Chapter 5 Co-Culture of Young Porcine Islets with Liver Microvascular Endothelial Cells in Fibrin Improves Insulin Secretion and Reduces Apoptosis

UNIVERSITE DE SHERBROOKE

Faculte de genie Departement de genie chimique et de genie biotechnologique

DEVELOPPEMENT ET VALIDATION DE SURFACES ET D’ECHAFAUDAGES PROPICES AU DEVELOPPEMENT ET AU MAINTIEN DES FONCTIONS DE CELLULES PANCREATIQUES BETA

TOWARDS THE DEVELOPMENT AND VALIDATION OF BIOMATERIAL SURFACES AND SCAFFOLDS SUITABLE FOR PANCREATIC BETA CELL DEVELOPMENT AND FUNCTION

These de doctorat Specialite: genie chimique

Evan Aiozie DUBIEL

Jury : Patrick VERMETTE (directeur) Jonathan LAKEY

Steven PARASKEVAS Marie-Josee BOUCHER

Marcel LACROIX (rapporteur)

Sherbrooke (Quebec) Canada Novembre 2012 Library and Archives Bibliotheque et Canada Archives Canada

Published Heritage Direction du 1+1 Branch Patrimoine de I'edition 395 Wellington Street 395, rue Wellington Ottawa ON K1A0N4 Ottawa ON K1A 0N4 Canada Canada Your file Votre reference ISBN: 978-0-494-93268-1

Our file Notre reference ISBN: 978-0-494-93268-1

NOTICE: AVIS: The author has granted a non L'auteur a accorde une licence non exclusive exclusive license allowing Library and permettant a la Bibliotheque et Archives Archives Canada to reproduce, Canada de reproduire, publier, archiver, publish, archive, preserve, conserve, sauvegarder, conserver, transmettre au public communicate to the public by par telecommunication ou par I'lnternet, preter, telecommunication or on the Internet, distribuer et vendre des theses partout dans le loan, distrbute and sell theses monde, a des fins commerciales ou autres, sur worldwide, for commercial or non support microforme, papier, electronique et/ou commercial purposes, in microform, autres formats. paper, electronic and/or any other formats.

The author retains copyright L'auteur conserve la propriete du droit d'auteur ownership and moral rights in this et des droits moraux qui protege cette these. Ni thesis. Neither the thesis nor la these ni des extraits substantiels de celle-ci substantial extracts from it may be ne doivent etre imprimes ou autrement printed or otherwise reproduced reproduits sans son autorisation. without the author's permission.

In compliance with the Canadian Conformement a la loi canadienne sur la Privacy Act some supporting forms protection de la vie privee, quelques may have been removed from this formulaires secondaires ont ete enleves de thesis. cette these.

While these forms may be included Bien que ces formulaires aient inclus dans in the document page count, their la pagination, il n'y aura aucun contenu removal does not represent any loss manquant. of content from the thesis. Canada To my siblings: Erin, Nathan, Matthew, and Kerrie RESUME

Le diabete mellitus de type I est une maladie de plus en plus abondante. Cette demiere est caracterisee par la destruction auto-immunitaire des ilots de Langerhans incluant les cellules de type p qui produisent de l’insuline dans le pancreas endocrinien. Une option de traitement pour les patients atteints de cette maladie est notamment une greffe des ilots de Langerhans. Ce traitement est limits du au nombre restreint de donneurs d’organes et aussi a la perte de fonctionnalite des ilots suite a la greffe. Les etudes effectuees tout au long de cette these ont pour optique d’adresser ces contraintes par le biais de la science des biomateriaux. La these debute avec un survol details des concepts de base et des complexites associes aux interactions de type cellules et surfaces trouvees dans la litterature. II s’agit specifiquement des interactions physiques et chimiques, des systemes experimentaux ainsi que des caracterisations et modifications associes aux interactions entre cellules et surfaces. La premiere etude de nature experimentale examine la morphogenese des cellules progenitrices ductales (PANC-1 cell line) vers des ilots qui produisent des agregats semblables a des ilots (ILA). Le tout est fait sur des surfaces de carboxymethyl dextrane (CMD) sur lesquelles le RGD est greffe via un lien covalent. L’expression des marqueurs d’lLAs (cytokeratin-19, Ki67, et E-cadherin) qui peuvent etre associes k un changement de phenotype de ces cellules a ete evaluee ainsi que la secretion et 1’expression de l’insuline. La seconde etude de nature experimentale a pour optique 1* immobilisation de la fibronectine (FN) sur les memes surfaces de CMD mentionnees auparavant sur lesquelles des cellules ayant un phenotype fi ( INS-1 cell line) ont prolifere. Lors du processus d’immobilisation, plusieurs solutions ont ete etudiees. L’immobilisation de la fibronectine sur des surfaces de CMD a ete validee par la spectrometrie de photoelectrons induits par rayons X. Le mecanisme d’immobilisation a ete determine par imagerie et mesures de force par microscopie a force atomique, la spectroscopie de dichroisme circulaire ainsi que par la diffusion dynamique de la lumiere. De plus, la croissance des cellules de type INS-1 et la secretion d’insuline ont ete evaluees. La demiere etude de nature experimentale visait l’etude de la coculture des cellules endotheliales et des ilots de pore dans un gel de fibrine. L’effet de la presence des cellules endotheliales sur la production d’insuline des ilots a ete evalue. De plus, l’apoptose cellulaire en coculture a ete evaluee et comparee aux cultures simples.

Mots cl6s: biomateriaux; forces de surface et intermoleculaires; modifications de surface; surfaces anti-adherentes; adsorption de proteines; diabete; ilots de Langerhans; secretion d ’insuline ABSTRACT

Type I diabetes mellitus is a disease of increasing abundance, characterized by the auto-immune destruction of insulin producing p-cells comprising islets of Langerhans in the endocrine pancreas. A treatment option for patients with severe type I diabetes mellitus is the surgical transplantation of islets of Langerghans. This treatment is limited because of pancreas donor scarcity and the progressive loss of islet graft function after both islet isolation from the donor pancreas and transplantation. The present studies within this thesis aim to address these issues by means of biomaterials science. It begins with a detailed broad review of general concepts and complexities associated with cell - surface interaction studies. More specifically physical & chemical interactions, experimental systems, and surface modification & characterization associated with cell - surface interactions are discussed. The first experimental work studies the morphogenesis of ductal-like progenitor cells (i.e., PANC-1 cell line) to insulin expressing and producing islet-like aggregates (ILA) on well-defined low-fouling carboxymethyl-dextran (CMD) surfaces upon which the ligand arginine-glycine-asparic acid is covalently grafted. ILA expression patterns of markers (i.e., cytokeratin-19, Ki67, and E-cadherin) typically associated with a change PANC-1 phenotype were assessed along with insulin expression and secretion. The second experimental study involves the immobilization of the protein fibronectin (FN) onto low-fouling CMD surfaces upon where cells of a P-cell-like phenotype (i.e., INS-1 cell line) were cultured. Various solutions were used during the immobilization process. The successful immobilization of FN on CMD was assessed via X-ray photoelectron spectroscopy and the mechanisms of immobilization were assessed via atomic force microscopy imaging & force measurements, circular dichroism spectroscopy, and photon correlation spectroscopy. INS-1 cell growth and insulin secretion were also assessed. The final experimental work studied the effect of the co-culture of endothelial cells with isolated porcine islets in fibrin gel. The effect endothelial cells had on the capacity of islets to secrete insulin was accessed. In addition, cellular apoptosis was accessed with co-cultures and compared to cultures without endothelial cells.

Keywords: Biomaterials; Surface and Intermolecular Forces; Surface Modification; Low- Fouling; Protein Adsorption; Diabetes; Islets o f Langerhans; Insulin Secretion ACKNOWLEGEMENTS

First and foremost, I would like to thank my mentor and research director Patrick Vermette for the opportunity to perform the boundless scientific research that enlivens me. The past years have been a pleasure and I look forward to our future work together.

I would also like to thank Dr. Rennian Wang and her research group at the University of Western Ontario for providing me with their laboratory and knowledge that gave me the training necessary to perform much of my lab work.

I am also grateful to Dr. Steven Paraskevas and Craig Hasilo at McGill University for providing me with a generous quantity of isolated islets of Langerhans that was used to access the feasibility of my final experimental study.

A very special thank you goes to Dr. Jonathan Lakey, Morgan Lamb, and their entire laboratory for their continuous and unrestrictive supply of islets throughout my final experimental study. Such work would not have been performed without such generosity.

I would also like to thank Mathew Evans for his assistance in the French language.

Finally, I would like to thank all my colleagues over the years at the Universite de Sherbrooke for providing a pleasant work environment and members of the jury (i.e., Dr. Jonathan Lakey, Dr. Steven Paraskevas, Dr. Marie-Josee Boucher, and Dr. Marcel Lacroix) for providing their insights and for their longevity in reading a particularly lengthy thesis. TABLE OF CONTENTS

RESUME...... i ABSTRACT...... ii ACKNOWLEGEMENTS...... iii LIST OF FIGURES...... viii LIST OF TABLES...... xii LIST OF SYMBOLS...... xiii LIST OF ACKRONYMS...... xv Chapter 1 INTRODUCTION ...... 1 Chapter 2 Bridging the Gap Between Physicochemistry and Interpretation Prevalent in Cell - Surface Interactions ...... 5 Foreword ...... 6 2.1 Abstract ...... 8 2.2 Introduction ...... 9 2.3 Cellular Adhesion to Surfaces ...... 13 2.3.1 Cell Surface...... 13 2.3.2 Cellular Adhesion ...... 15 2.3.3 Protein Mediation between Cells and Surfaces ...... 17 2.4 Interactions in Cellular Adhesion ...... 21 2.4.1 Chemical Interactions ...... 22 2.4.2 Physical Interactions ...... 23 2.4.3 Interaction Measurements ...... 33 2.5 Systems for Studying Cell-Material Interactions ...... 34 2.5.1 In Vivo Systems...... 34 2.5.2 Cells in a Three-Dimensional Environment ...... 35 2.5.3 Differences between Two-Dimensional and Three-Dimensional Cell Culture Systems 37 2.5.4 Properties of Cell Culture Systems ...... 40 2.5.4.1 Mechanical Properties ...... 40 2.5.4.2 Biochemical Properties ...... 45 2.5.4.3 Transport Properties ...... 48 2.5.5 Materials for Three-Dimensional Cell Culture Systems ...... 49 2.5.5.1 Natural Materials...... 49 2.5.5.2 Synthetic Materials ...... 51 2.5.6 Micropattemed Cell Cultures ...... 52 2.5.7 Current State and Future Direction of Three-Dimensional Cell Culture Systems.. 54 2.6 Modification, Properties, and Analysis of Surfaces ...... 55 2.6.1 Modified Surfaces in Biomaterials Science ...... 55 2.6.2 Surface Fabrication ...... 56 2.6.3 Self-Assembled Monolayers (SAMs) ...... 59 2.6.4 Plasma Polymerization ...... 62 2.6.5 Properties and Analysis of Modified Surfaces in Cell Science ...... 65 2.6.6 Contact Angle in Correlating Cell Responses toward Biomaterial Surfaces 67 2.6.6.1 Hydrophobic Interactions ...... 69 2.6.6.2 Contact Angle Measurements ...... 71 2.6.6.3 Contact Angle Hysteresis ...... 72 2.6.6.4 Meaning and Relevance of Contact Angle Measurements ...... 73 2.6.7 Low-Fouling Surfaces ...... 76 2.7 Conclusions ...... 83 2.8 Acknowledgment ...... 88 Chapter 3 In Vitro Morphogenesis of PANC-1 Cells into Islet-Like Aggregates Using RGD- Covered Dextran Derivative Surfaces ...... 89 Foreword ...... 90 3.1 Abstract ...... 92 3.2 Introduction ...... 93 3.3 Materials and Methods ...... 95 3.3.1 Surface Preparation ...... 95 3.3.2 X-ray Photoelectron Spectroscopy ...... 97 3.3.3 Cell Culture...... 97 3.3.4 Immunofluorescence ...... 98 3.3.5 Morphometric Analysis ...... 99 3.3.6 Phase Contrast Microscopy ...... 99 3.3.7 Insulin Secretion ...... 100 3.3.8 Proliferation Assay ...... 100 3.3.9 Glycosaminoglycan Staining ...... 100 3.3.10 Data and Statistical Analysis ...... 101 3.4 Results...... 101 3.4.1 X-ray Photoelectron Spectroscopy ...... 101 3.4.2 RGD-CMD Surfaces Promote Islet-Like Aggregate (ILAs) Formation ...... 102 3.4.3 RGD-CMD Surfaces Result in Insulin-Secreting ILAs ...... 105 3.4.4 RGD-CMD Surfaces Promote Cell Adhesion and Increase Cell Proliferation.... 107

3.4.5 Integrins as and avP 3 are Expressed in RGD-CMD ILAs ...... 108 3.5 Discussion ...... 108 3.6 Conclusion ...... 112 3.7 Acknowledgment ...... 112 Chapter 4 Solution Composition Impacts Fibronectin Immobilization on Carboxymethyl-Dextran Surfaces and INS-1 Insulin Secretion ...... 113 Foreword ...... 114 4.1 Abstract ...... 116 4.2 Introduction...... 117 4.3 Materials and Methods ...... 119 4.3.1 Surface Preparation ...... 119 4.3.2 X-ray Photoelectron Spectroscopy ...... 120 4.3.3 Atomic Force Microscopy ...... 120 4.3.4 Circular Dichroism Spectroscopy ...... 121 4.3.5 Photon Correlation Spectroscopy ...... 122 4.3.6 Cell Culture...... 122 4.3.7 Cell Number ...... 122 4.3.8 Glucose-Stimulated Insulin Secretion ...... 123 4.3.9 Statistical Analysis ...... 123 4.4 Results and Discussion ...... 124 4.4.1 Fibronectin Grafting onto CMD Surfaces Occurs Only with Solutions of Low Salt Concentration ...... 124 4.4.2 Solution Composition Does Not Significantly Alter Fibronectin Conformation/Size 125 4.4.3 Topography of Fibronectin Surfaces Made in CaCl2 Solution Differed Subtly From Those Made in MgCl2 and MnCl2 ...... 127 4.4.4 Solution Composition During Fibronectin Immobilization Does Not Affect INS-1 Cell Growth But Does Affect Insulin Secretion ...... 130 4.5 Conclusion ...... 134 4.6 Acknowledgment ...... 135 Chapter 5 Co-Culture of Young Porcine Islets with Liver Microvascular Endothelial Cells in Fibrin Improves Insulin Secretion and Reduces Apoptosis ...... 136 Foreword ...... 137 5.1 Abstract ...... 139 5.2 Introduction ...... 140 5.3 Materials and Methods ...... 141 5.3.1 Islet Isolation and Maturation of Porcine Islets...... 141 5.3.2 Cell Culture, Islet Culture, Fibrin Gels, and Time Lapse Microscopy ...... 142 5.3.3 Glucose-Stimulated Insulin Secretion ...... 143 5.3.4 Immunohistological Staining and Morphometric Analysis ...... 144 5.4 Results and Discussion ...... 144 5.4.1 Time Lapse Microscopy ...... 144 5.4.2 Insulin Content in Media and Response to Glucose Stimulation ...... 145 5.4.3 Porcine Islet Insulin Expression and Cellular Apoptosis ...... 149 5.5 Conclusion ...... 153 5.6 Acknowledgements ...... 153 Chapter 6 CONCLUSIONS ...... 154 LIST OF REFERENCES...... 161 APPENDIX A ...... 207 APPENDIX B ...... 211 APPENDIX C ...... 214 LIST OF FIGURES

Figure 2.1: (A) Traditional cell-surface experimental setup. (B) Cross section of generic eukaryotic cell surface...... 10 Figure 2.2: Structure and dynamics of cellular adhesion to surfaces ...... 17 Figure 2.3: Primary structure of fibronectin ...... 19 Figure 2.4: Double-layer like interaction between a modified substrate bearing a charge and a cell surface with charge bearing receptors ...... 27 Figure 2.5: Steric repulsion of proteins from a polymer-grafted surface ...... 31 Figure 2.6: Interaction potentials of surface and intermolecular forces in biological systems 32 Figure 2.7: Key conceptual differences between 2D and 3D culture systems ...... 36 Figure 2.8: Human fibroblasts project a dendritic network of extensions in collagen matrices but not on collagen-coated cover slips ...... 38 Figure 2.9: In vivo 3D matrix adhesions differ from focal or fibrillar adhesions on 2D substrates...... 46 Figure 2.10: Self-assembled monolayer exposed to proteins and cells ...... 59 Figure 2.11: Surface immobilized polymer conformations ...... 77 Figure 2.12 Dependence of CMD conformation on experimental conditions and its effect on cellular resistance ...... 80 Figure 2.13: Primary and secondary adsorption of proteins on low-fouling surfaces ...... 82

Figure 3.1: Biomaterial surface preparation scheme ...... 96 Figure 3.2: XPS survey and high resolution C Is spectra ...... 102 Figure 3.3: Immunofluorescence staining for Ki67 (A-C), CK-19 (D-F), and E-cadherin (E-Cad) (G-I) of PANC-1 cells on selected culture conditions over time ...... 104 Figure 3.4: Expression level of Ki67 (A), CK-19 (B), and E-cadherin (E-Cad) (C) of PANC-1 cells cultured on the different surfaces over time ...... 105 Figure 3.5: (A) ILA insulin secretion after 14 days. (B) ILA insulin secretion per million cells after 14 days. (C) ILA insulin and glucagon expression after 14 days for CMD ...... 106 Figure 3.6: (A) Phase contrast microscopy of PANC-1 aggregation to ILAs and expansion in SFM. (B) Total cellular content measured by the CyQUANT® assay during SFM culture ...... 107 Figure 3.7: Representative ILA as integrin expression after 14 days for TCPS (A). av03 integrin expression after 14 days for RGD-CMD (B) ...... 108

Figure 4.1: (A) Elemental surface composition via XPS survey. (B) High resolution C Is XPS spectra of FN-CMD surfaces and intermediate layers with EDC/NHS activation . (C) High resolution C Is XPS spectra of FN-CMD surfaces and intermediate layers without EDC/NHS activation ...... 125 Figure 4.2: (A) Circular dichroism spectra of FN in various aqueous solutions. (B) Mean hydrodynamic diameter, Dh of lOOpg/mL FN in various aqueous solutions measured via PCS. 126 Figure 4.3: (A-C) 750nm x 750nm AFM images in liquid (i.e., lx PBS) using Tapping™ mode of FN-CMD grafted in the presence of lOpM CaCL, IOjiM MgCL, and lOpM MnCL, respectively. (D) RMS roughness of FN-CMD surfaces, CMD surfaces in various solutions, and HApp-modified and clean glass substrates ...... 128 Figure 4.4: (A) INS-1 cell numbers on surfaces measured via CyQUANT®. (B) INS-1 cell insulin secretion following glucose stimulation after 7 days of culture. (C) Stimulation index for INS-1 cells from 2.8mM Glucose to 28mM Glucose ...... 131

Figure 5.1 : Time lapse microscopy of: (A) PLMEC+PI(in Fibrin) - porcine islets co-cultured with liver microvascular endothelial cells embedded in fibrin. (B) PI(in Fibrin) - porcine islets embedded in fibrin. (C) PI-PLMEC monolayer - porcine islets on top of a liver microvascular endothelial cell monolayer ...... 145 Figure 5.2: Insulin content of basal media measured via ELISA ...... 146 Figure 5.3: Insulin content and stimulation index of glucose-stimulated insulin secretion assay for (A) 3D PLMEC+PI(in Fibrin) and PI(in Fibrin) cultures and (B) 2D PI-PLMEC monolayer and PI-TCPS cultures ...... 148 Figure 5.4: Sample immunofluorescence images of insulin and glucagon expression and immunohistochemistry images of apoptosis (i.e., TUNEL) of porcine islets for all culture condition at 2 days and 7 days ...... 149 Figure 5.5: (A) Expression levels of insulin and glucagon and 0 day, 2 days, and 7 days. (B) Percent loss of insulin expressing P-cells from 2 days to 7 days. (C) Expression of apoptotic cells measured via TUNEL...... 151

Figure A. 1: Phase contrast microscopy of adherent (RGD-CMD) surfaces and non-adherent (RGE-CMD, CMD, and TCPS) surfaces ...... 207 Figure A.2: Stained sections for GAG using Alcian Blue of (A) PANC-1 ILAs (14 days) and (B) mouse intestine as a positive control ...... 208 Figure A.3: Immunofluorescence staining of as integrin of cells grown on RGD-CMD (A-C), RGE-CMD (D-F), CMD (G-I), and TCPS (J-L) ...... 209

Figure A.4: Immunofluorescence staining of avp 3 integrin of cells grown on RGD-CMD (A-C), RGE-CMD (D-F), CMD (G-I), and TCPS (J-L) ...... 210

Figure B.l: High resolution C Is XPS spectra outlining differences between successful (lOpM CaC^) and unsuccessful (lx PBS and 150mM NaCl) FN grafting to CMD surfaces (A) and to CMD surfaces activated (act.) and non-activated (no act.) with (B) CaCh, (C) MgCL, and (D). MnCl2...... :...... 211 Figure B.2: CMD AFM force - separation distance profiles in various solutions ...... 212 Figure B.3: 750nm x 750nm AFM images in liquid using Tapping™ mode of (A-B) intermediate surfaces and (C-F) CMD layers in various solutions ...... 213

Figure C.l: Time lapse microscopy of porcine liver microvascular endothelial cells (PLMEC) and porcine islets embedded in fibrin gel over 6 days [i.e., PLMEC+PI(in Fibrin)] and PLMEC embedded in fibrin gel after 24h (i.e., 24h-PLMEC) and 6 days (i.e., 6 Days-PLMEC)...... 214 Figure C.2: Time lapse microscopy of porcine islets embedded in fibrin gel [i.e., PI(in Fibrin)] over 6 days ...... 215

x Figure C.3: Time lapse microscopy of porcine islets seeded over a monolayer of porcine liver microvascular endothelial cells (i.e., PI-PLMEC monolayer) ...... 216 Figure C.4: Time lapse microscopy of porcine islets seeded over tissue culture polystyrene (i.e., PI-TCPS). Single cells of fibroblastoid morphology appear early and proliferate throughout the culture period ...... 217 Figure C.5: Immunofluorescence image of insulin and glucagon expression at 0 days (left). Immunohistochemistry image of apoptosis detected via TUNEL at 0 days (right) ...... 218 Figure C.6: Level of cells expressing glucagon in all culture conditions ...... 218 LIST OF TABLES

Table 2.1: Cell-surface studies ...... 12 Table 2.2: Cellular attachment to fibronectin domains ...... 20 Table 2.3 Surface modification substrates and chemicals ...... 57 Table 2.4: Properties of cell culture surfaces ...... 58 Table 2.5: Static water contact angle and cell/protein responses ...... 68 Table 2.6: Advancing, receding, and dynamic water contact angle and cell/protein responses... 69

Table 3.1: The total number and mean diameter of islet-like aggregates (ILAs) obtained from PANC-1 cells cultured on the different surfaces after 7 days on average ...... 103 LIST OF SYMBOLS

Chapter 2 a Length of monomer comprising polymer chain

C e s Physical constant (incorporating surface geometry, charge, and solution conditions) of Double Layer interaction

C v d w Geometric constant (analogous to Hamaker constant) of van der Waals interaction

8 Diameter of solvent molecules

D Distance between adsorption points of polymer on a surface e Dielectric permittivity of interacting medium e 0 Permittivity of free space

0 Water contact angle

0a d v Advancing water contact angle

0d y n Dynamic water contact angle

0r e c Receding water contact angle

0stat Static water contact angle

Eo Contact energy k Debye length

Lo Thickness of polymer adlayer in free form q Magnitude of electric charge r Separation distance

Rg Radius of gyration

a Surface density of adsorbed polymer

w(r) Interaction potential Chapter 3 a Equivalent diameter of prolate spheroid b Polar diameter of prolate spheroid c Equatorial diameter of prolate spheroid Chapter 4 c Molar concentration

Dh Hydrodynamic diameter

1 Path length

0 Ellipicity

[0] Molar ellipicity

xiv LIST OF ACKRONYMS

2D Two Dimensional

3D Three Dimensional

Act. Activated with EDC( 1 -ethyl-3 -(3 - dimethylaminoproplyl)carbodiimide) and NHS (N- hydroxysuccinimide)

ADV Advancing

AFM Atomic Force Microscopy

ANOVA Analysis of Variance

Au( 111) Miller Oriented (lmn) Gold Surface

BM Basement Membrane

BSA Bovine Serum Albumin

CAD Computer-Aided Design

CCBD Central Cell Binding Domain

CD Circular Dichroism

CMD Carboxymethyl-Dextran

CMRL-1066 Connaught Medical Research Laboratories medium

CK-19 Cytokeratin-19

DAPI 4\6-diamidino-2-phenylindole, dihydrochloride

DL Double-Layer

DLVO Deijaguin-Landau-Verway-Overbeek

xv DMEM Dulbecco’s Modified Eagle Medium

DYN Dynamic

E-Cad E-Cadherin

ECM Extracellular Matrix

EDC l-ethyl-3-(3-dimethylaminoproplyl)carbodiimide

ELISA Enzyme-Linked Immunosorbent Assay

EMET Epithelial to Mesenchymal to Epithelial Transition

EMT Epithelial to Mesenchymal Transition

ERK Extracellular Signal-Regulated Kinase

FAK Focal Adhesion Kinase

FBS Foetal Bovine Serum

FCCS Fluorescent Cross-Correlation Spectroscopy

FCS Fluorescent Correlation Spectroscopy

FEP Fluorinated Ethylene Propylene

FGF Fibroblast Growth Factor

Fn Fibronectin (Chapter 2)

FN Fibronectin (Chapter 4)

FN-CMD Fibronectin covalently grafted to a Carboxymethyl Dextran surface

FRAP Fluorescence-Recovery After Photobleaching

GAG Glycosaminoglycan

GBD Gelatin Binding Domain

xvi GRGDS Glycine- Arginine-Glycine-Asparic Acid-Serine

GRGES Glycine-Arginine-Glycine-Glutamic Acid- Serine

GSIS Glucose-Stimulated Insulin Secretion

HApp w-heptylamine Plasma Polymerization

HBD Heparin Binding Domain

H-bond Hydrogen Bond

HBSS Hank’s Balanced Salt Solution

HGF Hepatocyte Growth Factor

'H-NMR Hydrogen-1 (Proton) Nuclear Magnetic Resonance

HUVEC Human Umbilical Vein Endothelial Cell

IC Immunochemistry

ICCS Image Cross-Correlation Spectroscopy

ICS Image Correlation Spectroscopy

ILA Islet-Like Aggregate

INS-1 Insulinoma Cell Line

IBMX 3-isobutyl-1 methylxanthine

IEQ Islet Equivalent

LPA Lysophosphatidic Acid

LR Ligand-Receptor

MALDI Matrix-Assisted Laser Desorption/Ionization

MAP Mitogen Activated Protein

xvii NHS N-hydroxysuccinimide

No Act. Not Activated with EDC( 1 -ethyl-3-(3- dimethylaminoproplyl)carbodiimide) and NHS (N- hydroxysuccinimide)

OWLS Optical Waveguide Lightmode Spectroscopy

PANC-1 Pancreatic Carcinoma Cell Line

PBS Phosphate Buffered Saline

PC Phosphatidylcholine

PCR Polymerase Chain Reaction

PCS Photon Correlation Spectroscopy

PDGF Platelet Derived Growth Factor

PEG Poly(Ethylene Glycol)

PEO Poly(Ethylene Oxide)

PFC Perfluorocarbon

PHSRN Proline-Histidine-Serine-Arginine-Asparagine

PI Porcine Islets

PI(in Fibrin) Porcine Islets embedded in Fibrin gel

PI-PLMEC(in Fibrin) Porcine Islets and Porcine Liver Microvascular Endothelial Cells embedded in Fibrin gel

PI-PLMEC monolayer Porcine islets on a Porcine Liver Microvascular Endothelial Cell monolayer

PI-TCPS Porcine islets on Tissue Culture Polystryrene

xviii PLMEC Porcine Liver Microvascular Endothelial Cells

PTFE Polytetrafluoroethylene

QCM Quartz Crystal Microbalance

REC Receeding

RF Radio Frequency

RGD Arginine-Glycine-Asparic Acid

RGD-CMD Arginine-Glycine-Asparic Acid covalently grafted to a Carboxymethyl Dextran surface

RGE Arginine-Glycine-Glutamic Acid

RGE-CMD Arginine-Glycine-Glutamic Acid covalently grafted to a Carboxymethyl Dextran surface

RMS Root Mean Square

RPMI Roswell Park Memorial Institute medium

SAM Self-Assembled Monolayer

SCM Serum Containing Media

S.E.M. Standard Error of the Mean

SFA Surface Force Apparatus

SFM Serum-Free Media

SIMS Secondary Ion Mass Spectrometry

SMC Smooth Muscle Cell

SPR Surface Plasmon Resonance

STICS Spatio-Temporal Image Correlation Spectroscopy

xix Sulfo-SMCC Sulfosuccinimidyl-4-(V-Maleimidomethyl)cyclohexane-1 - carboxylate

TCPS Tissue Culture Polystyrene

TUNEL Terminal deoxynucleotidyl transferase dUTP Nick End Labelling

UV Ultra-Violet

VDW van der Waals

VEGF Vascular Endothelial Growth Factor

Vn Vitronectin

XPS X-ray Photoelectron Spectroscopy

YIGSR Tyrosine-Isoleucine-Glycine-Serine-Arginine

xx Chapter 1 INTRODUCTION

i Diabetes mellitus largely exists in two forms; type I diabetes characterized by the lack of insulin production and type II characterized by ineffective insulin use. According to the World Health Organization, approximately 346 million people worldwide have diabetes, 90% being Type II. Symptoms of diabetes include excessive thirst, frequent urination, hunger, and fatigue. Complications of diabetes include heart disease, stroke, blindness, neuropathy, and kidney failure to name a few. It is uncommon but of increasing occurrence, that cases of type I and type II diabetes are diagnosed in the same patient [1]. The Canadian Diabetes Association predicts diabetes to cost Canada $16.9 billion in the year 2020 alone.

Type I diabetes results in the auto-immune destruction of insulin producing P-cells that comprise islets of Langerhans in the endocrine pancreas. Treatment of type I diabetes often involves insulin therapy which involves rigorous monitoring of glucose levels in the blood and subsequent exogenous injections of insulin. Avoidance of the numerous complications associated with type I diabetes requires particular patient compliance to monitor blood glucose levels and inject insulin when necessary. However, despite insulin therapy, for some patients, blood-glucose levels remain uncontrollable. Such extreme cases can require islet transplantation as an option of therapy.

Islet transplantation has shown particular promise [2]. Islet transplantation involves the isolation of islets from a donor pancreas, typically followed by infusion into the hepatic portal vein of the patient [3]. The primary limitation of islet transplantation is a severe shortage of donor islets for transplantation. Another limitation is that islet isolation results in the loss of islet function and eventual cell death [4]. This may contribute to the progressive loss of islet graft function (i.e., insulin secretion) in transplanted patients. The present thesis aims to help address these limitations through two means: through a broad review on the fundamental intermolecular and surface forces involved in cell - surface interactions, and application of such fundamentals to biomaterial surface and scaffold design upon which the response of cells of pancreatic endocrine relevance are assessed.

Three types of cells/tissue of endocrine pancreatic relevance were utilized. First, Pancreatic Carcinoma (PANC-1), a cell line believed to be analogue to pancreatic ductal progenitor cells that arguably differentiate into islets [5]. PANC-1 cell response to biomaterial surfaces was thus accessed with the aim of producing surrogate insulin secreting tissue; an attempt to address the shortage of donor pancreases. Second, Insulinoma (INS-1) were utilized, a cell line that expresses and secretes insulin in a glucose responsive manner, analogue to those of P-cells comprising islets [6]. INS-1 cell response towards biomaterials surfaces was assessed with the aim of improving insulin secretion to glucose stimulation. This was done with the hope of addressing the reduced functionality of p-cells comprising islet grafts through the identification of an additional factor (i.e., solution composition during protein adsorption) typically not considered when designing biomaterial surfaces to improve isolated islet functionality. Specifically the hope was that this would help further improve insulin secretion to glucose stimulation. Finally, islets isolated from porcine pancreases were utilized. Isolated islets were co-cultured with endothelial cells in a biomaterial scaffold with the aim to improve insulin secretion and reduce cell death by apoptosis. The use of isolated islets attempts to address again, the progressive loss of islet insulin secreting capacity after isolation from the pancreas.

Chapter 2 of this thesis titled, “ Bridging the Gap Between Physicochemistry and Interpretation Prevalent in Cell - Surface Interactions ”, consists of a broad review on the fundamental intermolecular and surface forces that are involved in cell - surface interactions. Such forces are the origin in a series of actions and reactions that ultimately govern cell behavior (i.e., adhesion, spreading, migration, proliferation, differentiation, and function). Therefore, the review is not only relevant to the application of biomaterials science as a therapeutic strategy to diabetes mellitus, as is done in this thesis, but also to the application of biomaterials science to many other therapeutic strategies. Within the review, the cell surface and cellular adhesion of cell surface receptors to ligands are discussed. Fundamental interactions such as Coulomb, van der Waals, double-layer, hydrophobic, H-bonding, and steric repulsion along with their relevance to biology are discussed. Also, systems [two-dimensional (2D) and three-dimensional (3D)] for studying cell - material interactions are presented. Finally, surface modification, properties, chemistry, and characterization is presented and discussed. Throughout the entire review, the advantages, disadvantages, and complexities of all topics are discussed.

In Chapter 3 titled, “/« vitro Morphogenesis o f PANC-1 Cells into Islet-Like Aggregates Using RGD-Covered Dextran Derivative Surfaces ”, PANC-1 cells were exposed to biomaterial surfaces with well-defined properties and their possible transformation into surrogate insulin expressing and producing tissue was assessed. Briefly, low-fouling carboxymethyl dextran

3 (CMD) surfaces bearing covalently grafted arginine-glycine-asparic acid (RGD) [i.e., RGD- CMD] appear to be the most promising as they not only express insulin (as the controls did as well) but had the most insulin in the culture media.

For Chapter 4 titled “ Solution Composition Impacts Fibronectin Immobilization on Carboxymethyl-Dextran Surfaces and INS-1 Insulin Secretion ”, fibronectin was covalently grafted onto low-fouling CMD surfaces using various solutions. It was found that not only does solution composition play a role in subsequent INS-1 insulin secretion on fibronectin-bearing CMD surfaces, but it also impacts whether or not fibronectin can in actual fact be grafted onto low-fouling CMD surfaces.

Chapter 5 titled, “ Co-Culture of Young Porcine Islets with Liver Microvascular Endothelial Cells in Fibrin Improves Insulin Secretion and Reduces Apoptosis ”, involves studying the effect endothelial cells have on isolated islet insulin secretion and programmed cell death (i.e., apoptosis). Briefly, it was found that endothelial cells have beneficial effects on insulin secretion whether or not one employs a 2D or 3D cell culture system. Nevertheless, there was a more pronounced improvement in insulin secretion capacity and further reduction in apoptosis when using 3D fibrin cultures in conjunction with endothelial cells.

Finally, Chapter 6, titled “ Conclusion ” presents general insights into each experiment. In addition, future direction is suggested.

4 Chapter 2 Bridging the Gap Between Physicochemistry and Interpretation Prevalent in Cell - Surface Interactions Foreword

Authors and Affiliation :

Evan A. Dubiel : Ph.D. Candidate, Universite de Sherbrooke, Departement de genie chimique et de genie biotechnologique

Yves Martin : Lecturer, Universite de Sherbrooke, Departement de genie chimique et de genie biotechnologique

Patrick Vermette : Professor, Universite de Sherbrooke, Departement de genie chimique et de genie biotechnologique

Date of Acceptance : October 22, 2010

State of Acceptance : Final Version Published

Jo u rn al: Chemical Reviews

Reference :

[Dubiel, E.A. and Vermette, P. (2009) Complexities in Understanding Cellular Interactions to Modified Surfaces, 27th Annual Conference o f the Canadian Biomaterials Society, Laval University, Quebec City, Quebec, Canada]

[Dubiel, E.A., Martin, Y., and Vermette, P. (2011) Bridging the Gap Between Physicochemsitry and Interpretation Prevalent in Cell - Surface Interactions, Chemical Reviews, volume 111, p. 2900-2936]

Contribution :

This article contributes to the literature review component of this thesis. The abstract was presented at the 27 th Annual Conference o f the Canadian Biomaterials Society. The content of the article was published in Chemical Reviews (note that Chemical Reviews does not publish abstracts). Section 2.5 was written by Yves Martin (except Section 2.5.2, 2.5.4.1-2.5.4.2, and 2.5.6 where Evan A. Dubiel also contributed). Section 2.6.6 was written by Patrick Vermette. All other sections were written by Evan A. Dubiel. All work was done under the direction and supervision of Patrick Vermette.

6 Titre en fran$ais :

Le rapprochement entre la physico-chimie et 1’interpretation des mecanismes gouvemant les interactions entre les cellules et les surfaces

R6sum6:

Les interactions entre les cellules et les surfaces modifiees sont d’une grande importance dans le domaine de la medecine. La comprehension de ces demieres implique plusieurs aspects s’etendant dans les domaines de la biologie, de la chimie ainsi que la physique. Le nombre de projets de recherche effectuee dans ce domaine a augmente depuis les dix demieres annees. Malgre ce fait, la comprehension des phenomenes d’interaction entre cellules et surfaces reste tres mauvaise. Ce manque est du notamment a la variation des parametres lors des experiences. Les interactions sont typiquement deduites suite a une periode d’exposition predeterminee des cellules a une surface modifiee. II existe typiquement trois couches d’interactions dans un systeme cellule - surface : (i) un substrat, (ii) une couche intermediaire, comprenant notamment des proteines ainsi que certaines macromolecules et (iii) la cellule elle-meme. II existe done trois interfaces colloi'dales d’interaction. Lors de l’analyse de la reponse des cellules, les forces surfaciques et intermoleculaires incluant notamment la force Coulomb, les liens van der Waals, la force de double - couche et la force sterique a l’interface et a l’interieur de l’interface sont souvent ignores. Par contre, ces forces se trouvent a l’origine d’une serie d’actions et de reactions gouvemant le comportement des cellules (l’adhesion, l'etalement, la migration, la proliferation ainsi que la differentiation). Les forces mentionnees auparavant influencent les proprietes du substrat modifie tout en influen?ant radicalement la conformation de la proteine adsorbee. Ces demieres agissent en tant que mediateur dans le processus de l’adhesion cellulaire et elles gouvement ainsi le comportement cellulaire. Malheureusement, du a la complexity des systemes biologiques, il est difficile d’etablir et de correler I’effet de ces forces sur le phenomene de l’adhesion cellulaire et le comportement subsequent de ces demieres. De plus, l’utilisation des monocouches auto assemblies et de la polymerisation par plasma comme outils de modification des proprietes de surface possede plusieurs contraintes nefastes souvent ignorees. Certains de ces effets nefastes proviennent du fait que les forces surfaciques et intermoleculaires sont souvent negligees et du fait que la caracterisation de surface est inappropriee. Ces actions entrainent souvent l’obtention des proprietes de surface inattendues menant k des resultats d’interactions cellulaires mal guidees. En resume, il existe une abondance marquee de publications impliquant des interactions cellulaires avec des surfaces modifiees, mais il existe peu ou aucune comprehension generale du phenomene decrit.

7 2.1 Abstract

Cellular interactions on modified surfaces have a wide variety of applications in medicine. Understanding cellular interactions to modified surfaces involves aspects of biology, chemistry, and physics. The past decade has seen an explosion of research in the area. However, despite such a growth in research, very little comprehensive understanding on cell interactions to modified surfaces exists. This is partly due to variations in experimental parameters. Typically, cell interactions are assessed after a randomly chosen period of exposure to a modified surface and culture condition. Generally, there are three layers of interaction in a cell - surface system: (i) an inert substrate, (ii) a multi-component intermediate layer consisting of proteins and other macromolecules and, (iii) the cell. Thus there are three colloidal interfaces of interaction. Surface and intermolecular forces including Coulomb, van der Waals, double - layer, and steric to name a few, at and within each interface, are often overlooked when studying cell response to modified surfaces. However, these forces are in actual fact the origin in a series of actions and reactions that ultimately govern cell behavior (i.e. adhesion, spreading, migration, proliferation, and differentiation). Such forces influence the properties of the modified substrate dictating the conformation of the adsorbed proteins that mediate cell adhesion thus governing subsequent cell behavior. Unfortunately, the complexity of biological systems makes it difficult to explicitly correlate how these forces effect the dynamics of cellular adhesion and subsequent cell behavior. Furthermore, the use of self - assembled monolayers and plasma polymerization to modify the properties of surfaces have complications that are often overlooked. Such complications manifest from neglecting surface and intermolecular forces, and inadequate surface characterization resulting in unintended surface properties leading to misleading results in cellular interactions with surfaces. Thus while there is an abundance of literature on cellular interactions to modified surfaces, little to no general understanding exists.

8 2.2 Introduction

Biomaterials science and tissue engineering involve the study and design of materials capable of eliciting a desired and controllable response from cells and biological environments. Research efforts studying the processes at interfaces between engineered synthetic materials and biological environments have exploded in the past two decades [7]. Such research applies to the design of surfaces for medical implantation ranging from dental implants [8-10], contact lenses [11-13], drug delivery [14-16] and the growth of tissue substitutes in scaffolds [17-19]. Particular emphasis is placed on engineering surfaces that would rapidly promote the integration of an implant {e.g., artificial joints) into the host to reduce healing time [20]. In conjunction, many efforts have also focused on developing surfaces for the prevention of inflammation and thrombosis of implanted tissue and foreign materials [21-25]. These studies highlight only some of the research in biomaterials science and tissue engineering. Within biomaterials science, tissue engineering, and the numerous other sub-disciplines that have emerged over the past two decades are cell-material and cell-surface interactions.

Many studies involving cell-surface and cell-material interactions focus solely on an end result after almost randomly chosen periods of times and culture conditions. Less emphasis is placed on understanding the mechanisms that influence cell behavior on a surface. As a result, there is an abundance of literature on information regarding the behavior and response of cells towards surfaces but limited understanding of the events and conditions that factor into such behavior. One of the aims of this article is to outline the influence surface and intermolecular forces have on the response and behavior of cells on surfaces. Particularly, emphasis is placed on the influence of electrostatic and van der Waals forces, hydrophobic interactions, hydrogen bonding, and steric repulsion.

A traditional cell-surface colloidal system generally consists of four interfaces: (i) substrate- surface modification, (ii) the surface modification-biomolecule, (iii) the biomolecule-cell, and (iv) the cell-culture medium. As shown in Figure 2.1, the system is typically engineered in a “bottom-up” fashion. For example, a substrate is modified by means of self-assembled monolayers (SAM) or plasma polymerization. Subsequently, proteins, extracellular matrix (ECM) components, or other biomolecules, such as growth factors are adsorbed to the surface

9 via chemisorption or physisorption, followed by cell seeding in culture media. While almost all studies employ such a colloidal system, its relevance to the in vivo reality is a credible issue. As observations suggest, the traditional cell culture approach is not necessarily representative of the architecture, mechanical forces, and mass transport properties of a physiological environment. As such, large efforts should be made in biomaterials and tissue engineering to design in vitro systems that further represent the physiological in vivo reality. While advances have been made and some studies have in actual fact elicited a desired cell response, much research is still required to truly understand cell-surface and cell-material interactions in order to significantly advance not only molecular biology, but also biomaterials and tissue engineering research.

(A)

Cell - culture medium

Biomolecule - cell

Surface modification - biomofecule

Substrate - surface modification

(B)

Carbohydrate Glycolipid Carbohydrate side chain V Integral protein Peripheral Celf Exterior protein

Plasma Membrane

Cholesterol Monotopic Cell Interior Phospholipids Transport integral protein Lipid anchored protein protein

Figure 2.1: (A) Traditional cell-surface experimental setup. (B) Cross section of generic eukaryotic cell surface (i.e., plasma membrane). Many of the components portray a localized net electric charge. Figures not drawn to scale.

10 Many studies have been performed on cell adhesion and responses to surfaces using a variety of phenotypes and surface modifications [26-36]. As shown in Table 2.1, results of previous studies on cell adhesion and responses to surfaces always vary somewhat between subsequent studies. The differences in results can be a consequence of varying phenotypes, surface modification, and/or a variety of other differences in experimental methods. For example, both

Webb et al. [36] and Keselowsky et al. [30] studied osteoblast adhesion to amine (NH 2)- and methyl (CH 3)-bearing surfaces. Webb et al. found osteoblast activity (adhesion and migration) followed CH 3 > NH2 while Keselowsky et al. found the opposite. Between the two studies there are many differences in the experimental processes. Webb et al. [36] modified substrates with organosilanes while Keselowsky et al. [30] used organothiol SAM. Webb et al. [36] did not cover the modified surface with proteins while Keselowsky et al. [30] coated the organothiol- modified surface with fibronectin (Fn). Both conclude that cell response and behavior is dictated by the surface chemistry of the modified substrate. While such a conclusion is logical and correct, due to the differences in experimental methods, it is next to impossible to draw a comprehensive understanding on the surface mechanisms that govern osteoblast response and behavior. One can only draw conclusions based on the specific surface composition employed, as these researchers have done. While these studies suggest a useful strategy for understanding and modulating cell behavior, they do not provide insight into the specific mechanisms that govern such behavior, nor strategies to control and modulate such behavior for biomedical applications. Limitations such as these just described can be extended to nearly all (including the few presented in Table 2.1) biomaterials and tissue engineering related literature. The result is a large quantity of literature with limited basis of general comprehensive knowledge and understanding.

In a recent keynote address [37], R. Nerem categorized biomaterials and tissue engineering research over the last two decades. The 1990’s were classified as “the go go years” in which it was seen as a potential scientific revolution and interest and research exploded. The years 2000 to 2005 were categorized as “the sobering years” as researchers began to realize the enormous problems and complexities associated with such research. The year 2005 to present is categorized as “back to the future” as some researchers are now focusing on overcoming the many hurdles to retain the potential that research in biomaterials, tissue engineering, and other related disciplines was believed to have in the 1990’s [37]. We agree with R. Nerem’s categorization and hope this article helps advancing the science to the potential it was initially believed to have.

Table 2.1: Cell-surface studies

Phenotype Modification Ref: Relative level of adherenCe/behavior fibroblast organosilane + SM° [27] NH2 a COOH > CH3 = OH organothiol + PM* [31] TCPSc>CH 3>COOH commercial TCPS [28] UPSrf > TCPS organosilane + SM [35] n h 2 > c h 3 commercial TCPS [33] Primaria™ > TCPS + PM endothelial Commercial TCPS [33] Primaria™ > TCPS + PM organosilane + PM [32] NH2 > glass organosilane [29] NH2>CH osteoblast organosilane [36] CH3 > NH2 organothiol + PM [30] NH2 > CH3 leukocyte organothiol [26] CH3 > OH > COOH organothiol + PM [34] OH>COOH>CH3 organothiol [34] CH3 > COOH > OH aSM = serum modification. PM = protein modification. TCPS = tissue culture polystyrene. ^UPS = untreated polystyrene

We begin with presenting biological aspects. The cell surface {i.e., the plasma membrane) is briefly discussed, followed by cellular adhesion of the cell surface to ligands. Physical aspects are then presented. Fundamental interactions including Coulomb, van der Waals, double-layer, hydrophobic, H-bonding, and steric repulsion are introduced and their relevance to biology is discussed. An extensive discussion on systems for studying cell-material interactions follows. Finally, surface chemistry is presented as it is essential to biomaterials science. Particular emphasis is placed on SAM and plasma polymerization modifications to enhance chemical activity, along with their advantages and disadvantages with respect to surface uniformity and stability. The design of low-fouling surfaces is also discussed.

12 Depending on the scientific background of the reader, sections of this article may appear trivial. However, the primary aim of this review is to collectively present literature relevant to cell- material and cell-surface interactions from all three fundamental scientific disciplines in order to present the reader with a wide spectrum of perspectives of such literature subsequently directing future research considerations and interpretation in the direction of general comprehension of cell behavior on surfaces. We believe that fundamental understanding is necessary to identify and design surfaces, materials, and systems to elicit and control a desired cell response.

This article covers a large timeline of scientific literature, as this was necessary to provide a full picture of the concepts behind cell-material interactions. Most studies directly related to cell- material interactions that are reviewed in our article have largely been published in the last 20 years. At the same time, we also review the fundamental concepts of surface science that are necessary to better understand and appreciate such interactions, some which originated nearly a century ago.

2.3 Cellular Adhesion to Surfaces

2.3.1 Cell Surface

Generally, the cell surface can be depicted as shown in Figure 2.1. The plasma membrane (i.e., cell surface) is selective in the reactions that are allowed to proceed and the molecules that are allowed to penetrate. The plasma membrane mainly consists of a lipid bilayer composed of phospholipids. Scattered throughout the lipid bilayer are membrane proteins and carbohydrates (generically known as polysaccharides).

The plasma membrane is a non-rigid body. Plasma membrane structure is largely represented using the Fluid-Mosaic model proposed by Singer and Nicolson [38]. The Fluid-Mosaic model utilized many previous studies in its formulation [39] and is continuously being redefined. In the Fluid-Mosaic model, lipids and proteins behave like a fluid and for all intensive purposes are free to move around, in a diffusive manner, within the two-dimensional structure of the plasma membrane [40, 41]. The lipid component of the cell surface gives the cell its fluid-like properties. The three primary classes of lipids composing the cell membrane are phospholipids, glycolipids, and cholesterol. The most abundant lipids are the phospholipids. Glycolipids are

13 lipids with a carbohydrate side chain. Cholesterols are located within the interior of the plasma membrane. Carboxyl groups close to the polar head of the phospholipids interact with hydroxyl groups of the cholesterol via H-bonding. The result is increased spatial separation of adjacent polar head groups of the phospholipids. During interaction between a plasma membrane protein (i.e., a receptor) with another ligand, there is an abundance of cholesterol within the vicinity of the membrane protein relative to the plasma membrane as a whole. This further isolates the membrane protein from the rest of the plasma membrane aiding in the ligand-receptor interaction. The formed structure is known as a lipid raft.

The two main functions of membrane proteins are to provide communication between the cell and its external environment (i.e., cell signaling) and to regulate the flow of substances into and out of the cell. A membrane protein (i.e., a cell receptor) interacts with a signaling protein (i.e., a ligand). This is the first in a series of events that dictate subsequent cell function and behavior. Transport proteins are membrane proteins that regulate the flow of substances in and out of the cell. To name a few, substances can include the intake of nutrients or the removal of waste from the cell.

There are many different types of membrane proteins. Classification is dependent on the structure and the manner in which it is integrated to the plasma membrane. For example, membrane proteins containing carbohydrate side chains are called glycoproteins and those that do not penetrate the plasma membrane are known as peripheral membrane proteins. Peripheral membrane proteins are anchored to the plasma membrane via weak electrostatic interactions and H-bonding. Proteins that are covalently bound to the lipid molecules within the plasma membrane are known as lipid-anchored membrane proteins.

The majority of membrane proteins are categorized as integral membrane proteins. Integrins and cadherins are classified as integral membrane proteins. Integral membrane proteins possess hydrophobic regions that interact with the hydrophobic interior of the plasma membrane (i.e., the tails of the phospholipids). As such, there is a high affinity of the protein for the plasma membrane making integral membrane proteins the most difficult to isolate from the plasma membrane. Integral membrane proteins also possess hydrophilic regions by which they interact with the external aqueous environment of the cell. Some integral membrane proteins protrude

14 from only one side of the cell, known as integral monotopic proteins while others span the entire thickness of the plasma membrane, known as transmembrane integral proteins.

This discussion has been overly simplified for the purpose of brevity. We have only discussed the plasma membrane as a barrier that defines the boundary of a cell and its external environment as that is the most relevant for this article. However, the plasma membrane also acts as a boundary for the separation of organelles in the intracellular compartment. For detailed discussion on the overall context of the plasma membrane and more detail into its structure, the reader is referred to widely used biology texts by Lodish et al. [42] and Becker et al. [43]. The interactions (H-bonding and hydrophobic) mentioned here are discussed in detail in Section 2.4.2.

2.3.2 Cellular Adhesion

It is generally believed within the scientific community that cells adhere to surfaces through an intermediate interaction between an extracellular matrix (ECM) protein and a surface. ECM proteins are introduced to the surface by two ways. In one way, cells synthesize and secrete ECM proteins such as fibronectin (Fn) and vitronectin (Vn) to promote adherence [44]. In the other way, proteins are pre-adsorbed to the surface independently prior to the introduction of cells to the surface. Subsequently, when cells are introduced to a protein-coated surface, less protein synthesis via the cells is necessary.

Cellular adhesion is a dynamic process that involves many different proteins [45, 46]. Adherence of cells to surfaces is achieved through an interaction between adhesive ligands on the ECM proteins and receptors on the cell surface. This interaction is mediated by integrins [47]. Integrins are heterodimeric transmembrane glycoproteins that consist of a and P subunits [48]. While we discuss only integrin cell receptors in this article, we note to the reader that there are other known types of cell receptors. In general, most cell receptor molecules fall into five categories: immunoglobins, integrins, selectins, mucins, and cadherins. For a concise review on the different types of cell receptors, the reader is referred to literature by Pierres et al. [49].

