this study demonstrates our ability to manipulate fabrication of our oxygen generating material to incorporate agents that can improve physical characteristics of the prototypes. Future studies should concentrate on determining an appropriate liquid carrier concentration that can enhance material reactivity while still retaining cyto-compatibility.

195

Figure 7-1. Addition of osmotic agents to OxySite material can enhance porosity. A) Cartoon representation of OxySite material with added osmotic agents. Inset represents pore formation after water dissolution of excipients from polymer. B) Pictogram representation of bioactive material with salt (bottom panel) and sugar (top panel) solid particles added at different w/w concentration. C) Glucose concentration in leaching buffer of salt containing oxygen biomaterials at different w/w concentration of water carriers after one and two days of leaching. D) Percent salt concentration remaining in solution (normalized to buffer) after 1 day of leaching oxygen materials with different salt concentrations.

196

Figure 7-2. Optimization of OxySite material by addition of PEG 200 as a porogen. A) Cartoon representation of OxyO material after addition of PEG 200. Inset represents pore formation after water dissolution of excipients from polymer. B) Pictogram representation of OxyO material after short (1 d) and long (30 d) leaching time. Clear differences in swelling can be observed in material after 30 d post leaching. C) Weight changes in OxyO material after hydration for 60 d. D) Dimensional changes in thickness (open circle) and diameter (black squares) of OxyO after hydration for 60 d.

197

Figure 7-3. Incorporation of PEG200 in oxygen generating material can improve matrix porosity. A) Table comparison of porosities of OxySite material compared to OxyO after 1 d post leaching. B) Histogram representation of pore size intruded in OxyO material after porosity assessments. C) Scanning electron micrographs of OxyO material surface (top panel) and cross section (bottom panel) after 1 d post hydration. Scale 100 µm.

198

Figure 7-4. Comprehensive oxygen kinetic measurement for OxyO sample of 10 mm in diameter. Oxygen tension measured via a non-invasive sensing system in a temperature controlled environment.

199

Figure 7-5. Long term co-culture of beta cells with OxyO and OxySite under low oxygen tension. A) MTT viability, expressed as optical density (OD), of 350×103 MIN6 cells cultured at 0.01 mM oxygen with a blank PDMS disk (open sqaures), with an OxySite disk (black square) or with an OxyO disk (black triangles). A normoxic control, cells cultured at 0.2 mM was implemented as a reference (red open square). Metabolic activity was determined every 5 days for 15 d. B) Hydrogen peroxide concentration in milleu collected from beta cells samples culture under low oxygen tension. Legend is similar to that described for panel A. C) Representative live/dead confocal images of MIN6 beta cells culture with OxySite or OxyO under low oxygen as well as controls under low and standard oxygen after 5 and 10 d of co-culture. Scale bar 50 µm.

200

Figure 7-6. Hydrogen peroxide concentration of Blank PDMS (open circles) disks, OxySite disks (black squares), and OxyO disks (Grey open triangles) at selected time points for 60 days.

201

CHAPTER 8 SUMMARY

Conclusion

Clinical islet transplantation has the potential to cure type I diabetes.

Nevertheless, the long-term success of this therapy has been hindered by the significant loss of the graft in the early post-transplant period, due to various challenges, whereby the lack of adequate oxygenation plays a prominent role. Thus, the advancement of islet transplantation would greatly benefit from strategies aimed at improving graft oxygenation in vivo.

The central hypothesis of this dissertation was the engineering an in situ oxygen generating device to mitigate hypoxia-mediated apoptosis and improve function in cellular based devices containing pancreatic islets and beta cells. Steps to achieve these goal were divided into three different aspects: 1) a complete optimization and characterization of the biomaterial device; 2), a comprehensive assessment of oxygenation effects on pancreatic cultures exposed to low oxygen conditions; and 3), the implementation of this oxygen delivery system into two distinct platforms for extrahepatic transplantation.

The first aim of this thesis focused on the extensive characterization of our in situ oxygen generating device. Our first contribution to this goal was the development of a fabrication protocol that greatly improved the reproducibility of our devices. Moreover, we provided, for the first time, a comprehensive analysis of the biomaterial properties, both at the micro and macro scale, as well as oxygen kinetic profiles for extensive time periods. This work culminated in the establishment of a highly accurate mathematical model that was able to predict the release kinetics of oxygen from the biomaterial, as

202

well as the sustainability of oxygen release. Of note, this model can be implemented with OxySite prototypes of various geometries with high precision. Lastly, we developed computational models investigating the effect of our oxygen release capabilities on pancreatic islet cultures of different loading densities; taking into account islet dimensions as an important parameter in oxygen diffusion. Moreover, these models examined the effect of environmental oxygen conditions with or without our oxygen generating system. Implementation of this tool can be highly useful for assessment of our devices under different conditions; narrowing down the best platforms that can then be implemented in in vitro experimental models.

The second part of this thesis focused on investigating the effect of OxySite on pancreatic cultures at low oxygen conditions. While we had previously reported on the ability of our device to mitigate apoptosis and maintain function [65], a deeper understanding on the cellular mechanism involved in this process was lacking. For this purpose, we cultured islets at high loading densities under low and standard oxygen conditions, either with or without the presence of our oxygen generator. Results demonstrated the ability of our device to mitigate activation of glycolytic pathways and maintain aerobic function in high density islet cultures. Moreover, we confirmed the downregulation of hypoxia mediated activation of pro-apoptotic proteins, as well as pro- inflammatory factors in OxySite cultures compared to low and standard oxygen controls.

Significantly, we verified that the presence of OxySite did not impair secretion of angiogenic factors, such as VEGFa, from pancreatic islets. Further, we demonstrated the ability of our OxySite device to sustain the oxygen consumption rate of the islets, under normal and high glucose levels; a parameter that has been strongly linked to islet

203

transplant outcomes [213]. Lastly, we validated the quality and effectiveness of the islets cultured under low oxygen conditions with our OxySite device in a diabetic animal model. Results from these transplants confirmed our observations that islets cultured with OxySite were highly functional, as observed by complete euglycemia achievement of animals transplanted with OxySite treated islets; in stark comparison to the high density standard oxygen controls, which remained hyperglycemic.

The next step in this thesis involved the translation of our OxySite device into an in vivo platform. For this purpose two distinct strategies were tested: (1) a completely macroporous platform; and (2) an immunoisolating device. For the macroporous scaffold system, the first challenge involved the incorporation of the two platforms

(scaffold and OxySite) into a single prototype. Various manufacturing procedures and strategies were tested to optimize the fabrication of our final prototype, without hindering the unique qualities of each system. After numerous testing, a final prototype was developed and the biocompatibility and efficacy in an animal model was evaluated.

While the first prototype presented problems in host responses, a final prototype constructed by masking the surface of our OxySite disk with two macroporous scaffolds yield more promising results. Biocompatibility and efficiency tests in vivo confirmed the capability of this improved prototype to circumvent challenges seen with the initial prototype. Moving forward, devices were rescaled to maximize the effects of islet loading and oxygen gradients in vivo. Characterization of the rescaled device confirmed its ability to generate oxygen at robust physiologically relevant levels. Significantly, testing of these platforms in an animal model revealed the capability of these prototypes to improve early graft function, when compared to control devices. Of note, a cohort of

204

recipients receiving OxySite containing devices were able to maintain euglycemia at significantly lower loading densities than initially transplanted (due to the increase in animal weight over the implantation time ). Herein, we demonstrated, for the first time, the ability of OxySite to enhance islet function in an in vivo model. This can prove significant in developing devices for extrahepatic transplantation at scales that are not only clinically relevant but minimally invasive.

Finally, OxySite disks were incorporated within hydrogels for the generation of completely immunoisolated devices for extrahepatic islet transplant. Herein, we sought to demonstrate the aptness of our OxySite platform to sustain cellular based graft function for extended periods of time within avascular systems. Agarose was chosen as our polymeric matrix, due to its ease in fabrication and various literature success in the context of pancreatic islet encapsulation [190, 192, 193]. Nevertheless, implementation of agarose with our OxySite device proved to be challenging due to structural issues, as initial attempts at encapsulating OxySite within agarose resulted in devices that easily fractured in vivo. We suspected that this instability was a consequence of the mechanical mismatch between our OxySite device and the hydrogel. Multiple agarose formulations, with different tensile strengths and at different concentrations, were implemented to increase the mechanically stability of these implants. In the end, a new fabrication procedure, agarose type, and concentration was optimized, resulting in stable, implantable constructs. Initial testing of these devices in a diabetic rat animal model revealed improved function from grafts containing an OxySite disks compared to controls; however, complete euglycemia could not be achieved. Assessments of explants demonstrated a significant foreign body reaction to the implants, irrespective

205

the presence of the OxySite disk. This was suspect as the reason for the observed graft instability. To deter the challenges observed in our rat models, we moved our platforms to the mouse scale and selected the poorly oxygenated implant site of the intraperitoneal space. Considerably, results in an allograft-like model (i.e. implanting mouse derived beta cell line) demonstrated, for the first time, the ability of OxySite to maintain cell viability and function in vivo, at levels significantly greater than controls.

Moving this to a xenograft platform, however, proved to be vexing, as an adequate critical mass to obtain enhanced efficiency in vivo remained elusive. Nevertheless, explants at selective time points exposed the positive effects of OxySite presence in vivo; in particular in maintaining islet glucose stimulated insulin responsiveness and decreasing the activation of hypoxia mediated factors.

As a final point, we sought to enhance the sustainability of oxygen release via the modification of the fabrication process of our OxySite platforms. Enhancing the temporal release from our devices to significantly greater timer periods (> 6 months) was highly desirable for avascular systems, such as those highlighted above. We implemented various hydrophilic water carriers to enhance water migration and porosity of the system. After screening several agents, we found PEG200 to be the most appropriate material for enhancing water uptake into our OxySite platforms. Characterization of the new device demonstrated significantly enhanced porosity when compared to original

OxySite disks. Moreover, increased water uptake, and polymer swelling was observed in these new prototype matrices. Further, oxygen generation at later time points was superior. Nevertheless, testing of these platform in an in vitro culture model bore negative and unexpected results on beta cell viability. It is suspected that the enhanced

206

porosity facilitated the escape and accumulation of toxic byproducts, with calcium decomposition leading to the accumulation of reactive agents that imparted cytotoxicity.

Nevertheless, results herein presented established a basis for OxySite modification that if carefully tested can be implemented to significantly augment oxygen delivery profiles.

Future Work

The overall aim of this work was to optimize and characterize an in situ oxygen generating device and translate this into an in vivo model in order to develop a superior platform for extrahepatic islet transplantation to cure type I diabetes. The results presented herein meet the majority of the goals set up at the beginning of this project; mainly, the optimization of an oxygen generating platform, the establishment of a mathematical model to predict release, and the translation into two distinct but equally challenging platforms for islet transplantation. Nevertheless, there is still room for the improvement and expansion of the results discussed within this thesis.

The pliability of our material based on variations in loading density and radial dimension has been explored; however, devices of different sizes and shapes have not been evaluated. Developing protocols for the fabrication of OxySite matrices of different forms greatly increases our ability to translate this devices to an innumerable number of platforms with a wide range of applications. Moreover, while a highly accurate mathematical model was developed to fit kinetics of release observed from our devices, further testing should be implemented to validate the robustness of the model.

Specifically, platforms at different calcium peroxide loading densities should be evaluated in order to prove our hypothesis that calcium dissolution is not a limiting factor in the release of oxygen from the system. Further, testing platforms of different thicknesses would be important in validating the assumption that radial release is a

207

trivial contribution to oxygen kinetics. Proving this would have the important consequence of significantly reducing the size of our oxygen platforms; a quality with great implications when scaling these devices to humans. Along this lines, the development of computational models that estimate not only the effect of environmental oxygen conditions on critical oxygen tensions, but that can also predict insulin responsiveness based on oxygen and glucose concentration can prove highly powerful in guiding and fast tracking our experimental research.

Of particular interest for this research is the development of islet delivery devices with our OxySite disks for extrahepatic islet transplantation. Herein, we developed and evidenced the impact that oxygen delivery in situ has on improving function and viability in vivo; however, complete efficacy was not obtained for the tested platforms. For the macroporous devices, platforms that can support not only islet function early on but for long-term should be developed. This can be achieved via the incorporation of enhanced platforms (i.e. smaller size and/or heightened porosity) that can fully support the require oxygen demands in vivo. Likewise, strategies to validate the ability of our oxygen platforms to generate oxygen in vivo should be explored. As oxygen tensions, fluid concentrations, and flow are vastly different in vivo than in our idealized chambers for oxygen measurement, quantification of in vivo oxygen levels can provide immense support in fabricating prototypes with desirable oxygen profiles. Further, consideration of superior polymers for encapsulation should be sought. Agarose provided a facile platform for proof of concept; however, it is a basic polymer lacking the bioactivity of

“smarter” polymers developed in recent years. Hence, in moving this platform forward, active polymers that can be modified to mitigate immune responses or deliver pro-

208

angiogenic factors and/or immunomodulating agents should be implemented in

combination with our OxySite material. Our current funding in collaboration with Dr.

