Citation
Endothelial Cell Response to Microengineered Surfaces Analyzed by Fluorescent and Atomic Force Microscopy

Material Information

Title:
Endothelial Cell Response to Microengineered Surfaces Analyzed by Fluorescent and Atomic Force Microscopy
Alternate Title:
Endothelial response to microengineered surfaces analyzed by fluorescent and atomic force microscopy
Creator:
Feinberg, Adam Walter
Copyright Date:
2004

Subjects

Subjects / Keywords:
Arteries ( jstor )
Cantilevers ( jstor )
Cells ( jstor )
Cultured cells ( jstor )
Diameters ( jstor )
Focal adhesions ( jstor )
Moduli of elasticity ( jstor )
Plasmas ( jstor )
Tissue grafting ( jstor )
Topography ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright the author. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
4/30/2005
Resource Identifier:
436097556 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

ENDOTHELIAL CELL RESPONSE TO MICROENGINEERED SURFACES ANALYZED BY FLUORESCENT AND ATOMIC FORCE MICROSCOPY By ADAM WALTER FEINBERG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2004

PAGE 2

Copyright 2004 by Adam Walter Feinberg

PAGE 3

This document is dedicated to my grandfather William Epstein for his loving guidance

PAGE 4

ACKNOWLEDGMENTS There are many people that I would like to thank for their help and encouragement during the research and writing of my PhD dissertation. Academically, I would like to thank my advisor Dr. Anthony Brennan for his guidance and straightforward manner. I am grateful for the opportunities I found in his research group and the financial support he provided me during my time at UF. I would also like to thank the other members of my committee, Dr. Winfred Phillips, Dr. Chris Batich, Dr. Dan Purich and Dr. C. Keith Ozaki for their time and valuable critique of my dissertation. The members of my research group have also been valuable to me as friends and coworkers. When I entered the research group, Jeanne McDonald, Jeremy Mehlem and Lee Zhao each supplied valuable guidance to me as a new graduate student and I watched each one graduate and continue on to promising careers. Wade Wilkerson started with me at UF and was a great friend and coworker. He has my thanks for making my time at UF an interesting and entertaining experience and I wish him and his wife Laura a long and happy life. Clay Bohn, Amy Gibson, Leslie Wilson, Brian Hatcher, Nikhil Kothurkar, Thomas Estes and Charles Seegert have each contributed their expertise when I required assistance. I also would like to thank Dr. Edward Block, and Mr. Zaher Abouhamze for their assistance in providing endothelial cells for my research. I would also like to thank Dr. L. Amelia Dempere, the director of MAIC, for her assistance with the Dimension 3100 AFM. Personally, many people have offered great support both in growing up and in graduate school. My parents have been nothing but a source of love and motivation for iv

PAGE 5

as long as I can remember; I could not ask for two better parents. They are both my role models and friends and I thank them. My brothers, Matt and Ben, have been great even though we disagree, fight and embarrass each other in front of girls. Of course, I would not have it any other way, and I thank them both for their support and for keeping life interesting. I dragged Peter Vinch, my best friend from high school, down here to sunny Florida for his graduate work and he has been my roommate, lifting partner and friend once again. Finally, no acknowledgement for this thesis would be complete without recognizing the love, support and motivation given to me by my fiance, Ms. Ayelet Tal. She has brought both happiness and perspective to my life and I can not imagine having spent the last 4+ years at UF without her. I hope I have not forgotten anyone, and if I have then please be assured that you have my deepest thanks. v

PAGE 6

TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................iv LIST OF TABLES ...............................................................................................................x LIST OF FIGURES ...........................................................................................................xi ABSTRACT ....................................................................................................................xvii CHAPTERS 1 INTRODUCTION......................................................................................................1 2 BACKGROUND........................................................................................................6 Introduction................................................................................................................6 Anatomy of a Blood Vessel.......................................................................................6 Cardiovascular Disease Linked to Endothelial Cell Dysfunction..............................8 Vascular Grafts – Natural, Synthetic and Tissue Engineered Replacements...........11 Analysis of Current Vascular Grafts.................................................................12 The Mechanics of the Cardiovascular System and Vascular Grafts.................15 Microscale Patterning of Topography and Chemistry on Surfaces..........................18 Atomic Force Microscopy and Its Application to Biological Systems....................21 AFM Force Curves............................................................................................25 Force Curve Anatomy.......................................................................................26 Interpreting Specific Interactions from AFM Force Curves.............................29 3 ULTRASTRUCTURAL ANALYSIS OF PORCINE PULMONARY ARTERY..31 Introduction..............................................................................................................31 Materials and Methods.............................................................................................31 Harvesting Porcine Pulmonary Artery..............................................................31 Scanning Electron Microscopy of Porcine Artery............................................33 Fluorescent Microscopy....................................................................................33 Optical Image Processing..................................................................................34 Atomic Force Microscopy of Porcine Artery....................................................35 Mounting the tissue samples......................................................................36 Examining topography in contact mode....................................................37 Examining topography and stiffness in force volume mode.....................39 vi

PAGE 7

AFM data analysis.....................................................................................39 Results and Discussion.............................................................................................41 SEM Analysis of Porcine Artery......................................................................41 Optical Microscopy of Endothelial Cell Monolayer.........................................44 AFM of Fresh Porcine Pulmonary Artery.........................................................45 Topography by contact mode.....................................................................45 Mechanical properties by force volume mode...........................................47 Conclusions..............................................................................................................54 4 EFFECT OF ARGON PLASMA TREATMENT ON PDMS ELASTOMER INVESTIGATED BY AFM.....................................................................................55 Introduction..............................................................................................................55 Materials and Methods.............................................................................................55 Sample Preparation...........................................................................................55 Argon RFGD Plasma Treatment.......................................................................56 AFM Operation.................................................................................................58 Analysis of Force Curves for Elastic Modulus.................................................59 Results and Discussion.............................................................................................60 Topography.......................................................................................................60 Force Curve Properties......................................................................................64 Polymer Elastic Modulus..................................................................................65 Change in elastic modulus with indentation depth....................................67 PDMSe viscoelastic properties..................................................................69 Force Volume Images.......................................................................................70 Conclusions..............................................................................................................71 5 ENDOTHELIAL CELLS CULTURED ON MICROENGINEERED SURFACES EVALUATED BY FLUORESCENT MICROSCOPY......................72 Introduction..............................................................................................................72 Materials and Methods.............................................................................................72 Substrate Production.........................................................................................72 Microtopographical surface patterns..........................................................72 Chemical surface micropatterns.................................................................77 Cell Culture.......................................................................................................78 Fluorescent Staining..........................................................................................80 Imaging.............................................................................................................81 Image Processing..............................................................................................82 Statistical Analysis............................................................................................86 Results and Discussion.............................................................................................87 Surface Analysis of Cell Culture Substrates.....................................................87 Topography of flat and micropatterned PS and PDMSe...........................87 Immunoflourescent staining of FN micropatterns on PS and PDMSe......90 Immunoflourescent Imaging of Cell Structure.................................................93 Qualitative summary of EC response to microengineered surfaces..................93 EC response to microtopographies and the disruption of cell spreading...94 vii

PAGE 8

Looking at EC adaptation to flat substrates with and without FN...........100 Quantifying The Cell Density, Percent Coverage And Cell Area..................105 Quantifying Focal Contact Adhesions To The Substrate................................112 Conclusions............................................................................................................121 6 CELL RESPONSE TO ENGINEERED SURFACES ANALYZED BY ATOMIC FORCE MICROSCOPY.......................................................................123 Introduction............................................................................................................123 Materials and Methods...........................................................................................124 Substrate Production.......................................................................................124 Cell Culture.....................................................................................................126 AFM Experimental Setup and Procedure.......................................................126 AFM setup...............................................................................................126 AFM operation.........................................................................................127 AFM data analysis...................................................................................128 Results and Discussion...........................................................................................129 EC topography on PS and PDMSe..................................................................129 Mechanical Properties of ECs.........................................................................131 Conclusions............................................................................................................142 7 SIMULTANEOUS QUANTIFICATION OF RECEPTOR-LIGAND LOCATION, BINDING FORCE AND INTERACTION DISTANCE ON LIVE CELLS..........................................................................................................144 Introduction............................................................................................................144 Materials and Methods...........................................................................................148 Cantilever Preparation.....................................................................................148 Cell Culture and Instrument Setup..................................................................150 Data Analysis..................................................................................................151 Results and Discussion...........................................................................................153 Cell Response to AFM Imaging......................................................................153 Quantifying Interaction Force.........................................................................153 Quantifying Interaction Distance....................................................................161 Mapping Receptor Location on the Cell Surface............................................163 8 CONCLUSIONS AND FUTURE RESEARCH....................................................166 Conclusions............................................................................................................166 Future Work...........................................................................................................169 Endothelial Cells, Circulating Endothelial Cells and Progenitor Endothelial Cells.............................................................................................................170 Endothelial Cell Adhesion Molecules.............................................................172 viii

PAGE 9

APPENDIX A ANALYSIS OF FORCE CURVES FOR ELASTIC MODULUS........................174 B STATISTICAL ANALYSIS OF CELL RESPONSE TO ENGINEERED SURFACES............................................................................................................178 Cell Density............................................................................................................178 Percent Coverage....................................................................................................182 Cell Area................................................................................................................187 Size of Focal Contact Adhesions...........................................................................191 Focal Contact Adhesions per Cell..........................................................................193 LIST OF REFERENCES.................................................................................................197 BIOGRAPHICAL SKETCH...........................................................................................208 ix

PAGE 10

LIST OF TABLES Table page 2-1. Types of small caliber vascular grafts and their performance (adapted from Teebken and Haverich)...........................................................................................................14 3-1. Maximum scan rates and scan speeds for selected scan sizes when examining fresh porcine artery by contact mode AFM......................................................................39 3-2. Elastic modulus for different regions of porcine artery lumen..................................53 4-1. Variation in RMS roughness of the different polymers examined............................60 4-2. Elastic modulus of polymers surfaces indented ~5 nm in to the substrate...............67 5-1. Different microengineered surfaces used as substrates for EC culture.....................79 5-2. Variation in RMS roughness of the different materials examined............................89 6-1. Elastic modulus for the different regions identified in figure 6-4...........................135 6-2. Elastic modulus for the different regions identified in figure 6-6..........................138 6-3. Elastic modulus for the different regions identified in figure 6-7..........................142 x

PAGE 11

LIST OF FIGURES Figure page 2-1. Anatomical structure of the blood vessel wall............................................................7 2-2. Fluorescent micrographs actin filaments and other cell components..........................8 2-3. Examples of cardiovascular disease resulting from dysfunction of the endothelium.11 2-4. Computer models showing the stress concentration due to sutures at (A) end-to-side and (B) end-to-end artificial graft to host artery anastomoses..........................17 2-5. Graphs showing the change in compliance in an artery-to-artery anastomosis due solely to the applied sutures.....................................................................................18 2-6. Optical microscopy of stained PVECs on a textured RFGD treated siloxane elastomer substrate...................................................................................................20 2-7. Effect of spreading on cell growth and apoptosis in bovine (A) and human (B) endothelial cells........................................................................................................21 2-8. Focal adhesion formation increases as cell spreading is promoted...........................22 2-9. Schematic of the Dimension 3100 AFM used for all analyses.................................24 2-10. Schematic of a silicon nitride substrate used for contact and force volume mode AFM. Each side has 0.6 m thick, 100 and 200 m long V-shaped cantilevers on it, with the wider leg and higher spring constant cantilevers on the bottom (not drawn to scale).........................................................................................................25 2-11. A force curve as output by the Nanoscope version 5 release 12 software. The white line is the extending (trace) movement and the yellow line is the retracting (retrace) movement..................................................................................................27 2-12. Schematic drawing representing the tip and sample interaction during trace and retrace seen in the force curve in Figure 2-11..........................................................28 2-13. Illustration of a receptor-ligand pair as it stretches and ruptures. The receptor-ligand pair begins in a compressed state with the protein in a globular form. As the pair is stretched the protein on ravels until it is taught and the C-C-C backbone is xi

PAGE 12

stretched. Upon reaching a critical force, the specific interaction ruptures and the extended proteins collapse back down to a globular form.......................................30 3-1. This is an example of a porcine heart where the pulmonary artery is labeled as the pulmonary trunk. The pulmonary artery can be seen originating from the right ventricle and would terminate at the lungs if they were still attached.....................32 3-2. A schematic of the way strips of tissue are cut from the artery, marked for blood flow direction and glued to a substrate for AFM or SEM analysis..........................33 3-3. An example of the methodology used to examine fluorescent images of the stained cell nuclei in fixed porcine artery. The well illuminated area is cropped and then small regions of interest (ROI) are examined. Each ROI is contrast and brightness enhanced, filtered to enhance nuclei edges and then the nuclei are hand counted. ImageJ tracks each tagged nucleus providing total number upon completion.........35 3-4. Photographs of the AFM experimental setup used to analyze fresh and fixed sections of porcine pulmonary artery.......................................................................36 3-5. A SEM micrograph of a pulmonary artery cross-section. The three layers of the arterial wall, the tunica intima, tunica media and tunica adventitia, can clearly be identified as changes in tissue morphology..............................................................41 3-6. SEM micrographs taken of the lumen of fixed sections of porcine pulmonary artery.43 3-7. Fluorescent image of lumen of a fixed porcine pulmonary artery. The cell nuclei of the ECs are stained blue using DAPI and the actin cytoskeleton network is stained red using phalloidin-TRITC. (100x magnification).................................................45 3-8. Contact mode AFM topographic scans of fresh porcine pulmonary artery imaged in HBSS at ambient conditions....................................................................................47 3-9. An example of the adhesions that form between the AFM tip and cell membrane when the tip becomes contaminated with membrane and protein fragments after too much force is applied...............................................................................................48 3-10. The topography of a section of artery reconstructed from the force volume image. Force curves from the circled regions were analyzed in order to determine the average elastic moduli. Corresponding plots from the force curves in each region are overlaid in the numbered graphs. Regions 1, 2 and 4 are taken on raised areas indicating nuclei and region 3 is a lower area of the cell body. There is a significant difference between the elastic modulus of regions 1, 2 and 4 compared to region 3 (P<0.05).................................................................................................50 3-11. The topography of a section of artery reconstructed from the force volume image showing a thin slice spanning across nuclei from two adjacent cells. Force curves from the circled regions were analyzed in order to determine the average elastic moduli. Corresponding plots from the force curves in each region are overlaid in xii