The two main classifications of cellular adhesions are focal contacts and fibrillar adhesions. Focal adhesion contacts are flat elongated structures usually located at the cell periphery, which

15 involve the cytoskeletal proteins tensin, vinculin, a-actinin, and talin as well as signaling proteins including focal-adhesion-kinase (FAK) and paxillin. Fibrillar adhesions are more centrally located and involve the cytoskeletal protein tensin [50].

The dynamics of cellular adhesion are hypothetically illustrated in an informative commentary by Zamir and Geiger [51]. a5pl and its interaction with Fn along with the aVp3 integrin appear to be the most studied complexes in conjunction with cellular adhesion. As shown in Figure 2.2, adhesions initially contain both a5pi and aVp3 integrins. Co-localization of a5/aV subunits appears early during adhesion, then over time, these alpha subunits segregate [52]. Moreover, experiments have shown that P3 integrins remain with the initial adhesions while pi undergoes translocation [53]. A contractile force provided by the protein myosin II translocates the a5pi along with tensin centripetally towards the interior of the cell and forms a fibrillar adhesion in a process known as fibrillogenesis. In the same manner as the a5 and pi integrin subunits, Fn has been shown to translocate during cellular adhesion [54]. Furthermore, covalent linking of Fn to a substrate can impede fibrillogenesis [55]. On the other hand, aVp3 does not translocate even though the same contractile force is present leaving what is referred to as a focal adhesion contact [51, 56]. Depending on the substrate properties and composition, focal contacts alone or both focal contacts and fibillar adhesions may be present in cellular adhesions [55, 57].

It should be mentioned, as was done in the studies just highlighted, that there are likely other integrin and ECM complexes involved in cellular adhesion. Most integrin receptors have the ability to bind to multiple adhesive ligands. In addition, an adhesive ligand can bind with multiple integrin receptors [58]. For a concise review on integin a and p subunits and their associated ligand receptors along with insight into the resultant signaling and cell function, we refer the reader to articles by Hynes [48, 59].

16 Figure 2.2: Structure and dynamics of cellular adhesion to surfaces. (A) Cellular adhesions initially contain both a5pi/Fn and avp3/Vn integrin-ECM adhesions. (B) A contractile force provided by myosin II translocates the a5pi/Fn centripetally towards the cell center to form a fibrillar adhesion leaving behind a focal contact. Abbreviations; a = actin, a = a-actinin, m = myosin II, p = parvin/actopaxin, pa = paxillin, ta = talin, te = tensin, and vi = vinculin. Reproduced with permission from reference [51]. Copyright 2001 The Company of Biologists.

2.3.3 Protein Mediation between Cells and Surfaces

As already mentioned, receptors on the cell surface interact with the adhesive ligands of the ECM proteins that are adsorbed to the underlying substrate to form adhesions (i.e., focal and fibrillar structures). The availability of such adhesive ligands for interaction with cells is dependent on interactions between the ECM protein and the underlying substrate. As a result of such interactions, ECM proteins will undergo conformational and orientation changes upon adsorption to a substrate. The nature of the interaction mechanisms (i.e., surface and intermolecular forces) involved is the next subject of this article.

Proteins are non-rigid bodies. Thus, while the primary structure remains intact upon interaction with a surface, the secondary, tertiary, and quaternary structures may undergo spatial changes from their natural conformation. This is known as a conformational change. An orientation is the axial and azimuthal angles of the protein with respect to an arbitrary axis. The resultant conformation and orientation of a protein due to interactions with a surface has a direct impact on which particular ligands are available for interactions with cell surface receptors, thus directly influencing the behavior of cells on the surface.

To provide a specific example of various ligands influencing cell behavior, let us consider Fn. Fn is a dimer glycoprotein of ~ 450kDa found in the ECM and blood. It reacts with many biomolecules and plays a wide variety of roles in physiological processes such as embryogenesis, angiogenesis, and wound healing [60-64].

Plasma and cellular Fn are similar but vary slightly. We assume the structure of plasma and cellular Fn as identical for the purpose of brevity. For differences in the structure of plasma and cellular Fn, we refer the reader to Hayashi and Yamada [65]. Fn has a primary structure as shown in Figure 2.3. There are two similar polypeptide chains of ~ 225kDa each. Each chain has a slight variation in amino acid sequence [66]. There is no absolute general consensus on the specific functions of each of the three domains presented in Figure 2.3 with respect to cellular function. Various phenotypes respond differently to the domains of Fn, particularly, the gelatin binding domain (GBD) and heparin II binding domain (HBD). Such differences are a result of different phenotypes having different cell surface receptors. While one phenotype may have a cell receptor specific for a particular domain of Fn, another phenotype may lack a receptor with such specificity. For example, as shown in Table 2.2, human bone cells show attachment to the HBD while bovine comeal epithelial cells do not [67]. The central cell binding domain (CCBD) is the primary ligand believed to be responsible for cellular adhesion to Fn. Within this domain (see Figure 2.3) is the 9th-10th type III repeat, which contains the proline-histidine-serine- arginine-asparagine (PHSRN) synergy site and the adhesive tri-peptide RGD sequence [68]. The PHSRN synergy site is not responsible for adhesion alone. Rather in unison with the RGD site,

18 cellular adhesive affinity is enhanced [69]. In general, the HBD produces a modest (if any) effect on cellular adhesion alone [67]. However, together with the CCBD, cells show enhanced migration and proliferation in the presence of the HBD [70, 71]. As shown in Table 2.2, the GBD does not contribute to cellular adhesion for both human bone and bovine comeal epithelial cells [67]. However, for monocytes, the GBD domain is believed to play a role in cellular adhesion [72]. We note that there are many other domains of Fn that we do not include here, again for the purpose of brevity. The reader is referred to a review by Hynes and Yamada for identification and the biological properties of such domains [73].

Type II PHSRN RGD O Typ®1,1 Type I Domains \ 1°-12 /

H2N-)1 COOH

HOOC Gelatin Binding Domain Central Cell Heparin II Binding (GBD) Binding Domain Domain (CCBD) (HBD)

Figure 2.3: Primary structure of fibronectin (Swiss Institute of Bioinformatics, ExPASy Proteomics Server. http://www.expasy.org/cgi-bin/protparam7P02751, accessed on July 28 , 2010). There are two nearly identical mirror segments of approximately 220-230 kDa joined by two disulfide bonds.

A single ligand has multiple influences on a single phenotype, and varying phenotypes have different responses to single ligands. More importantly, since protein conformation and orientation influences which adhesive ligands are available for interactions with cells, we explicitly see how protein conformation and orientation directly influences cell behavior. Because properties of the substrate influence protein conformation and orientation, the substrate has a direct influence on cell behavior.

19 Table 2.2: Cellular attachment to fibronectin domains

Phenydtype CCBD HBD GBD Reference human bone + + - [67] bovine comeal epithelial + -- [67] endothelial + + n/a [71] osteocarcinoma + + n/a [70] monocyte n/a n/a 4- [72] n/a = not available as the domain of Fn in the literature was not studied.

It is important to note that the information outlined in the previous paragraph has limitations. Many of the studies presented drew conclusions from either utilizing synthetic domain fragments of Fn or through the blocking of particular domains via antibodies. The complete absence of certain domains as a result of using domain fragments likely has drastic effects on cell behavior. Cellular function is rarely the result of one ligand on a protein. Rather it is a result of simultaneous cell receptor interactions with multiple surface ligands. For example, it is believed that GBD stimulates fibrillogenesis in fibroblast adhesion but is not a direct contributor to cell adhesion. Rather fibrillogenesis is stimulated through a cooperative mechanism of the cell surface receptor transglutaminase with both the CCBD and GBD [74]. In conclusion, while synthetic fragments of Fn may give some information regarding the role of particular ligands, such results should be interpreted with caution as in a physiological environment, such synthetic fragments do not exist. Furthermore, even when not using synthetic Fn fragments, one still likely has Fn fragments and even contaminants within their system. Fn is usually purified from blood plasma using chromatography. Regardless of purification technique, the end product contains trace amounts of contaminants such as proteolytic fragments, fibrinogen, and plasma gelatinase [75-77]. Thus, all results should be interpreted with caution as some may not be completely physiologically representative.

The orientation and conformation of surface adsorbed Fn determines which particular ligands (CCBD, GBD, and HBD) are available for potential interactions with cell receptors, ultimately eliciting cell function. The conformation of Fn is dependent on parameters such as the type of Fn (e.g., plasma, cellular) and the conditions of the solution environment (i.e., pH, salt presence, etc.) [78], For the case of surface adsorption, the properties of the surface and/or conjugation mode has influence on Fn conformation [79, 80]. Regardless, depending on such parameters, Fn

20 takes a compact or extended conformation. In a compact form, the RGD ligand is not accessible [78]. As such, this conformation is not favorable for cellular adhesion. Alternatively, an extended conformation better exposing the adhesive ligands is more favorable for cellular adhesion.

For all intensive purposes, a particular conformation/orientation of Fn or any protein is the resultant of a superposition of protein-substrate and protein-protein interactions [81]. In the initial stages of adsorption, protein-substrate interactions are of most relevance. As the surface coverage increases, interactions between neighboring surface protein-protein interactions also become relevant. These interactions are dependent on the properties of the substrate and the protein. We will now address the physical nature of such interactions.

2.4 Interactions in Cellular Adhesion

Often, little emphasis is placed on the surface when studying cell-material interactions. Regularly, the unrestricted word “immobilized” is used to describe a biomolecule on a surface, essentially marginalizing the importance of the chemical or biochemical surface modification employed prior to cell seeding. However, the modification is in fact the precedence of cell behavior on a surface. There are many different genres of interactions that can occur in cell- material studies. Interactions that are of particular importance are those at and within the vicinity of the four interfaces of the colloidal system (shown earlier in Figure 2.1 A). To describe a molecule that is immobilized onto a surface, the term adsorption universally applies. Chemisorption (i.e., sharing/exchange of electrons) and physisorption (i.e., no sharing/exchange of electrons) are the two generic interactions that occur in cell-surface studies. Within these genres are many different subgenres of surface and intermolecular interactions, not all of which are completely understood.

The interactions between cells, biomolecules, and surfaces are very complex. Such complexity arises from the dynamics inherent in physiological and biological systems. Often multiple genres and subgenres of interactions are involved in a cell-surface system, none of which occurs at the same time. Furthermore, such interactions are not necessarily additive in space as there is

21 a dependence on the properties of the molecule (i.e., polarizability, dielectric permittivity, etc.) and its surroundings (i.e., nearest neighbor distance, solvents present, etc.) [82].

The thermodynamics of biological systems are very complex. Biological systems are not necessarily even approaching thermodynamic equilibrium. Numerous liquid-liquid and liquid- solid phase boundaries are involved, to name only two. Thermodynamic analysis of such interfaces usually assumes flatness and homogeneity. However, such assumptions are unrealistic as biological interfaces can be significantly curved and are rather heterogeneous. Nevertheless, some insight can be gained from thermodynamic analysis that utilizes such assumptions. We refer the reader to concise review by Lyklema for such a discussion on the thermodynamics of biological interfaces [83].

In this section, we briefly introduce the intermolecular and surface interactions (i.e., chemisorption and physisorption) relevant in cell-surface science. For comprehensive treatment of such concepts the reader is referred to excellent texts by Israelachvili [84] and Hunter [85]. In addition, Leckband and Israelachvili [86] and Baszkin and Norde [87] provide excellent reviews on the same concepts with a special emphasis placed on biology. Throughout this section, we refer to such works.

2.4.1 Chemical Interactions

Physics, chemistry, and bulk materials science generally use the terms covalent or chemical interaction/bond to describe the interaction between two molecules that share/exchange charge. In surface science, the term chemisorption is used to describe the same principle for the more specific case of when a molecule is bound to a surface under such properties. Typical features of chemisorption are high affinity with the surface and uniform directional bonding of the adsorbate (relative to one another) throughout the surface [84], Often, one will also encounter the term “grafted” to describe molecules chemisorbed to a surface.

A common form of chemisorption employed in surface science is amide bonding via carbodiimide chemistry. The surface can be composed of chemically available COOH groups from SAM [88], carboxymethyl-dextran (CMD) grafting [24], or glutaric anahydride [79]. With the help of catalysts, biomolecules are then exposed to the surface in a buffer solution and

22 become immobilized through the formation of amide bonds between available NH 2 groups in the biomolecules and the available COOH groups on the surface. Another type of chemisorption is employing a free sulfur atom in the biomolecule for interaction with a maleimide functional surface thereby forming maleimide chemisorption [79, 80, 89, 90]. Consideration must be taking into account if the chemical moieties are available in the biomolecule to be chemisorbed. For example, Fn has approximately 75 lysine residues with available NH 2 functionalities for chemisorption. While Fn has 65 cysteine residues (a sulfur containing amino acid), all but two are available for maleimide chemistry [79]. The others are already employed for disulfide bonding in the tertiary structure. The RGD motif is commercially available in the form of GRGDS, the tail amino acids present for the purpose of chemical grafting whether it is chemisorption to a surface or another application.

Chemisorption is employed in the formation of SAM of organosulfurs and organosilanes (discussed later in Section 2.6.3). Organosulfur SAM are chemisorbed to gold surfaces via Au-S covalent bonds. Organosilanes are chemisorbed to glass or other silicon-based material surfaces.

2.4.2 Physical Interactions

All long range interactions that do not involve the sharing/exchange of electrons in surface science technically fall under the categoiy of physisorption. In physics, chemistry, and bulk materials science such interactions are referred to as physical interactions or sometimes physical bonds. While electrostatics is the nature of nearly all physisorbed interactions, some physical interactions (such as hydrophobic) are not completely understood. There is no universally accepted interaction potential to describe every interaction. In this section, where possible, we describe interactions quantitatively with the interaction potential. The resultant force of an interaction can be found by taking the derivative of the interaction potential with respect the spatial dependence (in this article the separation distance, r).

23 We begin with the simplest interaction, that being the Coulomb interaction. The interaction potential of the Coulomb interaction is given by,

’*<'•> = IT-—4 nee0 r (2-» where qi and q 2 are the charge magnitudes of the two interacting bodies, e is the dielectric permittivity of the interacting medium, e0 is the permittivity of free space, and r is the separation distance. If the two charges have the same polarity (i.e., both negative or positive), the interaction potential becomes positive thus the two charges experience a repulsive force. If the two charges have opposite polarities (i.e., one negative and one positive), the interaction potential becomes negative thus the two charges experience an attractive force. For all interactions, a negative interaction potential corresponds to an attractive force and a positive interaction potential corresponds to a repulsive force. We also note that if one takes the derivative of Equation 2.1 with respect to r, the result is Coulomb’s Law. For simplicity, we present the interaction between two point charges only. For other geometrical configurations we refer the reader to the classic texts of Griffiths [91] and Jackson [92].

The Coulomb interaction is relevant in cell-surface science as often surfaces with an associated charge are usually correlated with cell behavior to a charge polarity. Large biomolecules such as proteins and growth factors also portray local charge in their amine (NH 2+) and carboxyl (COOH') groups. The net charge of a large biomolecule is dependent on the pH of the solvent and can be estimated by knowing its isoelectric point. The cell surface also has a charge association. For example, hyaluronan is an osteoblast cell surface matrix protein of high negative charge [93], Thus, hypothetically, an osteoblast cell resists binding to a negatively charge surface (e.g., a COOH-SAM modified substrate) compared to a positively charged

surface (e.g., a NH2-SAM modified substrate). For further insight on the role of electrostatic interactions at biological interfaces, the reader is referred to literature by Kleijn and van Leeuwen [94]. The Coulomb interaction is often discussed indirectly in cell-surface literature as cell behavior correlation to substrate charge polarity is often discussed. While it is correct that a simplistic Coulomb interaction is relevant in a cell-surface system, it is somewhat naive to believe that it is the sole relevant physical interaction in the overall system.

24 A well known and studied interaction is known as the van der Waals (VDW) interaction. The interaction potential is generally given by,

w(r) = -%£. (2.2) r where r is the separation distance and the constant C v d w is dependent on the properties and geometry of the interacting bodies. In 1873, J.D. van der Waals proposed the successful “equation of state” [95]. The success of van der Waals’ equation of state inspired theoretical research on the origin of the forces in a gaseous system. Such origins were originally described quantitatively by Keesom, Debye, and London. Thus, the VDW force has three distinct sources of molecular interaction, (i) The first source is between molecules with permanent dipoles (or more generally multi-poles) and is known as the orientation or Keesom interaction [96], (ii) The second source occurs when the dipole moment of one molecule induces a dipole moment in another. Such dipole induction is dependent on the polarizability of the molecule whose dipole moment is induced. This is known as the induction or Debye interaction [97]. (iii) The third interaction source is the result of interaction of the natural dipole moments as a result of electron orbiting between two molecules. All molecules express such an “instantaneous” dipole moment. This is perhaps the most important interaction of the three VDW contributors as it is the strongest and is always present no matter the molecules interacting. This force is known as the dispersion or London interaction [98]. Together these three interactions make up the total VDW interaction. More quantitative detail on the individual Keesom, Debye, and London interactions can be found in the texts of Israelachvili [84] and Hunter [85]. Applying the theoretical and conceptual principles of VDW interactions to biological processes is discussed in detail by Israelachvili [99].

While one or more of the Keesom, Debye, or London interactions are always present in any interaction, in cell-surface research, macroscopic as opposed to microscopic descriptions of the VDW interaction are often more convenient and relevant to cell-material interactions. The VDW interaction between two macroscopic bodies can be approximated using pair-wise summation known as Hamaker theory [100]. Using Lifshitz theory [101], the Hamaker constant (analogous to C v d w ) can be determined for two macroscopic bodies interacting through a medium. The

25 VDW interaction potential and force between various macroscopic geometries in terms of the Hamaker constant are presented by Leckband and Israelachvili [86].

The interaction between a cell and a surface can intuitively be thought of as the interaction of two different, charged surfaces in an electrolytic solution. While gravity is the primary force responsible for bringing the cell from the suspension medium to the surface, there is in actual fact another interaction consequent of the electric charge the plasma membrane and modified surface portray in the electrolytic culture medium in which they interact. Such an interaction is generally known as the double-layer (DL) interaction. The DL interaction results when two charged surfaces approach one another in an electrolyte solution, charges can be attracted, repulsed, or bound to the surfaces. Such an interaction is given quantitatively by,

M r) * +CESe~Kr (2.3) where Ces is a constant and the factor k ' 1 is known as the Debye length. The constant Ces is dependent on the properties (i.e., surface charge density, potential, geometry, etc.) of the surfaces and the Debye length depends on the properties of the liquid (i.e., solute concentration, type, solution temperature, etc.). C e s can be determined by solving the Poisson-Boltzman equation or other trying theories.

Cell culture media contain a wide variety of solutes such as proteins like albumin, carbohydrates like glucose, dipolar amino acids, and electrolytic salts like NaCl and KC1 all of which influence the Debye length. As mentioned earlier, substrates, biomolecules, and cell surfaces often portray a local net charge. Thus, DL interactions are very relevant in cell-surface science as initial cell- surface or biomolecule-surface interactions can intuitively be thought of as two charged surfaces approaching one another in an electrolyte solution (i.e., culture media or buffer solutions) as shown in Figure 2.4. This initial interaction likely influences the morphology and behavior of cells on surfaces. Such an interaction is rarely mentioned in cell-surface literature. Analysis is usually performed several hours to days after initial cell seeding to the surface. As mentioned earlier in 2.3.2, cell adhesion is a dynamic process. Initial DL-like interactions as just described likely influence the dynamic mechanisms that result in the later observed cellular behavior such as morphology, proliferation, migration, etc. Focus is often placed on observations chosen at a given period of time after cell seeding to the surface, thus often neglecting the dynamic

26 molecular mechanisms that lead to such observations. Some DL interaction potentials and forces for various macroscopic geometries are given by Leckband and Israelachvili [86]. Cells and large biomolecules are non-rigid bodies. Thus, while these interaction potentials for DL in macroscopic geometries are relevant, they would not truly represent a cell-surface DL interaction. This adds complexity to an already complicated system making a quantitative description challenging.

Cell Interior

Cell surface

Positively c h a r g e d ■* + + receptor

Negatively charged I 1 1 i I 1 i i i receptor

Electrolytic cell culture media

Positively charged amine modified substrate

Figure 2.4: Double-layer like interaction between a modified substrate bearing a charge and a cell surface with charge bearing receptors. Figure not drawn to scale. The arrows indicate that the cell is approaching the surface.

For surfaces in aqueous solutions it is common for the DL and VDW interactions to be paired together in what is known as DLVO (Deijaguin-Landau-Verwey-Overbeek) theory [102, 103]. The VDW force is affected very little by changes in the solution or surface environment while the DL force is very sensitive going from attraction to repulsion depending on conditions. Generally, the DL force is of greater magnitude than the VDW force. The VDW force becomes more dominant at smaller separation distances. However, one must also consider that at smaller

27 separation distances, other forces can become dominant [86]. The fundamental relevance of DL interactions to biological interfaces is further discussed by Kleijn and van Leeuwen [94].

A force that arises when two surfaces approach one another in a solution is known as the oscillatory force. Such a force oscillates between attractive and repulsive and the periodicity is dependent on the diameter of the confined molecules between the surfaces. The interaction of the oscillatory force is generally given by,

(2.4) where Eo is the contact energy and <5 is the diameter of the solvent molecules. A requirement for the oscillatory force is a rigid smooth surface. In biology and physiology, cells are non-rigid and large biomolecules are rarely if ever rigid. As such, the oscillatory force should be less relevant in biological systems [86].

An interaction specific for the presence of hydrogen is known as a hydrogen bond (H-bond). The H-bond exists between a hydrogen atom and an electronegative atom or molecule. It pictorially appears to be a VDW-Keesom dipole-dipole interaction. However, quantitatively the affinity of an H-bond is approximately an order of magnitude higher than an electrostatic VDW- Keesom interaction yielding it a special name. It has been accepted that there is no exchange/sharing of charge in H-bonding. However, the unique characteristics of H-bonding have them classified by many as quasi-covalent [84].

The unique behavior of materials in the presence of water has led to the widely studied topic of hydrophobicity. While the nature of hydrophobic effects and interactions are unknown, their importance in biology and physiology is unprecedented. Hydrophobic materials are usually non polar. The first detailed theory of hydrophobicity was the work of Frank and Evans [104]. When a hydrophobic molecule (i.e., solute) is placed in water, water molecules surround the molecule and orient into an ordered structure in the vicinity of the molecule while not losing any of their H-bond sites. Such a phenomenon is not entropically favorable and is known as the hydrophobic effect. This is an entropically unfavorable interaction because the water molecules in close vicinity of the solute hydrophobic molecule are well-ordered in comparison to those not in the vicinity. When multiple hydrophobic molecules and/or two hydrophobic surfaces are placed in water, the molecules cluster together and/or the surfaces attract one another. Such interactions are far stronger than if the bodies were not in water (i.e., VDW interactions using Lifshitz theory [105]), yielding it a special name. Such a phenomenon is known as the hydrophobic interaction.

Quantitative descriptions of the hydrophobic interaction between surfaces have been developed using force-distance measurements [106-108]. It was found that a hydrophobic interaction generally follows a somewhat exponential decay [109]. We revisit these experiments on the hydrophobic effect later in Section 2.6.6.

Despite numerous experimental and theoretical studies on hydrophobic interactions, its nature and a general quantitative interpretation still remain elusive [105], It has been suggested that the hydrophobic interaction is a special type of VDW interaction that is a result of the unique dielectric properties of water [84]. Despite the vast unknowns on the origin of hydrophobic interactions, without question these interactions play an important role in biology and physiology due to the presence of water in such systems. They are believed to be responsible for stabilizing biomolecules such as DNA and proteins and are always implied to be involved in the adhesion of cells to surfaces [110].

A common term that arises in science is hydrophilicity. Such a term arises because many molecules are highly soluble in water (as opposed to non-soluble or hydrophobic) and are thus coined “hydrophilic”. Hydrophilic molecules are believed to have a disordering effect (i.e., entropically favorable), the opposite of hydrophobic molecules. However, one must note that no such hydrophilic interaction exists [84], Rather, hydrophilic matter in water has a weak hydrophobic interaction in comparison to hydrophobic matter. Hydrophilicity is not a polar opposite characteristic to hydrophobicity as some seem to imply.

The presence of long polymeric and large molecules in biological and physiological systems results in a very significant repulsive force known as a steric interaction. There is no single interaction potential that can quantitatively describe steric interactions. As discussed later in Section 2.6.7, the steric interaction potential is dependent on the conformation and density of the polymeric material. Furthermore, the interaction potential is an approximation as opposed to an exact expression. Discussions doubting the validity of such approximations still remain. The

29 magnitude of steric interactions is dependent on polymer density and the molecular weight of the body undergoing interaction.

The nature of steric repulsion is due to entropically unfavorable compression of polymer chains or large molecules upon contact with another large molecule. As shown in Figure 2.5, the system is compressed due to the encroachment of a large molecule. For the system to return to its entropically favorable state, the polymer chain is extended, repulsing the large molecule. When polymeric materials such as dextran-derivatives or poly(ethylene glycol) (PEG) are chemisorbed to surfaces at an optimal density, the magnitude of steric repulsion can in fact prevent not only protein adsorption, but also cellular adhesion. Thermal fluctuation in physiological and biological systems also can lead to oscillatory motion of polymer brushes, enhancing repulsion. Steric repulsion is discussed in more detail later in this article (Section 2.6.7) as it pertains to low-fouling surfaces (i.e., little to no nonspecific protein adsorption).

The interactions discussed thus far in this section are vital to protein adsorption. As discussed earlier in Section 2.3 of this article, protein adsorption also plays an influential role in cellular adhesion. For a discussion on physical interactions and protein adsorption, the reader is referred to a series of literature composed by Horbett, Brash, and Hoffmann [111-113]. Very briefly, it is believed that proteins adsorb with a high affinity to hydrophobic surfaces resulting in unfolding of the protein (i.e., drastic conformational change) at the surface followed by the release of water from the interface. Cationic proteins bind to anionic surfaces and vice versa. Thus, attractive DVLO and hydrophobic interactions are the primary forces responsible for the adsorption of proteins to surfaces [81,114].

30 Protein approaching a surface covered M by a low-fouling polymer ™ /

Protein compressing grafted polymer Repulsion of protein via extension of grafted chains creating an entropically polymer chains returning the system to its unfavorable state entropically favorable state

Figure 2.5: Steric repulsion of proteins from a polymer-grafted surface.

An interaction between biological molecules that is somewhat contradictory is the ligand- receptor (LR) interaction. The short range, high affinity, well ordered geometric structure, and specificity of this interaction make it unique from other types of physical interactions. The most studied LR interaction is that between biotin and streptavidin and/or avidin [115]. Biotin- avidin/streptavidin and other LR interactions involve electrostatic, VDW, H-bonding, and hydrophobic interactions in sequence or simultaneously. LR interactions also have a finite life span. The complexity of LR interactions is vast. However, these are perhaps the frontier between physical {i.e., VDW, DL, etc.) interactions and biological function {e.g., diseases). For

31 example, apoptosis and proliferation of cells is regulated by LR interactions. Dysfunction of such LR interactions can result in hyperproliferation, often leading to cancer [116]. For a thorough review of the physics of LR interactions in both a theoretical and experimental context, the reader is referred to an article by Bongrand [117].

Many of these interactions can be present in a biological system [82], It is difficult to draw the interaction potential of a force in a biological system for reasons mentioned in Section 2.4. To give the reader a semi-quantitative representation of these forces, Figure 2.6 shows the interaction potential of some forces as a function of separation distance.

DLVO

Steric

Oscillatory re o . Electrostatic Repulsive

O O. c Separation o Distance '•6 res

Electrostatic Attractive

VDW

Figure 2.6: Interaction potentials of surface and intermolecular forces in biological systems. While there is no universal interaction potential to describe the steric interaction, it is generally semi-quantitatively represented as shown [82, 86]. 32 In summary, this section has shown that cell adhesion is a dynamic complex process involving an array of physical interactions, both specific and non-specific, occurring at varying stages during cellular adhesion. Furthermore, the nature of some of these interactions remains unknown. It is difficult to describe macroscopic phenomena such as cell adhesion and subsequent behavior from a molecular perspective. Some attempts have been made and are still in the works to unify molecular and macroscopic phenomena associated with cell adhesion. For a discussion on this, the reader is referred to an article by Simson and Sackmann [118]. Briefly, it appears that the elasticity of the cell membrane is the parameter that largely dictates cell adhesion. The interactions of elastic lipid bilayers that comprise the cell membrane with similar or other types of surfaces are described as undulation forces (first coined by Helfrich [119]).

2.4.3 Interaction Measurements

There are many experimental methods that exist to measure such forces just discussed. Some include surface force apparatus (SFA), atomic force microscopy (AFM), optical tweezers, osmotic pressure measurement, micropipette aspiration, and shear flow detachment measurements, to name a few. For the appropriation of a particular experimental method, the reader is referred to Leckband and colleagues [86, 120]. Unfortunately, these techniques are in vitro and are not representative of a physiological environment. Nevertheless, valuable information can be obtained when employing one or more of these techniques.

The two most widely used techniques in characterizing biological forces are the SFA and AFM. The SFA is a very powerful tool for measuring the force between two surfaces at a known separation distance. Modifications by Israelachvili and Adams [121] to the original design of Tabor and Winterton [122] extended the capability of the SFA to liquid environments making SFA a very powerful instrument in measuring biological forces. The two surfaces can be tailored to meet the specific interests of the user. Force is measured via the deflection of a spring and distance is measured via optical interferometry. SFA can measure the electrostatic profiles (albeit local not global) of proteins and the force-distance profiles between cell membrane receptors and proteins [123, 124]. The AFM was developed by Binnig et al. in 1986 [125]. The ability of the AFM to achieve high resolution topographic images and force measurements at the

33 nanometric scale in air and particularly liquids makes it an attractive tool in biology [126]. Force interactions between a molecule and a surface are achievable. A molecule is immobilized onto the cantilever of the AFM; the cantilever is then brought towards the surface. The deflection (measured with a laser) of the cantilever is then used to profile the interaction between the molecule of interest and the surface of interest. Interactions as weak as a single H-bond have been reported [127].

Valuable information has been acquired using AFM and SFA to measure interactions in biological systems. Such experiments provide insight in the underlying mechanisms and dynamics of cell-surface interactions that are not achievable using static and fortuitous labeling of cells on modified surfaces.

2.5 Systems for Studying Cell-Material Interactions

2.5.1 In Vivo Systems

The in vivo environment of cells is far more complex than any controlled 2D or 3D in vitro system. Depending on the tissue type, in vivo cells can have surroundings consisting of other cell phenotypes (e.g., cells from major organs, muscle cells) and ECM molecules. Moreover, the ECM is composed of a restricted number of molecules that have profound effects on the mechanical and chemical environment of cells. The ECM is able to bind soluble growth factors such as vascular endothelial growth factor (VEGF) and hepatocyte growth factor (HGF), and thus create molecular gradients that subsequently result in localized interactions with surrounding cells [128].

The most common ECM molecules are collagens. Other proteins, such as elastin and proteoglycans (such as hyaluronic acid), also contribute in defining mechanical and chemical niches. It is not the aim of this section to summarize ECM and cellular environments of different tissues, although this information is vital when attempting to create representative 3D in vitro culture environments. For example, the study of chondrocytes in vitro necessitates an environment which possesses inelasticity, is highly hydrated, is sparsely vascularized, and where

34 cells are located at relatively large distance from one another. Hepatic tissue, on the other hand, requires ECM elasticity, high vascularization, and cell aggregation to maintain differentiation. For specific information on different tissue types relatively to ECM, vascularisation, and cell-cell interactions, the reader is referred to histology texts by Kessel [129] and Stevens and Lowe [130]. Cells in more relevant 3D in vitro culture systems behave much differently than in traditional 2D systems.

2.5.2 Cells in a Three-Dimensional Environment

The growth of cells in a three-dimensional (3D) in vitro environment is not a novel idea in biology. More than 50 years ago [131], experiments showed differences between the shape of fibroblasts cultured in blood plasma clots relative to the orientation of the fibrous network of the clot, hinting at the importance of the 3D surroundings on cell behavior. Around the same time, mesenchymal cells cultured on glass cover slips showed stress fibers seldom encountered in vivo [132]. Despite these results, 2D systems have been until recently the principle vehicle for in vitro cell and biomaterials studies, principally due to their relative ease of use. The current knowledge in cellular physiology, genetics, and proteomics has in large part been influenced by experiments in which cells were artificially grown on flat surfaces, often plastics.

Key conceptual differences between in vitro 2D (i.e., traditional) and 3D cell culture systems are outlined in Figure 2.7. Traditional 2D cell culture systems consist of adhesive ligands spatially

distributed on a surface (e.g., on the jc - z plane in Figure 2.7). This 2D localization of ligands results in adhesive asymmetry of the cellular organism. Furthermore, the rigid non-porous surface upon which the adhesive ligands are supported creates a mass transfer limitation zone. Conversely, in a 3D cell culture system the adhesive ligands are spatially distributed in all geometric planes. Thus, all areas of the cellular organism are equally exposed to adhesive ligands. Because the adhesive ligands are distributed in a porous gel scaffold, a 3D cell culture system lacks the mass transfer limitation zone (i.e., the cell-substrate interface) of the 2D system.

35 2D system 3D system y

Surface Z & X Mass transfer limitation zone u

Localized interactions of cell surface with adhesive ligands Porous gel scaffold 1 matrix

Figure 2.7: Key conceptual differences between 2D and 3D culture systems.

The emergence of 3D cell culture systems has outlined the importance of many variables that are not considered when culturing cells in 2D. In addition to usual considerations such as nutritive medium composition, cell density, dissolved gas composition, and surface chemistry, variables such as ECM chemical composition and mechanical properties define 3D culture systems. 3D systems can be fabricated from materials derived from in vivo constituents, such as ECM and synthetic materials. Differences in mechanical properties, for example, the elasticity of the support matrix and that of cells, create a boundary. Changes in the flux of metabolites and gases in 3D systems, relative to 2D, also greatly contribute to define cell response.

We first present key results obtained in 3D environments. Then, the important parameters of 3D systems including, the mechanical, biochemical, and transport properties of the matrix scaffold used to support cells, will be discussed. Finally, important culture systems made from natural and synthetic materials will be reviewed, along with reported uses.

36 2.5.3 Differences between Two-Dimensional and Three-Dimensional Cell Culture Systems

A number of experiments have undoubtedly demonstrated dramatic variations in cellular response when culturing cells in a 3D environment compared to 2D flat culture plates. Important differences in key biochemical events, cell shapes, and cell motility have been observed and characterized. Morphological differences often strike the imagination and perhaps best convince the huge gap between cell physiology in 2D and 3D. For example, fibroblasts grown in collagen matrices compared to collagen-coated cover slips present a widely different morphology when exposed to selected biochemical niches. Specifically, platelet derived growth factor (PDGF), known to stimulate dendritic network formation in vivo, and lysophosphatidic acid (LPA), known to cause cellular retraction as shown in Figure 2.8 [133, 134]. Moreover, fibroblasts grown in 3D collagen matrices show important differences to growth factor stimulation that impacts the cell-generated contractile forces [135]. Mechanically loaded matrices stimulate the differentiation of fibroblasts into myofibroblasts [136].

Cell migration in 3D is widely different from distinct focal contacts and stress fibres observed in haptokinetic cell migration in 2D culture systems [137]. Spindle-shaped cells such as fibroblasts, many tumor cells, and endothelial cells constantly reorganize their surrounding ECM through, for example, serin proteases and metalloproteinase mediated proteolysis. These cells also interact constantly with the ECM through the large number of integrins located on their surface to migrate in 3D systems [138, 139].

37 \ 'In

Figure 2.8: Human fibroblasts project a dendritic network of extensions in collagen matrices but not on collagen-coated cover slips. Fibroblasts were incubated 5 h on collagen-coated surfaces (A-C) or in collagen matrices (D-F). After 1 h, 50 ng/mL PDGF (B and E) or 10 pM LPA (C and F) was added to the incubations. At the end of the incubations, samples were fixed and stained for actin. Bar: 80 pm. Reprinted with permission from reference [134]. Copyright 2003 The American Society for Cell Biology.

Another example is angiogenesis, the formation of new capillaries from existing blood vessels, which is present in various physiological processes whether they be normal, such as embryogenesis and wound healing, or abnormal such as in developing cancer masses. Large developments have been achieved in understanding important cell-material interactions linked to angiogenesis. However, angiogenesis models in 3D have recently been shown to exemplify the

38 importance of studying cells in the third dimension rather than on flat surfaces [140, 141]. The importance of ECM for angiogenesis has been documented for 2D models. Evidence suggests that 3D models yield dramatic differences in cell behavior. Important cellular signalling pathways for angiogenesis, such as mitogen activated protein (MAP) kinases and extracellular signal-regulated kinase (ERK) pathways have both been shown to be strongly influenced by the use of 3D collagen and fibrin gel environments compared to 2D models [142]. Moreover, in the same study, by comparing 2D cell culture systems with collagen and fibrin matrices, dramatic morphological differences were reported when culturing human umbilical vein endothelial cells (HUVEC) in a 3D matrix compared to 2D culture plates. For example, in 3D, HUVEC vacuoles rapidly grow in size and adjacent cells coalesce to form tube-like structures after only 4 hours [142],

The onset of cancer cell dedifferentiation such as the transition from epithelial to mesenchymal phenotype is now considered an ensemble of complex processes in which extracellular cues play a major role [143]. The importance of signalling pathways leading to dedifferentiation has been studied widely in vitro. However, data from plastic 2D culture plates show important differences from results obtained with cells cultured on porous substrates or 3D collagen gels. The differences range from important morphological differences upon biochemical exposition [144] to variations in apoptosis levels [145].

3D models have recently been used to study the mechanisms of apoptosis in adult tissues. It has been known for a long time that very important differences exist between the survival of cancer cells embedded in 3D spheroid bodies compared to tumors grown on flat surfaces when exposed to death stimuli [146, 147]. Also, the presence of a large number of adhesion molecules for cancer cells in a 3D environment is suspected to be responsible for the drug resistance of certain types of cancers [148]. These phenomena have been an obstacle for the in vitro discovery of novel anti-cancer agents.

Aberrations in the apoptosis pathway in adult tissues are known to be an important factor in the pathogenesis of cancer, among other diseases [149]. Although thorough apoptosis studies have been carried out using traditional 2D cell cultures, studies in more relevant 3D culture models are now thought to lead to a more conclusive understanding of its mechanisms, and thus of cancer [150], The loss of important cell-ECM interactions has been associated with apoptosis for many

39 tissue types. For example, the loss of specific integrin mediated interactions has been shown to result in apoptosis for neural and mammary gland tissue, among others [151, 152]. While 2D cultures have been used to reveal important adhesion mechanisms for apoptosis, experiments conducted using 3D systems have pointed towards the likelihood of more complex apoptosis signalling regulation. For example, lumen formation by cell apoptosis during mammary morphogenesis was shown not to be simply due to the loss of ECM interactions as suggested by 2D studies, but to much more intricate phenomena [151]. Among others, the polarity induced by 3D epithelial tissue architecture is thought to be a critical aspect of apoptosis regulation [149], as are the differences in mitochondria homeostasis modulation in 3D compared to 2D cultures [153].

Although the causes of the previous observations for different cell types in 3D compared with traditional 2D culture systems are not always understood, some general conclusions can be drawn as to the underlying mechanisms guiding these events. 3D environments have the capacity to convey a large spectrum of mechanical information, as well as to present a large array of textures and shapes to attached cells, when compared to flat surfaces. As such, mechanical properties and textures both are considered driving phenomena that help explain the large variations observed between flat surfaces and matrices along with two other key aspects of 3D environments, the chemical composition of scaffold material and its gas, nutrient, and waste permeability. These parameters can often have dramatic and overlooked effects in many experimental schemes of biomaterials science and have likely resulted in misleading observations. We will now attempt to outline the known effects of these phenomena and their mechanisms of action in cells.

2.5.4 Properties of Cell Culture Systems

2.5.4.1 Mechanical Properties

Mechanical properties of cell culture systems elicit and regulate, in an indirect manner, a large number of cell responses in both 2D and 3D systems and should be considered as important for cell physiology as biochemical properties. For example, such properties influence morphogenesis, bone formation, neutrophil activation, and are responsible for several stress related pathologies [154-156]. They affect cellular migration, morphology, and adhesion [157].

40 Mechanical loading is the imposition of stresses and/or strains on a body through a mechanical force. Should the force on the body be of appreciable magnitude and the body behaves in an elastic manner, deformation occurs. There are three ways to deform an elastic body via mechanical loading; through tensile extension, bulk compression, and shear stress. Quantitatively, the elasticity of a body is defined by the elastic modulus of the material; Young’s modulus in the case of tensile loading, the bulk modulus in the case of bulk compression, and the shear modulus in the case of shear stress. Cells undergo each type of mechanical loading in both 2D and 3D cell culture systems [158]. Furthermore, many materials used in tissue engineering and cellular scaffolds possess viscoelastic behavior. Therefore, to describe these systems, it appears necessary to investigate viscoelastic material functions (e.g., stress relaxation and creep and recovery).

In an exclusively intuitive sense, a feedback phenomenon appears to be present between the matrix elasticity and the cell. Cells modify their intracellular signaling according to the elasticity of their local environment. Studies show the elasticity of the 3D surrounding porous gel matrix of the cells has an impact on cell shape and cytoskeletal tensions, which in turn dramatically influences the fate of cells [159]. Likewise, cell plasma membrane elasticity equally influences cell behavior and fate, as discussed by Sackmann and Bruinsma [160]. Thus, 3D cell culture systems must be designed to provide mechanical loads with a magnitude, orientation, and type relevant to those of a physiological system.

The nature of the porous ECM scaffold, noticeably its elasticity, has a profound impact on the behavior of entrapped cells. Specifically, the nature and affinity of focal adhesions between cells and scaffold material is a key to force retroaction. Cells, in hydrogels such as collagen and fibrin portray many focal adhesions while cells in synthetic substrates, such as polyacrylamide gels, portray fewer focal adhesions [157]. Ligand-coated synthetic gels may improve in some cases, cell-material focal adhesions.

Techniques such as AFM and micropipette aspiration can be used to study the effect of mechanical loading on cells. Typically, mechanical loads are applied on macroscopic scaffolds in which cells are entrapped. This can be accomplished on flat planes (e.g., tissue culture wells), yielding a 2D system [158]. Systems able to apply tensile loads on cell monolayers at known magnitudes are available (e.g., Bio-Flex plate, Flexcell Inti Corp.) and have been used to grow

41 smooth muscle cells (SMC) in strained collagen layers [155]. 2D surfaces are relevant to the study of endothelial cells lining blood vessels. However, important efforts are conducted to use 3D scaffolds which are more representative of the in vivo reality for most tissue types. In such systems, scaffolds can be “loaded” through the attachment of the whole scaffold to vessel walls or “unloaded” in free-floating scaffolds [133]. Furthermore, complex bioreactor systems can also be used to control shear stresses around entrapped cells [161].

Evaluation of the effects that mechanical loads have on cell behavior is challenging. In 2D systems, compression of silicone sheets or membranes is often used [135, 157]. Silicone sheet elasticity can be varied for such systems, but must be quantified, which is a tedious procedure. Furthermore, while silicone sheets aptly illustrate unidirectional forces, multidirectional forces are unquantifiable in this system. Apparata that monitor gel deformation have been developed for 3D systems [135].

Fibroblasts remain predominantly studied to correlate the mechanical loading to cell response within a system. Fibroblasts are typically seeded in mechanically-loaded or unloaded type I collagen gels in which specific growth factors can be added, or cultured as a monolayer on a gel- coated well, with some form of force sensing device. Fibroblasts were found to be highly sensitive to the mechanical loading of their environment, which is logical considering their role in tissue homeostasis i.e., in synthesis of ECM components such as collagens and proteoglycans and in tissue remodeling and repair [158]. For example, fibroblast gene expression and growth factor responses in multiple mechanically defined environments, both 2D and 3D, show very important variations, as presented in a review by Wang et or/.[158]. Also, collagen fibril density was shown to modulate fibroblast proliferation and morphology [134, 162]. Typically, fibroblasts are known to proliferate when placed in unloaded matrices or 2D culture systems compared to mechanically loaded systems. Finally, force monitoring has shown that when growth factors are present in a 3D system, fibroblasts generate contractile forces [135].

The influence mechanical loading on the behavior of other types of cells including SMC, endothelial cells, myoblasts, hepatocytes, osteoblasts, neurons, and neutrophils has been studied in the past. SMC show increased proliferation and altered response to growth factors when subjected to mechanical strain in vitro [163, 164]. SMC behavior has also been shown to be dramatically different in 2D collagen-coated surfaces compared to 3D collagen gels, noticeably

42 protein expression and contractile activity [165]. Myoblasts plated on collagen-coated polyacrylamide gels in vitro present normal myotube striation only at intermediate matrix elasticity that corresponds to in vivo healthy muscle tissue [166]. Hepatocytes maintain a differentiated phenotype and aggregate only in relatively elastic materials (such as Matrigel) [167]. Neuron regeneration after injury in the central and peripheral nervous system is enhanced by the use of elastic hydrogels that are hypothesized to lower the density of an otherwise impenetrable scar tissue at injury [168]. Alkaline phosphatase activity, a marker of osteoblast activity, is enhanced up to seven-fold when exposed to dynamic flow (i.e., elevated shear stresses) [169]. When mechanically deformed in vitro in narrow channels to mimic in vivo deformation occurring in capillaries, neutrophils activate and extend pseudopods, thus increasing their migratory tendencies [156].

It appears that the influence mechanical loading has on cell behavior in culture systems, whether it be 2D or 3D, is substantial. However, understanding is incomplete. The underlying difficulty is that the mechanical loading of a system is macroscopic in nature while cell response is molecular in nature. Drastic variations in materials (e.g., cell phenotype, scaffold material, etc.) between subsequent studies complicate conclusions. Nevertheless, analytically, one can draw some general conclusions, using fluid mechanics and the Buckingham-Ji theorem (presented in a later paragraph), relating how material mechanical properties influence cell response and behavior (i.e., cell spreading and morphology) to mechanical loading. Observations in the physics community suggests that, as Singer and Nicolson’s Fluid-Mosaic model discussed earlier eludes [38], a cell can be considered a viscoelastic body. As such, the cell simultaneously has the properties of a liquid (i.e., viscosity) and a solid (i.e., elasticity). Furthermore, materials composing gel matrix scaffolds (e.g., fibrin, collagen, etc.) portray the same properties.

During spreading (i.e., the process of adhesion) on a 2D surface, the cell behaves like a viscoelastic fluid body undergoing wetting and dewetting transitions [160, 170]. Specifically, when a cell is interacting with a material, there is a disruption in its molecular architecture [171]. This disruption is a result of the generated force by the cell through specific short range LR and non-specific long range molecular interactions (discussed earlier in Section 2.4) between the cell and the material. For the relations of such forces to the elastic behavior of a cell in a molecular and theoretical context, the reader is referred to a thorough review by Sackmann and

43 Goennenwein [172]. Likewise, in a 3D matrix scaffold, cell spreading and/or migration is somewhat parallel to the shape and/or movement of drop of fluid in a surrounding fluid medium.

The Buckingham-rc theorem provides a means of using dimensional analysis to identify dimensionless numbers that describe the physical attributes of a system [173, 174]. It has been extensively used in fluid mechanics. If a cell has fluid-like properties, then some of the dimensionless numbers of fluid mechanics may be applicable to cell behaviors that resemble those of a fluid in contact with a solid, namely spreading and subsequent morphology. Utilization of the Buckingham-Ji theorem can perhaps shed some insight into the nature of macroscopic cell behaviors as cell adhesion is somewhat parallel to the wetting of a fluid on a surface. Perhaps dimensionless numbers can be developed in the future that yield a quantitative basis for cell behaviors much like the well established Reynold’s number gives a quantitative basis between turbulent and laminar flow regimes in fluid mechanics. Such analysis could presumably then be utilized to design systems and select materials that elicit and control a desired cell response and be invaluable tools to the biomaterials and tissue engineering community.

The molecular nature of the viscoelastic cell behavior has been under intense investigation over the last decade within the physics community. Two progressive experimental techniques that have been developed are magnetic bead microrheometry [175] and reflection interference contrast microscopy [176]. Theoretical studies have also been performed by Bruinsma and Sackmann [170] and Nardi et a/. [177], Experimentally, viscoelastic parameters (e.g., elastic moduli, relaxation time, etc.) of fibroblasts and HUVEC have been measured [175, 178], In another study, materials analogue to the cell membrane were fabricated and their viscoelastic properties measured [179]. From a molecular perspective, it appears that the molecules actin and myosin (heavily involved in cellular adhesion, as discussed earlier in Section 2.3.2) give cells their viscoelastic properties as addressed by Uhde et al. [180-182]. The viscoelastic properties of scaffold materials have also been studied experimentally and parameters (e.g., Young’s modulus, Poisson ratio, etc.) measured. In addition to more natural scaffold materials such as collagen [183] and soft hyaluonic acid gels [184, 185], synthetic polyelectrolyte multilayer films [186] have been studied. For a comprehensive review on progress in studies of this manner, the reader is referred to recent articles by Smith and Sackmann [187] and Trepat et a/. [188].

44 The literature on the viscoelastic properties of cells and scaffolding materials is particularly relevant to the biomaterials community. Such studies give more analytical insight as to how mechanical properties of a system influence cell behavior. A particularly relevant study is that of Feneberg et a f [178]. Here, a confluent monolayer of HUVEC was subjected to a mechanical load and their viscoelastic properties measured using magnetic bead microrheometry. Such an experiment is designed to be analogous to the shear load HUVEC experience in the lumen of blood vessels. Thus, this study is directly relevant to relentless in vitro efforts to develop perfused microvessels by the biomaterials and tissue engineering communities.

2 .5.4.2 Biochemical Properties

Cell-material adhesions are governed by short range specific LR interactions and long range nonspecific molecular interactions that are dependent on the chemical properties of both the cell and material (a 2D modified surface or a 3D gel matrix scaffold). Different phenotypes have different cell receptors and different matrix materials have different ligands. Together, these are the biochemical parameters of a cell culture system. Examples of cell receptors include the integrins a5pi and avp3 (discussed earlier in Section 2.3.2), while an example of an adhesive ligand is the RGD sequence found in fibrin and collagen gel scaffold materials, among others.

Adhesion receptors on the cell surfaces bind to various ligands of the ECM or other cells, regulating cell signalling cascades (e.g., through kinase and phosphatase dependent pathways) and yielding the final properties of living cells and tissues. Cell-material and cell-cell adhesion sites, whether specific or not, may be largely different in in vitro 2D systems compared to 3D systems. For example, focal and fibrillar adhesions in 3D differ from adhesions on 2D substrates in their integrin/paxillin content and localization, and in tyrosine phosphorylation of FAK, as illustrated by Figure 2.9 [189]. Moreover, a number of adhesion molecules present in natural ECM materials are sometimes difficult to obtain and localize in in vitro systems. For example, Matrigel, a collagen derivative used to mimic the in vivo basal membrane, is known to possess a large number of trace molecules, including growth factors, which have the capacity to elicit substantial effects on seeded cell populations. Matrigel is widely used in 3D culture systems as a matrix scaffold and can lead to irreproducible results as the trace molecules are not all known. Novel polymeric materials can also create unexpected responses from cells. Therefore, it is

45 important to be cautious when comparing cell behavior on a plastic dish and in different scaffolds, whether natural or synthetic.

ots Integrin ______paxillin as Integrin ♦ paxillin fibronectin ______m w gad_____

i ■ £ 1■ ■HiH

Figure 2.9: In vivo 3D matrix adhesions differ from focal or fibrillar adhesions on 2D substrates. (A to E) Confocal images of indirect immunofluorescence staining of NIH-3T3 mouse fibroblasts in vitro on a 2D fibronectin-coated cover slip; (F to J) transverse cryostat craniofacial mesenchyme sections of an E13.5 mouse embryo. a5 integrin [(A) and (F), green] and paxillin [(B) and (G), red] colocalize in a fibrillar organization in mesenchymal tissue [(H), yellow in merged image indicates overlap of red and green labels], but not on a 2D substrate in vitro (C). Fibronectin [(D) and (I), blue] localizes to fibrillar structures in vivo, and merged images indicate substantial overlap of all three molecules [(J), white compared to (E)]. Note that focal adhesions (filled arrowheads) and fibrillar adhesions (open arrowheads) show differential localization of the a5 integrin and paxillin markers only on traditional flat 2D substrates in vitro. The 3D matrix adhesions (arrows) identified by triple localization are present in 3D environments in vivo. Scale bar: 5 pm. Reprinted with permission from reference [189]. Copyright 2001 American Association for the Advancement of Science.

Cell adhesion from chemical niches varies widely depending on cell phenotype. Moreover, physiologic state, ligand distribution, concentration, and synergistic effects of numerous non adhesive niches effect cell response. Chemical niches can consist of proteins (such as growth factors), ions, sugars, dissolved gases, or synthetic materials, among others.