Garcia at Georgia tech is moving in this direction. For this purpose, agarose hydrogels will be replaced with a functionalized PEG-malemide within a double encapsulation strategy that provides ECM cues to reduce islet “anoikis” and pro-angiogenic factors to hasten the vascularization process surrounding the graft. I look forward to seeing the results obtained from implementation of this “intelligent” delivery system.

Finally, I believe there is great potential in seeking strategies to enhance the long-term oxygen delivery from our OxySite devices. Despite the great improvement seen with the implementation of PEG200 on material properties, we were not able to translate this to upgraded cell sustainability in vitro. Nevertheless, we were able to identify some of the challenges that could be instigating this undesirable responses.

This knowledge can be implemented in the future to fabricate superior devices able to provide sustainable long-term oxygenation in vivo.

209

APPENDIX A LOADING DENSITY CALCULATION FOR COMSOL MODELING

For ease in computation time the COMSOL modeling analyses of in vitro cultures of the OxySite material at different oxygen tensions were performed in 2D instead of 3D.

One of the main issues in decreasing the complexity of the modelling is calculating an appropriate loading density that is still representative of the whole culture volume.

For our purpose a characteristic subdomain of islet loading density for the 2D cultures was obtained by assuming a homogeneous distribution of islets within the surface of the insert (Figure A-1 Panel A). Islet IEQ loading density was converted to area (normalized to an islet of 150 µm in diameter) as follows:

(75 10 ) (A-1) 1500 ( ) = 0.27 1 −6 2 𝜋𝜋 𝑥𝑥 𝐼𝐼𝐼𝐼𝐼𝐼 ∙ Moreover, a percent𝐼𝐼𝐼𝐼𝐼𝐼 area of islets to insert was calculated for the total loading density versus the surface area of the insert.

0.27 (A-2) % = = 0.23 1.13 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 A cross section from the total surface was extracted and defined as a rectangle of the same length as the diameter of the insert and the same height of an islet of 150

µm (Figure A-1 Panel B). The total area of this cross section was then calculated and multiplied by the %area occupied by the islets to obtain an area corresponding to the islet tissue within the cross section. Finally, an islet number was obtained by diving the area occupied by the islets by the area of an islet of 150 µm, to obtain a final islet number normalized to this size.

= (0.015 1.2) = 0.018 (A-3) 2 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 ∙ 𝑐𝑐𝑐𝑐

210

0.018 23% = 0.00414 (A-4) 2 0.00414∙ 𝑐𝑐𝑚𝑚 (A-5) = 23.4 0.000177 2 𝑐𝑐𝑚𝑚 2 As oxygen diffusion𝑐𝑐𝑚𝑚 limitations worsen in proportion to islet size, a more realistic

model was built by representing islet density based on the different islet size

distributions. Data collected from isolations performed in our lab demonstrate a

consistent distribution of 51% of the islets between 50-100 µm range and 49% between

100-250 µm. Based on this we can represent the final islet number, as 10 islets of 100

µm and 5 islets of 250 µm, totaling the same IEQ. Same calculations were performed for the free islet group using a lower islet loading density consisted to those currently implemented in the literature for islet culture.

211

Figure A-1. A) Schematic representation of homogeneous distribution of islets within surface area of insert. B) Representative cross section of insert.

212

APPENDIX B BIOCOMPATIBILITY OF SCAFFOLD OXYSITE COMPOSITE MATERIAL

Application of the porous scaffold as a biomaterial for extrahepatic islet transplant had been previously demonstrated in our lab [113]. Yet, the combination of the two platforms, scaffold and OxySite, had never been evaluated in vivo. First composite material fabricated consisted of one scaffold with an OxySite disk attached to the bottom of the scaffold via a PDMS layer. Material characterization via SEM demonstrated no visual differences in the porosity of the macroporous scaffold due to the addition of the OxySite disk. Moreover, no changes were observed over the PDMS layer used to combine both platforms, even at the interface between the two materials

(Figure B-1).

In vivo behavior of this material was evaluated in a biocompatibility study both in the omental pouch and skin of non-diabetic female Lewis rats. In order to compare to previously reported data on the scaffold, composite materials of the same diameter (10 mm) were fabricated. Composite material was removed after acute (7 days) and chronic time points (30 days). Macroscopic examination of the implantation site revealed no discernible inflammatory tissue responses to the material in the OP site. Conversely, in some of the skin transplants hematomas were found in the skin pocket created to place the composite material, in particular at acute time points. We hypothesize that failure to fully cauterize any capillary bleeding that occur during surgery in addition to the constant mechanical friction of the material under the skin could lead to the formation and exacerbation of this response. Nevertheless, histological analysis of the macroporous scaffold depicts favorable host integration with the scaffold material, with robust collagen deposition and cell infiltration (Figure B-2).

213

Material impact in vivo was also evaluated in an efficacy experiment. For this purpose composite materials were loaded with the same IEQ, 1800, implemented in previous successful scaffold transplants [113] and transplanted into the OP site of diabetic female Lewis recipients. Figure B-3 panel A demonstrates the inability of the platform to restore normal non-fasting blood glucose in recipients. This observation was true for both composite platforms implemented, Scaffold-PDMS blank and scaffold-

OxySite. These results deviate entirely from those previously published on our scaffold material. Moreover, they hint at a possible detrimental interaction of the solid disk with the tissue and not necessarily the presence of peroxide as an active oxygen generating material. Pictures of the explanted graft (Figure B-3 panel C) demonstrate a significant inflammatory reaction to the solid portion of the composite material, further confirming our theory. Finally, histological analysis of the explanted grafts demonstrate good infiltration and host ECM deposition on the scaffold portion of the graft while the solid

PDMS was completely encapsulated by a fibrotic cap (Figure B-4). Again, this assessment was true for both platforms.

The response to our composite material while unexpected, due to the high biocompatibility of PDMS, can be explained by the material surface roughness. The response of tissue to implanted material is highly dependent on surface topography.

Texture and/or porous implants have been shown to provide a greater surface area for cellular infiltration and integration [214], while smooth surface implants have been demonstrated to promote fibrosis [174]. Nevertheless, these results demonstrated our ability to fabricate composite materials based on the unification of two platforms developed in our laboratory. Finally, the unexpected material response in vivo lead us to

214

the development of a new optimized composite platform for the evaluation of our

OxySite material as a bridge to revascularization which is addressed in Chapter 5.

215

Figure B-1. SEM images of surface and cross section of composite scaffold material. No direct effects on pore structure was observed by the addition of an OxySite disk to the bottom of the scaffold.

216

Figure B-2. Histological analysis of composite scaffolds transplanted in the omental pouch and skin of normal Lewis female rats after 30 days post-transplant. Good host integration and collagen deposition was observed on the scaffold surface. Scale 100 µm.

217

Figure B-3. In vivo assessment of scaffold-OxySite composite material. Scaffolds were loaded with 1800 IEQ and transplanted in the omental pouch of diabetic Lewis female recipients. A) Non fasting blood glucose of transplanted animals demonstrates inability of composite material to restore normoglycemia. B) No difference was observed in the body weight of the recipients. C) Images of graft during explant. An adverse foreign body reaction was observed to the solid part of the graft irrespective of the presence of peroxide.

218

Figure B-4. Histological analysis of explanted graft 35 d post-transplant. Good host integration was observed on the scaffold side (top). However, solid disk (bottom) was completely surrounded by a fibrotic band in both control PDMS and OxySite group.

219

APPENDIX C LOADING TITRATION OF AGAROSE XENOGRAFT TRANSPLANT MODEL

Islet mass is a critical factor in the outcome of islet transplantation for the treatment of diabetes [215]. This is particularly true for intraperitoneal transplants due to the poor vascularization of the graft, and the less efficient release of insulin [216]. Thus, in order to translate our immunoisolating device into a mouse xenograft model we had to perform loading titration experiments to determine a loading density in which the effects of oxygenation could be seen early in the post-transplant period.

Figure C-1 summarizes all loading densities evaluated and their respective volume/volume percentage within our grafts. Higher loading densities in line with those reported in the literature for macroencapsulation devices [216, 217] were implemented at first. As seen in Figure C-2 panel A, normoglycemic non fasting blood glucose levels were observed for both controls and treated group. Based on these results lower loading densities were implemented (900, 700, and 600 IEQ), with the idea of finding a critical loading mass that will render the grafts inadequate unless a favorable oxygen environment was provided (Figure C-2 Panel B and C). Nevertheless, albeit multiple efforts, at all packing densities no difference in mean reversal time to normoglycemia was observed between the control groups and those with the OxySite material. Thus, despite the decrease oxygen levels experienced by the graft in the IP space, remaining beta cell mass was adequate to regulate blood glucose levels in the transplanted mice.

Moreover, destabilization of the grafts around day 10 post-implantation was observed in the majority of the transplants. This is in stark contrast to our allogeneic implants were no such destabilization was observed; hinting perhaps at a xenogeneic mediated reaction to the graft.

220

Overall, these experiments provided us with valuable insight as to in vivo responses to our biomaterial in a xenograft model. Moreover, one of the platforms herein developed was implemented in Chapter 7 as a model to assess the effects of in vitro supplemental oxygenation on graft viability and function ex vivo.

221

Figure C-1. Summary of all islet loading implemented (yellow inset) and the corresponding vol/vol loading density in the graft (red inset).

222

Figure C-2. Representative non fasting blood glucose (top panel) and body weight (bottom panel) of all the different loading densities tested in vivo. Images are representative of one experiment for each loading density however multiple transplants were performed to confirm results.

223

APPENDIX D LOADING TITRATION OF AGAROSE ALLOGRAFT TRANSPLANT MODEL

Based on results obtained in Chapter 7 and appendix C we decided to test our

OxySite material within an allograft model in order to decrease the influence of immunological responses on graft efficacy. While it is within the capabilities of our lab to implement mouse islets as the cell source for our transplants the limited yield obtained from mouse isolations would greatly hinder our ability to assess this model effectively.

Thus, for this purpose we decided to implement MIN6, a beta cell line, as our cell source. MIN6, are a transgenic cell line developed from C57/B6 mice [218]. MIN6 have been implemented widely as an alternative cell source in the diabetes community due to their ability to deliver insulin in a glucose responsive manner, albeit limited compared to islets [218]. Herein, we present the results of a titration experiment of MIN6 cells within our grafts, with the idea of determining a loading density at which MIN6 cells would mature within the graft, given a beneficial oxygen environment.

All loading densities explored as well as their equivalent vol/vol packing densities within our device are represented in Figure D-1. Initial experiments were performed at higher loading densities (3.2x106 cells) that those needed for primary cell xenograft models to perform in the IP space (see Appendix C). A higher loading density was chosen due to the diminished glucose responsiveness observed in MIN6 cells. Results demonstrated that at this packing density both control and treated groups quickly revert animals non-fasting blood glucose to normoglycemic levels, as observed in Figure D-2 panel A. Likewise, hyperglycemia reversal was stable for both groups at all time points tested. Furthermore, no difference in body weight was observed between the two groups through the duration of the experiment.

224

Follow up experiments were performed at a reduced loading density (1.5x106 cells). The idea being that delaying reversal times would provide us with a sufficient interval to start observing differences between both groups. While, there is a leaning towards improved blood glucose regulation in the OxySite containing group compared to controls, complete normoglycemia was not achieved in either group (Figure D-2 panel B). This results suggests a determent of MIN6 growth in vivo due to a suboptimal loading density. Exposure of MIN6 to stressful and highly hyperglycemic environments can hinder cell division and lead to cell death, even in oxygen rich environments [219].

Hence, moving forward experiments were performed on a loading density in a range between those tested herein. Transplant results for those studies are discussed at length in Chapter 6.

225

Figure D-1. Summary of all MIN6 loading implemented (yellow inset) and the corresponding vol/vol loading density in the graft (red inset). MIN6 loading was normalized to an IEQ using the conversion 1EQ=2000 cells.