PAGE 13

the numbered graphs. Regions 1, 2, 3 and 4 are moving from a raised to lower area and region 5 is a raised area. There is a significant difference between the elastic modulus of regions 1 and 4 (P<0.05) and regions 2 and 4 (P<0.05)........................52 4-1. An illustration of the two different types of plasma systems used to treat the PDMSe. The home built system flows argon through an RF coil and then directs the plasma onto the sample. The reactive ion etcher uses two plates on the top and bottom of a chamber filled with argon to create a plasma that surrounds the entire sample.......................................................................................................................57 4-2. A schematic of the AFM experimental setup used to image the PDMSe submerged in HBSS. The AFM scanner was lowered into the liquid filled Petri dish and scanned a coating of plasma treated PDMSe on the bottom....................................58 4-3. Topographical images of the PDMSe (A) before and (B) after treatment with an argon plasma. The 100 m 2 areas were imaged in liquid using contact mode.......61 4-4. AFM contact mode topographic scans of PDMSe treated with (A) 1 minute exposure and (B) 10 minute exposure to argon plasma in the RIE. For the same size scan area of 10 m, a distinct change in surface morphology is observed even though the surface roughness remains essentially unchanged.................................62 4-5. The 1 minute RIE plasma treated PDMSe had a surface morphology (A) that appeared similar in morphology to phase separated spinoidal decomposition seen in block copolymers. A close-up of this region (B) shows an ~4 nm height variation and the size of these ‘fingers’ can be quantified using (C) a cross-section plot......64 4-6. Representative force curves from (a) PDMSe and (b) argon plasma treated PDMSe. Force curves were recorded during force volume imaging......................................66 4-7. Drawing illustrating PDMS oligomers (oils) coating the surface of the bulk elastomer and the silicon nitride AFM tip. This is believed to occur during imaging of the PDMS samples and is responsible for the large jump-on and pull-off adhesive forces seen in the force curves..................................................................66 4-8. The elastic modulus of PDMS elastomer cured against glass, air cured, air cured with 10 sec argon plasma treatment and air cured with 1 min argon plasma treatment. All samples were examined under liquid buffer to minimize capillary effects. There was only a significant difference between the untreated and plasma treated samples (P<0.05). Force curve indentation were performed at 1 Hz for all samples.....................................................................................................................69 4-9. Dependence of the elastic modulus on indentation speed for both trace (indentation into the sample) and retrace (withdrawal from the surface). There are clear trends of a higher elastic modulus for trace compare to retrace, but no significant difference between indentation rates........................................................................70 xiii

PAGE 14

5-1. These are optical micrographs of photomasks printed on standard transparency using a high-resolution 5000 DPI image setter........................................................74 5-2. CAD drawings of the patterns used for the chrome on glass photomask..................74 5-3. This diagram outlines the steps to produce micrometer scale topographical features75 5-4. Samples for cell culture with micropatterned surfaces are prepared by placing a cured micropatterned substrate in the center of a 35 mm diameter Petri dish and curing in place by surrounding it with PDMSe prepolymer. ECs in media are poured into the remaining volume of the Petri dish and cultured for 4, 7 or 14 days.76 5-5. An illustration of the CP process where FN is adsorbed to a PDMSe stamp and transferred to another surface...................................................................................78 5-6. An illustration of the image processing performed in ImageJ in order to quantify the cell properties for the actin (A, C and E) and vinculin (B, D and F) stains.............84 5-7. An illustration of the image processing used to count cell nuclei and nuclear area.86 5-8. (A) Fluid tapping mode image of PS (625 m 2 area). (B) PDMS elastomer imaged with tapping mode in fluid (100 m 2 area). This sample was the standard Silastic T-2............................................................................................................................88 5-9. AFM tapping mode topographic images of the hexagonally patterned silicon wafer master.......................................................................................................................89 5-10. AFM tapping mode images showing the topography of the PDMSe substrates....91 5-11. Fluorescent images of CP FN patterns..................................................................92 5-12. Fluorescent images taken at 4, 7 and 14 days for PVECs cultured on all the experimental surface conditions used......................................................................97 5-13. ECs cultured on microtopographically patterned PDMSe for 14 days..................99 5-14. AFM topographic image of PDMSe patterned with 20 m wide raised hexagons. This surface prevented adhesion of ECs with cell growth restricted to the flat surface surrounding the patterned area...................................................................100 5-15. (A) PS 4 day. (B) PS 4 day. (C) PS with FN 4 day. (D) PS with FN 14 day....102 5-16. (A) PDMSe plasma 14 day. (B) PDMSe plasma with FN 14 day.......................103 5-17. PVECs cultured 2 days on plasma treated PDMSe patterned with 20 m FN hexagons. Hexagons visible from small cross reaction with primary or secondary antibody. This is the edge of the patterned region as the growing monolayer approaches the non-patterned area in black...........................................................104 xiv

PAGE 15

5-18. PVECs grown on PS patterned with FN stars for periods of (A) 4 days and (B) 14 days in culture........................................................................................................105 5-19. PVECs are grown on FN micropatterned plasma PDMSe...................................106 5-20. A plot of cell density on different engineered surfaces at 4, 7 and 14 days culture time. Represented are cell numbers per 1,527,295 m 2 , the field of view at 100x, error bars represent one standard deviation........................................................108 5-21. Plotted is the number of cells on PDMSe_FN_circle as a function of culture time. The dashed line represents a 2 nd order polynomial best fit to the data with the corresponding equation shown. Error bars represent one standard deviation and cell number is given per 100x field of view (1,527,295 m 2 )...............................109 5-22. A plot of percent coverage of ECs on the different engineered substrates as a function of culture time. Error bars have been omitted to prevent clutter. There are no data points for PDMSe_FN_hex at 7 and 14 days or for PDMSe_FN_star at 14 days because cell cultures had died........................................................................111 5-23. A plot of the average cell area as a function of the engineered substrate ECs were cultured on and the length of time they were cultured. Cell areas need to be compared with cell density and cell morphology to understand if the response is favorable or not. Error bars represent one standard deviation...........................113 5-24. A plot of the variation in the diameter of focal contact adhesions between ECs and engineered substrates. For each substrate type, values are pooled from 4, 7 and 14 days culture. Error bars represent one standard deviation.................................114 5-25. A plot of number of focal contact adhesions per cell as a function of the engineered substrate ECs were cultured on and the length of time they were cultured. Error bars represent one standard deviation........................................115 5-26. A close-up of an EC cultured on different surfaces showing the variation in the number and arrangement of focal contact adhesion (green immunofluorescent stain for vinculin) and organization of the actin cytoskeleton (red phalloidin-TRITC fluorescent stain for actin)......................................................................................119 5-27. Examining the alignment of focal contact adhesions of ECs on PDMSe_FN_circle to the underlying FN micropattern. The images are..............................................120 5-28. Lines overlaid on top of an image of an EC on the PDMSe_FN_circle surface to show alignment to the underlying FN micropattern...............................................121 5-4. Samples for cell culture are prepared by placing a cured micropatterned substrate in the center of a 35 mm diameter Petri dish and curing in place by surrounding it with PDMSe prepolymer. ECs in media are poured into the remaining volume of the Petri dish and cultured for 4, 7 or 14 days........................................................125 xv

PAGE 16

6-1. The AFM tip is lowered directly into a Petri dish with cultured ECs on the bottom..127 6-2. AFM contact mode topographic images of PVECs imaged while living in HBSS with corresponding fluorescent optical micrographs of PVECs cultured under the same conditions......................................................................................................130 6-3. A comparison between contact mode and force volume images.............................132 6-4. Comparison of the same region of ECs on PS imaged in both contact mode and force volume mode.................................................................................................134 6-5. Comparison of the mechanical properties of PVECs grown on PS or FN coated PDMSe...................................................................................................................137 6-6. An analysis of the topography and mechanical properties of ECs grown on PS for 4 days.........................................................................................................................139 6-7. An analysis of the topography and mechanical properties of ECs grown on PDMSe_FN for 4 days...........................................................................................141 7-1. Illustrations of the interaction between the sLe X coated AFM tip and the PVEC surface....................................................................................................................148 7-2. Topographic AFM contact mode images of PVECs in HBSS revealing the effects of contact force and imaging time on the cell membrane and cytoskeleton..........154 7-3. The force curves were analyzed for adhesion peak forces, the values were sorted into histograms and then an autocorrelation function was applied........................158 7-4. A histogram of the specific interaction distance from 0 to 150 nm split into 12 nm bins.........................................................................................................................163 A-1. 3-dimensional drawings representing (a) the conical tip used in the Sneddon model and (b) the actual tip shape, which is a four-sided pyramid...................................174 A-2. Sketch illustrating the method for analyzing the linear portion of the force curve in order to calculate the elastic modulus of the sample. The points (z 1 , d 1 ) and (z 2 , d 2 ) are taken from the linear contact region of the curve. (Adapted from Radmacher et al)...........................................................................................................................176 xvi

PAGE 17

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ENDOTHELIAL RESPONSE TO MICROENGINEERED SURFACES ANALYZED BY FLUORESCENT AND ATOMIC FORCE MICROSCOPY By Adam Walter Feinberg May 2004 Chair: Anthony Brennan Cochair: Winfred Phillips Major Department: Biomedical Engineering Dysfunction of the endothelium leads to a number of cardiovascular diseases including atherosclerosis and intimal hyperplasia. As a first step toward engineering a synthetic small diameter vascular graft, the strategic use of micrometer scale topography and patterned chemistry has been used to control endothelial cell (EC) adhesion, spreading and overall function. Polystyrene (PS) Petri dishes and polydimethylsiloxane elastomer (PDMSe) were micropatterned with topographies and fibronectin (FN) using a combination of microprocessing and microcontact printing techniques. ECs were cultured for periods of 4, 7 and 14 days on the substrates and analyzed using immunoflourescent staining and atomic force microscopy (AFM). It was found that 1 m vertical step heights in PDMSe disrupt EC adhesion and spreading for lateral spacings from 2 to 100 m. Flat PS supported the formation of EC monolayers independent of FN coating while PDMSe coated with FN did not support xvii

PAGE 18

robust growth. PDMSe with FN micropatterns 20 to 40 m in width supported rapid initial growth, but cells became stuck resulting in apoptosis by 14 days. Square packed 3 m diameter FN circles spaced at 6 m demonstrated a cell density significantly higher (P<0.05) than all other tested surfaces by 14 days. On the FN circle patterned PDMSe cell density was 1536 247 cells/mm 2 , half the cell density of 3196 336 cells/mm 2 found in vivo, with an average of 41 12 focal contact adhesions per cell. ECs cultured on PS had a dense cytoskeleton with uniform cell elasticity across the cell body as found by AFM. In contrast, ECs cultured on PDMSe coated with FN had a cytoskeleton that was less dense and increased in stiffness from the periphery of the cell to the nuclear region. AFM was further used to quantify the expression of P-selectin on the surface of ECs mapping both the location of these receptors as well as measuring the receptor-ligand bond rupture force. Results demonstrate that P-selectin forms 1 to 2 m diameter patches at the periphery of the cell when ECs are cultured on PS. SLe X -(P-selectin) bond rupture force was calculated as 183 40 pN at a pulling rate of 12 m/s. xviii

PAGE 19

CHAPTER 1 INTRODUCTION In today’s aging society, advanced technologies are improving the quality and duration of life for people across the world. As traditional sciences such as biology and chemistry merge with the fields of medicine and applied engineering, multidisciplinary teams have mapped the human genome, created drugs to treat a host of diseases and manufactured medical devices to replace joints and organs such as the heart. Great progress has been made, but the future still holds the most exciting discoveries in the biomedical field. That is because some of the very fundamental issues in understanding biology and its interaction with the manmade world are still poorly known. The building block of the human body, a single cell, is an amazingly complex machine that regulates a myriad of complex biochemical reactions through its metabolism. Though a cell can be seen easily with an inexpensive optical microscope, visualizing the interaction between proteins, lipids and carbohydrates inside the cell requires advanced techniques ranging from immunochemical assays and fluorescent, electron and atomic force microscopy to DNA and proteomic arrays. Ideally, understanding how any one cell in the body functions would allow the same findings to be applied to other cell types. Unfortunately, the difficulty in standardizing cellular interaction with environmental stimuli is complicated by highly differentiated cell types with specialized functions. While many cellular processes are standard for most mammalian cells such as mRNA transcription and actin based motility, most cells have unique functions such as the phagocytic behavior of leukocytes or bone 1