Chemical niches are presented to embedded cells in 3D scaffolds through two means: (i) diffusion of molecules through the porous scaffold and (ii) ligands that constitute the composition of the scaffold. While the first mechanism is dependent on transport properties (discussed in the next section), the second mechanism depends on the chemical composition of the scaffold. For example, collagen and fibrin gels contain the RGD sequences and Matrigel

46 (composed of both laminin and collagen) contains RGD sequences of collagen and the YIGSR (tyrosine-isoleucine-glycine-serine-arginine) sequence of laminin. Both the RGD and YIGSR sequences are responsible for initial cell adhesion to scaffold materials composed of such sequences. Natural polymers such as collagen will often contain trace molecules that are likely to interact with entrapped cells. On the other hand, synthetic materials can be devoid of such residues; although commercial plastics often contain plastic additives and those can affect cell responses [190]. In the case of mass transport, angiogenic growth factors such as VEGF and fibroblast growth factor (FGF), known to stimulate migration and proliferation, can be supplemented into culture media and delivered to the embedded cells via convection and/or diffusion. As an example, consider angiogenesis. A relevant scaffold material is fibrin gel as it is well established that it stimulates angiogenesis [191, 192], Suppose that the response of the endothelial cells in the system is inadequate, one could then supplement their culture media with VEGF or FGF to further simulate an angiogenic response.

Deciphering how chemical niches (i.e., parameters) influence cell response and behaviour is not a trivial task from an analytical standpoint. Traditionally, techniques involve cell labelling using antibodies or other markers specific for a chemical niche. Nearly all labelling requires that the cells be fixed and/or permeabilized.

In addition to the microscopic techniques that were used to obtain the insights into cellular adhesion discussed in Section 2.3.2, spectroscopic methods also exist to allow quantitative assessment of membrane protein dynamics in living cells. These include fluorescent-correlation spectroscopy (FCS), image-correlation spectroscopy (ICS), and fluorescence-recovery after photobleaching (FRAP) [193]. Such techniques have yet to be exploited by the biomaterials and tissue engineering community. FCS tracks fluctuations in fluorescence intensity of labeled molecules in a very small volume (~ lfL) and correlates such fluctuations to the respective diffusion coefficient [194]. Two membrane proteins can be labeled simultaneously and their interaction assessed [195]. This particular case is known as fluorescence cross-correlation spectroscopy (FCCS). Cell membrane proteins can be relatively immobile making FCS and FCCS inapplicable. When such a condition arises, FRAP applies [196]. Fluorescence fluctuations of FCS and FCCS can be analyzed as a function of space producing an image in the techniques of ICS and image cross-correlation spectroscopy (ICCS), respectively [197, 198].

47 Studies have been published that provide some further insight into how cells respond to materials and biochemical niches with respect to traditional cell labeling. ICCS has recently been used to quantitatively measure the fraction of actin that interacts with a-actinin at the boundary of a migrating Chinese hamster ovarian cell on fibronectin-coated glass [199]. In separate studies, ICS was used to track the dynamics of the a5 integrin and its association with a-actinin and paxillin [200, 201]. An extension of ICS, known as spatio-temporal image correlation spectroscopy (STICS), allows one to quantify the direction and flow rate of membrane proteins during cellular processes, such as migration and proliferation [ 2 0 2 ].

The full potential of these spectroscopic techniques is beginning to be unraveled. Such techniques would be very powerful in understanding the true influence of chemical niches and other material properties on cell adhesion and behavior.

2.5.4.3 Transport Properties

2D and 3D cell culture systems are governed by drastically different mass transport phenomena. 3D matrix scaffolds drastically reduce gas and nutrient exchange rates between the cells and their surrounding medium. This has already been illustrated in Figure 2.7. In 3D systems, oxygen has first to transfer from air to the culture medium, and then must diffuse through the scaffold to reach the cells. While the first transfer step is present in 2D and 3D systems, the last step is only present in 3D systems and drastically effects cell behavior and in many cases, results in significant cell death. Furthermore, if introduction of a chemical niche into a 3D system via medium supplementation is desired, transport properties must be assessed to ensure that such a niche is delivered to the embedded cells in an adequate fashion.

A review of mass transfer phenomena and requirements for 3D cell culture systems is given by Martin and Vermette [161]. Very briefly, mass transport via convection (i.e., flow of nutrient

medium) is required for gel scaffold thickness ranging from 1 0 0 pm to 1 mm, as diffusion alone is not sufficient to provide oxygen and nutrients. As such, for larger thickness and 3D systems, reactor configurations must provide flow through the scaffold, while minimizing possible damage or undesired shear stress to the cells. Engineering principles must be used to evaluate convective and diffusive effects and shear stresses imposed on cells, and then to design 3D cell culture systems that provide sufficient mass transfer in a physiologically adequate fashion. 48 Hollow fiber reactors and hypothetical cell assemblages with functional “blood” vessels broadly constitute the most adequate reactor systems to achieve optimal mass transfer and minimize shear forces.

Oxygen is only sparsely soluble in typical culture media (e.g., human blood carries 50 times more oxygen due to hemoglobin). As such, oxygen carriers need to be developed to allow dense tissue growth in 3D culture systems. Most research on oxygen carriers is done as a potential for blood substitutes as opposed to increasing the oxygen content in culture media for cell culture systems. Oxygen content in culture media can be increased by employing, for example, perfluorocarbon-based (PFC) or hemoglobin-based oxygen carriers. A thorough review on their preparation and progress is provided by Centis and Vermette [203]. Briefly, while they dramatically increase oxygen solubility, some oxygen carriers appear to be somewhat cytotoxic. PFC-based carriers have been shown to be highly toxic to fibroblasts and thus do not appear to be applicable in biomaterials and tissue engineering applications [204]. Furthermore, liposome- encapsulated hemoglobin shows significant toxicity to HUVEC monolayers in 2D with 50% cell death in the first 24h [205], On the other hand, when cells were embedded in a 3D fibrin gel and grown using a culture medium containing liposome-encapsulated hemoglobin, no cytotoxicity was observed and even cell proliferation increased compared to controls [206]. It appears that if one wishes to enhance the oxygen solubility of culture media, a fine balance is required to ensure an adequate increase in oxygen delivery and minimal toxic impact to cells.

2.5.5 Materials for Three-Dimensional Cell Culture Systems

2.5.5.1 Natural Materials

Ideally, in vitro systems should be fabricated from the same molecules present in in vivo systems, while adjusting the exact composition and localized growth factors to mimic specific cell responses. Because of difficulties in purifying ECM molecules for scaffolds with well defined local mechanical and chemical properties, this requirement is somewhat unrealistic. However, ECM gels, such as fibrin, collagen and Matrigel can be used, only to some extent, to acquire desired matrix rigidity [157]. Furthermore, the exact ECM composition including trace chemical factors and local mechanical properties for most tissue types is not well known.

49 The protein collagen is known to be present in living tissues in over a dozen types and in various fibrillar forms depending on the tissue. However, collagens can be difficult to purify while maintaining their architecture. Small amounts of impurities may cause unexpected cell responses in culture systems fabricated from extracted natural ECM. Accordingly, collagen gels have been noted to show important differences from in vivo systems relative to cell attachment, migration, and proliferation [189]. The 3D movement of lymphocytes in native lymph nodes appears random and significantly faster than what was observed in vitro in collagen gels [207]. Moreover, cell-cell interactions are seemingly prolonged in vivo compared to in vitro model systems [208]. Cells can be mixed with a collagen solution prior to polymerization in order to embed cells within the matrix. Importantly, the density of collagen fibers and their orientation has a large effect over cell migration and morphology [209]. Moreover, the mechanical loading of the material impacts cell fate. Mass transfer limitations for collagen-based systems have until now prevented any reported growth of large tissues. Collagen matrices may be blended with fibronectin or glycoaminoglycans such as hyaluronan in order to achieve in v/vo-like matrices [210, 211]. However, such blends are also difficult to reproduce and can yield different collagen structures as well as varying cell responses [212]. The supernatant of a Schwannoma which predominantly contains laminin, collagen IV, and nidogen, known as Matrigel, is often used to mimic the basement membrane. However, as a large number of molecules can be found in Matrigel, it is often difficult to reproduce observations made using this material [213]. As such, care must be taken when interpreting results obtained from collagen-based systems as the fabrication method and exact source of the material is likely to have an important influence on the observations.

Fibrin is used mainly in applications where very low levels of trace chemicals are desired as it is possible to obtain relatively pure reactives (i.e., fibrinogen from blood plasma and recombinant thrombin). Fibrin can be covalently modified to incorporate growth factors or other biochemical niches [214, 215]. Furthermore, it is possible, to some extent, to modulate the porosity and mechanical properties of fibrin by adding other compounds. For example, hyaluronic acid, a natural proteoglycan, increases fibrin gel porosity and hydration thus promoting cell migration [216]. Aligned poly-L-lactic fibers can be added prior to fibrin gel formation to guide fibroblast and keratinocyte migration [217], while biodegradable knitted elastomeric fabric can be added to improve the mechanical properties of tissue-derived cardiac constructs designed for in vivo

50 implantation [218]. Magnetic beads coated with thrombin can be used to align fibrin fibers in 2D at the nanoscale level using magnetic fields [219]. The main drawback in using fibrin is its mechanical properties change over time and degradation occurs. Aprotinin, a plasmin inhibitor, may be added to slow degradation [ 2 2 0 ].

2.5.5.2 Synthetic Materials

Synthetic materials are being studied as replacements of certain natural materials. Such materials have the mandate to behave in a similar fashion to natural materials. As such, synthetic materials must allow specific localized interactions with certain types of cells, be non-toxic, and permit cell migration. In some cases, synthetic polymers are desirable as they allow the isolation of specific niches for cell study. For example, to some extent, using ligand-coated polyacrylamide gels for 3D cell culture allows simultaneous tuning of chemical and mechanical niches [157]. Because frequent novel materials are reported in the literature, a very large number of materials can be used. Degradation of the synthetic material is sometimes preferred to allow cell migration and tissue development. Three vital requirements for the materials are non toxicity, ease of sterilization, and the promotion of cellular adhesion. However, while some of these materials can achieve well defined porosities and other mechanical properties, many of these materials lack adequate chemical properties to promote cellular adhesion and thus require some type of additional chemical modification.

Transformed natural materials, such as cross-linked polysaccharides (e.g., agarose and alginate) may be used to encapsulate cells. Simple 3D constructs made of encapsulated cells in alginate or agarose beads are often encountered in biomaterials and tissue engineering literature, noticeably for the culture of chondrocytes, pancreatic islets, and hepatic cells. Encapsulation is reported to increase cell viability, tolerance to shear stress, and proliferation as well as inducing in vzvo-like behavior [167]. Systems made from these polymers show flexibility in their mechanical properties as their elastic modulus is dependent on the molecular mass of the polymer [157]. They have been widely used in 3D systems where high shear stresses are present [161].

In cancer research, tumor cells are frequently studied to assess the fundamental differences between 2D and 3D cultures. The cells are usually studied in the form of spherical assemblies termed spheroids [221]. These assemblies can be obtained by spontaneous aggregation from the 51 2D cultures of certain cell lines (e.g., breast cancer). Cancer cells can also be grown on agar, agarose, or recombinant basement membrane to help induce the formation of spheroids. Both techniques have yielded useful data, although inadequate mass transport limits spheroid size and viability of inner cells. Moreover, the interaction of cancer cells with stromal cells cannot be modeled with these systems. However, novel scaffolds are being developed to allow such co cultures [2 2 1 ].

Bone cells necessitate rigid (i.e., inelastic) scaffolds to mimic the in vivo ECM. As such, composite systems made of polymers and ceramics are actively being developed to combine the mechanical properties of ceramics with the advantages of polymers (e.g., biodegradability, flexibility, and available chemical sites) [222, 223]. A review of materials developed to culture bone cells in 3D is presented by Rezwan et al. [224].

In most studies, cells are cultured in synthetic organic polymer scaffolds of various shapes, adhesion characteristics, mechanical properties, and porosities. An example is hydrogels that combine receptor binding sites to exert cell traction and protease-sensitive degradation sites to allow cell migration [225]. These hydrogels are fabricated from multi-arm vinyl sulfone functionalized poly(ethylene glycol) backbones with either mono-cysteine adhesion peptides or bis-cysteine protease-sensitive peptides to allow localized degradation. This scheme allows a complex interaction between cells and synthetic materials somewhat akin to the interactions present in native tissues. An extensive list of polymers currently used is reviewed by Hutmacher [226]. Large efforts are also devoted to scaffold creation with predetermined architecture using computational topology design and solid free-form fabrication [227].

2.5.6 Micropatterned Cell Cultures

The majority of in vitro cell culture systems, whether 2D or 3D, involve dispersing cells in a homogenous fluid within a largely homogenous construct such as a Petri dish or well plate. Unfortunately, cells in in vivo systems are surrounded by a complexity of physical and chemical microenvironments [228]. As such, large efforts have been made in the patterning of cells to surfaces at micron resolution in a technique known as micropatteming.

52 Micropatteming has a variety of potential applications. In fundamental biology, micropatteming can be used to position biochemical parameters to understand how the quantity and geometry of localized ligand distributions effects cell response and any subsequent behavior [229, 230]. Biomaterials scientists and tissue engineers exploit any understanding gained to control cell response, growth, and behavior [231, 232].

Numerous techniques can be used to produce micropattemed surfaces. The most widely used is that of soft lithography [233, 234]. Within soft lithography are the techniques of microcontact printing and microfluidic patterning. Another common technique is that of photolithography, where ultraviolet illumination is used to pattern a substrate. For a concise review of the engineering approaches to produce micropattemed surfaces, the reader is referred to an article by Falconnet et or/. [235].

Few techniques exist to spatially localize biochemical niches in synthetic materials to genuinely mimic in vivo tissue architecture. This currently constitutes a limitation for the design of 3D systems to produce viable human tissue substitutes. Lithography-based deposition methods have been developed to produce well defined 2D surfaces for cell study [236]. However, these techniques are not transferable to 3D systems. Layer-by-iayer printing techniques have been reported to enable 3D positioning of matrix material. For example, 3D printing of porous hydroxyapatite scaffolds for bone engineering achieves well-defined dimensions, although lacking localized biochemical parameters [237, 238], Polylactic acid 3D printing from CAD models was also reported, again lacking localized biochemical parameters [239].

Mircropatteming has also found use in further understanding how cells respond to surface topography [240]. However, quantitatively defining surface topography is challenging [241]. Early reports indicated that cell adhesion is enhanced when the roughness of a surface is increased [242]. Later, numerous studies would show that adsorbed protein conformation is sensitive to nanoscale topography features of the surface [243-245]. The nanoscale is a similar length scale to that of the proteins themselves. As mentioned earlier in Section 2.3, protein adsorption directly impacts cell adhesion and subsequent response. Topographical features at the microscale can be too large to directly impact protein adsorption [246] and thus, likely would have no direct impact on cell adhesion and response.

53 Cell migration has been shown to be somewhat controlled via the use of micropatteming. On flat surfaces, cell migration is essentially random in the absence of biochemical stimuli [247]. Later, micropatteming proteins showed that one can in actual fact guide migratory cells [248- 251]. As suggested by Curtis and Wilkinson [252], the guidance of migratory cells along micropattemed grooves and ridges can be explained by the fact that the actin in the cell cytoskeleton is organized in a way that requires minimal work. Thus, the cells migrate along the path of the grooves and ridges. However, controlling the direction of migration has yet to be achieved. While guiding the migration of cells is a step forward from that of adhesion only, far more control of cell behavior is needed to apply such micropattemed systems for tissue engineering applications. Patterning cell migration in a 3D environment can be achieved by fibres disposed in a given pattern within a gel, as it was reported by us [140, 141].

When micropatteming techniques for modulating cell response first surfaced, it apprears they were touted as potential revolutionary tools for biology, biomaterials, and tissue engineering research. In biology, it appears micropatteming has substantiated the fact that ligand composition, localization, and concentration dictates cell adhesion and subsequent behaviors such as morphology and survival. However, over a decade since its introduction in biomaterials and tissue engineering research, micropatteming appears to still remain a “potential” tool as opposed to a “valuable” tool. Perhaps the reason for this is that miropattemed systems do not directly alter cell-cell interactions or transport properties, both being major contributors to tissue growth and cell response to biomaterials surfaces. As such, the majority of culture systems based on micropattemed surfaces are more an extension of 2D systems rather than a truly 3D approach.

2.5.7 Current State and Future Direction of Three-Dimensional Cell Culture Systems

Cells cultured in 3D systems clearly demonstrate large differences, for example, relative to morphology and protein expression compared to those grown in 2D environments. This should not be surprising considering the extent to which biochemical, mechanical, and nutrient and waste transfer properties differ for the two types of systems. However, large efforts still need to be deployed in order to achieve the fabrication of 3D environments in vitro that can successfully

54 reproduce the in vivo reality with low variability and relative ease of use to allow everyday use. 3D culture systems which have been reported in the literature often present limitations in nutrient transfer, give inadequate or unmeasurable biochemical cues to cells, or fail to give important mechanical signals. Progress is made every year to improve nutrient transfer through bioreactor design and attempts to induce channels in lumps of cells (e.g., by angiogenesis).

Petri dishes and multi-well plates currently constitute the simplest and most widely used “bioreactors” for 3D cultures. Cells are embedded within a polymerized porous scaffold material in a multi-well plate or Petri dish and nutrient-rich supernatant is supplemented onto the matrix and changed during culture. Mass transfer in such systems is strongly limited due to inadequate diffusion of nutrients through the porous scaffold. As such, only thin sections of embedded cells can be used. Cells located closer to the supernatant beneficiate from enhanced oxygen and nutrient access. Mass transfer in such systems may be improved by the use of well inserts that suspend the embedded cells on a highly porous membrane that allows diffusion through the top and bottom of the scaffold containing embedded cells. Without convective mass transport, however, only thin sections of embedded cells can be used. More complex bioreactor systems are needed to enhance mass transfer [161].

Novel materials are constantly reported to allow cell growth in vitro. The next few years should witness the emergence of integrated systems. However, some aspects of the in vivo environment should rapidly be addressed. For example, spatial localisation of key biochemical niches is very likely to be important to reproduce the highly non-uniform in vivo tissues. While developments are generated on flat surfaces to achieve this goal, very few 3D systems address this critical issue.

2.6 Modification, Properties, and Analysis of Surfaces

2.6.1 Modified Surfaces in Biomaterials Science

An approach to understand and subsequently control the cell-surface interaction is the modification of surfaces with physical and chemical properties that result in protein adsorption with conformation/orientation favorable for a particular cell response. For example, the design 55 of a surface using adsorbed VEGF and Fn to promote the migration and proliferation of endothelial cells [ 8 8 ]. Such surfaces would be beneficial to better understand the complex process of angiogenesis and revascularization of new and/or damaged tissues [253].

Surface modification can employ plasma polymerization and/or SAM, two approaches that will be further discussed below. Another approach is the design of low-fouling surfaces [254]. The term low-fouling relates to the resistance of a surface towards protein adsorption, which subsequently results in little to no cellular adhesion to the surface. Common low-fouling modifications involve the attachment of PEG or the equivalent poly(ethylene oxide) (PEO) [254- 257], dextran-derivatives [23-25,258-260], or phospholipids [261-265] to surfaces.

In this section, we discuss the modification of substrates using SAM and plasma polymerization. Different types and the advantages, disadvantages, and complexities of each are discussed. The surface properties that result due to such modifications and methods of analysis are outlined. Particular emphasis is placed on the low-fouling properties of modified surfaces and the limitations of surface analysis using contact angle measurements.

2.6.2 Surface Fabrication

As shown earlier in Figure 2.1 A, prior to the introduction of proteins and/or cells, substrates can undergo modification for the purpose of enhancing their chemical and biological activity. Such modifications typically alter the chemistry and topography of the substrate [246], As shown in Table 2.3 and Table 2.4, there are a wide variety of materials used for surface modification in biomaterials science [266]. Materials that are typically used as substrates are composed of alloys or metals such as titanium or gold. Silicon-based materials such as glass and mica are also very common. In addition, the use of polymers such as polytetrafluoroethylene (PTFE) and fluorinated ethylene propylene (FEP) have been used as substrates.

56 Table 2.3 Surface modification substrates and chemicals

Surface Modification - Substrate - , Modification Compounds Self-Assembled Monolayers Organothiol Noble Metals -gold, Au(lll) - copper - organothiols (/?-SH) - silver - organosulfides (R-S) Semiconductors - organodisulfides (R-S-S-R) - indium phosphide - gallium arsenic Organosilane Silicon-based Materials - glass - quartz - OTS (octadecyltrichlorosilane) - mica - APTMS [(3-aminopropyl)trimethoxysilane] - silicon oxide - alkyltrichlorosilane - Tridecafluoro-1,1,2,2-tetrahydroochyl-l- Metallic Compounds dimethylchlorosilane - aluminum oxide - zinc selenide Plasma Polymerization Amine Functional Monomers - glass - n-heptylamine - mica - ehtylene diamine - FEP - propylamine - PTFE - allylamine - Ti Alloys Other Functionalities - acetaldehyde (CHO) Plasma Treatment - glass -At - mica - n 2

-FEP - 0 2 -PTFE -He - metals - H 2

Low-Foulim Phospholipids - PC (phosphatidylcholine) - glass - Phosphatidylserine - mica Synthetic polymers -FEP - PEG (poly(ethylene glycol)) -PTFE - PEO (poly(ethylene oxide)) Dextran-derivatives - CMD (carboxymethyl-dextran)

57 Table 2.4: Properties of cell culture surfaces Surface Modification Functionality Properties charee wettabilitv

Self-Assembled Monolavers CH3 neutral hydrophobic OH neutral hydrophilic

n h 2 + hydrophilic COOH - hydrophilic

c f 3 neutral hydrophobic Commercial Substrates Untreated polystyrene n/a neutral hydrophobic Tissue culture grade n/a - hydrophilic polystyrene Primaria™ NH2/COOH +/- hydrophilic Polylysine-coated plastic n/a + n/a Polvmer Grafted

PEG-grafted n h 2 COOH Dextran-derivatives COOH All low-fouling" (CHO) Note: For the row pertaining to SAM, the functionality corresponds to the tail-group." The property “low-fouling” is somewhat interpretive as discussed in Section 2.6.7. For low-fouling behavior, one must define the tolerance of acceptable adsorption for their particular application and then use the appropriate experimental method and conditions to produce a low-fouling surface.

Two available chemical surface modifications of substrates are the use of plasma polymerization [93, 267] or SAM [27, 50]. Following the modification of a substrate using plasma polymerization or SAM, the modified surface can be conjugated with a protein, chemical spacer {e.g., PEG, dextran-derivatives), or any compound that will affect the surface activity.

58 2.6.3 Self-Assembled Monolayers (SAMs)

Perhaps the most popular surface modifications in cell-materials research is the use of SAM [268-271]. A SAM consists of molecules each containing a head and tail group. The head-group is adsorbed to a substrate while the tail-group is available for further interaction. This is illustrated in Figure 2.10. Proteins can then be physisorbed or chemisorbed onto the SAM- modified surface, followed by the seeding of cells.

Cells

Biomolecules

Tail - group

monolayer "n

iubstrate

Figure 2.10: Self-assembled monolayer (SAM) exposed to proteins and cells. Figure not drawn to scale. The head-group is chemisorbed to the substrate. The tail-group (in this example, a hydrophobic methyl, CH 3 functionality) is available for interaction. In this particular example, the SAM is exposed to biomolecules (chemisorption or physisorption) prior to cell seeding. Alternatively, cells can be seeded directly to the SAM. When this is the case, the cell will synthesize and secrete ECM proteins to aide in adhesion.

The popularity of SAM in cell-materials research stems from the commercial availability of SAM molecules with specific head-tail group combinations. The availability of different tail- groups allows researchers to alter the chemistry (see Table 2.4 for specific examples) of surfaces rapidly. Unfortunately, the popularity of SAM is also due to a perceived notion that preparation can be performed with ease and simplicity. Such a notion should be taken with caution, as there is still much research currently being performed to understand the mechanisms of self-assembly and to identify the experimental conditions necessary to produce well ordered homogeneous

59 SAM in a reproducible fashion. As shown in Table 2.3, two popular SAM systems include the use of organosilanes on hydroxylated glass, silicon oxides, or aluminum oxides and organosulfurs on noble metals.

The most popular and studied organosulfur SAM system is the organothiol (7?-SH) on Miller oriented Au(lll) surfaces. Here R represents the tail-group and the remaining chemical functionality represents the head-group. Other organosulfur-noble metal systems include organosulfides (R-S) and organodisulfides (i?-S-S-J?). Other noble metal substrates include silver and copper. The popularity of organothiol SAM on Au(l 11) is a result of the strong affinity of the S-Au bond and the impeccable well defined structure of the S adsorbed to the Au(lll) surface [272].

The formation of organosulfur SAM is a complex and not completely understood phenomenon. It has generally been proven that the formation of organothiol SAM on Au(l 11) follows a two- step process. The first step occurs rather quickly (i.e., in a few minutes) which involves rapid chemisorption of the head SH group to the Au(lll) surface in a diffusive fashion. Upon completion of this initial step, the monolayer has reached 80-90% of its final thickness. The second step is less rapid, taking hours. In this step, intermolecular Coulomb and VDW interactions occur between both the adsorbate-substrate and adsorbate-adsorbate, resulting in the formation of an organized 2D crystal surface structure [273-275].

There is still much controversy regarding the influence of intermolecular electrostatic interactions to the degree of disorder [276]. Furthermore, even when assuming an ideal organomonolayer, one must still proceed with caution. It has been shown that upon completion of an organosulfur SAM, assumed homogeneous, the strength of H-bonding between COOH or

NH2 tail-ends of the chemisorbed SAM and organosulfur molecule tail-ends that remain in suspension results in a partial multilayer [277]. This results in a surface of significant heterogeneity and roughness. Care must also be taken if one is preparing their own gold surfaces. Proper annealing must be performed to ensure a high quality Au(l 11) surface with large, flat, virtually defect-free terraces. Together, this indicates that organosulfur SAM preparation and fabrication must be pursued with caution as surface homogeneity is not guaranteed.

60 Organosilanes on modified glass surfaces are another popular SAM system in cell-materials research. They sometimes appear to be favorable over organothiols. The primary reason being that glass can be used as the underlying substrate. For organothiols, vapor deposition of gold or another noble metal onto a silicon-based material such as mica or glass is required. Gold-coated substrates can also be purchased commercially. However, their high cost and partial transparency makes them unattractive. Shipping time can be lengthy as the gold-coated substrates are usually produced upon order to provide customers with oxide-free gold substrates. On the other hand, organosilane monolayers require bare glass surfaces and various chemical reagents both of which are readily available commercially, further enhancing their convenience.

Organosilanes contain a reactive silane group on one end (i.e., head-group) and a functional group (i.e., tail-group) on the other which depends on the application or surface desired. The formation of organosilane SAM has been extensively studied [278]. Many different silinization chemical reactions exist. Most simplistically, silinization is made on hydroxylated or amine surfaces. The head-group of the silane SAM molecule contains a Si residue that interacts with the hydroxylated/amine surface to form a Si-O bond and subsequently release an intermediate. Different types of silanes are classified by the intermediate.

While the transparency of glass, the availability of materials and reagents, and the perceived simplicity of silanization make organosilane SAM attractive, the preparation and reproducibility of such surfaces are extremely sensitive to experimental conditions. For example, alkyltrichlorosilanes are sensitive to moisture conditions. The absence of water results in incomplete monolayer formation while an excess in water results in solution polymerization [279-283]. Non-ideal moisture conditions result in heterogeneous surfaces. To provide an example of the consequences of an incomplete monolayer, consider that COOH groups adsorb on metal oxide surfaces (e.g., aluminum oxides) [272]. Thus, an incomplete monolayer could result in nonspecific adsorption of the carboxylic acid groups of a protein with the exposed aluminum oxide surface resulting in a modified surface with a significant degree of heterogeneity, nonspecific adsorption, and variable roughness, leading to a possible misinterpretation of the cell-surface interaction. The stability of some widely used alkyl- terminated organosilanes for biomedical applications such as cell culture and medical implantation has also been questioned by researchers [284, 285].

61 Regardless of whether or not one is using organothiols or organosilanes to study cell behavior, careful consideration must be taken regarding the formation, homogeneity, and stability of SAM. In general, the belief that SAM consisting of longer chains between the head-tail lead more rapidly to stable self-assembly is consistent, regardless of the SAM system under study. This is likely because of a larger quantity of electrostatic and VDW adsorbate-adsorbate interactions. The general difficulty amongst the development of SAM is reproducibility, stemming from an incomplete understanding of the dynamics behind their formation. The aim of theoretical work on SAM today is to better understand the surface and intermolecular interactions necessary for the formation of uniform reproducible SAM. Experimental characterization of SAM to ensure reproducibility also still remains a difficult task [286].

Self-assembly today, in general, remains a widely studied field [271, 287]. The complexity and dynamics of SAM formation has attracted much interest and research in areas of experimental and theoretical physics, chemistry, and nanotechnology [288]. However, observed cell responses and functions sometimes may be due to unintentional surface heterogeneity, nonspecific adsorption, and roughness.

2.6.4 Plasma Polymerization

Plasma polymerization is a widely used technique to chemically modify inert substrates. Plasma polymerization can be used to introduce a thin film of amine functionalities. Subsequently, additional compounds such as PEG spacers [93, 254, 257], polysaccharides such as CMD [23- 25], or any heterobifunctional cross-linking compounds [79, 89] can be conjugated to the amine- functionalized substrate. This is usually followed by the covalent immobilization of ECM proteins and peptides to further enhance surface activity to biological species. The addition of spacers such as CMD and PEG with low-fouling properties is advantageous as many biomolecules (i.e., ECM proteins) undergo a significant reduction in activity upon adsorption directly to a substrate [289]. The spacer acts as a means of simultaneously retaining the activity of the biomolecule and preventing nonspecific adsorption of proteins and biomolecules.

Numerous methods of plasma surface modification exist, such as plasma treatment. However, in plasma treatment, the gas comprising the plasma consists of a non-polymerizing gas. Thus,

62 plasma polymerization should not be confused with plasma treatment. In plasma polymerization the plasma state is driven by electric glow discharge usually via RF power generators. Because the plasma state is not generated via heating, plasma generated by electric glow discharge is often referred to as “low-temperature plasma”. In general, plasma polymerization is often referred to as “glow discharge plasma polymerization” where organic monomer vapors are polymerized. Furthermore, it should be pointed out that glow discharge plasma polymerization should not be confused with “plasma graft polymerization” and “plasma radiation polymerization”. The distinct features of the many plasma modifications are outlined in a nice review by Yasuda [290]. A comprehensive review on surface modification using plasma methods for biomolecule and cell immobilization can be found in an excellent review by Siow et al. [291].

For the production of reliable high quality plasma polymerization modified substrates, tedious optimization of the plasma parameters is required. The quality and quantity of the thin film is highly dependent on experimental parameters such as monomer pressure, input power, deposition time, and the nature of the monomer itself. Unfortunately, there are no universal quantitative values for one specific monomer or application. All parameters are dependent on the geometry and spatial dimensions of the chamber, all of which vary among laboratories. As a result, it is recommended that users perform optimization experiments. A very effective simple method is the use of a factorial experimental design for the determination of optimal plasma polymerization parameters [292].

There are disadvantages when performing plasma polymerization on a surface [266]. Due to the complex composition of the plasma and a variety of reaction possibilities, when an organic monomer is introduced to a surface via plasma polymerization, there in actual fact exist many unknown functionalities on the surface [293, 294], Thus, for an amine monomer such as n- heptylamine, while the modified surface is primarily composed of amine functionalities, there in actual fact exists other nitrogen containing non-amine (e.g., NH) functionalities. Post experimental plasma polymerization conditions also can affect the properties of a modified surface. Immediately after plasma polymerization, the substrate can be placed in a liquid solution containing spacers (e.g., PEG, dextran-derivatives, etc.) or heterobifunctional cross- linking molecules (e.g., sulfo-SMCC, etc.). Unfortunately, plasma polymerized thin films can be

63 unstable in solution [295]. When some plasma polymerized thin films are exposed to a liquid solution, they can undergo swelling. This can lead to the breaking of adhesive bonds resulting in partial delamination of the thin film [93, 289]. Extreme cases of plasma polymerized thin film swelling have also resulted in significant reduction in thin film density and in the portrayal of porosity. As a result, molecules exposed to the thin film can interact with amine functionalities buried several layers in the thin film resulting in surface roughness [296]. Another problem is that many surface deposited plasma polymers undergo rapid oxidation, reducing their reactivity [297]. Swelling and oxidation problems can be somewhat reduced by using a rapid covalent mechanism immediately after plasma polymerization. Carbodiimide chemistry is often used for such a purpose. Some plasmas, particularily n-heptylamine in a study by Martin et al, have been shown not to swell when immersed in aqueous solutions [292]. Conversely, Vasilev et al. performed a detailed study on n-heptylamine plasma and found swelling that was dependent on the radio frequency power of the plasma process [298]. The reason for these somewhat contradictory results is perhaps due to the choice of underlying substrate. Martin et al. used FEP substrates while Vasilev et al. used silicon wafers. In our personal experience, we found plasma polymer layers of M-heptylamine on silicon wafers to undergo significant delamination upon exposure to aqueous solution. An interesting study would be investigating the dependence of n- heptylamine plasma layer characteristics on various substrate compositions. A debate also exists on whether microwave frequency is superior to generate plasma as opposed to the commonly used radio frequency [93, 299].

The disadvantages of plasma polymerization manifest from the fact that technical exploitation of the process far outweighs any scientific research of the physics and chemical reactions that result in the modified surface. Any research into such mechanisms is largely based on speculation to explain empirical observations. Unfortunately, the physics of plasmas has largely been studied by physicists using a non-polymerizing gas as plasma. Little theoretical and analytical work has been done on polymerizing plasmas. Also, few analytical instruments are commercially available to diagnose in situ plasma polymerization processes, probably owing to the often very corrosive nature of the generated by-products. As a result a complete picture of the process of glow discharge plasma polymerization does not exist.

64 Despite difficulties in the optimization, instability of the modified substrates in solution, and a sparse theoretical picture, plasma polymerization modified substrates remain of interest. The primary benefit of plasma polymerization modified surfaces is their making is significantly quicker when compared to SAM-modified surfaces. The lengthy time required for the formation of SAM can lead to problems as if an inadequate time is allotted for their formation; the result is a heterogeneous surface. Of course, the choice of modification is equally dependent on the application and aim of the experiment being performed. If well fabricated and characterized, a SAM-modified surface, particularly an organosulfur, is superior to plasma polymerization modified surfaces due to the absence of unknown functionalities. With SAM, the only surface functionality should be that of the tail-group.

As previously outlined, plasmas composed of non-polymerizing gases can also be used for modification in a process known as plasma treatment. In biomaterials, plasma treatment can be used for surface cleaning (usually metallic materials) and modification [300]. Plasma treatments for cleaning are typically composed of N 2, O2, Ar, He, or H 2 . Unlike plasma polymerization where the surface acts as mechanical support for the thin film [301], plasma treatment can etch atoms from the surfaces and replace them with new functionalities [291]. The most common plasma treatment is that using ammonia plasma to functionalize the surface with amines. Plasma-treated surfaces can significantly alter the physical morphology of the substrate. Thus, one’s observations may be due to such morphological changes as opposed to chemical modification. Users of plasma treatment often do not address the resulting heterogeneity and morphological changes of the surface as being the contributors to varying cell response and/or biomolecule interaction. It is almost always assumed that the cell response and/or biomolecule interactions are a result of an assumed chemical substrate modification. However, this assumption is not always true as roughness and heterogeneity simultaneously contribute to such responses. Furthermore, plasma-treated films can undergo delamination exposing cells and/or biomolecules to the underlying substrate and not to the chemical modification.

2.6.S Properties and Analysis of Modified Surfaces in Cell Science

Cell behavior to a surface is dependent on the surface properties. Two of the most common surface properties considered by the researchers are the surface polarity (i.e., charge) and

65 wettability. Wettability can be thought of as the tendency of a liquid drop to spread (i.e., wet) a surface. Table 2.4 summarizes the properties of common surfaces and their modifications. Included are common functional tail-ends (i.e., surface chemistry) of SAM, spacers used post plasma polymerization, and commercially available surfaces unmodified by the user, along with their properties. Unmodified substrates are usually immediately used for standard cell culture practice. Plasma polymerization modified substrates with grafted CMD or PEG are further modified via the chemisorption of ECM proteins or other biomolecules with amide bonds [23] or Schiff base linkages [302, 303]. For SAM-modified surfaces, the ECM proteins are introduced to the surface without the use of a catalyst to entice chemisorption relying solely on physisorption mechanisms [50, 304]. However, chemisorption is possible for the SAM-modified surfaces should that be applicable [88].

A great deal of emphasis has been placed on surface wettability and charge, as shown in Table 2.4. It is generally believed that hydrophilic surfaces are superior to hydrophobic for cell adhesion, the reason being that proteins adsorb to hydrophilic surface with less affinity, allowing greater freedom of conformation and orientation. Subsequently, a greater number of cells adhere to a hydrophilic substrate relative to hydrophobic. Proteins and other organic molecules portray a net charge that is dependent on their isoelectric point. A protein or organic molecule can interact with a surface of opposite charge through Coulomb interactions.

In addition to AFM and SFA discussed earlier, surface characterization typically involves combinations of X-ray photoelectron spectroscopy (XPS), contact angle measurements, immunochemistry (IC) and immunofluorescence (IF), secondary ion mass spectrometry (SIMS), and matrix-assisted laser desorption/ionization (MALDI). XPS is employed to ensure the addition of a thin film whether it be, plasma polymerization, SAM, or a protein film as the spectrum is dependent on the atomic composition of the surface [305]. XPS has also been used to determine surface coverage of adsorbed proteins [306], although the use of XPS to characterize surface coverage can have serious limitations. Contact angle measurement is used to determine the wettability of the substrate. While contact angle measurements are very common in biomaterials surface science, the information they provide is ill defined. Furthermore, the contact angle of a surface is highly influenced by surface heterogeneity and roughness. The applicability of contact angle to biomaterials surface science is very

66 questionable. We address this matter in detail in the next section. IC/IF employs the use of antibodies for the specific recognition of proteins or fragments thereof as a semi-quantitative measurement of the surface protein density. Two other common methods of measuring the surface density of a protein are radio labeling and fluorescent conjugation using, for example, fluorescein or rhodamine. SIMS and MALDI can be used to determine the chemical composition of a surface [307, 308]. Each of these characterization techniques are not without limitations. One must determine which techniques are most applicable to their research.

Mechanical properties of modified surfaces are often overlooked. One possible reason for this is that mechanical properties are generally difficult to explicitly quantitatively define [309]. For example, while modification of surface roughness has been studied thoroughly in the past, it is difficult to quantitatively define “roughness”. Many studies aim to chemically modify a surface while not considering any resultant roughness. Such roughness is unintentional [252], and this may lead to studies in which one might believe they are observing cell responses to chemical modification, while they could be observing responses to both roughness and the chemical modification. Furthermore, the flexibility and the elasticity of a modified surface also have been shown to effect cell response, particularly migration and the formation of focal adhesion contacts [310]. In conclusion, one must consider that cell response is dependent on each the chemistry, topography, and elasticity of the modified surface, not to mention the elasticity of the cell itself. We believe that disregarding any of these individual contributions towards cell response has resulted in confusion.

2.6.6 Contact Angle in Correlating Cell Responses toward Biomaterial Surfaces

It is difficult to extract any relationship between cell responses and biomaterial contact angle measurements by water. Table 2.5 and Table 2.6 summarize some results of establishing cell and protein interaction as a function of biomaterial water contact angle, 9 [27, 35, 311-328]. Some studies tend to indicate that cell responses can be related to the water contact angle found on specific biomaterials, but overall, no clear links can be found. Comparison of results from different studies is complicated by the fact that experimental parameters often vary from one study to another. For example, cell phenotype, growth conditions, and cell densities are often

67 different. Furthermore, ranges of water contact angle values between surfaces are often too narrow, making it difficult to establish meaningful correlation between water contact angle and cell response. More importantly, the analysis of Table 2.5 and Table 2.6 points out that far more important material characterization should be considered when trying to correlate biomaterial properties to cell responses. These can include surface chemistry, surface molecular architecture approximated by surface force analyses, and surface roughness, only to name a few.

Table 2.5: Static water contact angle and cell/protein responses

cells/proteins material cell response ref \o 00 O stromal various materials no clear relationship between [316] chondrocyte polystyrene-modified with plasma 39°-107° cell number seemed to decrease as 0 [329] polymerization increased chondrocyte, fibroblast films of poly(lactic acid) (PLA), 37°-85° cell number increased as 9 decreased [319] polyfglycolic acid) (PGA), and poly(lactide-co-glycolide) (PLGA) treated with chloric acid fibronectin, MG63 cells Modified titanium-aluminum-vanadium 4°-43° No clear relationship between 8 and binding [321] alloy of fibronectin or cell attachment osteoblast CO2 laser treated 316 LS stainless steel 78°-89° Cell proliferation increased as 9 decreased [315] O endothelial Plasma treated polystyrene N> -vl 00 © Different adhesion between nontreated and [325] plasma treate polystyrene but no relationship between 9 and cell adhesion for treated samples osteoblast Titanium alloy substrates (rod and sheet of 75° (rod) Cell density seemed to decrease as 8 [313] Ti-6Al-4 V) 85°(sheet) decreased 0 0 fibroblast SAMs of different alkyl chain lengths (Si 00 O Cell coverage and detachment rate decreased [311] as 8 increased fibroblast Silanes with SH, NH2, and CH, 0°-92° No clear relationship between 8 and cell [35] attachment osteoblast Titanium surface oxides 48°-83° No clear relationship between 8 and cell [328] attachment or proliferation Smooth muscle Plasma polymerized polyethylene 42°-103° No clear relationship between 8 and cell [320] terephthalate) (PET) and PTFE surfaces density or growth yeast Glass, polypropylene, polystyrene, 33°-103° Yeasts seemed to be more firmly attached to [314] stainless steel hydrophobic supports yeast Membranes made of polyacrylonitrile, 0 00 0 No clear relationship between 8 and initial [317] polysulfone, modified polysulfone, cell adhesion rate polypropylene, and Teflon

68 The usefulness of measuring water contact angles on solid biomaterials surfaces is therefore opened to question. Can water contact angle on a biomaterial surface predict protein adsorption and/or cell responses? This section briefly addresses such a question.

Table 2.6: Advancing, receding, and dynamic water contact angle and cell/protein responses

cells/proteini^ < material cell response ref endothelial individual and mixed heparin/albumin #adv - 25 ° - 90° no clear relationship between 0 and cell [322] layer on poly(acrylic acid) attachment albumin, fibrinogen semi-interpenetrating polymer 9 a m = 25 ° - 90" protein adsorption decreased as 0 decreased [312] networks with PEG-diol hepatoblastoma Membranes made from 0a d v = 53°-80° no clear relationship between 0 and cell adhesion [318] polyetherimide, polyacrylonitrile- N-vinylpyrollidone copolymer, polyacylonitrile, and polyvinylidendifluoride 10% FBS in Poly(3-hydroxybutyrate) and poly(3- 9rec = 46°-62° no clear relationship between 0 and protein [327] DMEM, hydroxybutyrate-co-3- adsorption or cell density chondrocyte hydroxyhexanoate) blends fibroblast various materials 0 adv - 60°-110° no clear relationship between 0 and cell number [323] fibronectin Surface-modified dADV = 60°-81° Fibronectin displacement of preadsorbed [324] poly(hydroxybutyrate) films fibronectin was not a function of 0

neutrophils phosphorylcholine-containing Sabv = 75 ° - 92° Neutrophil attachment rate constants decreased [326] polyureathanes as 0 increased; however, very high hystersis was observed (up to 92°)

FBS, fibroblast CH3> Br, CHCH2, NH2, COOH, 0 * d v = 15 ° - 95 ° No clear relationship between 0 and FBS proteins [27] PEG, and OH terminated alkyl desorbed from the surfaces, cell attachment silane SAMs (%), or cell spread areas “ For 0dyn , both 0Adv and f e were measured and were within the range as shown.

2.6.6.1 Hydrophobic Interactions

When investigating water contact angle results, in part, we look at the wettability of a solid surface. The wettability is categorized as either hydrophobic or hydrophilic. As mentioned earlier in Section 2.4.2, the origin of hydrophobic interactions is poorly understood.

In an excellent review paper, Meyer et a l [105] reported the most recent progress in understanding hydrophobic interactions. The hydrophobic effect refers to the very low solubility of non-polar solutes in an aqueous phase. For example, fluorocarbons are considered as being one of the most hydrophobic molecules. Many self-assembly processes are driven by

69 hydrophobic forces including micelle, vesicle, and bilayer formation, and protein (un)folding [84].

Molecules or surfaces can be hydrophilic, as they interact with water by having either Lewis acid and/or Lewis base character. H-bonds can also be referred to as Lewis acid (electron acceptor) and Lewis base (electron donor) interactions [330, 331]. Some molecules can behave both as a Lewis acid or base and water is an example. Surfaces bearing chemical moieties having Lewis acid and/or base characteristics should be more hydrophilic, if other surface properties are ignored.

As mentioned by Meyer et al. [105], no deep quantitative understanding of the exact origin and nature of hydrophobic forces across water exist. Non-polar groups are barely soluble in water. The strength of hydrophobic interactions is larger than that described by the Lifshitz theory of VDW forces [105, 332, 333]. It is very difficult to estimate forces between hydrophobic surfaces due to: (i) the possible presence of gas bubbles that can result in a density-depleted water layer around these surfaces, which can potentially bridge the surfaces and affect force measurement profiles [334]; (ii) the use of different experimental techniques and protocols to produce hydrophobic surfaces and/or to measure the resulting surface forces; and (iii) surface heterogeneity, surface roughness, as well as surface molecular structure and composition, which can result in different surface molecular rearrangements when hydrophobic surfaces are immersed in water.

An example involving hydrophobic forces in protein interactions is the use of hydrophobic interaction chromatography. In fact, protein separation can be carried out by hydrophobic interaction chromatography, a technique using hydrophobic interactions between non-polar regions on the surface of proteins and hydrophobic ligands immobilized on the adsorbant [335]. It is reported that adsorption increases with mobile phases having high ionic strength. Elution is done by decreasing subsequently the eluent salt concentration [335]. The type of salts is known to affect hydrophobic interactions [335]. The main properties influencing protein retention during hydrophobic interaction chromatography experiments are salt concentration and type as well as the surface density of the hydrophobic ligands [336]. However, the concepts behind hydrophobic interaction chromatography have not been applied to study proteins-biomaterial interactions.

70 2.6.6.2 Contact Angle Measurements

Wettability involves the spreading of a liquid onto a solid substrate. In a sense, wettability is related to the water structure adopted over a solid surface. Wetting or spreading can be categorized into non-reactive or reactive. Considering Zisman’s rule [337, 338], wetting surfaces are associated to high surface energy and non-wetting ones are referred to as low surface energy. The level of wetting is represented by the contact angle formed at the interface between a solid and a liquid [339].

In reactive systems, such as those involved in cell-material interactions, the process of wetting also involves mass transport of molecules at the solid/liquid interface. A biochemical reaction and/or adsorption can occur between the surface and the molecules dispersed/dissolved in the contacting medium, and the produced surface species can also influence wetting [339].

The contact angle between a liquid and a solid is the angle formed between the liquid and the solid surface. Contact angles can be measured by different methods including the static sessile drop method, the dynamic sessile drop, and the dynamic Wilhelmy, only to name a few [339]. A contact angle of 180° indicates that there is little to no wetting between the liquid and the solid surface. It is generally accepted that a liquid is said to wet a solid surface when the contact angle is < 90° [339], although no exact consensus is applied in the biomaterials literature. With a contact angle of zero, the liquid is considered to completely wet the solid. Superhydrophobic surfaces (contact angle > 150°) have been developed. These surfaces find applications for their potential anti-sticking, anti-contamination, and self-cleaning properties [338, 340, 341]. Micro- and nano-structures appear to play a crucial role in achieving superhydrophobic surfaces, again pointing out to the importance of surface structure on wettability.

Contact angles measured on solid surfaces can vary depending on how the material has been processed or cleaned. For example, the advancing contact angle of water on Teflon can vary from 98° to 130°, as summarized by Yektafard and Ponter [342]. Although sometimes subtle, the environment can influence the contact angles on solid surfaces. Furthermore, the drop size can affect the contact angles [342].

71 From an experimental side, incorrect experimental methods can be found in the literature dealing with contact angle measurements [342]. For example, solid surfaces are not always prepared and/or processed adequatly {i.e., cleaning, fabrication, characterization, etc.). Often, many papers do not report the measurements of both the advancing and receding contact angles. Surface roughness also influences the contact angle [339, 343], although the effect of surface roughness remains unclear.

2.6.6.3 Contact Angle Hysteresis

Contact angle hysteresis is almost always observed when measuring contact angles on solid surfaces. The three main sources of hysteresis are surface roughness, chemical contamination, and solutes in the liquid [344], The difference between advancing and receding angles (i.e., hysteresis) can be around tens of degrees [345]. Hysteresis of one or two degrees can be considered within the uncertainty associated with contact angle measurements [330]. However, it is possible to encounter hysteresis up to 50° [330]. With these errors, it becomes difficult (nearly impossible) to draw any conclusions, even qualitative, on thermodynamic parameters that one may try to deduce from contact angle measurements. This situation might explain, in part, why no exact correlations have been found between contact angle experiments and protein adsorption or cell response.

As outlined above, surface roughness and heterogeneity are two main factors influencing contact angle hysteresis [330, 339, 342], For example, commercial plastics can be heterogeneous and contain many additives used to facilitate their shaping and processing [190], Even plastic made of only one polymer can have surface heterogeneity because of the mosaic formed by crystalline and amorphous regions [342]. Another cause of hysteresis with polar polymers is the possible reorientation of molecules or groups at the polymer surface under the influence of the liquid exposure. Hydroxyl groups on polymer chains can be buried within the bulk of the material when exposed to a non-polar environment. But when contacting water, these molecules can move to the surface due to interaction with the polar aqueous environment. This process of molecular reorientation can be time-dependent [346]. It is thus likely that the speed at which a liquid wets a solid would also affect this process.

72 Many researchers in the biomaterials literature do not measure advancing and receding contact angles. It can be argued that it would be necessary to measure both advancing and receding angles if someone wishes to extract some information from these measurements, if any information is in fact available. However, analysis of the results listed in Table 2.5 and Table 2.6 reveals that this information does not even provide more insight to establish a relationship between cell responses and water contact angle measurements on biomaterials.

2.6.6.4 Meaning and Relevance of Contact Angle Measurements

As stated by E. Vogler, "Even the coarsest categorization of biomaterials into fully-water- wettable and not fully-water-wettable fails to bring the protein adsorption picture into focus" [347, 348]. This statement can be extended to cell-material interactions, as revealed by results listed in Table 2.5 and Table 2.6.

The above discussion reveals a clear advantage that justifies the use of surface force measurements by SFA and/or AFM combined with chemical analyses over contact angle measurements. By measuring contact angle, we see the wetting of a solid surface by water in the presence of a gaseous phase. Unfortunately, this situation is rarely encountered in cell culture experiments. As mentioned by Yaminsky and Ninham [345], only surface force experiments can reveal how interactions vary over separation distance between solid surfaces. Different forces can exist between surfaces and these include VDW and Coulomb (treated together via DLVO theory), hydrophobic, oscillatory, only to name a few. Knowing or approximating, just partly, these interactions is far more instructive for designing biomaterials surfaces aiming to modulate cell responses. The development of the SFA [121] and the AFM [125] has paved the way to the characterization of surface properties at the molecular level for hydrated systems that are not accessible by any other instruments. Unfortunately, SFA and AFM in force measurements mode do not appear sufficiently exploited by biomaterial surface scientists, as many still prefer the use of contact angle measurements to characterize and design biomaterials surfaces.

Many researchers apply the Gibbs function (or Gibbs energy) to explain contact angle or surface tension measurements. Even more audacious is attempting to disentangle or to predict the interactions between complex solid surfaces and even more complex biological fluids {e.g., blood) and living cells. The Gibbs energy behind the interaction between a solid surface and a 73 contacting medium includes the chemical composition of the solid surface, the molecular architecture of the interface, the roughness and homogeneity of the solid surface, and the composition and properties of the contacting medium.

When a solid surface is exposed to or immersed into an aqueous solution, the molecular architecture of the solid-liquid interface is dictated by how the molecules attached on the solid surface and those dissolved or dispersed in the contacting media are disposed within the interface. It is necessary to treat this molecular architecture within the interface between a solid surface bearing hydrated layers and an aqueous medium, but this level of information is limited by the instruments available to probe such a dynamic interface. The molecular architecture of these interfaces can be depicted by molecules attached to a solid surface stretching more or less into the contacting aqueous phase and by some gradients of ions, water molecules and biomolecules (e.g., amino acids, peptides, proteins, lipids, etc.) depending on the aqueous solution involved. In fact, the molecules on the solid surface can extend into the contacting medium forming a hydrated network that can be penetrated by some dissolved (or dispersed) molecules contained in the contacting medium, depending on sizes of both the molecules attached on the surface and those in solution (or in dispersion). It is also important to consider that molecules in solution or in dispersion can form complex concentration gradients within the interface between the solid material and the bulk contacting medium. For example, we agree with E. Vogler who suggested that one should appreciate the difference between proteins in direct contact with a surface and proteins associated or entrapped in a bound hydration layer [348]. In fact, we believe that both should be assessed to correlate cell-material interactions.