226

Figure D-2. Representative non fasting blood glucose (top panel) and body weight (bottom panel) of all the different loading densities tested in vivo. Images are representative of one experiment for each loading density however multiple transplants were performed to confirm results.

227

APPENDIX E OXYGEN GENERATING BIOMATERIAL SUPPORTS SCHWANN CELL VIABILITY IN A MODEL OF SPINAL CORD INJURY

A study based on the use of our OxySite material to support Schwann cell (SC) viability in a model of spinal cord injury was investigated. The ability of the SC to substantially influence regeneration in both the peripheral and the central nervous system derives in part from their remarkable ability to produce a variety of trophic factors, as well as from expressing a number of cell adhesion molecules known to influence neurite growth, on their surfaces. Studies in the past decade have shown that

SCs are a source of a remarkable number of the known neurotrophic molecules [220].

Moreover, studies by Bunge et al. have demonstrated elevation of cAMP normal after

SC transplantation. The presence of cAMP can help overcome the inhibitory aspect of the injured spinal cord milieu by influencing growth cone turning behavior, affecting guidance towards chemoattractive cues and enabling nerve cells from mature animals to extend fibers across inhibitory substrates, such as myelin or myelin-associated glycoprotein [221]. Nevertheless, a poor survival rate has been reported for these cells when transplanted into the epi-center cavity, although the endogenous SC will fill up the cavity [222]. Griffiths and McCulloch [223] showed progressive axonal changes and the development of necrotic zones during the first few days after injury. Thus, by 24 to 48 hours after major trauma, the injury site is necrotic, especially the central zone previously occupied by hemorrhage [224]. Several days later, the hemorrhagic zone shows cavitation and the adjacent areas exhibit patchy necrosis, often with sharply defined margins. These progressive changes, consisting of cavitation and coagulative necrosis at the injury site and in adjacent areas [225]. Thus, it is our hypothesis that the

228

presence of the OxySite material could be beneficial in increasing cell viability by providing a temporary increase in the surrounding oxygen concentration.

After 1 week of performing the contusion collagen rolls containing Schwann cells with or without the oxygen generator were transplanted into Harlan rats. 4 weeks post implantation the constructs were perfused and explanted and histological assessments were performed (Figure E-1). The spinal cord appeared to have structural continuity with the implant in the groups containing collagen only or collagen plus the oxygen generator. Myelin was constant across the implant-spinal cord in the implant group B and D at 4 weeks postoperatively. The central portion of the implant was occupied by a cystic cavity in the cell-implanted groups A and C at 4 weeks postoperatively. Sever scar formation was not seen at the implant-spinal cord interface in any of the groups. In longitudinal sections of both proximal and distal segments of a SC graft attached to the host spinal cord, numerous immunostained (GFAP positive) axons were seen to exit both the rostral and caudal host spinal cord and enter the graft (Figure E-1). However, cysts still remained within the collagen + PDMS and collagen + O2 groups. The in- growing axons were mostly associated with SCs as evidenced by double labelling of tissue sections with GFP and hoeschet (Figure E-2). It is evident that the cell concentration is increased in the group with the oxygen generator as seen by the increase in the hoescht stained in this group (Figure E-2 Panel C). Moreover, there appears to be no difference between the collagen control and the collagen with the

PDMS blank attesting to the biocompatibility of the biomaterial.

Greater magnification of the immunohistochemical slides showed a homogeneous distributions of cells as seen by the hoescht dye on the oxygen

229

containing scaffold compared to its sham controlled (Figure E- 3 panel A and B).

Moreover greater distribution of GFP positive cells and S100 protein is seen on the oxygen group with distinct distributions of the neurofillament that aligns with the cellular content of the graft as seen by Figure E-3.

Metabolic assay results however are not in agreement with the histological results shown above. MTT assay shows that there is a decrease in cellular viability in the hypoxic controlled compared to its normoxic counterpart. However, the metabolic viability in the construct is not rescued by the introduction of the oxygen generator as seen in Figure E-4. Nevertheless, this unexpected drop in viability could be a result of an interaction of the calcium peroxide in the oxygen generator with the assay dye, since the rods were not removed for the preparation of this assay.

Overall, these data provide evidence that introduction of an oxygen generator within a Schwann cell 3 dimensional scaffold can be beneficial in enhancing cell survival and myelination. Future experiments should be performed to assess the rate of oxygenation provided by this biomaterial as well as assays to measure nerve formation in the graft. More mechanistic studies regarding the effect of oxygen on these cells should also be performed, to assess cytokine and growth factor production in this oxygenating environment since this is one of the main functions of Schwann cells in tissue repair.

230

Figure E-1. Luxol fast blue (LFB) staining shows the morphology of myelin in the graft site. (A) sham experiment no material, (B) collagen only, (C), collagen+PDMS and (D) Collagen+O2

231

Figure E-2. Immunofluorescence of Schwann cell GFP positive cells, with nuclear staining and a neurofilament marker. (A) Collagen only, (B) Collagen +PDMS and (C) collagen + O2.

232

Figure E-3. Immunofluorescence of Schwann cell GFP positive cells, with nuclear staining and a neurofilament marker and a neurofilament protein Top (A) Collagen +PDMS and (B) collagen + O2. Bottom includes a GFP marker S- 100 and nuclear staining, (c) Collagen +PDMS and (B) collagen + O2.

233

Figure E-4. Metabolic activity of 3D scaffolds containing embedded schwann cells in the presence of the oxygen generator culture under normoxic or hypoxic conditions.

234

APPENDIX F OPTIMIZATION OF AGAROSE CONCENTRATION AND CASTING PROCESS FOR THE FABRICATION OF IMMUNOISOLATING DEVICES

One of the first challenges we encountered in fabricating our immunoisolating devices was the choice of polymer to implement. There are numerous hydrogel polymers in the literature proven effective for cellular encapsulation [157, 164, 212,

226]. Nevertheless, agarose was chosen due to the extensive literature as an encapsulating polymer for beta cells, as well as the facile fabrication, shown cytocompatibility, and casting process [190, 192, 193]. First method of fabrication for our agarose immunosiolated devices consisted of a two-step fabrication in which a cell agarose mixture would be poured into a casting mold, subsequently an OxySite disk was placed in the center; and a second layer of agarose-cell mix was added on top.

Post gelation constructs would be removed from mold and implemented either for transplantation. Implantation of these devices in vivo yielded no difference in glycemic levels for animals transplanted either with an OxySite disk or a control blank. Upon retrieval it became evident that graft failure was due to the absence of an intact graft.

It was suspected that the difference in strength between the PDMS disk and the agarose, was causing a fracture between the two agarose layers post fabrication. With this hypothesis in mind we set to test different agarose, at various tensile strengths and concentrations to find a formulation that could improve the stability of our devices. A summary of all the polymers tested, as well as their respective tensile strengths and concentrations are presented in Figure F-1. Early on it became evident that concentration higher than 3% could not be implemented as it would significantly increasing the gelation temperature of the gel resulting in the polymer solidifying in the

235

tube prior to casting. This was true regardless of the tensile strength or original gelation temperature of the hydrogel polymer. Initial testing of the agarose hydrogels at the different types and concentrations was carried out as explained above and placed in the

IP space of diabetic mice. As seen in Figure F-2, all of these formulations failed to restore normoglycemia in the mice. Upon explantation again all of the devices were either fragmented in pieces; or had separated between a top and bottom layer leaving the PDMS disk free to float within the IP space.

This new evidence of instability within our devices even at higher loading densities lead us to hypothesize that the two-step casting method was perhaps hindering the hydrogel mixture to gel as a whole. Thus, in moving forward the actual fabrication procedure was modified to allow for a one step pouring of the hydrogel mixture. Custom made molds were fabricated out of PDMS at the desired geometry. A solid PDMS layer mold was made separately and attached to the cylindrical mold via a

PDMS layer. Subsequently, pins were placed at the geometrical center of each well within the mold to allow for suspension of the PDMS disk prior to addition of the agarose-cell mixture. Moreover, molds were placed within the top plate of a petri dish and placed in a hot plate set at 37oC to further prevent gelation prior to casting. From all the hydrogels tested a low gelation temperature agarose, at a concentration of 2% w/w was determined to provide the best constructs, with high reproducibility and no issues associated with gelation prior to casting. Constructs fabricated in this manner were transplanted into naïve mice to test for mechanical stability in vivo. As observed in

Figure F-3, constructs fabricated in this manner remain fully intact upon implantation for

236

extended periods of time. This optimized fabrication protocol was implemented to fabricated all the immunoisolated devices described in Chapter 6.

237

Figure F-1. Summary of agarose polymers implemented in optimization of fabrication protocol with respective tensile strengths and gelling temperature.

238

Figure F-2. Non-fasting blood glucose levels for mice receiving agarose immunoisolated devices from a subset of agarose polymers tested. Animals failed to revert to a normoglycemic state, and upon implantation device fracture was observed in all samples.

239

Figure F-3. Images illustrating improved agarose immunoisolated device fabricated using optimized manufacturing protocol. A) Agarose construct prior to transplant. B) Construct being placed in IP space of naïve mice. C) Construct upon removal after 30 d post transplantation.

240

REFERENCES

[1] F.B. Barton, M.R. Rickels, R. Alejandro, B.J. Hering, S. Wease, B. Naziruddin, J. Oberholzer, J.S. Odorico, M.R. Garfinkel, M. Levy, F. Pattou, T. Berney, A. Secchi, S. Messinger, P.A. Senior, P. Maffi, A. Posselt, P.G. Stock, D.B. Kaufman, X. Luo, F. Kandeel, E. Cagliero, N.A. Turgeon, P. Witkowski, A. Naji, P.J. O'Connell, C. Greenbaum, Y.C. Kudva, K.L. Brayman, M.J. Aull, C. Larsen, T.W. Kay, L.A. Fernandez, M.C. Vantyghem, M. Bellin, A.M. Shapiro, Improvement in outcomes of clinical islet transplantation: 1999-2010, Diabetes care 35(7) (2012) 1436-45.

[2] N. Zhang, D. Su, S. Qu, T. Tse, R. Bottino, A.N. Balamurugan, J. Xu, J.S. Bromberg, H.H. Dong, Sirolimus is associated with reduced islet engraftment and impaired beta- cell function, Diabetes 55(9) (2006) 2429-36.

[3] M. McCall, R. Pawlick, T. Kin, A.M.J. Shapiro, Anakinra Potentiates the Protective Effects of Etanercept in Transplantation of Marginal Mass Human Islets in Immunodeficient Mice, American Journal of Transplantation 12(2) (2012) 322-329.

[4] M. Koulmanda, M. Bhasin, Z. Fan, D. Hanidziar, N. Goel, P. Putheti, B. Movahedi, T.A. Libermann, T.B. Strom, Alpha 1-antitrypsin reduces inflammation and enhances mouse pancreatic islet transplant survival, Proceedings of the National Academy of Sciences 109(38) (2012) 15443-15448.

[5] S. Merani, C. Toso, J. Emamaullee, A.M. Shapiro, Optimal implantation site for pancreatic islet transplantation, Br J Surg 95(12) (2008) 1449-61.

[6] G. Christoffersson, J. Henriksnäs, L. Johansson, C. Rolny, H. Ahlström, J. Caballero- Corbalan, R. Segersvärd, J. Permert, O. Korsgren, P.-O. Carlsson, M. Phillipson, Clinical and Experimental Pancreatic Islet Transplantation to Striated Muscle: Establishment of a Vascular System Similar to That in Native Islets, Diabetes 59(10) (2010) 2569-2578.

[7] D.W. Scharp, C.J. Swanson, B.J. Olack, P.P. Latta, O.D. Hegre, E.J. Doherty, F.T. Gentile, K.S. Flavin, M.F. Ansara, P.E. Lacy, Protection of encapsulated human islets implanted without immunosuppression in patients with type I or type II diabetes and in nondiabetic control subjects, Diabetes 43(9) (1994) 1167-70.

[8] A. Pileggi, R.D. Molano, C. Ricordi, E. Zahr, J. Collins, R. Valdes, L. Inverardi, Reversal of diabetes by pancreatic islet transplantation into a subcutaneous, neovascularized device, Transplantation 81(9) (2006) 1318-24.

[9] A.J. Hussey, M. Winardi, X.L. Han, G.P. Thomas, A.J. Penington, W.A. Morrison, K.R. Knight, S.J. Feeney, Seeding of pancreatic islets into prevascularized tissue engineering chambers, Tissue engineering. Part A 15(12) (2009) 3823-33.