PAGE 20

2 readsorbtion by osteoclasts. As a result, any valid study of the interaction between cells and materials needs to clearly address the cell types involved and the important environmental stimuli; be it cytokines, pharmaceuticals, extracellular matrix, manmade materials or other cells. In this research, endothelial cells (ECs) and the interaction with microengineered surfaces is examined to determine conditions that elicit normal EC growth and function on these manmade materials. This is an area of critical interest because ECs are the interface between blood and the rest of the body throughout the cardiovascular system. Referred to as the intima, the endothelial cell monolayer is on the interior of all blood vessels from the largest artery to the smallest capillary. Dysfunction of the intima leads to the development of occlusions that reduce or eliminate blood flow or the formation of small clots that break free and form emboli. A number of clinical procedures exist to correct these conditions including drugs that dissolve clots and plaques, endovascular techniques that use angioplasty balloons and stents to reopen and maintain patent vessels and bypass grafts that circumvent damaged areas. The research presented in this dissertation has direct application to maintaining the patency of these stents and bypass grafts. The ability to elicit and control normal EC growth on manmade materials is imperative because current stents and bypass grafts fail after a period of time due to the failure to form or maintain an endothelium. Autologous vein grafts have a viable endothelial cell layer but eventually fail in the arterial circulation due to formation of a new occlusion or mechanical failure, such as an aneurysm, at the higher arterial pressures. Stents and artificial vascular grafts fail to form an EC monolayer at all leaving

PAGE 21

3 metal and plastic as the long-term blood contacting surface. Extensive research has been conducted trying to improve small diameter vascular grafts ranging from entirely synthetic to entirely tissue engineered biological replacements. Unfortunately, there has yet to be any successful designs that have proven safe and effective for clinical use. Therefore, it becomes critical to determine why human tissues, and specifically ECs, fail to attach and proliferate on artificial graft surfaces. To produce a synthetic surface that will promote tissue regeneration of a functional EC monolayer in vitro or in vivo requires the engineering of a biomimetic interface. It is theorized that this may be achieved by fabricating a surface that resembles the basement membrane that ECs are supported by in blood vessels. Specifically, this will require topography that orients cells and chemistry that provides anchor points for adhesion. A combination of fabrication techniques can be used to engineer the graft surface on the cellular level. Designing micrometer scale topographical features to direct cell growth has become commonplace using micromolding techniques adapted from the microelectronics industry. A related method known as microcontact printing (CP) can be used to add chemical functionality to a surface in specific well-defined patterns. Combining these techniques results in a system for controlling cell behavior and can be used in conjunction with tailored mechanical properties to replicate a natural extracellular matrix (ECM). In addition to building a biomimetic surface to support EC monolayer growth, there needs to be a valid technique to evaluate how ECs are in fact responding to the manmade substrate. Recent advances in scanning probe microscopy, and particularly the subset of techniques known as atomic force microscopy (AFM), provide the ability to examine

PAGE 22

4 biological systems on the micrometer and nanometer scale and determine mechanical and chemical properties of the biointerface. Studies have demonstrated the ability to determine the elastic modulus and topography of living cells and synthetic surfaces in simulated physiologic fluids. The capability of the AFM has been expanded to map the location of specific cell surface receptors on the membrane and to quantify the binding force to the complementary ligand grafted to the AFM tip [1]. This technique allows the cellular response to a synthetic surface to be evaluated and compared to the cell as it exists in its natural state in vivo. While the AFM can be used to evaluate cellular response, it is proposed that a unique approach to design and fabrication of a small diameter vascular graft by evaluating EC response is found in combining the AFM based techniques with the microtopographical and chemical surface modification methods. The research proposed here is not intended to solve the small diameter artificial vascular graft problem in one fell swoop. Rather, it is aimed at attacking the problem from a new angle using a combination of techniques rarely combined in the same project. As a brief outline, a combination of electron, immunofluorescent and atomic force microscopy will be used to quantify the morphology, density and mechanical properties of ECs in fresh porcine pulmonary artery. This will be used as a basis to determine how ECs cultured on microengineered surface in vitro compare to the in vivo state. Then, the basic topographical and mechanical properties of the polystyrene (PS) and polydimethylsiloxane elastomer (PDMSe) used as cell culture substrates will be evaluated. This will be followed by an in depth evaluation of EC growth on PDMSe and PS surfaces engineered with microtopographies and/or micropatterned fibronectin (FN). The ECs will be immunofluorescent labeled to identify the cell nuclei, actin cytoskeleton

PAGE 23

5 and focal contact adhesions. Digital images captured on an epifluorescent microscope will be analyzed using image processing software to determine cell density, spreading, cell area and focal contact adhesions per cell as a function of culture substrate and incubation time (4, 7 or 14 days). Then, AFM will be used to find the variation in morphology and mechanical properties of ECs on the basic substrates. This will be followed by an advanced AFM technique that probes for the expression of membrane receptors (P-selectin in this example). The location of these receptors on the cell surface will be mapped and the receptor-ligand binding force to a ligand functionalized AFM tip will be calculated. Tying all this together will be a unified picture of how microengineered surfaces can be used to grow an EC monolayer on an elastomer, how the EC properties vary on the different surfaces and the relevance to EC properties found from fresh porcine artery.

PAGE 24

CHAPTER 2 BACKGROUND Introduction Engineering an endothelial cell (EC) monolayer requires an in-depth understanding of the fundamental issues involved in directing cell adhesion and function. This includes a thorough understanding of the physiology and anatomy of the blood vessels and the diseases that lead to dysfunction of the endothelium and underlying basement membrane. To design a surface intended to be used in small diameter vascular grafts, it is also necessary to review existing artificial vascular grafts and advantages and disadvantages associated with each type. Once a design is realized, the fabrication techniques must be fully understood and optimized to ensure the designs are successfully transferred to a functional device. Finally, the techniques used to evaluate the microengineered surfaces and EC growth must provide an in vitro model that is relevant to the in vivo performance in order to understand and improve the device. Anatomy of a Blood Vessel Blood vessels are composed of three layers known as the tunica intima, tunica media and tunica adventitia (figure 2-1) [2]. The intima is composed of a monolayer of ECs supported by an extracellular matrix known as the basement membrane or basal lamina membrane. This intimal layer is the lumen of the blood vessel and acts as the interface between the blood and the rest of the body. It moderates hemodynamics, clotting and mass transport of nutrients and waste into and out of the blood. The media is composed of smooth muscle cells along with elastin. This layer is responsible for elastic 6

PAGE 25

7 and contractile properties of the vessel. The compliance and constrictive properties of artery during pulsatile flow are due to SMC contraction and the elastic properties of the elastin layer. Also present in the media is collagen that acts as a structural reinforcement to prevent excessive dilation of the blood vessel. The third layer is the adventitia and is composed mostly of fibroblasts and loose connective tissue. This layer provides a vascular bed to supply nutrients to the tissue of larger blood vessels and is absent for small vessels where diffusion can maintain cell viability. n c ize LumenLumenInternal elastic Internal elastic LaminaLaminaExternal elastic External elastic LaminaLaminaEndotheliumEndotheliumSubendothelialSubendothelialLayerLayerTunica MediaTunica MediaTunica Tunica AdventitiaAdventitia LumenLumenInternal elastic Internal elastic LaminaLaminaExternal elastic External elastic LaminaLaminaEndotheliumEndotheliumSubendothelialSubendothelialLayerLayerTunica MediaTunica MediaTunica Tunica AdventitiaAdventitia B A Figure 2-1. Anatomical structure of the blood vessel wall. (A) An illustration of the different tissue and extracellular matrix types. (B) In this optical micrograph, elastic tissue is colored deep purple, collagen is pale pink and cytoplasm (ismooth muscle and nerve) is purple. Note the distribution of elastin in the wall of this artery. A continuous sheet of elastin comprises the internal elastilamina (white highlight in image at right). Scattered elastic fibers characterthe adventitia. The intima consists of little more than endothelium, barely visible between the internal elastic lamina and the lumen ( http://www.siumed.edu/~dking2/crr/images/CR023b.jpg ).

PAGE 26

8 Figure 2-2. Fluorescent micrographs actin filaments and other cell components. (A) Confocal fluorescence microscopy using a thin specimen: bovine pulmonary artery ECs, triple stained with BODIPY FL-labelled tubulin antibody, Texas Red -X phalloidin staining the actin filaments and DAPI staining the chromosomes in the nucleus (not visible) ( http://www.brc.cornell.edu/brcinfo/mif/Gallery/2color.jpg ). (B) Pulmonary artery smooth muscle cells labelled for monomeric G-actin (green) and F-actin (red) ( http://www.ucl.ac.uk/medicine/bhflabs/fluroimages/Image-1.gif ). B A Cardiovascular Disease Linked to Endothelial Cell Dysfunction A number of cardiovascular diseases result from dysfunction of the EC layer and subsequent occlusion or clot formation. ECs (figure 2-2A) are rounded and well spread cells with a developed actin cytoskeleton. Because these cells are in the blood stream they are designed to exist in a state of constant shear flow, which is needed to prevent clot formation on the surface. Furthermore, the ECs normally produce chemokines like nitrous oxide (NO) and antithrombin that act as anticoagulants and maintain normal physiologic blood flow [3]. When the EC layer is disrupted, a number of problems occur. In atherosclerosis, fatty deposits called plaques develop on the interior of the blood vessel at tears of gaps in the endothelium. As these plaques grow a number of problems occur as the ECs fail to cover the deposit. The primary issue is occlusion of the blood vessel from undamaged (figure 2-3A) to mostly occluded (figure 2-3B). As the plaques grow

PAGE 27

9 they become progressively harder causing a high modulus mechanical mismatch between the plaque filled region and the surrounding vessel. This can lead to both rupture of the blood vessel and excessive smooth muscle cell growth that further occludes the blood flow. In addition to restricted blood flow, the plaque and fatty deposit surface may be thrombogenic producing clot formation and occlusion. In some situations, these clots may break free and form emboli in the blood stream resulting in subsequent stroke or heart attack. When blood vessels do become occluded, surgical intervention is routinely needed to reopen them. In coronary artery disease, the coronary artery becomes blocked preventing the heart itself from receiving adequate blood supply (figure 2-3C). Endovascular techniques use a catheter that enters the body through the femoral artery at a groin incision and snakes through the body. Typically angioplasty is performed where the restricted section of the artery is expanded using a balloon. Often, a stent is also placed in the blocked area serving as a cylindrical metal cage that physically holds the vessel open. Though initially successful, reocclusion is common due to a relapse of the initial disease state. In the stented area the endothelium is damaged or missing due to the original atherosclerosis or the physical trauma caused by the balloon angioplasty and stent placement. The smooth muscle cells typically begin to grow thickening the neointima until occlusion occurs. This process known as intimal hyperplasia is due to a missing endothelium and the compliance mismatch caused by the metal stent. Intimal hyperplasia also occurs at the anastomosis between natural vessels due to the stiff sutures and at the junctions between natural vessel and artificial vascular grafts. The best solution to a blocked artery is to attach another vessel that circumvents the damaged area

PAGE 28

10 (figure 2-3D). This is done using the internal mammary artery or removing sections of the saphenous vein from the leg and using it to run from the aorta to regions of the coronary artery distal to the blockage. Unfortunately, many people do not have good veins or have already used them in a previous bypass procedure. While artificial vascular grafts made of polyethylene terephthalate (PET) and expanded-polytetrafluoroethylene (ePTFE) work well at large diameters for abdominal aortic aneurysm and similar applications, they fail at smaller diameters. This is due to a highly thrombogenic surface that passivates through formation of an adherent clot layer. At larger diameters, this is not an issue because the high shear flow rates prevent excessive thickening. However, smaller vessels have lower shear rates and, combined with a failure to re-endothelialize, leads to occlusion from clotting and intimal hyperplasia. Clearly, an artificial graft that remains patent at small diameters of 6 mm in diameter is needed. In addition to coronary artery bypass grafts, small diameter artificial vascular grafts are need for below the knee surgery to prevent the loss of limbs. In the initial disease and subsequent surgical interventions, gaps in the endothelial or its failure to regenerate leads to vessel occlusion. The need to elicit normal ECs function on natural and artificial vascular grafts and stents is obvious but difficult to achieve. That is the goal of this research and, to understand why, it is necessary to look at how and why current graft designs and surface engineering techniques fail.

