Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Magnetically actuated MEMS devices for active control of cell migration Khademolhosseini, Farzad 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2015_may_khademolhosseini_farzad.pdf [ 49.42MB ]
Metadata
JSON: 24-1.0135703.json
JSON-LD: 24-1.0135703-ld.json
RDF/XML (Pretty): 24-1.0135703-rdf.xml
RDF/JSON: 24-1.0135703-rdf.json
Turtle: 24-1.0135703-turtle.txt
N-Triples: 24-1.0135703-rdf-ntriples.txt
Original Record: 24-1.0135703-source.json
Full Text
24-1.0135703-fulltext.txt
Citation
24-1.0135703.ris

Full Text

MAGNETICALLY ACTUATED MEMS DEVICES FOR ACTIVE  CONTROL OF CELL MIGRATION  by  Farzad Khademolhosseini  M.A.Sc., The University of British Columbia, 2009  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Mechanical Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  April 2015  © Farzad Khademolhosseini, 2015 ii  Abstract The ability of living cells to sense and respond to mechanical cues from the surrounding environment has been the subject of much study. Over the past two decades, a variety of techniques have been used to apply mechanical stimuli and investigate cell response. Recently, with advances in the field of Microelectromechanical Systems, devices incorporating microscale actuators have been developed to apply forces and study the cell response of individual cells. Among these microdevices, micropillar arrays incorporating remotely actuated magnetic pillars have shown some success as combined actuation and sensing platforms for cell strain studies. However, issues associated with the complex fabrication techniques used, the low actuation forces generated and the high magnetic field gradients required for pillar actuation have hindered the wide-scale adoption of these devices by the general research community. Consequently, investigation into the use of these active micropatterned surfaces in eliciting or controlling specific cellular response on a multicellular level has yet to be undertaken.  This thesis aims to investigate the application of remotely actuated micropillar surfaces in controlling the migration behavior of cells on a multicellular level. First, using a novel custom-made magnetically actuated cell strain assessment tool, conventional tests are performed on endothelial cells to determine the minimum strain requirements for eliciting cell response. Then, a new technique for fabrication and patterning of magnetic micropillar arrays is developed to overcome the complexities of previous fabrication methods. Using the newly developed fabrication technique, magnetic micropillar arrays of various dimensions are fabricated and their mechanical, magnetic and material properties are characterized. The fabricated magnetic micropillars generate forces of several hundred nanonewtons at moderate magnetic fields of iii  100mT and are favorable to previous state-of-the-art. Finally, a cell migration chip comprising various micropillar topologies is developed and the migration behavior and migration rates of sheets of cells on the micropillar surfaces in the presence and absence of micropillar actuation is studied using in-vitro experiments. We show that actuated micropillar surfaces significantly impede cell migration, reducing cell migration rates by up to 85%. The magnetically actuated micropillar surfaces could have possible in-vivo applications for preventing cell-migration induced biofouling of medical implants.   iv  Preface The research presented in this dissertation was carried out by the author at the University of British Columbia under the supervision of Dr. Mu Chiao at the Department of Mechanical Engineering. Some part of the research were conducted at the Child and Family Research Institute of the B.C. Women’s and Children’s Hospital in collaboration with Dr. James Lim from the Department of Pediatrics at the University of British Columbia.  Chapter 2 of the thesis presents the design and characterization of a new membrane based cell- strain device, and the utilization of said device to conduct substrate-strain experiments on cells. Part of the material presented in chapter 2 is based on the following publication:   F. Khademolhosseini, M. Chiao, "Magnetically actuated cellular strain assessment tool: a study on the strain induced directional migration of human endothelial cells", Proc. of the 16th int. Conf. on Solid-state Sensors, Actuators and Microsystems, pp. 2275-2278, 2011. All aspects of the work presented in chapter 2, including literature review, design, construction and fabrication of the cell-strain device, design and conducting of cell-strain experiments, processing and statistical analysis of data, writing and presentation of the conference paper and writing of the thesis manuscript were performed by the author of this thesis. The author received help and initial training in cell culture, cell staining and microscopy techniques from Dr. James Lim, Mr. Jack Liu and Mr. John Jackson at the University of British Columbia. v  Chapter 3 of the thesis presents a novel technique for the fabrication and patterning of polymer micropillar structures with embedded functional microparticles, i.e. micropillars with magnetic, electrical and/or fluorescent properties. The fabricated devices are characterised and compared with previous state-of-the-art techniques to demonstrate the advantages of the technique presented in the thesis. Part of the material presented in chapter 3 has been published in the IEEE Journal of Microelectromechanical Systems:	 F. Khademolhosseini and M. Chiao, "Fabrication and Patterning of Magnetic Polymer Micropillar Structures Using a Dry Nanoparticle Embedding Technique," IEEE/ASME Journal of Micro Electro Mechanical Systems, vol. 22 (1), pp. 131-139, 2013 Parts of chapter 3 were presented in the IEEE MEMS 2012 conference and published in the conference proceedings:		 F. Khademolhosseini and M. Chiao, "A dry nanoparticle embedding technique for fabrication of magnetic polymer micropillars," Proc. of IEEE 25th Int. Conf. on Micro-Electro-Mechanical-Systems, pp. 212-215, 2012 An 8-page manuscript based on the material presented in Chapter 3 and Appendix E, detailing the theoretical modeling and experimental characterization of the base tilting effect for the two-phase magnetic micropillar structures fabricated, has been prepared and is ready for submission to the Journal of Micromechanics and Microengineering:   F. Khademolhosseini and M. Chiao, "Experimental Characterization of the Base Tilting Effect and Determination of the Tilting Coefficient of PDMS Micropillar Structures".  vi  All aspects of the work presented in chapter 3, including literature review, design, development and optimization of the novel microfabrication technique, fabrication of the magnetic micropillar arrays, characterization of the micropillar properties, processing and statistical analysis of data, writing of the Journal and Conference papers and writing of the thesis manuscript were performed by the author of this thesis. The author provided training for and in-return received help from undergraduate students, Mr. Carl Xing and Mr. Andrew Wang, with conducting repetitive experimental measurements of the base tilting of magnetic micropillars, once the general procedures and processes were developed and established by the author. Chapter 4 presents proof-of-concept cell migration experiments based on wound healing assays conducted on micropillar migration chips, where we demonstrate the efficacy of actuated micropillar arrays in controlling cell migration. A paper based on the results obtained in chapter 4 was presented at the IEEE MEMS 2015 conference and published in the conference proceedings:  F. Khademolhosseini, C-C Liu, C.J. Lim and M. Chiao, "Application of Periodic Loads on Cells from Magnetic Micropillar Arrays Impedes Cellular Migration," Proceedings of IEEE 28th International Conference on Micro-Electro-Mechanical-Systems, pp. 624-627, 2015. All aspects of the work presented in chapter 4, including literature review, design, fabrication and optimization of the multi-region cell-migration chips, conducting of the cell migration experiments and wound healing assays, processing and statistical analysis of data, writing of the conference paper and writing of the thesis manuscript were performed by the author of this vii  thesis. The author received help from Mr. Parham Pournazari, an undergraduate student working for Dr. Chiao, with the purchasing and setup of parts for a permanent magnet actuator required for conducting of cell strain experiments, and help with imaging techniques and cell fixing and staining procedures from Dr. James Lim and Mr. Jack Liu. The author provided training for and in-return received help from undergraduate students, Mr. Carl Xing and Mr. Andrew Wang, with post-processing of images acquired from cell migration experiments, once the general procedures and processes were developed and established by the author. The author has made the following major contributions in the work presented in this dissertation: 1. Developed a novel fabrication technique, i.e., the dry nanoparticle embedding technique, for the fabrication and patterning of polymer micropillar structures having micropillars embedded with functional particles, i.e., magnetic, electrically conductive or fluorescent particles. 2. Characterised the magnetic and mechanical properties, and the material composition of magnetic micropillar arrays developed using the dry nanoparticle embedding technique and demonstrated the advantages of the technique compared to previous state-of-the-art approaches in obtaining higher particle loading ratios in the micropillars, thus significantly enhancing their magnetic properties. 3. Conducted in-vitro cell migration experiments and wound healing assays to study the effects of periodic loads from magnetic micropillar arrays on cell migration behavior. This included: a. Design, development, fabrication and optimization of a multi-region micropillar cell-migration chip to study the effect of force application for different micropillar topologies. viii  b. Design of various cell migration experiments and the design and construction of multiple experimental setups required for conducting the experiments, including custom-made permanent-magnet actuators, custom-made electromagnet actuators and custom made housings for cell-culture. c. Conducting of short term and long-term wound healing assays to study the short-term and long-term efficacy of the micropillar devices in controlling cell migration. 4. Design and development of a new, low-cost and easily replicable device for conventional cell strain experiments. 5. Conducting of conventional cell-strain experiments on Human Umbilical Vein Endothelial Cells to study the cell response to strain and to find the minimum strain thresholds required for triggering cell response. ix  Table of Contents  Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iv Table of Contents ......................................................................................................................... ix List of Tables .............................................................................................................................. xiii List of Figures ............................................................................................................................. xiv List of Symbols .............................................................................................................................xx List of Abbreviations ................................................................................................................. xxi Acknowledgements ................................................................................................................... xxii Dedication ................................................................................................................................. xxiii Chapter 1: Introduction ................................................................................................................1 1.1  Cellular Mechanotransduction ........................................................................................ 3 1.2  Cell Migration and Proliferation ..................................................................................... 4 1.3  Studying the Effects of Mechanical Stimuli on Cells ..................................................... 6 1.3.1  Conventional Devices for Cell-Strain/Stress .............................................................. 7 1.3.2  MEMS Devices for Cell-Strain/Stress ........................................................................ 9 1.3.3  Considerations for Development of MEMS Devices for Cell-Mechanics ............... 11 1.4  Micropillar Arrays for Cell-Mechanics Studies ............................................................ 12 1.4.1  Micropillar Arrays as Sensors ................................................................................... 12 1.4.2  Micropillar Arrays as Actuators ................................................................................ 13 1.4.3  Limitations of Current Magnetic Micropillar Actuators ........................................... 14 1.4.3.1  Fabrication of Magnetic Polymer Micropillar Actuators .................................. 14 x  1.4.3.2  Magnetic Field Requirements ........................................................................... 16 1.4.4  Micropillar Actuators as a Tool to Govern Cellular Behavior in Vivo .................... 16 1.5  Thesis Objectives and Overview of Thesis ................................................................... 18 Chapter 2: Magnetically Actuated Cellular Strain Assessment Tool: A Study on the Effects of Cyclic Substrate Strains on Human Endothelial Cells.........................................................22 2.1  The Magnetically Actuated Cellular Strain Assessment Tool (MACSAT) .................. 24 2.1.1  MACSAT Design...................................................................................................... 24 2.1.2  Strain Distribution in MACSAT Membranes ........................................................... 26 2.1.3  Directions of Minimum Deformation in MACSAT Membranes ............................. 29 2.2  Cell Strain Experiments with MACSAT ...................................................................... 32 2.2.1  HUVEC Culture ........................................................................................................ 32 2.2.2  Preparation of Culture Dishes ................................................................................... 32 2.2.3  Cell Re-orientation Experiments ............................................................................... 33 2.2.4  Actin Re-orientation and Alignment Experiments ................................................... 37 2.2.5  Discussion of Results ................................................................................................ 43 2.3  Conclusion .................................................................................................................... 47 Chapter 3: Dry Nanoparticle Embedding Technique for Fabrication and Patterning of Magnetic Micropillar Arrays ......................................................................................................48 3.1  Fabrication .................................................................................................................... 51 3.1.1  The Dry Particle Embedding Technique................................................................... 51 3.1.2  Physical Embedding of Nonmagnetic Particles ........................................................ 55 3.1.3  Controlled Patterning of the Micropillar Device ...................................................... 56 3.2  Characterization ............................................................................................................ 59 xi  3.2.1  Material Composition of the Micropillars ................................................................ 59 3.2.2  Young’s Modulus of FeC-PDMS ............................................................................. 61 3.2.3  Magnetization Properties of the Micropillars ........................................................... 63 3.2.4  Bending Performance of the Micropillars ................................................................. 67 3.3  Conclusion .................................................................................................................... 71 Chapter 4: Controlling Cell Migration with Magnetic Polymer Micropillar Arrays ...........72 4.1  Design Considerations for Cell-Migration Chip ........................................................... 72 4.1.1  Micropillar Size and Spacing .................................................................................... 73 4.1.2  Controlling for Direction of Force Application vs. Cell Migration .......................... 76 4.1.3  Controlling for the Net Effects of Force Application from Micropillars .................. 76 4.1.4  The Multi-region Cell-Migration Chip ..................................................................... 77 4.2  Fabrication of Cell-Migration Chip .............................................................................. 78 4.3  Experiments .................................................................................................................. 81 4.3.1  Cell Culture and Chip Preparation ............................................................................ 81 4.3.2  Application of Strains on Cells from Magnetic Pillars ............................................. 82 4.3.3  Simulated Scratch-Wound Assay ............................................................................. 85 4.3.4  Long-Term Cell-Migration Assay ............................................................................ 94 4.3.5  Experiments with Live Imaging................................................................................ 97 4.3.6  Discussion of Results .............................................................................................. 102 4.4  Concluding Remarks ................................................................................................... 104 Chapter 5: Conclusions and Future Work ..............................................................................106 5.1  Summary ..................................................................................................................... 107 5.2  Future Work ................................................................................................................ 110 xii  5.2.1  Micropillar Chip Design Variations........................................................................ 111 5.2.2  Variations in Actuation Parameters ........................................................................ 112 5.2.3  Cell Proliferation Studies ........................................................................................ 112 5.2.4  Long-term Biocompatibility Studies ....................................................................... 113 5.2.5  In-vivo Application Studies .................................................................................... 113 Bibliography ...............................................................................................................................115 Appendices ..................................................................................................................................132 Appendix A :How to Culture Cells on Plateau Region of MACSAT Membrane .................. 132 Appendix B : OrientationJ Plugin for Angular Orientation Analysis ..................................... 133 Appendix C : Axial Strain Threshold for Triggering Actin Reorientation ............................. 136 Appendix D : Masking Technique for Controlled Patterning of Micropillar Structures ........ 137 Appendix E : Theoretical Modeling and Experimental Characterization of the Base Tilting Effect of FeC-PDMS Micropillar Structures .......................................................................... 138 Appendix F : Permanent Magnet Actuator for Cell Migration Tests ..................................... 157 Appendix G : Electromagnet and Microscope Setup for Live Imaging ................................. 158 Appendix H : Path and Velocity Data Obtained from Live-Imaging Experiments ................ 159 Appendix I : Distributions of Cell Instantaneous Velocities .................................................. 161  xiii  List of Tables  Table 1.1 Advantages and disadvantages of the various types of MEMS devices and actuation mechanisms used for cell mechanics studies. Devices based on magnetic field actuation and devices using fluid flows have the most favorable s set of properties. ......................................... 11 Table  3.1: Comparison of magnetic particle to PDMS loading ratios for various techniques used to date ............................................................................................................................................ 60 Table  3.2: Maximum Displacement & Equivalent Force/ Equivalent Bending Moment on Magnetic Pillars ............................................................................................................................ 70 Table  4.1: Comparison of distances traveled and average velocities of HUVECs migrating among non-magnetic and actuated magnetic micropillar arrays (square pattern, 110 µm spacing)....................................................................................................................................................... 99 Table  4.2: Comparison of percentages of total time spent at different movement speeds based on calculated instantaneous velocities I.V. of 25 cells migrating among non-magnetic and 25 cells migrating among actuated magnetic micropillar arrays (square pattern, 110 µm spacing) ........ 101 Table H1: Path length and velocity data for 25 cells migrating among actuated magnetic pillars as obtained from live-imaging experiments ................................................................................ 159 Table  H2: Path length and velocity data for 25 cells migrating among non-magnetic pillars as obtained from live-imaging experiments .................................................................................... 160 xiv  List of Figures  Figure  1.1: Schematic of cell-ECM interaction and transmission of mechanical cues (mechanical loads, stresses and strains) .............................................................................................................. 4 Figure  1.2: Schematic of the cell migration phases for endothelial cells ....................................... 5 Figure  2.1: Schematic design of MACSAT. ................................................................................. 25 Figure  2.2: FEM Simulation results and experimental measurement (n=3) of strain distribution in the silicone-elastomer membrane of the MACSAT. ..................................................................... 27 Figure  2.3: Directions of minimum deformation in a elastic membrane undergoing uniaxial stretching ....................................................................................................................................... 30 Figure  2.4: Operating steps for cell culture and application of tensile strains (in the radial direction) on cells growing on flexible substrates using the MACSAT. ...................................... 33 Figure  2.5: Raw and color coded images showing reorientation of HUVECs due to 6% cyclic strain @ 1 Hz ................................................................................................................................ 35 Figure  2.6: Time evolution of the distribution of cell angles for cells undergoing 6% cyclic strain @ 1 Hz .......................................................................................................................................... 36 Figure  2.7: Comparison of actin alignment in a control cell (no cylcic strain) compared to a cyclically strained cell ................................................................................................................... 38 Figure  2.8: Distribution of actin orientation angles for the control cell and the strained cell of Figure 2.7 ...................................................................................................................................... 40 xv  Figure  2.9: Distribution of actin orientation angles for strained cells vs. control cells ................ 41 Figure  2.10: Coherency of actin alignment for strained cells vs. control cells ............................. 42 Figure  2.11: Reorientation behavior of a closely spaced group of HUVECs having cell-cell contact ........................................................................................................................................... 43 Figure  3.1: Schematic diagram of fabrication steps using the dry nanoparticle embedding technique ....................................................................................................................................... 52 Figure  3.2: SEM images of fabricated magnetic PDMS pillars.................................................... 54 Figure  3.3: Application of the dry nanoparticle embedding technique to fabricate fluorescent PDMS pillars and Electrically conductive PDMS pillars, on a transparent PDMS base ............. 56 Figure  3.4: Devices consisting of 40 μm magnetic FeC-PDMS pillars (black) alongside non-magnetic pure PDMS pillars (clear) patterned using the described masking technique and the dry nanoparticle embedding method ................................................................................................... 58 Figure  3.5: Material composition of FeC-PDMS magntic pillars obtained using Energy-dispersive X-ray spectroscopy analysis ........................................................................................ 59 Figure  3.6: Stress-strain curves and Young's moduli of pure PDMS and FeC-PDMS. ............... 62 Figure  3.7: Schematic of magnetic tranlational forces and magnetic bending moments applied to a magnetic polymer pillar in a uniform magnetic field ................................................................. 64 Figure  3.8: Anisotropic magnetization behavior of FeC-PDMS pillars for magnetic fields  applied orthogonal to and parallel to the pillar axis ...................................................................... 66 xvi  Figure  3.9: Simultaneous actuation of multiple magnetic PDMS pillars using an externally applied magnetic field. .................................................................................................................. 68 Figure  3.10: Measured pillar tip horizontal displacement and calculated equivalent horizontal tip force on pillars vs. externally applied magnetic field and magnetic field gradient ...................... 69 Figure  4.1: Schematic of cell attachement and force application when cells are cultured on top of a magnetic micropillar array ......................................................................................................... 74 Figure  4.2: Schematic of cell attachement and force application when cells are cultured among the pillars of a magnetic micropillar array .................................................................................... 75 Figure  4.3: Schematic of cell-migration chip, designed with nine different regions .................... 77 Figure  4.4: Schematic of fabrication steps for cell-migration chip .............................................. 79 Figure  4.5: Microscope images of a fabricated multi-region cell-migration chip having magnetic and non-magnetic pillars and a combination of various micropillar topologies ........................... 80 Figure  4.6: Pillars coated with a fluorescently labeled Fibronectin (FN) coating demonstrate uniform coating of all micropillar regions. ................................................................................... 82 Figure  4.7: Microscope image of an endothelial cell attached to a magnetic pillar on one side undergoing a stretch of approximately 5%  .................................................................................. 83 Figure  4.8: Fluorescently stained HUVECs growing among non-magnetic and actuated magnetic pillars show different morphologies ............................................................................................. 84 Figure  4.9: Representative fluorescent microscopy images of a 48 hour scratch-wound assay experiment looking at cell migration in the X-direction for the case of X-actuation. .................. 86 xvii  Figure  4.10: Representative fluorescent microscopy images of a 48 hour scratch-wound assay experiment looking at cell migration in the X-direction for the case of Y-actuation ................... 87 Figure  4.11: Measured data (n=3) for the average X-direction cell migration rates for the case of no actuation, i.e., when no external magnetic field was applied.. ................................................ 88 Figure  4.12: Measured data (n=3) for the average X-direction cell migration rates for X-actuation and Y-actuation of magnetic pillars.. ............................................................................................ 89 Figure  4.13: Non-dimensional migration rate (NDMR) of HUVECs among magnetic pillar arrays of various topologies for the three different actuation scenarios, i.e., no actuation (no external magnetic field applied), X-actuation and Y-actuation .................................................................. 91 Figure  4.14: Sample fluorescent microscopy images of a long-term cell-migration experiment looking at cell migration in the X-direction for the case of X-actuation ...................................... 95 Figure  4.15: Short-term (1-day) and Long-term (12-day) non-dimensional migration rate of HUVECs among magnetic and non-magnetic pillar arrays with square pattern and interpillar spacing of 110 µm (n=3) ............................................................................................................... 96 Figure  4.16: Sample images of a 24 hour wound-healing assay performed on a micropillar array chip (square pattern @ 110 µm spacing) ...................................................................................... 98 Figure  B1: OrientationJ dialogue box. ........................................................................................ 134 Figure C1: Magnitude of axial strain vs. angular orientation in the plateau region of the MACSAT .................................................................................................................................... 136 Figure  D1: Fabrication steps for patterning of magnetic micropillars among non-magnetic micropillars in one polymer casting step .................................................................................... 137 xviii  Figure  E1: Schematic of magnetic forces and moments acting on a magnetic pillar in the presence of a magnetic field B .................................................................................................... 138 Figure E2: The total deformation of a polymer pillar structure (elastic pillar on elastic base) under the action of a bending moment consists of pillar bending and base tilting ..................... 139 Figure E3: Schematic drawing of the electromagnet design. ..................................................... 145 Figure  E4: Side view of a magnetic FeC-PDMS micropillar on a pure PDMS substrate: application of the magnetic field causes pillar bending and base tilting .................................... 146 Figure  E5: (a) Stress-strain curves for pure PDMS and 40% w/w FeC-PDMS (b) Young’s Moduli of pure PDMS, 20% w/w FeC-PDMS and 40% w/w FeC-PDMS obtained from the slopes of stress-strain curves of part (a). ..................................................................................... 