241

[10] D.M. Berman, J.J. O'Neil, L.C. Coffey, P.C. Chaffanjon, N.M. Kenyon, P. Ruiz, Jr., A. Pileggi, C. Ricordi, N.S. Kenyon, Long-term survival of nonhuman primate islets implanted in an omental pouch on a biodegradable scaffold, Am J Transplant 9(1) (2009) 91-104.

[11] T.R. Hefty, C.S. Kuhr, K.T. Chong, D.G. Guinee, W. Wang, J.A. Reems, C.J. Greenbaum, Omental Roll-Up: A Technique for Islet Engraftment in a Large Animal Model, Journal of Surgical Research 161(1) (2010) 134-138.

[12] S. Shah, E. Lowery, R.K. Braun, A. Martin, N. Huang, M. Medina, P. Sethupathi, Y. Seki, M. Takami, K. Byrne, C. Wigfield, R.B. Love, M. Iwashima, Cellular Basis of Tissue Regeneration by Omentum, PLoS One 7(6) (2012) e38368.

[13] Z. Kakabadze, K. Shanava, C. Ricordi, A.M. Shapiro, S. Gupta, E. Berishvili, An isolated venous sac as a novel site for cell therapy in diabetes mellitus, Transplantation 94(4) (2012) 319-24.

[14] J. Komori, L. Boone, A. DeWard, T. Hoppo, E. Lagasse, The mouse lymph node as an ectopic transplantation site for multiple tissues, Nat Biotech 30(10) (2012) 976-983.

[15] V. Perez, A. Caicedo, D. Berman, E. Arrieta, M. Abdulreda, R. Rodriguez-Diaz, A. Pileggi, E. Hernandez, S. Dubovy, J. Parel, C. Ricordi, N. Kenyon, N. Kenyon, P. Berggren, The anterior chamber of the eye as a clinical transplantation site for the treatment of diabetes: a study in a baboon model of diabetes, Diabetologia 54(5) (2011) 1121-1126.

[16] J.C. Stendahl, D.B. Kaufman, S.I. Stupp, Extracellular matrix in pancreatic islets: relevance to scaffold design and transplantation, Cell transplantation 18(1) (2009) 1-12.

[17] P.P. Tak, G.S. Firestein, NF-κB: a key role in inflammatory diseases, The Journal of Clinical Investigation 107(1) (2001) 7-11.

[18] G. Nikolova, N. Jabs, I. Konstantinova, A. Domogatskaya, K. Tryggvason, L. Sorokin, R. Fässler, G. Gu, H.-P. Gerber, N. Ferrara, D.A. Melton, E. Lammert, The Vascular Basement Membrane: A Niche for Insulin Gene Expression and ² Cell Proliferation, Developmental cell 10(3) (2006) 397-405.

[19] F. Thomas, J. Wu, J.L. Contreras, C. Smyth, G. Bilbao, J. He, J. Thomas, A tripartite anoikis-like mechanism causes early isolated islet apoptosis, Surgery 130(2) (2001) 333-338.

[20] D. Borg, E. Bonifacio, The Use of Biomaterials in Islet Transplantation, Current Diabetes Reports 11(5) (2011) 434-444.

242

[21] R.F. Gibly, X. Zhang, M.L. Graham, B.J. Hering, D.B. Kaufman, W.L. Lowe, Jr., L.D. Shea, Extrahepatic islet transplantation with microporous polymer scaffolds in syngeneic mouse and allogeneic porcine models, Biomaterials 32(36) (2011) 9677-84.

[22] D.M. Berman, J.J. O'Neil, L.C.K. Coffey, P.C.J. Chaffanjon, N.M. Kenyon, P. Ruiz, A. Pileggi, C. Ricordi, N.S. Kenyon, Long-Term Survival of Nonhuman Primate Islets Implanted in an Omental Pouch on a Biodegradable Scaffold, American Journal of Transplantation 9(1) (2009) 91-104.

[23] T. Kin, J.J. O'Neil, R. Pawlick, G.S. Korbutt, A.M. Shapiro, J.R. Lakey, The use of an approved biodegradable polymer scaffold as a solid support system for improvement of islet engraftment, Artificial organs 32(12) (2008) 990-3.

[24] E. Pedraza, A. Brady, C. Fraker, R. Molano, S. Sukert, N. Kenyon, A. Pileggi, C. Ricordi, C. Stabler, Macro-porous PDMS Scaffolds for Extrahepatic Islet Transplantation, Cell transplantation in press (2012).

[25] N. Kenyon, E. Pedraza, D.M. Berman, M. Willman, N. Kenyon, E. Margolles-Clark, P. Buchwald, P. Latta, C. Ricordi, C. Stabler, Scaffolds within an Omental Pouch Site Support Long Term Survival of Allogeneic, Nonhuman Primate Islets, Amer J Transplant 11(S2) (2011) 117.

[26] J. Daoud, M. Petropavlovskaia, L. Rosenberg, M. Tabrizian, The effect of extracellular matrix components on the preservation of human islet function in vitro, Biomaterials 31(7) (2010) 1676-82.

[27] J.T. Daoud, M.S. Petropavlovskaia, J.M. Patapas, C.E. Degrandpre, R.W. Diraddo, L. Rosenberg, M. Tabrizian, Long-term in vitro human pancreatic islet culture using three-dimensional microfabricated scaffolds, Biomaterials 32(6) (2011) 1536-42.

[28] L.M. Weber, K.S. Anseth, Hydrogel encapsulation environments functionalized with extracellular matrix interactions increase islet insulin secretion, Matrix biology : journal of the International Society for Matrix Biology 27(8) (2008) 667-73.

[29] L.M. Weber, K.N. Hayda, K.S. Anseth, Cell-matrix interactions improve beta-cell survival and insulin secretion in three-dimensional culture, Tissue engineering. Part A 14(12) (2008) 1959-68.

[30] D. Salvay, C. Rives, X. Zhang, F. Chen, D.B. Kaufman, W.L. Lowe, Jr., L.D. Shea, Extracellular matrix protein-coated scaffolds promote the reversal of diabetes after extrahepatic islet transplantation., Transplantation 85(10) (2008) 1456-1464.

[31] C.-C. Lin, K.S. Anseth, Cell–cell communication mimicry with poly(ethylene glycol) hydrogels for enhancing β-cell function, Proceedings of the National Academy of Sciences 108(16) (2011) 6380-6385.

243

[32] D.J. Lim, S.V. Antipenko, J.M. Anderson, K.F. Jaimes, L. Viera, B.R. Stephen, S.M. Bryant, B.D. Yancey, K.J. Hughes, W. Cui, J.A. Thompson, J.A. Corbett, H.W. Jun, Enhanced rat islet function and survival in vitro using a biomimetic self-assembled nanomatrix gel, Tissue engineering. Part A 17(3-4) (2011) 399-406.

[33] T. Kheradmand, S. Wang, R.F. Gibly, X. Zhang, S. Holland, J. Tasch, J.G. Graham, D.B. Kaufman, S.D. Miller, L.D. Shea, X. Luo, Permanent protection of PLG scaffold transplanted allogeneic islet grafts in diabetic mice treated with ECDI-fixed donor splenocyte infusions, Biomaterials 32(20) (2011) 4517-4524.

[34] J. Su, B.-H. Hu, W.L. Lowe Jr, D.B. Kaufman, P.B. Messersmith, Anti-inflammatory peptide-functionalized hydrogels for insulin-secreting cell encapsulation, Biomaterials 31(2) (2010) 308-314.

[35] N. Lifson, C.V. Lassa, P.K. Dixit, Relation between blood flow and morphology in islet organ of rat pancreas, American Journal of Physiology - Endocrinology And Metabolism 249(1) (1985) E43-E48.

[36] K.E. Dionne, C.K. Colton, M.L. Yarmush, Effect of hypoxia on insulin secretion by isolated rat and canine islets of Langerhans, Diabetes 42(1) (1993) 12-21.

[37] W. Moritz, F. Meier, D.M. Stroka, M. Giuliani, P. Kugelmeier, P.C. Nett, R. Lehmann, D. Candinas, M. Gassmann, M. Weber, Apoptosis in hypoxic human pancreatic islets correlates with HIF-1alpha expression, Faseb J 16(7) (2002) 745-7.

[38] K. Cheng, K. Ho, R. Stokes, C. Scott, S.M. Lau, W.J. Hawthorne, P.J. O'Connell, T. Loudovaris, T.W. Kay, R.N. Kulkarni, T. Okada, X.L. Wang, S.H. Yim, Y. Shah, S.T. Grey, A.V. Biankin, J.G. Kench, D.R. Laybutt, F.J. Gonzalez, C.R. Kahn, J.E. Gunton, Hypoxia-inducible factor-1alpha regulates beta cell function in mouse and human islets, J Clin Invest 120(6) (2010) 2171-83.

[39] J. Cantley, S.T. Grey, P.H. Maxwell, D.J. Withers, The hypoxia response pathway and beta-cell function, Diabetes Obes Metab 12 Suppl 2 (2010) 159-67.

[40] S. Dubois, A.M. Madec, A. Mesnier, M. Armanet, K. Chikh, T. Berney, C. Thivolet, Glucose inhibits angiogenesis of isolated human pancreatic islets, J Mol Endocrinol 45(2) (2010) 99-105.

[41] B. Ludwig, B. Zimerman, A. Steffen, K. Yavriants, D. Azarov, A. Reichel, P. Vardi, T. German, N. Shabtay, A. Rotem, Y. Evron, T. Neufeld, S. Mimon, S. Ludwig, M.D. Brendel, S.R. Bornstein, U. Barkai, A novel device for islet transplantation providing immune protection and oxygen supply, Horm Metab Res 42(13) (2010) 918-22.

244

[42] B. Ludwig, A. Rotem, J. Schmid, G.C. Weir, C.K. Colton, M.D. Brendel, T. Neufeld, N.L. Block, K. Yavriyants, A. Steffen, S. Ludwig, T. Chavakis, A. Reichel, D. Azarov, B. Zimermann, S. Maimon, M. Balyura, T. Rozenshtein, N. Shabtay, P. Vardi, K. Bloch, P. de Vos, A.V. Schally, S.R. Bornstein, U. Barkai, Improvement of islet function in a bioartificial pancreas by enhanced oxygen supply and growth hormone releasing hormone agonist, Proceedings of the National Academy of Sciences 109(13) (2012) 5022-5027.

[43] E. Pedraza, M.M. Coronel, C.A. Fraker, C. Ricordi, C.L. Stabler, Preventing hypoxia-induced cell death in beta cells and islets via hydrolytically activated, oxygen- generating biomaterials, Proceedings of the National Academy of Sciences 109(11) (2012) 4245-4250.

[44] M. Shimoda, S. Chen, H. Noguchi, S. Matsumoto, P. Grayburn, In vivo non-viral gene delivery of human vascular endothelial growth factor improves revascularisation and restoration of euglycaemia after human islet transplantation into mouse liver, Diabetologia 53(8) (2010) 1669-1679.

[45] N.E. Davis, L.N. Beenken-Rothkopf, A. Mirsoian, N. Kojic, D.L. Kaplan, A.E. Barron, M.J. Fontaine, Enhanced function of pancreatic islets co-encapsulated with ECM proteins and mesenchymal stromal cells in a silk hydrogel, Biomaterials 33(28) (2012) 6691-6697.

[46] T. Ito, S. Itakura, I. Todorov, J. Rawson, S. Asari, J. Shintaku, I. Nair, K. Ferreri, F. Kandeel, Y. Mullen, Mesenchymal Stem Cell and Islet Co-Transplantation Promotes Graft Revascularization and Function, Transplantation 89(12) (2010) 1438-1445 10.1097/TP.0b013e3181db09c4.

[47] Y. Ding, D. Xu, G. Feng, A. Bushell, R.J. Muschel, K.J. Wood, Mesenchymal stem cells prevent the rejection of fully allogenic islet grafts by the immunosuppressive activity of matrix metalloproteinase-2 and -9, Diabetes 58(8) (2009) 1797-806.

[48] T.Y. Yeung, K.L. Seeberger, T. Kin, A. Adesida, N. Jomha, A.M. Shapiro, G.S. Korbutt, Human mesenchymal stem cells protect human islets from pro-inflammatory cytokines, PLoS One 7(5) (2012) e38189.

[49] D.M. Berman, M.A. Willman, D. Han, G. Kleiner, N.M. Kenyon, O. Cabrera, J.A. Karl, R.W. Wiseman, D.H. O'Connor, A.M. Bartholomew, N.S. Kenyon, Mesenchymal Stem Cells Enhance Allogeneic Islet Engraftment in Nonhuman Primates, Diabetes 59(10) (2010) 2558-2568.