PAGE 29

11 B A Figure 2-3. Examples of cardiovascular disease resulting from dysfunction of the endothelium. (A) A normal artery. (B) An artery that is narrowed due tofatty deposits. (C) Example of a blockage in a coronary artery. (D) Example of mammary artery and saphenous vein being used as grafts in coronary artery bypass graft surgery. Images from: A and B the Internet Pathology Laboratory for Medical Education: Florida State University College of Medicine, C the www.merck.com/pubs/mmanual_home/illus/i27_1.gif website and D the http://www.learnaboutbypass.com/images/bypass.gif website. D C Vascular Grafts – Natural, Synthetic and Tissue Engineered Replacements In cardiovascular disease, the frequent treatment for many conditions is the resection of the damaged tissue and replacement with an artificial vascular graft. These grafts are made primarily from PET (Dacron) or ePTFE (Teflon) tubes ranging in size from 8 to 30 mm in diameter. Artificial grafts have proven to be successful at diameters larger than 6 mm in diameter and remain patent for upwards of 10 years. Unfortunately,

PAGE 30

12 as the vessel diameter decreases in size the rate of complications increases. The problems extend beyond known surgical risks such as infection and excessive bleeding. An example is the seal between the native vessel and the replacement artificial graft. The anastomosis in artificial graft placement is characterized by a significant disruption in the mechanics of the vessel wall, the blood flow dynamics and the native biological environment. Analysis of Current Vascular Grafts Artificial vascular grafts have had a long and successful history as viable replacements for diseased and damaged blood vessels. The two materials that currently have FDA approval are Dacron and Teflon woven and knitted grafts for aortic, pulmonary and other large caliber vessels. While these grafts work well at large diameters, they have extremely poor patency at or below 6 mm in diameter. There are a multitude of underlying reasons for the high rate of occlusion, but the main cause is neointimal hyperplasia, or excessive growth of smooth muscle cells (SMCs). As discussed, mechanical mismatch between natural and artificial vessels creates a region of high stress concentration that elicits an adaptive response by the body. The SMCs at the anastomoses respond by growing and thickening the vessel wall, and at small diameters blood flow is stopped. In addition to the mechanical stress, SMCs grow due to poor re-endothelialization of the lumen, and a lack of chemical cues that maintain quiescent SMCs. From these facts, it is proposed that a functional small diameter vascular graft will be able to prevent neointimal hyperplasia and promote re-endothelialization of the artificial graft lumen as an anti-thrombogenic surface. This design goal has been recognized by many researchers and a number of small diameter vascular grafts have been fabricated and tested both in vitro and in vivo. The

PAGE 31

13 artificial vascular grafts fall into different categories, and for some the term artificial does not mean synthetic. A number of thorough review papers have evaluated the different artificial grafts approaches, and there are many strengths and weaknesses to build upon [4-6]. Clinically evaluated grafts for medical procedures such as coronary artery bypass grafts (CABG) are either biological or synthetic in origin. Table 2-1 lists the main types of grafts used, their origin and the 5-year patency rates. As mentioned, the synthetic ePTFE and PET grafts have low patency at small diameters with intimal hyperplasia, thrombosis and suture aneurysm being the major complications. Biological grafts are much more successful at small diameters and are harvested from different sources. Xenografts typically come from cow arteries, but suffer from degeneration and calcification when placed in humans. Furthermore, there exists a risk of disease transmission between species (the possibility of which has been demonstrated with mad cow’s disease). Allografts are typically harvested from cadavers and organ donors and include both vein and artery grafts. Due to cryopreservation methods and host immune response, these grafts also fail due to degeneration, calcification and poor healing. Autografts are currently the best grafts and are widely used for CABG and other small diameter graft procedures. Because they are from the host and are used immediately after harvesting, these grafts due not suffer from degeneration, calcification or long-term chronic immune response. Arterial grafts show a ~95% 5-year patency thus representing the design goal of any artificial vascular graft. Due to limited supply of artery, vein grafts are more commonly used but suffer from intimal hyperplasia due to compliance mismatch and break down from an inability to remodel to the higher pressure when

PAGE 32

14 placed in arterial circulation. The main problem with autografts is the limited supply, and this is the reason an artificial graft is needed for patients that can not provide arterial or venous autografts. Table 2-1. Types of small caliber vascular grafts and their performance (adapted from Teebken and Haverich) [4]. Autograft Vein Autograft Artery Allograft Different Sources Xenograft Different Sources Prosthesis Plastic Examples Great saphenous vein, arm veins, popliteal veins, superficial femoral vein Internal and external iliac artery, superficial femoral artery, internal thoracic artery Artery, great saphenous vein, umbilical vein, cryopreserved vein segments Bovine carotid/internal thoracic artery Dacron and ePTFE Availability Limited; diameter <1 to >6 mm Very limited; diameter <1 to >8 mm Limited/good; diameter 4-6 mm Good; diameter 4-8 mm Very good; diameter 6 to >30 mm Long-term results Complete healing, degeneration and aneurysms are rare, intimal hyperplasia Very good No complete healing, degenerative disease; calcifications; transmission of disease Degeneration, calcifications, intimal hyperplasia; transmission of disease Intimal hyperplasia, thrombosis, obstruction, suture aneurysms 5-year Patency ~75% (great saphenous vein); ~65% (arm veins) ~95% ~60% (umbilical vein) ~59% (bovine xenografts) ~40% (ePTFE); ~80% (EC seeded) ePTFE) Compliance Mismatch Favorable Favorable Less favorable Less favorable Unfavorable Rather than using the limited supply of autologous vein and artery, many researchers have investigated the use of other tissues and cells for rebuilding blood vessels [4]. Harvested biological materials include pericardium, cutis, fascia, peritoneum, small intestine, ureter, diaphragm, subcutis, cutis and various allogenic and xenogenic decellularized arteries. While these grafts have seen varying degrees of

PAGE 33

15 success, they all fail due to poor adaptation to the mechanical stress of arterial circulation and poor healing. Instead of adapting mature tissues for use as vascular grafts, cells can be used to buildup the grafts in a process termed tissue engineering. There is a lot of potential to this process and it has already achieved impressive results in laboratory settings. A tissue engineered blood vessel (TEBV) uses in vitro cell culture techniques to build up an intima composed of ECs, a media composed of smooth muscle cells and an adventitia composed of fibroblasts. Different bioreactor designs build the layers separately and then combine them or construct all three layers simultaneously [7-9]. These TEBVs have equivalent mechanical properties to natural artery because the media and adventitia produce collagen and elastin. Also, they are non-thrombogenic because of the well formed EC layer that produces characteristic markers such as von Willebrand factor and anti-thrombin. Despite these excellent properties, these TEBVs are disadvantageous because of their biological nature. The only way to prevent immunogenic response is to use autologous cells, and use of autologous cells requires harvesting, in vitro expansion and culture in bioreactors for at least 2 to 3 months [7, 9]. Thus, TEBVs are not appropriate for acute applications reinforcing the need for a suitable small diameter artificial vascular graft with no biological antigens to trigger the immune system. The Mechanics of the Cardiovascular System and Vascular Grafts In the context of the cardiovascular system, one needs to address both the mechanics of the vasculature and the fluid dynamics of the blood to ensure proper biological function. Therefore, when engineering a device it is necessary to understand the forces that will be encountered and to select the appropriate materials to withstand those stresses. All FDA approved artificial grafts are based on PET or ePTFE, both of

PAGE 34

16 which are glassy polymers at ambient and physiologic temperatures. As such, they due not match the viscoelastic mechanical properties of natural tissue. This is thought to be one reason that artificial graft patency is severely reduced at smaller diameters. Conversely, autologous vein grafts demonstrate excellent patency even at very small diameters. Vein grafts however still fail due to poor remodeling to higher arterial pressures overtime leading to arteriosclerotic plaques and eventually occlusion. If an artificial graft can match the patency rates of vein grafts, then it will be deemed an incredible success, if it can improve on the patency rates even further then it will be a medical breakthrough. To that end, it is critical that a thorough understanding of mechanics of the cardiovascular system, and specifically small diameter arteries and blood flow through them, be achieved. Current artificial vascular grafts are composed of a woven or knit fabric in the case of Dacron or a continuous, porous tube in the case of Teflon. The materials are in a glassy polymer state with a relatively high elastic moduli of ~1 GPa. Artificial grafts have minimal compliance and are the main contributing factor to the drastic change in the vessel wall mechanical properties. In addition, the sutures connecting the artificial graft to the natural vessel also contribute to the compliance mismatch and act as stress concentration points at the anastomotic line. The change in mechanical properties at the anastomosis is thought to be one cause of intimal hyperplasia. The mechanical properties of the natural artery or vein are quite different from that of the Dacron or Teflon found in artificial grafts. There are differences in mechanical properties between veins and arteries as well. Arterial pressure is higher and therefore the elastic modulus is

PAGE 35

17 higher than for veins. This translates to a thicker and stronger media in arteries. This structure is quite different from that found in artificial grafts. B A Figure 2-4. Computer models showing the stress concentration due to sutures at (A) end-to-side and (B) end-to-end artificial graft to host artery anastomoses [10]. The stress concentrations caused by the sutures can be seen in figure 2-4, which is a mathematical model of an artificial graft to host artery (A) end-to-side anastomosis and (B) end-to-end anastomosis [10]. It is believed that the smooth muscle cells react to these concentrated stresses and proliferate as some sort of biological compensation. At larger vessel diameters, the high shear rate of the flowing blood prevents excessive thickening and vessel occlusion. However, at smaller diameters the smooth muscle cells continue to grow causing intimal hyperplasia. The compliance mismatch is evident even when no artificial graft is used. In total natural vessel anastomoses, the compliance mismatch due to the sutures alone is significant. Figure 2-5 shows graphs of (left) the compliance coefficient and (right) the elastic modulus along the vessel axis moving from the proximal to distal side of the anastomosis. From these graphs, it is clear that a significant discontinuity exists at the anastomotic line, and is believed to contribute to intimal hyperplasia [11].

PAGE 36

18 B A Figure 2-5. Graphs showing the change in compliance in an artery-to-artery anastomosis due solely to the applied sutures [11]. Microscale Patterning of Topography and Chemistry on Surfaces Directing cell growth and function is critical for the successful development of tissue engineered scaffolds for both in vivo and in vitro tissue and organ repair. Similarly, the coatings and surface treatments of implantable medical devices must be incorporated seamlessly into the host or they will be rejected by an acute or chronic inflammatory immune response. To address this issue, engineering the biointerface between cells and tissues and synthetic materials has seen a rapid growth in research during the past decade. The current paradigm seeks to engineer the biomaterial to incorporate seamlessly into the physicochemical properties of the host tissue and to meet the metabolic requirements of specific cell types. This includes controlling the topography, chemistry, surface energy and mechanical properties of the biomaterial surface. The ability to add topography and chemistry to surfaces has expanded in the last decade. Using photolithographic methods adapted from the microelectronics industry, both chemistry and topography can be added to polymer surfaces at the micrometer and nanometer scale in complex spatial arrangements. For example, topography is added to

PAGE 37

19 polydimethylsiloxane elastomer (PDMSe) by photolithographically patterning a silicon wafer and curing the polymer against it [12, 13]. This technique may be used with other polymers directly or using the PDMSe negative as a mold. These PDMSe molds have been pioneered by Whitesides et al as microfluidic devices for pattering surface using protein solutions [14-23]. The microfluidic devices can be used for patterning, precise mixing of biologic fluids, cell culture or any combination there of. Similar to the microfluidic process is using patterned PDMSe substrates to stamp other surfaces using micro-contact printing [24, 25]. Whitesides has also introduced a number of microlens array techniques that allow the rapid prototyping of new microtopography and microfluidic system designs [26]. Another approach that looks promising is the adaptation of standard inkjet printers to use protein solutions instead of standard inks [27-29]. Careful tuning of solution properties allows a simple CAD program to design and then print complex protein patterns on polymer sheets. These techniques combine to allow a wide array of methods for topographical and chemical surface patterning, all of which can be done inexpensively in a laboratory setting. The study of defined patterns of topography have stemmed from observations that roughness of a biomaterial surfaces influences cell and tissue response. Cells also been shown to move along well defined topographical features, and drastically change their morphology in response to this texture [30-35]. Clark et al found cell response to a single 5 m step that groove depth increased cell alignment substrata of varying widths and depths (4 to 24 m repeat width, 0.2 to 1.9 m depth) [36, 37]. Astroglial cells showed preferential adhesion to pillars and wells on the scale of 0.5 to 2.0 m in width and 1.0 m in height [38]. Walboomers and Jansen used rat dermal fibroblasts on

PAGE 38

20 polystyrene, polylactic acid, silicone, and titanium coated polystyrene substrates to show that the microtextures influence cell guidance, while surface chemistry influences morphology [34]. Porcine vascular ECs can be polarized by 5 mm wide, 5 mm tall ridges (figure 2-6 parts A and B) as well as and 5 mm pillars can be used to decrease cell spreading (figure 2-6B) [39]. From these studies it is clear that the height, width and spacing of micrometer scale topographic features greatly affect cell growth and function. Figure 2-6. Optical microscopy of stained PVECs on a textured RFGD treated siloxane elastomer substrate. (A) smooth and 5mm high, 1000 mm long textured ridges (10X), (B) 5 mm high textured pillars (10X), (C) smooth and 5mm high, 60 mm long textured ridges (10X) [39]. In addition to topography, surface chemistry appears to play a large role in the field of contact guidance for controlling cell proliferation and metabolism. Cells will exhibit a preference to hydrophilic areas of patterned cellulose acetate and palladium metal surfaces [40]. Similarly, patterns of alternating hydrophobic dimethyl-trichlorosilane groups with aminosilane groups on glass slides cause cells to preferentially adhere to the hydrophilic areas [41, 42]. Biomedical polymers are modified with gas plasmas to increase hydrophilicity or to graft specific proteins and polymers to the surface [43, 44]. The size of chemical patterns can also influence cell death with a dependence of pattern area on the percentage of apoptosis in bovine and human ECs (figure 2-7) [45]. In