148 Figure  E6: Measurements obtained for the magnitude of the magnetic field in the 6 mm long air gap of the electromagnet ............................................................................................................. 149 Figure E7: Side-view of FeC-PDMs magnetic micropillar on a pure PDMS base as seen under an optical microscope. ..................................................................................................................... 149 Figure  E8: Measurements obtained for the tilting angles θ of a group of micropillars (L=120µm, D=24µm, L/D=5) at various bending angles α ........................................................................... 150 Figure  E9: Measurements obtained for the ratio of bending and tilting angles for micropillars with various aspect ratios ............................................................................................................ 151 Figure  E10: Contributions of (a) the pillar bending angle α and (b) the pillar tilting angle θ, to the total angular deformation β of the pillar structure ...................................................................... 152 xix  Figure  E11: Correction factor needed to obtain the correct values of bending moment or substrate stiffness in studies where the base tilting effect has not been accounted for .............................. 155 Figure  F1: Permanent magnet actuator setup for application of periodic magnetic field on micropillar chips ......................................................................................................................... 157 Figure  G1: Image of incubated microscope setup, the custom made electromagnet and the custom-designed and 3D printed microchip housing used for cell migration tests with live imaging ....................................................................................................................................... 158 Figure  I1: Distribution of cell velocities based on percentage of time spent at a certain instantaneous velocity ................................................................................................................. 161 Figure  I2: Boxplots showing the distribution of cell velocities based on percentage of time spent at a certain instantaneous velocity .............................................................................................. 162  xx  List of Symbols  DM  Membrane diameter  DPM  Permanent magnet diameter DX Total dist. traveled in X-direction  DY Total dist. traveled in Y-direction  E Young’s modulus F Magnetic translational force I Area moment of Inertia I.V.  Instantaneous velocity of cells l length M Magnetic bending moment m Magnetization  RXY Ratio of DX to DY t Membrane thickness  V Vertical displacement of center of membrane  ε Axial strain εr Radial strain  εθ Circumferential strain  ݒ Poisson’s ratio    xxi  List of Abbreviations  DIC Differential Interference Contrast EGM Endothelial Growth Media FeC Carbonyl Iron FeC-PDMS Carbonyl Iron- Polydimethylsiloxane FITC Fluorescein Isothiocyanate FN Fibronectin HMDS Hexamethyldisilazane HUVECs Human Umbilical Vein Endothelial Cells MACSAT Magnetically Actuated Cellular Strain Assessment Tool NDMR Non-Dimensional Migration Rate PBS Phosphate Buffered Saline PDMS Polydimethylsiloxane PU Polyurethane  xxii  Acknowledgements  I would like to thank my supervisor, Dr. Mu Chiao for giving me the opportunity to be part of the MEMS lab at UBC and to conduct my research under his supervision. I am grateful for his continuous support, patience and encouragement, without which completion of this dissertation would not have been possible. I also offer my enduring gratitude to Dr. James Lim who provided me with invaluable support and guidance with my project, as well as access to facilities without which many of the experimental work for this thesis would not have been possible. I owe particular thanks to Dr. Hongshen Ma, Dr. Karen Cheung, Dr. Helen Burt and Dr. Urs Hafeli for providing access to their labs and facilities, to Dr. York Hsiang for fruitful discussions on the research project and dissertation, and to Mr. John Jackson for providing training and help on various test equipment. I am grateful to my coworkers and lab-mates, Hadi, Nazly, Kevin, Payam, Aurora, Colin, Eric, Hongbin, Kaiwen, Jack, Eva, Daniel and Pascal, and to my dear friends, Ashkan, Sina, Danial, Masih, Ehsan and Ali for their friendship, support and encouragement throughout the past years. Most importantly, I would like to thank my dear family for their never ending support and encouragement, specially my parents whose sacrifice and unconditional love knows no bounds, and my siblings, Farnaz, Fardis and Farzin, whose companionship and moral support has been a constant source of hope and motivation. Finally I would like to acknowledge financial support from the NSERC Vanier Canada Graduate Scholarship, the Izaak Walton Killam Doctoral Fellowship and the UBC four-year Doctoral Fellowship. xxiii  Dedication     To My Dear Parents1  Chapter 1: Introduction  CHAPTER 1  Introduction   The ability to control the behavior and function of living cells has been the subject of much interest. Various cell behavior, such as cellular migration (how cells move), cellular proliferation (how cells divide and multiply) and cellular differentiation (how cells become specialized to perform a specific function) have a critical role in the proper function of living organisms and changes in the above behavior play an instrumental role in various pathologies. It has been found that cells respond to various types of external stimuli, i.e., stimuli in the form of chemical [1], electrical [2]-[4] or mechanical signals [5, 6]. It is therefore possible to use these external stimuli to trigger certain biological functions/responses in a cell and control cell behavior. Among the various forms of external stimuli, mechanical stimuli, in the form of mechanical stresses and strains, and their effect on cell behavior has been the subject of much study. It has been shown that controlled mechanical stresses and strains applied on cells from an underlying substrate [7, 8], as well as shear stresses applied on cells from a fluid flow elicit changes in the migratory and proliferatory behavior of cells [9, 10].  2  Conventionally, cell strain studies were done at the macro-scale. For example, groups of cells were cultured on a flexible membrane and the membrane was then stretched to apply a controlled strain on the adhered cells and study the cell response [7, 11]-[15]. More recently, with the advancement of available technologies and specifically the advancement of Micro-Electro-Mechanical Systems (MEMS), it has been possible to apply loads on individual cells at the microscale to study their response [16]-[20].  Various MEMS devices have been developed to date and used in the study of cell response to mechanical stimuli [17, 19]-[21]. These devices differ in terms of their functional density, easiness of implementation, actuator size, cell medium compatibility and the known side effects on cells. Among the various MEMS devices developed, micropillar arrays, especially those actuated magnetically [22]-[26], have specific properties that make them a uniquely suitable candidate for cell-strain applications. Micropillar arrays have a high functional density, are easy to implement, have great cell medium compatibility, do not cause any known adverse side effects on cells and can integrate individual actuators as small as a few microns. Magnetic micropillar actuators have recently been used in cell-strain studies, to study the response of cells to mechanical stimuli [22, 26]; however their effectiveness in controlling cellular behavior and their possible use for medical implant applications has yet to be investigated.  In this chapter, first, the concept of mechanotransduction and how cells sense external mechanical stimuli is briefly described. Then, brief descriptions of cell migration and cell proliferation and a review of conventional and MEMS devices used in cell mechanics studies are presented, focusing on currently available micropillar arrays, their properties and their shortcomings. The chapter is concluded by providing an overview of the thesis and describing 3  the objectives of the thesis with regards to assessing the functionality of micropillar arrays in controlling cell behavior.  1.1 Cellular Mechanotransduction Mechanotransduction is the term used to describe the ability of cells and living tissue to sense mechanical stimuli and respond by tissue remodeling [27]. This includes the sensation of the mechanical stimulus (such as mechanical loads, stresses or strains) by cells, translation of the mechanical stimulus into one or more biochemical signals and the sequence of biological responses that follow [27].  Figure 1.1 shows a schematic of a cell attached to the extracellular matrix through focal adhesion sites. The cell skeleton (cytoskeleton), which defines cell shape and locomotive abilities, is comprised of a number of mechanical elements, a network of actin filaments, microtubules and intermediate filaments [19]. The filaments of the cytoskeleton provide mechanical coupling among different regions of the cell and connect the cell membrane to its nucleus. Across the cell plasma membrane, receptors called integrins provide a physical connection between the cytoskeleton and the extracellular matrix at focal adhesion sites [19]. Due to the presence of such physical connections, mechanical loads applied to the extracellular matrix can be transferred to the cytoskeleton, which is a dynamic entity and remodels itself accordingly [10, 28]. Since many signaling molecules are associated with the cell cytoskeleton, mechanical stimulation of the cytoskeleton can be coupled with chemical responses [29, 30]. It has been shown that these simple changes in the shape of the cytoskeleton and the ensuing mechanochemical transduction can alter the proliferation, differentiation and migration behavior of cells [8, 30, 31].   Figure  1.1connectionmechanica1.2 CeCell miganother. multicellstructuregenerate are accom(cells thaextensioncontractimechanisforces is schemati: Schematic o between thl cues (mechall Migrationration, referMigration pular organiss and organforces in orplished thrt have a nu/protrusion on of the cms of movalmost alwc of the diffef a how a cell e extracellulanical loads, s and Prolifs to the prolays an imms. Cells ms or to accder to moveough differecleus) exhibof the leadytoskeletonement diffeays an essenrent phasesconnects to thr matrix (ECtresses and streration cesses invoportant roligrate for diomplish goa during mignt mechanisit the folloing front, , release ofr for varioutial step for in endothele extra-cellulM) and theains) betweenlved in the te in the prfferent reasols such as ration. Forcems based owing distincintegrin-med the cell rs cells, the achievingial cell migrar matrix. Fo cell body. T the ECM anranslation ooper develons, i.e., in owound hea generationn the type ot phases duiated attacear and rec generationmovement ation duringcal adhesionshis allows td the cell.   f cells frompment and rder to feedling [32, 33 and the resf cell. Manyring migrathment of thycling [32, and transfe[35, 36]. Fig angiogenes  provide a phhe transmissi one locatimaintenanc, to generate]. Cells neulting move eukaryoticion; sensinge leading  34]. Whiler of mechaure 1.2 shois. 4 ysical on of on to e of  new ed to ment  cells  and front,  the nical ws a  Figure  1.2go throughCell proldivision. replenishand prodeukaryotor mass, the cell karyokinforces pla: Schematic o a similar seqiferation refIn normal h the tissue. Cuces two cic cells is dra sequence nucleus), cyesis (divisioy an importf the cell miguence of phaers to an incealthy tissuell divisionells, knowniven by sizeof events is tokinesis (dn of the nuant role in cration phasesses during mirease in thee, cell prol is the proce as daugh-dependent initiated thaivision of tcleus). Mecell division for endotheligration   number of iferation anss where onter cells. Ctransitions [t results in he cell cytohanical forcand proliferal cells. Manycells due to d cell divise cell, knowell division37]-[39].  Amitosis (divplasm, orges, in the foation [40, 4 other adhercell growthion is restrin as the mo and prolis cells growision of the anelles andrm of pulli1].      ent eukaryoti followed bycted to cellsther cell, diferation in to a certainchromosom membrane)ng or contr5 c cells  cell  that vides most  size es in  and actile 6  1.3 Studying the Effects of Mechanical Stimuli on Cells The effect of mechanical stimuli (in the form of mechanical stresses and strains) on cells and the resulting changes in cell migration and cell proliferation has been a subject of special interest to the research community. Mechanical stresses and strains are readily seen in the body under physiological conditions and have been shown to be important regulators and modulators for various cellular functions. This mechanosensitive behavior of cells was first reported by Julius Wolff in the nineteenth century [42] when he observed that bones remodeled based on the stresses they experienced, a phenomenon which later became known as Wolff’s Law. It is now widely accepted that not only bones, but various tissues and most types of cells in the body are capable of sensing and responding to mechanical stimuli [27]. For example, the cyclic expansion and contraction of arteries attributed to blood pressure changes corresponding to the systolic and diastolic phases, leads to the exertion of a cyclic stretch on the endothelium and the underlying smooth muscle layer [43].This cyclic strain affects the migration of endothelial cells (ECs) and regulates angiogenic sprouting in human vascular cells [31]. Changes to the thickness of the arterial wall in response to changes in the circumferential stress have also been reported [44]. Mechanical forces also play an important role in cell division as well. It has been demonstrated that for cells growing on two dimensional substrates, the orientation of the cell division axis is guide by physical forces [45]. It has also been shown that the mechanical interactions between cells and the extracellular matrix are an important mechanism by which cell division in three dimensional environments is regulated [40]. 7  In order to study the effect of mechanical strains and stresses on cells in vitro, many different devices, and experimental setups have been developed. These devices can be divided into two distinct groups; (a) conventional devices (b) MEMS devices  1.3.1 Conventional Devices for Cell-Strain/Stress  Conventional devices for application of cell strain/stress can be divided into two groups; devices applying strains and stresses on cells from an underlying substrate to which the cells are adhered, and devices applying fluid shear stresses on cells attached to a substrate. The devices in the first group have two main components; a flexible membrane on which cells are cultured and to which cells adhere, and a means or method of force application to stretch the flexible membrane. The means of force application to the membrane can be pneumatic, i.e., by applying positive or negative pressures to stretch the membrane [13], or (electro)mechanical, i.e., using dynamic indenters [11, 12, 15] or moving clamps [7, 14] to stretch the membrane. In such devices, the boundary conditions of the membrane, the thickness and surface profile of the membrane, and/or the directions and locations of force application to the membrane can be designed in such a way as to achieve different strain profiles (uniaxial, equi-biaxial, etc…) on the culture surface of the membrane. Furthermore, parameters such as the magnitude of membrane deformation and strain magnitude as well as time duration of stretch can be controlled to obtain desired loading scenarios.  While conventional cell strain devices based on membrane stretching provide a useful means for the study of cellular response to substrate strains experiments, certain factors limit their availability to the general research community. Custom devices made by different research 8  groups are often composed of custom machined parts which are hard to replicate. Their operation requires access to computer programs and controllers which are not readily available. Furthermore, most such devices include parts such as clamping mechanisms that have to be reused. Since such parts are in direct contact with cell media or cell substrates during experiments, maintaining a sterile culture environment is a tedious task. Commercially available alternatives (such as those offered by Flexcell Int.) are expensive and cost in the thousands of dollars. It is therefore of much interest to design new membrane stretching devices that are easy to replicate and maintain. An ideal device for this purpose would be: a) made with off-the-shelf components and/or components readily available in most research labs b) operated with generic/multipurpose signal generators and devices available in most research labs c) able to easily maintain sterility of cell cultures required for long term experiments d) easily integratable with incubators and microscopes already available in cell culture labs The second group of conventional cell strain devices mostly consists of parallel plate flow chambers, variable height flow chambers and variable width flow chambers [46]. In these devices, cells are cultured on a surface of interest, and a fluid flow is applied on the surface of the cultured cells [9]. Consequently the effect of the shear stresses from fluid flow on the cells can be studied. Parameters such as fluid viscosity and fluid velocity can be controlled to obtain desired values of applied shear stress on the surface of the cells.  9  1.3.2 MEMS Devices for Cell-Strain/Stress  MEMS devices for application of mechanical loads on cells vary both in terms of the means of actuation and the number of cells that can be stimulated. Stimulation of single cells has been achieved using magnetic (electromagnetic), electrical, electrothermal or optical means. Stimulation of small groups has been achieved using electrical and electromagnetic actuators while stimulation of large groups of cells has been mostly performed using microfluidics and fluid flows. Electromagnetic stimulation of single cells is achieved by using magnetic tweezers [47, 48]. Magnetic tweezers operate by applying magnetic forces on a magnetic microparticle (usually a magnetic microsphere) that has been functionalized and adhered to specific cell receptors.  Forces applied to the microparticle are transferred to the cell through the receptor and allow for application of mechanical loads on the cells. Electrical stimulation of single cells or electro-deformation is achieved by applying a non-uniform electric field to a cell, causing the interfacial polarization of the cell membrane and creation of a dipole, resulting in the mechanical deformation of the cell due to the stresses created at the interface [17, 49]. Electrothermal actuators are based on the concept of thermal expansion due to electric currents heating up a structure. In these devices, usually one or more microsized beams of specific shape having direct or non-direct physical contact with a cell are heated up with a controlled electrical current, resulting in the controlled thermal expansion of the beam and application of mechanical loads on the cell [16, 17]. Optical MEMS actuators for application of mechanical loads on cells are mostly in the form of optical tweezers using the concept of laser optical trapping [17]-[19]. In these devices optical fibers guide divergent laser beams and trap individual cells. By changing the 10  parameters of the laser beam, the trapped cell can be elongated through the generation of a mechanical stretching force. Current MEMS devices for mechanical stimulation of small groups of cells are based upon arrays of microactuators that can be actuated by electrical or electromagnetic means or through the use of micro positioning stages [17, 19]. An electroactive polymer actuator for stretching of multiple cells was recently demonstrated [21]. The actuator consisted of a flexible Poly-Di-Methyl-Siloxane (PDMS) membrane attached to a rigid PDMS structure with an array of equally spaced channels containing compliant gold electrodes. Application of a voltage to the gold electrodes resulted in the stretching of the PDMS membrane which allowed for application of uniaxial strains of up to 20% to 128 individual cells. Soft polymer electromagnetic actuators have also been used to apply mechanical loading on groups of cells [22]-[26]. Such actuators typically consist of polymer micropillars containing magnetic material. Application of a magnetic field results in micropillar deformation and the application of mechanical loads to cells adhered to those substrates. Compared to membrane stretching techniques where loads are applied continuously along the entire cell, these micropillar structures apply loads/strains at discrete locations of the cells allowing for the study of phenomena such as the propagation of forces in the cell cytoskeleton. These structures will be studied in detail in section 1.4. MEMS devices for application of mechanical loads on large groups of cells have also been demonstrated. Simultaneous application of loads on large groups of cells has been achieved using fluid flows in parallel flow micro channels to apply hydrodynamic shear forces on cells [50, 51], by using large arrays of micro-wells with flexible membranes that can be stretched to apply loads on cells adhered to the membrane surfaces [20, 52, 53] or by application of magnetic 11  forces on cells dosed with magnetic nanoparticles [54]. Serial stimulation of large groups of cells has also been achieved using microfluidic channels [17, 55]. 1.3.3 Considerations for Development of MEMS Devices for Cell-Mechanics When developing MEMS technologies for the mechanical stimulation of cells, several considerations have to be taken into account. In this regards, it is favorable to use or develop technologies that have low economical cost and technological complexity, but are able to study large population of cells [17]. Furthermore, MEMS actuators which have fewer side effects on cells, a high functional density and high cell medium compatibility, and allow for greater miniaturization while being easy to implement, are a more favorable option [17]. Table 1.1 summarizes the characteristics, advantages and disadvantages of the different MEMS devices and actuation mechanisms used for cell mechanics studies based on the findings of Desmaele et al. [17].  MEMS Device Actuation Type Known Side Effects on CellsFunctional Density Actuator Size Cell Medium Compatibility Easiness of ImplementationMagnetic Fields  None High µm High Simple Electric Fields Potential High µm High Simple Piezo-electric  None High µm Medium Simple Electrothermal  Potential High µm Limited Complex Positioning Stage None Low mm High Very Simple Fluid Flow None High µm High Simple  Table 1.1 Advantages and disadvantages of the various types of MEMS devices and actuation mechanisms used for cell mechanics studies. Devices based on magnetic field actuation and devices using fluid flows have the most favorable s set of properties. 12  1.4 Micropillar Arrays for Cell-Mechanics Studies Among the various MEMS actuators discussed in section 1.3 micropillar arrays, especially those actuated through magnetic means, have a good combination of attributes that make them favorable for the application of mechanical stresses and strains on cells. Micropillar arrays have a medium to high functional density, are relatively easy to implement, have high cell medium compatibility, do not cause any known adverse side effects on cells and can integrate individual actuators as small as a few microns. It is for these reasons that in recent years, researchers have been increasingly more interested in using micropillar arrays for the study of cell mechanics [19, 56]-[63]. To date, micropillar devices have been used as a means to study cell-substrate and cell-cell interactions [56, 64], the loads generated by cells in their contractile phase [65], the traction loads applied by cells on their adherent substrates [62, 66], and the changes in cell migration and differentiation based on substrate stiffness [57, 67]. A majority of these devices are comprised of arrays of micrometer to nanometer sized pillars (or posts) made from various polymeric or composite materials [23, 68, 69]. These devices can be divided into two distinct categories based on their applications; passive devices and active devices. 1.4.1 Micropillar Arrays as Sensors Passive micropillar devices are used mainly as sensors, i.e., there are no actuation mechanisms on the devices [56, 57, 66, 70]. In such devices, cells are cultured on a bed of polymer micropillars and the forces they exert on their underlying substrates are studied through observation of micropillar deformation. As cells attach to the pillar surfaces and focal adhesions are created, contractions in the cells, either stimulated externally or generated internally by the cells, cause the pillars to bend, and this bending can be measured and used to calculate the 13  magnitude of the contractile forces generated in the cells. Passive micropillar devices have also been used to study cell response to variations in substrate stiffness [57, 67]. In these devices, arrays of micropillars fabricated with various diameters/heights make it possible to have substrates with identical material composition but with different substrate stiffnesses. When cells adhere and grow on the top of a bed of such micropillars, their substrate-stiffness-dependent behavior can be studied. Using such devices, it has been demonstrated that the migration of cells or the differentiation of primary cells (stem cells) is affected by substrate stiffness [67].  1.4.2 Micropillar Arrays as Actuators Active micropillar devices are those with pillars that can be actuated. Usually, such devices are used as actuators combined with sensors, i.e., selective groups of micropillars function as actuators while the rest function as sensors [22, 24, 26]. Actuation is commonly achieved through magnetic means, that is, magnetic particles are randomly distributed across some pillars, and those pillars will actuate in the presence of a magnetic field. These devices are used in a similar manner to passive micropillar devices, with the difference that stimulation of adherent cells is achieved by application of loads to the cells from active pillars on the device. When these pillars are actuated, they apply loads on the order of nanonewtons to specific regions of the cell, stimulating a response in the cell which causes the passive micropillars to bend, and as such the propagation of contractile forces along the cytoskeleton (i.e., the internal elements of the cell) can be studied. Active micropillar devices present great possibilities for the field of experimental cell-mechanics. While passive micropillar devices have been used in cell-mechanics studies for some 14  time, active micropillar devices, specifically those where actuation is achieved magnetically, have only recently been used. 1.4.3 Limitations of Current Magnetic Micropillar Actuators  The use of magnetic polymer micropillar arrays for cell-mechanics research is a new and growing field. New applications of these magnetic micropillar devices can be postulated, however, several factors have prevented their wide-scale adoption by the research community. These factors include complications with currently available fabrication techniques and the magnetic field requirements for pillar actuation. 1.4.3.1 Fabrication of Magnetic Polymer Micropillar Actuators In general, most polymer micropillar arrays, both passive and active devices, are fabricated using micro-molds and the replica molding technique. Micro-molds are first generated using standard microfabrication procedures based on the photolithography technique. Polymers are then cast on the micro-molds, allowed to cure, and de-molded to obtain the final structure. For active micropillar devices, an extra step is required before or in conjunction with the polymer casting step in order to introduce functional particles, i.e., particles with magnetic properties, into the structure of the micropillars. Previous fabrication methods for magnetic polymer micropillar structures rely on the solvent-casting technique for embedding of magnetic particles in the body of the micropillars. The solvent casting technique is based on the dispersion of magnetic particles in a fluidic medium [22, 24, 26]. Two different approaches to the solvent casting method are commonly used. In the first approach, magnetic particles are mixed and sonicated with a polymer base to obtain a 15  uniform ferrofluid [26]. In the case of highly viscous polymers, compatible solvents are used to dilute the polymer base. The ferrofluid is then mixed with a polymer hardener and excessive solvent is evaporated, allowing the polymer to be cast into a mold and cured to obtain the final device. Examples of such devices are magnetic polyacrylamide (PAM) pillars developed by le Digabel et al. [26]. It is important to emphasize that this approach is limited by agglomeration/ aggregation of magnetic particles, especially when using highly viscous polymers like PDMS, which prevents mold casting with very fine features [26]. Particle aggregation can be partially overcome through the use of magnetic particles with specialized surfactants in conjunction with strong solvents [71], however the use of hazardous solvents and the excessive sonication time required for obtaining a uniform ferrofluid does not make this approach an attractive option for many applications.  A second approach to the solvent-casting technique has been to disperse particles in low-viscosity solvents, guide them inside the fine features of molds using external magnetic fields, and then allow the solvents to evaporate, leaving behind the magnetic particles inside the fine features of the mold [22, 24]. This is followed by casting a host polymer on the mold to encompass the magnetic particles and provide the final micropillar structure. Sniadecki et al. were able to use this approach to fabricate PDMS micropillars with embedded cobalt nanowires [22, 24]. This second approach poses benefits and has fewer limitations compared to the first approach, however, it still requires the use of solvents and sonication to obtain a uniform ferrofluid and it is difficult to reproduce multiple micropillars of uniform properties on one device. Furthermore, magnetic particle placement is a purely random process, this affects design 16  of devices used for cell studies as there is no control over which pillars will act as actuators and which pillars will act as sensors. 1.4.3.2 Magnetic Field Requirements Many of the magnetic polymer micropillar structures developed to date require the use of strong magnetic fields with extremely high magnetic field gradients (on the order of 103 mT/mm) in order to actuate the magnetic pillars [23, 26]. These high magnetic field gradients are usually achieved by focusing the magnetic field using sharp magnetized tips. While these focused magnetic fields are able to provide the high magnetic field gradients required for magnetic pillar actuation, the high magnetic field gradients only exist within the immediate vicinity of the sharp magnetized tip, i.e., distances of a few tens of microns [23, 26]. This prevents the simultaneous and uniform actuation of multiple pillars covering a large surface area and/or truly remote actuation from a long distance away.  In order to overcome current limitations with available magnetic micropillar arrays, any new fabrication technique should be simple, replicable and integratable with viscous polymers such as PDMS without the need to use harsh solvents. Furthermore, it should allow for greater control over magnetic particle placement, and produce uniform and homogenous pillars. Finally, the magnetic pillars should be actuatable with easily producible magnetic fields of moderate magnitudes without the need for high magnetic field gradients. 