[50] E.A. Phelps, A.J. Garcia, Update on therapeutic vascularization strategies, Regenerative medicine 4(1) (2009) 65-80.

245

[51] M.M. Martino, F. Tortelli, M. Mochizuki, S. Traub, D. Ben-David, G.A. Kuhn, R. Muller, E. Livne, S.A. Eming, J.A. Hubbell, Engineering the growth factor microenvironment with fibronectin domains to promote wound and bone tissue healing, Sci Transl Med 3(100) (2011) 100ra89.

[52] E.A. Phelps, N. Landázuri, P.M. Thulé, W.R. Taylor, A.J. García, Bioartificial matrices for therapeutic vascularization, Proceedings of the National Academy of Sciences 107(8) (2010) 3323-3328.

[53] L.W. Chow, L.-j. Wang, D.B. Kaufman, S.I. Stupp, Self-assembling nanostructures to deliver angiogenic factors to pancreatic islets, Biomaterials 31(24) (2010) 6154-6161.

[54] J.C. Stendahl, L.J. Wang, L.W. Chow, D.B. Kaufman, S.I. Stupp, Growth factor delivery from self-assembling nanofibers to facilitate islet transplantation, Transplantation 86(3) (2008) 478-81.

[55] R.L. Carrier, M. Papadaki, M. Rupnick, F.J. Schoen, N. Bursac, R. Langer, L.E. Freed, G. Vunjak-Novakovic, Cardiac tissue engineering: cell seeding, cultivation parameters, and tissue construct characterization, Biotechnology and bioengineering 64(5) (1999) 580-589.

[56] M.C. Lewis, B.D. MacArthur, J. Malda, G. Pettet, C.P. Please, Heterogeneous proliferation within engineered cartilaginous tissue: the role of oxygen tension, Biotechnology and bioengineering 91(5) (2005) 607-615.

[57] H. Wu, E.S. Avgoustiniatos, L. Swette, S. BONNER‐WEIR, G.C. Weir, C.K. Colton, In situ electrochemical oxygen generation with an immunoisolation device, Annals of the new York Academy of Sciences 875(1) (1999) 105-125.

[58] K. Bloch, E. Papismedov, K. Yavriyants, M. Vorobeychik, S. Beer, P. Vardi, Photosynthetic oxygen generator for bioartificial pancreas, Tissue engineering 12(2) (2006) 337-344.

[59] G. Camci‐Unal, N. Alemdar, N. Annabi, A. Khademhosseini, Oxygen‐releasing biomaterials for tissue engineering, Polymer international 62(6) (2013) 843-848.

[60] J.D. McMillan, D.I. Wang, Mechanisms of Oxygen Transfer Enhancement during Submerged Cultivation in Perfluorochemical‐in‐Water Dispersionsa, Annals of the New York Academy of Sciences 589(1) (1990) 283-300.

[61] A.P. Piccionello, A. Guarcello, A. Calabrese, I. Pibiri, A. Pace, S. Buscemi, Synthesis of fluorinated oxadiazoles with gelation and oxygen storage ability, Organic & biomolecular chemistry 10(15) (2012) 3044-3052.

246

[62] M. Koulmanda, M. Bhasin, Z. Fan, D. Hanidziar, N. Goel, P. Putheti, B. Movahedi, T.A. Libermann, T.B. Strom, Alpha 1-antitrypsin reduces inflammation and enhances mouse pancreatic islet transplant survival, Proc Natl Acad Sci U S A 109(38) (2012) 15443-8.

[63] S.H. Oh, C.L. Ward, A. Atala, J.J. Yoo, B.S. Harrison, Oxygen generating scaffolds for enhancing engineered tissue survival, Biomaterials 30(5) (2009) 757-762.

[64] C.L. Stabler, B.S. Harrison, C.A. Fraker, E. Pedraza, C. Ricordi, Silicone-peroxide compositions for long-term, controlled oxygen release, Google Patents, 2015.

[65] E. Pedraza, M.M. Coronel, C.A. Fraker, C. Ricordi, C.L. Stabler, Preventing hypoxia-induced cell death in beta cells and islets via hydrolytically activated, oxygen- generating biomaterials, Proceedings of the National Academy of Sciences of the United States of America 109(11) (2012) 4245-50.

[66] J.-M. Vergnaud, Controlled drug release of oral dosage forms, CRC Press1993.

[67] H. Lapidus, N. Lordi, Drug release from compressed hydrophilic matrices, Journal of Pharmaceutical Sciences 57(8) (1968) 1292-1301.

[68] L. Shen, Z. Chen, Critical review of the impact of tortuosity on diffusion, Chemical Engineering Science 62(14) (2007) 3748-3755.

[69] J. Siepmann, N. Peppas, Modeling of drug release from delivery systems based on hydroxypropyl methylcellulose (HPMC), Advanced drug delivery reviews 48(2) (2001) 139-157.

[70] M. Cox, B. Dunn, Oxygen diffusion in poly (dimethyl siloxane) using fluorescence quenching. I. Measurement technique and analysis, Journal of Polymer Science Part A: Polymer Chemistry 24(4) (1986) 621-636.

[71] R. Vervoort, S. Cattle, Linking hydraulic conductivity and tortuosity parameters to pore space geometry and pore-size distribution, Journal of Hydrology 272(1) (2003) 36- 49.

[72] Q. Zeng, K. Li, T. Fen-Chong, P. Dangla, Analysis of pore structure, contact angle and pore entrapment of blended cement pastes from mercury porosimetry data, Cement and Concrete Composites 34(9) (2012) 1053-1060.

[73] D. Bentz, O.M. Jensen, A. Coats, F. Glasser, Influence of silica fume on diffusivity in cement-based materials: I. Experimental and computer modeling studies on cement pastes, Cement and Concrete research 30(6) (2000) 953-962.

247

[74] C.E. Salmas, G.P. Androutsopoulos, A novel pore structure tortuosity concept based on nitrogen sorption hysteresis data, Industrial & engineering chemistry research 40(2) (2001) 721-730.

[75] P. Buchwald, FEM-based oxygen consumption and cell viability models for avascular pancreatic islets, Theor Biol Med Model 6 (2009) 5.

[76] G. Di Colo, V. Carelli, E. Nannipieri, M.F. Serafini, D. Vitale, Effect of water-soluble additives on drug release from silicone rubber matrices. II. Sustained release of prednisolone from non-swelling devices, International Journal of Pharmaceutics 30(1) (1986) 1-7.

[77] V. Carelli, G. Di Colo, E. Nannipieri, M.F. Serafini, Evaluation of a silicone based matrix containing a crosslinked polyethylene glycol as a controlled drug delivery system for potential oral application, Journal of controlled release 33(1) (1995) 153-162.

[78] P. Maechler, L. Jornot, C.B. Wollheim, Hydrogen peroxide alters mitochondrial activation and insulin secretion in pancreatic beta cells, Journal of Biological Chemistry 274(39) (1999) 27905-27913.

[79] X. Fu-Liang, S. Xiao-Hui, G. Lu, Y. Xiang-Liang, X. Hui-Bi, Puerarin protects rat pancreatic islets from damage by hydrogen peroxide, European journal of pharmacology 529(1) (2006) 1-7.

[80] G.E. Kass, S. Orrenius, Calcium signaling and cytotoxicity, Environmental Health Perspectives 107(Suppl 1) (1999) 25-35.

[81] P.O. Carlsson, F. Palm, A. Andersson, P. Liss, Markedly decreased oxygen tension in transplanted rat pancreatic islets irrespective of the implantation site, Diabetes 50(3) (2001) 489-95.

[82] J. Siepmann, N.A. Peppas, Hydrophilic Matrices for Controlled Drug Delivery: An Improved Mathematical Model to Predict the Resulting Drug Release Kinetics (the “sequential Layer” Model), Pharmaceutical Research 17(10) (2000) 1290-1298.

[83] L. Bambrick, Y. Kostov, G. Rao, In vitro cell culture pO2 is significantly different from incubator pO2, Biotechnology progress 27(4) (2011) 1185-1189.

[84] Y. Zhou, Y. Wang, Z. Zhou, Y. Li, J. Lu, Preparation and Oxygen Release Properties of Calcium Peroxide-based Oxygen Slow-release Composite, China Water & Wastewater 28(7) (2012) 64-67.

[85] A.E.P.A.G. Top, Hydrogen peroxide use to increase oxidant capacity for in situ bioremediation of contaminated soils and aquifers: A review, Environmental Pollution 82(213) (1993).

248

[86] A.J. Waite, J.S. Bonner, R. Autenrieth, Kinetics and stoichiometry of oxygen release from solid peroxides, Environmental engineering science 16(3) (1999) 187-199.

[87] E. Pedraza, Engineering an optimal bioartificial pancreas for islet transplantation using bioactive scaffolds, (2011).

[88] A. Mashak, A. Rahimi, Silicone polymers in controlled drug delivery systems: a review, Iran Polym J 18(4) (2009) 279-295.

[89] Y. Fu, W.J. Kao, Drug release kinetics and transport mechanisms of non- degradable and degradable polymeric delivery systems, Expert opinion on drug delivery 7(4) (2010) 429-444.

[90] R. Sitruk-Ware, P. Inki, The levonorgestrel intrauterine system: long-term contraception and therapeutic effects, (2005).

[91] H. Maeda, M. Brandon, A. Sano, Design of controlled-release formulation for ivermectin using silicone, International journal of pharmaceutics 261(1) (2003) 9-19.

[92] A.M. Shapiro, C. Ricordi, B.J. Hering, H. Auchincloss, R. Lindblad, R.P. Robertson, A. Secchi, M.D. Brendel, T. Berney, D.C. Brennan, E. Cagliero, R. Alejandro, E.A. Ryan, B. DiMercurio, P. Morel, K.S. Polonsky, J.A. Reems, R.G. Bretzel, F. Bertuzzi, T. Froud, R. Kandaswamy, D.E. Sutherland, G. Eisenbarth, M. Segal, J. Preiksaitis, G.S. Korbutt, F.B. Barton, L. Viviano, V. Seyfert-Margolis, J. Bluestone, J.R. Lakey, International trial of the Edmonton protocol for islet transplantation, N Engl J Med 355(13) (2006) 1318-30.

[93] C.R. Group, 2007 update on allogeneic islet transplantation from the Collaborative Islet Transplant Registry (CITR), Cell Transplant 18(7) (2009) 753-67.

[94] E.A. Ryan, B.W. Paty, P.A. Senior, D. Bigam, E. Alfadhli, N.M. Kneteman, J.R. Lakey, A.M. Shapiro, A.M. Shapiro, J.R. Lakey, B.W. Paty, P.A. Senior, D.L. Bigam, E.A. Ryan, Five-year follow-up after clinical islet transplantation Strategic opportunities in clinical islet transplantation, Diabetes 54(7) (2005) 2060-9.

[95] M.R. Rickels, M.H. Schutta, J.F. Markmann, C.F. Barker, A. Naji, K.L. Teff, β-Cell Function Following Human Islet Transplantation for Type 1 Diabetes, Diabetes 54(1) (2005) 100-106.

[96] F.M. Gribble, Intolerant of glucose and gasping for oxygen, Nat Med 15(3) (2009) 247-249.

[97] K.E. Dionne, C.K. Colton, M.L. Yarmush, Effect of oxygen on isolated pancreatic tissue, ASAIO Trans 35(3) (1989) 739-41.

249

[98] C.A. Fraker, S. Cechin, S. Alvarez-Cubela, F. Echeverri, A. Bernal, R. Poo, C. Ricordi, L. Inverardi, J. Dominguez-Bendala, A physiological pattern of oxygenation using perfluorocarbon-based culture devices maximizes pancreatic islet viability and enhances beta-cell function, Cell transplantation 22(9) (2013) 1723-33.

[99] R. Olsson, J. Olerud, U. Pettersson, P.-O. Carlsson, Increased Numbers of Low- Oxygenated Pancreatic Islets After Intraportal Islet Transplantation, Diabetes 60(9) (2011) 2350-2353.

[100] D. Nyqvist, S. Speier, R. Rodriguez-Diaz, R.D. Molano, S. Lipovsek, M. Rupnik, A. Dicker, E. Ilegems, E. Zahr-Akrawi, J. Molina, M. Lopez-Cabeza, S. Villate, M.H. Abdulreda, C. Ricordi, A. Caicedo, A. Pileggi, P.-O. Berggren, Donor Islet Endothelial Cells in Pancreatic Islet Revascularization, Diabetes 60(10) (2011) 2571-2577.