PAGE 39

21 addition, fibronectin patterns can be used to direct the formation of focal adhesions to surfaces using techniques similar to that employed in the research to be presented (figure 2-8) [46]. Figure 2-7. Effect of spreading on cell growth and apoptosis in bovine (A) and human (B) endothelial cells. Percentage of cells undergoing apoptosis or DNA synthesis were plotted as a function of projected cell area. Apoptosis was measured 24 h after plating. DNA synthesis detected by the incorporation of BrdU either over the first 24 h after plating (A) or between 20 and 24 h after plating, during S-phase (B) [45]. Atomic Force Microscopy and Its Application to Biological Systems The traditional role of the AFM has been to measure the topography of surfaces with near atomic resolution. A standard AFM setup is shown in figure 2-9. A wide variety of materials have been imaged including metals, ceramics, polymers and biological systems. Depending on the material and microscope setup, morphology, grain structure, phase structure, electronic structure, magnetic structure, conductivity and mechanical properties may be determined on the nanometer and micrometer scale [47]. There are a number of imaging modalities that can be used depending on material type,

PAGE 40

22 Figure 2-8. Focal adhesion formation increases as cell spreading is promoted. (A) Diagrams of adhesive patterns (left), phase contrast images of cells (middle), and immunofluorescence images of vinculin in a series of micropatterned substrates where single cells can spread across multiple, small adhesive islands. The bottom row shows a cell on an unpatterned FN substrate. Scale bars indicate 10 m in length. (B) Quantitation of total contact area of FN per cell (ECM per cell), projected cell area (cell spreading), and amount of vinculin per cell on three different micropatterned substrates containing 3, 5, or 10 m circular islands. (C) Quantitation of total vinculin and phosphotyrosine per cell plotted against projected area of cells for cells cultured on many different micropatterned geometries, including squares, circles, lines, and unpatterned substrates [46]. but the two most common are tapping mode and contact mode. Contact mode imaging applies constant force to the surface and is in continuous contact. Tapping mode imaging vibrates the cantilever and thus the tip makes intermittent contact with the surface at the maximum amplitude of each cycle. There are advantages and disadvantages to each

PAGE 41

23 method that depend on the material being imaged, the tip material and geometry and the environment. For example, to actually achieve atomic resolution requires using contact mode in a vacuum with an atomically sharp tip on a hard, relatively flat, sample. When evaluating biomaterials, these criteria are impossible to meet and resolution is on the nanometer rather than Angstrom scale. The real advantage of the AFM is its ability to operate in diverse environments. The best example of this is imaging in liquid environments for in vitro and in situ examination of biological systems. There is also another type of imaging called force volume that is used to get topographic data that uses a 2-dimensional map of force curves to reconstruct the surface. Each cantilever has a specific stiffness that requires a given force to bend it a set distance. This relation is expressed in the familiar Hook’s Law, F = kx, where the force, F, is equal to the displacement, x, times a proportionality constant, k. The constant k is referred to as the spring constant. Typically, when using the AFM, distances are expressed in nm, forces in nN and the spring constant in N/m. The standard AFM contact mode tip substrate model D-NP (Digital Instruments, Veeco Metrology Group) in figure 2-10 is micromachined from silicon nitride with a layer of gold deposited on the backside of the cantilever to improve laser beam reflection. There are four cantilevers on the substrate, each with a different overall length and leg width with a corresponding spring constant. It is important to use a cantilever with a spring constant that is sensitive enough to measure the surface in question. AFM cantilevers are available with spring constants ranging from less than 0.01 N/m to as high as 100 N/m. This wide range allows imaging of nearly any type of solid material in both tapping and contact modes. For low T g polymers and cells, the relatively low modulus and softness requires a minimum of forces

PAGE 42

24 be used to prevent damage to the surface. Selecting the appropriate cantilever geometry and spring constant is based on both the imaging modality to be used and the anticipated modulus of the material. Figure 2-9. Schematic of the Dimension 3100 AFM used for all analyses. A small probe with nanometer radius of curvature is controlled by XYZ piezoelectrics and brought dawn to the surface with a stepper motor. Deflection of the cantilever is measured by reflecting a laserbeam off it to a photodetector. A feedback loop controls the piezoelectrics to maintain contact with surface at a specified contact force. A video camera is used to align the laser beam on the cantilever and to define where the AFM tip will contact the experimental surface. 4 QuadrantPhoto DetectorVideo Camera forTip Alignment andSample ViewingLED Based LaserX-Y Translation Stage –SoftwareControlled Stepper MotorsSampleAFM Control Headwith Z Stepper Motorand PiezoelectricScannerX-Y PiezoelectricScannerLaser BeamZ PiezoelectricScannerAFM tip Force measurements for elastic modulus are almost always performed using silicon nitride (Si 3 N 4 ) contact mode tips because of the variety of available spring constants. Tapping mode tips such as model TESP, (Digital Instruments, Veeco Metrology Group) have a tip radius of 5-10 nm and a high cantilever spring constant of 20-100 N/m. The

PAGE 43

25 force from this tip would easily penetrate soft polymeric samples or be blunted by harder samples if used in a contact mode. Both results are unacceptable because plastic deformation of either the tip or sample will cause a change in contact area and give erroneous data. The Si 3 N 4 probes have a tip radius of 40 nm and lower spring constant that has been shown to work well with polymers such as PDMSe and PS. Figure 2-10. Schematic of a silicon nitride substrate used for contact and force volume mode AFM. Each side has 0.6 m thick, 100 and 200 m long V-shaped cantilevers on it, with the wider leg and higher spring constant cantilevers on the bottom (not drawn to scale) (adapted from [1]). AFM Force Curves The AFM is used to measure elasticity and adhesion using force curves. Rather than rastering the AFM probe across the surface in an XY plane, the tip is moved only in the Z-axis. The tip is moved from a non-contact region above the surface down into the sample until a set tip deflection or depth is reached and then retraces its motion. All measurements are recorded as voltages and then converted to more useful units, such as nanometers using conversion factors. Therefore, the force curve is really a graph comparing the voltages obtained from the scanner z-axis position vs. the tip deflection. The horizontal axis is the voltage applied to the z piezoelectric as the tip-substrate is moved towards the sample. The vertical axis is the voltage read from the photo detector 0.12 N/m 0.06 N/m 0.68 N/m 0.32 N/m

PAGE 44

26 array recording the position of the laser beam bouncing off the cantilever as the tip makes contact with the sample and the cantilever bends. Force Curve Anatomy It is necessary to u nderstand the anatomy of a standard force curve obtained with AFM drawi before a more detailed analysis can be done. Figure 2-11 shows a force curve obtained on a PDMSe sample showing the common components of the graph. The force curve records the extension and retraction, more commonly referred to as the trace and retrace, of the tip starting from a non-contact position above the surface. The tip extends down into the surface at a defined speed and then retracts after either a set distance or tip deflection is reached. The force curves produced by the Nanoscope IIIa version 5 release 12 software use white to designate the trace and yellow for the retrace as a convention. To illustrate the tip/sample interactions during the force measurements, each ng in figure 2-12 illustrates the forces acting on the tip and its resulting deflection. As the tip begins approaching the surface there are no forces acting on it and the cantilever beam has no deflection. The reflected laser beam is centered in the photo detector array. As the tip gets closer to the surface, forces interacting between the tip and the sample become great enough to over come the spring force of the cantilever and the tip suddenly jumps into contact with the surface seen in Figure 2-12B. This causes the sudden dip in trace curve seen at point b of figure 2-11. These attractive forces include Vander walls, electrostatic and capillary in air. As the z piezoelectric continues to move the tip substrate towards the sample, the cantilever bend changes from concave down, to flat and then to concave up as seen in Figure 2-12B to 2-12D. This change in cantilever bend is seen as the change from a positive slope in the force curve at point b to a negative slope in the force curve at point c of figure 2-11. The area with negative slope is referred

PAGE 45

27 to as the contact region and involves both cantilever deflection measured by the photo detector and sample deformation (Figure 2-12D). It is the slope of this region from which elastic modulus values are extracted. The z piezoelectric continues to move down until specified distance is traveled or the tip deflection reaches a set value and then reverses direction, i.e., retraces. The retrace is similar to the trace until the tip begins to separate from the surface. As seen in Figure 2-12E, the tip holds on to the surface even as the z piezoelectric continues to move away due to adhesive forces. This is well illustrated at point e in Figure 2-11 where the large adhesive force seen for the PDMSe extends well below the zero deflection position. Eventually the spring force of the cantilever exceeds the adhesive force between tip and sample and the tip breaks free and returns to its resting deflection as seen in 2-12F and at point f, Figure 2-11. Figure 2-11. A force curve as output by the Nanoscope version 5 release 12 s a b d c e f oftware. The white line is the extending (trace) movement and the yellow line is the a) retracting (retrace) movement. The horizontal axis is the z piezoelectric position in nm and the vertical axis is tip deflection in nm. The numbered areas correspond to the schematic drawings in figure 2-12 and represent (noncontact, (b) attractive jump-on, (c) extending linear contact region, (d) retracting linear contact region, (e) adhesive pull-off and (f) noncontact regions of the force curve. Force curve from a PDMSe sample (adapted from[1]).

PAGE 46

28 Figure 2-12. Schematic drawing representing the tip and sample interaction during trace and retrace seen in the force curve in Figure 2-11. In each picture the tip position and cantilever deflection are seen as well as the laser beam reflecting off the back of the tip to the photodetector array. The arrow in each picture represents the direction the z piezoelectric is moving. The interactions arenoncontact, (B) attractive jump-on, (C) extending linear contact region, (D) retracting linear contact region, (E) adhesive pull-off and (F) non-contact regions of the force curve (adapted from [1]). (A) Photo Detector Sample Cantilever Photo Detector Sample Cantilever Photo Detector Laser Sample Cantilever Photo Detector Laser Sample Cantilever Photo Detector Laser Sample Cantilever Photo Detector Laser Sample Cantilever Laser Laser a) b) c) d) e) f)

PAGE 47

29 InterpretinThe ce and distance are the force pical force curves exhibit non-s2-13. When the AFM tip touches the surface a receptor-ligand g Specific Interactions from AFM Force Curves basis for detecting the specific interaction for curves measured by the AFM between the tip and sample. Ty pecific interactions characterized by a single adhesion upon retraction from the surface. The cause of this non-specific adhesion is a combination of electrostatic and van der Waals interactions, capillary forces and molecular entanglements. Specific interactions are due to chemical bonding between the tip and sample and these bonds must be ruptured in order to retract from the surface. Proper functionalization of the tip and surface allow the mechanics and binding between single molecules to be examined and has been demonstrated for DNA strands, Avidin to Biotin, antigen to antibody and PSGL-1 to P-selectin interactions [48-61]. These papers provide the basic methodology for interpreting the adhesion events seen in the retrace portion of the force curve and analyzing them appropriately. The basis of the adhesion events is the uncoiling and rupture of single molecules interactions as shown in Figure bind forms and is compressed on the surface as the proteins adopt a globular form. As the tip retracts a stress is applied to the complex and the protein chains begin to unravel through low-energy rotations about the carbon backbone and scission of intra-chain bonds. When the chains are fully unraveled the high-energy covalent bonds between backbone carbon atoms is stressed causing the weak point, the specific receptor-ligand bond formed through hydrogen bonding, to fail. This bond rupture occurs for each receptor-ligand pair and a single or multiple bonds may fail at one time depending on the interaction geometry. These force measurements are replicated hundreds of times across

PAGE 48

30 the cell surface and then statistically analyzed to find the force and distance of the interaction. Figure 2-13. Illustration of a receptor-ligand pair as it stretches and ruptures. The receptor-ligand pair begins in a compressed state with the protein in a globular he CReceptor-Ligand is CompressedMolecules Unravel C-C-C Backbone is StretchedBond Rupture Receptor-Ligand is CompressedMolecules Unravel C-C-C Backbone is StretchedBond Rupture form. As the pair is stretched the protein on ravels until it is taught and tC-C backbone is stretched. Upon reaching a critical force, the specific interaction ruptures and the extended proteins collapse back down to a globular form (adapted from [1]).