1.4.4 Micropillar Actuators as a Tool to Govern Cellular Behavior in Vivo To date, available micropillar devices have mostly been used to study cellular behavior, such as the generation and propagation of forces during cell locomotion and migration, but not to 17  actively control cell behavior, i.e., controlling the migration and proliferation of cells. With advances in fabrication procedures that allow the fabrication of polymer micropillar chips containing tens of thousands of magnetic micropillars that can be remotely actuated in a controllable manner, new applications for these devices can be postulated. It might be possible for example to apply mechanical loads and forces on groups/ sheets of cells to actively control the migration and proliferation behavior of the cells, as previously demonstrated using conventional macro-sized devices based on membrane stretching techniques [7, 14, 72]. In contrast to membrane stretching techniques however, magnetic micropillar arrays that allow for remote actuation could possibly be integrated with implants for in-vivo medical applications.  One possible future application for magnetic micropillar arrays could be the prevention of cell-migration induced biofouling of medical implants. One example is the case of intra-ocular lenses (IOLs), where the migration of lens epithelial cells from surrounding tissue onto the posterior capsule causes biofouling and clouding of the lens, a condition referred to as posterior capsular opacification (PCO), resulting in the degradation of the implant and its eventual failure [73, 74].  Much effort has been put into finding ways to prevent the migration of lens epithelial cells between the posterior capsule and the IOL surface [75]-[78]. One possible solution might be to integrate magnetic micropillar arrays on the extremities of the IOL, and through micropillar actuation, direct the migration of lens epithelial cells away from the IOL and thereby prevent or reduce the occurrence of PCO. A similar integration of micropillar arrays with MEMS drug delivery devices currently under development can be postulated.  Another possible application for magnetic micropillar arrays could be the control of wound healing and scar formation. It is known that both cell proliferation and cell migration play an 18  important role in wound healing and tissue repair [79] and that rates of cell migration and proliferation directly affects scar formation [33, 80, 81]. It might thus be possible to use adhesive patches integrated with magnetic micropillar arrays that are actuated with a magnetic field, to apply controlled loads to the wound surface and control/change cell migration and proliferation patterns, with the aim of controlling the wound healing process and reducing scar formation.  Similarly, if magnetic micropillar arrays prove effective in promoting directed cell migration, materials incorporating these micropillar surfaces could be integrated with vascular stents to promote migration of endothelial cells from the proximal end of the stent towards the distal end, promoting the full recovery of an intact endothelium on the luminal surface of the stent. In order to assess the usability of magnetic micropillar arrays for future integration with medical implants, in vitro experiments that study the effectiveness of magnetic micropillar arrays at controlling cellular migration and proliferation need to be undertaken. To date, no such studies have been reported. 1.5 Thesis Objectives and Overview of Thesis The main objective of this thesis is to develop a robust and easily replicable technique for the fabrication of homogenous magnetic micropillar arrays which can be actuated using low to medium level magnetic fields, and to study the effect of the actuated magnetic micropillar arrays on cell migration behavior in-vitro to assess their usefulness in future medical implant applications.  The following criteria have to be met with regards to the fabricated micropillar arrays: 19  i) The fabrication technique needs to be easy, cost effective, replicable and scalable, and allow complete control over the placement and embedding of functional particles. ii) The micropillar arrays need to be homogenous in nature, with minimal variance in pillar properties. iii) The micropillar array should allow for the simultaneous actuation of pillars over a large region/surface area. iv) The pillars should be actuatable with low to medium level magnetic fields that are easy to produce as to allow remote external actuation across large distances required for medical implant applications. Once magnetic micropillar arrays with the above characteristics have been fabricated, the following questions need to be answered with regards to their effectiveness in controlling cellular migration through in vitro experiments with live cells; i) How does cyclic actuation of the magnetic micropillars affect cell migration? ii) Do changes in micropillar configuration (topographical patterns) or direction of actuation cause changes in performance with respect to controlling cell migration? This thesis is presented in five chapters. This chapter gave an introduction on the concept of mechanotransduction and how cells sense and respond to external mechanical stimuli. The role of mechanical forces in cellular migration and proliferation was established and a literature review on the currently available devices for application of stresses and strains on cells was presented. Furthermore, various conventional and MEMS devices for the application of strains on cells and their limitations were discussed, and suggestions for improvements were made. Finally, it was postulated that improved magnetic micropillar arrays that meet certain criteria 20  could have a possible use in future medical implant applications, and the need for in-vitro experiments in order to assess the effectiveness of magnetic micropillar arrays in controlling cell migration behavior was established. Chapter 2 presents a novel device developed for performing conventional substrate strain experiments on adherent cells. The Magnetically Actuated Cellular Strain Assessment Tool (MACSAT) is made with off-the-shelf components and is easily replicable, providing a cheap and reliable alternative for other researchers interested in studying the effect of substrate strains on cells. The MACSAT is used to perform substrate strain tests on Human Umbilical Vein Endothelial Cells, see their response to temporal and spatial variations in substrate strains and find the minimum strain values that elicit changes in the cells. The results obtained are later used in chapter 4 to assess the effectiveness of the micropillar arrays as cell strain devices. Chapter 3 presents a novel method for the fabrication of homogenous micropillar arrays containing functional material, i.e., particles with magnetic, electrical or fluorescent properties, in the structure of the pillars. Using this method, which we have named the Dry (nano)Particle Embedding Technique, homogenous arrays of magnetic micropillars that can be remotely actuated with low to medium level magnetic fields are fabricated. Experimental results for the characterization of the material composition and magnetic properties of the micropillars arrays and the generated actuation forces are presented and the favorable properties of the new devices compared to previous state-of-the-art are demonstrated. Chapter 4 presents cell culture studies performed on the magnetic micropillar arrays. Short-term and long-term migration assays are performed to look at how micropillar actuation effects the movement of cells and cell migration rates. Both endpoint experiments and experiments with live 21  imaging and tracking of cells are conducted. The migration patterns and the migration speeds of cells are then analyzed and the effectiveness of actuated magnetic micropillar arrays in controlling cell migration is demonstrated.  Chapter 5 summarizes the work presented and the conclusions reached in the thesis. Recommendations for future work are also presented. 22  Chapter 2: Magnetically Actuated Cellular Strain Assessment Tool: A Study on the Effects of Cyclic Substrate Strains on Human Endothelial Cells CHAPTER 2  Magnetically Actuated Cellular Strain Assessment Tool: A Study on the Effects of Cyclic Substrate Strains on Human Endothelial Cells   To simulate physiologically relevant strains for the purpose of experimental studies, various conventional devices and setups have been built to apply uniaxial and biaxial strains on flexible/elastic substrates on which cells are cultured [8, 11, 14, 82]. These devices employ pneumatic or electromechanical means to stretch flexible membranes on which cells have been cultured. When the elastic substrates are stretched, strain is partially transferred to cells adhered on the substrates, and as such, the effect of substrate induced strains on cells can be studied.  In-vitro experiments using these devices have shown that cyclic tensile strain in cell culture substrates has a significant effect on cellular orientation/alignment and cellular migration of different cell types, such as smooth muscle cells, endothelial cells and fibroblasts. When subject to substrate stretching, these cell types reorient transverse to the stretching direction and the actin filaments within those cells remodel into bundles of stress fibers aligned with the direction of 23  minimum deformation [72, 83, 84]. In one study, it was reported that endothelial cells subjected to three types of substrate strain, i.e., pure elongation, uniaxial stretching and biaxial stretching re-oriented towards a specific direction of minimal deformation for each type of load [7, 72]. In another study, the effect of two parameters, i.e., substrate strain amplitude and frequency of strain application, on the re-orientation rate of fibroblasts were examined [7]. Results showed that for strain amplitudes below 16% and for strain frequencies below 1 Hz, these two parameters are positively correlated with the re-orientation rate. Substrate induced cyclic tensile strain also caused directed migration of endothelial cells, with observed changes in migration rates of up to 15 fold [31].   While conventional cell strain devices based on membrane stretching provide a very convenient and useful means for the study of cellular response to substrate strains experiments, certain factors limit their availability to the general research community. Custom devices made by different research groups are often composed of custom machined parts which are hard to replicate. In addition, operation of some of the devices available requires access to controllers/computer programs which are not readily available. Furthermore, most of the available devices include parts such as clamping mechanisms that have to be reused. Since such parts are in direct contact with cell media or cell substrates during experiments, maintaining a sterile culture environment is a tedious task. Alternatively, commercial devices for cell-strain (such as the pneumatic devices offered by Flexcell Int.) are expensive and cost in the thousands of dollars. It is therefore of much interest to design new membrane stretching devices that are cheap, reliable and easy to replicate and maintain by others interested in performing cell strain experiments.  24  In this chapter a magnetically actuated cellular strain assessment tool (MACSAT), a new device for the application of cyclic stretch on membrane-like substrates for the purpose of cell strain studies, is demonstrated. The MACSAT is made with off-the-shelf components readily available in most research labs or easily obtainable from general hardware stores, it can be operated with generic/multipurpose signal generators available in most labs, it allows for easy maintenance of cell culture sterility using low-cost commercially available culture plates and it is small enough to integrate with standard incubators and microscopes already available in cell culture labs. The strains produced in the MACSAT membranes are characterized using finite element analysis and experimental measurements. Finally, the performance of the MACSAT is demonstrated by using it to apply strains on and study the re-orientation behavior of Human Umbilical Vein Endothelial Cells (HUVECs).  2.1 The Magnetically Actuated Cellular Strain Assessment Tool (MACSAT)  2.1.1 MACSAT Design The MACSAT consists of a commercial culture plate (Bioflex Plate, Flexcell Int. Co), an electromagnet, an aluminum base functioning as a heat sink and a supporting outer Plexiglas housing (Figure 2.1). The commercial culture plate has 6 circular wells, each having a flexible silicon elastomer membrane base as the cell culture substrate. Two centrally-placed, rare-earth magnets are added to the top and bottom surface of the substrate in each well. The magnets have their opposite poles facing each other and firmly clamp down on the center portion of the substrate. An electromagnet connected to a signal generator and a voltage amplifier is used to apply a controlled magnetic field at a controlled frequency on the permanent magnets.  25   Figure  2.1: Schematic design of MACSAT. As the permanent magnets are attracted by the electromagnet, vertical motion of the central part of the substrate is produced. This vertical motion induces a tensile strain in the radial direction and a compressive strain in the circumferential direction of the circular substrate. The amplitude and frequency of the magnetic field can be adjusted to obtain the desired magnitude and frequency of cyclic strain in the substrate.  26  2.1.2 Strain Distribution in MACSAT Membranes To determine the strain magnitude and strain distribution in the flexible substrates of the MACSAT, the deformation of the substrates was simulated using FEA in COMSOL software. In the FEA model, the permanent magnets had a diameter of DPM =6mm (equal to the diameter of the permanent magnets used in our cell strain experiments), and based on manufacturer specifications, values of t =0.5mm, DM=35mm and E=1 MPa were used for the thickness, diameter and Young's modulus of the flexible membrane, respectively. The outer edge of the membrane, which is clamped by the stationary walls of the culture well, was modeled as fixed boundary conditions and the membrane center, which is in contact with the permanent magnets, was modeled as clamped boundary conditions with motion only in the vertical direction.  As expected, it was found that the magnitude of strain developed in the MACSAT substrate is a function of the maximum vertical displacement V of the permanent magnets. FEA results obtained for a 3.5 mm vertical displacement of the central magnet are presented in Figure 2.2. 27   Figure  2.2:  FEM Simulation results and experimental measurement (n=3) of strain distribution in the silicone-elastomer membrane of the MACSAT for a vertical central displacement of V=3.5mm.  Figure 2.2 (a) shows the axisymmetric strain distribution in the membrane. We observe that two types of strains are developed on the top side of the membrane where the cells are to be cultured, 28  a tensile strain in the radial direction and a compressive strain in the circumferential direction. Moving away from the edges of the permanent magnets, the tensile radial strain developed in the substrate increases until it reaches a plateau. In the plateau region, which is several millimeters wide, the magnitude of radial strain has approximately a 10% variation. Also, in this region, the radial strain magnitude is one order of magnitude larger than the average compressive strain magnitude within the same region. For experiments where a relatively uniform strain needs to be applied on a group of cells, the plateau region can be used to culture cells.  We also performed experimental measurement of the strains developed in the substrate to verify the simulation results. We achieved this by laser-marking a rectangular grid on the substrate and optically measuring the deformation of the grid in radial and circumferential directions as the substrate was stretched. To begin, an Olympus SZ61 stereo microscope mounted on a YZ tilting platform, and a Basler A102 digital camera were used to take an image of the grid on the unstretched membrane. Permanent magnets were then clamped to the center of the membrane and the center was deflected by 3.5 mm. The microscope was then tilted to bring the grid in the focal plane of the lens, making sure that grid is parallel to the focal plane, and an image of the deformed grid was taken. The resulting images were then post-processed in ImageJ and the deformation of the grid was used to calculate the strain in the membrane. Results from experimental measurements of the strain in the plateau region are presented in Figure 2.2 (a). Three separate measurements were taken for each data point, the average was chosen as the value of strain and the standard deviation was used to construct error bars. Results obtained from experimental measurements showed strain distributions and strain magnitudes agreeing with those obtained from simulations. 29  Based on the results obtained for the strains in the MACSAT membranes, it is seen that the strains developed in the plateau regions of the flexible membranes of the MACSAT resemble that of uniaxial stretching. In uniaxial stretching a membrane is stretched in direction 1 and undergoes a contraction in direction 2 perpendicular to direction 1. As shown in the strain contour plots of Figure 2.2 (b), at any point in the plateau region, there is a net membrane stretch in the radial direction (direction 1) and a net contraction of the membrane in the circumferential direction (direction 2), perpendicular to the radial direction.  2.1.3 Directions of Minimum Deformation in MACSAT Membranes To determine the direction of minimum deformation on the MACSAT membranes, the theory of strain transformation can be used [85]. We define the direction of minimum deformation as the direction in which a one dimensional element (such as an actin filament) oriented in that direction experiences the lowest magnitude of axial deformation/axial strain.  Figure 2.3(a) shows a substrate (or a subsection of a substrate) undergoing uniaxial stretching. Any one dimensional element ‘A’ adhered to the substrate and oriented in the direction of stretching (direction 1) undergoes a tensile axial strain, while any one dimensional element ‘D’ oriented in the direction of contraction (direction 2) undergoes a compressive axial strain. Using the theory of strain transformation and with the help of visual aids such as Mohr’s circle for plane strain [85], we can find the directions of minimum deformation for the substrate. Figure 2.3(b) shows Mohr’s circle for the uniaxial stretching scenario depicted in Figure 2.3(a). 30   Figure  2.3:  Directions of minimum deformation in a membrane undergoing uniaxial stretching: (a) In uniaxial stretching a membrane is stretched in direction 1 and undergoes a contraction in direction 2 perpendicular to direction 1. Element ‘A’ is oriented in the direction of stretching and undergoes an axial tensile strain, while element D is oriented in the direction of contraction and undergoes a compressive axial strain. (b) Mohr’s circle for the plane strain scenario depicted in part (a). It is seen from Mohr’s circle that there are two unique angular orientations (‘B’ and ‘C’) in the plane of strain where the axial strain ε is zero. These directions of zero axial strain divide the plane of strain into multiple angular regions, with either compressive or tensile axial strain 31  It is seen from Mohr’s circle that there are two unique angular orientations (‘B’ and ‘C’) in the plane of strain where the axial strain (ε) is zero. Denoting the tensile strain in direction 1 with ε1 and the compressive strain in direction 2 with -ε2, the angle θ between the directions of zero axial strain and direction 2 can easily be determined from the following equation:  121212121111cos2/1cos2/1                                                             (1.1) As seen in Figures 2.3(a) and 2.3(b), at any point on a uniaxially stretched substrate, the directions of zero axial strain divide the plane of strain at that point into multiple angular regions, with either compressive or tensile axial strain. Therefore, a 1D element (such as an actin filament or a stress fiber) not oriented in the directions of zero axial strain (directions ‘B’ and ‘C’), will either experience a compressive axial strain, or a tensile axial strain, depending on its angle of orientation. Since at any point in the plateau region of the MACSAT membrane, there is a tensile strain in the radial direction and a compressive strain in the circumferential direction, equation 1.1 can be used to determine the direction of zero axial strain with respect to the circumferential direction. For a 3.5mm vertical displacement of the membrane center, corresponding to an average tensile strain of 6% in direction 1 (radial direction) and an average compressive strain of -1% in direction 2 (circumferential direction), θ is approximately ± 22 degrees. That is, for any point in the plateau region of the membrane, there are two directions of zero axial strain at that point which are located symmetrically about the circumferential direction at that point. If θ=0º is chosen to denote the circumferential direction at a specific point, the two directions of zero axial strain for that point are located at θ= ±22º. 32  2.2 Cell Strain Experiments with MACSAT 2.2.1 HUVEC Culture Human Umbilical Vein Endothelial Cells (HUVECs) were chosen as the cell line for our cell strain experiments.  Endothelial cells have been shown to be very responsive to substrate strains and previously published research reporting their reorientation and migration response in the presence of substrate strains is available. They are therefore a good candidate for cell strain experiments assessing the functionality and usefulness of the MACSAT.  For our cell-strain experiments, HUVECs (EGM-2, cryo amp, code cc-2517A) and culture media (EGM-2 BulletKit, code cc-3162) were obtained from Lonza. The cells were expanded and passaged in culture flasks in the EGM-2 culture media. Cells from passages 3 to 7 were used for conducting the cell strain experiments in the MACSAT. 2.2.2 Preparation of Culture Dishes Six-well culture plates (BioFlex Plate, Flexcell int.) were obtained and used for cell strain experiments. Each well in the culture plate had a thin elastomeric membrane base with a Fibronectin coating. To prepare the flexible membranes for cell strain analysis, we used a laser micromachining center (New Wave Research, QuikLaze 50 STII) to mark a circular grid on the reverse side of the substrates (the side not in contact with the cells). The grid was subsequently used as a guide to visualize and correctly determine the radial and circumferential directions at any point on the substrate once viewed under the microscope. It also served the purpose of making sure we were looking at the same region of the substrate when gathering data from the time-lapse images obtained during the experiments.   33  2.2.3 Cell Re-orientation Experiments To determine the effect of substrate strain on the reorientation of cells, HUVECs were transferred from culture flasks, seeded on the plateau region of flexible substrates of a 6 well Bioflex culture plate (see Appendix A) and incubated in EGM-2 under 5% CO2 until the substrate surfaces reached an approximate confluency of 50% (Figure 2.4).   Figure  2.4:  Operating steps for cell culture and application of tensile strains (in the radial direction) on cells growing on flexible substrates using the MACSAT. First, cells are cultured in the plateau region of the flexible membrane. Then permanent magnets are then added to the top and bottom of the membrane and pulled by a controlled cyclic force from an electromagnet. This results in the development of a periodic cyclic stretch in the membrane and the application of cyclic strain on the cells adhered on the membrane. 34  One of the wells was then chosen for the strain test and the rest of the wells were used as controls. At this point permanent magnets were added to the substrate of the well chosen for the strain test and the substrate was cyclically stretched for 10 hours at a frequency of 1Hz and a central displacement of 3.5mm corresponding to an average radial tensile strain of 6% in the plateau region. At different time intervals during the experiment cells at multiple locations on the substrates were imaged using a Nikon Eclipse TS 100 inverted microscope equipped with a Nikon plan 10x/0.25 Ph1 objective and a Lumenera infinity2 ccd-camera. Post-acquisition image analysis was performed in ImageJ using the OrientationJ plugin. Details about the parameters used in OrientationJ and the mathematical derivation of direction vectors are presented in Appendix B. It is important to note here that the actuation frequency of 1Hz and the strain magnitude of 6% used in our experiments were chosen because of their physiological relevance. In the human body, a resting heartbeat of 60 bpm imposes a cyclic force on vascular cells at a frequency of 1 Hz, and cells in arteries undergo tensile strains between 2-18% due to blood pressure changes [27]. Figure 2.5 shows micrographs of the same group of cells at the start of strain application and after 10 hours of strain application. It is observed that initially, before application of the cyclic strain, many cells did not have a well-defined longitudinal axis. Furthermore, the cells that had a well-defined longitudinal axis were oriented in random directions. In comparison, after 10 hours of cyclic strain application, most of the cells became elongated and had well-defined longitudinal axes. Furthermore, the longitudinal axes of the cells were oriented transverse to the stretching (tensile strain) direction.  This result is consistent with previous reports on the reorientation of 35  endothelial cells away from the stretching direction in response to cyclic uniaxial stretching (11, 12, 20).  Figure  2.5: Raw and color coded images showing reorientation of HUVECs due to cyclic strain. The cyclic tensile strain (6% at 1Hz) is applied in direction of arrow; images show the same group of cells before and after strain application. Top: before application of strain cells have random orientations. Bottom: after 600 minutes of cyclic strain application cells re-align transverse to direction of applied cyclic tensile strain. We analyzed the effect of the existence of two directions of zero axial strain on the reorientation behavior of the cells. It was expected that both these directions of minimal deformation would be preferred directions of reorientation for the cells. Figure 2.6 shows the time evolution of the distribution of cell angles for a group of cells undergoing 6% cyclic strain. The two directions of  zero axiastrain (-9Figure  2.6direction, direction (± 22º.Withcompressil strain, loca0º <θ<-22º a : Time evoluwhich is the dmax. comprein 10 hours ove axial strainFraction of Cells ted at θ= ±2nd +22º <θtion of the direction of stssive axial strf cyclic stretc. 2º, divide th<+90º) and cistribution ofretching (maxain) is at 0º, ahing, about 8e plane of sompressive cell angles fimum tensilend the minim6% of cells hatrain into an axial strainor cells unde axial strain)um deformatve orientatiogular region (-22º <θ<22rgoing cyclic, is at ± 90º. Tion angle (zern angles in ths of tensile º).     strain. The he circumfero axial straine angular reg36 axial radial ential ) is at ion of 37  While at the start of the experiment there was a random distribution of cell orientation angles, after two hours of strain application, many cells had started to reorient away from the angular region of tensile strain and towards the directions of minimum deformation located at θ= ±22º. After two, three, six and ten hours of strain application, more than 53%, 58%, 61% and 86% of the cells had orientation in the region of axial compressive strain (-22º <θ<22º), respectively, whereas at the start of the experiment only 20% of cells had orientation angles within this range. 2.2.4 Actin Re-orientation and Alignment Experiments To determine the effect of substrate strain on the reorientation and alignment of actin filaments inside the cell, HUVECs were strained using the MACSAT. After 6 hours of cyclic strain application the cells were rinsed once in phosphate buffered saline (PBS) and fixed in 3.7% formaldehyde/ PBS (Sigma Aldrich, St. Louis, MO) for 15 minutes. Following three rinses in PBS to wash off excess formaldehyde, cellular F-actin was stained by incubation with 0.2 μg/mL of Rhodamine conjugated phalloidin (R-phalloidin, Sigma Aldrich) in 0.05% TX100/1X PBS for 30 minutes at room temperature. The substrate was rinsed a further 3 times in PBS and finally in 10 mM Tris-Cl pH 7.4 before mounting onto a coverslip with Prolong Gold with DAPI (Invitrogen). The mounted substrate was left to set overnight before imaging. The fixed cells were imaged on an Olympus IX81 microscope equipped with a 40x 0.75 NA UPlanFLN air objective, a CoolSnap HQ2 camera (Photometrics), X-Cite Exacte (Lumen Dynamics) light source and Semrock Quadband Sedat filter set. Post-acquisition image analysis was done on ImageJ using OrientationJ plugin. Strained cells showed significant actin reorientation and alignment compared to unstrained control cells. Results from immunofluorescence staining of actin filaments are presented in Figure 2.7. 38   Figure  2.7: Comparison of actin alignment in a control cell compared to a strained cell. (a) Control cell with no cyclic strain (b) Strained cell after 6 hours of 6% cyclic strain (c & d) Color coded images showing the direction of actin elements of the control cell and strained cell (e & f) Reconstructed images of the cells based on the computed directionality vectors obtained from ImageJ and OrientationJ (g & h) Magnified inset regions of reconstructed images of the control and the strained cell showing direction of actin alignment.  39  Figure 2.7 (a & b) shows a side by side comparison of actin alignment in a control cell compared to a strained cell. Figure 2.7 (c & d) show the same cells where actin filaments have been color coded based on their orientation angles (using ImageJ and OrientationJ). The existence of multiple colors in the color coded image of the control cell suggests that actin elements are oriented in many directions. In comparison, the existence of a dominant color in the color coded image of the strained cell demonstrates that there is a dominant/favorable orientation angle for the actin elements in the strained cell. Figure 2.7 (e, f, g & h) show the same cells where the images were reconstructed based on the OrientationJ computed directionality vectors of the actin elements. It is observed that the actin elements in the control cell appear in small local bundles with random directions. In contrast, the actin elements of the strained cells are aligned in one dominant direction.  Quantitative analysis of actin alignment angles shows reorientation of actin bundles of strained cells away from the angular region of tensile axial strain and towards the angular region of compressive axial strain. Figure 2.8 shows the distribution chart of actin orientation angles for the sample cells of Figure 2.7. It is observed that the control cell has a random distribution of actin orientation angles (Figure 2.8(a)). In comparison, for the strained cell, more than 91% of the actin elements have orientation angles corresponding to the region of compressive axial strain in between the two directions of zero axial strain (Figure 2.8(b)).   40   Figure  2.8: Distribution of actin orientation angles for (a) the control cell and (b) the strained cell of Figure 2.7. While the actin bundles in the control cell have random orientations, more than 91% of actin elements in the strained cell have aligned in between the two directions of zero axial strain, in the angular region of compressive axial strain. Figure 2.9 shows the data obtained from comparing the actin orientation angles of 28 control samples and 28 strained samples. It presents the angular distance of the mean and peak actin orientation angles from the circumferential direction.  