[101] G. Mattsson, L. Jansson, P.-O. Carlsson, Decreased vascular density in mouse pancreatic islets after transplantation, Diabetes 51(1362-1366) (2002).

[102] W. Moritz, F. Meier, D. Stroka, M. Giuliani, P. Kugelmeier, P. Nett, R. Lehmann, D. Cadninas, M. Gassmann, M. Weber, Apoptosis in hypoxic human pancreatic islets correlates with HIF-1α expression, The FASEB Journal 16(7) (2002) 745-747.

[103] J.D. Gordan, C.B. Thompson, M.C. Simon, HIF and c-Myc: sibling rivals for control of cancer cell metabolism and proliferation, Cancer Cell 12(2) (2007) 108-13.

[104] B.D. Kelly, S.F. Hackett, K. Hirota, Y. Oshima, Z. Cai, S. Berg-Dixon, A. Rowan, Z. Yan, P.A. Campochiaro, G.L. Semenza, Cell type-specific regulation of angiogenic growth factor gene expression and induction of angiogenesis in nonischemic tissue by a constitutively active form of hypoxia-inducible factor 1, Circ Res 93(11) (2003) 1074-81.

[105] D.J. Manalo, A. Rowan, T. Lavoie, L. Natarajan, B.D. Kelly, S.Q. Ye, J.G.N. Garcia, G.L. Semenza, Transcriptional regulation of vascular endothelial cell responses to hypoxia by HIF-1, 2005.

[106] E.P. Cummins, E. Berra, K.M. Comerford, A. Ginouves, K.T. Fitzgerald, F. Seeballuck, C. Godson, J.E. Nielsen, P. Moynagh, J. Pouyssegur, C.T. Taylor, Prolyl hydroxylase-1 negatively regulates IkappaB kinase-beta, giving insight into hypoxia- induced NFkappaB activity, Proc Natl Acad Sci U S A 103(48) (2006) 18154-9.

[107] G.L. Semenza, Regulation of mammalian O2 homeostasis by hypoxia-inducible factor 1, Annu Rev Cell Dev Biol 15 (1999) 551-78.

[108] W.G. Kaelin, P.J. Ratcliffe, Oxygen sensing by metazoans: the central role of the HIF hydroxylase pathway, Mol Cell 30(4) (2008) 393-402.

250

[109] N.S. Kenyon, M. Chatzipetrou, M. Masetti, A. Ranuncoli, M. Oliveira, J.L. Wagner, A.D. Kirk, D.M. Harlan, L.C. Burkly, C. Ricordi, Long-term survival and function of intrahepatic islet allografts in rhesus monkeys treated with humanized anti-CD154, Proceedings of the National Academy of Sciences of the United States of America 96(14) (1999) 8132-7.

[110] P. Buchwald, S.R. Cechin, J.D. Weaver, C.L. Stabler, Experimental evaluation and computational modeling of the effects of encapsulation on the time-profile of glucose-stimulated insulin release of pancreatic islets, BioMedical Engineering OnLine 14(1) (2015) 1-14.

[111] K.K. Papas, A. Pisania, H. Wu, G.C. Weir, C.K. Colton, A stirred microchamber for oxygen consumption rate measurements with pancreatic islets, Biotechnol Bioeng 98(5) (2007) 1071-82.

[112] J.D. Wikstrom, S.B. Sereda, L. Stiles, A. Elorza, E.M. Allister, A. Neilson, D.A. Ferrick, M.B. Wheeler, O.S. Shirihai, A Novel High-Throughput Assay for Islet Respiration Reveals Uncoupling of Rodent and Human Islets, PLoS ONE 7(5) (2012) e33023.

[113] E. Pedraza, A.C. Brady, C.A. Fraker, R.D. Molano, S. Sukert, D.M. Berman, N.S. Kenyon, A. Pileggi, C. Ricordi, C.L. Stabler, Macroporous three-dimensional PDMS scaffolds for extrahepatic islet transplantation, Cell Transplant 22(7) (2013) 1123-35.

[114] E. Pedraza, A. Brady, C. Fraker, R. Molano, S. Sukert, N. Kenyon, A. Pileggi, C. Ricordi, C. Stabler, Macro-porous PDMS Scaffolds for Extrahepatic Islet Transplantation, Cell transplantation 22(7) (2013) 1123-1135.

[115] A. Brady, M. Martino, E. Pedraza, S. Sukert, A. Pileggi, C. Ricordi, J. Hubbell, C. Stabler, Pro-Angiogenic Hydrogels within Macroporous Scaffolds Enhances Islet Engraftment in an Extrahepatic Site, Tissue Engineering 19(23-24) (2013) 2544-52.

[116] A.C. Brady, M.M. Martino, E. Pedraza, S. Sukert, A. Pileggi, C. Ricordi, J.A. Hubbell, C.L. Stabler, Proangiogenic hydrogels within macroporous scaffolds enhance islet engraftment in an extrahepatic site, Tissue Eng Part A 19(23-24) (2013) 2544-52.

[117] G.C. Weir, S. Bonner-Weir, J.L. Leahy, Islet Mass and Function in Diabetes and Transplantation, Diabetes 39(4) (1990) 401-405.

[118] T. Samuel, P. Cockwell, Islet cell transplantation, Journal of the Royal Society of Medicine 95(1) (2002) 31-33.

[119] A. Kim, K. Miller, J. Jo, G. Kilimnik, P. Wojcik, M. Hara, Islet architecture: A comparative study, Islets 1(2) (2009) 129-36.

251

[120] S. Bonner-Weir, B.A. Sullivan, G.C. Weir, Human Islet Morphology Revisited: Human and Rodent Islets Are Not so Different After All, The journal of histochemistry and cytochemistry : official journal of the Histochemistry Society (2015).

[121] D.S. Luciani, S.A. White, S.B. Widenmaier, V.V. Saran, F. Taghizadeh, X. Hu, M.F. Allard, J.D. Johnson, Bcl-2 and Bcl-xL suppress glucose signaling in pancreatic β- cells, Diabetes 62(1) (2013) 170-82.

[122] R.J. Boohaker, G. Zhang, A.L. Carlson, K.N. Nemec, A.R. Khaled, BAX supports the mitochondrial network, promoting bioenergetics in nonapoptotic cells, Am J Physiol Cell Physiol 300(6) (2011) C1466-78.

[123] I. Papandreou, R.A. Cairns, L. Fontana, A.L. Lim, N.C. Denko, HIF-1 mediates adaptation to hypoxia by actively downregulating mitochondrial oxygen consumption, Cell Metab 3(3) (2006) 187-97.

[124] C.T. Taylor, E.P. Cummins, The Role of NF‐κB in Hypoxia‐Induced Gene Expression, Annals of the New York Academy of Sciences 1177(1) (2009) 178-184.

[125] M.Y. Donath, M. Böni-Schnetzler, H. Ellingsgaard, P.A. Halban, J.A. Ehses, Cytokine production by islets in health and diabetes: cellular origin, regulation and function, Trends in Endocrinology & Metabolism 21(5) (2010) 261-267.

[126] D.M. Berman, R.D. Molano, C. Fotino, U. Ulissi, J. Gimeno, A.J. Mendez, N.M. Kenyon, N.S. Kenyon, D.M. Andrews, C. Ricordi, A. Pileggi, Bioengineering the Endocrine Pancreas: Intraomental Islet Transplantation Within a Biologic Resorbable Scaffold, Diabetes 65(5) (2016) 1350-1361.

[127] J. Lau, P.O. Carlsson, Low revascularization of human islets when experimentally transplanted into the liver, Transplantation 87(3) (2009) 322-5.

[128] J. Cantley, S.T. Grey, P.H. Maxwell, D.J. Withers, The hypoxia response pathway and β-cell function, Diabetes Obes Metab 12 Suppl 2 (2010) 159-67.

[129] A. Pileggi, E.S. Fenjves, D. Klein, C. Ricordi, R.L. Pastori, Protecting pancreatic beta-cells, IUBMB Life 56(7) (2004) 387-94.

[130] C.M. Girgis, K. Cheng, C.H. Scott, J.E. Gunton, Novel links between HIFs, type 2 diabetes, and metabolic syndrome, Trends in Endocrinology & Metabolism 23(8) (2012) 372-380.

[131] M. Bensellam, B. Duvillie, G. Rybachuk, D.R. Laybutt, C. Magnan, Y. Guiot, J. Pouyssegur, J.-C. Jonas, Glucose-induced O 2 consumption activates hypoxia inducible factors 1 and 2 in rat insulin-secreting pancreatic beta-cells, PLoS One 7(1) (2012) e29807.

252

[132] J.A. Beckman, M.A. Creager, P. Libby, Diabetes and atherosclerosis: epidemiology, pathophysiology, and management, JAMA 287(19) (2002) 2570-81.

[133] M. Nagata, Y. Mullen, S. Matsuo, M. Herrera, M. Clare-Salzler, Destruction of islet isografts by severe nonspecific inflammation, Transplant Proc 22(2) (1990) 855-6.

[134] U. Johansson, A. Olsson, S. Gabrielsson, B. Nilsson, O. Korsgen, Inflammatory mediators expressed in human islets of Langerhans: implications for islet transplantation, Biochem Biophys Res Commun 308 (2003) 474-479.

[135] M.Y. Donath, J. Størling, K. Maedler, T. Mandrup-Poulsen, Inflammatory mediators and islet β-cell failure: a link between type 1 and type 2 diabetes, Journal of Molecular Medicine 81(8) (2003) 455-470.

[136] I.L. Campbell, T.W. Kay, L. Oxbrow, L.C. Harrison, Essential role for interferon- gamma and interleukin-6 in autoimmune insulin-dependent diabetes in NOD/Wehi mice, Journal of Clinical Investigation 87(2) (1991) 739-742.

[137] K.T. Chambers, J.A. Unverferth, S.M. Weber, R.C. Wek, F. Urano, J.A. Corbett, The role of nitric oxide and the unfolded protein response in cytokine-induced β-cell death, Diabetes 57(1) (2008) 124-132.

[138] F. Jung, L.A. Palmer, N. Zhou, R.A. Johns, Hypoxic Regulation of Inducible Nitric Oxide Synthase via Hypoxia Inducible Factor-1 in Cardiac Myocytes, Circulation Research 86(3) (2000) 319-325.

[139] I.S. Grewal, B.J. Rutledge, J.A. Fiorillo, L. Gu, R.P. Gladue, R.A. Flavell, B.J. Rollins, Transgenic monocyte chemoattractant protein-1 (MCP-1) in pancreatic islets produces monocyte-rich insulitis without diabetes: abrogation by a second transgene expressing systemic MCP-1, J Immunol 159(1) (1997) 401-8.

[140] M.A. Kriegel, C. Rathinam, R.A. Flavell, Pancreatic islet expression of chemokine CCL2 suppresses autoimmune diabetes via tolerogenic CD11c+ CD11b+ dendritic cells, Proceedings of the National Academy of Sciences of the United States of America 109(9) (2012) 3457-62.

[141] L. Piemonti, B.E. Leone, R. Nano, A. Saccani, P. Monti, P. Maffi, G. Bianchi, A. Sica, G. Peri, R. Melzi, L. Aldrighetti, A. Secchi, V. Di Carlo, P. Allavena, F. Bertuzzi, Human pancreatic islets produce and secrete MCP-1/CCL2: relevance in human islet transplantation, Diabetes 51(1) (2002) 55-65.

[142] D.J. Manalo, A. Rowan, T. Lavoie, L. Natarajan, B.D. Kelly, S.Q. Ye, J.G. Garcia, G.L. Semenza, Transcriptional regulation of vascular endothelial cell responses to hypoxia by HIF-1, Blood 105(2) (2005) 659-69.

253

[143] B. Vasir, L.P. Aiello, K.H. Yoon, R.R. Quickel, S. Bonner-Weir, G.C. Weir, Hypoxia induces vascular endothelial growth factor gene and protein expression in cultured rat islet cells, Diabetes 47(12) (1998) 1894-1903.

[144] K. Masuda, K. Abdelmohsen, M. Gorospe, RNA-binding proteins implicated in the hypoxic response, Journal of Cellular and Molecular Medicine 13(9a) (2009) 2759-2769.

[145] L.P. McQuillan, G.K. Leung, P.A. Marsden, S.K. Kostyk, S. Kourembanas, Hypoxia inhibits expression of eNOS via transcriptional and posttranscriptional mechanisms, American Journal of Physiology - Heart and Circulatory Physiology 267(5) (1994) H1921-H1927.