One cannot assume continuity of the macroscopic physical properties at interfaces. Gradients of density and other thermodynamic properties in the normal direction can be significant [345]. The ion environment and solvation can affect the water structure at solid surfaces. The Hofmeister series has been suggested to classify ions in order of their ability to affect water structure. Some salts can modify the solubility of proteins and their structure stability. Again, the use of water contact angle values does not reveal the relative impact of these parameters on cell-material interactions. The interface thickness and solute concentrations within such interface appear to be elegant parameters to use as one of the criteria to design low-fouling surfaces. However, very few techniques allow such characterization in complex biological fluids

74 containing proteins, peptides, ions, and other molecules. Rather than contact angle measurements, perhaps quartz crystal microbalance (QCM) with dissipation monitoring, some spectroscopic methods, and AFM force measurements can be used to approximate, to some extent, these properties. To make the problem even more complex, often the exact composition of many biological fluids is unknown. Foetal bovine serum (FBS) is a good example where the exact composition is unknown along with some possible batch-to-batch variations. Consequently, it is very difficult, almost impossible to rigorously apply thermodynamic concepts to the design of biomaterials surfaces with the aim to modulate cell-material interactions using these often said "thermodynamic" analyses.

Perhaps water contact angle has been interpreted with over enthusiasm due to the fact that by mass human cells primarily consist of water? Thus, one may believe water contact angle measurements to be applicable contrary to our personal belief. However, as we have discussed, it is obvious that in general the interaction between water and matter remains poorly understood. An alternative explanation for the overuse of contact angle is the influence of the polywater theory. In fact, polywater was a hypothetical polymerized form of water [349]. This concept created scientific controversy and has been invalidated.

As proposed by Fowkes [350-352], the thermodynamic free energy, which can also be referred to as the exergy of the system (i.e., the maximal amount of useful work that can be extracted from a system as it is brought to equilibrium with the surrounding environment) should be separated into several components corresponding to forces dictating the possible work of adhesion. These can include VDW forces, H-bonds, acid-base forces, Coulomb forces, and steric forces, only to name a few. Knowing or approximating these forces is far more instructive than trying to speculate the optimization of the surface free energy from contact angle measurements with the aim to modulate very complex cell-material interactions. No reliable method for the direct measurements of surface free energy of solids exists for such complex systems encountered in biomaterials [353]. Nevertheless, many still continue to extract these values from contact angle measurements obtained from various liquids on solids. Perhaps, it would be interesting to investigate the application of the Buckingham n theorem [173, 174] to develop dimensionless analyses of the forces involved in cell-material interactions.

75 Thorough thermodynamic concepts and contact angle measurements can hardly be applied when analyzing cell-material interactions, particularly if someone wishes to apply Gibbs energy to the design of biomaterial surfaces. Reaction kinetics and path selections should be considered using perhaps a constrained optimization strategy to direct a particular reaction while minimizing or delaying other unwanted reactions.

2.6.7 Low-Fouling Surfaces

Low-fouling surfaces reject non-specific adsorption. As mentioned earlier, such surfaces are of particular interest in the development of materials suitable for medical implants as protein adsorption is believed to be the premier event in cell adhesion. Three particular low-fouling materials that have received much attention are phospholipids, particularly phosphatidylcholine (PC) and its derivatives, synthetic polymers such as PEG, and finally polysaccharides, such as dextran-derivatives. Other types of “low-fouling” materials for surface modification exist. For such materials, the reader is referred to a brief report by Hoffman [354].

PC is a zwitterionic phospholipid making up a major portion of the outer cellular surface [42]. Thus, PC-treated surfaces have attracted much attention to mimic cell membranes [261-265, 355- 361]. However, many of these studies only report a reduction in protein adsorption and thus should not be considered low-fouling [254]. Furthermore, studies by Malmsten, in which the comparison of low-fouling characteristics of PC and PEG were made, revealed that PEG was found to be superior in terms of low-fouling characteristics [362, 363]. Also, in two separate studies, Ruiz et al. [364] reported an 80% reduction in fibrinogen adsorption using PC while for the same protein Huang et al. [365] reported a 96-98% reduction in adsorption using PEG. While experimental conditions will vary between different studies, PEG seems to be superior in terms of protein rejection over PC.

The polymer PEG has been widely studied in biomedical applications [254], The attractiveness of PEG is manifested by its ability to conjugate covalently to proteins while decreasing immunogenic response and serum lifetime, as first reported by Abuchowski et al. [366, 367]. PEG is the most studied low-fouling polymer. While many advances have been made in understanding the physical and chemical mechanisms that result in low-fouling PEG surfaces, a clear picture remains elusive. 76 It is generally believed that PEG layers have steric repulsive forces of greater magnitude than the attractive VDW and DL forces. Such dominance in repulsive magnitude is dependent on the molecular architecture of the PEG layer. The factors governing protein attraction or repulsion are discussed in more detail later in this section. The most important parameter in preventing protein adsorption appears to be PEG surface density. As shown pictorially in Figure 2.11, high- density “brush” and “mushroom” PEG confirmations have superior low-fouling capabilities in comparison to low-density “pancake” conformations. Conformation is dependent on the radius of gyration of the polymer (Rg) and the distance between adsorption points on the surface (D). Steric repulsion of high magnitude results for the case when Rg>D due to the “brush” conformation. At Rg~D, “mushroom” conformations form. Finally, for Rg

Pancake

Mushrooms Brush

Figure 2.11: Surface immobilized polymer conformations. Figure not drawn to scale. Adapted with permission from reference [254]. Copyright 2003 Elsevier. The pancake, mushroom, and brush conformations were originally coined by de Gennes [368-372].

77 The production of PEG layers in a dense “brush” conformation is somewhat of a difficult task. This is due to steric hindrance of PEG chains as they come in contact with chains that are already adsorbed onto the surface. Producing PEG coatings under “cloud point” conditions seems an effective way of eluding steric hindrance to produce high density PEG coatings thus preventing protein adsorption [255]. PEG surfaces produced under “cloud point” conditions also possess chemically available amine groups for specific biomolecule immobilization making them ideal for biomaterial applications [257]. Such a feature is very beneficial as an ideal implant should simultaneously promote maximal cell adhesion for integration of the implant into the host while minimizing undesired host responses. For example, a PEG surface with chemically available amine groups allows the specific covalent grafting of biomolecules (e.g., Fn, RGD, etc.) for enhancing cellular adhesion (i.e., integrating the implant into the host) while minimizing adsorption favorable for inflammatory responses.

The abundance in literature of different PEG surface preparations and their interactions with various biological systems is vast. For a comprehensive review on the subject, the reader is referred to a detailed review by Vermette and Meagher [254]. Generally, one must perform surface characterization using techniques such as AFM, QCM, and surface plasmon resonance (SPR). Such a practice is not always followed. There are many different methods to graft PEG molecules on surfaces. Covalent immobilization of PEG chains is recommended as physisorbed chains have been shown to be unstable [374]. However, while silanated (a chemisorption of the polymer) PEG surfaces effectively reduced protein adsorption [375], one should be cautious as silanated surface stability is questionable as previously discussed in Section 2.6.3 of this article.

Although PEG has extraordinary abilities in protein repulsion subsequently resulting in the inability of cells to adhere to PEG, it has very limited capabilities for the dense grafting of peptides, proteins, etc. for engineering a specific reaction [376]. An example of a specific reaction could be the chemisorption of an ECM protein on the surface thereby promoting cellular adhesion and thus implant integration. Such a limitation inspired Massia et al. [259] to develop surfaces with the polysaccharide dextran immobilized on a surface. The potentialy dense peptide grafting capabilities of dextrans result from their flexibility and the abundant presence of hydroxyl groups throughout the long polysaccharide chains [258]. Such hydroxyl groups can be modified into carboxyl [24] and aldehyde [302, 303] functionalities.

78 The literature on dextran-derivative surfaces is not nearly as vast as that of PEG. It has been shown that the method of immobilization of CMD affects the low-fouling properties of the surface. CMD is a dextran upon which some of the abundant hydroxyl groups have been transformed to a chemically available carboxyl group. Carboxyl groups are more convenient for the chemisorption of biomolecules through the use of amide bonding, as shown in Figure 2.12B. Prior to CMD immobilization, surfaces premodified using acetaldehyde plasma polymerization and polyamine spacers had superior low-fouling capabilities to those treated with heptylamine plasma polymerization only [260, 377]. The effect of immobilization conditions over CMD layer conformation and grafting density has also been investigated (Figure 2.12A). The electrolyte concentration upon initial adsorption affects the conformation with a high electrolyte concentration resulting in unfavorable CMD conformation [24]. The ratio of carbodiimide catalyst concentration to CMD carboxylation affects the final density. The optimized conditions were found to be a carboxylation degree of 70% and a carbodiimide catalyst molar ratio to carboxylation degree of five (i.e., EDC + NHS/COOH = 5). The resultant conformations are shown in Figure 2.12. Employing such conditions resulted in serum and cellular repellency similar to that of PEG (see Figure 2.12C) [23, 25] while simultaneously allowing the grafting of biomolecules to study specific cell adhesion [23]. The dependence of cellular resistance to CMD conformation is shown in Figure 2.12. While dextran derivatives have some benefits over PEG, they have their limits depending on the application. For example, it has been shown that dextran-derivatives activate the complement system in hematological studies [378]. PEG-grafted surfaces have proven to be more adequate for hematological studies [379, 380], For other types of polysaccharide surface compositions with low-fouling capability the reader is referred to literature by Morra and Cassineli [381].

79 (A) (B) Conditions of favorable CMD conformation Biomolecule grafting to CMD surface

Low EDC

Low electrolyte High EDC concentration

Conditions of unfavorable CMD conformation

Low EDC

High electrolyte Hlgh EDC concentration (C) Day 3 Day 4 Day 5

High electrolyte concentration Low EDC

Low electrolyte High EDC

Figure 2.12 Dependence of CMD conformation on experimental conditions and its effect on cellular resistance. (A) Conformational dependence of surface immobilized CMD. Adapted with permission from reference [24]. Copyright 2007 American Chemical Society. (B) Biomolecule grafting to CMD surface. Figure not drawn to scale. (C) Resistance of fibroblast adhesion to spots of CMD. Adapted with permission from reference [25]. Copyright 2007 John Wiley & Sons, Inc. Experimental details for both studies can be found within the references.

It is worthwhile discussing in more detail the general physical mechanisms that govern protein repulsion. In general, for a low-fouling surface there exist two protein-surface separations at which adsorption can occur [382]. There is one very close and one further from the underlying surface, as shown in Figure 2.13A and Figure 2.13B. As shown in Figure 2.13C and Figure 2.13D, such is the result of the attractive VDW forces of the underlying surface and the repulsive steric force from the polymer/polysaccharide adlayer. We restrict the attractive interaction potential to that of the VDW for simplicity. However, other attractive forces (e.g., hydrophobic, electrostatic, etc.) can be present depending on the properties of the underlying substrate. The

80 interaction potential for the VDW interaction is given by Equation 2.2. As mentioned in Section 2.4.2, there is no accepted interaction potential for the steric force. However, it has been shown previously that perhaps adsorbed polymer brushes yield a parabolic profile (as shown in Figure 2.13C) given by,

( 2 (2.5) M r) * wo l ~ L l V LoJ where w0 ~ (a/a)2/3. Here, r is the separation distance, a is the surface density of the surface immobilized polymer, a is the length of the monomer comprising the polymer chains, and Lo is the thickness of the polymer adlayer in its free form {i.e., no compression from an adsorbate) [383]. The first separation known as primary adsorption occurs very close to the surface {i.e., < lnm). Secondary adsorption occurs at the boundary of the polymer/polysaccharide adlayer when the attractive VDW force overcomes the magnitude of the repulsive steric force. It should be pointed out that the boundary of the polymer/polysaccharide adlayer is not always a well defined position in space [84]. Also, for this discussion we employ the VDW interaction potential between two small atoms or molecules. As mentioned earlier in Section 2.4.2, the interaction potential of the VDW force is in actual fact dependent on the geometry and is different for macroscopic surfaces and macromolecules. Despite these technicalities, this model is still relevant for explaining as quantitatively and simply as possible, primary and secondary adsorption of molecules onto “brush” polymer surfaces.

As discussed by Leckband et al. [384], secondary adsorption is a direct result of the “brush” conformation outlined earlier. The ideal low-fouling surface has neither the primary and secondary adsorption profiles in its interaction potential. Approaches to eliminating primary and secondary adsorption on a surface are discussed in excellent detail by Leckband et al. [384]. Briefly, eliminating primary adsorption is generally very difficult as a result of the multiplicity of surface compositions of both proteins and the bare surface [385], Thus, a realistic approach is to alter the interaction potential such that secondary adsorption at distances > lnm is eliminated and the primary maximum is of such high magnitude that the protein never reaches the underlying surface in proximity of < 1 nm.

81 (A) Primary Adsorption (C) Individual Potentials

Steric (brush conformation)

(B) Secondary Adsorption (D) Total Potential

Primary -Secondary Adsorption Adsorption

Figure 2.13: Primary and secondary adsorption of proteins on low-fouling surfaces. Figure not drawn to scale. (A) Primary adsorption of protein. (B) Secondary adsorption of protein. (C) Interaction potentials of individual attractive (e.g., VDW) and repulsive steric forces on a low- fouling surface. (D) Superposition of individual attractive and repulsive forces yielding the total interaction potential of a low-fouling surface [382-384].

Despite the advances in understanding the picture of low-fouling interactions between surfaces and proteins, an overall interaction picture remains elusive. As discussed in a review by Vermette and Meagher [254], the surface grafting density of the adsorbed polymer is believed to be the most important parameter governing the repulsion of proteins. However, in many systems such a parameter cannot be properly characterized. The VDW interaction between proteins and PEG also requires more explicit attention. Furthermore, all PEG literature leaves absent the consideration of H-bonding in such a system while the solubility of PEG is based on H-bonding.

82 Interpretation of the term “fouling” also leads to confusion. It is advised that one draws caution when declaring their surface “non-fouling”. The definition of “non-fouling” is zero adsorption. However, it can be argued that such a characteristic is impossible to theoretically verify as a result of a finite sensitivity of the technique used. For example, optical waveguide lightmode spectroscopy (OWLS) has a sensitivity of 1-2 ng/cm2, thus an adsorption density lower than the sensitivity would go undetected leading to a false “non-fouling” categorization. Because of such controversy with the definition of “zero”, the term “low-fouling” is deemed more appropriate [254], Furthermore, a surface should also be tested with a wide range of proteins before concluding low-fouling behavior. Some surfaces may be resistant towards one protein but not to another. For example, in a study by Malmsten et al. [386], it was shown that the smaller protein insulin had more adsorption relative to fibrinogen, a larger protein, on the same surface. Thus, one must test their surface with a wide array of proteins to ensure it is truly low-fouling.

The terms low and non-fouling also require consideration of the timeframe of the experiment or application [254]. Some surfaces simply delay the adsorption of proteins rather than completely eliminate it. In one study, the surfaces were only exposed to proteins for 4 min [387], which is an inadequate length if one is considering biomedical applications. For an application with short-term exposure to a physiological environment such as the case of a contact lens, a shorter time frame is adequate. However, the same surface would likely be inadequate for a long-term application. Thus, one must clearly define a tolerable threshold of adsorption when designing and deeming if the surface is adequate for their particular application.

2.7 Conclusions

The scientific discourse of cell-surface and cell-material interactions is vast, covering fundamental aspects of physics, chemistry, and biology. There appears to be significant segregation between many researchers with varying scientific backgrounds studying to some extent, systems related to cell-surface and cell-material interactions. We believe that such segregation is the primary contributing factor in the wide variety of results reported throughout this article and what we believe to be very limited understanding of cell behavior {i.e., adhesion,

83 migration, proliferation, and differentiation) to modified surfaces and materials. Other contributing factors are limitations in the experimental and theoretical framework.

Extensive limitations exist in both quantitatively modeling and measuring surface and intermolecular forces in cell-surface science. Such limitations stem from the complexities of biological systems. Biological systems involve many large molecules, non-linear behavior, and somewhat chaotic thermodynamics as such systems are never in or even approaching equilibrium. Together with limitations in computational and force measurement technology, it is somewhat unrealistic to attempt to quantify all aspects of a cell-surface or cell-material interaction in regards to surface and intermolecular forces. Furthermore, the origins of fundamental interactions such as hydrophobic have no quantitative basis and are very poorly understood. Nevertheless, we believe such forces require more attention in the future than they are currently attracting. Perhaps a good starting point would be to quantitatively determine to what extent Coulomb and DL interactions coerce with gravity dining initial cell adhesion. Subsequently, either delaying or accelerating the LR interaction, the frontier between physical interactions such as Coulomb, VDW, double-layer, etc. and biological functions such as proliferation, apoptosis, etc.

As discussed in Section 2.5, a traditional in vitro cell culture system is not representative of the in vivo physiological reality. Differences in observations between 2D and 3D cell culture systems manifest from alterations in the mechanical environment, cell signaling pathways (i.e., chemical niches), and transport properties between the two systems. The design of a system that adequately represents the mechanical and transport properties of the human body is extremely challenging. This being the reason for 2D systems being the principle vehicle used to study cell- material interactions. Furthermore, imaging cells in a 3D environment, at the same resolution than that in 2D systems, is extremely difficult.

Efforts in tissue engineering have advanced experimental simulation of the mechanical and transport properties of the human body in vitro through the use of gels composed of the ECM elements and bioreactors capable of dynamic mass transport, both being more representative of the in vivo reality with respect to traditional 2D cell culture systems. In 3D cell culture systems, the mechanical environment is simulated using gels composed of ECM materials, namely fibrin, collagen, and Matrigel, to name a few. We strongly believe that Matrigel should be avoided as

84 its composition is complex and somewhat unknown. Furthermore, it is often difficult to reproduce observations using Matrigel. If possible, collagen should also perhaps be avoided as it is difficult to purify. The architecture of purified collagen varies based on the purification technique employed. Regardless of purification technique, collagen always contains impurities, this being the underlying reason for large variations in the results of previous literature that employs the use of collagens. Marine collagens are perhaps an alternative. Nevertheless, collagen is superior relative to Matrigel and finds use in biomaterials research. In our opinion, fibrin is the most reliable natural ECM gel material available today. While problems with degradation have been reported, fibrin can be freer of impurities and can be modified with growth factors or other biochemical niches allowing one to assess their role in 3D tissue development and function. The design of a bioreactor to simulate mass transport properties of the human body requires one to evaluate whether diffusive or convective transport is adequate for the delivery of nutrients and the removal of wastes while simultaneously minimizing shear stress onto the cells and tissues. Despite many advances in in vitro cell culture more representative of the in vivo reality, the limitations mentioned here are significant and one must factor such limitations into their future research.

While protein interactions are known to be the underlying mechanisms of cellular adhesion to materials, more detail into such mechanisms is necessary to better understand cell-surface and cell-material interactions. As discussed in Section 2.3, cellular adhesion is a complex and dynamic process. Such dynamics are dependent on the particular availability of ligands on the modified surface. The availability of particular ligands is dependent on the physical and chemical properties from modification of the underlying substrate. Very little is known about how modifications to the substrate leading to the availability of ligands on the modified surface affect the dynamics of cellular adhesion. Such intercellular dynamics govern subsequent behaviors such as morphology, proliferation, and differentiation, to name a few. Cell labeling of adhesion proteins, which is fortuitous in itself, fixes the cells, resulting in an unprecedented loss of how cells behave to modified surfaces. Furthermore, it is performed after a random time (varying from hours to days) between different studies in the literature. Techniques capable of analyzing live cells are perhaps methods that can provide far more extensive insight into cell behavior to biomaterials than those of traditional fixation and labeling.

85 The specific influence of substrate modifications on the conformation of biomolecules adsorbed to the substrate is not well known. Most literature restricts the resultant protein conformation to a substrate as either “extended” or “compact”. Little research mentions or even suggests what the resultant secondary, tertiary, or quaternary structure of the adsorbed protein is. Furthermore, literature implies there is a conformational change of the protein from its “natural” conformation upon adsorption to a surface. However, the “natural” conformation is not explicitly defined giving such statements no basis. Also, the proteins used in such studies are not ideal. Purification results in trace amounts of contaminants such as proteolytic fragments, to name one, and these contaminants effect cell adhesion and subsequent behavior, something that is not usually accounted for in the literature. In summary, attention to cell adhesion dynamics, adsorbed protein architecture, and considerations in the imperfections of purified proteins is required in future research.

Surface modifications also complicate research, as discussed in Section 2.6. Methods such as SAM and plasma polymerization for surface modification are not without disadvantages. For SAM, organosilanes pose numerous problems due to their sensitivity to moisture conditions. Organosulfur SAM pose fewer problems than organosilanes, but one still must be cautious of H- bonding between COOH and NH 2 tail groups of free and adsorbed organosulfur SAM molecules resulting in a partial multilayer. Otherwise, one might be deceived believing that exposed cells and/or biomolecules are interacting with a flat COOH or NH 2 surface, when they are in actual fact interacting with a partial multilayer composed of a combination of the tail and head groups of the SAM molecule with variable roughness. Fortunately, methods provided by Wang et al. [277] exist that removes such a partial multilayer. We believe that plasma treatment as a means of chemical substrate modification should not be employed as the resulting morphological alterations and significant delamination only weakens the merit of resultant cell response and/or biomolecule interaction being in fact due to the chemical modification of the substrate. The

presence of unwanted chemical moieties {e.g., NH when ideally only NH 2) remains the premier problem when using plasma polymerization. Unfortunately, this is difficult, if not impossible to control. We thus believe that the best substrate modification currently available for model studies is the use of organosulfur SAM on Au(l 1 l)surfaces, while if someone wishes to study a coating that can be applied to a wide range of substrate chemistries, types, and geometries,

86 plasma polymerization appears to be a good choice. However, we strongly urge adequate surface characterization prior to further modification using biomolecules or direct cell seeding.

More often than not, modified surfaces are not adequately characterized prior to the adsorption of biomolecules and/or the seeding of cells. Typically, the most common surface characterization technique used is XPS, which provides excellent information on surface composition but no information on surface architecture. Thus, problems such as unintentional surface roughness and heterogeneity can go unnoticed with XPS. The second most common surface analysis technique used is water contact angle measurements. The misconceptions and complexities of water contact angle measurements are vast. The largest being their relevance to biological systems. When measuring contact angle, one is observing how a solid interface interacts with an aqueous liquid phase in the presence of air (a third phase). This third phase is rarely encountered in biology. Surface roughness, chemical contamination, and heterogeneity all cause contact angle hysteresis, something that is often overlooked by researchers in cell-material interactions who in most cases use static water contact angle measurements. Furthermore, when a surface is wetted, H-bonding and Coulomb interactions can force molecules from the surface, creating a gradient of ions, thus influencing the contact angle. Finally, surface wettabilty, the property that is assessed via contact angle measurements, is a direct result of the hydrophobic properties of the surface, as discussed in Section 2.4 and Section 2.6, very little quantitative understanding of the origin and nature of hydrophobic interactions are known. With all these complexities, it is no wonder that no direct correlation can be made between contact angle and predicting the behavior of cells and/or biomolecule interactions on a surface as shown in Table 2.5 and Table 2.6. Nevertheless, contact angle use remains rampant today in cell-surface and cell-material research. We strongly discourage the use of contact angle as a means of surface characterization. If one wishes to employ low-fouling surfaces in their research, the degree of acceptable adhesion must be established and considerations on the finite sensitivity of adsorption measurement techniques must be taken into account.

We recommend employing AFM along with XPS, thus acquiring both surface composition and architecture for biomaterial surface characterization. To access the physical properties of the surface, techniques such as AFM force measurements or SFA should be used. QCM with dissipation monitoring is one of the more advantageous means of assessing biomolecule

87 adsorption to a surface. QCM with dissipation monitoring allows one to assess the quantity of biomolecules adsorbed to the surface while simultaneously giving information regarding the rigidity of the adsorbed biomolecules.

Cellular adhesion and behavior to modified surfaces is in fact governed by, for all intensive purposes, the superposition of surface and intermolecular interactions. We understand the difficulty, both experimentally and theoretically, in resolving the individual contributions of such interactions within the system. Nevertheless, such interactions at the molecular level described in Section 2.4 govern macroscopic cell behavior and response. A fundamental understanding of this influence would provide invaluable tools for designing and utilizing in vitro systems and materials that elicit a desired and controllable cell response, thereby advancing not only fundamental cell biology, but also biomaterials science, tissue engineering, and the numerous other related sub-disciplines.

The purpose of this article was not to hunt down errors, but to establish the general concepts behind cell-material interactions. We do not pretend to better understand or to marginalize the work that has been presented throughout this article. Rather, we wished to present the reader a wide scientific spectrum of literature relevant to the subject of cell-material interactions as we believe many of the limitations and complexities of physics, chemistry, and biology are not known sufficiently across disciplines. We hope that this article directs future research considerations and interpretation towards general comprehension of cell behavior to surfaces and materials as opposed to expanding an overabundant mass of literature.

2.8 Acknowledgment

The authors are grateful to Prof. Nathalie Faucheux and Prof. Marcel Lacroix for their thoughts and insights on this review. This work was supported by the National Sciences and Engineering Research Council of Canada (NSERC, Discovery grant) and by the Universite de Sherbrooke.

88 Chapter 3 In Vitro Morphogenesis of PANC-1 Cells into Islet- Like Aggregates Using RGD-Covered Dextran Derivative Surfaces

89 Foreword

Authors and Affiliation:

Evan A. Dubiel : Ph.D. Candidate, Universite de Sherbrooke, Departement de genie chimique et de genie biotechnologique

Carina Kuehn : Ph.D. Candidate, Universite de Sherbrooke, Departement de genie chimique et de genie biotechnologique

Rennian Wang : Associate Professor, University of Western Ontario, Department of Physiology & Pharmacology, Medicine

Patrick Vermette : Professor, Universite de Sherbrooke, Departement de genie chimique et de genie biotechnologique

Date of Acceptance : September 4th, 2011

State of Acceptance : Final Version Published

Jo u rn al: Colloids and Surfaces, B: Biointerfaces

Reference :

[Dubiel, E.A., Kuehn, C., Wang, R., and Vermette, P. (2012) In vitro Morphogenesis of PANC-1 Cells into Islet-Like Aggregates Using RGD-Covered Dextran Derivative Surfaces, Colloids Surfaces B: Biointerfaces, volume 89, p. 117-125]

Contribution:

This article contributes to the first experimental work of this thesis. The contents of this work were published in Colloids and Surfaces B: Biointerfaces. All experimental work was performed by Evan A. Dubiel with the assistance of Carina Kuehn. Initial experiments were planned and performed at the University of Western Ontario under the guidance of Rennian Wang. Data analysis was performed by Evan A. Dubiel. The article was written by Evan A. Dubiel. All work was performed under the direction and supervision of Patrick Vermette.

90 Titre en frangais :

La morphogenese in vitro des cellules PANC-1 vers des agregats semblables a des ilots par le biais des surfaces de dextrane derivees recouvertes de RGD

R6sum£ :

Des substrats anti-adherents sur lesquels du carboxymethyl dextrane (CMD) a ete depose et ensuite sur lesquelles le RGD a ete greffe via un lien covalent ont ete valides dans l’optique d'etudier la differentiation des cellules de type PANC-1 dans un milieu de culture sans serum. Lors de la periode d’exposition des cellules a une surface de type RGD-CMD, il a ete observe que ces cellules adherent a la surface et forment des agregats semblables a des ilots (ILA - Islet- Like Aggregates) tout en maintenant un contact avec la surface en question. Les cellules PANC- 1 n’adherent pas a des surfaces de type RGE-CMD, CMD et de polystyrene de calibre culture cellulaire. Lors de ces experiences, il a ete observe que les agregats qui se maintiennent en suspension forment des ILAs. Lors de 1’utilisation de la surface de type RGD-CMD, la formation des ILAs de taille reduite par rapport aux experiences mentionnees precedemment a ete observee. L’expression de Ki67 et CK-19 a diminue suite a des periodes d’incubation de 3, 7 et 14 jours. L’expression d’E-cadherin n’a pas change durant les periodes de 3 et 7 jours, mais a diminue pour la periode de 14 jours. Pour chaque marqueur biologique, il n’y a pas eu de difference marquee concemant l’expression entre les surfaces. Apres 14 jours de culture cellulaire, les ILAs ont faiblement exprime l’insuline ainsi que le glucagon. De plus, le niveau de production de I’insuline n’etait aucunement fonction des surfaces employees. Aucune quantite remarquable d’insuline ou de glucagon n’a ete detectee suite a l'immunomarquage aux jours 3 et 7 d’incubation. Au jour 14, un niveau superieur total d’insuline secretee ainsi que le niveau d’insuline par cellule ont ete mesures pour les cellules ayant prolifere sur des surfaces de type RGD-CMD. Au quatorzieme jour de culture, il y avait trois fois plus de cellules sur les surfaces de type RGD-CMD par rapport aux autres conditions de culture. De plus, il a ete observe que 1’expression de l’integrine as etait en concentration abondante a travers les ILAs pour toutes les surfaces. Par contre, l’expression de l’integrine av03 a seulement ete remarquee lorsque les surfaces de type RGD-CMD etaient utilisees.

91 3.1 Abstract

Substrates bearing low-fouling carboxymethyl dextran (CMD) upon which RGD is covalently grafted were validated to study PANC-1 cell differentiation in serum free medium. When exposed to RGD-CMD, cells lightly adhered to the surface and formed islet-like aggregates (ILAs) in contact with the surface. PANC-1 were non-adherent on RGE-CMD, CMD, and tissue culture polystyrene surfaces and aggregated in suspension forming ILAs. RGD-CMD resulted in smaller ILAs. Ki67 and CK-19 expression decreased after 3, 7, and 14 days. E-cadherin expression did not change from 3 to 7 days, but decreased after 14 days. For each marker, there was no difference in the expression between the surfaces. After 14 days, ILAs weakly expressed insulin and glucagon and the expression level was independent of the surfaces. No convincing level of insulin or glucagon was detected by immunostaining at 3 and 7 days. A higher level of total and per cell insulin release was detected in the media after 14 days for cells grown on RGD- CMD. After 14 days, there were three-fold more cells on RGD-CMD surfaces compared to cell numbers found in other culture conditions. a5 integrin expression was rampant throughout ILAs for all four surfaces. avP 3 integrin expression was only noticeable for ILAs cultured on RGD- CMD surfaces.

92 3.2 Introduction

Surgical transplantation of islets of Langerhans is currently a treatment for Type I Diabetes Mellitus [2]. However, many challenges such as a limited islet supply from cadaver donors and a marked loss in islet function and viability post-transplantation have yet to be overcome [388]. To that end, there is considerable interest to generate insulin-producing tissues from alternative sources.

To date, biomaterials and tissue engineering research applied to the surgical transplant of islets of Langerhans has focused on many areas of interest. One is the use of biocompatible semi- permeable membranes to encapsulate islets for protection from immune responses while simultaneously maintaining efficient insulin release [389-391]. Another is the use of natural and synthetic scaffolds to improve islet viability [392, 393]. Other investigations include the study of cell differentiation, as exemplified by two separate articles in which one reported human embryonic stems cells seeded in scaffolds and the other foetal pig islet-like cell clusters encapsulated in barium alginate that were both implanted in mice [394, 395]. Some tissue engineering studies have investigated pancreatic precursor cell differentiation into endocrine-like lineage using natural and synthetic scaffolds [396-399]. In this study, biomaterial surfaces with well defined molecular surface properties were used as a strategy to acquire insulin-producing islets from precursor cells in vitro.

One particular type of such precursor cells are believed to be adult pancreatic ductal cells [400- 404]. Isolated human pancreatic ductal tissue has been shown to differentiate into endocrine- hormone positive tissue in vitro when cultured in serum-free media (SFM) and overlayed with Matrigel [405]. In a separate in vitro study, pancreatic ductal tissue from adult mice was isolated and cultured on rat tail collagen in SFM after which endocrine-hormone expressing islet-like clusters resulted [406]. These studies indicate that pancreatic ductal cell differentiation can be induced in vitro and that extracellular matrix (ECM) constituents play a role in such differentiation.

Pancreatic carcinoma (PANC-1) cells have ductal characteristics such as the expression of cytokeratin-19 (CK-19) [407, 408]. Furthermore, PANC-1 cells have been shown to have the

93 ability to differentiate into endocrine-hormone producing islet-like aggregates (ILAs) when cultured in SFM [409]. Such differentiation appears to undergo an epithelial-to-mesenchymal- to-epithelial transition (EMET) [410]. Other studies have utilized PANC-1 cells to study islet development [411-414]. PANC-1 cells can thus be a useful cell line to study pancreatic endocrine differentiation.

The ECM in its interstitial and basement membrane (BM) form can influence both islet function and formation [415]. The epithelial-mesenchymal interface is established by the BM. The BM is composed of a wide variety of ECM cues and other biologically active molecules including laminins, collagens, fibronectin, as well as heparin and chondroitin sulphate [416], Cells and tissues communicate with their ECM via integrins composed of heterodimeric a|3 subunits [59]. Such interactions have been shown to induce morphogenesis and cytodifferentiation of the undifferentiated pancreatic epithelial progenitors to islets [417, 418], One particular constituent of the ECM is the RGD epitope. It has been shown that the RGD peptide is critical in islet development in vivo. When fragments of human foetal pancreas transplanted into mice were blocked with RGD analogues, a marked disruption in islet architecture and a significant reduction in the number of insulin-positive cells were observed [419].

In the present study, we utilize biomaterial surfaces consisting of substrates grafted with low- fouling carboxymethyl dextran (CMD) upon which RGD is covalently grafted. The rational behind such biomaterial surfaces is that they have particularly well defined properties in comparison with previous studies to investigate the effect of RGD-bearing surfaces on the development and function of PANC-1 cells. For example, Matrigel contains numerous chemical moieties, some of which are not identified, while collagen extraction from rat tail can yield a variety of collagen subtypes [420, 421], Here, the RGD peptide is covalently grafted to a low- fouling CMD surface making the system free of impurities and other ECM constituents allowing one to probe the direct influence of RGD on pancreatic development. Furthermore, the low- fouling nature of CMD surfaces allows one to limit non-specific physical interactions between the undifferentiated cells and the surface, thus only interactions between the RGD ligand and ILA/cell receptors are present. Therefore, the components of such a system are better defined and can shed some further insight into the specific role of RGD ligands as strategies to support cell differentiation into insulin-secreting tissue for surgical transplantation.

94 3.3 Materials and Methods

3.3.1 Surface Preparation

The synthesis of dextran into 25% carboxylated dextran (CMD) has been described in detail elsewhere [23]. Briefly, 5g of 70kDa MW dextran from Leuconostoc mesenteroides (D4751, Sigma-Aldrich, Oakville, ON, Canada) was dissolved in 25mL of 1M NaOH solution. Bromoacetic acid was added to yield a final concentration of 0.5M. The solution was allowed to stir overnight and then dialysed against Milli-Q water, 0.1M HC1, and finally Milli-Q water for 24h each. The dialysis membranes (132 115, Spectrum Labs, Rancho Dominguez, CA, USA) had a MWCO of 8,000. Finally, solutions were lyophilized. The degree of carboxylation was measured using 'H-NMR and was found to be approximately 25%.

The preparation scheme of the biomaterial surfaces is shown in Figure 3.1. For RGD-CMD surfaces, 12mm2 99% pure borosilicate glass substrates (Cover Glass No. 990, Assistent, Sondheim, Germany) were soaked overnight in 35% nitric acid, thoroughly rinsed in Milli-Q water, and dried with 0.22pm filtered air. The substrates were fimctionalized with amines using plasma polymerization of H-heptylamine (126802, Sigma-Aldrich) in a custom-built reactor as described elsewhere [292]. Briefly, rc-heptylamine monomer pressure was stabilized at 0.040torr and glow discharged for 45s at an ignition power of 80W and an excitation frequency of 50kHz. The distance between the electrodes was 10cm. For RGE-CMD and CMD surfaces, 12-well tissue culture polystyrene (TCPS) plates were treated for 70s under the same plasma conditions.

95 n-heptylamine (HApp) Plasma Polymerization

(B) nh 2nh 2 nh 2 nh 2 nh 2 n h 2 MAM

COOH

(c)

Figure 3.1: Biomaterial surface preparation scheme. (A) Clean borosilicate glass surfaces and TCPS 12-well plates amine functionalized via n-heptylamine plasma polymerization (HApp). (B) Amine-functionalized surfaces are grafted with CMD via carbodiimide chemistry. (C) GRGDS or GRGES were grafted to available COOH groups of the CMD surface via carbodiimide chemistry.

CMD was then immediately grafted on the amine-functionalized substrates as described elsewhere [23]. Briefly, CMD was dissolved in Milli-Q water to yield a 2mg mL‘1 solution. The solution was then filtered through a 0.45 pm filter. The CMD solution was activated by the addition of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC, E l769, Sigma-Aldrich) and N-hydroxysuccinimide (NHS, 130672, Sigma-Aldrich) to a concentration of 17mM for lOmin. Fresh n-heptylamine plasma polymer (HApp)-covered surfaces were immersed in the activated solution and the reaction was allowed to proceed for 2h at ambient temperature under agitation. CMD-grafted surfaces were then rinsed in 1M NaCl solution (2 x 12h) followed by rinsing in Milli-Q water (2 x 12h). After rinsing, surfaces were autoclaved and stored at 4°C until use.

96 Such surfaces have been shown to portray superb low-fouling behaviour towards cells and proteins [25, 377].

Peptide grafting was performed as described previously [23]. Briefly, GRGDS and GRGES (44- 0-23 and 44-0-51, respectively, American Peptide, Sunnyvale, CA, USA) solutions were prepared to a final concentration of 50p.gmL~’ in IX phosphate buffered saline (PBS, BP6651, Fisher Scientific, Ottawa, ON, Canada). CMD surfaces were activated in a 200mM aqueous solution of EDC+NHS, rapidly rinsed in Milli-Q water and then covered in peptide solution under agitation. The reaction was allowed to proceed for 2h at ambient temperature followed by rinsing twice in PBS. All peptide grafting was performed under sterile conditions.

3.3.2 X-ray Photoelectron Spectroscopy

Surfaces were analysed using an AXIS ULTRA DLD spectrometer (Kratos Analytical Ltd., Manchester, UK). The spectrometer is equipped with a 225W Al Ka source. Elemental composition was analysed with a survey spectrum at a pass energy of 160eV. Atomic concentrations were calculated using CasaXPS (Casa Software Ltd.) by determining the relevant integral peak areas, and applying the sensitivity factors supplied by the instrument manufacturer using a Shirley background. High resolution C Is spectra were acquired at 20eV. For comparison of C Is high resolution data, spectra peaks were normalized to unity with respect to their CHX maxima intensity. Leading edges of high resolution spectra were also shifted to ensure the CHX (285.0eV) maxima were all coincident. Three separate areas of each surface were analysed.

3.3.3 Cell Culture

Pancreatic ductal carcinoma (PANC-1, CRL-1469, ATCC, Manassas, VA, USA) cells were expanded in a serum containing medium (SCM) composed of Dulbecco’s Modified Eagle Medium (DMEM, 12100-46, Invitrogen, Burlington, ON, Canada) with 10% foetal bovine serum (FBS, 12483, Invitrogen). Culture media were changed on alternate days. Cells were cultured in

a 5% CO2 environment at 37°C.

97 For differentiation, cells were trypsinized upon confluence and re-suspended in serum-free medium (SFM) composed of an equal v/v mixture of low glucose DMEM (11885, Invitrogen) and F12 hamster nutrient supplement (11765, Invitrogen), 1% w/v bovine serum albumin (BSA, A7906, Sigma-Aldrich), lOng mL'1 insulin-like growth factor I (100-11, Peprotech, Dollard des Ormeaux, QC, Canada), S.Smg m L1 transferrin (T8158, Sigma-Aldrich), and 6.7ngmL'’ sodium selenite (S5261, Sigma-Aldrich). Cells were suspended at a density of 80,000 cellsmL'1. After trypsinization, cells were not exposed to FBS or other trypsin inhibitors. For non-adherent cultures (i.e., RGE-CMD, CMD, and TCPS), cells were re-suspended on 12-well plates (modified plates for RGE-CMD and CMD). For adherent cultures RGD-CMD cultures, modified 12mm2 surfaces were placed inside 12-well plates and cells were re-suspended inside the well. Media change of non-adherent cultures involved transferring the entire volume from the 12-well plate into a 1.5mL falcon tube. ILAs were allowed to settle to the bottom of the falcon tube for 5min. The supernatant was then removed and ILAs were re-suspended in lmL of fresh SFM and transferred back to the same well. For adherent RGD-CMD cultures, media change involved transferring the surface with sterile tweezers to a new well already containing lmL of fresh SFM.

Cells were cultured in a 5% CO 2 environment at 37°C. SFM was changed after 24h, then on alternate days.

3.3.4 Immunofluorescence

Non-adherent ILA cultures (i.e., RGE-CMD, CMD, and TCPS) were collected into centrifuge tubes and aggregates were allowed to sediment for 3min, Adherent ILA cultures (i.e., RGD- CMD) were loosely attached to the surface. Surfaces were rinsed thrice in PBS and solution was collected. In both cases, single and dead cells remained in suspension and were removed.

Cultured ILAs (RGD-CMD, RGE-CMD, CMD and TCPS) at day 3, 7 and 14 were harvested, fixed in 4% paraformaldehyde and processed for paraffin embedding and sectioning using standard protocols [422]. Paraffin sections were cut at a thickness of 4pm. Primary antibodies to insulin (ab7842), glucagon (abl8461), CK-19 (abl5463), Ki67 (abl6667), E-cadherin (ab53033), avPa (ab7166) were purchased from Abeam (Cambridge, MA, USA). The primary antibody for the as integrin (sc-10729) was purchased from Santa Cruz (Santa Cruz, CA, USA). Secondary antibodies were purchased from Invitrogen. All sections were blocked in 10% goat

98 serum and counterstained with 4’,6-diamidino-2-phenylindole, dihydrochloride (DAPI, D1306, Invitrogen). Manufacturer recommendations were used for pre-treatment, dilutions and incubation time, and temperature. Samples were imaged using an epifluorescence microscope (Nikon Eclipse).

3.3.5 Morphometric Analysis

The percentages of cells positive from individual histological sections were manually counted from fluorescent images taken at 400X magnification using Simple PCI software (Hamamatsu Corporation). Time zero is defined as the instant before the cells are exposed to SFM and the surfaces (i.e., RGD-CMD, RGE-CMD, CMD, and TCPS). Time zero for Ki67 and CK-19 consisted of a 95% confluent PANC-1 monolayer cultured in SCM which was trypsinized and processed for paraffin embedding and sectioning, as described in Section 3.3.4. Data are presented as an average ± standard error of the mean (S.E.M) of 3 individual cell passages (n = 3) for each time point and surface condition. All averages contain between 200 and 2000 cells.

3.3.6 Phase Contrast Microscopy

Images were recorded at 100X magnification on live cell and tissue cultures using a Nikon Eclipse TE2000-S microscope. All images were recorded directly after a medium change. ILA shape was assumed to be either that of a regular or prolate spheroid. In the case of a prolate spheroid, the polar and equatorial diameters were normalized to the equivalent of a regular spheroid using the following relation

where a is the diameter of a regular sphere, b the polar diameter of a prolate spheroid, and c the equatorial diameter of a prolate spheroid. Diameter measurements were measured using ImageJ software (National Institute of Health). Data are presented as an average ± S.E.M of n > 3.

99 3.3.7 Insulin Secretion

Insulin secretion in the medium was measured using a human Ultrasensitive Insulin EIA kit (80- INSHUU-E01, Alpco Diagnostics, Salem, NH, USA). SFM was changed 24h prior to time points of interest (3, 7, and 14 days) and then collected for analysis. SFM contains 7.8mM of D- glucose. For RGD-CMD surfaces on glass substrates (1.2cm2), insulin content was normalized to 12-well plate surface area (i.e., 3.8cm2), which were used for RGE-CMD, CMD, and TCPS culture conditions. Absorbance was measured using a Synergy™ HT Multi-Detection Microplate Reader (Bio-Tek, Winooski, VT, USA). Data are presented as an average ± S.E.M of n = 3.

3.3.8 Proliferation Assay

Cell number was quantified using a CyQUANT® assay (C35006, Invitrogen). Prior to carrying out the assay, ILAs were collected into centrifuge tubes as described in Section 3.3.4. Here, tubes were centrifuged for 5min at 1,000 rpm. Supernatant was removed and sediment was disaggregated via re-suspension in trypsin and placed under agitation for 3min. The suspension was centrifuged again then re-suspended in hanks balanced salt solution provided with the assay. Fluorescence was measured using a Synergy™ HT Multi-Detection Microplate Reader (Bio- Tek). Time zero is defined as the instant before the cells are exposed to SFM and the surfaces (i.e., RGD-CMD, RGE-CMD, CMD, and TCPS). Data are presented as an average ± S.E.M of n = 3.

3.3.9 Glycosaminoglycan Staining

Glycosaminoglycan (GAG) staining was performed using lOg-L'1 Alcian Blue (A3157, Sigma- Aldrich) in a 3% acetic acid aqueous solution. 4pm histological sections (see Section 3.3.4) were rehydrated and stained for 5min in the solution described above. Sections were counterstained in Harris Hematoxylin (HHS16, Sigma-Aldrich) for 5min.

100 3.3.10 Data and Statistical Analysis

Analysis of variance (ANOVA) was used to compute if there were any statistical differences between means. When p < 0.05, differences in the means were considered statistically significant.

For insulin release per million cells, the mean insulin release via ELISA of a surface after 14days was divided by the respective mean cell number via CyQUANT®. Since individual errors are random and independent, uncertainty was computed following the standard quotient rule of addition in quadrature. Two quantities were considered statistically different if their absolute difference was greater than the sum of their uncertainties.

3.4 Results

3.4.1 X-ray Photoelectron Spectroscopy

The elemental composition acquired by XPS is shown in Figure 3.2A. Freshly deposited HApp coatings resulted in polymerized adlayers rich in nitrogen-containing species and hydrocarbons. After incubating HApp-covered substrates in an activated CMD solution, there was a marked increase in the atomic concentration of oxygen and in the O/C ratio, typical of CMD grafting to a HApp layer via carbodiimide chemistry [24]. For both RGD and RGE, there was a small increase in the nitrogen content. There was no augmentation in the N/C ratio for RGD and RGE. Thus, one cannot verify the successful grafting of the peptides from the XPS survey.

High-resolution spectra of the bioactive surfaces are also presented in Figure 3.2B and Figure 3.2C. The maxima at 285eV correspond to C-C and CHX moieties. For each surface, the C Is spectra of CMD indicated the presence of a carbon-oxygen species at 286.5eV (C-O), 288.0eV (C=0), and 289. leV (0-C=0) relative to HApp-covered surfaces, confirming the successful grafting of CMD [24], Furthermore, the RGD and RGE spectra show a substantial increase in the signal at O-C-O relative to CMD verifying the successful grafting of each peptide to the surface.

101 (A) Atomic Concentration, % Ratios C 1* 0 1 s N 1s o/c NIC HApp 90.9 1.7 7.4 0.02 0.08 HApp + CMD 76.3 19.5 4.2 0.26 0.06

HApp * CMD + RGE 76.4 19.2 4.4 0.25 0.06 HApp + CMD + RGD 77.1 18.1 4.8 0.23 0.06 (B) (C) RGD Immobilized on CMD RGE immobilized on CMD HApp 1.6 - — HApp CMD 0.9 CM) RGOCMD RGE-CMD 0 .8 0.7 0.6 0.5 9 0.4 0.3

0.1 0.0 282 283 284 285 286 287 288 289 290 291 292 282 283 284 285 286 287 290 291 292 Binding Energy (eV) Binding Energy (eV)

Figure 3.2: XPS survey and high resolution C Is spectra. (A) Elemental composition of the different surfaces. (B-C) High resolution C Is XPS spectra of RGD-CMD and RGE-CMD surfaces and interlayers.

3.4.2 RGD-CMD Surfaces Promote Islet-Like Aggregate (ILAs) Formation

PANC-1 cells exposed to RGD-CMD in SFM lightly adhered (see Figure A .l) to the surface and formed ILAs in constant contact with the surface. In contrast, PANC-1 cells re-suspended in SFM were non-adherent on RGE-CMD, CMD, and TCPS surfaces and aggregated in suspension to form ILAs.

The total number of ILAs and their average diameters are listed in Table 3.1. Cells cultured on RGD-CMD formed smaller ILAs with a significantly increased number of ILAs within the entire well in comparison to the CMD and TCPS (Table 3.1). Conversely, RGE-CMD ILAs had an average diameter in the range (albeit still significantly different) of that of RGD-CMD. The total number of ILAs for each condition corresponded well to the diameter of the ILAs. Conditions

102 with larger ILAs coincided to a fewer number and conditions with smaller ILAs correlated to a larger overall number.

Table 3.1: The total number and mean diameter of islet-like aggregates (ILAs) obtained from PANC-1 cells cultured on the different surfaces after 7 days on average. All surfaces showed a significant difference in the total number of ILAs (p < 0.00016) except CMD and TCPS (p = 0.63). Mean ILA diameter was significantly different between all surfaces (p < 0.018). Data are presented as average ± S.E.M of n > 3.

Surface Total ILAs* Mean Diameter (pm) RGD-CMD 853 + 35 82 + 2

RGE-CMD 368 ± 6 95 + 2

CMD 92 ± 12 178 + 5

TCPS 83 + 12 209 + 23 a Total number of ILAs per well.

The expression of Ki67, CK-19, and E-cadherin in ILAs was analyzed for each surface condition, as shown in Figure 3.3 and Figure 3.4. The percentage of Ki67- and CK-19-positive cells in the ILAs decreased progressively at 3, 7, and 14 days of culture on all the experimental groups (Figure 3.4A and Figure 3.4B). However, the expression of E-cadherin showed no change from 3 to 7 days, but after 14 days there was a marked decrease (Figure 3.4C). For each individual marker, there was no significant difference in expression between surface conditions.

103 3 Days 7 Days 14 Days

(A) (B) (C)

Ki67

(D) (E) (F)

CK-19

(G) (H) (I)

E-Cad

Figure 3.3: Immunofluorescence staining for Ki67 (A-C), CK-19 (D-F), and E-cadherin (E-Cad) (G-I) of PANC-1 cells on selected culture conditions over time. Nuclei are DAPI stained in blue. Presented are selected representative images, as all culture conditions had equivalent Ki67, CK- 19, and E-cadherin immunofluorescence staining at their respective time points (see Figure 3.4).

104 (A) (B)

3 days 7 days 14 days

Figure 3.4: Expression level of Ki67 (A), CK-19 (B), and E-cadherin (E-Cad) (C) of PANC-1 cells cultured on the different surfaces over time. Nuclei are DAPI stained in blue. Ki67 and CK-19 expression was not significantly different between surfaces at all individual culture times, p > 0.12 and p > 13 respectively. However, Ki67 and CK-19 expression was significantly decreased for all surfaces at all different culture times analyzed, p < 0.00079 and p < 0.0059 respectively. E-Cad expression was not significantly different between surfaces at all individual culture times, p > 0.41. For different culture times, from 3 days to 7 days there was no significant difference in E-Cad expression for all surfaces, p > 0.75. However, from 7 days to 14 days there was a significant decrease in E-Cad expression for all surfaces, p < 0.0000017. Presented are Ki67 images from CMD, CK-19 images from RGE-CMD, and E-Cadherin from RGD-CMD after 3 days and RGE-CMD after 7 days and 14 days. Data are presented as an average ± S.E.M. of n = 3. 3.4.3 RGD-CMD Surfaces Result in Insulin-Secreting ILAs

Neither insulin nor glucagon was detected by immunostaining at 3 days and 7 days for all surfaces tested. The same applied to insulin secretion after 3 days and 7 days. After 14 days,

105 low levels of insulin were detected in the media (Figure 3.5A). The RGD-CMD culture condition showed a 2-fold increase in the insulin secretion compared to the other culture conditions. In addition, RGD-CMD showed the best insulin secretion per cell (Figure 3.5B). After 14 days of culture in SFM, both insulin and glucagon expression were weakly detected via immunostaining in the ILAs (Figure 3.5C).

(A)

RGO-CM3 RGE-CMD CMD TCPS RGD-CMD RGE-CMD CMD TCPS Surface Corvftion Surface Condition

Figure 3.5: (A) ILA insulin secretion after 14 days [no significant difference between RGE-CMD & CMD and TCPS and CMD, p > 0.16. All other comparisons showed significant differences, p < 0.0085]. Data are presented as an average ± S.E.M. of n = 3. (B) ILA insulin secretion per million cells after 14 days. Only RGE-CMD and CMD showed no significant difference. The mean insulin release via ELISA of a surface after 14 days was divided by the respective mean cell number via CyQUANT®. Uncertainty was computed following the standard quotient rule of addition in quadrature. (C) ILA insulin and glucagon expression after 14 days for CMD. Nuclei are stained in blue. All culture conditions had equivalent insulin and glucagon staining after 14 days. No insulin and glucagon expression was detected at 3 days and 7 days.

106 3.4.4 RGD-CMD Surfaces Promote Cell Adhesion and Increase Cell Proliferation The total counted cell numbers found in the different culture conditions are presented in Figure 3.6. For the adherent RGD-CMD surfaces, upon the formation of ILAs, single cells that may or may not make up the ILAs (Figure 3.6A) began to significantly proliferate after 7 days. After 14 days of culture, there were nearly three-fold more cells for the adherent RGD-CMD surfaces (Figure 3.6B) in comparison to all non-adherent surfaces. The non-adherent RGE-CMD, CMD, and TCPS all had similar proliferation rates (Figure 3.6B). For all conditions, no detectable levels of GAG expression were observed (Figure A.2).