41   Figure  2.9: Distribution of actin orientation angles for strained cells vs. control cells. Maximum tensile axial strain was applied in the radial direction. Data was obtained from 28 control samples and 28 strain samples. The 5 vertical lines in each boxplot (from left to right) represent the minimum value, first quartile, median value, third quartile and maximum value, respectively. While the actin bundles in the control cell have random orientation angles, actin elements in the strained cell have aligned in the region of compressive axial strain.  In each sample, the mean and peak orientation angle of the actin elements was calculated separately. The resulting data points representing the mean actin orientation angles of the samples were then used to obtain the distribution presented in the boxplots of Figure 2.9. It was observed that for the control cells, the mean value of the actin orientation angles did not favor a specific direction. In contrast to the control samples, more than 75% of the strained samples had actin orientation angles (both mean and peak) that were within the angular region of compressive axial strain.   We obsestrained. samples aFigure  2.1control anvalue, firsdeviation strained cstrained cThe meacontrol c4 fold inindependstrained actin elemrved that thFigure 2.10nd strained0:  Coherencd strained sat quartile, meof actin alignells, respectivells comparedn and standells, and 0.4crease in thent-samplesand control ents betwee coherenc shows the d samples.  y of actin alimples. The 5 dian value, thment coherenely, pointing t to control celard deviatio81 and 0.09e extent of  t-test was cells. Thereen the two gy of actin aata obtainedgnment for svertical lines ird quartile cy are 0.124 o an approxils.  n of actin 1 for the stractin alignmconducted  was a signroups of cellignment in for the cohtrained cells in each boxpland maximumand 0.069 formate 4 fold inalignment cained cells, ent in straito compareificant diffels; t(30)=12creased sigerency of avs. control ceot (from left  value, respe the control ccrease in theoherency wrespectivelyned cells co the coherrence in th.19, p<0.000nificantly actin alignmells. Data wasto right) reprctively. The mells, and 0.48 coherency ofere 0.124 a, pointing tompared to ency of ace coherency1.  s the cells nt of the co  obtained froesent the minean and stan1 and 0.091 f actin alignmnd 0.069 fo an approxicontrol cellstin alignme of alignme42 were ntrol m 28 imum dard or the ent in r the mate . An nt in nt of 43  We also observed that cell-cell contact did not inhibit cellular reorientation or actin alignment in strain experiments. Figure 2.11shows a group of closely spaced cells with cell-cell contacts. Similar to isolated cells, after 6 hours of cyclic strain application all the cells in the group and the internal actin filaments within the cells had reoriented away from the stretching direction towards the directions of zero axial strain.   Figure  2.11:  A closely spaced group of HUVECs having cell-cell contact was subjected to stretching in the direction of the arrow. All the cells in the group reoriented away from the stretching direction and actin filaments within the cells formed stress fibers in the direction of minimum axial strain.   2.2.5 Discussion of Results  The existence of two directions of zero axial strain in the case of uniaxial stretching results in the separation of the plane of strain into multiple angular regions with tensile and compressive axial strains (Figure 2.3). That is, based on the orientation angle of an element in the plane of strain, it 44  can be undergoing tensile or compressive axial strain, or zero axial strain if it is exactly aligned with one of the directions of minimum deformation. Due to the symmetry of the different regions of strain, two elements that are not oriented in the same direction, and thus are not aligned, could still experience similar strain conditions. The distribution of cell orientation angles among these different regions of strain should be taken into account when studying the reorientation and alignment behavior of cells subject to substrate strains. Time-dependent distribution graphs of cell reorientation angles for a large number of cells undergoing cyclic stretching show that from the early stages of strain application (first three hours), cells start to reorient towards the directions of zero axial strain (Figure 2.6). At the end of the experiments however, we observe that not all the cells are oriented exactly in one of the directions of zero axial strain. Many cells have re-oriented in between these two directions and are undergoing compressive axial strain along their long axis, while some cells remain in the angular region of tensile axial strain close to but not fully aligned with the directions of minimum deformation. This distribution of cell orientation angles after cyclic stretching can be explained by considering that although cyclic elongation and cyclic compression are unfavorable conditions for cell alignment [72, 86], a certain threshold of strain, i.e., a minimum of 1% to 3.5% strain, has to be applied to the cells in order to initiate cell re-orientation [72, 86]. In our experiments, the maximum strain developed in the angular region of compressive axial strain is only 1%, smaller than the reported threshold values of strain required for re-orientation, while the maximum strain developed in the angular region of tensile axial strain is 6%,  larger than the threshold values [72]. As cells rotate away from the directions of maximum tensile strain (θ=±90º) towards the directions of minimum deformation (θ=±22º), the magnitude of axial strain 45  they experience decreases, reaching the threshold value of 1% at θ=±32º (see Appendix C). Therefore, within the angular regions of -90º <θ<-32º and +32º <θ<+90º cells experience axial strain magnitudes that are larger than the threshold strain and cellular reorientation away from the strain direction is triggered. Once cells have reoriented to the angular region of -32º <θ<+32º which includes the angular region of compressive axial strain (-22º <θ<+22º), the maximum applied axial strain on cells is below the 1% threshold and further reorientation is not triggered, thus the angular orientation of the cells remains within this region. Consequently, the whole angular region of -32º <θ<+32º presents a favorable direction for final orientation of the cells.   Looking at the orientation angles of actin filaments within strained cells we have observed that the actin filaments form bundles of stress fibers oriented in the angular region of compressive axial strain or aligned with a direction of minimum deformation (Figures 2.8 & 2.9). Our observation that that many actin elements have final angular orientations spread in the region of compressive axial strain, matches predictions from mathematical models describing actin reorientation due to uniaxial stretching. One such model predicts that for a varied distribution of pre-tension magnitudes in the actin filaments prior to stretching, two outcomes are possible for their final reorientation directions based on the magnitude of applied uniaxial strain [87]. For high uniaxial strains (e.g. 12%) the model predicts that actin filaments will have final orientation angles within a small spread (±5 degrees) of the direction of minimum deformation. In contrast, for low uniaxial strains (e.g. 4%) the model predicts that actin filaments will have a much wider angular spread, covering most of the compressive axial strain region. This spreading of actin filaments in the angular region of compressive axial strain for low uniaxial strain magnitudes is attributed to the fact that for actin filaments with larger pretention, the compressive strains in the 46  said angular region are smaller than the threshold strains required to overcome actin pre-tention and cause actin disassembly, making this angular region of compressive axial strain suitable for orientation of such elements [87]. The actin reorientation behavior predicted by the mathematical model is thus consistent with the spread of actin orientation angles over the region of compressive axial strain observed in our experiments for 6% uniaxial strain (Fig. 2.8).  Building on previous literature investigating the formation and alignment of actin bundles and stress fibers, we have used the coherency parameter to quantify a threshold for the occurrence of actin alignment in strained cells. By observing the average coherency values for images obtained from control and strained samples we find an average 4-fold increase in the degree of alignment of actin filaments within strained cells compared to control cells. Furthermore, the minimum coherency of alignment for any one of the strained cells in our experiment (0.36) is more than the maximum coherency of alignment in any of the control cells (0.26). Thus a coherency value such as 0.3 can be chosen as a threshold value above which the actin filaments of a cell are considered aligned. While these coherency values might differ based on the resolution of images obtained and the parameters defined for the mathematical analysis of orientation, once such parameters have been chosen for a set of experiments, threshold values can be determined and used with consistent results. We therefore propose that coherency of actin alignment can be used as a new readout to determine the occurrence of alignment in cells. We propose conducting further experiments to determine whether the extent of actin alignment in a strained cell, and thus the magnitude of coherency parameter, is proportional to the magnitude of applied cyclic strain. It has been questioned whether the confluency of cell cultures in cell strain experiments affects their reorientation response. The reason for this is twofold; first, different cell densities might be 47  the result of cells being in different growth cycles which might affect their strain response [72], and second, the extracellular environment has been shown to affect the cell reorientation response to mechanical stretching [88]. For instance, confluent and subconfluent endothelial cell cultures showed a different alignment/ reorientation response when subject to shear stress from fluid flows [89, 90]. While our experiments were conducted at subconfluent cell levels, we were able to observe regions of the cell culture with isolated cells and compare them to regions of the cell culture with groups of closely spaced cells having cell-cell contact (Figure 2.11). We have not observed any difference in the reorientation or alignment response between the two. As is the case for isolated cells, groups of closely spaced contacting cells and their internal actin elements reoriented away from the stretching direction and towards the directions of zero axial strain. We therefore deduce that cell-cell contact and confluency of the cell culture does not significantly inhibit actin reorientation in response to substrate strain. 2.3 Conclusion The MACSAT is a reliable, low cost, versatile, easy-to-replicate and easy-to-maintain device for cell strain studies. Using a magnetic means to accomplish membranes stretching, the MACSAT is able to simulate physiological substrate strains with a variety of strain magnitudes and a wide range of cyclic loading rates, making it an accessible alternative to currently available custom made or commercially available devices.  Using the MACSAT, application of uniaxial substrate stretch (also known as pure elongation) on Human Umbilical Vein Endothelial Cells was demonstrated through a set of experiments. A systematic study of the results obtained from the MACSAT experiments confirms findings from previous literature with regards to cell and actin reorientation response of human endothelial cells. 48  Chapter 3: Dry Nanoparticle Embedding Technique for Fabrication and Patterning of Magnetic Micropillar Arrays CHAPTER 3  Dry Nanoparticle Embedding Technique for Fabrication and Patterning of Magnetic Polymer Micropillar Arrays   Over the past decade, micro-structured/micro-patterned surfaces have found a niche use in cell-mechanics studies [19, 20, 23, 56]-[58, 68]. A majority of the devices developed are comprised of arrays of micrometer to nanometer sized pillars made from various polymer or composite materials [23, 58, 68, 69]. These devices can be divided into two distinct groups based on their applications; passive devices and active devices. Passive devices are used mainly as sensors, i.e., there are no actuation mechanisms on the devices [56, 57, 66]. Active devices are used as actuators combined with sensors [22, 24, 26].  Active micropillar devices present great possibilities for the field of experimental cell-mechanics, but unlike passive micropillar devices which have been used in cell-mechanics studies for a more than a decade, active micropillar devices, specifically those where actuation is achieved magnetically, have only recently been used [22, 24, 26]. Several factors have prevented 49  the wide-scale adoption of magnetic micropillar actuators for cell-mechanics applications by the research community. These factors include complications with currently available fabrication techniques and the magnetic field requirements for pillar actuation. Previous fabrication methods for magnetic polymer structures rely on the solvent-casting technique. This technique is based on the dispersion of magnetic particles in a fluidic medium [22, 24, 26, 71]. Two different versions of the solvent-casting technique are commonly used. In one approach, magnetic particles are mixed and sonicated with a polymer base to obtain a uniform ferrofluid [26, 71]. In the case of highly viscous polymers, compatible solvents are used to dilute the polymer base. The ferrofluid is then mixed with a polymer hardener and excessive solvent is evaporated, allowing the polymer to be cast into a mold and cured to obtain the final device. Examples of such devices are magnetic PDMS membranes previously developed by our group [71] and magnetic polyacrylamide (PAM) pillars developed by le Digabel et al. [26]. It is important to emphasize that this approach is limited by agglomeration/ aggregation of magnetic particles, especially when using highly viscous polymers like PDMS, which prevents mold casting with very fine features [26, 71]. Particle aggregation can be partially overcome through the use of magnetic particles with specialized surfactants in conjunction with strong solvents [71], however the use of hazardous solvents and the excessive sonication time required obtaining a uniform ferrofluid does not make this approach an attractive option for many applications.  A second approach to the solvent-casting technique has been to disperse particles in low-viscosity solvents and guide them inside the fine features of molds using external magnetic fields. The solvents are then allowed to evaporate leaving behind the magnetic particles inside the finer features of the mold [22, 24]. This is followed by casting a host polymer on the mold to 50  encompass the magnetic particles and provide the final micropillar structure. Sniadecki et al. were able to use this approach to fabricate PDMS micropillars with embedded cobalt nanowires [22, 24]. This second approach poses benefits and has fewer limitations compared to the first approach, however, it still requires the use of solvents and sonication to obtain a uniform ferrofluid and it is difficult to reproduce multiple micropillars of uniform properties on one device. Also, magnetic particle placement is a purely random process, this affects design of devices used for cell-mechanics studies as there is no control over which pillars will act as actuators and which pillars will act as sensors. In this chapter, we present the dry particle embedding technique, a new method for fabrication of active magnetic micropillar structures aimed at experimental cell-mechanics studies. In the dry nanoparticle embedding technique, simple modifications to conventional fabrication processes omit problems arising from particle agglomeration, and allow us to achieve great uniformity of material properties across the whole micropillar array. We first present our new fabrication process, its attributes and capabilities. We then demonstrate a variety of micropillar devices containing different functional particles in the structure of the pillars, built using the processes described. Finally we present a characterization of magnetic and mechanical properties of our fabricated devices and demonstrate their advantages compared to previous state-of-the-art alternatives.  51  3.1  Fabrication  3.1.1 The Dry Particle Embedding Technique To overcome the limitations of the solvent-casting technique, we propose a dry nanoparticle embedding technique for magnetic polymer micropillar fabrication. The fabrication steps are presented in Figure 3.1.  First, photolithography is used to make an SU-8 master of the micropillar structure (step 1). PDMS polymer base (Sylgard 184 Silicone Elastomer, Dow Corning Corporation) is mixed with the cross-linker based on the manufacturer suggested ratio of 10:1 polymer base to cross-linker, poured over the SU-8 master, allowed to cure and peeled off to obtain a PDMS negative of the master (step 2). A Parylene coater (PDS 2010 Labcoter 2, Specialty Coating Systems) is then used to coat the PDMS negative with a 0.5 to 1 μm thick layer of Parylene-C (step 3). In the next step, magnetic nanoparticles are poured over the PDMS negative (step 4). We used magnetic nanoparticles with diameters in the range of 0.1 to 1 μm provided to us by UBC Pharmaceutical Sciences. The particles consist of elemental iron cores with an outer carbon coating. The total iron content of the particles is 95-98% by weight. It has previously been shown that iron/iron-oxide and iron/carbon core/shell nanoparticles are biocompatible, possess excellent magnetization properties and are stable against oxidation, making them a suitable candidate for use in biomedical applications [91, 92]. An external magnetic field from a 0.25 inch diameter, 0.25 inch thick, axially magnetized, cylindrical Neodymium (NdFeB) permanent magnet (with surface field of 6600 Gauss) is then used to guide the magnetic particles into the patterns of the Parylene-C-coated PDMS negative (step 5).  52       Figure  3.1: Magnetic micropillar fabrication steps using the dry nanoparticle embedding technique, final device has a transparent, non-magnetic PDMS base with magnetic PDMS Pillars. Step 1:  Standard photolithography to make an SU-8 master of final micropillar device     Step 2:  Pour PDMS over SU-8 master and cure to obtain a PDMS negative of master       Step 3: Parylene-C coating of PDMS negative    Step 4: Apply dry magnetic micro/nano-particles on the PDMS negative   Step 5: Induce movement of magnetic particles into features using magnetic field and wipe off excess particles     Step 6: Pour PDMS polymer on PDMS negative and magnetic particles    Step 7: Peel off final device once PDMS cures Magnetic PDMS Pillars UV light SU-8 PDMS PDMS Negative Parylene-C Coating PDMS Negative Magnetic Nanoparticles Magnetic Field PDMS Polymer Transparent, Pure PDMS Base 53  To achieve uniform and adequate filling of the holes, the permanent magnet is pressed against the underside of the mold and moved across the region of interest several times, effectively pulling the magnetic nanoparticles resting on the top surface of the mold into the holes. The mold is then checked under an optical microscope to verify that all holes have been adequately filled with the magnetic powder. Excess particles are then removed from the surface of the PDMS negative by wiping off with a cotton applicator. Next, a second PDMS solution (10:1 mixing ratio) is prepared and poured over the PDMS negative, the PDMS seeps into the features of the mold and encompasses the magnetic particles (step 6). Curing is performed at room temperature to increase curing time (24 to 48 hours). The low curing temperature reduces PDMS shrinkage [93, 94] and prevents any trapped bubbles in the PDMS from expanding, inhibiting the formation of unwanted cavities in the pillar structure [95]. The increased curing duration allows more time for the diffusion of the magnetic nanoparticles in the PDMS medium. This allows the PDMS to fully encompass the magnetic particles resulting in a uniform and homogenous micropillar structure. Once cured, the top PDMS layer with magnetic particles embedded in the micropillars is peeled off and the final device is obtained (step 7). The final device consists of a transparent, nonmagnetic PDMS base and magnetic PDMS pillars. The application of Parylene-C coating to the PDMS negative in step 3 serves two distinct purposes. First, it provides an intermediate layer between the PDMS negative and the top PDMS layer so that no PDMS-PDMS bonding occurs during the second PDMS casting in step 6. Previously, (tridecafluoro- 1,1,2,2,- tetra- hydrocytl)-1- trichlorosilane has been used for the silanization of PDMS surfaces to allow casting of PDMS on PDMS molds [68]. However, the aforementioned chemical is toxic and the silanization procedure takes a considerable amount of 54  time. The use of Parylene-C as a coating serves the same purpose, is safe and requires less time to complete. Similar to silanization, the Parylene coated PDMS negatives can be re-used for the fabrication of several devices from the same mold allowing us to study the consistency of the fabrication technique. Second, the Parylene-C layer provides a smooth slippery surface from which excess magnetic particles can easily be removed (wiped off) in step 5. An alternative fabrication technique based on PU molds that does not require the Parylene-C coating step is presented in chapter 4.  Figure 3.2 presents Scanning Electron Microscope (SEM) images of micropillar devices we have fabricated using the dry particle embedding technique. Two sets of devices having diameters of 40 and 8 μm and heights of 154 and 18 μm, respectively, are shown.   Figure  3.2: SEM images of magnetic PDMS pillars (a) & (b) 40 μm in diameter, 154 μm in height, (c) & (d) 8 μm in diameter, 18 μm in height. An example of a defect in the form of a missing or torn-off pillar during release of final device from mold is shown in (d). (a) (b) (c) (d) 55  The 8-μm diameter pillars have a center-to-center spacing of 16 μm and the 40 μm diameter pillars have a center-to-center spacing of 120, 160 or 200 μm. These dimensions were chosen to accommodate studies where cells will be grown on top of the 8 μm diameter pillars or in between the 40 μm diameter pillars. 3.1.2 Physical Embedding of Nonmagnetic Particles One of the great advantages of the dry nanoparticle embedding technique compared to the solvent casting method is the ability to integrate the technique with non-magnetic particles through a simple modification to the fabrication procedure. In step 5 of the fabrication procedure shown in Figure. 3.1, it is possible to induce movement of the particles into the features of the mold using physical pressure instead of external magnetic field, therefore the magnetic particles can be replaced with any other particle of interest, e.g., electrically conductive particles, fluorescent particles, etc. The only requirement is that the particles be small enough to be able to move into the features of the mold as the external pressure is applied. In contrast to the simplicity of the dry nanoparticle embedding technique, physical embedding of nonmagnetic particles is difficult to achieve with solvent-casting techniques. This is due to the fact that the particles are inside a solution; therefore a sedimentation process has to be used. As an example, centrifugation can be used to enhance sedimentation of the particles in the features of the mold [26], but it is an inefficient technique as it results in higher agglomeration rates of the particles in the solution. We have used nonmagnetic materials such as fluorescent or electrically conductive particles to make micropillar devices using the physical embedding technique. These devices are presented in Figure 3.3. Figure 3.3(a) shows a device consisting of a pure PDMS base with fluorescent pillars achieved through embedding of Fluorescein Sodium Salt (C20H10Na2O5, Fisher 56  Scientific) in the PDMS structure. Figure 3.3(b) shows a device consisting of a pure PDMS base with electrically conductive pillars achieved through embedding of carbon black particles (Vulcan XCV72R, CABOT) in the PDMS structure.  Figure  3.3:  Application of dry nanoparticle embedding technique to fabricate (a) Fluorescent PDMS pillars and (b) Electrically conductive PDMS pillars, on a transparent PDMS base. A cotton applicator was used to apply pressure on particles and guide them inside features of the mold.   3.1.3 Controlled Patterning of the Micropillar Device As mentioned in the introduction of this chapter, magnetic micropillar devices can be used as combined sensors and actuators. In such devices, magnetic and nonmagnetic pillars are built into one device. Previous fabrication of such devices, using the solvent casting methods, was based on the random dispersion of magnetic particles in the mold [22, 24, 26]. As such, there was no control over the placement/location of the magnetic pillars compared to non-magnetic pillars. Control over the relative location of magnetic and non-magnetic pillars is extremely useful, allowing the careful design of devices where the direction of forces and the number of forces (b) (a) 500 µm57  applied on cells can be pre-determined through design. This is where the dry-particle embedding technique poses another significant benefit compared to the solvent casting technique. It allows us to precisely determine the placement of magnetic and non-magnetic pillars in one device, i.e., it allows the patterning of the micropillar array.  To achieve controlled placement of magnetic and nonmagnetic pillars side by side on one device a simple masking technique is added before step 4 of the fabrication process presented in Figure 3.1. To fabricate the required mask, PDMS is spin coated on a glass slide and cured at 70ºC for 3 hours to obtain an 80 μm thick layer. The glass slide and the PDMS layer are then mounted in a laser micromachining center (Quick-lase 50 ST2, New-wave research Inc.) and a green laser beam (with a wavelength of 532 nm) is used to create a predefined pattern of through holes in the PDMS layer using laser ablation. We have observed that the green wave-length gives better quality cuts in our PDMS mask compared to IR and UV3 wavelengths.  The PDMS layer with ablated through holes is then released from the glass slide and serves as a mask for placement of magnetic particles into specified holes of the PDMS negative in step 4. To achieve this, a flip-chip bonding is used to align and mount the PDMS mask on the PDMS negative. The PDMS negative and the mask are then put in a vacuum chamber for 15 minutes to enhance the stiction of the mask to the negative. Once the magnetic (or nonmagnetic) particles are applied in step 4, the mask on the PDMS negative only allows for the movement of the particles into specific holes of the PDMS negative (Appendix D). After the particles are placed at the desired locations, the mask is removed and PDMS polymer is poured over the entire mold. Once cured, the final device is peeled off the mold. Figure 3.4 demonstrates devices with magnetic FeC-PDMS pillars patterned among pure PDMS pillars, fabricated using the described patterning technique. 58  It is important to emphasize several points here. First, the patterning technique described above is achieved with only one polymer casting step. Second, the patterning technique can be combined with the physical embedding technique to pattern electrically conductive or fluorescent pillars. Furthermore, multiple masks can be used successively to pattern magnetic, fluorescent and electrically conductive pillars, side-by-side, on one device. Finally, although it might be possible to combine a similar masking technique with solvent casting methods to pattern micropillar structures, adequate care has to be taken to make sure that the solvents used do not react with the mask applied on the mold. For example, toluene and xylene, which are commonly used solvents for preparation of PDMS ferrofluid for solvent casting [71], would react with the PDMS mask used in our patterning technique. In this case, the PDMS mask would need to be replaced with a mask made from a compatible material.  Figure  3.4: Devices consisting of 40 μm magnetic FeC-PDMS pillars (black) alongside non-magnetic pure PDMS pillars (clear) patterned using the described masking technique and the dry nanoparticle embedding method, (a) & (b)  'check board' pattern (c) & (d) pattern spelling 'UBC'. (d) (a) (b) (c) 300 µm300 µm 59  3.2 Characterization 3.2.1 Material Composition of the Micropillars We used the Energy-Dispersive X-ray spectroscopy (EDX) technique to determine the material composition of cross-sections of multiple micropillars. A sample EDX measurement is shown in Figure 3.5(a).         Figure  3.5: (a) Energy-dispersive X-ray spectroscopy analysis of one magnetic micropillar to obtain relative composition of elements. (b)Weight percentage of Iron in the final micropillar structures, data obtained from EDX analysis of 30 individual pillars. (a) (a) (b) 60  Thirty pillars on four different devices were randomly chosen for EDX analysis. The results showing the distribution of weight percentages of Iron loading in the micropillar structures of the thirty pillars are presented in Figure 3.5(b). Weight percentage results show a range of 35% to 43% by weight Fe loading of the PDMS polymer, with a median of 38%, an average of 38.55% and a standard deviation of 2.05%. The low value of the standard deviation compared to the mean suggests uniform material composition among the pillars of our device. It is important to note that the 38% by weight loading of elemental Iron in our PDMS micropillar structure is larger than or comparable to the highest particle loadings previously achieved for flexible microstructures of similar dimensions fabricated using solvent casting techniques [22, 24]-[26, 71]. Table 3.1 presents a summary of the particle loading ratios for magnetic polymer microstructures achieved to date by other research groups.   Group Magnetic Composite Type Minimum Feature  size Magnetic Particle Loading Ratios Evans et al. [23] Maghemite+PDMS 200 nm diam., 10 µm  height 18% wt. Sniadecki et al.[24]  Co Nanowires + PDMS 3 µm diam., 10 µm  height < 1% wt. Fahrni et al.[25] FeC + PDMS 300 µm x 100 µm x 15 µm 2-3% vol. LeDigabel et al. [26] Iron Oxide + PAM Gel 10 µm diam., 20 µm height 5% wt. Pirmoradi et al.[71] Iron Oxide + PDMS 7mm diam., 37 µm thick 32% wt. Khademolhosseini et al. (This work) FeC + PDMS 40 µm diam., 154 µm  height 35%-43% wt. 7%-8.5% vol.  Table  3.1: Comparison of magnetic particle to PDMS loading ratios for various techniques used to date 61  3.2.2 Young’s Modulus of FeC-PDMS  We determined the Young’s Modulus of the FeC-PDMS using a thermo mechanical analyzer (TA instruments Q400). First, 20% and 40% w/w FeC-PDMS polymer samples were prepared by adding FeC particles to glass vials of Sylgard 184 pre-polymer based on calculated weight ratios, and sonicated for 1 hour in a water bath to get a fully homogenous mixture. Sylgard 184 cross-linking agent was then added at the manufacturer recommended 10:1 by weight ratio and manually mixed for 10 minutes. Pure PDMS was also prepared at 10:1 polymer base to cross-linker ratio. The 20% and 40% w/w FeC-PDMS mixtures and the pure PDMS were then poured on glass trays to a height of 0.6 mm and oven-cured at 70 ºC for 3 hours. After curing, rectangular strips 25 mm long, 1.2 mm wide and 0.6 mm thick were cut from the FeC-PDMS and pure PDMS samples and used for the strain measurements with the thermo mechanical analyzer. To obtain force vs. displacement data, the samples were preloaded at 0.001N and loading was increased at a constant rate of 0.02 N/min to a maximum of 0.15 N at which point the loading was decreased back to 0 N at the same rate of 0.02 N/min. This procedure was repeated on three different samples of each of the materials. The resulting displacement and force data was used to obtain strain-stress curves for the samples. The slope of the obtained stress-strain curves was then used to calculate the Young’s modulus of each of the samples. Fig. 3.6(a) shows sample stress-strain curves for 40% FeC-PDMS and pure PDMS, obtained from the strain tests performed with the thermo mechanical analyzer. Fig. 3.6(b) shows the young’s moduli of pure PDMS, 20% and 40% w/w FeC-PDMS obtained by finding the slopes of their respective stress-strain curves.  