[146] Y. Lai, D. Schneider, A. Kidszun, I. Hauck-Schmalenberger, G. Breier, D. Brandhorst, H. Brandhorst, M. Iken, M.D. Brendel, R.G. Bretzel, T. Linn, Vascular endothelial growth factor increases functional beta-cell mass by improvement of angiogenesis of isolated human and murine pancreatic islets, Transplantation 79(11) (2005) 1530-6.

[147] R. Bottino, A.N. Balamurugan, H. Tse, C. Thirunavukkarasu, X. Ge, J. Profozich, M. Milton, A. Ziegenfuss, M. Trucco, J.D. Piganelli, Response of human islets to isolation stress and the effect of antioxidant treatment, Diabetes 53(10) (2004) 2559-68.

[148] A.N. Balamurugan, B. Naziruddin, A. Lockridge, M. Tiwari, G. Loganathan, M. Takita, S. Matsumoto, K. Papas, M. Trieger, H. Rainis, T. Kin, T.W. Kay, S. Wease, S. Messinger, C. Ricordi, R. Alejandro, J. Markmann, J. Kerr-Conti, M.R. Rickels, C. Liu, X. Zhang, P. Witkowski, A. Posselt, P. Maffi, A. Secchi, T. Berney, P.J. O'Connell, B.J. Hering, F.B. Barton, Islet Product Characteristics and Factors Related to Successful Human Islet Transplantation From the Collaborative Islet Transplant Registry (CITR) 1999–2010, American Journal of Transplantation 14(11) (2014) 2595-2606.

[149] C. Ricordi, J.S. Goldstein, A.N. Balamurugan, G.L. Szot, T. Kin, C. Liu, C.W. Czarniecki, B. Barbaro, N.D. Bridges, J. Cano, W.R. Clarke, T.L. Eggerman, L.G. Hunsicker, D.B. Kaufman, A. Khan, D.-E. Lafontant, E. Linetsky, X. Luo, J.F. Markmann, A. Naji, O. Korsgren, J. Oberholzer, N.A. Turgeon, D. Brandhorst, A.S. Friberg, J. Lei, L.-j. Wang, J.J. Wilhelm, J. Willits, X. Zhang, B.J. Hering, A.M. Posselt, A.M.J. Shapiro, NIH-sponsored Clinical Islet Transplantation Consortium Phase 3 Trial: Manufacture of a Complex Cellular Product at Eight Processing Facilities, Diabetes (2016).

[150] M.M. Coronel, C.L. Stabler, Engineering a local microenvironment for pancreatic islet replacement, Current Opinion in Biotechnology 24(5) (2013) 900-908.

[151] B. Ludwig, A. Reichel, A. Steffen, B. Zimerman, A.V. Schally, N.L. Block, C.K. Colton, S. Ludwig, S. Kersting, E. Bonifacio, M. Solimena, Z. Gendler, A. Rotem, U. Barkai, S.R. Bornstein, Transplantation of human islets without immunosuppression, Proc Natl Acad Sci U S A 110(47) (2013) 19054-8.

254

[152] B. Ludwig, A. Reichel, A. Steffen, B. Zimmermann, C.K. Colton, S. Ludwig, S. Kersting, E. Bonifacio, M. Solimena, Z. Gendler, A. Rotem, U. Barkai, S.R. Bornstein, Transplanting allo-islets without immunosuppression, Xenotransplantation 21(2) (2014) 189-189.

[153] E. Pedraza, A.-C. Brady, C.A. Fraker, R.D. Molano, S. Sukert, D.M. Berman, N.S. Kenyon, A. Pileggi, C. Ricordi, C.L. Stabler, Macroporous Three Dimensional PDMS Scaffolds for Extrahepatic Islet Transplantation, Cell transplantation 22(7) (2013) 1123- 1135.

[154] P.W. Burridge, A.M.J. Shapiro, E.A. Ryan, J.R.T. Lakey, Future trends in clinical islet transplantation, Transplantation Proceedings 34(8) (2002) 3347-3348.

[155] W. Bennet, C.-G. Groth, R. Larsson, B. Nilsson, O. Korsgren, Isolated Human Islets Trigger an Instant Blood Mediated Inflammatory Reaction: Implications for Intraportal Islet Transplantation as a Treatment for Patients with Type 1 Diabetes, Upsala Journal of Medical Sciences 105(2) (2000) 125-133.

[156] N.R. Barshes, S. Wyllie, J.A. Goss, Inflammation-mediated dysfunction and apoptosis in pancreatic islet transplantation: implications for intrahepatic grafts, Journal of Leukocyte Biology 77(5) (2005) 587-597.

[157] D. Dufrane, R.-M. Goebbels, P. Gianello, Alginate Macroencapsulation of Pig Islets Allows Correction of Streptozotocin-Induced Diabetes in Primates up to 6 Months Without Immunosuppression, Transplantation 90(10) (2010) 1054-1062.

[158] O.M. Sabek, S. Ferrati, D.W. Fraga, J. Sih, E.V. Zabre, D.H. Fine, M. Ferrari, A.O. Gaber, A. Grattoni, Characterization of a nanogland for the autotransplantation of human pancreatic islets, Lab on a Chip 13(18) (2013) 3675-3688.

[159] G. Warnock, R. Rajotte, A. Procyshyn, Normoglycemia after reflux of islet- containing pancreatic fragments into the splenic vascular bed in dogs, Diabetes 32(5) (1983) 452-460.

[160] A. Balamurugan, Y. Gu, Y. Tabata, M. Miyamoto, W. Cui, H. Hori, A. Satake, N. Nagata, W. Wang, K. Inoue, Bioartificial pancreas transplantation at prevascularized intermuscular space: effect of angiogenesis induction on islet survival, Pancreas 26(3) (2003) 279-285.

[161] P. Witkowski, H. Sondermeijer, M.A. Hardy, D.C. Woodland, K. Lee, G. Bhagat, K. Witkowski, F. See, A. Rana, A. Maffei, Islet grafting and imaging in a bioengineered intramuscular space, Transplantation 88(9) (2009) 1065.

255

[162] P. Soon-Shiong, E. Feldman, R. Nelson, R. Heintz, Q. Yao, Z. Yao, T. Zheng, N. Merideth, G. Skjak-Braek, T. Espevik, Long-term reversal of diabetes by the injection of immunoprotected islets, Proceedings of the National Academy of Sciences of the United States of America 90(12) (1993) 5843-5847.

[163] R.B. Elliott, L. Escobar, R. Calafiore, G. Basta, O. Garkavenko, A. Vasconcellos, C. Bambra, Transplantation of micro- and macroencapsulated piglet islets into mice and monkeys, Transplant Proc 37(1) (2005) 466-9.

[164] E.S. O'Sullivan, A. Vegas, D.G. Anderson, G.C. Weir, Islets Transplanted in Immunoisolation Devices: A Review of the Progress and the Challenges that Remain, Endocrine Reviews 32(6) (2011) 827-844.

[165] D.M. Berman, R.D. Molano, C. Fotino, U. Ulissi, J. Gimeno, A.J. Mendez, N.M. Kenyon, N.S. Kenyon, D.M. Andrews, C. Ricordi, Bioengineering the endocrine pancreas: intraomental islet transplantation within a biologic resorbable scaffold, Diabetes 65(5) (2016) 1350-1361.

[166] A.R. Pepper, B. Gala-Lopez, R. Pawlick, S. Merani, T. Kin, A.J. Shapiro, A prevascularized subcutaneous device-less site for islet and cellular transplantation, Nature biotechnology 33(5) (2015) 518-523.

[167] B. Ludwig, A. Rotem, J. Schmid, G.C. Weir, C.K. Colton, M.D. Brendel, T. Neufeld, N.L. Block, K. Yavriyants, A. Steffen, S. Ludwig, T. Chavakis, A. Reichel, D. Azarov, B. Zimermann, S. Maimon, M. Balyura, T. Rozenshtein, N. Shabtay, P. Vardi, K. Bloch, P. de Vos, A.V. Schally, S.R. Bornstein, U. Barkai, Improvement of islet function in a bioartificial pancreas by enhanced oxygen supply and growth hormone releasing hormone agonist, Proc Natl Acad Sci U S A 109(13) (2012) 5022-7.

[168] R. Gibly, J. Graham, X. Luo, W. Lowe Jr, B. Hering, L. Shea, Advancing islet transplantation: from engraftment to the immune response, Diabetologia 54(10) (2011) 2494-2505.

[169] A.M. Smink, M.M. Faas, P. de Vos, Toward engineering a novel transplantation site for human pancreatic islets, Diabetes 62(5) (2013) 1357-1364.

[170] W.F. Hiller, J. Klempnauer, R. Lück, B. Steiniger, Progressive deterioration of endocrine function after intraportal but not kidney subcapsular rat islet transplantation, Diabetes 40(1) (1991) 134-140.

[171] H.P. Selawry, K. Whittington, Extended allograft survival of islets grafted into intra- abdominally placed testis, Diabetes 33(4) (1984) 405-406.

[172] A.M. Posselt, C.F. Barker, J.E. Tomaszewski, J.F. Markmann, M.A. Choti, A. Naji, Induction of donor-specific unresponsiveness by intrathymic islet transplantation, Science 249(4974) (1990) 1293-1295.

256

[173] T.R. Hefty, C.S. Kuhr, K.T. Chong, D.G. Guinee, W. Wang, J.A. Reems, C.J. Greenbaum, Omental roll-up: a technique for islet engraftment in a large animal model, J Surg Res 161(1) (2010) 134-8.

[174] J.M. Anderson, Biological responses to materials, Annual review of materials research 31(1) (2001) 81-110.

[175] A.M.J. Shapiro, J.R.T. Lakey, E.A. Ryan, G.S. Korbutt, E. Toth, G.L. Warnock, N.M. Kneteman, R.V. Rajotte, Islet Transplantation in Seven Patients with Type 1 Diabetes Mellitus Using a Glucocorticoid-Free Immunosuppressive Regimen, New England Journal of Medicine 343(4) (2000) 230-238.

[176] E.a. Ryan, B.W. Paty, P.a. Senior, D. Bigam, E. Alfadhli, N.M. Kneteman, J.R.T. Lakey, a.M.J. Shapiro, Five-year follow-up after clinical islet transplantation., Diabetes 54 (2005) 2060-9.

[177] R.C. Froud T, Baidal DA, et al., Islet transplantation in type 1 diabetes mellitus using cultured islets and steroid-free immunosupression: miami experience, Am J Transplant 5(8) (2005).

[178] C.K. Colton, Implantable biohybrid artificial organs, Cell transplantation 4(4) (1995) 415-436.

[179] J.H. Brauker, V.E. Carr‐Brendel, L.A. Martinson, J. Crudele, W.D. Johnston, R.C. Johnson, Neovascularization of synthetic membranes directed by membrane microarchitecture, Journal of biomedical materials research 29(12) (1995) 1517-1524.

[180] E.A. Phelps, N. Landázuri, P.M. Thulé, W.R. Taylor, A.J. García, Bioartificial matrices for therapeutic vascularization, Proc Natl Acad Sci U S A 107(8) (2010) 3323- 8.

[181] P. Adlercreutz, B. Mattiasson, Oxygen supply to immobilized cells, European journal of applied microbiology and biotechnology 16(4) (1982) 165-170.

[182] P. De Vos, J.-L. Hillebrands, B.J. De Haan, J.H. Strubbe, R. Van Schilfgaarde, Efficacy of a prevascularized expanded polytetrafluoroethylene solid support system as a transplantation site for pancreatic islets1, Transplantation 63(6) (1997) 824-830.

[183] D.R. Burnett, L.M. Huyett, H.C. Zisser, F.J. Doyle, B.D. Mensh, Glucose sensing in the peritoneal space offers faster kinetics than sensing in the subcutaneous space, Diabetes 63(7) (2014) 2498-2505.

[184] P. De Vos, D. Vegter, B.J. De Haan, J.H. Strubbe, J.E. Bruggink, R. Van Schilfgaarde, Kinetics of intraperitoneally infused insulin in rats: functional implications for the bioartificial pancreas, Diabetes 45(8) (1996) 1102-1107.

257

[185] F. Wang, S.S. Li, R. Segersvärd, L. Strömmer, K.-G. Sundqvist, J. Holgersson, J. Permert, Hypoxia inducible factor-1 mediates effects of insulin on pancreatic cancer cells and disturbs host energy homeostasis, The American journal of pathology 170(2) (2007) 469-477.

[186] D.P. Bartel, MicroRNAs: target recognition and regulatory functions, cell 136(2) (2009) 215-233.