^ 3 Days 7 Days 10 Days 14 Days

Adherent

Non-Adherent

^ T m Z i r o (B) 2BM-. ■■RGO-CMD ■ ■ R O E - O I D 226 * c m CMO ■ ■ T C P S * £ M M .

& 1 7 5 S

01 MM 1 7 M

£3 , - i d Tim 2mm S Days

Figure 3.6: (A) Phase contrast microscopy of PANC-1 aggregation to ILAs and expansion in SFM. Arrows indicate cells spread and in focus with ILAs, thus adhered to the surfaces. (B) Total cellular content measured by the CyQUANT® assay during SFM culture. RGD showed the most significant cell growth relative to all other surfaces (p < 0.0059). Data are presented as an average ± S.E.M. of n = 3.

107 3.4.5 Integrins 0 5 and avP 3 are Expressed in RGD-CMD ILAs

In this study, the RGD binding as and avP 3 integrins were assessed. As shown in Figure 3.7A and Figure A.3, as expression is rampant throughout ILAs for all the surfaces tested. Furthermore, in all surface conditions, there was no noticeable difference in as expression over time from 3 to 14 days. However, avp 3 integrin expression was only noticeable for ILAs cultured on RGD-CMD surfaces (Figure 3.7B and Figure A.4). Overall, only a minority of ILAs on RGD-CMD surfaces expressed the avP 3 integrin. Furthermore, in contrast to as, within the

ILAs that had the avP 3 integrin, few individual cells had expression.

(A) (B)

Figure 3.7: Representative ILA as integrin expression after 14 days for TCPS (A). All culture conditions had equivalent as integrin expression. No changes in expression levels were seen at 3 days, 7 days, and 14 days. ayp 3 integrin expression after 14 days for RGD-CMD (B). Arrows indicate cells with avP 3- Nuclei are DAPI stained in blue.

3.5 Discussion

When PANC-1 cells were exposed to RGD-CMD surfaces in SFM, cells adhered to the surface and aggregated to form ILAs. Conversely, when they were exposed to low-fouling RGE-CMD and bare CMD as well as TCPS, cells remained in suspension and aggregated to form ILAs. These results (Figure A.l and Figure 3.6A) indicate that cellular adhesion is due solely to ligand- receptor interactions between the RGD surface and PANC-1/ILA surface receptors. It is

108 somewhat surprising that PANC-1 cells cultured on TCPS in SFM were non-adherent as TCPS unlike RGE-CMD and bare CMD, has adhesiveness. It should be noted that when PANC-1 cells are cultured on TCPS in SCM, cells are adherent and proliferative (expansion culture as described in Section 3.3.3). Therefore, when employing SFM, RGD-CMD surfaces appear to provide some of the benefits of serum (i.e., adherence and proliferation) while simultaneously providing differentiation (modestly improved differentiation as indicative from Figure 3.5B).

Our observation of a marked decrease in CK-19 expression over time (Figure 3.3 and Figure 3.4) is in good agreement with previous studies of PANC-1 differentiation in SFM [410, 414]. Past studies on human foetal p-cell differentiation have shown that when an epithelial progenitor cell loses its ductal CK-19 expression, it gains insulin expression [423]. A marked down regulation of E-cadherin expression (Figure 3.3 and Figure 3.4) is befitting for two reasons. First, epithelial-mesenchymal transition (EMT) corresponds to cellular E-cadherin down regulation [424-426]. Second, it has been shown in mice that E-cadherin is necessary for the initial aggregation of developed P-cells [427]. Together, these results indicate that PANC-1 cells are in actual fact losing their epithelial phenotype.

As shown in Figure 3.5C, insulin-expressing cells were found largely in the ILA core while glucagon-expressing cells were found primarily at the ILA periphery [414]. Some cells co express insulin and glucagon (Figure 3.5C), which perhaps indicates an immature phenotype and an incomplete EMET. Such co-expression has been observed not only in attempts to differentiate PANC-1 cells, but also when attempting to differentiate primary human ductal cells [405, 414]. After 14 days of culture, ILAs to some extent had the capacity to secrete small amounts of insulin (Figure 3.5A). Not only did RGD-CMD surfaces show the most total insulin secretion (Figure 3.5A), they also showed the most insulin secretion per cell (Figure 3.5B).

The supplementary determinant for the superior insulin release of RGD-CMD ILAs may be their adherence to the peptide motif. It has been recently shown that isolated islets cultured in vitro in the presence of the ECM cues, namely fibronectin, laminins, and collagens, have improved functionality [392]. Islet attachment to ECM cues in vitro seems to involve the RGD motif, implying an integrin-mediated process [422]. Therefore, it is plausible that such attachment is the reason for the predominant ILA functionality on RGD-CMD surfaces. Consequently, as and

ctvP3 integrin expression on ILAs was assessed, as they regulate adhesion and migration during

109 human foetal islet development [419, 428]. In addition, such expression was found to coincide with that of fibronectin and vitronectin, as expression was prevalently found in all ILAs and its level was constant throughout the culture period (Figure 3.7A and Figure A.3). Constant levels of as expression have been observed in the development of human and rat foetal pancreata [428,

429]. In contrast, avp 3 expression was scarce and only expressed in a small number of ILAs on RGD-CMD only (Figure 3.7B and Figure A.4). Such an expression pattern implies that perhaps avP3 expression in our system is solely due to ILA attachment to the RGD-CMD surface. The as expression is likely also because of ILA attachment to RGD-CMD. However, the prevalence of as in all ILAs for all surface conditions implies that the expression is also due to the presence of ECM, such as fibronectin synthesised by the cells, in the ILA interstitial space. It should be noted that RGD binding integrins are the most promiscuous of the whole family [58]. Therefore, it is rational that numerous other RGD binding integrins are present inside the system; not to mention non-RGD binding integrins, such as those of laminin attributes, which may be synthesized by cells in the interstitial ILA space. Increased binding of such integrins could have also contributed to the down regulation of E-cadherin (Figure 3.3G - Figure 3.31 and Figure 3.4C) [430]. The resulted E-cadherin down regulation may have also led to some ILA disassociation as what appears to have progressively occurred in the non-adherent cultures (Figure 3.6A).

The noticeable decrease in proliferative PANC-1 cells comprising ILAs we observed (Figure 3.3 and Figure 3.4) is consistent with that of previous studies [412, 414]. It is well established that the differentiation process in general results in the down regulation of cell replication [431-433]. While cellular proliferation, as determined by Ki67, was down-regulated within ILAs (Figure 3.3 and Figure 3.4), overall proliferation, as measured by the CyQUANT® assay, was observed for each culture condition as shown in Figure 3.6. Similarly, a replenishment of pancreatic progenitors in the epithelium was reported [434, 435]. It should be mentioned that in addition to possible differentiation, the Ki67 down regulation observed may also be due to oxygen limitations within ILAs. RGD-CMD cultures had extraordinary levels of overall proliferation in comparison with all others. The resulting cells at 7 and 10 days had an epithelial-like morphology resembling a PANC-1 monolayer grown in SCM as shown in Figure 3.6A. After 14 days of culture, immature ILAs appeared to form from the epithelial-like monolayer. In the case of RGD-CMD, daughter cells have a foundation upon which they can attach and an ECM ligand

110 upon which they can bind via integrins and continually proliferate. This is in contrast to the non adherent RGE-CMD, CMD, and TCPS culture conditions where cells have no such foundation. This can explain the significantly higher overall cell content of RGD-CMD cultures. The appearance of an epithelial-like monolayer during the middle to late stages (i.e., 7 and 10 days) may be because migration is inhibited by saturation of cell receptors (Figure 3.6A) by the RGD- CMD surface. As proliferation ensues, perhaps daughter cell receptors are no longer saturated (Figure 3.6A at 14 days), permitting migratory properties and the subsequent aggregation to ILAs. In addition, it is likely such ILAs are also resulting from proliferation. No GAG, which are believed to play a role in mesenchyme-related regulation of endocrine differentiation [436], were detected using Alcian Blue staining (Figure A.2).

Three observations in this study suggest a possible reason for the higher insulin content on RGD- CMD surfaces. First, ILA diameter is the smallest in RGD-CMD cultures. Small islets have been shown to undoubtedly have superior function and transplantation outcomes as smaller diameter islets have better insulin secretion which is believed to be due to more efficient diffusion of products in and out of the islet center [437,438]. Like islets of Langerhans that have undergone enzymatic disruption, ILAs have no vasculature and rely solely on diffusion for the transport of glucose, oxygen, and other nutrients to the islet center and the removal of wastes and other byproducts. However, since RGE-CMD ILAs had inferior insulin release and their diameter on average was at least in the vicinity to that of RGD-CMD, other factors than ILA diameter alone are contributing to the higher insulin release. Also, the down regulation of E- cadherin and possible subsequent disassociation may have also contributed to the higher insulin release with RGD-CMD ILAs via improved diffusivity. Here the presence of the RGD-CMD surface means the disassociated cells are not contact inhibited and overgrown on the surface. The result would be less of an aggregate/spheroid form and more of a spread form meaning more favourable diffusivity and thus better insulin release. Second, ILAs on RGD-CMD inherently have a relationship with ECM cues. It has been decisively proven that isolated islets with an established islet-matrix relationship function better and longer than those without [392, 422], However, in our study, we do not have mature isolated islets. Thus, we cannot exclude the fact that ILA-RGD signalling cascades in our system may not have the exact same effect. Lastly, RGD-CMD had massive proliferation of epithelial-like cells and later what appeared to be further aggregation to ILAs. Thus, while it appears to be plausible that the higher insulin

111 secretion could be due to simply the presence of a larger quantity of ILAs, even with modestly better insulin secretion per cell (Figure 3.5B), it is difficult to decisively determine what level of functionality the additional ILAs have, if any.

3.6 Conclusion

This study demonstrates that biomaterial surfaces with well-defined molecular surface properties could be useful as a strategy to help generate a supply of insulin-producing tissues for diabetic therapy. Generally, insulin-producing tissues generated from precursor cells typically have limited function. Our results show that ILAs generated on RGD-CMD surfaces have improved, albeit still limited, functionality in comparison with RGE-CMD, CMD, and TCPS controls. Thus such surfaces could perhaps someday, as fundamental knowledge of endocrine pancreatic development continues to progress, be implemented to push the functionality of “manufactured” insulin-producing tissues closer to that of tissue acceptable for diabetic therapy. Further implementation could include employing a primary pancreatic progenitor source and/or the immobilization of other biochemical niches, to name a few.

3.7 Acknowledgment

The authors are grateful to Georges Sabra, Laurent Olivier, Normand Pothier, Irene Kelsey, and Gilles Grondin for technical assistance. Dr. R. Wang is supported by a New Investigator Award from the Canadian Institutes of Health Research and by the Canadian Institutes of Health Research operation grant.

112 Chapter 4 Solution Composition Impacts Fibronectin Immobilization on Carboxymethyl-Dextran Surfaces and INS-1 Insulin Secretion

113 Foreword

Authors and Affiliation:

Evan A. Dubiel : Ph.D. Candidate, Universite de Sherbrooke, Departement de genie chimique et de genie biotechnologique

Patrick Vermette : Professor, Universite de Sherbrooke, Departement de genie chimique et de genie biotechnologique

Date of Acceptance : September 4th, 2011

State of Acceptance : Final Version Published

Jo u rn a l: Colloids and Surfaces, B: Biointerfaces

Reference :

[Dubiel, E.A., and Vermette, P. (2012) Solution composition impacts fibronectin immobilization on carboxymethyl-dextran surfaces and INS-1 insulin secretion. Colloids Surfaces B: Biointerfaces, volume 95, p. 266-273]

Contribution :

This article contributes to the second experimental work of this thesis. The contents of this work were published in Colloids and Surfaces B: Biointerfaces. All experimental work and data analysis was performed by Evan A. Dubiel. The article was written by Evan A. Dubiel. All work was performed under the direction and supervision of Patrick Vermette.

114 Titre en frangais:

L’effet de la composition des solutions sur 1’immobilisation de la fibronectine sur des surfaces de carboxymethyl dextrane ainsi que sur la secretion d’insuline a partir de cellules INS-1

R6sum6 :

II est demontre dans cet article que la composition chimique d’une solution joue un role face aux proprietes de surface de la fibronectine (FN) ainsi que sur la bioactivite des surfaces vis-a-vis la fonction des cellules insulinoma INS-1. II a et6 observe par spectrometrie de photoelectrons induits par rayons X (XPS) que la fibronectine se greffe sur des surfaces de carboxymethyl dextrane (CMD) avec succes dans des solutions de 10pM de CaCl2, 10pM de MgCl2, lOpM de MnCl2, ainsi qu’avec des solutions de lOpM et ImM de NaCl. II a ete observe, par contre, qu’il n’y avait pas d’attachement de la proteine lorsque les solutions de 150mM NaCl ou bien de lx PBS etaient utilisees. Aucune modification de conformation de la proteine FN a ete observee avec le dichro'isme circulaire ainsi qu’avec la diffusion dynamique de la lumiere et ce peu importe la solution utilisee. II a aussi ete mesure, par microscopie a force atomique qu’il n’y a pas de difference quantifiable entre les rugosites de surface, mais qu’il existe des differences qualitatives entre les topologies des surfaces. L’immobilisation de la fibronectine n’influence aucunement la croissance des cellules INS-1 durant des temps d’incubation de 3 et 7 jours, peu importe la nature du substrat ou la composition chimique de la solution utilisee. Par contre, la secretion d’insuline des cellules INS-1, en reponse au glucose, est affectee par la nature du substrat et par la composition chimique de la solution lors de l’immobilisation de la FN.

115 4.1 Abstract

It is shown that solution composition dining immobilization plays a critical role in the properties of fibronectin (FN) surfaces and their bioactivity towards insulinoma (INS-1) cell function. X- ray photoelectron spectroscopy revealed FN grafting onto low-fouling carboxymethyl-dextran (CMD) surfaces was successful with solutions composed of lOpM CaCl2, 10pM MgCl2, lOpM MnCl2, and lOpM and ImM NaCl, but unsuccessful with those made of 150mM NaCl or lx PBS. Circular dichroism and photon correlation spectroscopy revealed that regardless of solution composition, no measurable differences in free FN conformation prevail. Atomic force microscopy (AFM) imaging of FN-CMD revealed, while there are no quantitative differences in surface roughness, there are some subtle qualitative differences in topography. FN surface immobilization scheme does not influence INS-1 cell growth after 3 and 7 days regardless of the underlying substrate or solution composition. INS-1 cell insulin secretion in response to glucose is affected by the substrate and solution composition during FN immobilization.

116 4.2 Introduction

P-cells comprising islets of Langerhans are known to lose their insulin secreting capacity (i.e., function) towards glucose stimulation following isolation from donors’ pancreas. It is well accepted that isolated islets function better upon re-establishment of islet-extracellular matrix (ECM) relations [422]. Such re-establishment is often achieved via the immobilization of ECM molecules to surfaces [392, 422, 439, 440]. The immobilization of biomolecules (i.e., proteins, growth factors, peptides, etc.) to surfaces has a wide array of applications including biosensors and medical implants [441, 442]. Undesired response to biological environments remains a major hurdle, amongst others, in the fabrication of such surfaces. Understanding the influence of physical and chemical mechanisms that govern the interaction of biomolecules with surfaces would be of great benefit to overcome such a hurdle. Those interactions determine the resultant conformation of the adsorbed biomolecule, dictating subsequent responses towards biological environments [441].

The immobilization of fibronectin (FN), a 450kDa glycoprotein and constituent of the ECM, to various modified substrata has been investigated extensively. Researchers have adsorbed FN to self-assembled monolayers (SAMs) or polymers presenting various surface chemistries [50, 443- 445]. FN has also been grafted to SAMs [88]. Others utilize plasma treatment to graft cross linkers to then immobilize FN [79]. Each of these studies outlined above assessed the adhesion of cells to FN-coated surfaces revealing the influence of biomaterial surface properties on adsorbed FN conformation. One common feature of all these studies is that the surface does not have “low-fouling” properties. Low-fouling surfaces are advantageous in that they minimize the unwanted adsorption of biological fluid-borne proteins and/or cells to the surfaces. The mechanisms of low-fouling surfaces are largely believed to be achieved primarily through steric repulsion which propels incoming proteins from the surface and a well structured water “hydration barrier” that prevents direct interaction of proteins with the surface [254,446].

With low-fouling surfaces, it is believed one can perhaps achieve cell/protein-surface interaction selectivity. Interaction selectivity is paramount in the design of biomaterial surfaces for medical implant and biosensor applications. Low-fouling surfaces can be produced through the grafting, under adequate conditions, of polymers to minimize non-specific physical interactions (i.e.,

117 electrostatic, van der Waals’, etc.). Two polymers that have particularly effective low-fouling capacity when grafted to surfaces are poly(ethylene glycol) (PEG) and carboxymethyl-dextran (CMD) [259,447].

Selectivity of such surfaces is achieved via the grafting of biomolecules to available functional groups of the low-fouling surface. Selective bioactive surfaces have been successfully fabricated in the past via the grafting of adhesive peptides. For example, biologically active RGD ligands have been successfully grafted on low-fouling surfaces bearing dextran and PEG as demonstrated by Massia et al. [258] and Hubbell et al. [448], respectively. RGD is believed to be the ligand primarily responsible for the adhesion of cells to not only FN, but to a wide variety of other ECM proteins such as vitronectin and laminin to name a few [449].

Ionic salts comprising the solution undoubtedly influence the immobilization of proteins to surfaces as they affect the electrostatic and hydration properties of the system. Attraction of mobile solution ions from oppositely charged moieties comprising the protein and surface results in overall charge neutralization between the two bodies, thereby reducing (i.e., screening) the overall electric field therefore reducing the interaction forces [84]. Such electrostatic screening is influenced by ion concentration and properties (i.e., valency, size, etc.). Hydrophilic bodies in aqueous environments inherently have a well-ordered water structure at their surface atoning to their hydration and solubility. Their solubility can become jeopardized via the addition of ions to the solution. Ions have an association with water classified as either “structure promoting” or “structure breaking” according to the Hofmeister series [348, 450]. Given these facts and the low-fouling nature of CMD, it is therefore of interest to investigate what effects solution ions have on the ability of FN to adsorb to low-fouling CMD surfaces.

In this study, we follow the past successes of Massia et al. [258] and Hubbell et al. [448] with those of the FN-surface studies previously outlined [50, 79, 88, 443-445]. We demonstrate the successful grafting of FN to low-fouling surfaces bearing CMD. It is shown that the ions of the aqueous solution used during grafting influence not only the successful immobilization of FN to low-fouling CMD surfaces, but also in the subsequent bioactivity of FN-covered surfaces, as observed with INS-1 cell insulin secretion. INS-1 cells represent a cell line in which insulin secretion is glucose responsive, somewhat parallel to P-cells comprising islets of Langerhans.

118 4.3 Materials and Methods

4.3.1 Surface Preparation

99% pure borosilicate glass substrates (Cover Glass No. 990, Assistent, Sondheim, Germany) were soaked overnight in 35% nitric acid, rinsed in Milli-Q water, and dried with 0.22pm filtered air. The surfaces were then grafted with active amine groups using radio frequency glow discharge plasma polymerization. The monomer used was q-heptylamine (126802, Sigma- Aldrich, Oakville, ON, Canada) in a custom built reactor chamber described elsewhere [292]. Briefly, monomer pressure was stabilized at 0.040Torr and glow discharged for 45s with 80W of ignition power at an excitation frequency of 50kHz. Electrode separation was 10cm.

The aminated surfaces were then immediately immersed in am activated 2mg/mL carboxymethyl- dextran (CMD) aqueous solution. The synthesis of CMD is described elsewhere [23]. The resultant CMD was measured to have a carboxylation degree of 25% via 'H-NMR. After CMD dissolution, the solution was filtered through a 0.45pm filter. Activation of the solution was achieved via the addition of l-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC, Sigma- Aldrich, E l769) and N-hydroxysuccinimide (NHS, Sigma-Aldrich, 130672) to a concentration of 17mM. The EDC and NHS was dissolved in the CMD solution and allowed to activate CMD for lOmin. The aminated surfaces were then immersed in the activated CMD solution for 2h under agitation followed by rinsing twice in 1M NaCl for 12h each and twice in Milli-Q water for 12h each. Surfaces were then autoclaved and stored at 4°C until use. Such surfaces have been proven to have remarkable low-fouling capacity towards cells and a wide variety of biological fluid-borne proteins [25, 377].

Human plasma fibronectin (FN, FC010, Millipore, Temecula, CA, USA) arrived as a lmg/mL solution in 150mM NaCl and lOmM sodium phosphate. FN was dissolved to a final concentration of lOOpg/mL in lOpM CaCl2 (Sigma-Aldrich, C7902), lOpM MgCl2 (AC413415000, Fisher Scientific, Ottawa, ON, Canada), lOpM MnCl2 (Sigma-Aldrich, 244589), lOpM, ImM, and 150mM NaCl (Sigma-Aldrich, S9888), or lx PBS. lx PBS is composed of 137mM NaCl, 2.7mM KC1, lOmM Na2HP04-2H20, and 1.76mM KH2P04. All FN solutions had a final pH of 7.4. CMD surfaces were activated with a 200mM EDC+NHS aqueous solution

119 for lOmin and then rinsed one time in Milli-Q water. FN solutions were then added to the surfaces at ambient temperature for 2h under agitation and rinsed thrice in PBS.

4.3.2 X-ray Photoelectron Spectroscopy

Samples were analysed via X-ray photoelectron spectroscopy (XPS) using an AXIS ULTRA DLD spectrometer (Kratos Analytical Ltd.) equipped with a 225W Al Ka source. Elemental compositions were analysed with a survey spectra at a pass energy of 120eV. Atomic concentrations of each element were calculated using CasaXPS (Casa Software Ltd.) by determining the relevant integral peak areas, and applying the sensitivity factors supplied by the instrument manufacturer using a Shirley background. High resolution C Is spectra were acquired at 20eV, normalized to unity with respect to their CHX maximal intensity, and leading edges were shifted to ensure the CHX (285.OeV) maxima were all coincident. Three separate areas of each sample were analysed.

4.3.3 Atomic Force Microscopy

Atomic force microscopy (AFM) was performed at ambient temperature using a Digital Nanoscope Ilia Bioscope (Bruker, Santa Barbara, CA, USA) positioned on an inverted microscope (Ziess Axiovert 200, Thomwood, NY, USA). Both AFM imaging and force measurements were performed at ambient temperature in liquid using silicon nitride cantilevers (Bruker, DNP-S). Cantilevers were cleaned for 30min in Liquinox (1201, Alconox Inc, White Plains, NY, USA), 30min in Milli-Q water, and 30min in 99% ethanol followed by 60min of UV exposure (Novascan Technologies, PSD-UV, Ames, IA, USA).

AFM force measurements were performed using the colloidal force method developed by Ducker et al. [451]. Silica colloidal spheres (SS05N, Bangs Laboratories Inc., Fishers, IN, USA) with a mean diameter of 4120nm were glued to cantilevers using a UV adhesive (ELC-4481, Electro-Lite Corporation, Bethel, CT, USA). The spring constant of the modified cantilever was 0.14N/m as measured using the resonance method proposed by Cleveland et al. [452]. Surfaces were allowed to equilibrate in the appropriate solution for at least lh prior to measurements. Deflection vs distance curves were acquired at a z-scan velocity of 150nm/s. At least 10 curves

120 were acquired at a single spot for four separate spots of the surface. Deflection vs distance curves were transposed to force vs separation distance using custom software. The linear regions of the deflection vs distance curves were used to define zero force and zero separation. Work was calculated by determining the area under the curve via application of the trapezoidal rule using Origin (OriginLab Corporation, Northampton, MA, USA). Data are represented as average ± S.E.M.

AFM imaging was performed in liquid using Tapping™ mode. Operation parameters were: drive amplitude of 0.100-0.300V, amplitude setpoint of 0.1000-0.2500V, drive frequency of 7- 10kHz, scan rate of l-2Hz, integral and proportional gain of 0.2 and 0.3, respectively, with a resolution of 512x512 and a scan size of 750x750nm. The RMS amplitude was fixed at 0.3V. The cantilever was initially pre-engaged and the drive frequency, scan rate, amplitude set point, and RMS voltage parameters were adjusted. Afterwards, the same parameters were adjusted at a distance of 25 pm from the surface. Imaging then commenced. All images were treated with a third degree flattening and low pass filter followed by measuring the RMS roughness using version V6.13R1 of the AFM software. For each condition, at least three separate images were used and roughness measurements are represented as an average ± S.E.M.

4.3.4 Circular Dichroism Spectroscopy

Circular dichroism (CD) spectroscopy in the far-UV region was performed using a Jasco J-810 spectropolarimeter coupled with a Jasco Peltier-type thermostat (Jasco Analytical Instruments, Easton, MD, USA). Spectra were acquired at ambient temperature. Samples were loaded in quartz cuvettes with a path length of 0.1cm. For each sample, three separate spectra were recorded between 200-250nm at O.lnm intervals and then averaged into a single spectrum. Spectra with samples composed of the background solutions (i.e., without FN) were acquired in the same manner and subtracted from that of the samples of interest. The measured ellipticity, 9 (in degrees), was converted to molar ellipticity, [9], using the following relation,

= 100 — c l (4.1)

where c is the molar concentration and / is the path length (in cm).

121 For an adequate CD signal, solutions had to be prepared with a concentration five-fold higher than that used during FN immobilization on CMD and glass surfaces. FN stock solution arrives at a concentration of lmg/mL in 150mM NaCl and lOmM sodium phosphate. For CD, FN stock solution was concentrated to 5mg/mL using Amicon® Ultra-0.5 centrifugal filters (Millipore, UFC510008) with a MWCO of lOOkDa. This concentrated solution was then diluted in the appropriate solutions to yield 5x that of the immobilization solutions.

4.3.5 Photon Correlation Spectroscopy

To measure the mean hydrodynamic diameter DH, photon correlation spectroscopy (PCS) was carried out using BI-200 goniometer, a BI-APD highly sensitive avalanche photodiode detector, a TurboCorr digital correlator, and a He-Ne laser (X=632.8nm) all from Brookhaven Instruments

Corp. (Holtsville, NY, USA). All measurements were carried out at ambient temperature. Dh was determined using cumulant analysis. All FN solutions were filtered with 0.22pm filters prior to measurements. For each solution, three measurements were carried out. Data are represented as average ± S.E.M.

4.3.6 Cell Culture

INS-1 cells were cultured in RPMI-1640 (Invitrogen, 31800-022, Burlington, ON, Canada) supplemented with 10% v/v fetal bovine serum (FBS, Invitrogen, 12483), lOmM HEPES (Sigma-Aldrich, H3375), ImM sodium pyruvate (Sigma-Aldrich, S8636), and 50pM P- mercaptoethanol (Sigma-Aldrich, M7522) [453]. Modified surfaces were placed inside 12-well plates where 100 x 103 cells per well were seeded. After 48h, surfaces were transferred to new well plates and then supplemented with fresh media. Media were changed on alternate days.

Cells were cultured in a 5% CO 2 and 37°C environment.

4.3.7 Cell Number

Cell number was quantified using a CyQUANT® assay (Invitrogen, C35006). Fluorescence was measured using a Synergy™ HT Multi-Detection Microplate Reader (Bio-Tek, Winooski, VT,

122 USA). For each condition and time point, three separate INS-1 passages (n=3) were measured. Data are represented as average ± S.E.M.

4.3.8 Glucose-Stimulated Insulin Secretion

INS-1 cells were cultured on various surfaces for 7 days. For glucose-stimulated insulin secretion (GSIS), samples were incubated in Krebs buffer solution composed of 25mM HEPES, 115mM NaCl, 24mM NaHC03 (EMD Chemicals, SX0320-1, Gibbstown, NJ, USA), 5mM KC1 (Fisher Scientific, BP366), ImM MgCU, 5mM CaCk, 0.1% bovine serum albumin (BSA, Sigma-Aldrich, A7906) supplemented either with 2.8mM D-glucose (Sigma-Aldrich, G6152), 28mM D-glucose, or 28mM D-glucose and 50pM 3-isobutyl-1-methylxanthine (IBMX, Sigma- Aldrich, 15879). Each incubation was done for lh at 37°C. Insulin content was measured using a High Range (Rat) Insulin EIA kit (Alpco Diagnostics, 80-INSRTH-E01, Salem, NH, USA) and normalized to cell number estimated via CyQUANT®. Absorbance and fluorescence were measured using a Synergy™ HT Multi-Detection Microplate Reader (Bio-Tek, Winooski, VT, USA). For each condition, three separate INS-1 passages (n=3) were measured. The stimulation index was calculated by dividing INS-1 insulin content at the high glucose concentration (i.e., 28mM) with the insulin content to initial low glucose concentration (i.e., 2.8mM). Data are represented as average ± S.E.M.

4.3.9 Statistical Analysis

Analysis of variance (ANOVA) was used to compute if there was any statistical difference between means. When p < 0.05, differences in the means were considered statistically significant.

123 4.4 Results and Discussion

4.4.1 Fibronectin Grafting onto CMD Surfaces Occurs Only with Solutions of Low Salt Concentration

XPS C Is spectra and surveys of the intermediate n-heptylamine plasma polymer (HApp) and CMD layers (Figure 4.1) were in excellent agreement with those previously reported [454]. The presence of FN was evaluated with and without EDC/NHS activation. With activation (i.e., act.), FN grafting to CMD was successful only with solutions of low salt concentration. In the presence of lOpM CaCh, lOpM MgCh, and lOpM MnCh, XPS survey and C Is spectra (Figure 4.1 A and Figure 4.IB, respectively) clearly showed an increase in the N/C ratios and the presence of peptide bonds (288.2eV), indicative of the addition of a nitrogen-rich layer. No such increases were present with lx PBS or 150mM NaCl indicating little to no FN immobilization (see Figure 4.1A to Figure 4.1B and Figure B.l). This is somewhat surprising as we have found in previous studies that GRGDS and GRGES peptides do in actual fact graft on CMD in a biologically active manner with lx PBS [454], This is likely attributed to the fact that FN has a much larger M.W. (~450kDa) than GRGDS (490.5Da) and GRGES (504.6Da). It has been observed in the past that smaller M.W. proteins can adsorb more readily to low-fouling surfaces than larger M.W. proteins [386]. Nevertheless, PBS contains numerous solutes making the system more complicated. Given the complexity of PBS and the unsuccessful immobilization of FN to CMD when using PBS, it was therefore decided to use simpler less complex solution compositions for the majority of this study. There was also successful FN immobilization when using lOpM NaCl and ImM NaCl. Without EDC/NHS activation (i.e., no act.), in solutions with low salt concentration, XPS survey (Figure 4.1 A) showed the presence of FN, but with a smaller N/C ratio compared to the respective conditions using catalysts. The effect seen in XPS C 1 s spectra at 288.2eV was less pronounced when comparing conditions involving low salt concentration solutions with and without catalysts (see Figure 4.1C and Figure B.1B to Figure B.1D).

124 (A) C1a O la N1a O/C N/C HApp 90.4 1.6 8.0 0.02 0.09 CMD 76.5 19.7 3.8 0.26 0.05 CMD + 1x PBS ♦ lOOua/mL FN (acL) 74.9 20.5 4.6 0.27 0.06 CMD+ 150mM NaCl+ 100iutfmLFN (act! 74.8 20.8 4.4 0.28 0.06

CMD + 10uM CaCU + 100ua/mL FN (act! 73.2 18.7 8.1 0.26 0.11 CMD ♦ 10uM CaCL + 100uo/mL FN (no a ct) 76.0 17.0 7.0 0.22 0.09 CMD * 10uM MaCL ♦ 100im/mL FN (act) 73.0 19.1 7.9 0.26 0.11 CMD + 10uM MaCL + 100ua/mL FN (no a ct) 75.6 17.7 6.7 0.23 0.09 CMD + 10uM MnCL * lOOuo/mL FN (act) 73.1 19.4 7.5 0.27 0.10 CMD * 10uM MnCL + 100ua/mL FN (no a ct) 75.7 18.3 6.0 0.24 0.08 (B) (C) HApp CMD 1x PBS + 100rtJ?mL FN (act) 10>W CaO , + lOOwfrrt. FN (no act) 150nnM N ad + 100^m L FN (act) 1(HW MgO, + KXHetaL FN (no act) 10^MCaa; + 100v^mLFN(ac(.) 1CHM MnQ2 + 100«frrL FN (no act) 10j/4 MgQ, + 100^mL FN (act) 1

284 286 288 (anting Energy (oV)' H ndng Energy (eV)

Figure 4.1: (A) Elemental surface composition via XPS survey. (B) High resolution C Is XPS spectra of FN-CMD surfaces and intermediate layers with EDC/NHS activation (act.). (C) High resolution C Is XPS spectra of FN-CMD surfaces and intermediate layers without EDC/NHS activation (no act.). Peptide bonds correspond to a binding energy of 288.2eV.

4.4.2 Solution Composition Does Not Significantly Alter Fibronectin Conformation/Size

It is plausible that the solution composition can impact the conformation/size of FN in solution and this could be the reason for its successful grafting to CMD surfaces. Thus, circular dichroism (CD) spectroscopy and photon correlation spectroscopy (PCS) were used to determine if solubilized FN in CaCh, MgCb, and MnCh solutions had an effect on FN conformation/size relative to NaCl and PBS. CD spectroscopy was used to measure any possible effects on FN secondary structure and PCS for particle sizes.

125 The CD spectrum is presented in Figure 4.2A. Overall, the spectra are in good agreement with those reported previously having negative amplitude at approximately 212nm corresponding to p-sheets and positive amplitude at approximately 227nm corresponding to aromatic residues [455]. CD spectra of FN solutions containing different salts are nearly identical and thus there were no measurable changes in the secondary structure of FN. This is not surprising given that it was previously found that a concentration of 10M urea is required before any FN unfolding can be observed via CD [456].

The mean hydrodynamic diameter, Dh was measured with PCS and the results are presented in

Figure 4.2B. Overall, there are no drastic differences in FN Dh solubilized in the different saline solutions. The DH values are in fairly good agreement with those reported by others with somewhat similar solutions [457].

(A) -50pMCaCL (B) 60000-j

40000-

100(ig/mL FN (solute) Dh (nm) 10pM CaCI2 24.2 ± 1.2 I 10pM MgCI2 21.7 ±1.8 10pM MnCI2 22.1 ± 2.2 1x PBS 24.6 ± 0.9 150mM NaCl 19.8 ±0.9 -40000-

-60000 205 210 215 220 225 230 235 240 245 250 Wavelength (nm)

Figure 4.2: (A) Circular dichroism spectra of FN in various aqueous solutions. (B) Mean hydrodynamic diameter, Dh of lOOpg/mL FN in various aqueous solutions measured via PCS. Data are presented as average ± S.E.M (n=3).

126 4.4.3 Topography of Fibronectin Surfaces Made in CaCl2 Solution Differed Subtly From Those Made in MgCl2 and MnCl2

Colloidal AFM force measurements were performed to measure how solutions affect the compression of CMD adlayers as such dynamics are believed to be largely responsible for their low-fouling capacity [446]. Measurements (Figure B.2) showed that a lOpM CaCl2 solution resulted in surfaces requiring more work (127 ± 14 aJ) for total compression (i.e., from zero- force to zero-distance) than those of 150mM NaCl (35 ± 1 aJ) and lx PBS (28 ± 7 aJ). Surfaces in Milli-Q water required the most work (366 ± 29 aJ). Together these results confirm that at such concentrations, CaCl2 solution does not alter CMD adlayer compression dynamics in a manner to affect their low-fouling capacity relative to those produced in NaCl and PBS.

Solution ions undoubtedly influence the conformation of surface-grafted CMD ultimately impacting its low-fouling capacity. CMD chains are polyelectrolytes having numerous negatively charged carboxyl moieties along the polymer chain. Electrostatic repulsion between these moieties results in extension of the polymer backbone. In the presence of solution ions, these repulsive interactions are reduced via electrostatic screening causing a partial collapse of the polyelectrolyte chains resulting in a thinner CMD adlayer with a less repulsive steric profile. This assertion has been verified in this study via colloidal AFM force measurements (see Figure B.2), as well as by other past studies [289, 458]. However, as shown in Figure B.2, lOpM CaCl2 divalent electrolytes result in a longer range repulsive force-separation distance profile than 150mM NaCl and lx PBS electrolytes. Therefore, partial collapse does not explain the successful immobilization of FN to low-fouling CMD surfaces.

AFM imaging in liquid was performed to determine what effect different solutions had on the structure of CMD surfaces to investigate if this may be the reason for the successful grafting of FN and if there was any differences in the topography of FN surfaces. As shown in Figure 4.3A to Figure 4.3C, in the presence of lOpM MgCl2 and lOpM MnCl2, the topography of the FN surfaces appears qualitatively similar each portraying large abrupt mounds. On the other hand, FN grafted in the presence of lOpM CaCl2 yields smaller broader mounds.

127 (A) FN-CMD (10|iM CaCI2) (B) FN-CMD (10|jM MgCI2)

FN-CMD 1OjjM CaCI.

10(jM MgCI.

10mM MnCI.

Milli-Q W ater Glass Milli-Q W ater

Figure 4.3: (A-C) 750nm x 750nm AFM images in liquid {i.e., lx PBS) using Tapping™ mode of FN-CMD grafted in the presence of lOpM CaCl 2, lOpM MgCU, and 10pM MnCU, respectively. All FN-CMD surfaces were imaged in lx PBS, as this solution is somewhat analogue to the conditions of cell culture in terms of pH and osmolarity. (D) RMS roughness of FN-CMD surfaces, CMD surfaces in various solutions, and HApp-modified and clean glass substrates. For the FN-CMD row, solute column indicates solution composition during FN grafting while for CMD, HApp, and glass rows, solute column indicates solution composition during AFM imaging. Data are presented as average ± S.E.M (n=3).

Surface topography was quantitatively accessed via RMS roughness measurements presented in Figure 4.3D. All FN-CMD surfaces had equivalent RMS values. The topography of CMD surfaces was largely unchanged when exposed to various solutions. CMD surfaces exposed to

10pM CaCh, lOpM MgCU, and lOpM CaCl 2 solutions, which resulted in successful FN grafting, had a similar topography to that of lx PBS. Therefore, it does not appear that roughness measurements yield an explanation as to why FN grafting occurs only with solutions

128 of low salt concentration. As expected, each stage of the surface modification process resulted in an increase in RMS surface roughness (Figure 4.3D and Figure B.3).

The underlying reason for the successful immobilization of FN on CMD is likely due to multiple mechanisms. With higher ionic concentrations like lx PBS and 150mM NaCl, any attractive electrostatic interactions are significantly weakened as a result of screening making repulsive steric forces more dominant. Note that even though screening damps steric forces because of collapsing of the CMD backbone, such forces remain relevant even at high ionic concentrations [374]. Such collapsing of the CMD backbone also likely embeds carboxyl moieties required for amide bonding via carbodiimide chemistry. Together, we hypothesize these are the reasons for the unsuccessful immobilization of FN to CMD in the presence of lx PBS and 150mM NaCl. At lower ionic concentrations, any attractive electrostatic interactions are less screened and therefore more relevant and the CMD backbone is less collapsed likely resulting in more available carboxyl moieties available for carbodiimide chemistry. Also, cross-linking between anionic moieties of the CMD surface and FN protein, mediated by the divalent cations (Ca2+, Mg2+, or Mn2+) also perhaps contributes to FN immobilization. This cross-linking hypothesis is in agreement with the XPS results of Figure 4.1 A and Figure 4.1C showing FN grafting to CMD without EDC/NHS surface activation. By analogy, recently reported experimental observations suggest divalent ions act as cross-linkers in vimentin intermediate filament networks [459, 460]. Together, we hypothesize these are the reasons for the successful immobilization of FN to CMD in the presence of lOpM NaCl, ImM NaCl, lOpM CaC^, lOpM MgCl 2, and lOpM MgCU.

It also should be mentioned that for all conditions, solution ions also cause the disjunction of water structure vicinal to the hydrophilic moieties of both the FN protein and CMD surface resulting in FN-CMD interfacial interaction. As mentioned earlier, a particular ion’s ability to disrupt water structure follows the Hofmeister series [348,450]. It is likely that the classification of a particular ion on the Hofmeister series and its physical size influence not only whether or not FN can immobilize onto CMD but also the resultant conformation and subsequent bioactivity of immobilized FN. To test this presumption, INS-1 cells were exposed to FN surfaces and their response accessed.

129 4.4.4 Solution Composition During Fibronectin Immobilization Does Not Affect INS-1 Cell Growth But Does Affect Insulin Secretion

Cell response to various FN surfaces produced using different immobilization schemes, both chemical (i.e., grafted) and physical (i.e., adsorbed), were assessed via INS-1 cell growth and insulin response to glucose concentration. Grafting consists of the immobilization of FN onto EDC/NHS-activated CMD surfaces in lOpM CaCl2 (FN-CMD[CaCl2]), 10pM MgCl2 (FN- CMD[MgCl2]), 10pM MnCl2 (FN-CMD[MnCl2]), lOpM NaCl (FN-CMD[NaCl]), and ImM NaCl (FN-CMD[lmM NaCl]) solutions. Adsorption consisted of exposing clean borosilicate glass surfaces to FN in 10pM CaCl2 (FN-Glass[CaCl2]), and lx PBS (FN-Glass[PBS]). A clean borosilicate glass surface was also used as a control (i.e., no immobilized FN).

INS-1 net cell growth was assessed by measuring cell numbers via CyQUANT®. As shown in Figure 4.4A, while solutions composed solely of NaCl resulted on average to less adhered cells after 7 days, there are no significant differences in the cell number between surfaces (p > 0.068) after 3 days and 7 days. This implies that the FN immobilization scheme, at least in the long term, does not influence net cell growth. This is in contrast to other studies reporting that cell number in the short-term (i.e., < 24h) is influenced by the FN immobilization scheme [80, 461, 462]. Moreover, these CyQUANT® results imply that solution composition during FN immobilization does not affect INS-1 cell net growth.

130 13 days (A) 17 days

(QOj foO j (atadj [NtaQf| [m m | iooj [tam]

HZSmMQucose (B) [ZZl 28mM Quoose 28mM Qucose + 5 0 ^ IBMX ■ ■ 2.8rrM Qucose

[aaj IMgOj (ttaj Ip*o | [mimcI [a aj [urm) (C)

m 1

f“»J &»“) [M

Figure 4.4: (A) INS-1 cell numbers on surfaces measured via CyQUANT®. (B) INS-1 cell insulin secretion following glucose stimulation after 7 days of culture. (C) Stimulation index for INS-1 cells from 2.8mM Glucose to 28mM Glucose. For each (A), (B), and (C), salt concentrations are lOpM unless stated otherwise. Data are presented as average ± S.E.M (n=3). 131 As shown in Figure 4.4B, in all conditions INS-1 cells secrete insulin in a glucose responsive manner. A significant increase in insulin secretion ip < 0.025) was observed when glucose concentration was changed from 2.8mM to 28mM. Upon the addition of 50pM IBMX to 28mM glucose, a modest but not statistically significant increase (p > 0.079) in insulin secretion was observed for most conditions. INS-1 cells on surfaces made in NaCl had significant increases (p < 0.033) in insulin release upon the addition of 50pM IBMX to 28mM glucose. The cells on the clean glass control did not show an increase in insulin secretion upon addition of 50pM IBMX to 28mM glucose. Lastly, a return to 2.8mM glucose resulted in a marked decrease of insulin secretion for all conditions ip < 0.031) except for that of the clean glass control ip - 0.071). We will now discuss the differences in insulin secretion between surface conditions and their significance when going from 2.8mM to 28mM glucose as this condition is perhaps parallel to initial P-cell response for the return to normoglycemia.

For FN grafting, INS-1 cells on FN-CMD [MgCl2] had significantly higher levels of insulin release than those on FN-CMD[CaCl 2] and FN-CMD[MnCl2] ip = 0.01 and p = 0.03, respectively). Here, the underlying adlayers are the same (i.e., CMD) and the salts used during the immobilization are different {i.e., CaCl2, MgCfe, and MnCL) and at the same concentration. Yet, insulin response to glucose is significantly different. For FN adsorption, cells on FN- Glass[PBS] had significantly higher insulin release than those on FN-Glass[CaCl2] ip = 0.002). Here again, the underlying substrate below FN is the same, glass. Yet with different solution composition, insulin secretion levels are significantly different. Together, these results confirm that the solution composition used during immobilization has a direct influence on the resultant activity of the immobilized FN and the subsequent INS-1 cell insulin secretion.

Cells on FN-CMD [MgCl2] and FN-Glass[PBS] had some of the highest levels of insulin secretion. Furthermore, there was no significant difference in insulin release between cells on FN-CMD[MgCl2] and FN-Glass[PBS] ip = 0.56). Here, solution composition during immobilization, the underlying substrate/adlayer, and the immobilization scheme are different. Yet there is no significant difference in INS-1 cell insulin secretion. While FN-CMD[NaCl]), and FN-CMD[lmM NaCl] appear to result in good quantities of insulin secretion per cell, particularly with IBMX secretagogues, as shown in Figure 4.4C, their response to an increase in

132 glucose concentration (i.e., stimulation index) is lower on average in comparison to all other conditions. Together these results further confirm solution composition has a drastic influence on the resultant bioactivity of FN surfaces.

Our cell response results show that while cell function can vary depending on whether or not one employs different immobilization schemes (chemical grafting or physical adsorption) and different underlying surfaces (CMD or glass), solute composition used during the immobilization process can have such a significant effect that variances in cell function become absent (see Figure 4.4B). These results are compelling as numerous previous studies report that FN chemically grafted and physically adsorbed has differences in its biological activity as observed via their interaction with cells [80, 461-463]. Some of these studies report that grafted FN is more biologically active than adsorbed [80, 463], while others report the opposite [461, 462]. Moreover, cells on FN surfaces with different underlying substrates typically have varying responses with respect to one another [441]. Together, our results and those previously reported, corroborate that one should consider solution composition during biomolecule immobilization as a factor when attempting to design surfaces to elicit a desired cellular response.

The observed variations in cell response are likely due to varying conformational states of immobilized FN which could be attributed to the solution ion’s classification on the Hofmeister series. It should be noted that unlike INS-1 cells, CD and PCS are perhaps not sensitive enough to detect such subtle conformational changes as the results of Figure 4.2 indicate. According to the Hofmeister series, Ca2+ and Mg2+ ions are more “structure breaking” relative to Na+ [348, 450]. Nevertheless, each Na+, Ca2+, and Mg2+ break water structure to some extent surrounding hydrophilic moieties. The breaking of the water structure likely results in some ion penetration into the FN core which screens some intermolecular electrostatic forces yielding a conformational change. Differences in INS-1 cell response are attributed in part to this conformational change, particularly for FN-CMD[CaCl2], FN-CMD[MgCl2], FN-CMD[CaCl2], FN-CMD[NaCl] where the underlying substrates are the same , CMD.

Differences in INS-1 cell response between FN-CMD surfaces may also be attributed to the size of the ion. Varying hydrated radii have quantitatively been reported for Mg2+, Ca2+, and Na+ depending on the method of measurement [84, 464]. However, it has consistently been reported that the size of the ions under hydration follows Mg2+ > Ca2+ > Na+. It is plausible that the

133 smaller the ion, the deeper it penetrates into the FN core thereby further extending intermolecular electrostatic screening. Our results indicate that smaller solution ions result in less of a favourable FN conformation as INS-1 cell response to FN-CMD surfaces in which NaCl was employed had on average less cell net growth and a lower stimulation index after 7 days as shown in Figure 4.4.

Finally, it is worth noting that in this study all solutions contain some NaCl as FN employed in this study arrives with NaCl in the stock solution (see Section 4.3.1). Thus FN solutions in lOpM CaCl2 and lOpM MgCl2 also contain some NaCl. In a solution containing a mixture of 2:1 (Mg2+ and Ca2+) and 1:1 (Na4) chloride electrolytes, it is believed that Mg2+ and Ca2+ are preferentially solubilized against hydrophilic surfaces [348]. Therefore, MgCl2 and CaCl2 interactions with FN that induce conformational changes supersede those of NaCl interactions with FN. Our results corroborate such a partitioning of ions in solution as FN surfaces from lOpM CaCl2 lOpM MgCl2 (also with some NaCl) solutions results in significantly different net cell growth and insulin secretion (see Figure 4.4).

It should be mentioned that varying experimental conditions between this study and those just cited [80, 461-463] can also somewhat contribute to the relative differences we observe. The biological activity of immobilized FN is largely correlated with the cellular attachment and the subsequent cell morphology on FN-bearing surfaces after a rather short culture time (i.e., 2h to 24h). Cellular adhesion and spreading to a biomaterial surface, which can be assessed with any cell phenotype, and cellular function, such as glucose-stimulated insulin secretion which is specific to beta-like cells, such as INS-1 cells, are two completely different analyses. For short term INS-1 cell attachment and morphological studies to FN and other ECM cues, the reader is referred to a detailed study by Krishnamurthy et al. [453].

4.5 Conclusion

We demonstrate the successful grafting of fibronectin (FN) to low-fouling surfaces made of carboxymethyl-dextran (CMD). It is shown that solution composition plays a crucial factor for the successful grafting of FN to CMD. Thus, in addition to the important usual considerations

134 such as surface chemistry, composition, topography, etc., one should also consider solution composition during the biomolecule immobilization process when designing and preparing biomaterial surfaces. This study shows that solution composition can not only impact the successful immobilization of a large biomolecule such as FN to a low-fouling surface, but also the subsequent activity of the protein-bearing surface towards cells, as exemplified here by the FN surface influence on the glucose-stimulated insulin secretion of INS-1 cells.

4.6 Acknowledgment

This research project was supported by the University de Sherbrooke and the National Science and Engineering Research Council of Canada (NSERC) through a Discovery Grant. The authors are grateful to Georges Sabra, Mathew Evans, Heidi Brochu, Carina Kuehn, Laurent Olivier, Charles Gaudreault, Julie Chouinard, Marie-Eve Beaulieu, Prof. Pierre Lavigne, Xia Tong, Prof. Yue Zhao, Sonia Blais, Normand Pothier, and Irene Kelsey.

135 Chapter 5 Co-Culture of Young Porcine Islets with Liver Microvascular Endothelial Cells in Fibrin Improves Insulin Secretion and Reduces Apoptosis

136 Foreword

Authors and Affiliation:

Evan A. Dubiel : Ph.D. Candidate, Universite de Sherbrooke, Departement de genie chimique et de genie biotechnologique

Jonathan R.T. Lakey : Associate Professor, University of California, Irvine, Department of Surgery and Biomedical Engineering

Morgan W. Lamb : Staff Research Associate, University of California, Irvine, Department of Surgery and Biomedical Engineering

Patrick Vermette : Professor, Universite de Sherbrooke, Departement de genie chimique et de genie biotechnologique

Date of Submission : October 25th, 2012

Journal: Acta Biomaterialia

Contribution :

This article contributes to the third and final experimental work of this thesis. All experimental work and data analysis was performed by Evan A. Dubiel. Porcine islets were isolated at the University of California, Irvine by Morgan Lamb and Jonathan Lakey. The article was written by Evan A. Dubiel. All work was performed under the direction and supervision of Patrick Vermette.

137 Titre en frangais :

L’amelioration de la secretion d’insuline et diminution de l’apoptose via la co-culture dans la fibrine d’ilots porcins jeunes avec des cellules endotheliales micro-vasculaires du foie

R 6sum £:

L’isolation des llots de Langerhans conduit a la destruction des interactions entre la matrice extracellulaire et les vaisseaux des flots. Ces demiers favorisent la viabilite et la fonctionnalite des ilots. Cette etude a pour but de retablir les interactions mentionnees precedemment de fagon in vitro en dispersant des ilots de pore (PI) dans un gel de fibrine a l’interieur duquel des cellules endotheliales micro-vasculaires de foie de pore (PLMEC - Porcine Liver Microvascular Endothelial Cells ) sont dispersees. Une augmentation de la concentration d’insuline presente dans le milieu de culture a ete mesuree dans le cas de la co-culture des PI et des PLMEC dans un gel de fibrine [PI + PLMEC (dans la fibrine)] suite a une periode de 2 jours d’incubation. Par contre, une diminution de cette meme concentration a ete mesuree pour une periode de 7 jours pour des PI dans un gel de fibrine [PI(dans la fibrine)], des PI sur une monocouche confluente de PLMEC [PI-PLMEC monolayer] ainsi que des PI cultives sur du polystyrene de grade de culture cellulaire. Une augmentation de la secretion de l’insuline en reponse au glucose a ete mesuree pour les cultures de monocouches de PI+PLMEC (dans la fibrine) et de PI-PLMEC suite a deux jours d’incubation. Suite a une periode de culture de 7 jours, la reponse de l’insuline au glucose demeure amelioree pour la culture PI+PLMEC (dans la fibrine) par rapport a la culture PI (dans la fibrine). Le pourcentage de cellules exprimant l'insuline a augmente pour la culture de PI+PLMEC(dans la fibrine) sur une periode de deux joins tandis que ce pourcentage n’a pas varie pour les cultures de PI (dans la fibrine) et de monocouche de PI+PLMEC (dans la fibrine) et diminue pour la culture de PI-TCPS. Une diminution du pourcentage de cellules exprimant l'insuline a ete mesuree pour toutes les cultures mentionnees precedemment pour une duree d’incubation de 7 jours. Un pourcentage inferieur de cellules apoptotiques a ete detecte suite a 2 jours d’incubation pour la culture PI-PLMEC (dans la fibrine) lorsque compare aux autres types de cultures.