Figure  3.6strain curto avoid clfrom the s: (a) Measurve for 20% wutter (b) Youlopes of stressed stress-stra/w PDMS, whng’s Moduli o-strain curvein curves for ich lies in bef pure PDMSs of part (a).pure PDMS tween the two, 20% w/w Fand 40% w/w curves showeC-PDMS an  FeC-PDMSn, is not presd 40% w/w Fe (n=3). The sented here in C-PDMS obt62 tress-order ained 63  We obtained a Young’s modulus of 1.13±0.02 MPa for pure PDMS, within the range of previously reported values [96], and Young’s moduli of 1.18±0.04 MPa and 1.29±0.04 for 20% w/w and 40% w/w FeC-PDMS, respectively. It is observed that adding FeC magnetic particles to pure PDMS results in the increase of the Young’s moduli. Furthermore, we observe that increasing the weight percentage of FeC particles in FeC-PDMS from 20% to 40% causes less than a 10% increase in the measured Young’s moduli of the material. That is, a 20% variation in FeC content only causes a 10% variation in the Young’s modulus. It can therefore be deduced that our fabricated FeC-PDMS micropillars which contain between 35-43% FeC particles by weight (an 8% variation) will only have an estimated 4% variation in their corresponding Young’s moduli. Thus, the Young’s moduli of 40% FeC-PDMS can be used to for all micropillars in our experiments without the introduction of significant error in calculations of the bending performance. 3.2.3 Magnetization Properties of the Micropillars We used a superconducting quantum interference device (SQUID) magnetometer to characterize the magnetization properties of the magnetic PDMS polymer in the micropillar structure. Knowing the magnetization properties of the magnetic PDMS polymer allows us to use theoretical models to calculate the magnetic forces applied on the pillars from a known external magnetic field [97]-[99], facilitating design and optimization of our micropillar structures. Furthermore, a comparison of the magnetization curve of the magnetic micropillars with the magnetization curve of pure PDMS and the magnetization curve of the magnetic particles provides a second method of calculating the magnetic particle loading ratio in the PDMS micropillars. 64  As shown in Figure 3.1, step 4, the nanoparticles are originally dispersed with random orientation on the mold surface. However, if a permanent magnet is used to guide the particles inside the features of the mold, the particles tend to align their magnetic moments with the direction of the magnetic field lines from the permanent magnet, which is in the direction of the pillar axis. In addition, the pillar structure itself is geometrically anisotropic with its longest dimension being that of the pillar axis. We therefore expect the pillars to have an easy axis of magnetization, that is, an energetically favorable direction of spontaneous magnetization, in the direction of the pillar axis [100].  The existence of an easy axis of magnetization in the direction of the pillar axis affects the magnitude of bending loads applied on the pillars by the externally applied magnetic field. When a magnetic field B is applied on the pillars as shown in Figure 3.7, there are two types of magnetic loads that cause the pillars to bend; the magnetic translational force F that pulls the pillars towards the positive field gradient, and the magnetic torque or bending moment M, which rotates the magnetization axis m towards the direction of the applied magnetic field.  Figure  3.7: When the external magnetic field B has a nonzero angle with the pillar magnetization vector M, the magnetic bending moment M causes the pillars to bend. 65  If the magnetic bending moment is large enough, the pillars will bend even when the magnetic translational force is negligible, that is, bending of the magnetically anisotropic pillars can be achieved at very low or zero magnetic field gradients as long as the magnitude of the magnetic field is non-zero. This was previously demonstrated by Sniadecki et al. who embedded cobalt nanowires in PDMS pillars, and aligned them with the pillar axis [22, 24]. They demonstrated that due to the shape anisotropy of the vertically aligned cobalt nanowires, when a horizontal magnetic field was applied as in Figure 3.7, the magnitude of magnetization was largest in the direction of the cobalt nanowires, that is, the magnetization vector was in the direction of the pillar axis. As a result, large magnetic bending moments acted on the pillars in a horizontally applied uniform magnetic field, and made it possible to obtain relatively large magnitudes of pillar bending. To determine whether our magnetic PDMS pillars have a preferred direction of magnetization, the magnetic pillars were first encased in pure PDMS to prevent them from bending in the external magnetic field applied by the SQUID. The magnetization of the magnetic PDMS pillars was then measured at a room temperature of 296 K and in two directions, parallel to and orthogonal to the pillar axis. The results of the measurements obtained from the SQUID magnetometer for pillars (40 μm in diameter, 154 μm in height) are presented in Figure 3.8.  66    Figure  3.8: (a) The external magnetic field was applied in two different directions, orthogonal to and parallel to the pillar axis. (b) Magnetization vs. applied magnetic field for the magnetic PDMS micropillars, measured using a SQUID magnetometer. We observe that for a specific magnetic field value, the magnetic pillars have larger magnetization in the direction of the pillar axis, confirming that one easy axis of magnetization is in this direction. It is important to point out that the aspect ratio of the pillars plays an important role in the observed magnetization characteristics, and the existence of an easy axis in the direction of the pillar axis may in this case only be due to shape anisotropy. We measure a (a) (b) 67  saturation magnetization of 95 emu/g for the magnetic FeC-PDMS and a saturation magnetization of 210 emu/g for the free-standing magnetic nanoparticles, slightly smaller than the 220-222 emu/g saturation magnetization of bulk iron (elemental iron) at room temperature.  [101, 102]. A comparison of the 95 emu/g saturation magnetization we have obtained for the magnetic pillars with the 222 emu/g saturation magnetization of bulk iron, points to a 42% wt. loading of Fe in the magnetic polymer. This is in line with the 35% to 43% wt. loading of Fe we obtained from EDX measurements and serves as a confirmation of those results. We also obtained hysteresis curves for the magnetic FeC-PDMS polymer which show a 0.35 emu/g remnant magnetization and a -8.9 Oersted coercive field.  3.2.4 Bending Performance of the Micropillars To characterize the bending performance of the pillars, a 0.5 by 0.5 by 1 inch Neodymium permanent bar magnet (grade N52, nickel plated, magnetized through thickness) was used to apply a magnetic field across a 6x6 mm2 region of our micropillar arrays (Figure 3.9).  For each device, multiple pillars were randomly chosen and their displacements were measured as the magnitude of the magnetic field was increased from 0 mT to 350 mT (while the magnetic field gradient increased from 0 to 50 mT/mm). The magnitudes of the magnetic field and the magnetic field gradient at the location of the pillars were carefully measured in real time using a Gauss meter (F.W. Bell, Model 6010) and a transverse Hall probe (F.W. Bell, Model STD61-0404-05). To measure displacements, an optical microscope was used to take multiple images of the pillars during bending and the relative position of the pillar tips compared to static markers on the device were translated into displacement using image processing techniques. 68   Figure  3.9: Simultaneous actuation of multiple magnetic PDMS pillars using an externally applied magnetic field. The following first order approximations from Euler-Bernoulli beam theory were then used to relate the displacement ν at the pillar tips to the equivalent bending moment M and equivalent concentrated load P applied at the pillar tip [22]: ܲ ൌ 3݈ݒܫܧଷ , ܯ ൌ2ݒܫܧ݈ଶ  (3.1) Here, l is the pillar length (height) and I is the area moment of inertia of the pillar cross-section. In using the above Euler- Bernoulli equations, we assume that the pillars act as linear cantilever beams and that the bending rigidity (EI) is constant throughout the pillar length. We have used a Young's modulus E of 1.29 MPa based on the strain-tests measurements presented in section 3.2.2 for magnetic FeC-PDMS polymers (with 40% wt. magnetic particle loading ratios). A more detailed analysis of micropillar bending due to a magnetic bending moment, which takes into account deformations of the micropillar due to tilting of the elastic pillar base, is presented in Appendix E. Magnetic Field  Figure 3equivalenmagneticFigure  3.1on pillars pillars and.10 presentt horizonta field and m0: Measured vs. externally (b) 8μm diams the resull load applagnetic fieldpillar tip hori applied magneter, 18μm hts for the ied at pillar gradient. zontal displacetic field andeight pillars.measured h tips for vement (n=6) magnetic fielorizontal darious valueand calculated gradient: (aisplacements of the ex  d equivalent ) 40μm diam and calcuternally aphorizontal tipeter, 154 μm h69 lated plied  force eight 70  It is interesting to compare the magnitudes of the forces we have obtained with those previously obtained from devices fabricated using solvent casting techniques. Le Digabel et al. used black iron oxide to fabricate arrays of magnetic PAM Gel micropillars (E=210 KPa) with heights of 20 μm and diameters of 10 μm [26]. Applying an extremely high magnetic field gradient of 23e3 mT/mm, they observed a maximum tip displacement of 6 μm, which corresponds to an equivalent horizontal tip force of 0.231 μN. In comparison, with our 8 μm diameter 18 μm high pillars, we were able to obtain a significantly larger force of 0.425 ± 0.109 μN using a much lower field gradient of 41.45 mT/mm (and a magnetic field magnitude of 286 mT).  The ability to obtain higher forces using much smaller fields is beneficial for cell mechanics studies were a bigger range of forces can be applied on individual cells to study their response. Also, we have obtained these larger forces using smaller magnetic fields and magnetic field gradients, which are more easily producible, and allow the actuation of multiple pillars covering a large surface area and across larger distances. Table 3.2 presents a summary of the bending moments and tip forces we have obtained with our magnetic PDMS pillars, the values of which are reported based on a 95% confidence level. Pillar Dimensions (μm) Diameter=8  Height=18 Diameter=40  Height=154 Magnetic Field (mT) 286 352.7 Field Gradient (mT/mm) 41.45 50.2 Pillar Tip Displacement (μm) 3.07±0.80 61.41±3.48 Bending Moment (pN.m) 4.99±1.30 839.4±47.5 Horizontal Tip Force (μN) 0.43±0.10 8.18±0.46  Table  3.2: Maximum Displacement & Equivalent Force/ Equivalent Bending Moment on Magnetic Pillars 71  3.3 Conclusion The dry nanoparticle embedding technique presented in this chapter provides a new approach to fabrication of magnetic (and non-magnetic) micropillar structures for cell-mechanics studies. Compared to conventional solvent-casting techniques, the dry particle embedding technique has several benefits. It can be integrated with highly viscous polymers since agglomeration of magnetic particles inside the polymer is not an issue. Also, there is much greater control over magnetic particle placement, i.e., particles are directly placed in parts of the device where magnetic actuation or pillar functionality is desired.  The magnetic particle loading values obtained using the technique presented here are better than or comparable to the best results previously achieved with solvent-casting techniques. Simple modifications to the dry particle embedding technique, allow the embedding of nonmagnetic particles in the pillar structure. Furthermore, magnetic and non-magnetic pillars can be patterned side-by-side on one device in one polymer casting step.  Finally, we have demonstrated pillar tip forces and bending moments larger than previously achieved at significantly smaller values of externally applied magnetic field gradients, significantly increasing the distances over which remote actuation can be achieved. The enhanced actuation properties of micropillar devices presented in this chapter, fabricated using the dry particle embedding technique, will facilitate new experimental cell mechanics studies, specifically those looking at the remote actuation of micropillar arrays for control of cell migration behavior on a multicellular level.  72  Chapter 4: Controlling Cell Migration with Magnetic Polymer Micropillar Arrays  CHAPTER 4  Controlling Cell Migration with Magnetic Polymer Micropillar Arrays  Active polymer micropillar arrays, especially those where actuation can be achieved using external magnetic fields, present great possibilities for the fields of experimental cell-mechanics and tissue engineering. It might be possible for example to integrate polymer micropillar structures in the outer coating of medical implants and apply controlled mechanical loads on cells in the surrounding tissues to elicit specific cellular behavior in vivo. Particularly, if magnetic polymer micropillar arrays prove useful in controlling cell migration, their integration with medical implants could provide a new mechanism to prevent cell-migration induced biofouling of implants or to control wound healing and tissue growth and repair. In this chapter, we present an experimental study on the application and efficacy of magnetic micropillar arrays in controlling cell migration behavior. 4.1 Design Considerations for Cell-Migration Chip  When looking at the migration of cells among magnetic micropillar surfaces, several factors can affect cell migration behavior and cell migration rates.  The first factor is the topology of the 73  micropillar surfaces, i.e., the size, surface density and relative locations of the magnetic micropillars.  The second factor is the application of forces on cells, i.e., actuated vs. non-actuated micro-pillars. The third factor is the direction of force application with respect to the direction of cell migration. To independently assess the net effect of each of these three factors on cell migration rates, multiple levels of control are needed. 4.1.1 Micropillar Size and Spacing Choosing the micropillar size and spacing depends on the cell size and whether the cells are grown on top of the bed of micropillars or in between micropillars. To allow for the attachment, growth, spreading and movement of cells on top of a bed of micropillars, generally, a densely packed array of submicron diameter micropillars is needed [61, 63]. The mechanism by which application of strain on the cells by the micropillars is achieved is also somewhat more complex when cells are grown on the top of the micropillars, since closely packed pillars with different bending rigidities or combinations of magnetic and non-magnetic pillars are required to induce differential bending deformations resulting in straining of the cells (Figure 4.1).  Fabrication of submicron micropillar arrays requires high precision photolithography equipment that is not readily available. Furthermore, since the force and bending moment generated by a magnetic pillar is directly proportional to its magnetic volume, smaller pillars generate smaller forces for the same magnitude of applied magnetic field.  Figure  4.1attachmenbending rbending dIn compahave largsimpler; stationarythe relatcompresscomplexiwould gr:  A denselt, growth, spigidities or coeformations arison, whener pillars thas cell sprea pillar baseive displaceive) strains ty of fabricaow at the boy packed arreading and mmbinations ond cause stra cells are grat create largd at the base and the moment of thin the cellstion, we chttom surfaceray of submovement of f magnetic anining of cells.own in-betwer forces. T of the micrving magnee moving a (Figure 4.2ose to desig of the micrMagicron diametcells on top od non-magneeen the pillahe mechaniopillars foctic pillars [7nd stationa). Due to thn migrationopillar array netic Fielder micropillaf a bed of mitic pillars arrs of a micrsm by whichal adhesions0]. As the mry focal ade benefits o chips with s, in betwe  rs is neededcropillars. Pie required to opillar array the cells ar connect theagnetic pilhesions indf this approlarger pillaren the pillar  to allow follars with difinduce differ, it is possibe strained is cells to botlars are actuuces tensilach and reds where thes. 74 r the ferent ential le to  also h the ated, e (or uced  cells 75        Figure  4.2: Cells growing among magnetic micropillars create adhesions to stationary micropillar base and the moving magnetic pillars. As the magnetic pillars are actuated, the relative displacement of the moving and stationary adhesions induces tensile (or compressive) strains in the cells  We use Human Umbilical Endothelial Cells (HUVECs) in our migration experiments. HUVECs are adherent cells that once spread out have a confluent density of 300-400 cells/mm2 corresponding to an average individual cell surface area of 2500-3300 µm2. Ideally we would like to allow the cells to fully spread out at the base of the micropillar arrays while simultaneously having contact with one or more micropillars for strain application. Based on these criteria, we chose three different center-to-center pillar spacings of 70, 110 and 150 µm corresponding to micropillar densities of 204, 83, 44 micropillars/mm2, respectively, for our micropillar arrays. 76  Initial HUVEC cell culture on the micropillar arrays showed that the arrays with the highest density (with 70 µm interpillar spacing) did not allow newly cultured cells to deposit and to fully spread at the micropillar base, resulting in eventual apoptosis (cell death) of the un-spread cells. Consequently, only the arrays with interpillar spacings of 110 and 150 µm were used in the final migration-chip design.  4.1.2  Controlling for Direction of Force Application vs. Cell Migration As we previously discussed in chapter 2, cells reorganize their internal actin elements and migrate transverse to the direction of force/strain application. To control for directionality effects in our cell migration chips, we designed the chip to confine the micropillar arrays in channels with vertical sidewalls. Consequently, cells growing in these channels were restricted to migrate along the length/axis of the channel (the X-direction) while force application or pillar bending could be applied parallel or transverse to the channel axis. To add a further level of control, two different topological patterns for relative location of micropillars, i.e., a square pattern and a diamond (or staggered) pattern were studied. 4.1.3 Controlling for the Net Effects of Force Application from Micropillars To quantify the net effect of force application from micropillar arrays on cell migration and to separate any changes in migration behavior due to variations in surface topology, chips were designed to have channels/regions with magnetic pillars, and channels/regions having the same topological patterns but with non-magnetic pillars. In the presence of a periodic external magnetic field, cells in the regions with magnetic pillars would undergo cyclic loading from  magneticexternal m4.1.4 TFigure 4design co4 have nspaced eias regionRegion 5net cell min the X oFigure  4.3magnetic p pillars, whiagnetic fiehe Multi-r.3 shows a nsiderationon-magnetither 110 or s 1-4, but a is flat and digration to r Y directio: Schematic oillars, four rele cells in thld and the saegion Cell Mschematic os presented. c pillars, w150 µm apare compriseoes not conthe X-directns.   f cell-migratigions/patterne regions wme topologigration Cf the finalizThe cell-miith pillars lrt (center tod of magnetain any pillion, whereaon chip, desigs with non-mith non-magical conditiohip ed design ogration chipocated eithe center).  Retic pillars thars. The chas pillar bend ned with nineagnetic pillarnetic pillarns, but no cf the cell–ms has nine dr in a squagions 6-9 hat allow apnnels in theing and for different regs and one regs would expyclic loads.igration chifferent regire or diamave the exaplication o micropillarce applicatioions; four reion with no pierience the  ip based oons; regionsond patternct same topof forces on  chip confinn can be ap gions/patternllars (flat PDM77 same n the  1 to , and logy cells. e the plied s with S).  78  4.2 Fabrication of Cell-Migration Chip  To fabricate the cell-migration chips containing both magnetic and non-magnetic micropillars, we used a variation of the dry nanoparticle embedding technique previously demonstrated in Chapter 3.  The dry nanoparticle embedding technique allows for the patterning of magnetic and non-magnetic polymer micropillars on one chip in one polymer casting step, and is ideally suited for the fabrication of our multi-region cell-migration chips.   To fabricate the molds required for the dry nanoparticle embedding technique, first, an SU8 master of the micropillar chip was developed using conventional photolithography, followed by casting and curing of a PDMS layer on the SU8 master to obtain a PDMS negative (Figure 4.4).  This PDMS negative was then treated with oxygen plasma and silanized with Hexamethyldisilazane (HMDS) overnight, and a second PDMS casting and curing was performed to obtain a PDMS replica of the SU8 master. A polyurethane (PU) mixture (Smoothcast 310, Smooth-On Corp.) was then cast on the PDMS replica, cured and de-molded to obtain the reusable mold for fabrication of the cell-migration chips.  Next, employing the dry nanoparticle embedding technique, masking tape was used to cover parts of the PU mold corresponding to regions 1 to 4 on the migration chip, to prevent embedding of magnetic particles in those regions.  Carbonyl Iron (FeC) magnetic particles were then applied to the surface of the mold, and a permanent magnet was used to pull the particles into the cavities of the unmasked regions (regions 6 to 9).  Excess particles were then removed by wiping off with a cotton applicator or kim-wipes.  Next, the masking tapes were removed and a PDMS polymer solution was cast on the PU mold and allowed to cure at room temperature.   Once curmigrationFigure  4.4is demonsted, the PDM chip conta: Schematic orated. S was de-ining regionf fabrication molded froms with magnsteps for cell- the PU metic and nonmigration chiold.  This f-magnetic pp. The dry nainal PDMS illars.    noparticle emlayer is thebedding tech79  cell-nique 80  Figure 4.5 shows an image of the cell-migration chip fabricated using the presented fabrication method. Magnified insets of the pillars of region 7 before and after magnetic actuation are also shown.   Figure  4.5: (Top) Image of a multi-region cell-migration chip having magnetic and non-magnetic pillars and a combination of various micropillar topologies. Chip was fabricated using standard photolithography procedure followed by replica molding and the dry nanoparticle embedding technique, (Bottom) Magnified images showing the magnetic pillars of region 7, before actuation (left) and after actuation in X-direction using external magnetic field (right).  81  4.3 Experiments 4.3.1 Cell Culture and Chip Preparation Human Umbilical Vein Endothelial Cells (HUVECs, EGM-2, cryo amp, code cc-2517A) and culture media (EGM-2 BulletKit, code cc-3162) were obtained from Lonza. The cells were expanded and passaged in culture dishes in the EGM-2 culture media. Cells from passages 3 to 8 were used for conducting experiments. To prepare the cell migration chips and facilitate cell attachment to the hydrophobic PDMS micropillar arrays, the arrays were treated with Oxygen plasma for 120 seconds and immersed in a 22 µg/ml human Fibronectin (FN)/PBS solution for 24 hours. The arrays were rinsed twice with PBS prior to cell culture. The concentration of a Fibronectin coating on a surface has been shown to affect cell-substrate adhesion and cell migration rates [103]. To make sure that all micropillar regions had similar Fibronectin coating densities, a FITC labeled Fibronectin solution was used to coat a sample micropillar chip and check the uniformity of Fibronectin coating across the various micropillar regions. Obtained fluorescence microscopy images were post-processed in ImageJ to find variations of the fluorescence intensity over the different regions of the micropillar array. These variations were found to be less than 5% confirming uniform coating of the arrays. Figure 4.6 shows sample differential interference contrast (DIC) microscopy and fluorescence microscopy images of a coated and non-coated array.  Figure  4.6of the mic4.3.2 ATo checkcultured apply a pto bend. in cell dbending transversfrom mag: Pillars coateropillar regionpplication the efficaat sub-confleriodic 1 HzMost cells inimensions agenerally she to the dirnetic pillard with a fluos. of Strains ocy of strainuent densiti external m physical cos the magnowed higheection of pils, images ofrescently laben Cells fro applicationes among magnetic fieldntact with aetic pillars r values of lar bending sample cellled Fibronectm Magnetic on cells bagnetic pill to the magt least one mbended. Cestretching (. To quantifs were takenin (FN) coati Pillars y the magnars. An elecnetic pillarsagnetic pillls aligned axial strainy the strain prior to anng demonstraetic pillarstromagnet  causing thelar showed with the d) compared  magnituded after actua te uniform co, HUVECs was then us magnetic pperiodic chairection of to those ali applied on tion of mag82 ating were ed to illars nges pillar gned cells netic  pillars. Paligned wpillar benFigure  4.7approximaA qualitamagneticmagneticHz for 24To betterpreviouslost-processiith the pillding are sho: Cell attachetely 5% in thtive study lo pillars wa and non-m hours.   visualize thy describedng of Imagar bending wn in Figurd to a magnee attached ceoking at mos conductedagnetic pillae cell morp in section 2es using Imdirections. e 4.7.  tic pillar on oll.  rphological. To this er arrays. Mahologies, ce.2.4. FixedageJ showeSample imane side. Bend changes innd, HUVECgnetic pillalls were fixand stainedd strain mages of a ceing of the ma HUVECs ss were culrs were thened and stain cells weregnitudes abll before angnetic pillar iubject to pertured to co actuated ated accordinthen imagedove 5% in d after mag nduces a streiodic loads nfluency am a frequencyg to the met on an Olym83 cells netic tch of from ong  of 1 hods pus  IX81 micCite Exaimages oAfter 24 and spreHUVECsapproximapproximFigure  4.8different mmore flattHUVECs roscope equcte (Lumenf cells amonhours, HUVad out mor among noately 250 cately 350 ce: Fluorescentorphologies.ened and spramong non-mipped with  Dynamics)g magnetic ECs subjectphology andn-magnetic ells/mm2 folls/mm2 forly stained HU After 24 houead out morpagnetic pillara 20x air ob light sourcand non-ma to loading  a more ppillars. Fur actuated m non-magneVECs growinrs of periodhology and as. jective, a Coe and Semrgnetic pillarfrom magneronounced rthermore, cagnetic pilltic pillars. g among nonic loading fro more pronouolSnap HQock Quadbs are showntic pillars dalignment oells had a ars, compar-magnetic andm magnetic nced alignm2 camera (Pand Sedat f in Figure 4emonstratedf stress fiblower confed to a con actuated mapillars, HUVEent of stress hotometricsilter set. Sa.8.   a more flatters compareluent densifluent densignetic pillarsCs demonstfibers compar84 ), X-mple ened d to ty of ty of   show rate a ed to 85  4.3.3 Simulated Scratch-Wound Assay To simulate a scratch wound, HUVECs were seeded at sub-confluent densities on the outer sides of two vertically placed, parallel, micro cover glasses spaced 0.9 mm apart and incubated in EGM-2 under 5% CO2 for 12 hours until confluent sheets of cells were observed.  To assist tracking of cell locations, HUVEC nuclei were fluorescently stained with Hoechst stain (Hoechst 33342, Thermo Scientific) for 20 minutes, followed by three washes with PBS to remove excess Hoechst. The micro cover glasses were then removed, allowing the cells to migrate onto the area of the simulated scratch wound.   Immediately after removal of the micro cover glass barriers and start of cell-migration into the wound, a custom designed actuation setup (see Appendix F) integrating moving permanent magnets was used to simultaneously actuate the magnetic pillars on the cell-migration chips in either the X or Y directions at a cyclic frequency of 1 Hz throughout the experiment duration (24 or 48 hours).  At different time intervals, the scratch wounds corresponding to regions 1 to 9 on each of the micropillar chips were imaged using an Olympus IX81 inverted microscope equipped with a 4x/0.13 NA objective and a CoolSnap HQ2 ccd camera. Post-acquisition image analysis was performed in ImageJ to track the cell-sheet front as cells migrated into the wound area.  To obtain an average migration rate for the cells, at different time intervals, the area of the wound in each channel was measured and divided by the constant channel width (dimension of channel in Y-direction) to obtain the average change in the width of the scratch wound (the change in the x-dimension of the wound). The resulting number was divided by the time duration of the experiment to obtain the average rate/speed of cell-migration in the X-direction. Figures 4.9 and 4.10 show sample images from experiments.  FtwTddl igure  4.9: Reprehe case of X-actuere used to trackhroughout the eirection @ 1Hz. ecrease in size (moading showed msentative fluorescation. The nuclei progression of cxperiment, an exAfter 24 and 48 ost regions showuch lower recoverent microscopy im of HUVECs werells into the wouternal periodic mhours, wound aring full recovery y rates. ages of a 48 houe fluorescently stnd. White lines shagnetic field waseas in regions witof the wound) whr scratch-wound ained with Hoechow the extremiti applied to all 9 h non-magnetic ereas wound areassay experimentst at the start of es of the wound regions, causing pillars and in theas in regions with looking at cell mthe experiment aas determined byperiodic bending  flat PDMS regio magnetic pillarsigration in the X-nd the locations  the location of thof magnetic pillan demonstrated  where cells unde86  direction for of the nuclei e cell-front. rs in the X-a significant rwent cyclic  FtwTdssigure  4.10: Reprhe case of Y-actuere used to trackhroughout the eirection @ 1Hz. Aize (with some rehowed lower recoesentative fluoresation. The nuclei progression of cxperiment, an exfter 48 hours, wgions showing fulvery rates. cent microscopy i of HUVECs werells into the wouternal periodic mound areas in regl recovery of the wmages of a 48 houe fluorescently stnd. White lines shagnetic field wasions with non-maound) whereas wr scratch-wound ained with Hoechow the extremiti applied to all 9 gnetic pillars andound areas in reassay experimentst at the start of es of the wound regions, causing  in the flat PDMSgions with magne looking at cell mthe experiment aas determined byperiodic bending  region demonsttic pillars where igration in the X-nd the locations  the location of thof magnetic pillarated a significancells underwent c87  direction for of the nuclei e cell-front. rs in the Y-t decrease in yclic loading  Figure 4.when no Figure  4.1i.e., when rates to noMicropillain the averspacing (1Figure 4.Y-actuati 11 shows thmagnetic fie1: Measured no external mn-magnetic pr arrays withage cell migr50 µm). 