[187] C. Koumenis, C. Naczki, M. Koritzinsky, S. Rastani, A. Diehl, N. Sonenberg, A. Koromilas, B.G. Wouters, Regulation of protein synthesis by hypoxia via activation of the endoplasmic reticulum kinase PERK and phosphorylation of the translation initiation factor eIF2α, Molecular and cellular biology 22(21) (2002) 7405-7416.

[188] A.R. Pepper, B. Gala-Lopez, O. Ziff, A.M.J. Shapiro, Revascularization of Transplanted Pancreatic Islets and Role of the Transplantation Site, Clinical and Developmental Immunology 2013 (2013) 13.

[189] T. Tun, K. Inoue, H. Hayashi, T. Aung, Y.-J. Gu, R. Doi, H. Kaji, Y. Echigo, W.-J. Wang, H. Setoyama, A newly developed three-layer agarose microcapsule for a promising biohybrid artificial pancreas: rat to mouse xenotransplantation, Cell transplantation 5(5) (1996) S59-S63.

[190] T. Kobayashi, Y. Aomatsu, H. Iwata, T. Kin, H. Kanehiro, M. Hisanaga, S. Ko, M. Nagao, Y. Nakajima, Indefinite islet protection from autoimmune destruction in nonobese diabetic mice by agarose microencapsulation without immunosuppression1, Transplantation 75(5) (2003) 619-625.

[191] R. Lanza, P. Soon-Shiong, Experimental xenotransplantation of encapsulated cells, Xenotransplantation, Springer1991, pp. 297-312.

[192] M.S. Shoichet, R.H. Li, M.L. White, S.R. Winn, Stability of hydrogels used in cell encapsulation: An in vitro comparison of alginate and agarose, Biotechnology and bioengineering 50(4) (1996) 374-381.

[193] H. Iwata, K. Kobayashi, T. Takagi, T. Oka, H. Yang, H. Amemiya, T. Tsuji, F. Ito, Feasibility of agarose microbeads with xenogeneic islets as a bioartificial pancreas, Journal of biomedical materials research 28(9) (1994) 1003-1011.

[194] Y. Sun, X. Ma, D. Zhou, I. Vacek, A.M. Sun, Normalization of diabetes in spontaneously diabetic cynomologus monkeys by xenografts of microencapsulated porcine islets without immunosuppression, Journal of Clinical Investigation 98(6) (1996) 1417.

258

[195] A.A. Tomei, V. Manzoli, C.A. Fraker, J. Giraldo, D. Velluto, M. Najjar, A. Pileggi, R.D. Molano, C. Ricordi, C.L. Stabler, Device design and materials optimization of conformal coating for islets of Langerhans, Proceedings of the National Academy of Sciences 111(29) (2014) 10514-10519.

[196] Y. Teramura, H. Iwata, Bioartificial pancreas: microencapsulation and conformal coating of islet of Langerhans, Advanced drug delivery reviews 62(7) (2010) 827-840.

[197] M. Kumagai-Braesch, S. Jacobson, H. Mori, X. Jia, T. Takahashi, A. Wernerson, M. Flodström-Tullberg, A. Tibell, The TheraCyte™ Device Protects Against Islet Allograft Rejection in Immunized Hosts, Cell Transplantation 22(7) (2013) 1137-1146.

[198] Y. Kawakami, H. Iwata, Y.J. Gu, M. Miyamoto, Y. Murakami, A. Balamurugan, M. Imamura, K. Inoue, Successful subcutaneous pancreatic islet transplantation using an angiogenic growth factor–releasing device, Pancreas 23(4) (2001) 375-381.

[199] S. Peng, P.G. Hartley, T.C. Hughes, Q. Guo, Controlling morphology and porosity of porous siloxane membranes through water content of precursor microemulsion, Soft Matter 8(40) (2012) 10493-10501.

[200] D.N. Soulas, K.G. Papadokostaki, A. Panou, M. Sanopoulou, Drug Release from Poly(dimethylsiloxane)-Based Matrices: Observed and Predicted Stabilization of the Release Rate by Three-Layer Devices, Industrial & Engineering Chemistry Research 51(20) (2012) 7126-7136.

[201] C. Stabler, K. Wilks, A. Sambanis, I. Constantinidis, The effects of alginate composition on encapsulated βTC3 cells, Biomaterials 22(11) (2001) 1301-1310.

[202] C.L. Stabler, A. Sambanis, I. Constantinidis, Effects of alginate composition on the growth and overall metabolic activity of βtc3 cells, Annals of the New York Academy of Sciences 961(1) (2002) 130-133.

[203] P.E. Lacy, O.D. Hegre, A. Gerasimidi-Vazeou, F.T. Gentile, K.E. Dionne, Maintenance of normoglycemia in diabetic mice by subcutaneous xenografts of encapsulated islets, Science 254(5039) (1991) 1782-1784.

[204] R.P. Lanza, D.H. Butler, K.M. Borland, J.E. Staruk, D.L. Faustman, B.A. Solomon, T.E. Muller, R.G. Rupp, T. Maki, A.P. Monaco, Xenotransplantation of canine, bovine, and porcine islets in diabetic rats without immunosuppression, Proceedings of the National Academy of Sciences 88(24) (1991) 11100-11104.

[205] R. Elliott, L. Escobar, P. Tan, O. Garkavenko, R. Calafiore, P. Basta, A. Vasconcellos, D. Emerich, C. Thanos, C. Bambra, Intraperitoneal alginate-encapsulated neonatal porcine islets in a placebo-controlled study with 16 diabetic cynomolgus primates, Transplantation proceedings, Elsevier, 2005, pp. 3505-3508.

259

[206] J. Folkman, D.M. Long, The use of silicone rubber as a carrier for prolonged drug therapy, Journal of surgical research 4(3) (1964) 139-142.

[207] A.S. Hoffman, The origins and evolution of “controlled” drug delivery systems, Journal of Controlled Release 132(3) (2008) 153-163.

[208] R. Lo, K. Kuwahara, P.-Y. Li, R. Agrawal, M.S. Humayun, E. Meng, A passive refillable intraocular MEMS drug delivery device, Microtechnologies in Medicine and Biology, 2006 International Conference on, IEEE, 2006, pp. 74-77.

[209] P.K. Yuen, H. Su, V.N. Goral, K.A. Fink, Three-dimensional interconnected microporous poly (dimethylsiloxane) microfluidic devices, Lab on a Chip 11(8) (2011) 1541-1544.

[210] E. Pedraza, A.-C. Brady, C.A. Fraker, C.L. Stabler, Synthesis of macroporous poly (dimethylsiloxane) scaffolds for tissue engineering applications, Journal of Biomaterials Science, Polymer Edition 24(9) (2013) 1041-1056.

[211] N.A. Alcantar, E.S. Aydil, J.N. Israelachvili, Polyethylene glycol-coated biocompatible surfaces, Journal of biomedical materials research 51(3) (2000) 343-351.

[212] X.Y. Liu, J.-M. Nothias, A. Scavone, M. Garfinkel, J.M. Millis, Biocompatibility investigation of polyethylene glycol and alginate-poly-L-lysine for islet encapsulation, ASAIO journal 56(3) (2010) 241-245.

[213] I.R. Sweet, M. Gilbert, S. Scott, I. Todorov, R. Jensen, I. Nair, I. Al-Abdullah, J. Rawson, F. Kandeel, K. Ferreri, Glucose-stimulated increment in oxygen consumption rate as a standardized test of human islet quality, Am J Transplant 8(1) (2008) 183-92.

[214] R.K. Alla, K. Ginjupalli, N. Upadhya, M. Shammas, R.K. Ravi, R. Sekhar, Surface roughness of implants: a review, Trends in Biomaterials and Artificial Organs 25(3) (2011) 112-118.

[215] M. Biarnés, M. Montolio, V. Nacher, M. Raurell, J. Soler, E. Montanya, β-cell death and mass in syngeneically transplanted islets exposed to short-and long-term hyperglycemia, Diabetes 51(1) (2002) 66-72.

[216] U. Siebers, A. Horcher, R. Bretzel, G. Klöck, U. Zimmermann, K. Federlin, T. Zekorn, Transplantation of free and microencapsulated islets in rats: evidence for the requirement of an increased islet mass for transplantation into the peritoneal site, The International journal of artificial organs 16(2) (1993) 96-99.

[217] R.P. Meier, J.D. Seebach, P. Morel, R. Mahou, S. Borot, L. Giovannoni, G. Parnaud, E. Montanari, D. Bosco, C. Wandrey, Survival of free and encapsulated human and rat islet xenografts transplanted into the mouse bone marrow, PloS one 9(3) (2014) e91268.

260

[218] H. Ishihara, T. Asano, K. Tsukuda, H. Katagiri, K. Inukai, M. Anai, M. Kikuchi, Y. Yazaki, J.-I. Miyazaki, Y. Oka, Pancreatic beta cell line MIN6 exhibits characteristics of glucose metabolism and glucose-stimulated insulin secretion similar to those of normal islets, Diabetologia 36(11) (1993) 1139-1145.

[219] W.-H. Kim, J.W. Lee, Y.H. Suh, S.H. Hong, J.S. Choi, J.H. Lim, J.H. Song, B. Gao, M.H. Jung, Exposure to chronic high glucose induces β-Cell apoptosis through decreased interaction of glucokinase with mitochondria downregulation of glucokinase in pancreatic β-Cells, Diabetes 54(9) (2005) 2602-2611.

[220] R.P. Bunge, Expanding roles for the Schwann cell: ensheathment, myelination, trophism and regeneration, Curr Opin Neurobiol 3(5) (1993) 805-9.

[221] D.D. Pearse, F.C. Pereira, A.E. Marcillo, M.L. Bates, Y.A. Berrocal, M.T. Filbin, M.B. Bunge, cAMP and Schwann cells promote axonal growth and functional recovery after spinal cord injury, Nat Med 10(6) (2004) 610-6.

[222] E. Eftekharpour, S. Karimi-Abdolrezaee, M.G. Fehlings, Current status of experimental cell replacement approaches to spinal cord injury, Neurosurgical focus 24(3-4) (2008) E19.

[223] I.R. Griffiths, M.C. McCulloch, Nerve fibres in spinal cord impact injuries. Part 1. Changes in the myelin sheath during the initial 5 weeks, J Neurol Sci 58(3) (1983) 335- 49.

[224] C.H. Tator, D.W. Rowed, Current concepts in the immediate management of acute spinal cord injuries, Can Med Assoc J 121(11) (1979) 1453-64.

[225] C.H. Tator, M.G. Fehlings, Review of the secondary injury theory of acute spinal cord trauma with emphasis on vascular mechanisms, J Neurosurg 75(1) (1991) 15-26.

[226] N.E. Davis, L.N. Beenken-Rothkopf, A. Mirsoian, N. Kojic, D.L. Kaplan, A.E. Barron, M.J. Fontaine, Enhanced function of pancreatic islets co-encapsulated with ECM proteins and mesenchymal stromal cells in a silk hydrogel, Biomaterials 33(28) (2012) 6691-7.

261

BIOGRAPHICAL SKETCH

Maria was born in Barranquilla, Colombia. Her father was a professor in epidemiology and her mother was a clinical dietician. Maria lived and finished her primary studies in Colombia before deciding to move to the US to pursue a bachelor’s degree in biomedical engineering. After two years of community college Maria transferred to the University of Miami where she completed her B.S. degree in

Biomedical Engineering in 2010. During her undergraduate years she was actively involved in research at the Diabetes Research Institute under the supervision of Dr.

Cherie Stabler. In 2011, she started to pursue a doctorate in Biomedical Engineering.

She received her PhD. from the University of Florida in the fall of 2016.

Maria’s research thesis was focused on the engineering of oxygen generating materials for addressing the universal challenge of hypoxia within three-dimensional tissue engineered implants. Her work span from material development and characterization, to primary cell evaluation, and preclinical model testing. Further, her graduate project, entitled “Development of Bioactive Oxygen Platforms for Extrahepatic

Islet Transplantation” was awarded an individual predoctoral F31 fellowship from the

National Institute of Health (NIH). Through the subsequent course of her doctoral tenure, she published in some of the most prestigious journals in the field such as

Current Opinions in Biotechnology, and in the Proceedings of the National Academic of

Science.

Maria is passionate about traveling and exploring new places, cultures and food.

She enjoys watching tennis, cooking, going out for dinner with friends, and having great intellectual conversations. She believes in success through hard work and dedication.

262

Her motto in life is “Success is no accident. It is hard work, perseverance, learning, studying, sacrifice and most of all, love of what you are doing or learning to do” (Pele).

263