PAGE 49

CHAPTER 3 ULTRASTRUCTURAL ANALYSIS OF PORCINE PULMONARY ARTERY Introduction To engineer a monolayer of endothelial cells as a blood contacting surface in artificial vascular grafts it is imperative to understand the properties of the endothelial cell lumen in vivo. The properties that are important constitute the complete metabolic and structural makeup of the cells. Morphology, mechanical properties, contact inhibition and expression of specific cell surface markers are used as gross indicators of endothelial cell function. These properties are also important in understanding a wide array of vascular related pathologies including arteriosclerosis, aneurysm, emboli and occlusions. Materials and Methods Harvesting Porcine Pulmonary Artery Porcine pulmonary artery was harvested and prepared for analysis using a combination of analytical techniques. The protocol is based on that used by Dr. Block’s (VA Hospital, Gainesville, FL) group for harvesting of primary porcine pulmonary vascular endothelial cells (personal communication). Adult pigs were slaughtered by Mr. Sanford in Trenton, FL. The pigs were sacrificed and then slaughtered for use in the meat industry. When the heart and surrounding tissue was removed from the pig, typically 30-40 minutes after the pig was killed, the required tissue was excised. As illustrated in figure 3-1, the pulmonary artery was identified and cut away from the heart and surrounding connective tissue. The inflow direction of the artery was marked on the 31

PAGE 50

32 outside of the vessel using a permanent marker. The artery section was then repeatedly washed with phosphate buffered saline solution (PBS) to remove any blood and then stored in formalin (10% formaldehyde in PBS), PBS or RPMI 1640 media (Sigma). Once stored in the appropriate solution, the samples were transported back to the University of Florida in a cooler on ice. Figure 3-1. This is an example of a porcine heart where the pulmonary artery is labeled as the pulmonary trunk. The pulmonary artery can be seen originating from the right ventricle and would terminate at the lungs if they were still attached. ( http://science.tjc.edu/images/fresh_heart_specimen/ant-labeled.jpg ) Smaller tissue sections were cut from the resected artery for AFM and SEM analysis. To look at the vessel lumen, longitudinal strips were cut from the vessel wall using a scalpel and a pair of tweezers to stabilize the tissue during cutting (Figure 3-2). The strips were ~2 cm long and 0.5 to 1 cm wide with long axis parallel to the length of the artery. The inflow direction was marked with a small spot of indelible ink on the

PAGE 51

33 adventitia side. To look at the composition of the arterial wall, cross-sectional rings ~3 mm wide were cut with a scalpel and glued down to SEM stubs. Glue on SEM stub or in Petri dish for AFM Mark inflow direction with indelible ink Cut longitudinal strip from artery Figure 3-2. A schematic of the way strips of tissue are cut from the artery, marked for blood flow direction and glued to a substrate for AFM or SEM analysis. Scanning Electron Microscopy of Porcine Artery Tissue samples were directly analyzed with SEM. Fixed samples were trimmed to the desired shape and dehydrated using an ethanol series. In detail, samples were removed from the fixative and placed in PBS, 70% ethanol and 95% ethanol for 15 minutes each. Samples were then placed in 100% ethanol overnight and dried in desiccator at 37C for 4 hours. After drying samples were mounted on SEM stubs, coated with Au/Pd for 3 minutes and imaged on a Jeol 6400 SEM equipped with an Oxford ISIS image capture system. Fluorescent Microscopy Fixed sections of porcine pulmonary artery were stained in order to visualize the morphology of the endothelial cell layer. Sections approximately 5x5 mm in size were

PAGE 52

34 super glued in the center of a 35 mm diameter Petri dish with the lumen facing up. The Petri dishes were then filled with PBS and the cells were stained. . For epifluorescent microscopy, cells were stained in a PBS solution of 4',6-Diamidino-2-phenylindole (DAPI) at a concentration of 330:1 to visualize the cell nuclei. Concurrently, to visualize the actin cytoskeleton cell were also stained with phalloidin-TRITC at a concentration of 5 M for 12 hours in an incubator at 37C. Both types of fluorescent stains were imaged on a Zeiss Axioplan 2 microscope using the appropriate epifluorescent filters. Images were acquired using an Axiocam 11 megapixel digital camera (Zeiss) at a resolution of 3900x3090 pixels using Axiovision 3.1 software. Optical Image Processing Images were post processed using ImageJ v1.31 (NIH) to enhance brightness/contrast and count cells. ImageJ is an open source software product sponsored by the NIH and written in Java so it will run on Windows, Macintosh and Linux platforms; it is available for download at http://rsb.info.nih.gov/ij/ . Microscope images captured at 3900x3090 pixels were opened in ImageJ and contrast/brightness enhanced to make the nuclei as clearly visible as possible. Due to the curved surface of the arterial wall and the microscopes depth of field, the entire image was not focused. The center of the image was always adjusted to be in focus and the surrounding out of focus areas were cropped out. Within this cropped area, 5 to 8 separate regions of interest (ROI) were examined in order to best view the cell nuclei. Each ROI was further enhanced by adjusting the brightness and contrast and using the sharpen filter to accentuate the edges of the nuclei. Nuclei were counted using the cell counter plugin for ImageJ by manually clicking on each nucleus in the ROI and recording the number of nuclei. The average

PAGE 53

35 area occupied by each cell was calculated by dividing the total area of the ROI by the number of nuclei counted in the same region (equation 3-1). This process is summarized in figure 3-3 where the images as processed in ImageJ are displayed including the cell counter and results windows. Equation 3-1 imageareacellareanumberofnuclei Figure 3-3. An example of the methodology used to examine fluorescent images of the stained cell nuclei in fixed porcine artery. The well illuminated area is cropped and then small regions of interest (ROI) are examined. Each ROI is contrast and brightness enhanced, filtered to enhance nuclei edges and then the nuclei are hand counted. ImageJ tracks each tagged nucleus providing total number upon completion. Atomic Force Microscopy of Porcine Artery All scanning was performed on a Dimension 3100 AFM (Veeco Metrology) equipped with a Nanoscope IIIa controller and version 5.12r3 software. Imaging was performed in Hanks balanced salt solution (HBSS) using a liquid tip holder. For all imaging modes, 200 m long and thin silicon nitride cantilevers (Veeco Metrology) with

PAGE 54

36 0.06 N/m spring constant were used. Contact mode was used to determine surface topography and force volume mode was used to map the mechanical properties. Mounting the tissue samples Porcine pulmonary artery was analyzed in small sections by atomic force microscopy (AFM) to determine the morphology and mechanical properties of the lumen. Pieces were cut from the artery wall with an approximate area of 1 to 2 cm 2 and transferred to a tissue holder in a 35 mm diameter Petri dish. This modified Petri dish had a short metal wire in the bottom bent in to a ‘C’ shape and epoxied to the dish in the middle. The tissue samples were placed under the free ends of the wire and clipped into place along the edges leaving nearly the entire lumen of the artery wall exposed face-up in the middle of the Petri dish. The Petri dish was then filled with HBSS so the liquid extended ~2 mm above the lumen. The Petri dish was placed in the AFM stage and held in place using double-sided scotch tape. 17.5 mm B A Figure 3-4. Photographs of the AFM experimental setup used to analyze fresh and fixed sections of porcine pulmonary artery. (A) The Petri dish containing the section of artery is taped to the AFM stage and the scanner is then lowered directly into the fluid until the tissue is contacted. (B) A close-up of the 35 mm diameter Petri dish showing the section of artery flattened against the bottom and held in place using a metal clip.

PAGE 55

37 Examining topography in contact mode Examination of the porcine artery in contact mode AFM required a number of special imaging parameters. This was due to the unique problems associated with imaging a section of tissue versus cells grown and immobilized on a standardized substrate such as glass or polystyrene culture dishes. The luminal surface of the artery is a nondescript white and is not easily focused on using the AFM optical alignment system. There are two methods that may be used to focus the surface, which is needed to properly align the AFM tip before it can be engaged. One option is to focus on the edge of the tissue section where the discontinuity between the tissue and the surrounding liquid media can be resolved. The other option is to quickly drain the liquid from the Petri dish, use a black indelible ink marker to put a spot in a region of the surface that will not be examined, and then quickly rehydrate the sample. This procedure takes less than 15 seconds and the black spot made by the marker is easily focused on. Once the luminal surface is focused on, the AFM can be engaged in standard contact mode. Contact force between the AFM probe and artery was minimized by setting the deflection setpoint at, or 0.1 to 0.5 volts below, the noncontact deflection position. Scan rates are kept low to prevent damage to the tissue with maximum scan rates dependent on the chosen scan size (Table 3-1). The scan speed is given as Equation 3-1 scan speed = 2*(scan size)*(scan rate) and never exceeds 100 m/s. The main issue when examining tissue samples and other non-flat substrates by AFM is the maximum change in the Z-axis that occurs during an XY scan. While the lumen is quite flat at the scan sizes used in these experiments, it is not possible to ensure

PAGE 56

38 that the lumen surface is in a plane completely orthogonal to the AFM tip. As such, the lumen surface will have some tilt resulting from the natural curvature of the vessel and the way it sits in the Petri dish tissue mount. This tilt can be automatically corrected using standard flattening algorithms built in to the AFM analysis software. However, when this tilt exceeds the maximum Z-axis range of the AFM, the piezoelectrics max out and the AFM can no longer track the surface. The maximum Z-axis range of the Dimension 3100 AFM used in these experiments is 6.14 m, and this was constantly exceeded when imaging the porcine artery at scan sizes greater than 10 m. To compensate for this, the Z-axis stepper motor was used to move the entire scan head up or down as needed in 0.5 to 1.0 m increments. Using this method, it was possible to image at scan sizes up to 100 m, equivalent to 10,000 m 2 , but not without causing other problems. Using the Z-axis stepper motor to move the scanner caused a momentary loss of contact between the AFM tip and sample that lasted 1 to 2 seconds. As the tip re-establishes contact there is piezoelectric creep that occurs on the X, Y and Z piezoelectrics that results in a bad scan line in the images and 0.1 to 1 m shift in the image position. As a result, many of these images can not be optimally post-processed using flattening and plane-fit algorithms to remove sample tilt and curvature. In addition, using this method requires the tedious procedure of constantly watching the AFM scan throughout a 30 minute scan in order to move the Z-axis stepper motor as soon as the Z-axis piezoelectric is maxed out. Due to this limitation, future studies should be conducted on an AFM with a minimum Z-axis piezoelectric range of 10 m, though 20 m would be optimal.

PAGE 57

39 Table 3-1. Maximum scan rates and scan speeds for selected scan sizes when examining fresh porcine artery by contact mode AFM. Scan Size [m] Scan Rates [Hz] Resulting Scan Speed [m/s] 1 2 4 5 1 10 10 1 20 50 0.5 50 100 0.33 to 0.5 67 to 100 Examining topography and stiffness in force volume mode Force volume images were acquired by forming a 2 dimensional array of force curves on the sample surfaces. As such, examining each force curve allows the substrate properties at that point to be determined. The topographic image produced by force volume was lower resolution than that produced by contact mode. This is the result of time constraints, as a 64x64 data point force volume image takes 2 hours to complete while a 512x512 data point contact mode image takes a maximum of 30 minutes to complete. Cells could only be imaged for ~6 hours before they began to die from lack of nutrients. The topographic image produced by force volume is constructed as a surface plot drawn through the maximum tip deflection of each force curve. This value is varied slightly but is typically set at 100 nm. The force curve rate is set at 2 Hz, which allows the maximum imaging speed to be achieved with out introducing imaging artifacts due to hydrodynamic drag of the AFM cantilever in the liquid medium. Force curve scan size (Z-axis) ranges from 1.5 to 2.0 m depending on the particular samples. Force volume scan size (X and Y axis) ranges from 5 to 100 m depending on the specific location being examined and the desired level of detail. AFM data analysis Contact mode images were post processed using WSxM software (Nanotec Inc.). This freeware application opens a variety of SPM data formats and has enhanced 3

PAGE 58

40 dimensional rendering capability as compared to the Nanoscope v5r12 software supplied with the Dimension 3100 AFM. All the contact mode topography images were opened in WSxM and then flattened as needed. A 0 th order flatten was always performed normalizing the relative position of each scan line to its neighbors so that a continuous surface was formed. This is a standard practice as an arbitrary Z-axis displacement exists for each scan line and this must be normalized to recreate the 3-dimensional surface. In addition to the 0 th order flatten, a 1 st order flatten was applied to those images where an obvious tilt to the surface existed. This included all images where an adjustment of the Z-axis stepper motor was needed to compensate for Z-axis piezoelectric max out as well as samples that had tilts within the Z-axis piezoelectric range. For the porcine artery, any small bowing of the surface was not corrected for as it was interpreted as a real feature of the artery wall and not an imaging artifact. Force volume images were analyzed using two different processes. The topographic image produced by the force volume mode was analyzed using the WSxM software by the same procedure just described for the contact mode topographic images. Analysis of the individual force curves of the force volume image were performed using a combination of software. The Nanoscope v5r12 software (Veeco Metrology) was used to open the force volume dataset and save individual force curves as ASCII text files. These text files were then opened in the WSxM software as force curves and the slope of the contact region, non-contact deflection, maximum deflection and etc were recorded in an excel spreadsheet. From these values, the elastic modulus of the cell was calculated using the appropriate model in Excel (using methods described in appendix A).

PAGE 59

41 Results and Discussion SEM Analysis of Porcine Artery SEM analysis of the fixed artery sections verified that the structure of the experimental tissue agreed with published results [62]. Figure 3-5 shows a cross-section of the artery wall illustrating the three different tissue layers of the artery wall. The visisble layers include the tunica intima, tunic media and tunica adventitia. This is the anticipated structure of the blood vessel and simply serves to confirm that the pulmonary artery has the expected morphology. Lumen Tunica Intima Tunica Media Tunica Adventitia Figure 3-5. A SEM micrograph of a pulmonary artery cross-section. The three layers of the arterial wall, the tunica intima, tunica media and tunica adventitia, can clearly be identified as changes in tissue morphology. SEM of the lumen verifies the known ultrastructure of the endothelial cell monolayer in general. Shown in figure 3-6, each subpart demonstrates the monolayer nature of the endothelium as well as overall cell morphology. Figure 3-6A is a low angle

PAGE 60

42 image at ~20 to the surface and reveals how the ECs form a carpet-like surface structure with no apparent gaps. A close-up of this surface shown in figure 3-6B indicates that the ECs in this location of the artery are elongated. The twisted cell bodies and variable heights are due to the fixation procedure, which causes desiccation and ~25% tissue contracture. Top down images of the lumen show that EC morphology and alignment vary with location on the artery lumen. It is apparent that even over short distances of 100 to 200 m, ECs can change alignment from parallel to nearly 45 off axis to the flow direction of the blood (figure 3-6C). Zooming in on this section (figure 3-6D) shows the elliptical shape of the cells and the continuous packing. Figure 3-5E shows a section of artery lumen where the ECs are more randomly aligned and more round in morphology. Figure 3-6F is interesting because 4 leukocytes can be seen sitting on the surface of the endothelium. The presence of these leukocytes, which usually adhere to activated ECs, is indicative of the stress induced in the tissue between the time the pig was killed and the time the pulmonary artery was actually removed, cleaned and placed in fixative. Ideally, the tissue would be fixed at the time of death. This is referred to as perfusion fixation and involves administration of a fixative intravenously at the time of death to preserve tissues and prevent necrosis.