12 shows thon.  e measured ld was appldata (n=3) foragnetic fieldillars. The ex higher densiation rates coe measured X-directionied to the ch the average X was appliedistence of micties and smalmpared to mX-direction cell migratiip.  -direction ce. Regions witropillars reduler interpillaricropillar arrcell migration rates for ll migration rh magnetic pced migratio spacing (110ays with loweon rates for the case of n ates for the cillars showedn rates comp µm) showed r densities anthe cases of o actuationase of no actu similar migared to flat Pa higher redud larger interX- actuation88 , i.e., ation, ration DMS. ction pillar  and  Figure  4.1actuation.periodic bactuated mmagnetic p2: Measured  In both casesending of maagnetic pillaillars. data (n=3) for the externalgnetic pillarrs showed si the average  periodic mags in either thgnificantly lo  X-direction cnetic field wae X or Y dirwer migratioell migration s applied to ections @ 1n rates comp  rates; (a) X-aall 9 regions oHz, respectivared to the ctuation and f the chip, caely.  Regionsnon-actuated89 (b) Y-using  with  non-90  In order to isolate the effect of cyclic loads from micropillar actuation on cell migration rates, and in order to compare and analyze results from multiple experiments, we define a non-dimensional migration rate or NDMR as follows; ܴܰܯܦ ൌ Migration	rate	among	magnetic	pillars	with	topology	݅Migration	rate	among	nonmagnetic	pillars	with	topology	݅ In the above equation, the migration rate of cells in a region with magnetic pillars is normalized with respect to the migration rate of cells in a region that: (a) is on the same cell migration chip, (b) has non-magnetic pillars and (c) has the same topology, i.e., same pattern and spacing of pillars. These criteria ensure that any variations in cell motility due to the passage number of the cell or due to possible variations in the Fibronectin coating between different chips are accounted and controlled for. Based on the definition of the NDMR, an NDMR value around 1 indicates no or minimal changes in cell migration rates due to the action of magnetic pillars, an NDMR value lower than 1 indicates a reduction in cell migration rates due to the action of magnetic pillars and an NDMR value higher than 1 indicates an increase in cell migration rates due to the action of magnetic pillars. Figure 4.13 shows the NDMR obtained for the cases of no actuation, X-actuation and Y-actuation based on experimental data from 3 sample measurements.  Figure  4.1topologiesactuation NDMR fowith that tfrom the spresented and that X3: Non-dimen for the three and Y-actuatr each topoloopology by thame cell-migr(n=3). It was-actuation wasional migradifferent actuion. In all cagy was obtaine migration ration chip. D observed thas more effecttion rate (NDation scenarises net migraed by dividinate of HUVEata was obtait actuation ofive in impedinMR) of HUVos, i.e., no acttion was cong the migratCs among nonned from 3 sa magnetic pilg migration iECs among muation (no extfined to and ion rate of H-magnetic pimple measurlars significann the X-direcagnetic pillaernal magnetstudied in thUVECs amonllars with theements for eatly reduced ction. r arrays of vaic field appliee X-directiong magnetic p same topologch of the topoell migration91  rious d), X-. The illars y and logies  rates 92  In the absence of an external magnetic field (Figure 4.11), i.e., when no periodic loads were applied on the cells from the magnetic pillars, HUVECs growing among magnetic pillars had average migration rates of 12.29±0.71 µm/hr compared to average migration rates of 11.75±0.74 µm/hr for non-magnetic pillars. The average NDMRs obtained for magnetic pillars of various topological patterns in the absence of micropillar actuation was 1.05±0.05, pointing to a minimal increase in cell migration rates among magnetic pillars compared to non-magnetic pillars (Figure 4.13). Furthermore, in the absence of an externally applied magnetic field, the migration rate of HUVECs among the micropillars was slightly lower than the migration rates of HUVECs on flat PDMS with no pillars (Figure 4.11). It was observed that the mere existence of micropillars reduced cell migration rates for all the different topologies studied, and topologies with a higher density of micropillars generally showed larger reductions in cell migration rates. While HUVECs migrating on Flat PDMS had average migration rates of 13.72±1.73 µm/hr, HUVECs migrating among micropillar patterns with 150 µm spacing had average migration rates of 12.55±0.64 µm/hr and HUVECs migrating among micropillar patterns with 110 µm spacing had average migration rates of 11.48±0.44 µm/hr.  The maximum reduction in cell migration rates solely due to the effects of surface topology was approximately 17% for micropillars with center to center spacing of 110 microns arranged in a square pattern (Figure 4.11). In all cases the observed migration rates for HUVECs were consistent with previously reported range of values for endothelial cell sheet migration rates [104]. In the presence of an external magnetic field, application of periodic loads on cells from magnetic pillars caused a significant reduction in cell migration rates (Figure 4.12). Since both magnetic and non-magnetic pillars in X-actuated and Y-actuated chips were subjected to the 93  same external magnetic field, this reduction in cell migration rates among the magnetic pillars compared to the non-magnetic pillars can be solely attributed to the effect of periodic loads applied on the cells by the magnetic pillars, not to the effects of the external magnetic field. Looking at the average NDMR values obtained from measurements (Figure 4.13) actuation of magnetic pillars on average caused more than a 50% reduction in cell migration rates  for all the different topologies studied (since NDMR <0.5), with some topologies such as the square pattern with 110 µm spacing proving more effective in impeding cell migration (by as much as 85%). An independent sample student t-test comparing the migration rates of cells among magnetic and non-magnetic pillars showed that the reduction of migration rate of HUVECs among actuated magnetic pillars was statistically significant (P<0.05).  Interestingly, all regions with magnetic pillars showed lower X-direction cell migration rates when pillars were actuated in X-direction compared to when pillars were actuated in Y-direction, i.e., X-actuation was more effective in impeding cell migration in the X-direction. This behavior is consistent with previous reports from membrane stretching experiments where endothelial cells showed a lower tendency to migrate in the direction of force application and a higher tendency to migrate transverse to the direction of strain/force application [31].  Among all the different patterns and directions of force application studied, magnetic pillars with center to center spacing of 110 µm, arranged in a square pattern and actuated in X-direction caused the largest reduction in X-direction cell migration rates (Figures 4.12 and 4.13). 94  4.3.4 Long-Term Cell-Migration Assay To see the long term effect of cyclic loading from the micropillar arrays on HUVEC migration rates, and to check whether migration rate of HUVECs among actuated magnetic pillars would remain low or recover to normal values over time, a second set of experiments were conducted. In these set of experiments only the X-actuation square pattern micropillar arrays with 110 µm spacing was studied as it proved to be the most effective topology and direction of force application in impeding cell migration based on results from the short-term wound healing assays.   To conduct the long-term migration study, HUVECs were cultured to confluency on the magnetic and non-magnetic micropillar arrays on one side of a micro cover glass barrier. The micro cover glass was then removed and the migration of HUVECs among the micropillars was studied over a 12 day period.  To assist locating of cells, HUVEC nuclei were fluorescently stained with Hoechst. The chips were imaged at the start of the experiment (Day 1) and at the end of day 12 of the experiments and the images were analyzed with ImageJ to find the locations of the cell fronts at those two time points. Figure 4.14 shows images obtained at day 1 and day 12 for a sample experiment.     Figure  4.1migrationwith Hoeccells sheetmagnetic pillars), canon-magnpillars.  4: Sample fluin the X-direhst at the sta. White lines field was appusing periodetic pillars deorescent micction for the rt of the expeshow the localied to all ric bending of monstrated aroscopy imagcase of X-actriment and thtion of the cegions (regiomagnetic pill significantlyes of a long-tuation. The ne locations oell front. Thrns with magars in the X- higher cell merm cell-migruclei of HUVf the nuclei woughout the enetic pillars direction @ 1igration comation experimECs were flere used to trxperiment, aand regions Hz. After 12pared to regent looking uorescently stack progressn external pewith non-ma days, regionions with ma95  at cell ained ion of riodic gnetic s with gnetic  Figure 4and 12-dmicropilland no acFigure  4.1magnetic Throughomigrationterm 1-datime perioHUVECsrates of 0to values.15 shows thay non-dimear arrays. Thtuation) are5: Short-termand non-magut the experchips, causiny experimentds (12-days).  migrating a.14±0.05 an of 0.98±0.e cumulativnsional mige non-dime also presen (1-day) and netic pillar aiment, an exg periodic bens, actuated mmong actuad 0.37±0.0906 and 1.24e results oration ratesnsional migted for compLong-term (1rrays with sternal periodding of magnagnetic pillarted magneti for the 1-d±0.03 for cbtained fromof HUVECration rates arison purp2-day) non-diquare patteric magnetic etic pillars ins proved effec pillars shoay and 12-dontrols. Wh 3 sample s among thefor control coses. mensional mn and interpfield was ap the X-directctive in impewed averagay experimile cells mimeasuremen magnetic ahips (with n igration rate illar spacingplied only tion @ 1Hz. Sding cell mige non-dimenents, respecgrating amots for the nd non-mago magneticof HUVECs a of 110 µm o the X-actuimilar to the ration over lsional migrtively, compng the mag96 1-day netic  field mong (n=3). ation short-onger ation ared netic 97  pillars showed a slight recovery of the migration rate in 12-day experiments compared to 1-day experiments, the magnetic micropillar array with X-actuation still proved effective in impeding cell migration over the 12-day time period. The total newly covered surface area and the cell migration rate of HUVECs for magnetic pillars was on average less than 40% of those for non-magnetic pillars in the 12-day experiments.  4.3.5 Experiments with Live Imaging A wound healing assay with live-imaging was performed to track movement of cells into the wound area and to study the differences in migration patterns as cells migrated among non-magnetic and actuated magnetic pillars. To perform the live-imaging experiment, first a simulated scratch wound was prepared on a micropillar chip using the culture methods and techniques described in sections 4.3.1 and 4.3.2. The chip was then placed in a custom-fabricated culture dish and loaded onto a custom designed electromagnet resting on the moving stage of an incubated microscope at 37ºC and 5% CO2 conditions (see Appendix G). Cells were then allowed to migrate into the wound areas among magnetic and non-magnetic pillars (square pattern, 110 µm spacing) for a 24 hour period.  Throughout the experiment a 100 mT periodic magnetic field was applied to the micropillar chip using the electromagnet, causing the bending of magnetic pillars in the X-direction (in the direction of net migration) at a frequency of 1 Hz. During the experiment, DIC images of the scratch wound were obtained at regular 1 minute time intervals to track movement of cells. Post processing of the images was performed in ImageJ and the cell trajectories were determined using the MTrackJ plugin. Figure 4.16 shows the movement of cells into a simulated scratch  wound owound haFigure  4.1pattern @The obtainthe traject25 samplen a micropive been plo6: Sample im 110 µm spaced set of 144ories of cells  cells are demllar array, wtted. ages of a 24 hing). Images 0 images wasas they migraonstrated ovehere the traour wound-hof the wound then post-prted to the edgr 24 hours.  jectories ofealing assay p were taken aocessed in Imes of the cell s 25 sampleerformed on t 1 min time ageJ using thheet and into cells as thea micropillarintervals ovee MTrackJ p the wound ay move int  array chip (sr a 24 hour plugin to deterea. Trajector98 o the quare eriod. rmine ies of 99  To compare the migratory behavior of cells among magnetic and non-magnetic pillars, for each of the magnetic and non-magnetic pillar regions, 25 cells were randomly chosen from the edges of the wound at 24 hours and their trajectories were tracked back to the start of the experiment. The time and location data obtained for the cell trajectories were then used to calculate the various quantities such as the total distance traveled and the average velocity of migration for each of the cells studied (see Appendix H). The resulting data obtained for the 25 cells migrating among magnetic pillar was compared to the data obtained for the 25 cells migrating among non-magnetic pillars. For each of the quantities studied, a two sample T-test was used to compare the distributions from the two groups to determine the significance of differences observed, where P<0.05 indicates significantly different distributions based on a 95% confidence level. Table 4.1 gives a summary of the results.  Table  4.1: Comparison of distances traveled and average velocities of HUVECs migrating among non-magnetic and actuated magnetic micropillar arrays (square pattern, 110 µm spacing) Quantity Studied Non-Mag. Pillars Magnetic Pillars t-Test Results Dist. cells trav. along their path (µm) 693.2 ± 131 555.9 ± 99.6 t(45)=4.08, P=1.8e-4 DX= Tot. dist. trav. in X-direct. (µm) 476.4 ± 84.3 355.3 ± 65.2 t(45)=5.57, P=1.4e-6 DY= Tot. dist. trav. in Y-direc. (µm) 412.1 ± 112.9 349.1 ± 85.9 t(45)=2.27, P=3.5e-2 RXY= Ratio of DX to DY  1.22 ± 0.28 1.06 ± 0.26 t(48)=2.05, P=4.6e-2 Net X-direction migration dist. (µm) 386.2 ± 92.44 276.8 ± 59.2 t(41)=4.87, P=1.7e-5 Average velocity along path (µm/hr) 29.5 ± 4.79 23.5 ± 4.19 t(47)=4.56, P=3.6e-5 Net X-direction migration rate 16.1 ± 3.85  11.53 ± 2.46 t(41)=4.87, P=1.7e-5  100  It was found that HUVECs migrating among non-magnetic pillars had significantly longer trajectories (P<0.001) and moved further distances compared to HUVECs migrating among magnetic pillars. Consequently the average velocity of the cells along their trajectories was significantly higher (by about 25%) for non-magnetic pillars compared to magnetic pillars. HUVECs migrating among non-magnetic pillars had significantly more movement in the direction of net migration (X-direction) than transverse to the direction of migration (Y-direction), whereas HUVECs migrating among magnetic pillars moved similar distances in both directions. RXY which is the ratio of DX, the total distance traveled in X, to DY, the total distance traveled in Y, can be used to compare migration trajectories of cells. A higher RXY value shows a more directed migration in the X-direction and towards the wound area whereas a lower RXY shows a more random movement. It was observed that cells migrating among non-magnetic pillars showed significantly higher (by about 15%) RXY values and a more directed migration towards the wound area, compared to cells migrating among magnetic pillars. The net X-direction migration distance and the net X-direction migration rate were significantly higher (by about 40%) for cells migrating among non-magnetic pillars compared to cells migrating among magnetic pillars. A qualitative analysis of the cell movements among micropillars showed that cells demonstrated a ‘stop and go’ type of movement, initiated by extension of the cell membrane and attachment of lamellipodia to one or more of the surrounding pillars, followed by movement of the cell body in between the pillars or towards one of the pillars (the ‘go’ portion of the movement). Proximity of a cell with a specific pillar was usually followed by a period of time where the cell remained adhered to that pillar and showed minimal movement (the ‘stop’ portion of the movement), 101  before detachment and further movement. To quantify the stop and go movement of the cells, for each of the 50 cells studied, the instantaneous velocity of the cells and the time spent at that instantaneous velocity was calculated for the whole duration of the 24 hour experiment. It was observed that cells that attached to a pillar had instantaneous velocities smaller than 0.1 µm/min (or 6 µm/hr) during the ‘stop’ portion of their movement and instantaneous velocities higher than 0.1 µm/min during the ‘go’ portion of the movement. Based on the distributions obtained for the time duration of various instantaneous velocities I.V. of the 50 cells (see Appendix I), cell movement speed during a time period was categorized into the following 4 groups; stationary for I.V.<0.1 µm/min, slow-moving for 0.1<I.V.<0.4 µm/min, fast-moving for 0.4< I.V.<1.0 µm/min and very-fast-moving for 1.0<I.V. µm/min. Table 4.2 shows the percentage of time spent at different movement speeds for the 50 cells studied. A two sample T-test was used to compare the distributions obtained for magnetic and non-magnetic pillars and determine the significance of differences observed, where P<0.05 indicates significantly different distributions based on a 95% confidence level. Table  4.2: Comparison of percentages of total time spent at different movement speeds based on calculated instantaneous velocities I.V. of 25 cells migrating among non-magnetic and 25 cells migrating among actuated magnetic micropillar arrays (square pattern, 110 µm spacing)  Non-Mag. Pillars Magnetic Pillars t-Test Results % of time I.V.<0.1 µm/min 6.7 ± 0.7 13.9 ± 0.9 t(44)=2.93, P=5.3e-3 % of time 0.1< I.V.<0.4 µm/min 41.7 ± 11.2 47.0 ± 10.7 t(48)=1.67, P=1.0e-1 % of time  0.4< I.V.< 1.0 µm/min 42.5 ± 8.4 33.8 ± 9.9 t(47)=3.22, P=2.3e-3 % of time  1.0< I.V. µm/min 9.1 ± 5.4 5.3 ± 3.1 t(38)=2.99, P=4.9e-3  102  It was found that cells migrating among magnetic pillars spent approximately 13.9% of the duration of the experiment in idle or stationary mode, significantly higher (more than twice) the 6.7% of time spent stationary by cells among non-magnetic pillars. Cells spent a marginally (P=0.1) longer duration at slow speeds and significantly shorter duration at fast and very fast speed among magnetic pillars compared to non-magnetic pillars.  4.3.6 Discussion of Results It has been previously reported that for in-vitro scratch-wound assays, migration and wound healing is accomplished through the mechanism of lamellipodial crawling, where cell migration is triggered by an initial injury or by the availability of free space for migration through various biochemical signaling processes, resulting in the crawling of the cells into the wound area [105]-[108]. This crawling motility in general requires at least three distinct actions by the cell; polymerization and extension of the leading edge, adhesion to the substrate and retraction of the cell rear [109]. Furthermore, the adhesive cell-substrate interactions which are required for sustained migration, affect the migration rates in a biphasic manner [105], i.e., cell migration rates peak at an intermediate adhesion value, below and above which cells migrate at slower rates [110]. We postulate that the slight changes in cell migration rates we have observed for the micropillar surfaces compared to the flat PDMS surface in the absence of micropillar actuation (Figure 4.11) can be attributed to two different underlying causes. First, cells migrating among the micropillar arrays generally observe a lower amount of free or available space for migration due to the presence of the pillars. The reduced availability of free space for migration could affect the biochemical signaling events that trigger crawling of cells into the wound area [105]. This is also 103  consistent with our observation that a higher density of micropillars generally resulted in lower cell migration rates, i.e., pillar arrays with 110 µm interpillar spacing showed lower migration rates compared to those with 150 µm interpillar spacing (Figure 4.11). Second, it is likely that attachment of cells to the pillars changes the cell morphology and cell-substrate adhesion strength compared to cells on flat PDMS. It was previously shown that cells growing on 2D substrates generally demonstrated distinct stress fibers formed in the middle of the cell whereas cells spreading among micropillar arrays had contractile actin belts that surrounded the micropillars, and the most prominent actin structures appeared on the periphery of the cells [70]. Furthermore, cells growing among micropillars demonstrated a high number of focal adhesions to the micropillars [70]. It has been reported that adhesive area strongly modulates adhesion strength [111] and that changes in cell-substrate adhesion effect cell migration speeds [110]. Changes in cell-substrate adhesion could thus be another factor causing reduced cell migration rates among micropillar arrays compared to Flat PDMS in the absence of micropillar actuation (Figure 4.11). We postulate that the significant reduction in cell migration rates among actuated pillars compared to non-actuated pillars (Figures 4.12 and 4.13) is due to changes in cell-substrate adhesion and interruptions in the cycle of events required for the crawling motility of cells. Our observation that cells subject to periodic loads from magnetic pillars have a more flat and spread-out morphology compared to cells among non-actuated pillars (Figure 4.8) suggests that the former group of cells exhibit higher cell-substrate adhesion values compared to the latter group of cells. This increase in cell-substrate adhesion can be a cause for reduced cell motility and cell migration speeds [110]. Further, our observation that cells migrating among actuated pillars 104  generally spend more than double the time in stationary or idle mode compared to cells among non-actuated pillars (Table 4.2), could be due to disruptions in the cell locomotion cycle, i.e., the periodic loads applied to a cell by an actuated pillar could be disrupting the normal adhesion, contraction and de-adhesion cycles of the cell as it tries to move. Disruptions of the normal cell locomotory cycle due to applied periodic loads could also be the reason behind our observation of a more directed migration of cells towards the wound area among non-actuated pillars compared to the more random movement of cells among actuated pillars (Table 4.1). While in the absence of pillar actuation haptotacitc and chemotactic migration of cells is directed toward the wound area due to cell crawling toward ECM proteins [32, 34], periodic loads affect actin polymerization rates and actin orientation angles, causing changes in cell migration rates and the direction of cell migration (with migration occurring transverse to applied loads) [31, 72, 112]. Application of loads from X-actuated micropillars would thus cause a preferential migration of cells in the Y-direction counteracting their haptotactic affinity for X-directed migration towards the wound area (Table 4.1). Consequently, a higher density of micropillars would mean an increased number of loading sites and a more pronounced change in cell migration behavior due to applied periodic loads, which is consistent with our experimental findings where migration was impeded more by micropillars with a lower interpillar spacing of 110 µm (Figure 4.13). 4.4 Concluding Remarks In this chapter the effectiveness of actuated magnetic micropillar arrays in controlling the migration rate of cells was demonstrated. Using a novel fabrication technique a micropillar chip having multiple magnetic and non-magnetic regions was fabricated and wound-healing assays were conducted on the micropillar chip to assess the effects of various micropillar topologies on 105  cell migration rates in the presence and absence of micropillar actuation. We found that periodic loads applied on cells from magnetic micropillars significantly reduced cell migration and wound closure rates (by as much as 85%), and that micropillar arrays with a higher density were more effective in impeding cell migration. Further, we found that cell migration is impeded more significantly when micropillar actuation is applied in the direction of net migration. Long-term experiments conducted over a 12-day duration confirmed results from short-term experiments, proving the efficacy of actuated micropillars in impeding cell migration over longer periods of time. Overall, based on the results obtained we conclude that micropillar actuation provides a useful technique for impeding cell migration and that actuated micropillar surface could possibly be used as a means to prevent the cell-migration induced biofouling of medical implants.  106  Chapter 5: Conclusions and Future Work  CHAPTER 5  Conclusions and Future Work  Cells sense and respond to mechanical stimuli in their surrounding environment and the extracellular matrix (ECM). Mechanical stimuli can be in the form of variations in the ECM’s mechanical properties such as the mechanical stiffness or in the form of mechanical loads, i.e., mechanical stresses and strains applied to the cells from the ECM. The cell response to mechanical stimuli is often observed in the form of changes in proliferation and migration behavior. This thesis presented the concepts, fabrication processes and proof-of-principle cell migration experiments for an active polymer micropillar device aimed at controlling cell migration through mechanical stimulation. The developed polymer micropillar device, which consists of Carbonyl Iron PDMS micropillars on a pure PDMS base, proved effective in impeding cell migration through application of controlled and directional periodic loads on migrating cells. Furthermore, actuation of the pillars was achieved through magnetic means using an externally applied magnetic field, eliminating the need for an on-board power source.  The effectiveness of the micropillar device in controlling cell migration through application of periodic mechanical stimuli presents great possibilities for the development of a new class of medical implants integrating active micropillar surfaces with anti-biofouling properties. 107  This thesis included chapters on the development and characterization of a new device for conventional cell-strain experiments (Chapter 2), a new technique for fabrication of magnetic polymer micropillar structures for cell-strain application (Chapter 3) and proof-of-concept in-vitro cell migration experiments on micropillar arrays (Chapter 4), a summary of which is presented here. 5.1 Summary In Chapter 2 we demonstrated the Magnetically Actuated Cellular Strain Assessment Tool (MACSAT), a new device we designed and built for the application of substrate strains on cells. Contrary to most of the previous membrane-stretching based cell-strain devices demonstrated by other groups, the MACSAT is made with low-cost off-the-shelf components, it is easily integrated with existing incubators and microscopes in cell culture labs and it can be easily replicated by other groups wishing to study the effects of substrate induced strains on cells. We then presented a characterization of the strain field developed in the MACSAT membrane, showed the existence of a plateau region in which the maximum strain magnitude had less than a 10% variation in magnitude, and demonstrated that cells in the plateau region of the MACSAT membrane experienced strain fields similar to that of uniaxial stretching. Having characterized the strain field in the MACSAT, we used it to conduct cell-strain experiments on Human Umbilical Vein Endothelial Cells (HUVECs) and study the reorientation of the cells and their internal actin filaments in response to cyclic stretching and uniaxial strains. Using a systematic and objective approach to measure orientation angles of cells and actin filaments, we found that HUVECs (and their internal actin filaments) undergoing cyclic stretch, re-oriented away from the stretching direction towards angular orientations with zero axial deformation, confirming 108  results from past studies. A further comparison of final actin orientation angles and the tensile strains at said orientation angles revealed a 1% strain threshold, above which cell and actin reorientation was triggered and below which cell and actin reorientation remained unchanged. We also found that actin filaments in strained cells were more coherently aligned compared to actin filaments within non-strained cells and proposed that the coherency of actin alignment could be used to determine the extent of actin alignment in strained cells. In Chapter 3 we presented the dry nanoparticle embedding technique, a novel method for the fabrication of polymer micropillar structures having micropillars with embedded functional particles, i.e., magnetic, electrically conductive or fluorescent particles. We showed that the dry nanoparticle embedding technique offers several unique advantages compared to previous solvent-casting methods used for the fabrication of magnetic micropillar arrays; the dry nanoparticle technique does not require the use of solvents or the sonication of particles in a fluidic medium to obtain a homogenous solution, it can be used with highly viscous polymers such as Polydimethylsiloxane (PDMS), and it allows the patterning of functional and non-functional (magnetic and non-magnetic) pillars on the same chip in one polymer casting step, significantly reducing the time and effort required for the fabrication of micropillar chips with functional pillars. Furthermore, contrary to previous solvent casting methods where functional particles were applied to microfabrication molds using a fluidic medium, in the dry nanoparticle embedding technique the functional nanoparticles are applied to the microfabrication mold in their dry particulate state. Therefore, embedding of particles in the mold features can be achieved using physical pressure without the need for complicated sedimentation techniques. This allows the technique to be used not just with magnetic particles, but also with other functional particles 109  such as electrically conductive and fluorescent particles. Using the novel method presented, we fabricated arrays of Carbonyl-Iron-PDMS polymer magnetic micropillar structures with a range of dimensions. We then proceeded to characterize the magnetic and mechanical properties of the magnetic micropillars arrays as well as the corresponding material composition. We demonstrated that compared to previous solvent casting techniques the dry nanoparticle embedding technique achieves much higher particle loading ratios in the structure of the micropillars, significantly enhancing the magnetization properties of the micropillars. We then compared the magnetic forces developed by our magnetic micropillars to previous state-of-the-art magnetic micropillars of similar dimensions made with solvent casting technique, and showed that our magnetic micropillar structures were able to develop much larger forces at significantly lower (by two orders of magnitude) magnetic field gradients. The favorable magnetization properties of our micropillar arrays make them a better candidate for cell strain experiments, as magnetic actuation can be achieved over larger surface areas and from larger distances, allowing truly remote actuation for the first time. In Chapter 4 we presented proof-of-concept cell migration experiments studying the effects of periodic loads from magnetic micropillar arrays on cell migration behavior in order to assess the usability of magnetic micropillar arrays in anti-biofouling and tissue engineering applications. We first demonstrated the design of a multi-region cell migration chip comprised of regions with magnetic and non-magnetic pillars with various micropillar topologies. The design of the multi-region cell migration chip allowed for the conducting of cell migration experiments and wound healing assays with multiple levels of control, in order to isolate the net effects of micropillar actuation on the migration behavior (migration rate) of cells. Having fabricated the cell-110  migration chips using the novel techniques developed in Chapter 3, we conducted initial cell culture and cell-strain experiments to prove the efficacy of the micropillars in application of strains on cells. We demonstrated that cell-strains of 5% and above were achievable using magnetic actuation of the fabricated micropillars, and that cells growing among actuated micropillars had a more flat and spread-out morphology compared to cells growing among non-actuated micropillars. We then presented various short and long term cell migration assays (wound healing assays) on the micropillar chips comparing the sheet migration rate of cells for actuated and non-actuated pillars. We showed that actuation of micropillars significantly impeded cell migration rates (by up to 85%) and that cell migration was impeded more significantly when micropillar actuation was in the direction of migration. Conducting experiments with live-imaging to track cell trajectories, we found that in the absence of micropillar actuation individual cells had higher migration velocities (by about 25%) and a more targeted/directed migration (by about 15%) towards the wound region. We also showed that cells among actuated pillars spent more time in idle or stationary mode compared to cells among non-actuated pillars, and argued that this was due to the stronger adhesion of cells and the interruption of the cell locomotory cycle due to the periodic loads applied from the pillars. Overall, we concluded that micropillar actuation provides a useful technique for impeding cell migration and that actuated micropillar surface could be used as a means to prevent the cell-migration induced biofouling of medical implants. 