138 5.1 Abstract

The isolation of islets of Langerhans results in disruption of both islet-extracellular matrix interactions and islet vasculature resulting in reduced viability and functionality. This study attempts to re-establish both disruptions simultaneously in vitro by embedding isolated young porcine islets (PI) in fibrin gel where porcine liver microvascular endothelial cells (PLMEC) are dispersed. PI co-cultured with PLMEC embedded in fibrin [PI+PLMEC(in Fibrin)] showed increases in insulin content of culture media after 2 days while it only declined over 7 days for PI embedded in fibrin [PI(in Fibrin)], PI on a confluent PLMEC monolayer [PI-PLMEC monolayer], and PI on tissue culture polystyrene [PI-TCPS]. PI+PLMEC(in Fibrin) and PI- PLMEC monolayer cultures showed improved insulin secretion in a glucose responsive manner after 2 days. After 7 days, insulin response to glucose remained significantly improved with PI+PLMEC(in Fibrin) cultures over PI(in Fibrin). The percentage of insulin expressing P-cells increased for PI+PLMEC(in Fibrin) after 2 days while it remained unchanged for PI(in Fibrin) and PI-PLMEC monolayer, and declined for PI-TCPS. All cultures had a marked decrease in 0- cells percentage after 7 days. After 2 days PI-PLMEC(in Fibrin) had significantly lower percentages of apoptotic cells in comparison to all other conditions corroborating all insulin secretion results.

139 5.2 Introduction

Surgical transplantation of islets of Langerhans is a treatment option for selective patients with severe Type I Diabetes Mellitus [2]. This procedure involves isolating islets of Langerhans from cadaveric donors and infusion into the hepatic portal vein [3]. Five year outcomes demonstrate approximately 5-20% of patients being able to remain exogenous insulin free. It has been well established that isolation of islets results in disruption of the islet-extracellular matrix (ECM) relationship and in the loss of islet vasculature [422, 465]. This may contribute to the progressive loss of islet graft function in transplanted patients. Therefore, the present study attempts to simultaneously re-establish islet-ECM interactions and vascularisation in vitro by embedding isolated porcine islets in fibrin, a natural ECM biodegradable scaffold widely used in vascular tissue engineering [192], where endothelial cells of liver microvascular origin are dispersed.

Past studies have characterized the effects of both two-dimensional (2D) and three-dimensional (3D) ECM scaffolds on isolated islets and found benefits [392, 422, 466]. Fibrin gels have also shown promising benefits to isolated islets including p-cell proliferation, reduced apoptosis, and beneficial transplantation delivery [467-470]. Furthermore, fibrin is known to play a critical role in angiogenesis, the formation of new blood vessels from pre-existing vessels [471]. Moreover, isolated islets co-cultured in vitro with endothelial-mesenchymal stem cell composites in fibrin showed capillary-like structures penetrating into islets [472], Fibrin was thus chosen as an ECM scaffold for this study.

Endothelial cells comprising blood vessels play a critical role in the physiology of the endocrine pancreas; inducing pancreatic development, supplying islets with oxygen and nutrients, and regulating insulin secretion and transport to target tissues [473-475]. Furthermore, endothelial cells have also been employed to counteract blood-mediated inflammatory reactions in transplanted islets [476]. It is thus of interest to study islet-endothelial cell interactions.

Endothelial cells of angiogenic capacity are believed to reside in isolated islets [477]. After transplantation, intra-islet endothelial cells of the donor along with those of the recipient have been shown to contribute to revascularization and the formation of vessels [478]. It has also

140 been shown that intra-islet endothelial cells disappear rapidly during culture (approximately 80% loss after 2 days) [479]. However, this may be beneficial as during an immune attack, intra-islet endothelial cells are believed to be involved [480, 481]. Moreover, a recent islet transplantation study reports that islets grafts originating from cultured isolated islets reverse diabetes more rapidly than those of freshly isolated islets, calling for a culture strategy to simultaneously package islets with both additional endothelial cells and angiogenic factors [482]. Therefore, it is of interest to study the co-culture of islets and endothelial cells in fibrin.

5.3 Materials and Methods

5.3.1 Islet Isolation and Maturation of Porcine Islets

Islet isolation was performed with the approval of the Institutional Animal Care and Use Committees. Pancreases from young Yorkshire pigs (14-22 days) are removed using rapid surgical procurement (< 5 minutes) and placed in University of Wisconsin (UW) preservation solution; cold ischemia time was limited to 10 minutes. Pancreases were trimmed of surrounding adipose and lymphatic tissue. Islets were then digested from the majority of surrounding acinar tissue by mincing in a low dose collagenase MA/BP protease enzyme mixture (Vitacyte, Indianapolis, IN, U.S.A.). Tissue was allowed to further digest for 15 ± 2 minutes at 37°C and 60rpms. Digestion was then inhibited using cold HBSS supplemented with HEPES and 10% BSA. Partial digested tissue was then filtered through a 500pm metal mesh funnel, to remove undigested tissue greater than 500pm. Islet tissue clusters, 50-500pm, were then cultured and allowed to mature into complete islets. Islets were cultured at 37°C, 5% CO 2 in T- 175 suspension flasks containing our novel culture media (Mediatech, Manassas, VA, U.S.A.) [Lamb, et al. Transplantation submitted 2012] and, supplemented with 10% porcine serum, and allowed to mature for 8 days, with media changes every 48 hours. After 8 days, porcine islets (PI) were shipped overnight to the Universite de Sherbrooke, in tissue culture media (room temperature).

141 5.3.2 Cell Culture, Islet Culture, Fibrin Gels, and Time Lapse Microscopy

Porcine liver microvascular endothelial cells (PLMEC, Cell Systems Corporation, Kirkland, WA, USA, CSC 2PL3) were cultured using media and reagents recommended by the manufacturer (CSC complete medium®, 4Z0-500; Culture Boost®, 4CB-500; Bac-Off®, 4Z0- 462; Attachment Factor®, 4Z0-2O1). For PI-PLMEC monolayer cultures, PLMEC were seeded at a density of lxlO4 cellscm'2 in 48-well tissue culture polystyrene plates (TCPS) and allowed to form a confluent monolayer 72 hours prior to the addition of islets. Upon the arrival of islets, CSC complete medium was then removed, the PLMEC monolayer was rinsed once with hanks balanced salt solution (HBSS, Sigma-Aldrich, Oakville, ON, Canada, H1387), and then exposed to ~200 islet equivalents (IEQ) in 200pL of CMRL-1066 media (Life Technologies, Burlington, ON, Canada, 11530) supplemented with 10% v/v foetal bovine serum (FBS, Life Technologies™, A12617) and 1% v/v penicillin-streptomycin (Life Technologies™, 15140). As a control to PI-PLMEC monolayer cultures, -200 IEQ of PI were exposed to clean 48-well TCPS plates in 200pL of supplemented CMRL-1066 (PI-TCPS).

Fibrin gels were composed of 2 mg mL'1 fibrinogen (Sigma-Aldrich, F8630), 1 U mL'1 thrombin (Sigma-Aldrich, T9549), and 0.25 TIU m L 1 aprotinin (Sigma-Aldrich, A3428). All gel components were dissolved/suspended in HBSS. All gels had a final volume of 200pL and were plated in 48-well TCPS plates. For three-dimensional (3D) PLMEC+PI(in Fibrin), PLMEC (2xl06cells mL'') and PI (200 IEQ per well) were dispersed in fibrin. As a control to 3D PLMEC+PI(in Fibrin) cultures, -200 IEQ of PI were seeded in fibrin [PI(in Fibrin)]. For both 3D fibrin cultures, gels were allowed to polymerize for 5min and then covered with 200pL of supplemented CMRL-1066.

For all conditions, cultures were incubated at 37°C in a 5% CO 2 environment. Supplemented CMRL-1066 was changed every 24h. Used supplemented CMRL-1066 was collected and stored at -20°C until analysis. Supplemented CMRL-1066 contains 5.56mM D-glucose. Insulin content of collected media was measured using enzyme-linked immunosorbent assay (ELISA) kits purchased from Mercodia (Winston-Salem, NC, USA, 10-1223-01 & 10-1200-01). Absorbance was measured using a Safire2 microplate reader (Tecan Group Ltd., Mannedorf, Switzerland). Measured insulin content was normalized to that of 24h. Data are represented as

142 average ± standard error of the mean (S.E.M.) of n > 3. Analysis of variance (ANOVA) was used to compute any possible significant differences between means, lip < 0.05, means were considered statistically different. For time lapse microscopy, phase contrast images were recorded at 200x magnification using a Nikon Eclipse TE2000-S microscope.

5.3.3 Glucose-Stimulated Insulin Secretion

Glucose-stimulated insulin secretion (GSIS) was performed on all culture conditions. For 2D PI- PLMEC monolayer and PI-TCPS cultures, Krebs buffer solution composed of 25mM HEPES

(Sigma-Aldrich, H3375), 115mM NaCl (Sigma-Aldrich, S9888), 24mM NaHCC >3 (EMD Chemicals, Gibbstown, NJ, USA, SX0320-1), 5mM KC1 (Fisher Scientific, Ottawa, ON, Canada, BP366), ImM MgCb (Fisher Scientific, AC413415000), 5mM CaCl2 (Sigma-Aldrich, C7902), 0.1% w/v bovine serum albumin (BSA, Sigma-Aldrich, A7906) supplemented either with 2.8mM D-glucose (low-glucose, Sigma-Aldrich, G6152) or 28mM D-glucose (high-glucose) was employed. Cultures were incubated for lh at 37°C for each glucose concentration. Samples were stored at -20°C prior to analysis.

Because of significantly different physical and mass transport properties between 2D and 3D cell culture systems [441], the GSIS protocol for 3D PI+PLMEC(in Fibrin) and PI(in Fibrin) was modified. RPMI-1640 (Life Technologies™, 118739) media supplemented with 1% Penstrep and either 2.8mM D-glucose (low-glucose) or 28mM D-glucose (high-glucose) was used. Cultures were initially incubated overnight with low glucose media. The next day, cultures were incubated for 12h at 37°C for each glucose concentration. Samples were stored at -20°C prior to analysis.

For all culture conditions, insulin content was measured using ELISA kits purchased from Mercodia. Absorbance was measured using a Safire2 microplate reader (Tecan Group Ltd.). The stimulation index was calculated by dividing the high-glucose insulin content by that of the low- glucose. Data are presented as average ± S.E.M. of n = 3. ANOVA {p < 0.05) was used to compute any possible significant differences between means.

143 5.3.4 Immunohistological Staining and Morphometric Analysis

PI were fixed overnight in 4% paraformaldehyde, (Sigma-Aldrich, P6148) at 4°C. Samples were then processed for paraffin embedding and sectioning using standard protocols [483]. Paraffin blocks were sliced into 4pm sections. Primary antibodies to insulin (ab9569) and glucagon (ab 18461) were purchased from Abeam (Cambridge, MA, USA). Alexa Fluor® 488 and Alexa Fluor® 555 conjugated secondary antibodies were purchased from Life Technologies™ (A 11001 & A21428 respectively). Nuclei counterstaining was performed using 4’6-diamidino-2- phynylindole, dihydrochloride (DAPI, Life Technologies™, D1306). Manufacturer guidelines were followed for pretreatment, dilutions, incubation times, and incubation temperatures. Samples were viewed using an epifluorescence microscope (Nikon Eclipse).

An APO-BRDU-IHC™ kit for measuring apoptosis via terminal deoxynucleotide transferase dUTP Nick End Labeling (TUNEL) was purchased from Phoenix Flow Systems (San Diego, CA, USA, AH1001). Nuclei counterstain was performed using methyl green supplied by the manufacturer. Apoptosis was viewed using an optical bright field microscope (Leica).

Cells positive for insulin, glucagon, or apoptosis from individual histological sections were manually counted and normalized to the total cell number from DAPI and methyl green counterstain. Data are presented as average ± S.E.M. of n > 3. ANOVA (p < 0.05) was used to compute any possible significant differences between means. Percent loss was calculated by subtracting the quantity of insulin positive cells at 7 days from that at 2 days and then dividing by the quantity at 2 days. The uncertainty in percent loss was calculated by averaging the uncertainties at 2 days and 7 days.

5.4 Results and Discussion

5.4.1 Time Lapse Microscopy

PLMEC began to spread and PI began to sprout simultaneously after 24h of culture (Figure 5.1 A) for PLMEC+PI(in Fibrin) [472]. Likewise, PI began to sprout after 24h (Figure 5.IB) for

144 PI(in Fibrin) [477]. Day by day images of both 3D PLMEC+PI(in Fibrin) and PI(in Fibrin) cultures are provided as supplemental materials (see Figure C.l and Figure C.2). After 24, PI attach to the PLMEC monolayer surface while PLMEC comprising a monolayer appear to change their morphology; PLMEC comprising the monolayer nearly disappear after 48h (Figure 5.1C). Single cells of fibroblastoid morphological features (i.e., elongated spindle-like shape) begin to appear and proliferate throughout both 2D PI-PLMEC monolayer and PI-TCPS cultures (see Figure C.3 and Figure C.4).

(A) PLMEC+PI (in Fibrin)

(B)

pi (in Fibrin)

5.4.2 Insulin Content in Media and Response to Glucose Stimulation

The capacity of PI to secrete insulin was assessed by measuring insulin content of the basal media. Note that supplemented CMRL-1066 media contains 5.56mM D-glucose. PLMEC+PI(in Fibrin) showed the highest levels of insulin content in basal media (Figure 5.2) of all culture conditions. In fact, while other cultures showed a continuous decline in insulin content after 1 day, PLMEC+PI(in Fibrin) cultures showed an increase in insulin content from 1 day to 2 days.

145 However, afterwards insulin content also began to decline sequentially in PLMEC+PI(in Fibrin) cultures. Overall, both 3D PLMEC+PI(in Fibrin) and PI(in Fibrin) cultures had significantly more insulin content than those of 2D PI-PLMEC monolayer and PI-TCPS cultures. In 3D cultures, PLMEC+PI(in Fibrin) cultures showed higher insulin content than those of PI(in Fibrin) throughout the culture period (p < 0.030). In 2D cultures, PI-PLMEC monolayer cultures showed significantly less of a decline in insulin content (p = 0.036) after the first 3 days of culture relative to PI-TCPS.

- m - 3D PLMEC + PI On Fibrin) 3D PI On Fibrin) 1.4 -| 2D PI-PLMEC monolayer 2D P I-TC P S

1.2 - O * 10- .a (O | 0.8- o 2 c 0.6 - 0c 1 04- £ — 0.2 -

0.0 1 2 3 4 56 7 Culture Time (days)

Figure 5.2: Insulin content of basal media measured via ELISA. PLMEC+PI(in Fibrin) show an increase in insulin content from 1 day to 2 days. Throughout the 7-day culture, both 3D fibrin gel cultures showed higher insulin content than 2D TCPS cultures. Data are represented as average ± S.E.M. of n=3.

PI functionality was accessed through a glucose-stimulated insulin secretion (GSIS) assay. For all culture conditions, insulin content was measured and the stimulation index is defined as each individual replicate's 28mM insulin content divided by 2.8mM insulin content.

146 For 3D PLMEC+PI(in Fibrin) and PI(in Fibrin) cultures, after 2 days increases in insulin content after a D-glucose increase from 2.8mM to 28mM (Figure 5.3A) were present. The stimulation index was significantly higher (p = 0.002) for 3D PLMEC+PI(in Fibrin) compared to PI(in Fibrin) cultures. After 7 days, both 3D PLMEC+PI(in Fibrin) and PI(in Fibrin) cultures showed increases in insulin content after an increase in D-glucose concentration (Figure 5.3A). Furthermore, the stimulation index showed a significant increase (p = 0.0002) for 3D PLMEC+PI(in Fibrin) compared to PI(in Fibrin) cultures after 7 days.

For 2D PI-PLMEC monolayer and PI-TCPS cultures, after 2 days, insulin content did increase when stimulated from 2.8mM to 28mM D-glucose (Figure 5.3B). The stimulation index for PI- PLMEC monolayer cultures was significantly higher (p - 0.04) than that for PI-TCPS cultures. After 7days, increases in insulin content due to glucose stimulation were present but had significant uncertainty (Figure 5.3B). Furthermore, there was no significant difference (p = 0.38) in stimulation index between PI-PLMEC monolayer and PI-TCPS cultures.

The considerable uncertainty in most of the insulin content measurements should be addressed. Islets are considerably different from one another. Islets can vary in diameter (smaller islets produce insulin more efficiently), contain varying amounts of insulin producing p-cells, and islet isolations can also vary in their degree of success [437, 484]. The end result is islets of varying ability to produce insulin before culture and thus varying quantitative measurements in insulin content. The stimulation index is often employed to normalize such variability in the ability of islets to produce insulin. This is evident as with the possible exception of 7 day 2D cultures, uncertainty of stimulation indices are more than acceptable (Figure 5.3).

In summary, these results confirm that the co-culture of PLMEC with PI improves insulin secretion capacity, regardless of whether one employs a 3D fibrin or 2D TCPS culture. 3D PLMEC+PI(in Fibrin) showed an increase in insulin content from 1 day to 2 days (Figure 5.2) while all others declined. 2D PI-PLMEC monolayer cultures began with less of a decline in insulin content after the first 3 days of culture in comparison with PI-TCPS cultures. However, afterwards, the two cultures were on par (Figure 5.2). Stimulation indices were significantly higher for 3D PLMEC+PI(in Fibrin) than those of 3D PI(in Fibrin) after 2 days and 7 days (Figure 5.3A). 2D PI-PLMEC monolayer cultures also had significantly higher stimulation indices than those of PI-TCPS after 2 days but not after 7 days (Figure 5.3B).

147 (A) 3D Fibrin Gel Cultures Insulin Content Stimulation Index

1 - Days |

f i t n a i p m i nm*

7 Days |

(B) 2D TCPS Cultures Insulin Content Stimulation Index

2 Days | I a l ri II 1+

7 Days

Figure 5.3: Insulin content and stimulation index of glucose-stimulated insulin secretion assay for (A) 3D PLMEC+PI(in Fibrin) and PI(in Fibrin) cultures and (B) 2D PI-PLMEC monolayer and PI-TCPS cultures. Data are represented as average ± S.E.M. of n=3.

148 5.4.3 Porcine Islet Insulin Expression and Cellular Apoptosis

The percentage of insulin expressing p-cells was accessed via immunofluorescence staining (Figure 5.4). For the time point 0 days, a sample of islets was taken and fixed (according to Section 5.3.4), immediately upon arrival at the Universite de Sherbrooke. Immediately afterwards, PLMEC+PI(in Fibrin), PI(in Fibrin), PI-PLMEC monolayer, and PI-TCPS culture conditions were prepared and culture commenced. An image of insulin and glucagon expression at 0 day is available as supplemental material (Figure C.5).

Insulin/Glucagon TUNEL/Nuolei

PLMEC+PI (in Fibrin)

PI- PLMEC monolayer

Figure 5.4: Sample immunofluorescence images of insulin and glucagon expression and immunohistochemistry images of apoptosis (i.e., TUNEL) of porcine islets for all culture condition at 2 days and 7 days. Images of insulin, glucagon, and TUNEL staining at 0 day are available as supplementary material (Figure C.5).

149 After 2 days, PLMEC+PI(in Fibrin) cultures contained a significantly higher percentage of insulin expressing cells ip < 0.0006) compared to all other conditions (Figure 5.5A). In fact, PLMEC+PI(in Fibrin) cultures, had a significantly higher percentage of insulin expressing cells at 2 days in comparison to 0 days {p = 0.004). On the other hand, PI(in Fibrin) had similar levels of insulin expressing cells (as did PI-PLMEC monolayers) from 0 day through 2 days (p > 0.62). The PLMEC+PI(in Fibrin) insulin expressing cell percentage increase may be due to proliferation. It has been previously reported that p-cells of human islets undergo proliferation only when embedded in fibrin and exposed to a cocktail of growth factors [468]. In this particular study, perhaps PLMEC provide the extra niche necessary for p-cell proliferation like the growth factor cocktail did in the study of Beattie et al. The insulin expressing cell percentage increase after 2 days for PLMEC+PI(in Fibrin) is also in accordance with the increase of insulin content observed in PLMEC+PI(in Fibrin) culture media after 2 days (Figure 5.2). Finally, only PI-TCPS cultures had a significantly lower percentage of insulin expressing cells from 0 days through 2 days ip = 0.05). PI(in Fibrin) and PI-PLMEC monolayer cultures had similar percentages of insulin expressing cells ip = 0.73) after 2 days, both significantly higher than that of the PI-TCPS culture (p < 0.005). In summary, these results in insulin expressing cell percentage from 0 days through 2 days (Figure 5.5A) indicate that initially PLMEC and fibrin can in a mutually exclusive manner {i.e., PI(in Fibrin) and PI-PLMEC monolayer), help maintain P-cell phenotype in porcine islets. In combination, PLMEC and fibrin helps increase the percentage of insulin expressing in porcine islets as evident in PLMEC+PI(in Fibrin) cultures.

After 7 days, PLMEC+PI(in Fibrin) cultures had significantly more insulin expressing cells (p < 0.05) than all other culture conditions (Figure 5.5A). However, all culture conditions had a marked loss in the percentage of cells expressing insulin from 2 days through 7 days (Figure 5.5A). All culture conditions had a significant percent loss of insulin expressing cells from 2 days through 7 days (Figure 5.5B). 3D PLMEC+PI(in Fibrin) and PI(in Fibrin) cultures had a higher percent loss than those of 2D PI-PLMEC monolayer and PI-TCPS cultures (Figure 5.5B). This is perhaps due to the fact that 3D fibrin culture systems have more limited mass transport properties than 2D TCPS systems [441], Over the period of culture, waste perhaps accumulates preventing the delivery of oxygen and other nutrients having a detrimental effect on p-cells which have high oxygen and nutrient requirements [485]. The levels of glucagon expressing cells for all culture conditions are available as supplemental data (Figure C.6).

150 (A) CZUO Days 60-|

45-

30- JQ 3 * 15-

flnRDrt* F u e c r o ti^ e r TCPS

(B) 0O-i

60-

JS2

3 40-

2 0 -

PI n

60-

30- JS 3 15-

oc*p (hFfc** FUECmonci*ar TCPS

Figure 5.5: (A) Expression levels of insulin and glucagon and 0 day, 2 days, and 7 days. Data are represented as average ± S.E.M. of n=3. (B) Percent loss of insulin expressing p-cells from 2 days to 7 days. (C) Expression of apoptotic cells measured via TUNEL. Data are represented as average ± S.E.M. of n=3.

151 The percentage of apoptotic cells was accessed (Figure 5.4). An image of apoptosis expression via terminal deoxynucleotide transferase dUTP Nick End Labeling (TUNEL) expression at 0 days is available as supplemental material (Figure C.5). After 2 days, PLMEC+PI(in Fibrin) cultures had a significantly less ip < 0.005 ) apoptotic cell percentage than all other conditions (Figure 5.5C). In fact, PLMEC+PI(in Fibrin) cultures had similar percentages (p = 0.75) of apoptosis to PI at 0 days of culture. PI(in Fibrin) had a significantly higher percentage of apoptosis at 2 days ip - 0.0008) relative to PLMEC+PI(in Fibrin) but significantly lower in comparison to 2D PI-PLMEC monolayer and PI-TCPS cultures ip <0.005). These results are in agreement with previous studies that observed reduced TUNEL and active caspase-3 expression in human islets that were embedded in fibrin gel [468, 470]. This implies that fibrin suppresses cell death mediated via cellular interactions. Moreover, cell death is further reduced in fibrin gel in the presence of PLMEC.

After 7 days, 3D PLMEC+PI(in Fibrin) had on average an increase in the percentage of apoptotic cells from 2 days through 7 days (Figure 5.5C). However, because of considerable uncertainty, there was no statistically significant difference between the means of the percentage of apoptotic cells at 2 days and 7 days ip = 0.07). 3D PI(in Fibrin) had no significant differences in apoptosis level from 2 days through 7 days ip = 0.85). 2D PI-PLMEC monolayer and PI-TCPS cultures had significant decreases in the percentage of apoptotic cells from 2 days through 7 days ip < 0.05). This decrease is likely attributed to the fact that far fewer endocrine cells are present after 7 days because of such high levels of apoptosis after 2 days (see Figure 5.5A). Likewise, the statistically identical percentages of apoptotic cells in 3D PLMEC+PI(in Fibrin) and PI(in Fibrin) from 2 days through 7 days may be attributed to the higher percentage of insulin expressing P-cells at 2 days and 7 days (see Figure 5.5A). p-cells of isolated islets are known to be susceptible to apoptosis [465]. Therefore, if more P-cells are present throughout the PLMEC+PI(in Fibrin) culture period, as the results of percent insulin expression indicate (see Figure 5.5A), one may expect a higher percentage of apoptotic cells at the end of the 7 day culture period. In summary, these results in the percentage of apoptotic cells indicates that when PI are embedded in 3D fibrin gels, apoptosis is initially suppressed (i.e., after 2 days) early in the culture period, but not after a longer culture period (i.e., 7 days).

152 5.5 Conclusion

This study demonstrates that isolated porcine islets co-cultured with porcine liver microvascular endothelial cells in fibrin gels improve insulin secreting capacity and reduce cellular apoptosis. With further improvements, maybe dynamic culture, such findings can perhaps aide in pre transplant culture to promote islet recovery from the isolation process.

5.6 Acknowledgements

This research project was supported by the Universite de Sherbrooke and the National Science and Engineering Research Council of Canada (NSERC) through a Discovery Grant awarded to P. Vermette (grant # 250296-07). Funding was also provided through the Department of Surgery, University California Irvine. We are grateful to Jocelyne Ayotte and Michael Alexander for their technical assistance.

153 Chapter 6 CONCLUSIONS

154 Conclusions - Frangais

Le travail de cette these visait a repondre a certaines des limitations de la transplantation d'ilots pancreatiques endocriniens en tant que therapie pour le diabete de type I. Grace a l'utilisation de surfaces biomimetiques et de biomateriaux, il a ete tente de proposer une piste de solution a la grave penurie de pancreas disponibles pour l'isolation d'ilots et la perte subsequente progressive de leurs fonctions. Un aper?u des experiences realisees dans cette these, leur interpretation et des suggestions de travaux futurs vont etre presentes ci bas.

Le premier travail experimental qui constitue le Chapitre 3 est intitule « La morphogenese in vitro des cellules PANC-1 vers des agregats semblables a des Hots par le biais des surfaces de dextrane derivees recouvertes de RGD ». Cette etude a permis d'etudier une piste de solution a la penurie d'ilots de pancreas endocriniens disponibles pour la transplantation. II a ete constate que les cellules PANC-1 sur des surfaces de RGD-CMD produisaient le plus d'insuline par rapport aux surfaces RGE-CMD, CMD, et les controles. Bien que ce resultat soit encourageant considerant que la surface biomimetique {i.e., RGD-CMD) a produit le meilleur resultat, l'expression de l'insuline dans les cellules et la liberation d'insuline dans les milieux de culture n'ont ete detectees qu'apres 14 jours. Ce resultat contraste avec une autre etude rapportant l'expression de l'insuline dans les cellules apres seulement 4 jours; la secretion d'insuline a ete rapportee dans l'etude mais les donnees n'ont pas ete presentees [409]. Cela implique que la comprehension de la differentiation des cellules PANC-1 en un phenotype pancreatique endocrinien est quelque peu controversee et incomplete. Une controverse parallele existe aussi quant a l'origine des progeniteurs des cellules du pancreas au cours des etapes de developpement du pancreas [5, 400, 486-488]. Cela rend difficile la tache de proposer des travaux futurs. Neanmoins, il serait interessant de voir comment des cellules epitheliales similaires a celles utilisees par Bonner-Weir et al. [404, 405] se comporteraient en termes de differentiation vers le systeme endocrinien lorsque cultivees sur des surfaces biomimetiques. Toutefois, il convient de noter que les cellules epitheliales ductales isolees du pancreas perdent leur caractere progeniteur lorsqu'elles sont cultivees in vitro [489]. Des experiences supplementaires pourraient egalement inclure l'utilisation d'autres niches biochimiques tels que des proteines completes (par exemple, la fibronectine) plutot que des fragments de peptides tels le RGD. Toutefois, comme le Chapitre

4 l'illustre, il peut etre difficile de reussir a greffer de fa 9 on covalente des proteines de grande 155 taille telles la fibronectine. L'immobilisation sur des surfaces antiadhesives n'est pas toujours garantie.

Le travail experimental du Chapitre 4, qui est intitule « L 'effet de la composition des solutions sur l ’immobilisation de la fibronectine sur des surfaces de carboxymethyl dextrane ainsi que sur la secretion d ’insuline a partir de cellules INS-1 », a ete realise avec l'espoir de proposer une solution a la fonctionnalite reduite des cellules p. L'etude de la composition de la solution lors de l'immobilisation et Tadsorption de proteines n’est generalement pas consideree lors de la conception de surfaces biomimetiques. II a ete constate que non seulement la composition de la solution joue un role dans la reponse {i.e., au niveau de la secretion d'insuline a la stimulation au glucose) des cellules INS-1 cultivees sur des surfaces CMD+fibronectine, mais elle determine le succes de l'immobilisation de la fibronectine sur les surfaces antiadhesives composees de CMD. L'observation que la fibronectine ne se greffe pas sur des surfaces CMD lors de l'utilisation de PBS lx est plutot surprenante, etant donne le fait que sans aucun doute le GRGDS se greffait aux surfaces CMD dans ces conditions comme cela a ete demontre au Chapitre 3 et dans une etude distincte realisee par Sabra et Vermette [454]. Malheureusement, la caracterisation rigoureuse des surfaces CMD par imagerie AFM et des mesures de forces intermoleculaires et de la solution de fibronectine par dichroi'sme circulaire et spectroscopie de correlation de photons n'a pas permis de reveler un mecanisme permettant d'expliquer le greffage de la fibronectine. Certains mecanismes ont ete proposes sans pouvoir les valider (voir Section 4.4.3). Comme il a ete constate que la composition de la solution a egalement un impact ulterieur sur les fonctions des cellules INS-1, il est implicite qu'il existe differentes conformations de la fibronectine immobilisee en fonction de la composition de la solution utilisee lors de l'immobilisation. Des travaux futurs permettant de detecter specifiquement la conformation variable de la proteine immobilisee seraient interessants. Cela pourrait peut-etre etre mesure a l'aide de la microbalance a cristal de quartz (QCM) en mode de dissipation comme cela a ete fait dans le passe pour etudier l'adsorption de la fibronectine et autres proteines a divers substrats [490-492]. Pour identifier plus precisement comment varie la conformation de la fibronectine immobilisee, il pourrait etre envisage de developper un ELISA utilisant des anticorps primaires specifiques a des segments de la fibronectine (voir Figure 2.3) de fa?on similaire a ce qui a ete fait dans l’etude de Vallieres et al. [80].

156 Le travail experimental final de cette these (Chapitre 5), intitule « L ’amelioration de la secretion d ’insuline et diminution de I’apoptose via la co-culture dans la fibrine d ’ilots pore ins jeunes avec des cellules endotheliales micro-vasculaires du foie », visait a remedier a la perte progressive de la capacite des ilots de secreter de l'insuline apres leur isolation. II a ete constate que la co-culture avec des cellules endotheliales d'ilots porcins dans la fibrine a entraine une meilleure secretion d'insuline et une diminution de l'apoptose apres 2 jours. Toutefois, ces avantages ont ete de courte duree car apres 7 jours, il y avait une perte marquee de la secretion d'insuline, une diminution significative du pourcentage de cellules exprimant l'insuline, et une augmentation de l'apoptose cellulaire. Des travaux futurs devraient viser a maintenir les avantages des co-cultures dans un gel de fibrine au-dela de 2 jours. Peut-etre l'ajout de facteurs de croissance comme le fibroblast growth factor (FGF) ou le vascular endothelial growth factor (VEGF) aiderait a augmenter la proliferation et la formation des cellules PLMEC en capillaires [493]. Une etude menee par Sukmana et Vermette a demontre que l'addition de FGF et de VEGF ameliorait la maturation des cellules endotheliales en micro-vaisseaux et ce dans des gels de fibrine [141]. II est possible que la maturation des cellules endotheliales en micro-vaisseaux permettrait d'ameliorer la fonctionnalite des ilots ou meme d'aider a maintenir les avantages de la co-culture au-dela de 2 jours. Tout au long de la culture, le milieu CMRL-1066 supplement^ avec 10% FBS v/v et 1% de penicilline-streptomycine v/v a ete utilise. Les cellules endotheliales PLMEC restaient vivantes dans ces conditions dans la fibrine (voir Figure Cl). Toutefois, dans les cultures monocouches 2D PI-PLMEC, les cellules PLMEC commen 9 aient a changer de morphologie pour finalement se detacher. Par consequent, il serait interessant de voir comment les ilots se comporteraient en utilisant le milieu de culture recommande par le foumisseur (voir la Section 5.3.2). Enfin, la perte significative du pourcentage de cellules-p exprimant l'insuline dans les gels de fibrine peut etre attribute a une mauvaise diffusion des elements nutritifs et a une mauvaise elimination des dechets. En employant une culture dynamique, il serait peut-etre possible d'ameliorer et de maintenir la fonctionnalite des ilots [494,495].

157 Conclusion - English

The experimental work of this thesis aimed to address some of the limitations of islet transplantation as therapy for severe type I diabetes mellitus. Through biomaterials science, it was attempted to address the severe shortage in donor pancreases for islet isolation and the subsequent progressive loss of islet function. Insights on each of the experimental works in this thesis and suggested future work will now be presented.

The first experimental work of Chapter 3 titled, "In Vitro Morphogenesis o f PANC-1 Cells into Islet-Like Aggregates Using RGD-Covered Dextran Derivative Surfaces" attempted to address the shortage of insulin producing islets from donor pancreases available for transplantation. It was found that PANC-1 cells on RGD-CMD surfaces in actual fact produced the most insulin (total and per cell) in comparison to RGE-CMD, CMD, and TCPS controls. While this result is encouraging in the sense that the biomaterials surface (i.e., RGD-CMD) produced the best result, insulin expression in ILAs and in the culture media was not detected until after considerable culture time (i.e., 14 days). This was in contrast to the initial study that reported insulin expression in ILAs after 4 days; insulin secretion was reported in the study but data was not presented [409], This implies that understanding of PANC-1 differentiation into an endocrine pancreatic phenotype is somewhat controversial and incomplete. Parallel controversy also exists regarding the origin of primary islet progenitors cells during stages of not only pancreatic development, but post-natal development as well [5, 400, 486-488]. This makes it somewhat difficult to suggest convincing future work. Nevertheless, it would be interesting to see how primary pancreatic ductal epithelial cells similar to those used by Bonner-Weir et al. fair in terms of possible differentiation into endocrine-like tissue on biomaterials surfaces such as RGD-CMD [404, 405]. However, it should be noted that isolated pancreatic ductal epithelial cells lose their progenitor capacity when in culture [489]. Further experiments could also include the use of other biochemical niches such as complete proteins (e.g., fibronectin) rather than peptide fragments such as RGD. However, as was discovered in the experimental work of Chapter 4, one should be cautious as successful covalent grafting of large proteins such as fibronectin to low-fouling surfaces is not always guaranteed.

158 The Experimental work of Chapter 4 titled, "Solution Composition Impacts Fibronectin Immobilization on Carboxymethyl-Dextran Surfaces and INS-I Insulin Secretion " was performed with the hope of addressing the reduced functionality of p-cells through the identification of an additional factor {i.e., solution composition during protein adsorption) typically not considered when designing biomaterial surfaces. It was found that not only does solution composition play a role in subsequent INS-1 cell response {i.e., insulin secretion to glucose stimulation) to fibronectin-bearing CMD surfaces, but it also impacts whether or not fibronectin can in actual fact be grafted onto low-fouling CMD surfaces. The observation that fibronectin does not graft to low-fouling CMD surfaces when using lx PBS as solution is rather surprising given the fact that GRGDS peptides undoubtedly graft to CMD as was demonstrated in Chapter 3 and in a separate study by Sabra and Vermette [454]. Unfortunately, rigorous characterization of both the CMD surface {i.e., AFM imaging and force measurements) and fibronectin in solution {i.e., circular dichroism and photon correlation spectroscopy) did not reveal a mechanism as to why only certain solutions result in covalent grafting of fibronectin to CMD. This resulted in only hypothesizing the possible mechanisms (see Section 4.4.3). Since it was found that solution composition also impacts subsequent INS-1 cell function, it is implied that there are varying resultant conformations of immobilized fibronectin depending on the solution composition used during immobilization. Future work to more specifically detect varying immobilized fibronectin conformation would be interesting. This could perhaps generally be measured using quartz crystal microbalance (QCM) in dissipation mode as has been done in the past to study the adsorption of fibronectin and other proteins to various substrates [490-492], To more specifically identify varying immobilized fibronectin conformation, perhaps one could develop a custom ELISA using primary antibodies specific to fibronectin ligands such as the GBD, CCBD, and HBD (see Figure 2.3) similar to what has been done in a previous study by Vallieres et al. [80],

The final experimental work of this thesis (Chapter 5) titled, "Co-Culture o f Young Porcine Islets with Liver Microvascular Endothelial Cells Improves Insulin Secretion and Reduces Apoptosis" aimed to address the progressive loss of islet insulin secreting capacity after isolation from the pancreas. It was found that co-culturing isolated porcine islets with endothelial cells in fibrin gel resulted in the best insulin secretion and the lowest levels of apoptosis after 2 days. However, these benefits were short lived as after 7 days, there was a marked loss in insulin secretion, a

159 significant decrease in the percentage of insulin expressing P-cells, and increased cellular apoptosis. Future work should aim to somehow maintain the benefits of PLMEC co-culture in fibrin gel after 2 days of culture. Perhaps the addition of growth factors like basic fibroblast growth factor (FGF) or vascular endothelial growth factor (VEGF) would help benefit the proliferation and formation of PLMEC into capillary-like structures [493]. A study by Sukmana and Vermette found that the addition basic FGF and VEGF improved the maturation of human umbilical vascular endothelial cells in fibrin gel into microvessel-like structures [141]. Perhaps improved maturation of PLMEC in fibrin into microvessel-like structures would further improve islet functionality or even help maintain the benefits of co-culture that were observed after 2 days of culture. Throughout the culture, CMRL-1066 supplemented with 10% FBS v/v and 1% penicillin-streptomycin v/v was used. PLMEC remained alive throughout the culture in fibrin (see Figure Cl). However, in 2D PI-PLMEC monolayer cultures, PLMEC began to change their morphology after 1 day, then began to disappear from the monolayer after 2 days implying death. Therefore, it would be interesting to see how islets perform when using the manufacturer recommended culture medium (see Section 5.3.2). Finally, since the significant loss in the percentage of insulin expressing P-cells in 3D fibrin gel cultures may be attributed to inadequate nutrient delivery and waste removal, employing a dynamic culture could further improve and maintain islet functionality [494,495].

In summary, this thesis demonstrates that biomaterial surfaces and scaffolds have potential towards developing surrogate insulin producing tissues and improving the functionality of isolated islets. It is hoped that perhaps one day, as fundamental scientific knowledge on the subjects presented within this thesis advances, so too will the development of biomaterial surfaces and scaffolds suitable for pancreatic P-cell development and function; to the extent that surgical islet transplantation will be a significantly more readily available option of treatment for patients with severe type I diabetes mellitus.

160 LIST OF REFERENCES

[1] Van Belle, T.L., Coppieters, K.T. and Von Herrath, M.G. (2011). Type 1 Diabetes: Etiology, immunology, and therapeutic strategies. Physiological Reviews, volume 91, p. 79-118. [2] Shapiro, A.M., Lakey, J.R., Ryan, E.A., Korbutt, G.S., Toth, E., Wamock, G.L., Kneteman, N.M. and Rajotte, R.V. (2000). Islet transplantation in seven patients with type 1 diabetes mellitus using a glucocorticoid-free immunosuppressive regimen. New England Journal of Medicine, volume 343, p. 230-238. [3] Merani, S., Toso, C., Emamaullee, J. and Shapiro, A.M.J. (2008). Optimal implantation site for pancreatic islet transplantation. British Journal o f Surgery, volume 95, p. 1449-1461. [4] Rosenberg, L., Wang, R., Paraskevas, S. and Maysinger, D. (1999). Structural and functional changes resulting from islet isolation lead to islet cell death. Surgery, volume 126, p. 393- 398. [5] Bonner-Weir, S. and Sharma, A. (2006). Are there pancreatic progenitor cells from which new islets form after birth? Nature Clinical Practice Endocrinology & Metabolism, volume 2, p. 240-241. [6] Asfari, M., Janjic, D., Meda, P., Li, G., Halban, P.A. and Wollheim, C.B. (1992). Establishment of 2-mercaptoethanol-dependent differentiated insulin-secreting cell lines. Endocrinology, volume 130, p. 167-178. [7] Kasemo, B. (2002). Biological surface science. Surface Science, volume 500, p. 656-677. [8] Bryers, J.D. and Ratner, B.D. (2006). Biomaterials approaches to combating oral biofilms and dental disease. Bio Med Central Oral Health, volume 6 Suppl 1, p. SI 5. [9] Michiardi, A., Aparicio, C., Ratner, B.D., Planell, J.A. and Gil, J. (2007). The influence of surface energy on competitive protein adsorption on oxidized NiTi surfaces. Biomaterials, volume 28, p. 586-594. [10] Ratner, B.D. (2001). Replacing and renewing: synthetic materials, biomimetics, and tissue engineering in implant dentistry. Journal o f Dental Education, volume 65, p. 1340-1347. [11] Danion, A., Arsenault, I. and Vermette, P. (2007). Antibacterial activity of contact lenses bearing surface-immobilized layers of intact liposomes loaded with levofloxacin. Journal o f Pharmaceutical Sciences, volume 96, p. 2350-2363.

161 [12] Danion, A., Brochu, H., Martin, Y. and Vermette, P. (2007). Fabrication and characterization of contact lenses bearing surface-immobilized layers of intact liposomes. Journal o f Biomedical Materials Research. Part A, volume 82, p. 41-51. [13] Danion, A., Doillon, C.J., Giasson, C.J., Djouahra, S., Sauvageau, P., Paradis, R. and Vermette, P. (2007). Biocompatibility and light transmission of liposomal lenses. Optometry and Vision Science, volume 84, p. 954-961. [14] Brochu, H., Polidori, A., Pucci, B. and Vermette, P. (2004). Drug delivery systems using immobilized intact liposomes: a comparative and critical review. Current Drug Delivery, volume 1, p. 299-312. [15] Vermette, P., Griesser, H.J., Kambouris, P. and Meagher, L. (2004). Characterization of surface-immobilized layers of intact liposomes. Biomacromolecules, volume 5, p. 1496- 1502. [16] Vermette, P., Meagher, L., Gagnon, E., Griesser, HJ. and Doillon, C.J. (2002). Immobilized liposome layers for drug delivery applications: inhibition of angiogenesis. Journal of Controlled Release, volume 80, p. 179-195. [17] Bramfeldt, H., Sarazin, P. and Vermette, P. (2007). Characterization, degradation, and mechanical strength of poly(D,L-lactide-co-epsilon-caprolactone)-poly(ethylene glycol)- poly(D,L-lactide-co-epsilon-caprolactone). Journal of Biomedical Materials Research. Part A, volume 83, p. 503-511. [18] Bramfeldt, H. and Vermette, P. (2008). Enhanced smooth muscle cell adhesion and proliferation on protein-modified polycaprolactone-based copolymers. Journal o f Biomedical Materials Research. Part A, volume 88, p. 520-530. [19] Linnes, M.P., Ratner, B.D. and Giachelli, C.M. (2007). A fibrinogen-based precision microporous scaffold for tissue engineering. Biomaterials, volume 28, p. 5298-5306. [20] Das, K., Bose, S. and Bandyopadhyay, A. (2008). TiO(2) nanotubes on Ti: Influence of nanoscale morphology on bone cell-materials interaction. Journal of Biomedical Materials Research. Part A, volume 90, p. 225-237. [21] Li, L., Chen, S., Zheng, J., Ratner, B.D. and Jiang, S. (2005). Protein adsorption on oligo(ethylene glycol)-terminated alkanethiolate self-assembled monolayers: The molecular basis for nonfouling behavior. Journal o f Physical Chemistry B, volume 109, p. 2934-2941.

162 [22] Liu, L., Chen, G., Chao, T., Ratner, B.D., Sage, E.H. and Jiang, S. (2008). Reduced foreign body reaction to implanted biomaterials by surface treatment with oriented osteopontin. Journal o f Biomaterials Science, Polymer Edition, volume 19, p. 821-835. [23] Monchaux, E. and Vermette, P. (2007). Bioactive microarrays immobilized on low-fouling surfaces to study specific endothelial cell adhesion. Biomacromolecules, volume 8, p. 3668- 3673. [24] Monchaux, E. and Vermette, P. (2007). Development of dextran-derivative arrays to identify physicochemical properties involved in biofouling from semm. Langmuir, volume 23, p. 3290-3297. [25] Monchaux, E. and Vermette, P. (2008). Cell adhesion resistance mechanisms using arrays of dextran-derivative layers. Journal o f Biomedical Materials Research, Part A, volume 85 A, p. 1052-1063. [26] Barbosa, J.N., Barbosa, M.A. and Aguas, A.P. (2003). Adhesion of human leukocytes to biomaterials: an in vitro study using alkanethiolate monolayers with different chemically functionalized surfaces. Journal o f Biomedical Materials Research. Part A, volume 65, p. 429-434. [27] Faucheux, N., Schweiss, R., Lutzow, K., Wemer, C. and Groth, T. (2004). Self-assembled monolayers with different terminating groups as model substrates for cell adhesion studies. Biomaterials, volume 25, p. 2721-2730. [28] Garcia, A.J., Vega, M.D. and Boettiger, D. (1999). Modulation of cell proliferation and differentiation through substrate-dependent changes in fibronectin conformation. Molecular Biology o f the Cell, volume 10, p. 785-798. [29] Kapur, R. and Rudolph, A.S. (1998). Cellular and cytoskeleton morphology and strength of adhesion of cells on self-assembled monolayers of organosilanes. Experimental Cell Research, volume 244, p. 275-285. [30] Keselowsky, B.G., Collard, D.M. and Garcia, A.J. (2003). Surface chemistry modulates fibronectin conformation and directs integrin binding and specificity to control cell adhesion. Journal o f Biomedical Materials Research. Part A, volume 66, p. 247-259. [31] McClary, K.B., Ugarova, T. and Grainger, D.W. (2000). Modulating fibroblast adhesion, spreading, and proliferation using self-assembled monolayer films of alkylthiolates on gold. Journal o f Biomedical Materials Research, volume 50, p. 428-439.

163 [32] Spargo, B.J., Testoff, M.A., Nielsen, T.B., Stenger, D.A., Hickman, J.J. and Rudolph, A.S. (1994). Spatially controlled adhesion, spreading, and differentiation of endothelial cells on self-assembled molecular monolayers. Proceedings of the National Academy of Sciences o f the United States of America, volume 91, p. 11070-11074. [33] Steele, J.G., Dalton, B.A., Johnson, G. and Underwood, P.A. (1995). Adsorption of fibronectin and vitronectin onto Primaria and tissue culture polystyrene and relationship to the mechanism of initial attachment of human vein endothelial cells and BHK-21 fibroblasts. Biomaterials, volume 16, p. 1057-1067. [34] Tegoulia, V.A. and Cooper, S.L. (2000). Leukocyte adhesion on model surfaces under flow: effects of surface chemistry, protein adsorption, and shear rate. Journal of Biomedical Materials Research, volume 50, p. 291-301. [35] Webb, K., Hlady, V. and Tresco, P.A. (1998). Relative importance of surface wettability and charged functional groups on NIH 3T3 fibroblast attachment, spreading, and cytoskeletal organization. Journal o f Biomedical Materials Research, volume 41, p. 422-430. [36] Webb, K., Hlady, V. and Tresco, P.A. (2000). Relationships among cell attachment, spreading, cytoskeletal organization, and migration rate for anchorage-dependent cells on model surfaces. Journal o f Biomedical Materials Research, volume 49, p. 362-368. [37] Nerem, R.M. (2009) Cell-based therapies: from benchtop research to replacement, repair, and regeneration. 27th Annual Conference o f the Canadian Biomaterials Society. Laval University, Quebec City, Canada. [38] Singer, S.J. and Nicolson, G.L. (1972). The fluid mosaic model of the structure of cell membranes. Science, volume 175, p. 720-731. [39] Rothstein, A. (1978). The cell membrane - a short historical perspective. Current Topics in Membranes and Transport, volume 11, p. 1-13. [40] Chan, P.Y., Lawrence, M.B., Dustin, M.L., Ferguson, L.M., Golan, D.E. and Springer, T.A. (1991). Influence of Receptor Lateral Mobility on Adhesion Strengthening between Membranes Containing Lfa-3 and Cd2. Journal o f Cell Biology, volume 115, p. 245-255. [41] Dustin, M.L., Ferguson, L.M., Chan, P.Y., Springer, T.A. and Golan, D.E. (1996). Visualization o f CD2 interaction with LFA-3 and determination of the two-dimensional dissociation constant for adhesion receptors in a contact area. Journal of Cell Biology, volume 132, p. 465-474.

164 [42] Lodish, H., Berk, A., Matsudaira, P., Kaiser, C.A., Krieger, M. and Scott, M.P. (2004) Molecular Cell Biology, 5th Edition, W.H. Freeman & Company, New York, United States, 973 p. [43] Becker, W.M., Kleinsmith, L.J. and Hardin, J. (2006) The World o f the Cell, 6th Edition, Pearson Education Inc., San Francisco, United States, 944 p. [44] Evans, M.D. and Steele, J.G. (1998). Polymer surface chemistry and a novel attachment mechanism in comeal epithelial cells. Journal o f Biomedical Materials Research, volume 40, p. 621-630. [45] Petit, V. and Thiery, J.P. (2000). Focal adhesions: structure and dynamics. Biologie Cellulaire, volume 92, p. 477-494. [46] Zamir, E. and Geiger, B. (2001). Components of cell-matrix adhesions. Journal of Cell Science, volume 114, p. 3577-3579. [47] Hynes, R.O. (1987). Integrins: a family of cell surface receptors. Cell, volume 48, p. 549- 554. [48] Hynes, R.O. (1992). Integrins: versatility, modulation, and signaling in cell adhesion. Cell, volume 69, p. 11-25. [49] Pierres, A., Benoliel, A. and Bongrand, P., (2000). Cell - cell interactions. In: Baszkin, A., Norde, W. (Eds.) Physical Chemistry of Biological Interfaces, Marcel Dekker Inc., Basel, Switzerland, p. 459 - 522. [50] Faucheux, N., Tzoneva, R., Nagel, M.D. and Groth, T. (2006). The dependence of fibrillar adhesions in human fibroblasts on substratum chemistry. Biomaterials, volume 27, p. 234- 245. [51] Zamir, E. and Geiger, B. (2001). Molecular complexity and dynamics of cell-matrix adhesions. Journal o f Cell Science, volume 114, p. 3583-3590. [52] Zamir, E., Katz, M., Posen, Y., Erez, N., Yamada, K.M., Katz, B.-Z., Lin, S., Lin, D.C., Bershadsky, A., Kam, Z. and Geiger, B. (2000). Dynamics and segregation of cell-matrix adhesions in cultured fibroblasts. Nature Cell Biology, volume 2, p. 191-196. [53] Pankov, R., Cukierman, E., Katz, B.-Z., Matsumoto, K., Lin, D.C., Lin, S., Hahn, C. and Yamada, K.M. (2000). Integrin dynamics and matrix assembly. The Journal o f Cell Biology, volume 148, p. 1075-1090.