PAGE 61

43 Figure 3-6. SEM micrographs taken of the lumen of fixed sections of porcine pulmonary artery. (A) A low angle image of the lumen showing the continuous, carpet-like EC monolayer of the lumen. (B) A close-up of part A illustrating the close-packed ECs, but also the twisted distortion of the cell body thought to be due to the SEM fixation process. (C) a top-down view of the lumen showing ECs elongated in the flow direction. (D) A close-up of part C showing more detail of the cell bodies. (E) A region where 4 leukocytes are stuck to the ECs. (F) A region where the ECs have a rounded, rather than elongated, morphology. E F D C B A

PAGE 62

44 Optical Microscopy of Endothelial Cell Monolayer EM analysis because the fixation proce The fluorescent microscopy complimented the S dure did not involve dehydration in ethanol and exposure to vacuum. Figure 3-7 shows an example of the fluorescently stained tissue sample with the blue stained nuclei and red stained actin images overlaid on top of each other. From the image it is clear that the cells are closely packed together as seen in SEM. Some nuclei were blurred in appearance from the surface moving in and out of the focal plane due to the curvature of the vessel wall. The phalloidin-TRITC stain shows the presence of actin in a continuous network throughout the surface. This suggests that the ECs form a confluent monolayer with close tight junctions between the membranes adjacent cells. This corresponds to known characteristics for ECs in blood vessels where the cells are coupled together to form a semi-permeable membrane between the blood and the rest of the body. Quantitative analysis of cell density reveals 3,215 336 cells per mm 2 corresponding to a cell area 316 35 m 2 . This larger cell area, than seen with SEM analysis, but is more realistic since the fixation process does not result in noticeable tissue contracture as preparation for SEM analysis does. Both the SEM analysis presented and research by others demonstrates that ECs in some areas are elongated to the direction of flow. However, they are still closely packed and the calculated cell area is a good target for engineered EC monolayers.

PAGE 63

45 Figure 3-7. Fluorescent image of lumen of a fixed porcine pulmonary artery. The cell nuclei of the ECs are stained blue using DAPI and the actin cytoskeleton network is stained red using phalloidin-TRITC. (100x magnification) AFM of Fresh Porcine Pulmonary Artery Topography by contact mode AFM performed on living sections of artery provided information on cell morphology and mechanical properties without having to worry about any artifacts introduced by fixation, staining or coating for SEM. The contact mode images demonstrated that the endothelium could be imaged and that the surface was extremely smooth with even and continuous transitions between cells. Figure 3-8 shows a series of scans of different random areas on the surface of the artery. These AFM images are qualitatively similar to published observations of ECs in fresh rat aorta and after exposure

PAGE 64

46 to shear flow in vitro [63, 64]. In figure 3-8A, identification of the different parts of the cell in the AFM scan are shown. The approximate region of intersection between adjacent cells is taken as the minima in height between cells. The lack on any clear seam in the cell membranes prevents a more accurate determination. The 25 m scan also shows a single EC nucleus and the presence of stress fibers extending from the top of the nucleus down to the cell periphery can be seen. This is important because it indicates the arrangement of the actin stress fibers in the cell at a level of detail not possible by fluorescent microscopy using the actin stain (figure 3-7). Morphometric analysis of the AFM topographic scans provides detailed information about the size of the ECs. Cell width and height for each cell were calculated from the same cross-section taken at the peak of the nuclear area. Given the seamless integration between adjacent cells, the cell periphery was defined as the height minima between them. Because some cells appeared elongated in a one direction, widths were always taken across the narrowest part of the cell. The average EC width on the fresh porcine artery was 40 17 m with the large standard deviation due to the variation in cell shape from more round (figure 3-8D) to more elongated (figure 3-8B). Cell heights were more consistent at 2.53 0.86 m measured as the difference in height between the cell periphery and the peak of the nuclear area. Cell dimensions are likely to change based on the location that the ECs are from. For example, areas of the lumen subjected to higher shear rates have ECs with cell bodies elongated in the flow direction. While all the ECs are taken from the same region of the vessel, the shear environment that existed in the artery in vivo is not known for these samples.

PAGE 65

47 Cell body (cytoplasm) Nucleus Stress fibers Imaging artifact Approximate junction between adjacent cells D C B A Figure 3-8. Contact mode AFM topographic scans of fresh porcine pulmonary artery imaged in HBSS at ambient conditions. (A) 25 m scan size with stress fibers visible extending from the raised nucleus down to the cell perimeter. (B) Image of a 50 m scan size showing the main cell body of 2 ECs and parts of others. (C) Image of a 25 m scan size centered on a region between 3 ECs. (D) A large 87.2 m scan size showing 3 closely spaced ECs and a depressed area devoid of cell nuclei. Mechanical properties by force volume mode In addition to contact mode, force volume mode was used to produce a topographic image. Figure 3-9A reveals 3 raised areas corresponding to cell nuclei within a 50 m scan size demonstrating the close packing of the ECs. This is a topographic image constructed from the z-axis piezoelectric height corresponding to the maximum tip deflection for each force curve. In other words it is topography based on an

PAGE 66

48 iso-force surface through the force volume data set. The occasional bright spots in the image are due to adhesions between the AFM tip and the surface. This problem is illustrated clearly in figure 3-9B where many of these adhesions occur. Figure 3-9B is the same area as 3-9A, but captured as an array of 64x64 force curves rather than 32x32 force curves. This decreases the spacing between subsequent force curves and increases the imaging time 4-fold allowing experimental drift in the system to increase the applied force. As a result, fragments of membrane and other debris accumulate on the AFM tip and cause adhesions to form to the cells. This corrupts the force curve data, and to avoid this only 32x32 force curve scans were analyzed. 5.0m 5.0m B A Figure 3-9. An example of the adhesions that form between the AFM tip and cell membrane when the tip becomes contaminated with membrane and protein fragments after too much force is applied. (A) A topographic scan produced from a force volume dataset showing 3 raised nuclei in close proximity. (B) Another topographic scan produced from a force volume set with artifactual streaks in the image due to adhesion between the AFM tip and the ECs. Force volume images were interpreted by comparing the force curves obtained as a function of the region of the surface they came from. Figure 3-10 shows a 3-dimensional rendering of the surface shown in figure 3-9A with 4 regions of interest

PAGE 67

49 identified with dashed circles. Regions 1, 2 and 4 are raised portions of the surface that indicate cell nuclei are there. The force curves from each region were overlaid in single plots to help visualize differences between the mechanical properties. Features of the force curves correspond to specific mechanical properties. The top curves of each plot are the approach curves as the AFM tip indents into the surface while the bottom curves are the retract curve as the AFM tip pulls away. Nearly all the approach curves are linear indicating that the interaction is entirely elastic. The elastic modulus is based on the slope of the approach curve using the methods described in appendix A. The retract curve indicates that the cell is a viscoelastic material as the non-linearity in the slope is due to viscous loss. The hysteresis between the approach and retract curves taken as the difference in the area under the curves can give the energy lost to viscous flow [65]. Examination of the force curves from across the cells in figure 3-10 reveal important information about the variation in elastic modulus as a function of location on the lumen. Exact values for the elastic modulus computed as the average from the force curves in each identified region are listed for figures 3-10 and 3-11 in table 3-2. The elastic modulus for regions 1, 2 and 4 is ~20 kPa showing that the stiffness of all three nuclei is the same. Each of these regions is significantly different than region 3 (P<0.05) indicating the mechanical contribution of the cell nuclei and associated cytoskeletal components as compared to the rest of the cell body. While the difference between the regions is statistically different, the magnitude is only ~2 kPa suggesting that the surface of the lumen is relatively homogenous in terms of elastic modulus. This is consistent with the AFM images of topography (figure 3-9A) and the fluorescent image of the actin cytoskeleton (figure 3-8) showing a qualitatively smooth and continuous lumen surface.

PAGE 68

50 Looking at the force curves also shows a difference in the viscoelastic properties. Regions 1, 2 and 4 have retract curves that are approximately linear when compared to the retract curves from region 4. Also, there is a larger historesis in region 4 suggesting more fluid/cytoplasm is contributing to viscous loss. Figure 3-10. The topography of a section of artery reconstructed from the force volume image. Force curves from the circled regions were analyzed in order to determine the average elastic moduli. Corresponding plots from the force curves in each region are overlaid in the numbered graphs. Regions 1, 2 and 4 are taken on raised areas indicating nuclei and region 3 is a lower area of the cell body. There is a significant difference between the elastic modulus of regions 1, 2 and 4 compared to region 3 (P<0.05). 1 2 3 4 Region 1-300-250-200-150-100-5003.9651.59914719424228933738443247952757462267071776581286090751002Separation [nm]Deflection 95 Region 2-300-250-200-150-100-5003.9651.59914719424228933738443247952757462267071776581286090751002Separation [nm]Deflection 1 95 Region 3-300-250-200-150-100-5003.9651.59914719424228933738443247952757462267071776581286090751002Separation [nm]Deflection 2 95 Regiom 4-300-250-200-150-100-50011427405366799210511813114415717018319620922228Separation [nm]Deflection 3524 3 4

PAGE 69

51 Smaller areas of the lumen were examined in order to determine if more subtle variations in elastic modulus existed. Figure 3-11 show a 3-dimensional image of an area spanning between two nuclei with regions of interest correlating to stress fibers indicated by dashed circles. The goal was to see gradations in elastic modulus moving from the nuclei down to the cell periphery. Calculated values for elastic modulus are recorded in table 3-2 and show vary little change between the regions. There is a significant difference between the elastic modulus in regions 1 and 4 (P<0.05) and regions 2 and (P<0.05). This re-emphasizes the conclusion that the elastic properties of the lumen are relatively homogeneous as the variations in elastic modulus seem to be 5-10% at most. An increase in the viscous loss is still apparent and is most noticeable when comparing region 1 directly over a nucleus to region 4 at the cell periphery. An important observation is that the elastic modulus calculated from regions in figure 3-10 are approximately twice the values calculated for region in figure 3-11. These images were captured with the same cantilever so the variation is not due to the AFM probe. Rather the differences suggest variations in elastic modulus with macroscale changes in position on the lumen. Unfortunately, the exact location of these images in terms of the overall pulmonary artery was not recorded, but these images were captured in areas of the same artery ~10 mm apart. Future studies should examine this difference more closely to determine the magnitude of long-range variation of elastic modulus and how it compares to the flow characteristics of blood in those areas. .

PAGE 70

52 igure 3-11. The topography of a section of artery reconstructed from the force volume e ng 125 3 4 Region 1-300-250-200-150-100-5005.9176.91482192903614325035736447157868579289991070114112121283135414251496Separation [nm]Deflection [nm ] Region 2-300-250-200-150-100-5005.9176.91482192903614325035736447157868579289991070114112121283135414251496Separation [nm]Deflection [nm ] Region 3-300-250-200-150-100-5005.9176.91482192903614325035736447157868579289991070114112121283135414251496Separation [nm]Deflection [nm ] Region 4-300-250-200-150-100-5005.9176.91482192903614325035736447157868579289991070114112121283135414251496Separation [nm]Deflection [nm ] Region 5-300-250-200-150-100-5005.9176.91482192903614325035736447157868579289991070114112121283135414251496Separation [nm]Deflection [nm ] 12345 F image showing a thin slice spanning across nuclei from two adjacent cells. Force curves from the circled regions were analyzed in order to determine thaverage elastic moduli. Corresponding plots from the force curves in each region are overlaid in the numbered graphs. Regions 1, 2, 3 and 4 are movifrom a raised to lower area and region 5 is a raised area. There is a significant difference between the elastic modulus of regions 1 and 4 (P<0.05) and regions 2 and 4 (P<0.05).