5.2 Future Work This section discusses some future work that can be pursued to complement the work done in the current thesis. 111  5.2.1 Micropillar Chip Design Variations The micropillar chip demonstrated in this thesis was designed so that the cells were grown in between the pillars. This design was chosen for two reasons; first, the design allowed for application of larger forces on cells attached to pillars due to the larger dimensions of the pillars and second, the microfabrication facilities available for conducting the fabrication for this work did not allow for the fabrication of submicron sized pillars. A minor problem with the current design is that while magnetic micropillars provide a means to apply loads on migrating cells the mere existence of the pillars presents an obstacle in the path of migrating cells. As discussed in chapter 4, variations of the micropillar chip that have densely packed sub-micron sized pillars could allow the growth and attachment of cells on top of the pillars instead of in-between the pillars. This new configuration might provide new dynamics for the migration of cells, since magnetic pillars can still be used for application of loads on adherent cells without the pillars being obstacles in the path of migrating cells.  A second variation in design that would allow cells to be grown on top of a bed of micropillars is to fill the empty space in between the micropillars with a low stiffness medium (a low stiffness polymer). This would allow the magnetic pillars to be actuated, deforming the medium in between the pillars and applying substrate loads to cells growing on top of the medium. We have had some success with this second approach, using a low stiffness silicone elastomer (Sylgard 527, Dow Corning) to fill in the voids between our Carbonyl-Iron-PDMS micropillar arrays, making a new class of cell migration chips with magnetic actuation where cells can grow on a flat surface on top of the micropillars. Further modification of the surface is required however, before cell migration experiments can be conducted on these newly developed chips. We 112  postulate that by using this new class of the micropillar chips it might be possible to apply more uniform periodic substrate strains on adherent cells allowing more a controlled and uni-directional migration similar to conventional cell-strain experiments. 5.2.2 Variations in Actuation Parameters The experiments conducted in this thesis looked at changes in cell migration rates due to periodic loads applied at 1 Hz. This loading frequency was chosen as it close to the natural frequency of periodic loads applied on endothelial cells (and most other vascular cells) in the body at resting conditions, and it is the most widely used frequency for periodic strain application on cells in conventional experiments based on substrate stretching. It is important to consider that previous membrane based cell-strain experiments have shown a frequency dependent behavior in the reorientation response of cells to cyclic substrate stretching [7]. In the case of our experiments with the micropillar arrays, we might find that by increasing or decreasing the frequency of periodic loads on the cells from the micropillars actuators, a more significant change in cell migration behavior is achieved. We therefore suggest a series of follow-up experiments at different actuation frequencies using the micropillar arrays. 5.2.3 Cell Proliferation Studies The experiments presented in this thesis were designed to study changes in cell migration rates in the presence of periodic loads from magnetic micropillar actuators. Further studies looking at the effect of micropillar actuation on cell proliferation (cell growth and cell division) could be pursued to complement the findings of this thesis.  113  5.2.4 Long-term Biocompatibility Studies  An interesting research direction would be to further investigate the long-term biocompatibility of the Carbonyl-Iron-PDMS micropillar arrays. It has previously been shown that iron-iron-oxide and iron-carbon core-shell nanoparticles are biocompatible and a suitable candidate for use in biomedical applications [91, 92]. PDMS microdevices have also previously been the subject of biocompatibility studies and various PDMS implants have been successfully deployed in-vivo [113]. However, mild to moderate inflammatory response two weeks after the implantation of PDMS implants have also been reported [114]. It is therefore of interest to study the long-term biocompatibility of the micropillar devices presented in this thesis to verify its suitability for in-vivo applications. An interesting addition to the biocompatibility studies of the magnetic micropillar arrays would be to replace the PDMS polymer in the micropillar arrays with other biocompatible, biodegradable, or bioabsorbable polymers. Any polymer that can be cast in fluid form and cured to a solid consistency with medium elasticity can be interchangeably used instead of PDMS based on our fabrication method. The fabrication of magnetic micropillar arrays with biodegradable and bioabsorbable polymers requires further study on the biocompatibility of the magnetic particles which would be released into the body as the encompassing polymer degrades. 5.2.5 In-vivo Application Studies Another interesting research direction would be to investigate the in-vivo application of the device using animal trials. Based on the results from our in-vitro experiments where we observed 114  a significant reduction in cell migration rates among actuated magnetic pillars, we have suggested that actuated micropillar arrays can be used as new surfaces for medical implants (such as drug delivery devices or ocular implants) to prevent cell-migration induced biofouling of the implant surface. While we have demonstrated the efficacy of the device in impeding cell migration in-vitro, biofouling in-vivo is a more complicated and complex process. Therefore, in-vivo experiments are required to determine the usability of actuated micropillar surfaces in applications pertaining to the in-vivo prevention of cell-migration induced biofouling of implants.    115  Bibliography  [1] J. Y. Lim and H. J. Donahue, "Cell sensing and response to micro-and nanostructured surfaces produced by chemical and topographic patterning," Tissue Eng., vol. 13, pp. 1879-1891, 2007.  [2] W. Schlote, "Responses of vessel walls to chronically applied electrical stimuli," Basic Res. Cardiol., vol. 74, pp. 10-20, 1979.  [3] M. Zhao, B. Song, J. Pu, T. Wada, B. Reid, G. Tai, F. Wang, A. Guo, P. Walczysko and Y. Gu, "Electrical signals control wound healing through phosphatidylinositol-3-OH kinase-γ and PTEN," Nature, vol. 442, pp. 457-460, 2006.  [4] W. E. Brownell, F. Qian and B. Anvari, "Cell membrane tethers generate mechanical force in response to electrical stimulation," Biophys. J., vol. 99, pp. 845-852, 2010.  [5] J. Sadoshima and S. Izumo, "The cellular and molecular response of cardiac myocytes to mechanical stress," Annu. Rev. Physiol., vol. 59, pp. 551-571, 1997.  [6] P. Ehrlich and L. Lanyon, "Mechanical strain and bone cell function: a review," Osteoporosis Int., vol. 13, pp. 688-700, 2002.  [7] S. Jungbauer, H. Gao, J. P. Spatz and R. Kemkemer, "Two characteristic regimes in frequency-dependent dynamic reorientation of fibroblasts on cyclically stretched substrates," Biophys. J., vol. 95, pp. 3470-3478, 2008.  116  [8] C. Neidlinger‐Wilke, H. J. Wilke and L. Claes, "Cyclic stretching of human osteoblasts affects proliferation and metabolism: a new experimental method and its application," Journal of Orthopaedic Research, vol. 12, pp. 70-78, 1994.  [9] E. A. Sprague, J. Luo and J. C. Palmaz, "Human aortic endothelial cell migration onto stent surfaces under static and flow conditions," J. Vasc. Interv. Radiol., vol. 8, pp. 83-92, 1997.  [10] B. J. Ballermann, A. Dardik, E. Eng and A. Liu, "Shear stress and the endothelium," Kidney Int., vol. 54, pp. S100-S108, 1998.  [11] H. H. Vandenburgh, "A computerized mechanical cell stimulator for tissue culture: effects on skeletal muscle organogenesis," In Vitro Cellular & Developmental Biology, vol. 24, pp. 609-619, 1988.  [12] H. H. Vandenburgh and P. Karlisch, "Longitudinal growth of skeletal myotubes in vitro in a new horizontal mechanical cell stimulator," In Vitro Cellular & Developmental Biology-Plant, vol. 25, pp. 607-616, 1989.  [13] T. Iba and B. E. Sumpio, "Morphological response of human endothelial cells subjected to cyclic strain in vitro," Microvasc. Res., vol. 42, pp. 245-254, 1991.  [14] Y. C. Yung, H. Vandenburgh and D. J. Mooney, "Cellular strain assessment tool (CSAT): Precision-controlled cyclic uniaxial tensile loading," J. Biomech., vol. 42, pp. 178-182, 2009.  117  [15] W. J. Richardson, R. P. Metz, M. R. Moreno, E. Wilson and J. E. Moore, "A device to study the effects of stretch gradients on cell behavior," J. Biomech. Eng., vol. 133, pp. 101008, 2011.  [16] G. F. Christopher, J. M. Yoo, N. Dagalakis, S. D. Hudson and K. B. Migler, "Development of a MEMS based dynamic rheometer," Lab on a Chip, vol. 10, pp. 2749-2757, 2010.  [17] D. Desmaële, M. Boukallel and S. Régnier, "Actuation means for the mechanical stimulation of living cells via microelectromechanical systems: A critical review," J. Biomech., vol. 44, pp. 1433-1446, 2011.  [18] J. Guck, R. Ananthakrishnan, C. C. Cunningham and J. Käs, "Stretching biological cells with light," Journal of Physics: Condensed Matter, vol. 14, pp. 4843, 2002.  [19] C. Moraes, C. A. Simmons and Y. Sun, "Cell Mechanics Meets MEMS," CSME Bulletin SCGM, pp. 15-18, Fall 2006.  [20] C. Moraes, J. Chen, Y. Sun and C. A. Simmons, "Microfabricated arrays for high-throughput screening of cellular response to cyclic substrate deformation," Lab on a Chip, vol. 10, pp. 227-234, 2010.  [21] S. Akbari, M. Niklaus and H. Shea, "Arrays of EAP micro-actuators for single-cell stretching applications," in SPIE Smart Structures and Materials Nondestructive Evaluation and Health Monitoring, 2010, pp. 76420H-76420H-10.  118  [22] N. J. Sniadecki, A. Anguelouch, M. T. Yang, C. M. Lamb, Z. Liu, S. B. Kirschner, Y. Liu, D. H. Reich and C. S. Chen, "Magnetic microposts as an approach to apply forces to living cells," Proceedings of the National Academy of Sciences, vol. 104, pp. 14553, 2007.  [23] B. Evans, A. Shields, R. L. Carroll, S. Washburn, M. Falvo and R. Superfine, "Magnetically actuated nanorod arrays as biomimetic cilia," Nano Letters, vol. 7, pp. 1428-1434, 2007.  [24] N. J. Sniadecki, C. M. Lamb, Y. Liu, C. S. Chen and D. H. Reich, "Magnetic microposts for mechanical stimulation of biological cells: Fabrication, characterization, and analysis," Rev. Sci. Instrum., vol. 79, pp. 044302, 2008.  [25] F. Fahrni, M. W. J. Prins and L. J. van IJzendoorn, "Micro-fluidic actuation using magnetic artificial cilia," Lab Chip, vol. 9, pp. 3413-3421, 2009.  [26] J. le Digabel, N. Biais, J. Fresnais, J. F. Berret, P. Hersen and B. Ladoux, "Magnetic micropillars as a tool to govern substrate deformations," Lab Chip, 2011.  [27] M. R. Mofrad and R. D. Kamm, Cellular Mechanotransduction: Diverse Perspectives from Molecules to Tissues. Cambridge University Press, 2009,  [28] N. Wang, J. P. Butler and D. E. Ingber, "Mechanotransduction across the cell surface and through the cytoskeleton," Science, vol. 260, pp. 1124-1127, 1993.  [29] M. D. Schaller and J. Thomas Parsons, "Focal adhesion kinase and associated proteins," Curr. Opin. Cell Biol., vol. 6, pp. 705-710, 1994.  119  [30] D. E. Ingber, "Mechanical signaling and the cellular response to extracellular matrix in angiogenesis and cardiovascular physiology," Circ. Res., vol. 91, pp. 877-887, 2002.  [31] Y. C. Yung, J. Chae, M. J. Buehler, C. P. Hunter and D. J. Mooney, "Cyclic tensile strain triggers a sequence of autocrine and paracrine signaling to regulate angiogenic sprouting in human vascular cells," Proceedings of the National Academy of Sciences, vol. 106, pp. 15279-15284, 2009.  [32] L. Lamalice, F. Le Boeuf and J. Huot, "Endothelial cell migration during angiogenesis," Circ. Res., vol. 100, pp. 782-794, Mar 30. 2007.  [33] P. Martin, "Wound healing--aiming for perfect skin regeneration," Science, vol. 276, pp. 75-81, Apr 4. 1997.  [34] R. J. Petrie, A. D. Doyle and K. M. Yamada, "Random versus directionally persistent cell migration," Nature Reviews Molecular Cell Biology, vol. 10, pp. 538-549, 2009.  [35] M. P. Sheetz, D. P. Felsenfeld and C. G. Galbraith, "Cell migration: regulation of force on extracellular-matrix-integrin complexes," Trends Cell Biol., vol. 8, pp. 51-54, 1998.  [36] X. Trepat, M. R. Wasserman, T. E. Angelini, E. Millet, D. A. Weitz, J. P. Butler and J. J. Fredberg, "Physical forces during collective cell migration," Nature Physics, vol. 5, pp. 426-430, 2009.  [37] I. Rupeš, "Checking cell size in yeast," Trends in Genetics, vol. 18, pp. 479-485, 2002.  120  [38] J. A. Madri, B. M. Pratt and J. Yannariello-Brown, "Matrix-driven cell size change modulates aortic endothelial cell proliferation and sheet migration," Am. J. Pathol., vol. 132, pp. 18-27, Jul. 1988.  [39] P. Jorgensen and M. Tyers, "How cells coordinate growth and division," Current Biology, vol. 14, pp. R1014-R1027, 2004.  [40] J. Lesman, D. Notbohm, A. Tirrell and G. Ravichandran, "Contractile forces regulate cell division in three-dimensional environments," J. Cell Biol., vol. 205, pp. 155-162, Apr 28. 2014.  [41] J. M. Scholey, I. Brust-Mascher and A. Mogilner, "Cell division," Nature, vol. 422, pp. 746-752, 2003.  [42] J. Wolff, P. Maquet and R. Furlong, The Law of Bone Remodelling. Springer-Verlag Berlin, 1986,  [43] P. Dobrin, "Mechanical properties of arteries," Physiol. Rev., vol. 58, pp. 397-460, 1978.  [44] R. J. Price and T. C. Skalak, "Circumferential wall stress as a mechanism for arteriolar rarefaction and proliferation in a network model," Microvasc. Res., vol. 47, pp. 188-202, 1994.  [45] J. Fink, N. Carpi, T. Betz, A. Bétard, M. Chebah, A. Azioune, M. Bornens, C. Sykes, L. Fetler and D. Cuvelier, "External forces control mitotic spindle positioning," Nat. Cell Biol., vol. 13, pp. 771-778, 2011.  121  [46] E. W. Young and C. A. Simmons, "Macro-and microscale fluid flow systems for endothelial cell biology," Lab on a Chip, vol. 10, pp. 143-160, 2010.  [47] K. C. Neuman and A. Nagy, "Single-molecule force spectroscopy: optical tweezers, magnetic tweezers and atomic force microscopy," Nat. Methods, vol. 5, pp. 491-505, Jun. 2008.  [48] I. De Vlaminck and C. Dekker, "Recent advances in magnetic tweezers," Annual Review of Biophysics, vol. 41, pp. 453-472, 2012.  [49] L. MacQueen, M. Thibault, M. Buschmann and M. Wertheimer, "Electro-deformation of individual mammalian cells in suspension," in Solid Dielectrics (ICSD), 2010 10th IEEE International Conference on, 2010, pp. 1-4.  [50] E. Tkachenko, E. Gutierrez, M. H. Ginsberg and A. Groisman, "An easy to assemble microfluidic perfusion device with a magnetic clamp," Lab on a Chip, vol. 9, pp. 1085-1095, 2009.  [51] Z. Tang, Y. Akiyama, K. Itoga, J. Kobayashi, M. Yamato and T. Okano, "Shear stress-dependent cell detachment from temperature-responsive cell culture surfaces in a microfluidic device," Biomaterials, vol. 33, pp. 7405-7411, 2012.  [52] Y. Kamotani, T. Bersano-Begey, N. Kato, Y. Tung, D. Huh, J. W. Song and S. Takayama, "Individually programmable cell stretching microwell arrays actuated by a Braille display," Biomaterials, vol. 29, pp. 2646-2655, 2008.  122  [53] K. Shimizu, A. Shunori, K. Morimoto, M. Hashida and S. Konishi, "Development of a biochip with serially connected pneumatic balloons for cell-stretching culture," Sensors Actuators B: Chem., vol. 156, pp. 486-493, 2011.  [54] P. Tseng, J. W. Judy and D. Di Carlo, "Magnetic nanoparticle-mediated massively parallel mechanical modulation of single-cell behavior," Nature Methods, vol. 9, pp. 1113-1119, 2012.  [55] M. S. Kim, C. Y. Bae, G. Wee, Y. Han and J. Park, "A microfluidic in vitro cultivation system for mechanical stimulation of bovine embryos," Electrophoresis, vol. 30, pp. 3276-3282, 2009.  [56] N. Tymchenko, J. Wallentin, S. Petronis, L. Bjursten, B. Kasemo and J. Gold, "A novel cell force sensor for quantification of traction during cell spreading and contact guidance," Biophys. J., vol. 93, pp. 335-345, 2007.  [57] J. Fu, Y. K. Wang, M. T. Yang, R. A. Desai, X. Yu, Z. Liu and C. S. Chen, "Mechanical regulation of cell function with geometrically modulated elastomeric substrates," Nature Methods, vol. 7, pp. 733-736, 2010.  [58] F. Zhang, S. Anderson, X. Zheng, E. Roberts, Y. Qiu, R. Liao and X. Zhang, "Cell force mapping using a double-sided micropillar array based on the moiré fringe method," Appl. Phys. Lett., vol. 105, pp. 033702, 2014.  123  [59] K. S. Bielawski and N. J. Sniadecki, "Microposts with embedded nanowires to control substrate stiffness," in ASME 2012 Summer Bioengineering Conference, 2012, pp. 1165-1166.  [60] Z. Pan, C. Yan, R. Peng, Y. Zhao, Y. He and J. Ding, "Control of cell nucleus shapes via micropillar patterns," Biomaterials, vol. 33, pp. 1730-1735, 2012.  [61] L. E. Dickinson, D. R. Rand, J. Tsao, W. Eberle and S. Gerecht, "Endothelial cell responses to micropillar substrates of varying dimensions and stiffness," Journal of Biomedical Materials Research Part A, vol. 100, pp. 1457-1466, 2012.  [62] A. Saez, E. Anon, M. Ghibaudo, O. Du Roure, J. Di Meglio, P. Hersen, P. Silberzan, A. Buguin and B. Ladoux, "Traction forces exerted by epithelial cell sheets," Journal of Physics: Condensed Matter, vol. 22, pp. 194119, 2010.  [63] S. Ghassemi, G. Meacci, S. Liu, A. A. Gondarenko, A. Mathur, P. Roca-Cusachs, M. P. Sheetz and J. Hone, "Cells test substrate rigidity by local contractions on submicrometer pillars," Proc. Natl. Acad. Sci. U. S. A., vol. 109, pp. 5328-5333, Apr 3. 2012.  [64] Z. Liu, J. L. Tan, D. M. Cohen, M. T. Yang, N. J. Sniadecki, S. A. Ruiz, C. M. Nelson and C. S. Chen, "Mechanical tugging force regulates the size of cell–cell junctions," Proceedings of the National Academy of Sciences, vol. 107, pp. 9944, 2010.  [65] S. J. Han and N. J. Sniadecki, "Simulations of the contractile cycle in cell migration using a bio-chemical–mechanical model," Comput. Methods Biomech. Biomed. Engin., vol. 14, pp. 459-468, 2011.  124  [66] M. T. Yang, N. J. Sniadecki and C. S. Chen, "Geometric Considerations of Micro‐to Nanoscale Elastomeric Post Arrays to Study Cellular Traction Forces," Adv Mater, vol. 19, pp. 3119-3123, 2007.  [67] A. Buxboim and D. E. Discher, "Stem cells feel the difference," Nature Methods, vol. 7, pp. 695, 2010.  [68] J. L. Tan, J. Tien, D. M. Pirone, D. S. Gray, K. Bhadriraju and C. S. Chen, "Cells lying on a bed of microneedles: an approach to isolate mechanical force," Proc. Natl. Acad. Sci. U. S. A., vol. 100, pp. 1484, 2003.  [69] Y. Zhang, C. W. Lo, J. A. Taylor and S. Yang, "Replica molding of high-aspect-ratio polymeric nanopillar arrays with high fidelity," Langmuir, vol. 22, pp. 8595-8601, 2006.  [70] M. Ghibaudo, J. Di Meglio, P. Hersen and B. Ladoux, "Mechanics of cell spreading within 3D-micropatterned environments," Lab on a Chip, vol. 11, pp. 805-812, 2011.  [71] F. N. Pirmoradi, L. Cheng and M. Chiao, "A magnetic poly (dimethylesiloxane) composite membrane incorporated with uniformly dispersed, coated iron oxide nanoparticles," J Micromech Microengineering, vol. 20, pp. 015032, 2010.  [72] J. H. C. Wang, P. Goldschmidt-Clermont, J. Wille and F. C. P. Yin, "Specificity of endothelial cell reorientation in response to cyclic mechanical stretching," J. Biomech., vol. 34, pp. 1563-1572, 2001.  [73] J. Ram and G. S. Brar, "Posterior Capsule Opacification," Small Incision Cataract Surgery: Manual Phaco, pp. 255, 2010.  125  [74] S. Saika, L. Werner and F. J. Lovicu, "Lens Epithelium and Posterior Capsular Opacification," 2014.  [75] E. J. Hollick, D. J. Spalton, P. G. Ursell and M. V. Pande, "Lens epithelial cell regression on the posterior capsule with different intraocular lens materials," Br. J. Ophthalmol., vol. 82, pp. 1182-1188, Oct. 1998.  [76] Q. Peng, N. Visessook, D. J. Apple, S. K. Pandey, L. Werner, M. Escobar-Gomez, R. Schoderbek, K. D. Solomon and A. Guindi, "Surgical prevention of posterior capsule opacification: Part 3: Intraocular lens optic barrier effect as a second line of defense," Journal of Cataract & Refractive Surgery, vol. 26, pp. 198-213, 2000.  [77] O. Nishi, N. Yamamoto, K. Nishi and Y. Nishi, "Contact inhibition of migrating lens epithelial cells at the capsular bend created by a sharp-edged intraocular lens after cataract surgery," Journal of Cataract & Refractive Surgery, vol. 33, pp. 1065-1070, 2007.  [78] C. Morris, L. Werner and M. Tetz, "PCO prevention: IOL material versus IOL design," in Lens Epithelium and Posterior Capsular Opacification Anonymous Springer, 2014, pp. 297-312.  [79] S. Guo and L. A. Dipietro, "Factors affecting wound healing," J. Dent. Res., vol. 89, pp. 219-229, Mar. 2010.  [80] G. C. Gurtner, S. Werner, Y. Barrandon and M. T. Longaker, "Wound repair and regeneration," Nature, vol. 453, pp. 314-321, 2008.  126  [81] J. M. Reinke and H. Sorg, "Wound repair and regeneration," Eur. Surg. Res., vol. 49, pp. 35-43, 2012.  [82] F. Winston, E. Macarak, S. Gorfien and L. Thibault, "A system to reproduce and quantify the biomechanical environment of the cell," J. Appl. Physiol., vol. 67, pp. 397-405, 1989.  [83] M. Moretti, A. Prina-Mello, A. Reid, V. Barron and P. Prendergast, "Endothelial cell alignment on cyclically-stretched silicone surfaces," J. Mater. Sci. Mater. Med., vol. 15, pp. 1159-1164, 2004.  [84] B. Liu, M. J. Qu, K. R. Qin, H. Li, Z. K. Li, B. R. Shen and Z. L. Jiang, "Role of cyclic strain frequency in regulating the alignment of vascular smooth muscle cells in vitro," Biophys. J., vol. 94, pp. 1497-1507, 2008.  [85] F. P. Beer, "Transformation of plane strain," in Mechanics of Materials ,5th ed.Anonymous McGraw-Hill, 2009, pp. 470-475.  [86] P. Dartsch and E. Betz, "Response of cultured endothelial cells to mechanical stimulation," Basic Res. Cardiol., vol. 84, pp. 268-281, 1989.  [87] J. H.C. Wang, "Substrate deformation determines actin cytoskeleton reorganization: a mathematical modeling and experimental study," J. Theor. Biol., vol. 202, pp. 33-41, 2000.  [88] K. Kanda and T. Matsuda, "Mechanical stress-induced orientation and ultrastructural change of smooth muscle cells cultured in three-dimensional collagen lattices." Cell Transplant., vol. 3, pp. 481, 1994.  127  [89] M. Masuda and K. Fujiwara, "Morphological responses of single endothelial cells exposed to physiological levels of fluid shear stress," Front. Med. Biol. Eng., vol. 5, pp. 79-87, 1993.  [90] M. Malek and S. Izumo, "Mechanism of endothelial cell shape change and cytoskeletal remodeling in response to fluid shear stress," J. Cell. Sci., vol. 109, pp. 713-726, 1996.  [91] Y. Qiang, J. Antony, A. Sharma, J. Nutting, D. Sikes and D. Meyer, "Iron/iron oxide core-shell nanoclusters for biomedical applications," Journal of Nanoparticle Research, vol. 8, pp. 489-496, 2006.  [92] H. Cao, G. Huang, S. Xuan, Q. Wu, F. Gu and C. Li, "Synthesis and characterization of carbon-coated iron core/shell nanostructures," J. Alloys Compounds, vol. 448, pp. 272-276, 2008.  [93] C. Moraes, Y. Sun and C. A. Simmons, "Solving the shrinkage-induced PDMS alignment registration issue in multilayer soft lithography," J Micromech Microengineering, vol. 19, pp. 065015, 2009.  [94] H. Wu, T. W. Odom, D. T. Chiu and G. M. Whitesides, "Fabrication of complex three-dimensional microchannel systems in PDMS," J. Am. Chem. Soc., vol. 125, pp. 554-559, 2003.  [95] U. B. Giang, D. Lee, M. R. King and L. A. DeLouise, "Microfabrication of cavities in polydimethylsiloxane using DRIE silicon molds," Lab on a Chip, vol. 7, pp. 1660-1662, 2007.  128  [96] D. T. Eddington, W. C. Crone and D. J. Beebe, "Development of process protocols to fine tune polydimethylsiloxane material properties," in 7th International Conference on Miniaturized Chemical and Biochemical Analysis Systems, Squaw Valley, California, USA, 2003, pp. 1089-1092.  [97] J. W. Judy and R. S. Muller, "Magnetic microactuation of torsional polysilicon structures," Sensors and Actuators A: Physical, vol. 53, pp. 392-397, 1996.  [98] J. W. Judy and R. S. Muller, "Magnetically actuated, addressable microstructures," Journal of Microelectromechanical Systems, vol. 6, pp. 249-256, 2002.  [99] J. W. Judy, R. S. Muller and H. H. Zappe, "Magnetic microactuation of polysilicon flexure structures," Journal of Microelectromechanical Systems, vol. 4, pp. 162-169, 2002.  [100] J. Kim, S. E. Chung, S. E. Choi, H. Lee, J. Kim and S. Kwon, "Programming magnetic anisotropy in polymeric microactuators," Nature Materials, 2011.  [101] K. S. Suslick, M. Fang and T. Hyeon, "Sonochemical synthesis of iron colloids," J. Am. Chem. Soc., vol. 118, pp. 11960-11961, 1996.  [102] Z. Guo, L. L. Henry, V. Palshin and E. J. Podlaha, "Synthesis of poly (methyl methacrylate) stabilized colloidal zero-valence metallic nanoparticles," J.Mater.Chem., vol. 16, pp. 1772-1777, 2006.  [103] S. P. Palecek, J. C. Loftus, M. H. Ginsberg, D. A. Lauffenburger and A. F. Horwitz, "Integrin-ligand binding properties govern cell migration speed through cell-substratum adhesiveness," Nature, vol. 385, pp. 537-540, Feb 6. 1997.  129  [104] P. Vitorino and T. Meyer, "Modular control of endothelial sheet migration," Genes Dev., vol. 22, pp. 3268-3281, Dec 1. 2008.  [105] E. Fong, S. Tzlil and D. A. Tirrell, "Boundary crossing in epithelial wound healing," Proc. Natl. Acad. Sci. U. S. A., vol. 107, pp. 19302-19307, Nov 9. 2010.  [106] P. J. Sammak, L. E. Hinman, P. O. Tran, M. D. Sjaastad and T. E. Machen, "How do injured cells communicate with the surviving cell monolayer?" J. Cell. Sci., vol. 110 ( Pt 4), pp. 465-475, Feb. 1997.  [107] M. Poujade, E. Grasland-Mongrain, A. Hertzog, J. Jouanneau, P. Chavrier, B. Ladoux, A. Buguin and P. Silberzan, "Collective migration of an epithelial monolayer in response to a model wound," Proc. Natl. Acad. Sci. U. S. A., vol. 104, pp. 15988-15993, Oct 9. 2007.  [108] E. R. Block, A. R. Matela, N. SundarRaj, E. R. Iszkula and J. K. Klarlund, "Wounding induces motility in sheets of corneal epithelial cells through loss of spatial constraints: role of heparin-binding epidermal growth factor-like growth factor signaling," J. Biol. Chem., vol. 279, pp. 24307-24312, Jun 4. 2004.  [109] C. W. Wolgemuth, "Lamellipodial contractions during crawling and spreading," Biophys. J., vol. 89, pp. 1643-1649, 2005.  [110] M. A. Schwartz and A. R. Horwitz, "Integrating adhesion, protrusion, and contraction during cell migration," Cell, vol. 125, pp. 1223-1225, 2006.  130  [111] N. D. Gallant, K. E. Michael and A. J. Garcia, "Cell adhesion strengthening: contributions of adhesive area, integrin binding, and focal adhesion assembly," Mol. Biol. Cell, vol. 16, pp. 4329-4340, Sep. 2005.  [112] P. Dartsch, H. Hämmerle and E. Betz, "Orientation of cultured arterial smooth muscle cells growing on cyclically stretched substrates," Cells Tissues Organs (Print), vol. 125, pp. 108-113, 1986.  [113] M. Bélanger and Y. Marois, "Hemocompatibility, biocompatibility, inflammatory and in vivo studies of primary reference materials low‐density polyethylene and polydimethylsiloxane: A review," J. Biomed. Mater. Res., vol. 58, pp. 467-477, 2001.  [114] A. Dalu, B. S. Blaydes, L. G. Lomax and K. B. Delclos, "A comparison of the inflammatory response to a polydimethylsiloxane implant in male and female Balb/c mice," Biomaterials, vol. 21, pp. 1947-1957, 2000.  [115] R. Rezakhaniha, A. Agianniotis, J. T. C. Schrauwen, A. Griffa, D. Sage, C. V. C. Bouten, F. van de Vosse, M. Unser and N. Stergiopulos, "Experimental investigation of collagen waviness and orientation in the arterial adventitia using confocal laser scanning microscopy," Biomechanics and Modeling in Mechanobiology, pp. 1-13, 2012.  [116] E. Fonck, G. G. Feigl, J. Fasel, D. Sage, M. Unser, D. A. Rüfenacht and N. Stergiopulos, "Effect of aging on elastin functionality in human cerebral arteries," Stroke, vol. 40, pp. 2552-2556, 2009.  131  [117] I. Schoen, W. Hu, E. Klotzsch and V. Vogel, "Probing cellular traction forces by micropillar arrays: contribution of substrate warping to pillar deflection," Nano Letters, vol. 10, pp. 1823-1830, 2010.  [118] F. P. Beer, R. E. Johnston and J. T. DeWolf, Mechanics of Materials. ,5th ed.New York: McGraw-Hill, 2005,  [119] F. Khademolhosseini and M. Chiao, "Fabrication and Patterning of Magnetic Polymer Micropillar Structures Using a Dry-Nanoparticle Embedding Technique," Microelectromechanical Systems, Journal of, vol. 22, pp. 131-139, 2013.   132  Appendices  Appendix A  : How to Culture Cells on Plateau Region of MACSAT Membrane To confine cell culture to plateau region of the MACSAT membrane, a circular PDMS mask with a radius corresponding to that of the MACSAT membrane was prepared. Using a mechanical punch, through holes were punched in the PDMS mask at a radial distance corresponding to the plateau region of the MACSAT membrane. The PDMS mask was then laid on top of the MACAST membrane and the cells were cultured in the through holes. After 24 hours, the PDMS mask was removed, leaving confluent groups of cultured cells in the plateau region of the MACSAT membrane.              133  Appendix B  : OrientationJ Plugin for Angular Orientation Analysis  ImageJ and OrientationJ were used to analyze the raw images obtained from experiments. OrientionJ (http://bigwww.epfl.ch/demo/orientation) uses structure tensors and directional derivatives to analyze an image and characterize the orientation and isotropy properties of the region of interest [115, 116]. To do this, a Gaussian weighting function is used to specify the region of interest and obtain a structure tensor. The direction of the largest eigenvector of the structure tensor is taken as the predominant orientation, and the ratio between the sum and the difference of the maximum and minimum eigenvalues of the structure tensor is taken as the coherency of the region of interest. Coherency is bound between 0 and 1 with 1 indicating highly oriented and 0 indicating isotropic structures. Figure B1 shows the OrientationJ dialogue box. 134   Figure B1: OrientationJ dialogue box allows the user to choose the size of the Gaussian window used for the analysis, the method for calculating gradients in the image and options for color coding the output results. The size of the Gaussian window depends on the size of the features being analyzed, i.e., features with finer resolutions require smaller Gaussian windows. Through a trial and error approach we found a Gaussian window of 5 pixels gave the best representation of cell and actin filament orientation for images obtained from our experiments.  