165 [54] Avnur, Z. and Geiger, B. (1981). The removal of extracellular fibronectin from areas of cell-substrate contact. Cell, volume 25, p. 121-132. [55] Katz, B.-Z., Zamir, E., Bershadsky, A., Kam, Z., Yamada, K.M. and Geiger, B. (2000). Physical state of the extracellular matrix regulates the structure and molecular composition of cell-matrix adhesions. Molecular Biology o f the Cell, volume 11, p. 1047-1060. [56] Zimerman, B., Volberg, T. and Geiger, B. (2004). Early molecular events in the assembly of the focal adhesion-stress fiber complex during fibroblast spreading. Cell Motility and the Cytoskeleton, volume 58, p. 143-159. [57] Singer, I.I., Scott, S., Kawka, D.W., Kazazis, D.M., Gailit, J. and Ruoslahti, E. (1988). Cell surface distribution of fibronectin and vitronectin receptors depends on substrate composition and extracellular matrix accumulation. The Journal o f Cell Biology, volume 106, p. 2171- 2182. [58] Humphries, J.D., Byron, A. and Humphries, M.J. (2006). Integrin ligands at a glance. Journal o f Cell Science, volume 119, p. 3901-3903. [59] Hynes, R.O. (2002). Integrins: bidirectional, allosteric signaling machines. Cell, volume 110, p. 673-687. [60] Hynes, R.O. (1990) Fibronectins, 1st Edition, Springer-Verlag, New York, United States, 546 p. [61] Ni, H., Yuen, P.S., Papalia, J.M., Trevithick, J.E., Sakai, T., Fassler, R., Hynes, R.O. and Wagner, D.D. (2003). Plasma fibronectin promotes thrombus growth and stability in injured arterioles. Proceedings o f the National Academy o f Sciences o f the United States o f America, volume 100, p. 2415-2419. [62] Sakai, T., Johnson, K.J., Murozono, M., Sakai, K., Magnuson, M.A., Wieloch, T., Cronberg, T., Isshiki, A., Erickson, H.P. and Fassler, R. (2001). Plasma fibronectin supports neuronal survival and reduces brain injury following transient focal cerebral ischemia but is not essential for skin-wound healing and hemostasis. Nature Medicine, volume 7, p. 324-330. [63] Tamkun, J.W. and Hynes, R.O. (1983). Plasma fibronectin is synthesized and secreted by hepatocytes. Journal o f Biological Chemistry, volume 258, p. 4641-4647. [64] Yi, M., Sakai, T., Fassler, R. and Ruoslahti, E. (2003). Antiangiogenic proteins require plasma fibronectin or vitronectin for in vivo activity. Proceedings of the National Academy o f Sciences o f the United States o f America, volume 100, p. 11435-11438.

166 [65] Hayashi, M. and Yamada, K.M. (1981). Differences in domain structures between plasma and cellular fibronectins. Journal o f Biological Chemistry, volume 256, p. 11292-11300. [66] Yamada, K.M. (1983). Cell surface interactions with extracellular materials. Annual Review o f Biochemistry, volume 52, p. 761-799. [67] Dalton, B.A., McFarland, C.D., Underwood, P.A. and Steele, J.G. (1995). Role of the heparin binding domain of fibronectin in attachment and spreading of human bone-derived cells. Journal o f Cell Science, volume 108 p. 2083-2092. [68] Cutler, S.M. and Garcia, A.J. (2003). Engineering cell adhesive surfaces that direct integrin alpha5betal binding using a recombinant fragment of fibronectin. Biomaterials, volume 24, p. 1759-1770. [69] Ochsenhirt, S.E., Kokkoli, E., McCarthy, J.B. and Tirrell, M. (2006). Effect of RGD secondary structure and the synergy site PHSRN on cell adhesion, spreading and specific integrin engagement. Biomaterials, volume 27, p. 3863-3874. [70] Kang, W., Park, S. and Jang, J.H. (2008). Kinetic and functional analysis of the heparin- binding domain of fibronectin. Biotechnology Letters, volume 30, p. 55-59. [71] Wijelath, E.S., Rahman, S., Namekata, M., Murray, J., Nishimura, T., Mostafavi-Pour, Z., Patel, Y., Suda, Y., Humphries, M.J. and Sobel, M. (2006). Heparin-II domain of fibronectin is a vascular endothelial growth factor-binding domain: enhancement of VEGF biological activity by a singular growth factor/matrix protein synergism. Circulation Research, volume 99, p .853-860. [72] Akimov, S.S. and Belkin, A.M. (2001). Cell surface tissue transglutaminase is involved in adhesion and migration of monocytic cells on fibronectin. Blood, volume 98, p. 1567-1576. [73] Hynes, R.O. and Yamada, K.M. (1982). Fibronectins: multifunctional modular glycoproteins. Journal of Cell Biology, volume 95, p. 369-377. [74] Akimov, S.S. and Belkin, A.M. (2001). Cell-surface transglutaminase promotes fibronectin assembly via interaction with the gelatin-binding domain of fibronectin: a role in TGFbeta- dependent matrix deposition. Journal o f Cell Science, volume 114, p. 2989-3000. [75] Amrani, D.L., Homandberg, G.A., Tooney, N.M., Wolfenstein-Todel, C. and Mosesson, M.W. (1983). Separation and analysis of the major forms of plasma fibronectin. Biochimica et Biophysica Acta, volume 748, p. 308-320.

167 [76] Ruoslahti, E., Hayman, E.G., Pierschbacher, M. and Engvall, E. (1982). Fibronectin: purification, immunochemical properties, and biological activities. Methods in Enzymology, volume 82 Pt A, p. 803-831. [77] Speziale, P., Visai, L., Rindi, S. and Di Poto, A. (2008). Purification of human plasma fibronectin using immobilized gelatin and Arg affinity chromatography. Nature Protocols, volume 3, p. 525-533. [78] Johnson, K.J., Sage, H., Briscoe, G. and Erickson, H.P. (1999). The compact conformation of fibronectin is determined by intramolecular ionic interactions. Journal of Biological Chemistry, volume 274, p. 15473-15479. [79] Vallieres, K., Chevallier, P., Sarra-Boumet, C., Turgeon, S. and Laroche, G. (2007). AFM imaging of immobilized fibronectin: does the surface conjugation scheme affect the protein orientation/conformation? Langmuir, volume 23, p. 9745-9751. [80] Vallieres, K., Petitclerc, E. and Laroche, G. (2007). Covalent grafting of fibronectin onto plasma-treated PTFE: influence of the conjugation strategy on fibronectin biological activity. Macromolecular Bioscience, volume 7, p. 738-745. [81] Malmsten, M. (1998). Formation of adsorbed protein layers. Journal of Colloid and Interface Science, volume 207, p. 186-199. [82] Israelachvili, J. (2005). Differences between non-specific and bio-specific, and between equilibrium and non-equilibrium, interactions in biological systems. Quarterly Reviews of Biophysics, volume 38, p. 331-337. [83] Lyklema, J., (2000). Interfacial thermodynamics with special reference to biological systems. In: Baszkin, A., Norde, W. (Eds.) Physical Chemistry of Biological Interfaces, Marcel Dekker Inc., Basel, Switzerland, p. 1 - 48. [84] Israelachvili, J.N. (1991) Intermolecular and Surface Forces, 2nd Edition, Academic Press, San Diego, United States, 450 p. [85] Hunter, R.J. (2001) Foundations of Colloid Science, 2nd Edition, Oxford University Press, New York, United States, 806 p. [86] Leckband, D. and Israelachvili, J. (2001). Intermolecular forces in biology. Quarterly Reviews of Biophysics, volume 34, p. 105-267. [87] Baszkin, A. and Norde, W. (2000) Physical Chemistry o f Biological Interfaces, 1st Edition, Marcel Dekker Inc., Basel, Switzerland, 848 p.

168 [88] Liu, L., Ratner, B.D., Sage, E.H. and Jiang, S. (2007). Endothelial cell migration on surface-density gradients of fibronectin, VEGF, or both proteins. Langmuir, volume 23, p. 11168-11173. [89] Gauvreau, V., Chevallier, P., Vallieres, K., Petitclerc, E., Gaudreault, R.C. and Laroche, G. (2004). Engineering surfaces for bioconjugation: developing strategies and quantifying the extent of the reactions. Bioconjugate Chemistry, volume 15, p. 1146-1156. [90] Zhang, Z., Yoo, R., Wells, M., Beebe, T.P., Jr., Biran, R. and Tresco, P. (2005). Neurite outgrowth on well-characterized surfaces: preparation and characterization of chemically and spatially controlled fibronectin and RGD substrates with good bioactivity. Biomaterials, volume 26, p. 47-61. [91] Griffiths, D.J. (1999) Introduction to electrodynamics, 3rd Edition, Prentice Hall, Upper Saddle River, United States, 576 p. [92] Jackson, J.D. (1998) Classical electrodynamics, 3rd Edition, John Wiley & Sons, New York, United States, 808 p. [93] Finke, B., Luethen, F., Schroeder, K., Mueller, P.D., Bergemann, C., Frant, M., Ohl, A. and Nebe, B.J. (2007). The effect of positively charged plasma polymerization on initial osteoblastic focal adhesion on titanium surfaces. Biomaterials, volume 28, p. 4521-4534. [94] Kleijn, J.M. and van Leeuwen, H.P., (2000). Electrostatic and electrodynamic properties of biological interphases. In: Baszkin, A., Norde, W. (Eds.) Physical Chemistry o f Biological Interfaces, Marcel Dekker Inc., Basel, Switzerland, p. 49 - 84. [95] van der Waals, J.D., On the continuity of the gas and liquid state, in, Leiden University, Leiden, Netherlands, 1873, pp. 301. [96] Keesom, W.H. (1921). van der Waals attractive forces. Physikalische Zeitschrift, volume 22, p. 129-141. [97] Debye, P. (1920). van der Waals’ cohesion forces. Physikalische Zeitschrift, volume 21, p. 178-187. [98] London, F. (1930). Theory and systematics of molecular forces. Physikalische Zeitschrift, volume 63, p. 245-279. [99] Israelachvili, J.N. (1973). Van der Waals forces in biological systems. Quarterly Reviews of Biophysics, volume 6, p. 341-387.

169 [100] Hamaker, H.C. (1937). The London-van der Waals attraction between spherical particles. Physica, volume 4, p. 1058-1072. [101] Lifshitz, E.M. (1956). The theory of molecular attraction forces between solid bodies. Soviet Physics, Journal o f Experimental and Theoretical Physics, volume 2, p. 73-83. [102] Deijaguin, B. and Landau, L.D. (1941). Theoiy of the stability of strongly charged lyophobic soils and of the adhension of stronghly charged particles in solutions of electrolytes. Acta Physiochimica, Union o f Soviet Socialist Republics, volume 14, p. 633- 662. [103] Verwey, E.G.W. and Overbeek, J.T.G. (1948) Theory of the stability of lyophobic colloids, 1st Edition, Elsevier, Amsterdam, Netherlands, 218 p. [104] Frank, H.S. and Evans, M.W. (1945). Free volume and entropy in condensed systems III. Entropy in binary liquid mixtures; Partial molal entropy in dilute solutions; Structure and thermodynamics in aqueous electrolytes. Journal of Chemical Physics, volume 13, p. 507- 532. [105] Meyer, E.E., Rosenberg, K J. and Israelachvili, J. (2006). Recent progress in understanding hydrophobic interactions. Proceedings of the National Academy o f Sciences of the United States o f America, volume 103, p. 15739-15746. [106] Claesson, P.M., Blom, C.E., Herder, P.C. and Ninham, B.W. (1986). Interactions between water—stable hydrophobic Langmuir—Blodgett monolayers on mica. Journal of Colloid and Interface Science, volume 114, p. 234-242. [107] Parker, J.L. and Claesson, P.M. (1994). Forces between hydrophobic silanated glass surfaces. Langmuir, volume 10, p. 635-639. [108] Pashley, R.M., McGuiggan, P.M., Ninham, B.W. and Evans, D.F. (1985). Attractive forces between uncharged hydrophobic surfaces: direct measurements in aqueous solution. Science, volume 229, p. 1088-1089. [109] Israelachvili, J. and Pashley, R. (1982). The hydrophobic interaction is long range, decaying exponentially with distance. Nature, volume 300, p. 341-342. [110] Israelachvili, J. and Wennerstrom, H. (1996). Role of hydration and water structure in biological and colloidal interactions. Nature, volume 379, p. 219-225.

170 [111] Horbett, T.A. and Brash, J.L. (1987) Proteins at interfaces, physiochemical and biochemical studies, volume 343, American Chemical Society, Washington, D.C., United States, 706 p. [112] Horbett, T.A. and Brash, J.L. (1995) Proteins at Interfaces II: Fundamentals and Applications, volume 602, American Chemical Society, Washignton, D.C., United States, 561 p. [113] Horbett, T.A. and Hoffman, A.S., (1975). Bovine plasma protein adsorption to radiation grafted hydrogels based hydroxyethylmethacrylate and N-vinyl-pyrrolidone. In: Baier, R. (Ed.) Applied Chemistry at Protein Interfaces, American Chemical Society, Washington, D.C., United States, p. 230-254. [114] Haynes, C.A. and Norde, W. (1994). Globular proteins at solid/liquid interfaces. Colloids and Surfaces B: Biointerfaces, volume 2, p. 517-566. [115] Helm, C.A., Knoll, W. and Israelachvili, J.N. (1991). Measurement of ligand-receptor interactions. Proceedings of the National Academy of Sciences of the United States of America, volume 88, p. 8169-8173. [116] Tummino, P.J. and Copeland, R.A. (2008). Residence time of receptor-ligand complexes and its effect on biological function. Biochemistry, volume 47, p. 5481-5492. [117] Bongrand, P. (1999). Ligand-receptor interactions. Reports on Progress in Physics, volume 62, p. 921-968. [118] Simson, R. and Sackmann, E., (2000). Mimicking physics of cell adhesion. In: Baszkin, A., Norde, W. (Eds.) Physical Chemistry o f Biological Interfaces, Marcel Dekker Inc., Basel, Switzerland, p. 401-430. [119] Helffich, W. (1978). Steric interactions of fluid membranes in multilayer systems. Zeitschriftfuer Naturforschung, volume 33a, p. 305-315. [120] Leckband, D. (2000). Measuring the forces that control protein interactions. Annual Review of Biophysics and Biomolecular Structure, volume 29, p. 1-26. [121] Israelachvili, J. and Adams, G. (1978). Measurement of forces between two mica surfaces in aqueous electrolyte solutions in the range 0-100 nm. Journal of the American Chemical Society, Faraday Transactions I, volume 74, p. 975-1001. [122] Tabor, D. and Winterton, R.H. (1968). Surface forces: direct measurement of normal and retarded van der Waals forces. Nature, volume 219, p. 1120-1121.

171 [123] Leckband, D. (1995). The surface apparatus--a tool for probing molecular protein interactions. Nature, volume 376, p. 617-618. [124] Leckband, D. (2001). Force as a probe of membrane protein structure and function. Current Opinion in Structural Biology, volume 11, p. 433-439. [125] Binnig, G., Quate, C.F. and Gerber, C. (1986). Atomic force microscope. Physical Review Letters, volume 56, p. 930-933. [126] Kasas, S., Thomson, N.H., Smith, B.L., Hansma, P.K., Miklossy, J. and Hansma, H.G. (1997). Biological applications of the AFM: from single molecules to organs. International Journal o f Imaging Systems and Technology, volume 8, p. 151-161. [127] Hoh, J.H., Cleveland, J.P., Prater, C.B., Revel, J.P. and Hansma, P.K. (1992). Quantized adhesion detected with the atomic force microscope. Journal o f the American Chemical Society, volume 114, p. 4917-4918. [128] Griffith, L.G. and Swartz, M.A. (2006). Capturing complex 3D tissue physiology in vitro. Nature Reviews Molecular Cell Biology, volume 7, p. 211-224. [129] Kessel, R.G. (1998) Basic Medical Histology: the Biology o f Cells, Tissues, and Organs, 1st Edition, Oxford University Press, New York, United States, 624 p. [130] Stevens, A. and Lowe, J.S. (2007) Human Histology, 3rd Edition, Mosby, Toronto, Canada, 464 p. [131] Weiss, P. (1959). Cellular dynamics. Reviews o f Modem Physics, volume 31, p. 11-20. [132] Lewis, W.H. and Lewis, M.R., (1924). Behavior of cells in tissue cultures. In: Cowdry, E.V. (Ed.) General Cytology, University of Chicago Press, Chicago, United States, p. 384- 447. [133] Grinnell, F. (2003). Fibroblast biology in three-dimensional collagen matrices. Trends in Cell Biology, volume 13, p. 264-269. [134] Grinnell, F., Ho, C.H., Tamariz, E., Lee, D.J. and Skuta, G. (2003). Dendritic fibroblasts in three-dimensional collagen matrices. Molecular Biology o f the Cell, volume 14, p. 384- 395. [135] Brown, R.A., Sethi, K.K., Gwanmesia, I., Raemdonck, D., Eastwood, M. and Mudera, V. (2002). Enhanced fibroblast contraction of 3D collagen lattices and integrin expression by TGF-betal and -beta3: mechanoregulatory growth factors? Experimental Cell Research, volume 274, p. 310-322.

172 [136] Grinnell, F. (2000). Fibroblast-collagen-matrix contraction: growth-factor signalling and mechanical loading. Trends in Cell Biology, volume 10, p. 362-365. [137] Friedl, P. and Brocker, E.B. (2000). The biology of cell locomotion within three- dimensional extracellular matrix. Cellular and Molecular Life Sciences, volume 57, p. 41-64. [138] Birkedal-Hansen, H. (1995). Proteolytic remodeling of extracellular matrix. Current Opinion in Cell Biology, volume 7, p. 728-735. [139] Shapiro, S.D. (1998). Matrix metalloproteinase degradation of extracellular matrix: biological consequences. Current Opinion in Cell Biology, volume 10, p. 602-608. [140] Sukmana, I. and Vermette, P. (2010). Polymer fibers as contact guidance to orient microvascularization in a 3D environment. Journal o f Biomedical Materials Research Part A, volume 92A, p. 1587-1597. [141] Sukmana, I. and Vermette, P. (2010). The effects of co-culture with fibroblasts and angiogenic growth factors on microvascular maturation and multi-cellular lumen formation in HUVEC-oriented polymer fibre constructs. Biomaterials, volume 31, p. 5091-5099. [142] Yang, B., Cao, D.J., Sainz, I., Colman, R.W. and Guo, Y.L. (2004). Different roles of ERK and p38 MAP kinases during tube formation from endothelial cells cultured in 3- dimensional collagen matrices. Journal o f Cellular Physiology, volume 200, p. 360-369. [143] Jechlinger, M., Grunert, S. and Beug, H. (2002). Mechanisms in epithelial plasticity and metastasis: insights from 3D cultures and expression profiling. Journal of Mammary Gland Biology and Neoplasia, volume 7, p. 415-432. [144] Niemann, C., Brinkmann, V., Spitzer, E., Hartmann, G., Sachs, M., Naundorf, H. and Birchmeier, W. (1998). Reconstitution of mammary gland development in vitro: requirement of c-met and c-erbB2 signaling for branching and alveolar morphogenesis. Journal o f Cell Biology, volume 143, p. 533-545. [145] Oft, M., Peli, J., Rudaz, C., Schwarz, H., Beug, H. and Reichmann, E. (1996). TGF-betal and Ha-Ras collaborate in modulating the phenotypic plasticity and invasiveness of epithelial tumor cells. Genes & Development, volume 10, p. 2462-2477. [146] Kunz-Schughart, L.A., Groebe, K. and Mueller-Klieser, W. (1996). Three-dimensional cell culture induces novel proliferative and metabolic alterations associated with oncogenic transformation. International Journal o f Cancer, volume 66, p. 578-586.

173 [147] Yu, J.L., Coomber, B.L. and Kerbel, R.S. (2002). A paradigm for therapy-induced microenvironmental changes in solid tumors leading to drug resistance. Differentiation, volume 70, p. 599-609. [148] Santini, M.T., Rainaldi, G. and Indovina, P.L. (2000). Apoptosis, cell adhesion and the extracellular matrix in the three-dimensional growth of multicellular tumor spheroids. Critical Reviews in Oncology/Hematology, volume 36, p. 75-87. [149] Zahir, N. and Weaver, V.M. (2004). Death in the third dimension: apoptosis regulation and tissue architecture. Current Opinion in Genetics & Development, volume 14, p. 71-80. [150] Zakeri, Z. and Lockshin, R.A. (2002). Cell death during development. Journal o f Immunological Methods, volume 265, p. 3-20. [151] Debnath, J., Mills, K.R., Collins, N.L., Reginato, M.J., Muthuswamy, S.K. and Brugge, J.S. (2002). The role of apoptosis in creating and maintaining luminal space within normal and oncogene-expressing mammary acini. Cell, volume 111, p. 29-40. [152] Goh, K.L., Yang, J.T. and Hynes, R.O. (1997). Mesodermal defects and cranial neural crest apoptosis in alpha5 integrin-null embryos. Development, volume 124, p. 4309-4319. [153] Igney, F.H. and Krammer, P.H. (2002). Death and anti-death: tumour resistance to apoptosis. Nature Reviews Cancer, volume 2, p. 277-288. [154] Chaqour, B. and Goppelt-Struebe, M. (2006). Mechanical regulation of the Cyr61/CCN1 and CTGF/CCN2 proteins. Federation o f European Biochemical Societies Journal, volume 273, p. 3639-3649. [155] Putnam, A.J., Cunningham, J.J., Dennis, R.G., Linderman, J.J. and Mooney, D.J. (1998). Microtubule assembly is regulated by externally applied strain in cultured smooth muscle cells. Journal o f Cell Science, volume 111, p. 3379-3387. [156] Yap, B. and Kamm, R.D. (2005). Mechanical deformation of neutrophils into narrow channels induces pseudopod projection and changes in biomechanical properties. Journal o f Applied Physiology, volume 98, p. 1930-1939. [157] Georges, P.C. and Janmey, P.A. (2005). Cell type-specific response to growth on soft materials. Journal o f Applied Physiology, volume 98, p. 1547-1553. [158] Wang, J.H., Thampatty, B.P., Lin, J.S. and Im, H J. (2007). Mechanoregulation of gene expression in fibroblasts Gene, volume 391, p. 1-15.

174 [159] McBeath, R., Pirone, D.M., Nelson, C.M., Bhadriraju, K. and Chen, C.S. (2004). Cell shape, cytoskeletal tension, and RhoA regulate stem cell lineage commitment. Developmental Cell, volume 6, p. 483-495. [160] Sackmann, E. and Bruinsma, R.F. (2002). Cell adhesion as wetting transition? ChemPhysChem, volume 3, p. 262-269. [161] Martin, Y. and Vermette, P. (2005). Bioreactors for tissue mass culture: design, characterization, and recent advances. Biomaterials, volume 26, p. 7481-7503. [162] Pizzo, A.M., Kokini, K., Vaughn, L.C., Waisner, B.Z. and Voytik-Harbin, S.L. (2005). Extracellular matrix (ECM) microstructural composition regulates local cell-ECM biomechanics and fundamental fibroblast behavior: a multidimensional perspective. Journal o f Applied Physiology, volume 98, p. 1909-1921. [163] Birukov, K.G., Shirinsky, V.P., Stepanova, O.V., Tkachuk, V.A., Hahn, A.W., Resink, T.J. and Smirnov, V.N. (1995). Stretch affects phenotype and proliferation of vascular smooth muscle cells. Molecular and Cellular Biochemistry, volume 144, p. 131-139. [164] Stegemann, J.P. and Nerem, R.M. (2003). Altered response of vascular smooth muscle cells to exogenous biochemical stimulation in two- and three-dimensional culture. Experimental Cell Research, volume 283, p. 146-155. [165] Stegemann, J.P., Hong, H. and Nerem, R.M. (2005). Mechanical, biochemical, and extracellular matrix effects on vascular smooth muscle cell phenotype. Journal of Applied Physiology, volume 98, p. 2321-2327. [166] Engler, A.J., Griffin, M.A., Sen, S., Bonnemann, C.G., Sweeney, H.L. and Discher, D.E. (2004). Myotubes differentiate optimally on substrates with tissue-like stiffness: pathological implications for soft or stiff microenvironments. Journal o f Cell Biology, volume 166, p. 877-887. [167] Selden, C., Shariat, A., McCloskey, P., Ryder, T., Roberts, E. and Hodgson, H. (1999). Three-dimensional in vitro cell culture leads to a marked upregulation of cell function in human hepatocyte cell lines~an important tool for the development of a bioartificial liver machine. Annals of the New York Academy o f Sciences, volume 875, p. 353-363. [168] Teng, Y.D., Lavik, E.B., Qu, X., Park, K.I., Ourednik, J., Zurakowski, D., Langer, R. and Snyder, E.Y. (2002). Functional recovery following traumatic spinal cord injury mediated by

175 a unique polymer scaffold seeded with neural stem cells. Proceedings of the National Academy o f Sciences o f the United States o f America, volume 99, p. 3024-3029. [169] Leclerc, E., David, B., Griscom, L., Lepioufle, B., Fujii, T., Layrolle, P. and Legallaisa, C. (2006). Study of osteoblastic cells in a micro fluidic environment. Biomaterials, volume 27, p. 586-595. [170] Bruinsma, R. and Sackmann, E. (2001). Bioadhesion and the dewetting transition. Comptes Rendus de I Academie des Sciences Series IVPhysics, volume 2, p. 803-815. [171] Sackmann, E. (2006). Thermo-elasticity and adhesion as regulators of cell membrane architecture and function. Journal o f Physics: Condensed Matter, volume 18, p. R785-R825. [172] Sackmann, E. and Goennenwein, S. (2006). Cell adhesion as dynamic interplay of lock- and-key, generic and elastic forces. Progress o f Theoretical Physics, volume 165, p. 78-99. [173] Buckingham, E. (1914). On physically similar systems; illustrations of the use of dimensional equations. Physical Review, volume 4, p. 345-376. [174] Buckingham, E. (1915). Model experiments and the forms of empirical equations. Transactions o f the American Society o f Mechanical Engineers, volume 37, p. 263-296. [175] Bausch, A.R., Ziemann, F., Boulbitch, A.A., Jacobson, K. and Sackmann, E. (1998). Local measurements of viscoelastic parameters of adherent cell surfaces by magnetic bead microrheometry. Biophysical Journal, volume 75, p. 2038-2049. [176] Wiegand, G., Neumaier, K.R. and Sackmann, E. (1998). Microinterferometry: three- dimensional reconstruction of surface microtopography for thin-film and wetting studies by reflection interference contrast microscopy (RICM). Applied Optics, volume 37, p. 6892- 6905. [177] Nardi, J., Bruinsma, R. and Sackmann, E. (1998). Adhesion-induced reorganization of charged fluid membranes. Physical Review E, volume 58, p. 6340-6354. [178] Feneberg, W., Aepfelbacher, M. and Sackmann, E. (2004). Microviscoelasticity of the apical cell surface of human umbilical vein endothelial cells (HUVEC) within confluent monolayers. Biophysical Journal, volume 87, p. 1338-1350. [179] Limozin, L., Roth, A. and Sackmann, E. (2005). Microviscoelastic moduli of biomimetic cell envelopes. Physical Review Letters, volume 95, p. 178101.

176 [180] Uhde, J., Feneberg, W., Ter-Oganessian, N., Sackmann, E. and Boulbitch, A. (2005). Osmotic force-controlled microrheometry of entangled actin networks. Physical Review Letters, volume 94, p. 198102. [181] Uhde, J., Keller, M., Sackmann, E., Parmeggiani, A. and Frey, E. (2004). Internal motility in stiffening actin-myosin networks. Physical Review Letters, volume 93, p. 268101. [182] Uhde, J., Ter-Oganessian, N., Pink, D.A., Sackmann, E. and Boulbitch, A. (2005). Viscoelasticity of entangled actin networks studied by long-pulse magnetic bead microrheometry. Physical Review E, volume 72, p. 061916. [183] Forgacs, G., Newman, S.A., Hinner, B., Maier, C.W. and Sackmann, E. (2003). Assembly of collagen matrices as a phase transition revealed by structural and rheologic studies. Biophysical Journal, volume 84, p. 1272-1280. [184] Ladam, G., Vonna, L. and Sackmann, E. (2003). Micromechanics of surface-grafted hyaluronic acid gels. Journal o f Physical Chemistry B, volume 107, p. 8965-8971. [185] Sengupta, K., Schilling, J., Marx, S., Fischer, M., Bacher, A. and Sackmann, E. (2003). Mimicking tissue surfaces by supported membrane coupled ultrathin layer of hyaluronic acid. Langmuir, volume 19, p. 1775-1781. [186] Picart, C., Sengupta, K., Schilling, J., Maurstad, G., Ladam, G., Bausch, A.R. and Sackmann, E. (2004). Microinterferometric study of the structure, interfacial potential, and viscoelastic properties of polyelectrolyte multilayer films on a planar substrate. Journal o f Physical Chemistry B, volume 108, p. 7196-7205. [187] Smith, A.S. and Sackmann, E. (2009). Progress in mimetic studies of cell adhesion and the mechanosensing. ChemPhysChem, volume 10, p. 66-78. [188] Trepat, X., Lenormand, G. and Fredberg, J.J. (2008). Universality in cell mechanics. Soft Matter, volume 4, p. 1750-1759. [189] Cukierman, E., Pankov, R., Stevens, D.R. and Yamada, K.M. (2001). Taking cell-matrix adhesions to the third dimension. Science, volume 294, p. 1708-1712. [190] Dube, N., Al-Malaika, S., Laroche, G. and Vermette, P., (2001). Additives in biomedical polyurethanes. In: Vermette, P., Griesser, H.J., Laroche, G., Guidoin, R. (Eds.) Biomedical applications of polyurethanes, Landes Bioscience - Eurekah.com Publishers, Georgetown, Texas, United States, p. 55-74.

177 [191] Shaikh, F.M., Brien, T.P.O., Callanan, A., Kavanagh, E.G., Burke, P.E., Grace, P.A., Badylak, S.F. and McGloughlin, T.M. (2007). Fibrin gel scaffold optimization for tissue engineering vascular grafts. Tissue Engineering, volume 13, p. 1696-1697. [192] Shaikh, F.M., Callanan, A., Kavanagh, E.G., Burke, P.E., Grace, P.A. and McGloughlin, T.M. (2008). Fibrin: A natural biodegradable scaffold in vascular tissue engineering. Cells Tissues Organs, volume 188, p. 333-346. [193] Bates, I.R., Wiseman, P.W. and Hanrahan, J.W. (2006). Investigating membrane protein dynamics in living cells. Biochemistry and Cell Biology, volume 84, p. 825-831. [194] Bacia, K. and Schwille, P. (2003). A dynamic view of cellular processes by in vivo fluorescence auto- and cross-correlation spectroscopy. Methods, volume 29, p. 74-85. [195] Bacia, K., Kim, S.A. and Schwille, P. (2006). Fluorescence cross-correlation spectroscopy in living cells. Nature Methods, volume 3, p. 83-89. [196] Lippincott-Schwartz, J., Altan-Bonnet, N. and Patterson, G.H. (2003). Photobleaching and photoactivation: following protein dynamics in living cells. Nature Reviews Molecular Cell Biology, volume Supplement, p. S7-S14. [197] Petersen, N.O., Hoddelius, P.L., Wiseman, P.W., Seger, O. and Magnusson, K.E. (1993). Quantitation of membrane-receptor distributions by image correlation spectroscopy - concept and application. Biophysical Journal, volume 65, p. 1135-1146. [198] Wiseman, P.W., Squier, J.A., Ellisman, M.H. and Wilson, K.R. (2000). Two-photon image correlation spectroscopy and image cross-correlation spectroscopy. Journal o f Microscopy, volume 200, p. 14-25. [199] Comeau, J.W.D., Kolin, D.L. and Wiseman, P.W. (2008). Accurate measurements of protein interactions in cells via improved spatial image cross-correlation spectroscopy. Molecular BioSystems, volume 4, p. 672-685. [200] Brown, C.M., Hebert, B., Kolin, D.L., Zareno, J., Whitmore, L., Horwitz, A.R. and Wiseman, P.W. (2006). Probing the integrin-actin linkage using high-resolution protein velocity mapping. Journal o f Cell Science, volume 119, p. 5204-5214. [201] Wiseman, P.W., Brown, C.M., Webb, D.J., Hebert, B., Johnson, N.L., Squier, J.A., Ellisman, M.H. and Horwitz, A.F. (2004). Spatial mapping of integrin interactions and dynamics during cell migration by Image Correlation Microscopy. Journal o f Cell Science, volume 117, p. 5521-5534.

178 [202] Hebert, B., Costantino, S. and Wiseman, P.W. (2005). Spatiotemporal image correlation Spectroscopy (STICS) theory, verification, and application to protein velocity mapping in living CHO cells. Biophysical Journal, volume 88, p. 3601-3614. [203] Centis, V. and Vermette, P. (2009). Enhancing oxygen solubility using hemoglobin- and perfluorocarbon-based carriers. Frontiers in Bioscience, volume 14, p. 665-688. [204] Centis, V., Doillon, C.J. and Vermette, P. (2007). Perfluorocarbon emulsions cytotoxic effects on human fibroblasts and effect of aging on particle size distribution. Artificial Organs, volume 31, p. 649-653. [205] Centis, V. and Vermette, P. (2008). Physico-chemical properties and cytotoxicity assessment of PEG-modified liposomes containing human hemoglobin. Colloids and Surfaces B: Biointerfaces, volume 65, p. 239-246. [206] Centis, V., Development of culture media capable of enhanced oxygen transport and transfer in three-dimensional cell culture systems, in, Universite de Sherbrooke, Sherbrooke, Canada, 2009, pp. 127. [207] Miller, M.J., Wei, S.H., Parker, I. and Cahalan, M.D. (2002). Two-photon imaging of lymphocyte motility and antigen response in intact lymph node. Science, volume 296, p. 1869-1873. [208] Stoll, S., Delon, J., Brotz, T.M. and Germain, R.N. (2002). Dynamic imaging of T cell- dendritic cell interactions in lymph nodes. Science, volume 296, p. 1873-1876. [209] Guido, S. and Tranquillo, R.T. (1993). A methodology for the systematic and quantitative study of cell contact guidance in oriented collagen gels. Correlation of fibroblast orientation and gel birefringence. Journal o f Cell Science, volume 105, p. 317-331. [210] Cidadao, A.J. (1989). Interactions between fibronectin, glycosaminoglycans and native collagen fibrils: an EM study in artificial three-dimensional extracellular matrices. European Journal of Cell Biology, volume 48, p. 303-312. [211] Maaser, K., Wolf, K., Klein, C.E., Niggemann, B., Zanker, K.S., Brocker, E.B. and Friedl, P. (1999). Functional hierarchy of simultaneously expressed adhesion receptors: integrin alpha2betal but not CD44 mediates MV3 melanoma cell migration and matrix reorganization within three-dimensional hyaluronan-containing collagen matrices. Molecular Biology o f the Cell, volume 10, p. 3067-3079.

179 [212] Birk, D.E. and Lisenmayer, T.F., (1994). Collagen fibril assembly, deposition, and organization. In: Yurchenco, P.D., Birk, D.E., Mecham, R.P. (Eds.) Extracellular Matrix Assembly and Structure, Academic Press, San Diego, United States, p. 91-128. [213] Hiraoka, N., Allen, E., Apel, I.J., Gyetko, M.R. and Weiss, S.J. (1998). Matrix metalloproteinases regulate neovascularization by acting as pericellular fibrinolysins. Cell, volume 95, p. 365-377. [214] Sakiyama-Elbert, S.E., Panitch, A. and Hubbell, J.A. (2001). Development of growth factor fusion proteins for cell-triggered drug delivery. Federation of American Societies for Experimental Biology Journal, volume 15, p. 1300-1302. [215] Schense, J.C., Bloch, J., Aebischer, P. and Hubbell, J.A. (2000). Enzymatic incorporation of bioactive peptides into fibrin matrices enhances neurite extension. Nature Biotechnology, volume 18, p. 415-419. [216] Hayen, W., Goebeler, M., Kumar, S., Riessen, R. and Nehls, V. (1999). Hyaluronan stimulates tumor cell migration by modulating the fibrin fiber architecture. Journal of Cell Science, volume 112, p. 2241-2251. [217] Sun, T., Norton, D., Ryan, A.J., MacNeil, S. and Haycock, J.W. (2007). Investigation of fibroblast and keratinocyte cell-scaffold interactions using a novel 3D cell culture system. Journal o f Materials Science: Materials in Medicine, volume 18, p. 321-328. [218] Boublik, J., Park, H., Radisic, M., Tognana, E., Chen, F., Pei, M., Vunjak-Novakovic, G. and Freed, L.E. (2005). Mechanical properties and remodeling of hybrid cardiac constructs made from heart cells, fibrin, and biodegradable, elastomeric knitted fabric. Tissue Engineering, volume 11, p. 1122-1132. [219] Alsberg, E., Feinstein, E., Joy, M.P., Prentiss, M. and Ingber, D.E. (2006). Magnetically- guided self-assembly of fibrin matrices with ordered nano-scale structure for tissue engineering. Tissue Engineering, volume 12, p. 3247-3256. [220] Willerth, S.M., Arendas, K.J., Gottlieb, D.I. and Sakiyama-Elbert, S.E. (2006). Optimization of fibrin scaffolds for differentiation of murine embryonic stem cells into neural lineage cells. Biomaterials, volume 27, p. 5990-6003. [221] Kim, J.B., Stein, R. and O'Hare, M.J. (2004). Three-dimensional in vitro tissue culture models of breast cancer— a review. Breast Cancer Research and Treatment, volume 85, p. 281-291.

180 [222] Maquet, V., Boccaccini, A.R., Pravata, L., Notingher, I. and Jerome, R. (2004). Porous poly(a-hydroxyacid)/Bioglass® composite scaffolds for bone tissue engineering. I: preparation and in vitro characterisation. Biomaterials, volume 25, p. 4185-4194. [223] Zhang, K., Wang, Y., Hillmyer, M.A. and Francis, L.F. (2004). Processing and properties of porous poly(l-lactide)/bioactive glass composites. Biomaterials, volume 25, p. 2489-2500. [224] Rezwan, K., Chen, Q.Z., Blaker, J.J. and Boccaccini, A.R. (2006). Biodegradable and bioactive porous polymer/inorganic composite scaffolds for bone tissue engineering. Biomaterials, volume 27, p. 3413-3431. [225] Lutolf, M.P., Raeber, G.P., Zisch, A.H., Tirelli, N. and Hubbell, J.A. (2003). Cell- responsive synthetic hydrogels. Advanced Materials, volume 15, p. 888-892. [226] Hutmacher, D.W. (2001). Scaffold design and fabrication technologies for engineering tissues—state of the art and future perspectives. Journal o f Biomaterials Science, Polymer Edition, volume 12, p. 107-124. [227] Hollister, S.J. (2005). Porous scaffold design for tissue engineering. Nature Materials, volume 4, p. 518-524. [228] Li, N., Tourovskaia, A. and Folch, A. (2003). Biology on a chip: microfabrication for studying the behavior of cultured cells. Critical Reviews in Biomedical Engineering, volume 31, p. 423-488. [229] Chen, C.S., Mrksich, M., Huang, S., Whitesides, G.M. and Ingber, D.E. (1997). Geometric control of cell life and death. Science, volume 276, p. 1425-1428. [230] Chen, C.S., Mrksich, M., Huang, S., Whitesides, G.M. and Ingber, D.E. (1998). Micropattemed surfaces for control of cell shape, position, and function. Biotechnology Progress, volume 14, p. 356-363. [231] Pearson, R.G., Bhandari, R., Quirk, R.A. and Shakesheff, K.M. (2002). Recent advances in tissue engineering: an invited review. Journal o f Long-Term Effects o f Medical Implants, volume 12, p. 1-33. [232] Thissen, H., Johnson, G., Hartley, P.G., Kingshott, P. and Griesser, H.J. (2006). Two- dimensional patterning of thin coatings for the control of tissue outgrowth. Biomaterials, volume 27, p. 35-43. [233] Kane, R.S., Takayama, S., Ostuni, E., Ingber, D.E. and Whitesides, G.M. (1999). Patterning proteins and cells using soft lithography. Biomaterials, volume 20, p. 2363-2376.

181 [234] Whitesides, G.M., Ostuni, E., Takayama, S., Jiang, X. and Ingber, D.E. (2001). Soft lithography in biology and biochemistry. Annual Review o f Biomedical Engineering, volume 3, p. 335-373. [235] Falconnet, D., Csucs, G., Grandin, H.M. and Textor, M. (2006). Surface engineering approaches to micropattem surfaces for cell-based assays. Biomaterials, volume 27, p. 3044- 3063. [236] Chiu, D.T., Jeon, N.L., Huang, S., Kane, R.S., Wargo, C.J., Choi, I.S., Ingber, D.E. and Whitesides, G.M. (2000). Patterned deposition of cells and proteins onto surfaces by using three-dimensional microfluidic systems. Proceedings of the National Academy of Sciences of the United States o f America, volume 97, p. 2408-2413. [237] Cima, L.G., Vacanti, J.P., Vacanti, C., Ingber, D., Mooney, D. and Langer, R. (1991). Tissue engineering by cell transplantation using degradable polymer substrates. Journal o f Biomechanical Engineering, volume 113, p. 143-151. [238] Seitz, H., Rieder, W., Irsen, S., Leukers, B. and Tille, C. (2005). Three-dimensional printing of porous ceramic scaffolds for bone tissue engineering. Journal of Biomedical Materials Research, volume 74, p. 782-788. [239] Giordano, R.A., Wu, B.M., Borland, S.W., Cima, L.G., Sachs, E.M. and Cima, M.J. (1996). Mechanical properties of dense poly lactic acid structures fabricated by three dimensional printing. Journal o f Biomaterials Science, Polymer Edition, volume 8, p. 63-75. [240] Curtis, A. and Richie, M. (2001). Tissue engineering: the biophysical background. Physics in Medicine and Biology, volume 46, p. R47-R65. [241] Curtis, A.S. and Wilkinson, C.D. (1998). Reactions of cells to topography. Journal o f Biomaterials Science, Polymer Edition, volume 9, p. 1313-1329. [242] Lydon, M.J. and Clay, C.S. (1985). Substratum topography and cell traction on sulphuric acid treated bacteriological-grade plastic. Cell Biology International Reports, volume 9, p. 911-921. [243] Lundqvist, M., Sethson, I. and Jonsson, B.H. (2004). Protein adsoiption onto silica nanoparticles: conformational changes depend on the particles' curvature and the protein stability. Langmuir, volume 20, p. 10639-10647.

182 [244] Roach, P., Farrar, D. and Perry, C.C. (2006). Surface tailoring for controlled protein adsorption: effect of topography at the nanometer scale and chemistry. Journal of the American Chemical Society, volume 128, p. 3939-3945. [245] Vertegel, A.A., Siegel, R.W. and Dordick, J.S. (2004). Silica nanoparticle size influences the structure and enzymatic activity of adsorbed lysozyme. Langmuir, volume 20, p. 6800- 6807. [246] Roach, P., Eglin, D., Rohde, K. and Perry, C.C. (2007). Modem biomaterials: a review- bulk properties and implications of surface modifications. Journal of Materials Science: Materials in Medicine, volume 18, p. 1263-1277. [247] Gail, M.H. and Boone, C.W. (1972). Cell-substrate adhesivity. A determinant of cell motility. Experimental Cell Research, volume 70, p. 33-40. [248] Britland, S., Clark, P., Connolly, P. and Moores, G. (1992). Micropattemed substratum adhesiveness: a model for morphogenetic cues controlling cell behavior. Experimental Cell Research, volume 198, p. 124-129. [249] Britland, S., Morgan, H., Wojiak-Stodart, B., Riehle, M., Curtis, A. and Wilkinson, C. (1996). Synergistic and hierarchical adhesive and topographic guidance of BHK cells. Experimental Cell Research, volume 228, p. 313-325. [250] Clark, P., Connolly, P. and Moores, G.R. (1992). Cell guidance by micropattemed adhesiveness in vitro. Journal of Cell Science, volume 103 ( Pt 1), p. 287-292. [251] Pritchard, D.I., Morgan, H. and Cooper, J.M. (1995). Microscale patterning of biological molecules. Angewandte Chemie volume 34, p. 91-93. [252] Curtis, A. and Wilkinson, C. (1997). Topographical control of cells. Biomaterials, volume 18, p. 1573-1583. [253] Hynes, R.O. (2007). Cell-matrix adhesion in vascular development. Journal o f Thrombosis and Haemostasis, volume 5 Supplement 1, p. 32-40. [254] Vermette, P. and Meagher, L. (2003). Interactions of phospholipid- and poly(ethylene glycol)-modified surfaces with biological systems: relation to physico-chemical properties and mechanisms. Colloids and Surfaces B: Biointerfaces, volume 28, p. 153-198. [255] Golander, C., Herron, J., Lim, K., Claesson, P., Stenius, P. and Andrade, J., (1992). Properties of immoblilized PEG films and the interaction with proteins: experiments and

183 modelling. In: Harris, J.M. (Ed.) Poly(ethylene glycol) Chemistry, Biotechnical and Biomedical Applications, Plenum Press, New York, United States, p. 221-245. [256] Harder, P., Grunze, M., Dahint, R., Whitesides, G.M. and Laibinis, P.E. (1998). Molecular conformation in oligo(ethylene glycol)-terminated self-assembled monolayers on gold and silver surfaces determines their ability to resist protein adsorption. Journal of Physical Chemistry B, volume 102, p. 426-436. [257] Martin, Y. and Vermette, P. (2006). Low-fouling amine-terminated poly(ethylene glycol) thin layers and effect of immobilization conditions on their mechanical and physicochemical properties. Macromolecules, volume 39, p. 8083-8091. [258] Massia, S.P. and Stark, J. (2001). Immobilized RGD peptides on surface-grafted dextran promote biospecific cell attachment. Journal of Biomedical Materials Research, volume 56, p. 390-399. [259] Massia, S.P., Stark, J. and Letbetter, D.S. (2000). Surface-immobilized dextran limits cell adhesion and spreading. Biomaterials, volume 21, p. 2253-2261. [260] McLean, K.M., Johnson, G., Chatelier, R.C., Beumer, G.J., Steele, J.G. and Griesser, H.J. (2000). Method of immobilization of carboxymethyl-dextran affects resistance to tissue and cell colonization. Colloids and Surfaces B: Biointerfaces, volume 18, p. 221-234. [261] Chapman, D. and Hayward, J.A. (1985). New biophysical techniques and their application to the study of membranes. Biochemical Journal, volume 228, p. 281-295. [262] Chapman, D., Lee, D.C. and Hayward, J.A. (1986). Physicochemical studies of vesicles and biomembranes. Spectroscopic studies and phospholipid polymers. Faraday Discussions o f the Chemical Society, volume 81, p. 107-116. [263] Ishihara, K., Aragaki, R., Ueda, T., Watenabe, A. and Nakabayashi, N. (1990). Reduced thrombogenicity of polymers having phospholipid polar groups. Journal of Biomedical Materials Research, volume 24, p. 1069-1077. [264] Ishihara, K., Ziats, N.P., Tierney, B.P., Nakabayashi, N. and Anderson, J.M. (1991). Protein adsorption from human plasma is reduced on phospholipid polymers. Journal o f Biomedical Materials Research, volume 25, p. 1397-1407. [265] Iwasaki, Y., Kurita, K., Ishihara, K. and Nakabayashi, N. (1996). Effect of reduced protein adsorption on platelet adhesion at the phospholipid polymer surfaces. Journal o f Biomaterials Science, Polymer Edition, volume 8, p. 151-163.

184 [266] Ratner, B.D., Hoffman, A.S., Schoen, F.J. and Lemons, J.E. (1996) Biomaterials Science: An Introduction to Materials in Medicine, Academic Press, San Diego, United States, 484 p. [267] Nebe, B., Finke, B., Luthen, F., Bergemann, C., Schroder, K., Rychly, J., Liefeith, K. and Ohl, A. (2007). Improved initial osteoblast functions on amino-functionalized titanium surfaces. Biomolecular Engineering, volume 24, p. 447-454. [268] Mrksich, M., Chen, C.S., Xia, Y., Dike, L.E., Ingber, D.E. and Whitesides, G.M. (1996). Controlling cell attachment on contoured surfaces with self-assembled monolayers of alkanethiolates on gold. Proceedings of the National Academy of Sciences of the United States o f America, volume 93, p. 10775-10778. [269] Mrksich, M., Dike, L.E., Tien, J., Ingber, D.E. and Whitesides, G.M. (1997). Using microcontact printing to pattern the attachment of mammalian cells to self-assembled monolayers of alkanethiolates on transparent films of gold and silver. Experimental Cell Research, volume 235, p. 305-313. [270] Mrksich, M. and Whitesides, G.M. (1996). Using self-assembled monolayers to understand the interactions of man-made surfaces with proteins and cells. Annual Review o f Biophysics and Biomolecular Structure, volume 25, p. 55-78. [271] Whitesides, G.M., Kriebel, J.K. and Love, J.C. (2005). Molecular engineering of surfaces using self-assembled monolayers. Science Progress, volume 88, p. 17-48. [272] Ulman, A. (1996). Formation and structure of self-assembled monolayers. Chemical Reviews, volume 96, p. 1533-1554. [273] Colin, D., Bain, E., Troughton, B., Tao, Y.T., Evall, J., Whitesides, G.M. and Nuzzo, R.G. (1989). Formation of monolayer films by the spontaneous assembly of organic thiols from solution onto gold. Journal o f the American Chemical Society, volume 111, p. 321 - 335. [274] Godin, M., Williams, P.J., Tabard-Cossa, V., Laroche, O., Beaulieu, L.Y., Lennox, R.B. and Grutter, P. (2004). Surface stress, kinetics, and structure of alkanethiol self-assembled monolayers. Langmuir, volume 20, p. 7090-7096. [275] Haehner, G., Woll, C., Buck, M. and GrUnze, M. (1993). Investigation of intermediate steps in the self-assembly of n-alkanethiols on gold surfaces by soft x-ray spectroscopy. Langmuir, volume 9, p. 1955-1958. [276] Sushko, M.L. and Shluger, A.L. (2007). Dipole-dipole interactions and the structure of self-assembled monolayers. Journal o f Physical Chemistry B, volume 111, p. 4019-4025.

185 [277] Wang, H., Chen, S., Li, L. and Jiang, S. (2005). Improved method for the preparation of carboxylic acid and amine terminated self-assembled monolayers of alkanethiolates. Langmuir, volume 21, p. 2633-2636. [278] Onclin, S., Ravoo, B.J. and Reinhoudt, D.N. (2005). Engineering silicon oxide surfaces using self-assembled monolayers. Angewandte Chemie. International Edition in English, volume 44, p. 6282-6304. [279] Angst, D.L. and Simmons, G.W. (1991). Moisture absorption characteristics of organosiloxane self-assembled monolayers. Langmuir, volume 7, p. 2236-2242. [280] Le Grange, J.D., Markham, J.L. and Kurkjian, C.R. (1993). Effects of surface hydration on the deposition of silane monolayers on silica. Langmuir, volume 9, p. 1749-1753. [281] Silberzan, P., Leger, L., Ausserre, D. and Benattar, J.J. (1991). Silanation of silica surfaces. A new method of constructing pure or mixed monolayers. Langmuir, volume 7, p. 1647-1651. [282] Tripp, C.P. and Hair, M.L. (1992). An infared study of the reaction of octadecyltrichlorosilane with silica. Langmuir, volume 8, p. 1120-1126. [283] Wasserman, S.P., Tao, Y. and Whitesides, G.M. (1989). Structure and reactivity of alkylsiloxane monolayers formed by reaction of alkyltrichlorosilanes on silicon substrates. Langmuir, volume 5, p. 1074-1087. [284] Bain, J.R. and Hoffman, A.S. (2003). Tissue-culture surfaces with mixtures of aminated and fluorinated functional groups. Part 1. Synthesis and characterization. Journal o f Biomaterials Science, Polymer Edition, volume 14, p. 325-339. [285] Wang, A., Tang, H., Cao, T., Salley, S.O. and Ng, K.Y. (2005). In vitro stability study of organosilane self-assemble monolayers and multilayers. Journal of Colloid and Interface Science, volume 291, p. 438-447. [286] Li, J., Thiara, P.S. and Mrksich, M. (2007). Rapid evaluation and screening of interfacial reactions on self-assembled monolayers. Langmuir, volume 23, p. 11826-11835. [287] Whitesides, G.M. and Grzybowski, B. (2002). Self-assembly at all scales. Science, volume 295, p. 2418-2421. [288] Ulman, A. (1991) An Introduction to Ultrathin Organic Films: From Langmuir-Blodgett to Self-Assembly, 1st Edition, Academic Press, San Diego, United States, 442 p.

186 [289] Vermette, P. and Meagher, L. (2002). Immobilization and characterization of poly(acrylic acid) graft layers. Langmuir, volume 18, p. 10137-10145. [290] Yasuda, H. (1981). Glow discharge polymerization. Journal of Polymer Science: Macromolecular Reviews, volume 16, p. 199-293. [291] Siow, K.S., Britcher, L., Kumar, S. and Griesser, H.J. (2006). Plasma methods for the generation of chemically reactive surfaces for biomolecule immobilization and cell colonization - A review. Plasma Processes and Polymers, volume 3, p. 392-418. [292] Martin, Y., Boutin, D. and Vermette, P. (2007). Study of the effect of process parameters for n-heptylamine plasma polymerization on final layer properties. Thin Solid Films, volume 515, p. 6844-6852. [293] Lopez, G.P. and Ratner, B.D. (1991). Substrate temperature effects on film chemistry in plasma deposition of organics. 1. Nonpolymerizable precursors. Langmuir, volume 7, p. 766- 773. [294] Lopez, G.P. and Ratner, B.D. (1992). Substrate temperature effects on film chemistry in plasma deposition of organics. II. Polymerizable precursors Journal o f Polymer Science, Polymer Chemistry Edition, volume 30, p. 2415-2425. [295] Chen, Q., Forch, R. and Knoll, W. (2004). Characterization of pulsed plasma polymerization allylamine as an adhesion layer for DNA adsorption/hybridization. Chemistry o f Materials, volume 16, p. 614-620. [296] Chu, L., Knoll, W. and Forch, R. (2006). Biological multifunctional surfaces using plasma polymerization methods. Plasma Processes and Polymers, volume 3, p. 498-505. [297] Gengenbach, T.R., Chatelier, R.C. and Griesser, H.J. (1996). Characterization of the ageing of plasma-deposited polymer films: Global analysis of x-ray photoelectron spectroscopy data. Surface and Interface Analysis, volume 24, p. 271-281. [298] Vasilev, K , Britcher, L., Casanal, A. and Griesser, H.J. (2008). Solvent-induced porosity in ultrathin amine plasma polymer coatings. Journal o f Physical Chemistry B, volume 112, p. 10915-10921. [299] Hamerli, P., Weigel, T., Groth, T. and Paul, D. (2003). Surface properties of and cell adhesion onto allylamine-plasma-coated polyethylenterephtalat membranes. Biomaterials, volume 24, p. 3989-3999.