135  The time-lapse images acquired during our experiment were processed with OrientationJ to obtain the time-dependent statistical distribution of cell orientation and actin orientation angles. To find the cell orientation angles a 5 pixel Gaussian window and the Gaussian gradient options provided by OrientationJ were used. Using these parameters, OrientationJ was able to precisely detect the edges of the cell membranes and the general direction of cell elongation. To find actin orientation angles, a 5 pixel Gaussian window and the cubic spline gradient options were used as they provided more precise actin filament detection. With these settings OrientationJ was able to precisely detect the directionality of actin filament bundles in the cells. For the images obtained from experiments, the coherency parameter was also calculated and used to find the order of cell/actin alignment. Statistical analysis of the data obtained from ImageJ for the cell and actin orientation angles and coherency of cell and actin alignment was performed using MATLAB software. Student t-test analysis was performed in Microsoft Excel to assess the significance of the results. It is important to note in the experiments demonstrated in Chapter 2, a proportion of the actin cytoskeleton in the cells studied existed as a fine meshwork of filaments that were not resolvable in the microscope images obtained in the experiments. The minimal size of the oriented structures detected by OrientationJ was therefore limited by the resolution (the pixel density) of the obtained images and the size of the Gaussian window used for image analysis. The minimal size of the oriented structures detected can be improved either by increasing the pixel density (i.e., using a higher magnification objective) or by reducing the size of the Gaussian window used for image analysis. In our analysis, we were able to detect and determine the orientation of actin bundles (stress fibers) with diameters as small as 300 nm.  Appendi Figure C1mm verticradial direare at ±22Based on2.8 a valu x C  : Axial: Magnitude al displacemction (±90º). º and the angu actin reoriee of 1% axi Strain Throf axial strainent of the meCircumferentlar region ofntation resual strain is e eshold for  vs. angular mbrane cential direction isub 1% axiallts obtained nough to triTriggering orientation iner corresponds at 0 º. The d strain (|ε|<1) in section 2gger actin reActin Reor the plateau ring to a mairections of mis -32º<θ<+32.2.4 and the orientationientation egion of the ximum 6% tinimum axiaº. results pres.  MACSAT forensile strain l deformationented in Fig136  a 3.5 in the  (ε=0) ure  AppendiSteps use Figure D1polymer cx D  : Maskd in the mas: Fabrication asting step. MagThin PDMing Techniking technisteps for patt. Field S Layer que for Conque for patte  erning of magA tandthroTheplaDrymamaintoExcTheexcPDcurOnfinapillon trolled Patrning of minetic micropihin PDMS l cured. Laseugh holes i PDMS maced on top o magnetic psk. A magnegnetic partic the mold feess particle PDMS maess particlesMS polymee ce cured, thel device conars placed inthe mask paterning of Mcropillar strullars among nayer is spin-r ablation isn the layer ask is alignedf the mold.articles are t/magnetic les through atures. s are removsk is remove are wiped r is cast onto micropillarsists of magtermittentlyttern.  icropillarctures are don-magnetic coated on a used to crend obtain a  with the madded to thefield is usedthe holes ofed by wipind and any roff.  the mold a device is dnetic and n among eac Structuresemonstratedmicropillars  glass waferate a patternPDMS masicro- mold a top side of  to direct the the mask ang off. emaining nd allowed e-molded. Tonmagnetich other base137  . in one   of k. nd the  d to he  d 138  Appendix E  : Theoretical Modeling and Experimental Characterization of the Base Tilting Effect of FeC-PDMS Micropillar Structures I) Theoretical Model A theoretical model for the total deformation of a polymer pillar structure under the effect of a concentrated tip force was recently presented [117], where the contributions of base tilting, base dislocation, pillar bending and pillar shear were taken into account. Here, we look at a different loading scenario where only a bending moment acts on the pillar. Such a loading scenario occurs when a homogenous magnetic pillar is placed inside a uniform magnetic field (Figure E1).  Figure E1: Schematic of magnetic forces and moments acting on a magnetic pillar in the presence of a magnetic field B. For a uniform magnetic field (સ࡮ሬሬԦ ൌ ૙), and assuming a uniform magnetization ࢓ሬሬሬԦ of the homogenous magnetic pillar, the magnetic translational force is zero (ࡲሬԦ ൌ સሺ࢓ሬሬሬԦ. ࡮ሬԦሻ ൌ ૙.  Due to the absence of a magnetic field gradient (ܤ׏ሬԦ ൌ 0) in a uniform magnetic field, and assuming a uniform magnetization ሬ݉ሬሬሬሬԦ of the magnetic pillar (׏ሬ݉ሬԦ ൌ 0), the magnetic translational forces acting on the pillar are zero (ܨሬሬሬԦ ൌ ׏ሺሬ݉ሬሬሬሬԦ.ܤሬሬሬሬԦሻ ൌ 0). However, as long as the magnetic field is not parallel to the direction of the magnetic pillar’s easy axis of magnetization [97]-[99], a magnetic bending moment will act upon the magnetic pillar [100]. In this pure bending scenario, 139  the total deformation consists of pillar bending and base tilting (Figure E2), i.e., pillar shear and base dislocation do not contribute to the total angular deformation or tip displacement of the pillar.   Figure E2: The total deformation of a polymer pillar structure (elastic pillar on elastic base) under the action of a bending moment consists of pillar bending and base tilting. In Fig. 2 δTot is the total tip displacement which is the sum of the displacements due to pillar bending (δbend) and base tilting (δtilt). Similarly, β is the total rotation angle of the pillar tip, and is the sum of the rotations due to pillar bending (α) and base tilting (θ). Assuming small deformations, the pillar bending angle α and the tip displacement due to pillar bending δbend can be calculated using the following equations [118]   ߙ ൌ ܧܮܯ௣ܫ௣ 												 		ߜ௕௘௡ௗ ൌܮܯଶ2ܧ௣ܫ௣ ,  (1)where M is the bending moment acting on the pillar and L, Ep and Ip are the length, the Young’s modulus and the bending moment of inertia of the pillar, respectively. Schoen et al. postulated the following expressions for the base tilting angle θ [117]: 140   ߠ ൌ T௧௜௟௧ ൬σ௠௔௫ܧ௕ ൰	,  (2)where Eb is the Young’s modulus of the base, σmax is maximum stress at the pillar base and Ttilt is the tilting coefficient of the pillar base. Based on the above definition of the tilting angle, a micropillar structure will tilt more at the base if Ttilt is larger, even if the maximum normal strain (σ௠௔௫/ܧ௕) is the same at the base. The maximum stress can be calculated from [118]  σ௠௔௫ ൌ 2ܦܯI௣ 	,  (3)where D is the pillar diameter. Combining equations (2) and (3) we have  ߠ ൌ T௧௜௟௧ ቆ 2ܦܯE௕I௣ቇ	.  (4)Using the mathematical theory of elasticity, Schoen et al. derived the following relation for the tilting coefficient as a function of the Poisson’s ratio ϑb of the base material [117]  T௧௜௟௧ ൌ a 1 ൅ ߴ௕2ߨ ൜2ሺ1 െ ߴ௕ሻ ൅ 1 െ14ሺ1 െ ߴ௕ሻൠ ,  (5)where the constant ‘a’ is a free fitting parameter. Based on finite element simulation results for the deformation profile and average tilt angle of the pillar base, Schoen et al. proposed a value of a=1.3 and Ttilt=0.465 for PDMS substrates [117].  Using equation (4) and assuming that the tilting angle is small, based on geometric relations, the pillar tip displacement due to base tilting can be calculated as 141   ߜ௧௜௟௧ ൌ ܮ. ߠ݊ܽݐ ≅ ܮ. ߠ ൌ T௧௜௟௧ ቆ2ܮܦܯE௕I௣ቇ .  (6)Therefore the total tip displacement and the total angular rotation of the pillar tip are  ߠ ൌ T௧௜௟௧ ൬σ௠௔௫ܧ௕ ൰	,  (7) ߚ ൌ ߙ ൅ ߠ ൌ ܧܮܯ௣ܫ௣ ൅ T௧௜௟௧ ቆܦܯ2E௕I௣ ቇ . (8)From equation (8) we find that for an elastic pillar on a stiff base, Eb→∞, θ→0 and β=α, therefore all the deformation is due to pillar bending. In contrast, for a stiff pillar on an elastic base, Ep→∞, α→0 and β=θ, and all the deformation is due to base tilting. From equations (1) and (2) we can find the ratio of the tilting angle to the bending angle  ߠߙ ൌ ቆT௧௜௟௧.ܦܯ2E௕I௣ቇ ቆܮܯܧ௣ܫ௣ቇ൘ 	ൌT௧௜௟௧2 ൬ܧ௣ܧ௕൰ ൬ܦܮ൰ .  (9)It is seen that the contributions of the pillar bending angle α and the base tilting angle θ to the total angular deformation β of the pillar tip are a function of the pillar aspect ratio (L/D)  ߙߚ ൌܮ ܦൗܥ ൅ ܮ ܦൗ  (10-a) ߠߚ ൌܥܥ ൅ ܮ ܦൗ	,	142  where ܥ ൌ ୘೟೔೗೟ଶ ቀா೛ா್ቁ is a function of the material properties of the pillar and the base and is independent of the pillar size. Equation (9) can be re-organized to write  T௧௜௟௧ ൌ 2ቆܧ௕ܧ௣ቇ ൬ܮܦ൰ ൬ߠߙ൰	.  (11)In contrast to equation (5) where the warping profile at the pillar base is needed to estimate the parameter ‘a’ and calculate Ttilt, equation (11) can be used to determine the tilting coefficient based on the geometric dimensions, the material properties and the bending and tilting angles obtained from experiments. II) Materials and Methods A) Materials: PDMS (Sylgard 184 Silicone Elastomer, Dow Corning Corporation) was used as the polymeric matrix. Pure PDMS was prepared at the manufacturer’s recommended ratio of 10:1 pre-polymer to cross-linker. FeC core-shell micro-spheres (obtained from UBC pharmaceutical Sciences) with diameters ranging from 0.1 to 1 µm were used as the magnetic filler to fabricate the magnetic FeC-PDMS micropillars. Magnetic Iron bars (Scientific Alloys inc.) and 22 AWG copper wire (EIS-inc.) were used to make the custom-built electromagnet for magnetic field generation.   143  B) Fabrication of magnetic PDMS micropillars using the Dry Particle Embedding Technique (DPET) Conventional photolithography was used to make an SU8 master of the micropillars. A PDMS negative of the SU8 micropillars was then generated, silanized and used as a mold for the final structure. FeC particles were then embedded in the holes of the mold and a final PMDS casting and curing is performed. Once removed from the mold, the final structure consisted of magnetic FeC-PDMS micropillars on a pure PDMS substrate. The contrast between the black FeC-PDMS micropillars and the clear PDMS substrate provided a clear boundary at the base of the micropillars which was used to measure the base tilting angle as the micropillars deform. We fabricated micropillars with diameter ranging from 15-40 µm and heights ranging from 80-200 µm for our experiments. C) Characterizing the material composition and mechanical properties of the FeC-PDMS The material composition of the magnetic micropillars was determined through energy-dispersive X-ray spectroscopy (EDX) measurements with a Hitachi S3000N scanning electron microscope. Alternatively, a SQUID magneto-meter was also used to measure the magnetization properties and the saturation magnetization of the FeC-PDMS micro-pillars, pure FeC and pure PDMS. This data was consequently used to estimate the weight ratio of FeC to PDMS in the FeC-PDMS micropillars. The young’s modulus of pure PDMS, 20% w/w FeC- PDMS and 40% w/w FeC-PDMS were measured using a thermo mechanical analyzer (TA instruments Q400). To prepare the 20% and 40% w/w FeC-PDMS polymers, FeC particles were added to glass vials of Sylgard 184 pre-144  polymer based on the calculated weight ratios, and sonicated for 1 hour in a water bath to get a fully homogenous mixture. Sylgard 184 cross-linking agent was then added at the manufacturer recommended 10:1 by weight ratio and manually mixed for 10 minutes. Pure PDMS was also prepared as previously described. The 20% and 40% w/w FeC-PDMS mixtures and the pure PDMS were then poured on glass trays to a height of 0.6 mm and oven-cured at 70 ºC for 3 hours. After curing, rectangular strips 25 mm long, 1.2 mm wide and 0.6 mm thick were cut from the FeC-PDMS and pure PDMS samples and used for the strain measurements with the thermo mechanical analyzer. To obtain force vs. displacement data, the samples were preloaded at 0.001 N and loading was increased at a constant rate of 0.02 N/min to a maximum of 0.15 N at which point the loading was decreased back to 0 N at the same rate of 0.02 N/min. This procedure was repeated on three different samples of each of the materials. The resulting displacement and force data was used to calculate strain-stress curves for the samples. The slope of the stress-strain curves was then used to calculate the Young’s modulus of each of the samples. D) Generation and measurement of magnetic field We designed and build a custom electromagnet to generate the magnetic field required for our experiments. Figure E3 shows a schematic of the electromagnet design. The magnetic field of interest is generated in a 6 mm long air gap between the two poles of the electromagnet. A glass sheet acting as a stage connects the two poles of the electromagnet. The central region of the glass stage, measuring an area 2x2 mm2, was used as the test area for the measurements. A Hall Effect probe and Guassmeter (FW Bell, Model 6010) were used to find the magnitude of magnetic field generated at different locations across test area. The obtained magnetic field values and location data were then used to calculate the corresponding magnetic field gradients. 145   Figure E3: Schematic drawing of the electromagnet design. A glass stage sits in between the poles of the electromagnet. The test area measuring 2x2 mm2 is located at the center of the glass stage (coincident with the center of the air gap in the electromagnet) where the magnetic field is most uniform and the magnetic field gradient is at its lowest. E) Measurement of tilting and bending angles and calculation of the tilt coefficient To visualize the tilting of the micropillars, samples comprising of single FeC-PDMS micropillars of various geometric dimensions sitting on pure PDMS substrates were cut out and mounted on the test area of the glass stage such that the micropillar axis was perpendicular to the direction of the magnetic field. The whole setup comprising of the electromagnet and micropillar sample was then placed on a tilt stage and under a microscope (Signatone, PSM 1000) with a scope mounted digital camera (Basler A100 series) and the tilt stage was adjusted so that the micropillar axis was parallel to the focal plane of the microscope. An initial image of the micropillar was taken at zero magnetic field and used as the base image for consequent measurement of micropillar 146  deformation. The magnetic field was then increased by regular intervals up to a maximum of 250 mT and images of the deformed micropillar were taken at each interval. ImageJ was used to measure the bending and tilting angles of the micropillars from the obtained images (Figure E4). Equation (10) was then used to calculate the tilting coefficient based on the measured angles, the measured mechanical properties and the geometric dimensions of the micropillars.   Figure E4: Side view of a magnetic FeC-PDMS micropillar on a pure PDMS substrate. Application of the magnetic field causes pillar bending and base tilting. To determine the tilting angle, a line is drawn perpendicular to the base of the tilted pillar. The angle between this line and the original pillar axis denotes the tilting angle. The bending angle α, tilting angle θ, and total deformation angle β are shown here.  III) Results and Discussion EDX measurements showed a 35-43 percent w/w loading ratio of FeC magnetic particles in the FeC-PDMS micropillars [119]. A comparison of the saturation magnetization of the FeC-PDMS micropillars and pure FeC obtained from SQUID measurements pointed to an average 40% w/w 147  loading ratio of FeC to PDMS for the FeC-PDMS micropillars, consistent with material composition results obtained from EDX measurements. Figure E5(a) shows sample stress-strain curves for 40% FeC-PDMS and pure PDMS, obtained from the strain tests performed with the thermo mechanical analyzer. Figure E5(b) shows the young’s moduli of pure PDMS, 20% and 40% w/w FeC-PDMS obtained by finding the slopes of their respective stress-strain curves. We obtained a Young’s modulus of 1.13±0.02 MPa for pure PDMS, consistent with previously reported values [96], and Young’s moduli of 1.18±0.04 MPa and 1.29±0.04 for 20% w/w and 40% w/w FeC-PDMS, respectively. It is observed that adding FeC magnetic particles to pure PDMS results in the increase of the Young’s moduli. Furthermore, we observe that increasing the weight percentage of FeC particles in FeC-PDMS from 20% to 40% causes less than a 10% increase in the measured Young’s moduli of the material. That is, a 20% variation in FeC content only causes a 10% variation in the Young’s modulus. It can therefore be deduced that our fabricated FeC-PDMS micropillars which contain between 35-43% FeC particles by weight (an 8% variation) will only have an estimated 4% variation in their corresponding Young’s moduli. Thus, the Young’s moduli of 40% FeC-PDMS can be used for all micropillars in our experiments without the introduction of significant error in our calculations.   Figure E520% w/w pure PDMof part (a)Figure Ethe test aat very lfield gradgradient magneticforces caof the air : (a) Stress-stPDMS, whichS, 20% w/w F. 6 shows therea at the ceow magnetiient is lessare two orde translationan be assume gap are acterain curves fo lies in betweeC-PDMS a magnetic finter of the ac field grad than 10 mTrs of magnil force in thd to be zerod upon soler pure PDMSen the two cnd 40% w/w Feld magnituir gap, we aients. More/mm. It catude smallere magnetic . Effectivelyly by a mag and 40% wurves shown, eC-PDMS obdes measurere able to ac specificallyn be shown than field gmicropillars, the micropnetic bendin/w FeC-PDMis not presentained from d in the air hieve magn, in the tes that these radients req, and as sucillars placedg moment.S (to avoid clted here) (b) the slopes of gap of the eetic fields at area the avalues of thuired to indh, the magn in the test  utter, the curYoung’s Modstress-strain clectromagns high as 27verage mage magnetic uce a signifetic translatarea at the c148  ve for uli of urves et. In 0 mT netic field icant ional enter 149   Figure E6: Measurements obtained for the magnitude of the magnetic field in the 6 mm long air gap of the electromagnet. Pillars are placed in the test area at the center of the air gap for tilting and bending measurements. In the test area, the magnetic field gradient is less than 10 mT/mm and the magnetic translational forces acting on the magnetic micropillars are insignificant.  Figure E7: Side-view of FeC-PDMs magnetic micropillar on a pure PDMS base as seen under an optical microscope. The contrasting colors of the FeC-PDMS micropillar (black) and the pure PDMS base (clear) allows for the observation of the pillar-base boundary during base tilting. As the magnetic field (and consequently the magnetic bending moment acting on the pillar) is increased (from left to right), both the bending and tilting angles increase. 150  Figure E7 shows the different instances of bending and tilting of a 24 µm diameter, 120 µm high FeC-PDMS micropillar as the magnetic field was increased from 0 to 250 mT. It is observed that as the magnetic field (and consequently the magnetic bending moment acting on the pillar) is increased, both the bending and tilting angles increase. Figure E8 shows the measurements obtained for the bending and tilting angles of a group of sample micropillars with dimensions L=120 µm and D=24 µm. Using Matlab to apply a least squares multiple linear regression analysis on the experimental data, we found a value of 0.2232±0.0197 for the slope of the θ vs. α curve based on a 95% confidence interval.  Using equation (10), values of Eb=1.14 MPa, Ep=1.288 MPa, L/D=5 and the slope for the θ vs. α curve, we find a tilt coefficient of Ttilt=1.97±0.174 for the data of Figure E8.  Figure E8: Measurements obtained for the tilting angles θ of a group of micropillars (L=120µm, D=24µm, L/D=5) at various bending angles α. 151  Figure E9 shows experimental measurements obtained for the ratio of bending and tilting angles of micropillars with different aspect ratios. For each of the 5 aspect ratios studied, at least three different micropillars were tested. Using equation (9), and a least squares multiple linear regression analysis in MATLAB, we obtain a tilting coefficient of Ttilt=1.90±0.184 for the data presented in Fig. 9, based on a 95% confidence level, consistent with our previously obtained value of Ttilt=1.97±0.174.  Figure E9: Measurements obtained for the ratio of bending and tilting angles for micropillars with various aspect ratios. For each aspect ratio shown, at least three different micropillars were tested. (Dots show the average values obtained with error bars showing the standard deviation of measurements for each group of micropillars tested). 152  Figures E10(a) and E10(b) show the contributions of the pillar bending angle α and the base tilting angle θ to the total angular deformation β of the pillar tip as the aspect ratio of the pillar is varied.   Figure E10: Contributions of (a) the pillar bending angle α and (b) the pillar tilting angle θ, to the total angular deformation β of the pillar structure. As the pillar aspect ratio becomes larger the contribution of pillar bending to the total deformation becomes larger while the contribution of pillar tilting to the total deformation becomes smaller. For each aspect ratio shown, at least three different micropillars were tested. (Dots show the average values obtained with error bars showing the standard deviation of measurements for each group of micropillars tested). For smaller aspect ratios (L/D≈3), pillar bending and base tilting account for 72% and 28% of the total angular deformation, respectively. As the pillar aspect ratio is increased, the contribution from pillar bending increases while the contribution from base tilting decreases. For the larger aspect ratios tested in our experiments (L/D≈7), pillar bending and base tilting account for 87% and 13% of the total angular deformation, respectively. As seen from Figures E10(a) (a) (b) 153  and C10(b), the theoretical relations depicted in equation (10), used along with Ttilt= 1.90 we obtained from the data in Figure E9, can predict the experimental values with good accuracy. When calculating the bending moment or concentrated force acting on a micropillar based on the measured total deformation, significant errors are introduced if the base tilting effect is overlooked. To obtain the correct spring stiffness for a micropillar structure, we have to note that the total angular deformation of the micropillar is the sum of two different angular deformations; the angular deformation due to bending and the angular deformation from base tilting. Therefore, the system acts as two springs that are connected in series. As is the case with any combination of springs connected in series, the total spring constant of the system will be smaller than that of any of the individual springs that make up the system. That is, the overall stiffness of the elastic pillar on elastic base configuration is smaller than the bending stiffness of the pillar alone.  Depicting the spring constants for bending and tilting with Kbend and Ktilt, respectively, the total spring constant Ktot of the system is  K௧௢௧ ൌ K௕௘௡ௗK௧௜௟௧K௕௘௡ௗ ൅ K௧௜௟௧	,  (12)where, using equations (1) and (4) we have  K௕௘௡ௗ ൌ E௣I௣L 																	K௧௜௟௧ ൌ2E௕I௣T௧ܦ .  (13)If the base tilting effect is ignored the calculated bending moment based on the total angular deformation β is 154   ܯ ൌ K௕௘௡ௗߚ	,  (14)whereas the actual bending moment taking into account both the pillar bending and the base tilting is  ܯ௔௖௧ ൌ K௧௢௧ߚ ൌ ሺ݂ܿሻܯ ൌ ሺ݂ܿሻK௕௘௡ௗߚ (15)Here, cf is a correction factor obtained from  ݂ܿ ൌ ܯ௔௖௧ܯ ൌK௧௢௧K௕௘௡ௗ ൌ11 ൅ T௧௜௟௧2 ൬ܧ௣ܧ௕൰ ቀܦܮቁ.  (16)In earlier studies where bending moments acting on micropillars were estimated based on the pillar deformations and the pillar bending stiffness, the results need to be multiplied by the correction factor cf of Equation (16) to obtain the correct bending moment values that account for both substrate tilting and pillar bending. Figure E11 shows the correction factor for PDMS-on-PDMS and FeC-PDMS-on-PDMS micropillar structures of various aspect ratios.  It is seen that for micropillars with small aspect ratios (L/D=3) the bending moment can be overestimated by as much as 35% if the base tilting effect is not taken into consideration. For micropillars with large aspect ratios (L/D > 10) the overestimation is much less pronounced (less than 10%). 155   Figure E11: To obtain the correct values of bending moment or substrate stiffness in studies where the base tilting effect has not been accounted for, the reported bending moments and substrate stiffness can be multiplied by the correction factor above to obtain the corrected values.  III) Summary To obtain the correct spring stiffness for a micropillar substrate consisting of polymer micropillars on a polymer base, both the stiffness of the micropillars and the stiffness of the base have to be accounted for. When a pure bending moment is applied to the micropillar, as is the case when a magnetic micropillar is placed inside a uniform magnetic field, the total deformation of the micropillar structure consist of pillar bending and base tilting. The total deformation angle of the micropillar is the sum of the base tilting and the pillar bending angles, which are found to be linearly proportional to each other. For a specific pillar and base combination, the ratio of the base tilting angle and the pillar bending angle depends on the young’s moduli of the pillar and 156  base materials, the aspect ratio of the pillar, and the base Tilting Coefficient Ttilt. We demonstrated a unique experimental setup to observe and measure the bending and tilting angles for polymer-pillar-on-polymer-base structures. Based on our experimental measurements of the bending and tilting angles of FeC-PDMS micropillars sitting on pure PDMS substrates, we found a value of Ttilt =1.90 for the tilting coefficient of micropillar structures consisting of a pure PDMS base. We showed that if the base tilting effect is overlooked, the bending moment calculated from the total angular deformation of a pillar will be overestimated. The resulting error can be as large as 35% for micropillars with small aspect ratios (L/D=3).     AppendiThe permactuationFigure F1The flywhmovementx F  : Permanent magn in cell-migr: Permanent meel and link  of the magneanent Magnet actuator ation experagnet actuaconvert rotat holders on tet Actuatosetup that wiments is shtor setup for tional movemhe rail guidesr for Cell Me designed own in Figuapplication ofent of the m.  igration Tand built fore F1.  periodic maotor shaft inests r the purpognetic field onto a sinusoidse of micro micropillar al back and 157 pillar  chips. forth  AppendiThe custhousing, imaging Figure G1and 3D prElectrom(Scientifiin betwemT for 1x G  : Electom-made eand the inare shown in: Image of ininted microchagnet was c Alloys incen the poles Amp input romagnet alectromagncubated mic Figure G1.cubated micrip housing usdesigned in.) and 22 A of the eleccurrent.   nd Microscet actuator,roscope se oscope setup,ed for cell mi FEMM soWG coppertromagnet sope Setup f the customtup used fo the custom mgration tests wftware and  wire (EIS-ihowed magor Live Im designed r cell-migrade electromith live imaghand wounnc.). Measunetic field vaging and 3D-pration experiagnet and thing.  d using Marement of thalues of apinted microments withe custom-desgnetic Irone magneticproximately158 chip  live  igned  bars  field  100 159  Appendix H  : Path and Velocity Data Obtained from Live-Imaging Experiments  Table H1: Path length and velocity data for 25 cells migrating among actuated magnetic pillars as obtained from live-imaging experiments. Cell No.  Length of Path (µm) DX (µm) DY (µm)  RXY Net Disp. X (µm) Avg. Path Vel. (µm/hr) Avg. X‐Vel. (µm/ hr) 1  737.4  486.8  428.2  1.14  278.2  30.7  11.6 2  716.5  379.3  505.7  0.75  302.0  30.0  12.6 3  551.2  304.7  372.7  0.82  227.3  23.6  9.5 4  606.7  298.0  467.3  0.64  252.7  25.7  10.5 5  538.3  425.3  254.0  1.67  317.3  22.9  13.2 6  486.8  288.7  328.7  0.88  212.7  20.3  8.9 7  559.2  399.7  297.3  1.34  341.0  24.7  14.2 8  547.1  384.7  234.0  1.64  271.3  23.3  11.3 9  373.0  252.7  239.3  1.06  242.0  15.5  10.1 10  617.2  428.7  363.0  1.18  332.3  25.7  13.8 11  457.7  311.2  281.5  1.11  216.5  19.6  9.0 12  680.0  376.3  499.5  0.75  281.0  28.3  11.7 13  524.8  288.5  382.5  0.75  149.5  21.9  6.2 14  657.5  415.5  438.0  0.95  355.5  27.5  14.8 15  561.1  400.5  299.5  1.34  351.5  23.7  14.6 16  462.9  321.5  280.0  1.15  257.5  19.5  10.7 17  702.7  421.0  487.0  0.86  334.0  29.7  13.9 18  495.5  331.7  302.0  1.10  250.3  20.6  10.4 19  577.4  386.2  340.8  1.13  328.8  24.3  13.7 20  628.9  438.0  366.7  1.19  352.7  28.5  14.7 21  500.8  352.5  289.5  1.22  307.5  21.3  12.8 22  344.5  226.5  205.0  1.10  178.5  14.5  7.4 23  567.4  367.5  382.0  0.96  343.5  23.6  14.3 24  585.7  348.5  416.5  0.84  259.5  24.4  10.8 25  416.6  249.2  269.0  0.93  176.5  18.1  7.4 Average  555.9  355.3  349.2  1.06  276.8  23.5  11.5 Std. Dev.  99.6  65.2  85.9  0.2561 59.2  4.2  2.5  160   Table H2: Path length and velocity data for 25 cells migrating among non-magnetic pillars as obtained from live-imaging experiments. Cell No.  Length of Path (µm) DX (µm) DY (µm)  RXY Net Disp. X (µm) Avg. Path Vel. (µm/hr) Avg. X‐Vel. (µm/ hr) 1  661.6  518.0  337.0  1.54  490.0  29.6  20.4 2  733.8  562.0  361.0  1.56  462.0  30.6  19.3 3  682.9  505.5  365.0  1.38  370.5  28.5  15.4 4  595.7  427.3  319.0  1.34  350.0  24.9  14.6 5  878.1  498.7  623.3  0.80  458.7  36.6  19.1 6  654.3  503.0  344.0  1.46  429.0  27.7  17.9 7  646.0  441.0  366.0  1.20  409.0  27.1  17.0 8  587.4  434.0  423.0  1.03  282.0  24.5  11.8 9  698.2  464.0  445.0  1.04  368.0  29.1  15.3 10  758.4  426.0  549.0  0.78  398.0  32.6  16.6 11  845.9  566.0  550.0  1.03  472.0  35.2  19.7 12  741.9  372.8  550.2  0.68  278.2  31.0  11.6 13  853.2  621.3  504.0  1.23  601.3  35.6  25.1 14  593.0  471.5  271.8  1.73  325.5  25.2  13.6 15  587.1  437.3  320.7  1.36  369.3  24.9  15.4 16  514.5  370.0  294.7  1.26  335.3  21.4  14.0 17  744.8  456.7  492.7  0.93  248.7  31.0  10.4 18  538.8  396.0  306.0  1.29  277.3  22.5  11.6 19  698.7  445.3  446.7  1.00  338.7  30.3  14.1 20  688.6  475.3  390.7  1.22  364.7  28.7  15.2 21  907.5  547.3  629.3  0.87  335.3  37.8  14.0 22  910.7  656.0  482.0  1.36  522.7  38.0  21.8 23  613.1  410.0  333.3  1.23  319.3  25.6  13.3 24  355.3  286.0  164.0  1.74  278.0  23.1  11.6 25  839.2  617.7  433.3  1.43  570.3  35.0  23.8 Average  693.1  476.4  412.1  1.22  386.2  29.5  16.1 Std. Dev.  131.4  84.3  112.9  0.2810 92.4  4.8  3.9   AppendiFigure I1:Cell amonmagnetic px I  : Distri Distribution g magnetic illars. butions of Cof cell velocitpillars spent ell Instanties based on pmore time maneous Veloercentage of igrating at lcities  time spent at ower speeds a certain instcompared toantaneous ve cells among161  locity.  non-  Figure I2:instantanequartile anand non-m  Boxplots shoous velocity. d maximum agnetic pillarwing the distFrom left tovalues, respecs show signifribution of ce right, vertictively, and ouicantly differell velocities bal lines showtliers are shont velocity di ased on perce the minimuwn with red pstributions. ntage of timm, first quarlus signs. Cee spent at a ctile, median, lls among ma162  ertain third gnetic 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0135703/manifest

Comment

Related Items