187 [300] Aronsson, B.O., Lausmaa, J. and Kasemo, B. (1997). Glow discharge plasma treatment for surface cleaning and modification of metallic biomaterials. Journal of Biomedical Materials Research, volume 35, p. 49-73. [301] Huber, F. and Springer, J. (1996). Construction of a plasma polymerization reactor system for laboratory use. Journal o f Applied Polymer Science, volume 62, p. 145-151. [302] Massia, S.P., Holecko, M.M. and Ehteshami, G.R. (2004). In vitro assessment of bioactive coatings for neural implant applications. Journal o f Biomedical Materials Research, Part A,

volume 6 8 A, p. 177-186. [303] Wang, T.T., Ehteshami, G., Massia, S. and Muthuswamy, J. (2006). Immobilization and characterization of gamma-aminobutyric acid on gold surface. Journal of Biomedical Materials Research, Part A, volume 79A, p. 201-209. [304] Keselowsky, B.G., Collard, D.M. and Garcia, A.J. (2004). Surface chemistry modulates focal adhesion composition and signaling through changes in integrin binding. Biomaterials, volume 25, p. 5947-5954. [305] Ratner, B.D. and Castner, D.G., (2000). Electron spectroscopy for chemical analysis. In: Vickerman, J.C. (Ed.) Surface Analysis: The principle techniques, John Wiley and Sons Ltd., London, United Kingdom, p. 43-98. [306] Paynter, R.W., Ratner, B.D., Horbett, T.A. and Thomas, H.R. (1984). XPS studies on the organization of adsorbed protein films on fluoropolymers. Journal of Colloid and Interface Science, volume 101, p. 233-245. [307] Hillenkamp, F. (2007) MALDI MS: A practical guide to instrumentation, methods and applications, Wiley, Weinheim, Germany, 346 p. [308] Vickerman, J.C. and Swift, A., (2000). Secondary ion mass spectrometry - the surface mass spectrometry. In: Vickerman, J.C. (Ed.) Surface Analysis: The principle techniques, John Wiley & Sons Ltd., London, United Kingdom, p. 135-214. [309] Wong, J.Y., Leach, J.B. and Brown, X.Q. (2004). Balance of chemistry, topography, and mechanics at the cell-biomaterial interface: Issues and challenges for assessing the role of substrate mechanics on cell response. Surface Science, volume 570, p. 119-133. [310] Pelham, R.J. and Wang, Y.L. (1997). Cell locomotion and focal adhesions are regulated by substrate flexibility. Proceedings o f the National Academy o f Sciences o f the United States o f America, volume 94, p. 13661-13665.

188 [311] Chang, C.H., Liao, J.D., Chen, J.J.J., Ju, M.S. and Lin, C.C.K. (2006). Cell adhesion and related phenomena on the surface-modified Au-deposited nerve microelectrode examined by total impedance measurement and cell detachment tests. Nanotechnology, volume 17, p. 2449-2457. [312] Drumheller, P.D. and Hubbell, J.A. (1995). Densely cross-linked polymer networks of poly(ethylene glycol) in trimethylolpropane triacrylate for cell-adhesion-resistant surfaces. Journal o f Biomedical Materials Research, volume 29, p. 207-215. [313] Faghihi, S., Azari, F., Li, H., Bateni, M.R., Szpunar, J.A., Vali, H. and Tabrizian, M. (2006). The significance of crystallographic texture of titanium alloy substrates on pre osteoblast responses. Biomaterials, volume 27, p. 3532-3539. [314] Guillemot, G., Vaca-Medina, G., Martin-Yken, H., Vemhet, A., Schmitz, P. and Mercier- Bonin, M. (2006). Shear-flow induced detachment of Saccharomyces cerevisiae from stainless steel: Influence of yeast and solid surface properties. Colloids and Surfaces B: Biointerfaces, volume 49, p. 126-135. [315] Hao, L., Lawrence, J., Phua, Y.F., Chian, K.S., Lim, G.C. and Zheng, H.Y. (2005). Enhanced human osteoblast cell adhesion and proliferation on 316 LS stainless steel by means of C02 laser surface treatment. Journal of Biomedical Materials Research, Part B: Applied Biomaterials, volume 73B, p. 148-156. [316] Hori, Y., Inoue, S., Hirano, Y. and Tabata, Y. (2004). Effect of culture substrates and fibroblast growth factor addition on the proliferation and differentiation of rat bone marrow stromal cells. Tissue Engineering, volume 10, p. 995-1005. [317] Kang, S., Hoek, E.M.V., Choi, H. and Shin, H. (2006). Effect of membrane surface properties during the fast evaluation of cell attachment. Separation Science and Technology, volume 41, p. 1475-1487. [318] Krasteva, N., Harms, U., Albrecht, W., Seifert, B., Hopp, M., Altankov, G. and Groth, T. (2002). Membranes for biohybrid liver support systems - investigations on hepatocyte attachment, morphology and growth. Biomaterials, volume 23, p. 2467-2478. [319] Lee, S.J., Khang, G., Lee, Y.M. and Lee, H.B. (2002). Interaction of human chondrocytes and NIH/3T3 fibroblasts on chloric acid-treated biodegradable polymer surfaces. Journal o f Biomaterials Science, Polymer Edition, volume 13, p. 197-212.

189 [320] Lerouge, S., Major, A., Girault-Lauriault, P.L., Raymond, M.A., Laplante, P., Soulez, G., Mwale, F., Wertheimer, M.R. and Hebert, M.J. (2007). Nitrogen-rich coatings for promoting healing around stent-grafts after endovascular aneurysm repair. Biomaterials, volume 28, p. 1209-1217. [321] MacDonald, D.E., Rapuano, B.E., Deo, N., Stranick, M., Somasundaran, P. and Boskey, A.L. (2004). Thermal and chemical modification of titanium-aluminum-vanadium implant materials: effects on surface properties, glycoprotein adsorption, and MG63 cell attachment. Biomaterials, volume 25, p. 3135-3146. [322] Magoshi, T. and Matsuda, T. (2002). Formation of polymerized mixed heparin/albumin surface layer and cellular adhesional responses. Biomacromolecules, volume 3, p. 976-983. [323] Pesakova, V., Kubies, D., Hulejova, H. and Himmlova, L. (2007). The influence of implant surface properties on cell adhesion and proliferation. Journal o f Materials Science: Materials in Medicine, volume 18, p. 465-473. [324] Pompe, T., Keller, K., Mothes, G., Nitschke, M., Teese, M., Zimmermann, R. and Werner, C. (2007). Surface modification of poly(hydroxybutyrate) films to control cell-matrix adhesion. Biomaterials, volume 28, p. 28-37. [325] van Kooten, T.G., Spijker, H.T. and Busscher, H.J. (2004). Plasma-treated polystyrene surfaces: model surfaces for studying cell-biomaterial interactions. Biomaterials, volume 25, p. 1735-1747. [326] Yung, L.Y.L. and Cooper, S.L. (1998). Neutrophil adhesion on phosphorylcholine- containing polyurethanes. Biomaterials, volume 19, p. 31-40. [327] Zheng, Z., Bei, F.F., Tian, H.L. and Chen, G.Q. (2005). Effects of crystallization of polyhydroxyalkanoate blend on surface physicochemical properties and interactions with rabbit articular cartilage chondrocytes. Biomaterials, volume 26, p. 3537-3548. [328] Zhu, X.L., Chen, J., Scheideler, L., Reichl, R. and Geis-Gerstorfer, J. (2004). Effects of topography and composition of titanium surface oxides on osteoblast responses. Biomaterials, volume 25, p. 4087-4103. [329] Tsai, W.-B., Wei, T.-C., Lin, M.-C., Wang, J.-Y. and Chen, C.-H. (2005). The effect of radio-frequency glow discharge treatment of polystyrene on the behavior of porcine chondrocytes in vitro. Journal o f Biomaterials Science, Polymer Edition, volume 16, p. 699- 714.

190 [330] Good, R.J. (1992). Contact-angle, wetting, and adhesion - a critical-review. Journal o f

Adhesion Science and Technology, volume 6 , p. 1269-1302. [331] van Oss, C.J. (1997). Hydrophobicity and hydrophilicity of biosurfaces. Current Opinion in Colloid & Interface Science, volume 2, p. 503-512. [332] Craig, V.S.J., Ninham, B.W. and Pashley, R.M. (1993). The effect of electrolytes on bubble coalescence in water. Journal o f Physical Chemistry, volume 97, p. 10192-10197. [333] Yaminsky, V.V., Ninham, B.W., Christenson, H.K. and Pashley, R.M. (1996). Adsorption forces between hydrophobic monolayers. Langmuir, volume 12, p. 1936-1943. [334] Yaminsky, V.V., Yuschenko, V.S., Amelina, E.A. and Shehukin, E.D. (1983). Cavity formation due to a contact between particles in a nonwetting liquid. Journal of Colloid and Interface Science, volume 96, p. 301-306. [335] Queiroz, J.A., Tomaz, C.T. and Cabral, J.M.S. (2001). Hydrophobic interaction chromatography of proteins. Journal o f Biotechnology, volume 87, p. 143-159. [336] Mahn, A. and Asenjo, J.A. (2005). Prediction of protein retention in hydrophobic interaction chromatography. Biotechnology Advances, volume 23, p. 359-368. [337] Fox, J. and Zisman, W. (1950). The spreading of liquids on low energy surfaces I. polytetrafluoroethylene. Journal o f Colloid and Interface Science, volume 5, p. 514-531. [338] Zisman, W., (1964). Relation of the equilibrium contact angle to liquid and solid constitution. In: Fowkes, F.M. (Ed.) Contact angle wettability and adhesion, volume 43, American Chemical Society, Washignton, D.C., United States, p. 1-51. [339] Kumar, G. and Prabhu, K.N. (2007). Review of non-reactive and reactive wetting of liquids on surfaces. Advances in Colloid and Interface Science, volume 133, p. 61-89. [340] Li, X.M., Reinhoudt, D. and Crego-Calama, M. (2007). What do we need for a superhydrophobic surface? A review on the recent progress in the preparation of superhydrophobic surfaces. Chemical Society Reviews, volume 36, p. 1350-1368. [341] Sun, T.L., Feng, L., Gao, X.F. and Jiang, L. (2005). Bioinspired surfaces with special wettability. Accounts o f Chemical Research, volume 38, p. 644-652. [342] Yektafard, M. and Ponter, A.B. (1992). Factors affecting the wettability of polymer

surfaces. Journal o f Adhesion Science and Technology, volume 6 , p. 253-277.

191 [343] Verplanck, N., Coffmier, Y., Thorny, V. and Boukherroub, R. (2007). Wettability switching techniques on superhydrophobic surfaces. Nanoscale Research Letters, volume 2, p. 577-596. [344] de Gennes, P.G. (1985). Wetting: statics and dynamics. Reviews of Modern Physics, volume 57, p. 827-863. [345] Yaminsky, V.V. and Ninham, B.W. (1999). Surface forces vs. surface compositions. Colloid science from the Gibbs adsorption perspective. Advances in Colloid and Interface Science, volume 83, p. 227-311. [346] Vermette, P., Levesque, S. and Griesser, H.J., (2001). Biomedical degradation of polyurethanes. In: Vermette, P., Griesser, H.J., Laroche, G., Guidoin, R. (Eds.) Biomedical applications o f polyureathanes, Landes Bioscience - Eurekah.com Publishers, Georgetown, Texas, United States, p. 97-159. [347] Vogler, E.A., Graper, J.C., Sugg, H.W., Lander, L.M. and Brittain, W.J. (1995). Contact activation of the plasma coagulation cascade .1. Procoagulant surface chemistry and energy. Journal o f Biomedical Materials Research, volume 29, p. 1005-1016. [348] Vogler, E.A. (1998). Structure and reactivity of water at biomaterial surfaces. Advances in Colloid and Interface Science, volume 74, p. 69-117. [349] Franks, F. (1981) Polywater, The MIT Press, Cambridge, Massachusetts, United States,

2 2 2 p. [350] Fowkes, F.M. (1962). Determination of interfacial tensions, contact angles, and dispersion forces in surfaces by assuming additivity of intermolecular interactions in surfaces. Journal

o f Physical Chemistry, volume 6 6 , p. 382. [351] Fowkes, F.M. (1963). Additivity of intermolecular forces at interfaces I. Determination of the contribution to surface and interfacial tensions of dispersion forces in various liquids. Journal o f Physical Chemistry, volume 67, p. 2538-2541. [352] Fowkes, F.M., (1964). Dispersion force contributions to surface and interfacial tensions, contact angles, and heats of immersion. In: Fowkes, F.M. (Ed.) Contact angle, wettability, and adhesion, volume 43, American Chemical Society, Washington, D.C., United States, p. 99-111.

192 [353] Kloubek, J. (1992). Development of methods for surface ffee-energy determination using contact angles of liquids on solids. Advances in Colloid and Interface Science, volume 38, p. 99-142. [354] Hoffman, A.S. (1999). Non-fouling surface technologies. Journal o f Biomaterials Science, Polymer Edition, volume 10, p. 1011-1014. [355] Hall, B., Bird, R.R., Kojima, M. and Chapman, D. (1989). Biomembranes as models for polymer surfaces. V. Thrombelastographic studies of polymeric lipids and polyesters. Biomaterials, volume 10, p. 219-224. [356] Hayward, J.A. and Chapman, D. (1984). Biomembrane surfaces as models for polymer design: the potential for haemocompatibility. Biomaterials, volume 5, p. 135-142. [357] Hayward, J.A., Johnston, D.S. and Chapman, D. (1985). Polymeric phospholipids as new biomaterials. Annals o f the New York Academy o f Sciences, volume 446, p. 267-281. [358] Ishihara, K., Hanyuda, H. and Nakabayashi, N. (1995). Synthesis of phospholipid polymers having a urethane bond in the side chain as coating material on segmented polyurethane and their platelet adhesion-resistant properties. Biomaterials, volume 16, p. 873-879. [359] Ishihara, K., Nomura, H., Mihara, T., Kurita, K., Iwasaki, Y. and Nakabayashi, N. (1998). Why do phospholipid polymers reduce protein adsorption? Journal of Biomedical Materials Research, volume 39, p. 323-330. [360] Ishihara, K., Oshida, H., Endo, Y., Watanabe, A., Ueda, T. and Nakabayashi, N. (1993). Effects of phospholipid adsorption on nonthrombogenicity of polymer with phospholipid polar group. Journal o f Biomedical Materials Research, volume 27, p. 1309-1314. [361] Iwasaki, Y., Sawada, S., Nakabayashi, N., Khang, G., Lee, H.B. and Ishihara, K. (1999). The effect of the chemical structure of the phospholipid polymer on fibronectin adsorption and fibroblast adhesion on the gradient phospholipid surface. Biomaterials, volume 20, p. 2185-2191. [362] Malmsten, M. (1995). Protein adsorption at phospholipid surfaces Journal of Colloid and Interface Science, volume 172, p. 106 -115. [363] Malmsten, M. (1999). Studies of serum protein adsorption at phospholipid surfaces in relation to intravenous drug delivery Colloids and Surfaces, A: Physicochemical and Engineering Aspects, volume 159, p. 77 - 87.

193 [364] Ruiz, L., Fine, E., Voros, J., Makohliso, S.A., Leonard, D., Johnston, D.S., Textor, M. and Mathieu, H.J. (1999). Phosphorylcholine-containing polyurethanes for the control of protein adsorption and cell attachment via photoimmobilized laminin oligopeptides. Journal o f Biomaterials Science, Polymer Edition, volume 10, p. 931-955. [365] Huang, N., Michel, R., Voros, J., Textor, M., Hofer, R., Rossi, A., Elbert, D., Hubbell, J. and Spencer, N. (2001). Poly(L-lysine)-g-poly(ethylene glycol) layers on metal oxide surfaces: Surface-analytical characterization and resistance to serum and fibrinogen adsorption. Langmuir, volume 17, p. 489 - 498. [366] Abuchowski, A., McCoy, J.R., Palczuk, N.C., van Es, T. and Davis, F.F. (1977). Effect of covalent attachment of polyethylene glycol on immunogenicity and circulating life of bovine liver catalase. Journal o f Biological Chemistry, volume 252, p. 3582-3586. [367] Abuchowski, A., van Es, T., Palczuk, N.C. and Davis, F.F. (1977). Alteration of immunological properties of bovine serum albumin by covalent attachment of polyethylene glycol. Journal o f Biological Chemistry, volume 252, p. 3578-3581. [368] Alexander, S. (1977). Polymer adsorption on small spheres. A scaling approach. Journal de Physique, volume 38, p. 977-981. [369] Alexander, S. (1977). Adsorption of chain molecules with a polar head a scaling description. Journal de Physique, volume 38, p. 983-987. [370] de Gennes, P.G. (1976). Scaling theory of polymer adsorption. Journal de Physique, volume 37, p. 1445-1452. [371] de Gennes, P.G. (1980). Conformations of polymers attached to an interface. Macromolecules, volume 13, p. 1069 - 1075. [372] de Gennes, P.G. (1987). Polymers at an interface; a simplified view. Advances in Colloid and Interface Science, volume 27, p. 189-209. [373] Fleer, G.J., Cohen Stuart, M.A., Scheutjens, J.M.H.M., Cosgrove, T. and Vincent, B. (1993) Polymers at Interfaces, Chapman & Hall, London, United Kingdom, 496 p. [374] Pasche, S., Voros, J., Griesser, H.J., Spencer, N.D. and Textor, M. (2005). Effects of ionic strength and surface charge on protein adsorption at PEGylated surfaces. Journal o f Physical Chemistry B, volume 109, p. 17545-17552. [375] Jo, S. and Park, K. (2000). Surface modification using silanated poly(ethylene glycol)s. Biomaterials, volume 21, p. 605-616.

194 [376] Cima, L.G. (1994). Polymer substrates for controlled biological interactions. Journal o f Cellular Biochemistry, volume 56, p. 155-161. [377] Dubois, J., Gaudreault, C. and Vermette, P. (2009). Biofouling of dextran-derivative layers investigated by quartz crystal microbalance. Colloids and Surfaces B: Biointerfaces, volume 71, p. 293-299. [378] Labarre, D., Vauthier, C., Chauvierre, C., Petri, B., Muller, R. and Chehimi, M.M. (2005). Interactions of blood proteins with poly(isobutylcyanoacrylate) nanoparticles decorated with a polysaccharidic brush. Biomaterials, volume 26, p. 5075-5084. [379] Deible, C.R., Petrosko, P., Johnson, P.C., Beckman, E.J., Russell, A.J. and Wagner, W.R. (1998). Molecular barriers to biomaterial thrombosis by modification of surface proteins with polyethylene glycol. Biomaterials, volume 19, p. 1885-1893. [380] Martin, Y., Lepine, M., Bannari, A. and Vermette, P. (2007). Instrument and technique for the in vitro screening of platelet activation from whole blood samples. Review of Scientific Instruments, volume 78, p. 054302-054309. [381] Morra, M. and Cassineli, C. (1999). Non-fouling properties of polysaccharide-coated surfaces. Journal o f Biomaterials Science, Polymer Edition, volume 10, p. 1107-1124. [382] Halperin, A., Fragneto, G., Schollier, A. and Sferrazza, M. (2007). Primary versus ternary adsorption of proteins onto PEG brushes. Langmuir, volume 23, p. 10603-10617. [383] Halperin, A. (1999). Polymer brushes that resist adsorption of model proteins: design parameters. Langmuir, volume 15, p. 2525-2533. [384] Leckband, D., Sheth, S. and Halperin, A. (1999). Grafted poly(ethylene oxide) brushes as nonfouling surface coatings. Journal o f Biomaterials Science, Polymer Edition, volume 10, p. 1125-1147. [385] Leckband, D. and Sivasankar, S. (1999). Forces controlling protein interactions: theory and experiment. Colloids and Surfaces B: Biointerfaces, volume 14, p. 83-97. [386] Malmsten, M., Emoto, K. and Van Alstine, J.M. (1998). Effect of chain density on inhibition of protein adsorption by poly(ethylene glycol) based coatings. Journal of Colloid and Interface Science, volume 202, p. 507-517. [387] Green, R.J., Davies, M.C., Roberts, C.J. and Tendler, S.J. (1998). A surface plasmon resonance study of albumin adsorption to PEO-PPO-PEO triblock copolymers. Journal o f Biomedical Materials Research, volume 42, p. 165-171.

195 [388] Ryan, E.A., Paty, B.W., Senior, P.A., Bigam, D., Alfadhli, E., Kneteman, N.M., Lakey, J.R.T. and Shapiro, A.M.J. (2005). Five-year follow-up after clinical islet transplantation. Diabetes, volume 54, p. 2060-2069. [389] Simpson, N.E., Grant, S.C., Gustavsson, L., Peltonen, V., Blackband, S.J. and Constantinidis, I. (2006). Biochemical consequences of alginate encapsulation: a NMR study of insulin-secreting cells. Biomaterials, volume 27, p. 2577-2586. [390] Weber, L.M., Hayda, K.N., Haskins, K. and Anseth, K.S. (2007). The effects of cell- matrix interactions on encapsulated beta-cell function within hydrogels functionalized with matrix-derived adhesive peptides. Biomaterials, volume 28, p. 3004-3011. [391] Teramura, Y. and Iwata, H. (2009). Islet encapsulation with living cells for improvement of biocompatibility. Biomaterials, volume 30, p. 2270-2275. [392] Daoud, J., Petropavlovskaia, M., Rosenberg, L. and Tabrizian, M. (2010). The effect of extracellular matrix components on the preservation of human islet function in vitro. Biomaterials, volume 31, p. 1676-1682. [393] Weber, L.M., Hayda, K.N. and Anseth, K.S. (2008). Cell-matrix interactions improve beta-cell survival and insulin secretion in three-dimensional culture. Tissue Engineering, Part A, volume 14, p. 1959-1968. [394] Foster, J.L., Williams, G., Williams, L.J. and Tuch, B.E. (2007). Differentiation of transplanted microencapsulated fetal pancreatic cells. Transplantation, volume 83, p. 1440- 1448 [395] Lees, J.G., Lim, S.A., Croll, T., Williams, G., Lui, S., Cooper-White, J., McQuade, L.R., Mathiyalagan, B. and Tuch, B.E. (2007). Transplantation of 3D scaffolds seeded with human embryonic stem cells: biological features of surrogate tissue and teratoma-forming potential. Regenerative Medicine, volume 2, p. 289-300. [396] Boretti, M.I. and Gooch, K.J. (2008). Effect of extracellular matrix and 3D morphogenesis on islet hormone gene expression by Ngn3-infected mouse pancreatic ductal epithelial cells. Tissue Engineering, Part A, volume 14, p. 1927-1937. [397] Mason, M.N., Arnold, C.A. and Mahoney, M.J. (2009). Entrapped collagen type 1 promotes differentiation of embryonic pancreatic precursor cells into glucose-responsive beta-cells when cultured in three-dimensional PEG hydrogels. Tissue Engineering, Part A, volume 15, p. 3799-3808.

196 [398] Mason, M.N. and Mahoney, M.J. (2009). Selective beta-cell differentiation of dissociated embryonic pancreatic precursor cells cultured in synthetic polyethylene glycol hydrogels. Tissue Engineering, Part A, volume 15, p. 1343-1352. [399] Mason, M.N. and Mahoney, M.J. (2010). Inhibition of gamma-secretase activity promotes differentiation of embryonic pancreatic precursor cells into functional islet-like clusters in poly(ethylene glycol) hydrogel culture. Tissue Engineering, Part A, volume 16, p. 2593- 2603. [400] Bonner-Weir, S., Toschi, E., Inada, A., Reitz, P., Fonseca, S.Y., Aye, T. and Sharma, A. (2004). The pancreatic ductal epithelium serves as a potential pool of progenitor cells. Pediatric Diabetes, volume 5, p. 16-22. [401] Gao, R., Ustinov, J., Pulkkinen, M.A., Lundin, K., Korsgren, O. and Otonkoski, T. (2003). Characterization of endocrine progenitor cells and critical factors for their differentiation in human adult pancreatic cell culture. Diabetes, volume 52, p. 2007-2015. [402] Gmyr, V., Kerr-Conte, J., Belaich, S., Vandewalle, B., Leteurtre, E., Vantyghem, M.C., Lecomte-Houcke, M., Proye, C., Lefebvre, J. and Pattou, F. (2000). Adult human

cytokeratin 19-positive cells reexpress insulin promoter factor 1 in vitro: further evidence for pluripotent pancreatic stem cells in humans. Diabetes, volume 49, p. 1671-1680. [403] Sharma, A., Zangen, D.H., Reitz, P., Taneja, M., Lissauer, M.E., Miller, C.P., Weir, G.C., Habener, J.F. and Bonner-Weir, S. (1999). The homeodomain protein IDX-1 increases after an early burst of proliferation during pancreatic regeneration. Diabetes, volume 48, p. 507- 513. [404] Yatoh, S., Dodge, R., Akashi, T., Omer, A., Sharma, A., Weir, G.C. and Bonner-Weir, S. (2007). Differentiation of affinity-purified human pancreatic duct cells to beta-cells. Diabetes, volume 56, p. 1802-1809. [405] Bonner-Weir, S., Taneja, M., Weir, G.C., Tatarkiewicz, K., Song, K., Sharma, A. and O'Neil, J.J. (2000). In vitro cultivation of human islets from expanded ductal tissue. Proceedings of the National Academy of Sciences of the United States o f America, volume 97, p. 7999-8004. [406] Katdare, M.R., Bhonde, R.R. and Parab, P.B. (2004). Analysis of morphological and functional maturation of neoislets generated in vitro from pancreatic ductal cells and their

197 suitability for islet banking and transplantation. Journal of Endocrinology, volume 182, p. 105-112. [407] Githens, S. (1988). The pancreatic duct cell: proliferative capabilities, specific characteristics, metaplasia, isolation, and culture. Journal o f Pediatric Gastroenterology and Nutrition, volume 7, p. 486-506. [408] Lieber, M., Mazzetta, J., Nelson-Rees, W., Kaplan, M. and Todaro, G. (1975). Establishment of a continuous tumor-cell line (PANC-1) from a human carcinoma of the exocrine pancreas. International Journal o f Cancer, volume 15, p. 741-747. [409] Hardikar, A.A., Marcus-Samuels, B., Geras-Raaka, E., Raaka, B.M. and Gershengom, M.C. (2003). Human pancreatic precursor cells secrete FGF2 to stimulate clustering into hormone-expressing islet-like cell aggregates. Proceedings of the National Academy of Sciences of the United States o f America, volume 100, p. 7117-7122. [410] Gershengom, M.C., Hardikar, A.A., Wei, C., Geras-Raaka, E., Marcus-Samuels, B. and Raaka, B.M. (2004). Epithelial-to-mesenchymal transition generates proliferative human islet precursor cells. Science, volume 306, p. 2261-2264. [411] Hui, H., Wright, C. and Perfetti, R. (2001). Glucagon-like peptide 1 induces

differentiation of islet duodenal homeobox -1 - positive pancreatic ductal cells into insulin- secreting cells. Diabetes, volume 50, p. 785-796. [412] Misiti, S., Anastasi, E., Sciacchitano, S., Verga Falzacappa, C., Panacchia, L., Bucci, B., Khouri, D., D'Acquarica, I., Brunetti, E., Di Mario, U., Toscano, V. and Perfetti, R. (2005). 3,5,3'-Triiodo-L-thyronine enhances the differentiation of a human pancreatic duct cell line (hPANC-1) towards a P-cell-Like phenotype. Journal o f Cellular Physiology, volume 204, p. 286-296. [413] Wei, C., Geras-Raaka, E., Marcus-Samuels, B., Oron, Y. and Gershengom, M.C. (2006). Trypsin and thrombin accelerate aggregation of human endocrine pancreas precursor cells. Journal of Cellular Physiology, volume 206, p. 322-328. [414] Wu, Y., Li, J., Saleem, S., Yee, S., Hardikar, A.A. and Wang, R. (2010). c-Kit and stem cell factor regulate PANC-1 cell differentiation into insulin- and glucagon-producing cells. Laboratory Investigation, volume 90, p. 1373-1384.

198 [415] Krishnamurthy, M. and Wang, R. (2009). Integrins and extracellular matrices in pancreatic tissue engineering. Frontiers in Bioscience (Scholar Edition), volume 1, p. 477- 491. [416] Kleinman, H.K., Graf, J., Iwamoto, Y., Kitten, G.T., Ogle, R.C., Sasaki, M., Yamada, Y., Martin, G.R. and Luckenbill-Edds, L. (1987). Role of basement membranes in cell differentiation. Annals o f the New York Academy o f Sciences, volume 513, p. 134-145. [417] Gittes, G.K., Galante, P.E., Hanahan, D., Rutter, W J. and Debase, H.T. (1996). Lineage- specific morphogenesis in the developing pancreas: role of mesenchymal factors. Development, volume 122, p. 439-447. [418] Crisera, C.A., Kadison, A.S., Breslow, G.D., Maldonado, T.S., Longaker, M.T. and Gittes, G.K. (2000). Expression and role of laminin-1 in mouse pancreatic organogenesis. Diabetes, volume 49, p. 936-944. [419] Cirulli, V., Beattie, G.M., Klier, G., Ellisman, M., Ricordi, C., Quaranta, V., Frasier, F., Ishii, J.K., Hayek, A. and Salomon, D.R. (2000). Expression and function of AlphaVBeta3 and AlphaVBeta5 integrins in the developing pancreas. Journal o f Cell Biology, volume 150, p. 1445-1460. [420] Kleinman, H.K. and Martin, G.R. (2005). Matrigel: basement membrane matrix with biological activity. Seminars in Cancer Biology, volume 15, p. 378-386. [421] Ricard-Blum, S. and Ruggiero, F. (2005). The collagen superfamily: from the extracellular matrix to the cell membrane. Pathologie Biologie, volume 53, p. 430-442. [422] Wang, R.N. and Rosenberg, L. (1999). Maintenance of beta-cell function and survival following islet isolation requires re-establishment of the islet-matrix relationship. Journal o f Endocrinology, volume 163, p. 181-190. [423] Piper, K., Brickwood, S., Turnpenny, L.W., Cameron, I.T., Ball, S.G., Wilson, D.I. and Hanley, N.A. (2004). Beta cell differentiation during early human pancreas development. Journal o f Endocrinology, volume 181, p. 11-23. [424] Arias, A.M. (2001). Epithelial mesenchymal interactions in cancer and development. Cell, volume 105, p. 425-431. [425] Cano, A., Perez-Moreno, M.A., Rodrigo, I., Locascio, A., Blanco, M.J., del Barrio, M.G., Portillo, F. and Nieto, M.A. (2000). The transcription factor Snail controls epithelial-

199 mesenchymal transitions by repressing E-cadherin expression. Nature Cell Biology, volume 2, p. 76-83. [426] Huber, O., Bierkamp, C. and Kemler, R. (1996). Cadherins and catenins in development.

Current Opinion in Cell Biology, volume 8 , p. 685-691. [427] Dahl, U., Sjodin, A. and Semb, H. (1996). Cadherins regulate aggregation of pancreatic beta-cells in vivo. Development, volume 122, p. 2895-2902. [428] Wang, R., Li, J., Lyte, K., Yashpal, N.K., Fellows, F. and Goodyer, C.G. (2005). Role for

betal integrin and its associated alpha3, alpha5, and alpha 6 subunits in development of the human fetal pancreas. Diabetes, volume 54, p. 2080-2089. [429] Yashpal, N.K., Li, J., Wheeler, M.B. and Wang, R. (2005). Expression of betal integrin receptors during rat pancreas development-sites and dynamics. Endocrinology, volume 146, p. 1798-1807. [430] Hirano, M., Hashimoto, S., Yonemura, S., Sabe, H. and Aizawa, S. (2008). EPB41L5 functions to post-transcriptionally regulate cadherin and integrin during epithelial- mesenchymal transition. Journal o f Cell Biology, volume 182, p. 1217. [431] Coffman, F.D. and Studzinski, G.P. (1999). Differentiation-related mechanisms which suppress DNA replication. Experimental Cell Research, volume 248, p. 58-73. [432] Lumelsky, N., Blondel, O., Laeng, P., Velasco, I., Ravin, R. and McKay, R. (2001). Differentiation of embryonic stem cells to insulin-secreting structures similar to pancreatic islets. Science, volume 292, p. 1389-1394. [433] Vinik, A., Rafaeloff, R., Pittenger, G., Rosenberg, L. and Duguid, W. (1997). Induction of pancreatic islet neogenesis. Hormone and Metabolic Research, volume 29, p. 278-293. [434] Attali, M., Stetsyuk, V., Basmaciogullari, A., Aiello, V., Zanta-Boussif, M.A., Duvillie, B. and Scharfmann, R. (2007). Control of beta-cell differentiation by the pancreatic mesenchyme. Diabetes, volume 56, p. 1248-1258. [435] Duvillie, B., Attali, M., Bounacer, A., Ravassard, P., Basmaciogullari, A. and Scharfmann, R. (2006). The mesenchyme controls the timing of pancreatic beta-cell differentiation. Diabetes, volume 55, p. 582-589. [436] Zertal-Zidani, S., Bounacer, A. and Scharfmann, R. (2007). Regulation of pancreatic endocrine cell differentiation by sulphated proteoglycans. Diabetologia, volume 50, p. 585- 595.

200 [437] Lehmann, R., Zuellig, R.A., Kugelmeier, P., Baenninger, P.B., Moritz, W., Perren, A., Clavien, P. A., Weber, M. and Spinas, G. A. (2007). Superiority of small islets in human islet transplantation. Diabetes , volume 56, p. 594-603. [438] MacGregor, R.R., Williams, S J., Tong, P.Y., Kover, K., Moore, W.V. and Stehno-Bittel, L. (2006). Small rat islets are superior to large islets in in vitro function and in transplantation outcomes. American Journal o f Physiology: Endocrinology and Metabolism, volume 290, p. E771-E779. [439] Bosco, D., Meda, P., Halban, P.A. and Rouiller, D.G. (2000). Importance of cell-matrix

interactions in rat islet beta-cell secretion in vitro: role of alpha 6 betal integrin. Diabetes, volume 49, p. 233-243. [440] Woods, E.J., Walsh, C.M., Sidner, R.A., Zieger, M.A.J., Lakey, J.R.T., Ricordi, C. and Critser, J.K. (2004). Improved in vitro function of islets using small intestinal submucosa. Transplantation Proceedings, volume 36, p. 1175-1177. [441] Dubiel, E.A., Martin, Y. and Vermette, P. (2011). Bridging the gap between physicochemistry and interpretation prevalent in cell-surface interactions. Chemical Reviews, volume 111, p. 2900-2936. [442] Wong, L.S., Khan, F. and Micklefield, J. (2009). Selective covalent protein immobilization: Strategies and applications. Chemical Reviews, volume 109, p. 4025-4053. [443] Guerra, N.B., Gonzalez-Garcia, C., Llopis, V., Rodriguez-Hemandez, J.C., Moratal, D., Rico, P. and Salmeron-Sanchez, M. (2010). Subtle variations in polymer chemistry modulate

substrate stiffness and fibronectin activity. Soft Matter, volume 6 , p. 4748-4755. [444] Lan, M.A., Gersbach, C.A., Michael, K.E., Keselowsky, B.G. and Garcia, A.J. (2005). Myoblast proliferation and differentiation on fibronectin-coated self assembled monolayers presenting different surface chemistries. Biomaterials, volume 26, p. 4523-4531. [445] Lee, M.H., Ducheyne, P., Lynch, L., Boettiger, D. and Composto, R.J. (2006). Effect of biomaterial surface properties on fibronectin-alpha5betal integrin interaction and cellular attachment. Biomaterials, volume 27, p. 1907-1916. [446] Morra, M. (2000). On the molecular basis of fouling resistance. Journal o f Biomaterials Science, Polymer Edition, volume 11, p. 547-569. [447] Kenausis, G.L., Voros, J., Elbert, D.L., Huang, N., Hofer, R., Ruiz-Taylor, L., Textor, M., Hubbell, J.A. and Spencer, N.D. (2000). Poly(L-lysine)-g-poly(ethylene glycol) layers on

201 metal oxide surfaces: Attachment mechanism and effects of polymer architecture on resistance to protein adsorption. Journal o f Physical Chemistry B, volume 104, p. 3298-3309.

[448] VandeVondele, S., V 6 r6 s, J. and Hubbell, J.A. (2003). RGD-grafted poly-L-lysine-graft- (polyethylene glycol) copolymers block non-specific protein adsorption while promoting cell adhesion. Biotechnology and Bioengineering, volume 82, p. 784-790. [449] Ruoslahti, E. (1996). RGD and other recognition sequences for integrins. Annual Review o f Cell and Developmental Biology, volume 12, p. 697-715. [450] Cacace, M.G., Landau, E.M. and Ramsden, J.J. (1997). The Hofmeister series: salt and solvent effects on interfacial phenomena. Quarterly Reviews o f Biophysics, volume 30, p. 241-277. [451] Ducker, W.A., Senden, T.J. and Pashley, R.M. (1991). Direct measurement of colloidal forces using an atomic force microscope. Nature, volume 353, p. 239-241. [452] Cleveland, J.P., Manne, S., Bocek, D. and Hansma, P.K. (1993). A nondestructive method for determining the spring constant of cantilevers for scanning force microscopy. Review o f Scientific Instruments, volume 64, p. 403-405. [453] Krishnamurthy, M., Li, J., Al-Masri, M. and Wang, R. (2008). Expression and function of api integrins in pancretic beta (INS-1) cells. Journal o f Cell Communication and Signaling, volume 2, p. 67-79. [454] Sabra, G. and Vermette, P. (2011). Endothelial cell responses towards low-fouling surfaces bearing RGD in a three-dimensional environment. Experimental Cell Research, volume 317, p. 1994-2006. [455] Sakai, K., Fujii, T. and Hayashi, T. (1996). Conformational change precedes the formation of multimeric fibronectin. Journal o f Biochemistry, volume 119, p. 58-62. [456] Patel, S., Chaffotte, A.F., Goubard, F. and Pauthe, E. (2004). Urea-induced sequential unfolding of fibronectin: a fluorescence spectroscopy and circular dichroism study. Biochemistry, volume 43, p. 1724-1735. [457] Nelea, V., Nakano, Y. and Kaartinen, M.T. (2008). Size distribution and molecular associations of plasma fibronectin and fibronectin crosslinked by transglutaminase 2. Protein Journal, volume 27, p. 223-233. [458] Guo, X. and Ballauff, M. (2000). Spatial dimensions of colloidal polyelectrolyte brushes as determined by dynamic light scattering. Langmuir, volume 16, p. 8719-8726.

202 [459] Koster, S., Lin, Y.-C., Herrmann, H. and Weitz, D.A. (2010). Nanomechanics of vimentin

intermediate filament networks. Soft Matter, volume 6 , p. 1910-1914. [460] Lin, Y.-C., Yao, N.Y., Broedersz, C.P., Herrmann, H., MacKintosh, F.C. and Weitz, D.A. (2010). Origins of elasticity in intermediate filament networks. Physical Review Letters, volume 104, p. 058101. [461] Barbucci, R., Magnani, A., Chiumiento, A., Pasqui, D., Cangioli, I. and Lamponi, S. (2005). Fibroblast cell behavior on bound and adsorbed fibronectin onto hyaluronan and

sulfated hyaluronan substrates. Biomacromolecules, volume 6 , p. 638-645. [462] Zhang, Y., Chai, C., Jiang, X.S., Teoh, S.H. and Leong, K.W. (2007). Fibronectin immobilized by covalent conjugation or physical adsorption shows different bioactivity on aminated-PET. Materials Science and Engineering: C, volume 27, p. 213-219. [463] Biran, R., Webb, K., Noble, M.D. and Tresco, P.A. (2001). Surfactant-immobilized fibronectin enhances bioactivity and regulates sensory neurite outgrowth. Journal o f Biomedical Materials Research, volume 55, p. 1-12. [464] Kiriukhin, M.Y. and Collins, K.D. (2002). Dynamic hydration numbers for biologically important ions. Biophysical Chemistry, volume 99, p. 155-168. [465] Paraskevas, S., Maysinger, D., Wang, R., Duguid, W.P. and Rosenberg, L. (2000). Cell loss in isolated human islets occurs by apoptosis. Pancreas, volume 20, p. 270-276. [466] Daoud, J.T., Petropavlovskaia, M.S., Patapas, J.M., Degrandpre, C.E., DiRaddo, R.W., Rosenberg, L. and Tabrizian, M. (2011). Long-term in vitro human pancreatic islet culture using three-dimensional microfabricated scaffolds. Biomaterials, volume 32, p. 1536-1542. [467] Andrades, P., Asiedu, C., Rodriguez, C., Goodwin, K.J., McCam, J. and Thomas, J.M. (2007). Subcutaneous pancreatic islet transplantation using fibrin glue as a carrier. Transplantation Proceedings, volume 39, p. 191-192. [468] Beattie, G.M., Montgomery, A.M.P., Lopez, A.D., Hao, E., Perez, B., Just, M.L., Lakey, J.R.T., Hart, M.E. and Hayek, A. (2002). A novel approach to increase human islet cell mass while preserving p-cell function. Diabetes, volume 51, p. 3435-3439. [469] Lim, J.-Y., Min, B.-H., Kim, B.-G., Han, H.-J., Kim, S.-J., Kim, C.-W., Han, S.-S. and Shin, J.-S. (2009). A fibrin gel carrier system for islet transplantation into kidney subcapsule. Acta Diabetologica, volume 46, p. 243-248.

203 [470] Maillard, E., Juszczak, M.T., Clark, A., Hughes, S.J., Gray, D.R.W. and Johnson, P.R.V. (2011). Perfluorodecalin-enriched fibrin matrix for human islet culture. Biomaterials, volume 32, p. 9282-9289. [471] van Hinsbergh, V.W.M., Collen, A. and Koolwijk, P. (2001). Role of fibrin matrix in angiogenesis. Annals o f the New York Academy o f Sciences, volume 936, p. 426-437. [472] Johansson, U., Rasmusson, I., Niclou, S.P., Forslund, N., Gustavsson, L., Nilsson, B., Korsgren, O. and Magnusson, P.U. (2008). Formation of composite endothelial cell- mesenchymal stem cell islets. Diabetes, volume 57, p. 2393-2401. [473] Lammert, E., Cleaver, O. and Melton, D. (2001). Induction of pancreatic differentiation by signals from blood vessels. Science, volume 294, p. 564-567. [474] Ballian, N. and Brunicardi, F.C. (2007). Islet vasculature as a regulator of endocrine pancreas function. World Journal o f Surgery, volume 31, p. 705-714. [475] Eberhard, D., Kragl, M. and Lammert, E. (2010). Giving and taking : endothelial and P- cells in the islets of Langerhans. Trends in Endocrinology and Metabolism, volume 21, p. 457-463. [476] Johansson, U., Elgue, G., Nilsson, B. and Korsgren, O. (2005). Composite islet- endothelial cell grafts: A novel approach to counteract innate immunity in islet transplantation. American Journal o f Transplantation, volume 5, p. 2632-2639. [477] Linn, T., Schneider, K., Hammes, H.P., Preissner, K.T., Brandhorst, H., Morgenstem, E., Kiefer, F. and Bretzel, R.G. (2003). Angiogenic capacity of endothelial cells in islets of Langerhans. Federation of American Societies for Experimental Biology Journal, volume 17, p. 881-883. [478] Brissova, M., Fowler, M., Wiebe, P., Shostak, A., Shiota, M., Radhika, A., Lin, P.C., Gannon, M. and Powers, A.C. (2004). Intraislet endothelial cells contribute to revascularization of transplanted pancreatic islets. Diabetes, volume 53, p. 1318-1325. [479] Nyqvist, D., Kohler, M., Wahlstedt, H. and Berggren, P.-O. (2005). Donor islet endothelial cells participate in formation of functional vessels within pancreatic islet grafts. Diabetes, volume 54, p. 2287-2293. [480] Bharat, A., Saini, D., Benshoff, N., Goodman, J., Desai, N.M., Chapman, W.C. and Mohanakumar, T. (2007). Role of intra-islet endothelial cells in islet allo-immunity. Transplantation, volume 84, p. 1316-1323

204 [481] Savinov, A.Y., Wong, F.S., Stonebraker, A.C. and Chervonsky, A.V. (2003). Presentation of antigen by endothelial cells and chemoattraction are required for homing of insulin-

specific CD 8 + T cells. Journal of Experimental Medicine, volume 197, p. 643-656. [482] Nyqvist, D., Speier, S., Rodriguez-Diaz, R., Molano, R.D., Lipovsek, S., Rupnik, M., Dicker, A., Ilegems, E., Zahr-Akrawi, E., Molina, J., Lopez-Cabeza, M., Villate, S., Abdulreda, M.H., Ricordi, C., Caicedo, A., Pileggi, A. and Berggren, P.-O. (2011). Donor islet endothelial cells in pancreatic islet revascularization. Diabetes, volume 60, p. 2571- 2577. [483] Dubiel, E.A., Kuehn, C., Wang, R. and Vermette, P. (2012). In vitro morphogenesis of PANC-1 cells into islet-like aggregates using RGD-covered dextran derivative surfaces. Colloids and Surfaces B: Biointerfaces, volume 89, p. 117-125. [484] London, N.J., Swift, S.M. and Clayton, H.A. (1998). Isolation, culture and functional evaluation of islets of Langerhans. Diabetes & Metabolism, volume 24, p. 200-207. [485] Buchwald, P. (2009). FEM-based oxygen consumption and cell viability models for

avascular pancreatic islets. Theoretical Biology and Medical Modelling, volume 6 , p. 5. [486] Gu, G., Dubauskaite, J. and Melton, D.A. (2002). Direct evidence for the pancreatic lineage: NGN3+ cells are islet progenitors and are distinct from duct progenitors. Development, volume 129, p. 2447-2457. [487] Dor, Y., Brown, J., Martinez, O.I. and Melton, D.A. (2004). Adult pancreatic beta-cells are formed by self-duplication rather than stem-cell differentiation. Nature, volume 429, p. 41-46. [488] Bonner-Weir, S., Inada, A., Yatoh, S., Li, W.C., Aye, T., Toschi, E. and Sharma, A. (2008). Transdifferentiation of pancreatic ductal cells to endocrine beta-cells. Biochemical Society Transactions, volume 36, p. 353-356. [489] Ouziel-Yahalom, L., Straussman, S., Li, W.-C., Levine, J., Shaoping, D. and Bonner- Weir, S. (2010) Expansion of pancreatic duct cells in culture results in loss of transcriptional profile that favors neogenesis. 7(fh Scientific Session o f the American Diabetes Association. Orlando, FL, United States. [490] Galli Marxer, C., Collaud Coen, M. and Schlapbach, L. (2003). Study of adsorption and viscoelastic properties of proteins with a quartz crystal microbalance by measuring the oscillation amplitude. Journal o f Colloid and Interface Science, volume 261, p. 291-298.

205 [491] Hemmersam, A.G., Foss, M., Chevallier, J. and Besenbacher, F. (2005). Adsorption of fibrinogen on tantalum oxide, titanium oxide and gold studied by the QCM-D technique. Colloids and Surfaces B: Biointerfaces, volume 43, p. 208-215. [492] Hemmersam, A.G., RechendorfF, K., Foss, M., Sutherland, D.S. and Besenbacher, F. (2008). Fibronectin adsorption on gold, Ti-, and Ta-oxide investigated by QCM-D and RSA modelling. Journal o f Colloid and Interface Science, volume 320, p. 110-116. [493] Bramfeldt, H., Sabra, G., Centis, V. and Vermette, P. (2010). Scaffold vascularization: A challenge for three-dimensional tissue engineering. Current Medicinal Chemistry, volume 17, p. 3944-3967 [494] Chouinard, J.A., Gagnon, S., Couture, M.G., Levesque, A. and Vermette, P. (2009). Design and validation of a pulsatile perfusion bioreactor for 3D high cell density cultures. Biotechnology and Bioengineering, w olume 104, p. 1215-1223. [495] Shkilnyy, A., Dubois, J., Sabra, G., Sharp, J., Gagnon, S., Proulx, P. and Vermette, P. (2012). Bioreactor controlled by PI algorithm and operated with a perfusion chamber to support endothelial cell survival and proliferation. Biotechnology and Bioengineering, volume 109, p. 1305-1313.

206 APPENDIX A

RGD-CT/'f) ” HF-CMO

Figure A.I: Phase contrast microscopy of adherent (RGD-CMD) surfaces and non-adherent (RGE-CMD, CMD, and TCPS) surfaces.

207 Figure A.2: Stained sections for GAG using Alcian Blue of (A) PANC-1 ILAs (14 days) and (B) mouse intestine as a positive control.

208 3 Days 7 Days 14 Days

RGD-CMO

RGE-CMD

CMD

TOPS

Figure A.3: Immunofluorescence staining of a 5 mtegnn of cells grown on RGD-CMD (A-C), RGE-CMD (D-F), CMD (G-I), and TCPS (J-L).

209 3 Days 7 Days 14 Days

RGD-CMD

RGE-CMD

Figure A.4: Immunofluorescence staining of Ovfc integrin of cells grown on RGD-CMD (A-C), RGE-CMD (D-F), CMD (G-I), and TCPS (J-L).

210 APPENDIX B

-HApp (A) HApp (B) -CMD CM) 1.00 lO ^C aQ j 1x PBS + lOOiflftrt. FN (act) 10^M C ad j + KXWmL FN (no act) ISOrrM N ad + 100^m L FN (act) 10v« C ad , + 1(XWrrtL FN (act) - - 10pM C ad 3 + lOOtgfrnL FN (a d ) 10»M MgOj * 100|^rrL FN (act) 0.75- 10jjM M O , + 100|^rrt_ FN (act)

050-

0.25-

0.00 282 2Si __ 280 282 284 288 290286 292 BlncSng Energy (eV) B ndng Energy (eV)

(C) -HApp (D) -HApp -CMD CMD 1.00-1 10*M MgCl2 — 10^N hC lj 1CHM M ga3 + 100pgfmL FN (no a d ) — 1 0 ^ Mnd2 + KXWmL FN (no act.) - — 1CHW MgQ2 + lOQm'mL FN (act) — 1 0 ^ MiClj + 100j^mL FN (act)

075- 0.75-

0.50-

0.25- 025-

0.00 000 280 282 284 286 288 290 292 280 284 286 288 290 292 an d n g Energy (eV) Blndng Energy («V)

Figure B.l: High resolution C Is XPS spectra outlining differences between successful (lOpM CaCl2) and unsuccessful (lx PBS and 150mM NaCl) FN grafting to CMD surfaces (A) and to CMD surfaces activated (act.) and non-activated (no act.) with (B) CaCl2, (C) MgCl2, and (D). MnCl2.

211 Milli-Q Water 10pM CaCI2 ♦ 150mM NaCI ★ 1x PBS

E z E, oc ir

175 Separation Distance (nm)

Figure B.2: CMD AFM force - separation distance profiles in various solutions, Force is normalized to the radius of the sphere. Data are represented as average ± S.E.M (n=4).

212 (A) Glass0 , -Q W ater) (B) HApp (Milli-Q W ater)

(E) CMD (10pM MgCI2) (F) CMD (10pM MnCI2)

Figure B.3: 750nm x 750nm AFM images in liquid using Tapping™ mode of (A-B) intermediate surfaces and (C-F) CMD layers in various solutions. Brackets indicate the solution in which AFM imaging was performed.

213 APPENDIX C

Figure C.l: Time lapse microscopy of porcine liver micro vascular endothelial cells (PLMEC) and porcine islets embedded in fibrin gel over 6 days [i.e., PLMEC+PI(in Fibrin)] and PLMEC embedded in fibrin gel after 24h (i.e., 24h-PLMEC) and 6 days (i.e., 6 Days-PLMEC). All images recorded at 200x magnification.

214 Figure C.2: Time lapse microscopy of porcine islets embedded in fibrin gel [i.e., PI(in Fibrin)] over 6 days. All images recorded at 200x magnification.

215 Figure C.3: Time lapse microscopy of porcine islets seeded over a monolayer of porcine liver microvascular endothelial cells (i.e., PI-PLMEC monolayer). Single cells of fibroblastoid morphology appear as early as 24h of culture and proliferate to confluence during culture period. All images recorded at lOOx magnification.

216 Figure C.4: Time lapse microscopy of porcine islets seeded over tissue culture polystyrene (i.e., PI-TCPS). Single cells of fibroblastoid morphology appear early and proliferate throughout the culture period. All images recorded at 40x magnification.

217 Figure C.5: Immunofluorescence image of insulin and glucagon expression at 0 days (left). Immunohistochemistty image of apoptosis detected via TUNEL at 0 days (right).

r 10 D av M 2 Day* ■ B ? Days

0 ( } « ry *

Figure C.6: Level of cells expressing glucagon in all culture conditions. Data is represented as average ± S.E.M. There were no significant differences in glucagon expression between any culture conditions at any culture times.