PAGE 71

53 T able 3-2. Elastic modulus for different regions of porcine artery lumen Figure Region Elastic Modulus [kPa] 1 19.92 1.00 3-10 2 19.91 0.51 3 18.08 1.33 4 20.28 0.91 1 10.56 0.21 2 10.59 0.29 3 10.33 0.17 4 10.17 0.32 3-11 5 10.45 0.47 he values for elastic modulus of the lumen compare well to those obtained by others T using a variety of techniques. The application of a Hysitron nanoindenter to fresh samples or porcine aorta revealed an elastic modulus of 0.232 0.078 MPa [66]. This technique uses a relatively large indenter with 100 m radius of curvature tip compared to the AFM tip with a 40 nm radius of curvature. As such, the nanoindenter probes multiple cells at one time and can not resolve sub-cellular mechanical properties. The order of magnitude difference in elastic modulus may be due to the different artery type; however it is more likely due to sample desiccation. Use of the nanoindenter requires exposing the tissue to air during imaging, a technique that questions the entire validity of the methodology for biological samples. AFM has also been used to directly measure EC mechanical properties from fresh rat aorta [64]. This study did not calculate the elastic modulus directly, however it showed variation in mechanical properties with location on the cell. The cell nuclei were significantly stiffer than the surrounding cell body with no significant difference in stiffness between nuclei of different cells. This same paper saw similar results in terms of topography with some cells elongated and other more rounded and with the same continuous transition between adjacent cells. Mechanical properties of ECs have been studied on synthetic surface in vitro as well [63, 65, 67-70]. For example,

PAGE 72

54 ECs grown on gelatin-coated polystyrene in static culture had an elastic modulus of 6.8 0.4 kPa over the nucleus and 1.4 0.1 kPa near the cell periphery [65]. ECs exposed to a 2 Pa shear force demonstrated a stiffer mechanical response indicating a possible reason why the examined cells had a higher elastic modulus than reported in vitro [63]. Similary, the same ECs grown in static culture demonstrated an increase in elastic modulus with static culture time ranging from 1 to 15 kPa [68]. Clearly, the elastic modulus values obtained for the ECs in the porcine artery agree well with the literature. The unique difference is the relatively homogeneous elastic modulus across the cell body not seen for ECs in culture. In addition, there is an increase in the viscous loss over the cell periphery probably caused by the flow cytoplasm through the porous cytoskeleton. Conclusions The ECs formed a continuous, colayer on the vessel surface with the ECs a arpet-like mon pproximately aligned in the direction of flow. EC shape varied from an elongated ellipse to approximately circular depending on the region of the artery examined. Fluorescent microscopy revealed a similar close packing of the ECs as well as a robust actin network within and connected throughout all the cells. From AFM, living ECs in fresh artery showed a smooth and continuous transition between cells with close packing similar to that seen in the other techniques. The mechanical properties were relatively homogeneous throughout the lumen though small, statistically significant differences were seen in the elastic modulus of regions over the cell nuclei compared to regions over the cell periphery.

PAGE 73

CHAPTER 4 EFFECT OF ARGON PLASMA TREATMENT ON PDMS ELASTOMER INVESTIGATED BY AFM Introduction Using the AFM to calculate elastic modulus and surface roughness allows nanometer scale resolution when examining material surfaces. This capability becomes important when examining the effect of surface treatments that change the chemical and physical structure of a material. When engineering biomaterials these surface properties become critical to the success of the biointerface between the synthetic and biological components. Polydimethylsiloxane elastomers (PDMSe) are an example of a widely used medical polymer, but often its properties need to be enhanced. We have examined the difference in surface properties caused by the use of a radio frequency glow discharge (RFGD) argon plasma treatment to increase hydrophilicity of the PDMSe. By using the AFM, the top surface can be examined on the same scale of nanometers and micrometers that proteins and cells function within. Materials and Methods Sample Preparation The PDMSe used was Silastic T-2 (Dow Corning). It has good dimensional stability, self-degassing properties and optical transparency. Hydrosilylation PDMSe maintains shape in aqueous environments. The transparency and uniform surface, which can be produced with minimal defects, is excellent for viewing through with standard light microscopy techniques. Silastic T-2 was also chosen over other PDMSe because 55

PAGE 74

56 there were minimal unknown or non-disclosed additives, allowing for comprehensive control of additives and chemical modifications. The two components of the Silastic T-2 were mixed together in 10:1 w/w base resin to curing agent ratio in ~200 g batches. Each batch was degassed for 20 minutes and then allowed to sit for another 5 minutes at atmospheric pressure for continued degassing. A 60cc syringe with a catheter end was filled with the PDMSe prepolymer and used to dispense ~2 mL of PDMSe per 35 mm diameter polystyrene culture grade Petri dish (Corning). Any bubbles that formed during pouring were popped with a needle and the Petri dish covers were put on to prevent dust from the air settling on the curing PDMSe. The PDMSe was allowed to cure for at least 24 hours at room temperature before being used. According to the manufacturer the PDMSe was cured after 24 hours at 25C [71]. The resulting PDMSe film thickness was ~2 mm. Argon RFGD Plasma Treatment Plasma treated PDMSe was prepared by exposing fully cured films to RFGD argon plasma. Two types of plasma systems were used; a home built bell jar system (BJS) and a commercial reactive ion etcher (RIE). Both experimental systems are illustrated in figure 4-1 and there are a number of significant differences between the systems. The BJS operates by flowing argon gas at low pressure through a narrow glass column around which is wrapped a metal RF coil. An AC current is passed through the RF coil to generate a magnetic field that ionizes the argon gas creating plasma. Below the RF coil the bell jar opens up into a larger chamber where the sample sits on a raised platform. A vacuum outlet at the bottom of the bell jar pulls the argon out maintaining a constant flow at a low pressure of ~50 mTorr. As the argon plasma flows from the RF coil region down to the outlet the plasma dissipates with the power dropping proportional to the

PAGE 75

57 inverse square of the distance. The separation distance between the sample surface and the RF coil was maintained at 2.5”. For all samples treated in the BJS, pressure was maintained at 50 mTorr, the argon gas flow rate was 8 sccm and the RF power was 50 W. In the RIE system, the sample chamber is completely filled with the plasma, which is generated by passing an AC current between large plates at the bottom and top of the chamber. This system was operated at the same conditions as the BJS; 7 sccm argon, 50 mTorr and 50 W. After treatment samples were immediately covered with Hanks balanced salt solution (HBSS) to prevent hydrophobic recovery and stored up to 7 days before AFM analysis. Figure 4-1. An illustration of the two different types of plasma systems used to treat the PDMSe. The home built system flows argon through an RF coil and then directs the plasma onto the sample. The reactive ion etcher uses two plates on the top and bottom of a chamber filled with argon to create a plasma that surrounds the entire sample. Home Built SystemHome Built System Argon Flow Controller RF Power Generator Vacuum Pump RF CoilsRF CoilsVacuum Vacuum ChamberChamberSampleSampleAdjustable Adjustable HeightHeight Reactive Ion EtcherReactive Ion Etcher Vacuum Pump Argon Flow Controller RF Power Generator SampleSampleVacuum Vacuum ChamberChamber

PAGE 76

58 AFM Operation A Dimension 3100 AFM with a Nanoscope IIIa controller (Digital Instruments) was used for all measurements. The silicon nitride cantilevers (Veeco Metrology) have a spring constant of 0.06 N/m that matches the low modulus of the PDMSe. Samples were imaged in HBSS to eliminate capillary forces while imaging the soft polymer surface and to maintain the plasma modified PDMSe surface. The AFM scanner was lowered directly into a 35 mm Petri dish coated with PDMSe on the bottom and filled with 1.5 mL of HBSS (figure 4-2). Contact mode imaging was used to obtain high resolution topographic images of the surface. Scans sizes varied from 1 to 100 m at scan rates of 1 to 2 Hz. The force curves used to measure the elastic modulus and adhesion, were obtained with the AFM operated in a contact force volume mode or standard force curve mode. Force volume mode allowed Figure 4-2. A schematic of the AFM experimental setup used to image the PDMSe for both the topographic images and the force data to be obtained in a single scan. submerged in HBSS. The AFM scanner was lowered into the liquid filled orce curves were also performed to look at the change in elastic modulus with indentation depth and indentation rate. These were performed with the AFM in force Scan Head Petri Dish Liquid AFM Probe Substrate Petri dish and scanned a coating of plasma treated PDMSe on the bottom. F

PAGE 77

59 calibrand sample interaction were modeled using a Hertzian approach to rce curves (see appendix A for details). The H ation mode. A 5x5 arrays of force curves were spaced at 1m intervals on an XY grid. Each row was either incremented in indentation depth or force curve rate. For increasing indentation depth, constant cantilever deflections of 34.5, 172.4, 345 and 700 nm were used. The actual indentation depths depend on both the cantilever deflection and the elastic modulus of the material, but may be estimated as 9, 20, 50 and 70 nm for the 34.5, 172.4, 345 and 700 nm cantilever deflections respectively. PDMSe as cast against air, cured against glass or as cast and plasma treated for 10 s or 1 min were examined to determine if there were differences in elastic modulus with depth. For increasing indentation rates, cantilever deflection of 700 nm (~70 nm indentation) at 0.50, 1.00, 3.49, and 7.00 Hz was used on as cast PDMSe. With a force curve range of 1500 nm, the corresponding indentation speeds were 1.5, 3.0, 10.5 and 21.0 m/sec respectively. Analysis of Force Curves for Elastic Modulus The tip determine the sample elastic modulus from the fo ertz model predicts elastic deformation when two spheres are brought together under load [72]. Sneddon extended this relation to include the case of an infinitely stiff tip with a conical shape indenting a soft, planar sample [73]. In reality, the tip is a four-sided pyramid, but the approximation of a cone is used to simplify the model and, due to the small indentation depth, the error is assumed to be negligible. The elastic modulus is calculated from the contact region of the force curve. For PDMSe at ambient conditions the Poisson’s ratio is assumed to be 0.5 as the polymer can be considered an ideal elastomer. Statistical analysis of the surface elastic modulus is performed using one-way ANOVA followed by Tukey’s pairwise comparisons.

PAGE 78

60 Results and Discussion Topography There were visible changee PDMSe after treatment with ma (Figure 4-3). The as cast PDMSe had a relatively smooth surface from curing. Material Roughness, Rq [nm] s in the topography of th the argon plas against air with no unique features present. After exposure to the argon plasma the PDMSe surface changed significantly with a reduced RMS roughness (table 4-1). It appears that the plasma treatment ablated the surface and smoothed out the roughness. There were the occasional small spikes believed to be dust particles. The most interesting features in the plasma treated PDMSe were the occasional cracks in the surface (visible in figure 4-3B as indicated by the arrow). It is thought that the cracks result from the mechanical stress induced while handling the samples. These were extremely infrequent in the samples cured in the 35 mm diameter Petri dishes but they were quite prevalent in free standing films after being handled normally. Table 4-1. Variation in RMS roughness of the different polymers examined Polystyrene 9.1 8.0 PDMS Elastomer (E = 1.4 MPa) 26.9 5.6 PDMS Ela stomer (plasma treated) 6.8 4.5 he Argon RFGD plasma treated PDMSe surface is heavily modified compared to the as cast PDMSe. It is well known that the plasma treatment used here can cause major surfac T e modification by altering the surface chemistry [74, 75]. The plasma breaks surface bonds and causes free radicals to form on the surface. Immediately following the plasma treatment, the chamber is repressurized with air and the oxygen reacts with the

PAGE 79

61 free radicals on the surface of the polymer. A silica like layer of Si x O y forms on the surface that is much stiffer than the PDMSe bulk and PDMSoligomers. Figure 4-3. Topographical images of the PDMSe (A) before and (B) after treatment with 2mode. crack B A an argon plasma. The 100 m areas were imaged in liquid using contact equipment used and the duration of the plasma exposure time. For the BJS, the only differ The morphology of the surface generated by the plasma treatment varied with the ences in morphology were the appearance of surface cracks that began to spontaneously appear after 5 to 10 minutes of Argon exposure and increased with any mechanical stress on the film. For the RIE, no cracking was observed even at the maximum exposure of 10 minutes; however there were still noticeable changes in surface morphology. After 1 minute exposure, a finger-like morphology arose that qualitatively resembles spinoidal decomposition (figure 4-4A). However, since this is a homopolymer that is not a possibility because there is no second phase. This instead is likely a buckling phenomenon resulting from the surfaces stresses as the silica-like layer grows. Increasing exposure time to 10 minutes in the RIE plasma causes a coarsening of this surface morphology (figure 4-4B). The cause of the coarsening may be preferential etching of low crosslink density areas, continued surface buckling (as seen in figure 4

PAGE 80

62 4A) or may be a rearrangment of Si x O y to reduce surface energy. There is no change in RMS roughness between the 1 min and 10 min treatment times. Figure 4-4. AFM contact mode topographic scans of PDMSe tre 10m 10m1 min, Rq= 6.26 nm10 min, Rq= 5.90 nm B A ated with (A) 1 minute exposure and (B) 10 minute exposure to argon plasma in the RIE. For the . closer look at the 1 minute RIE plasma treatment sample reveals that the finger-like p same size scan area of 10 m, a distinct change in surface morphology is observed even though the surface roughness remains essentially unchanged A rojections snake around on the surface for 2 to 20 m (figure 4-5 A and B). Cross-section analysis of this area shows that the finger-like features are 3.2 0.5 nm high and 1.4 0.1 m wide (figure 4-5C). These dimensions indicate that this is indeed the silica-like layer and not PDMS oligomers because the width is much larger than that of a single polymer chain. Furthermore, it can not be a bundle of PDMS oligomers either because the low T g and lack of a T m indicate that PDMS does not form crystals and the chains only crosslink at the vinyl functionalized end groups rather than along the backbone.

PAGE 81

63 In general, these surfaces have RMS roughness values that are <10 nm. However, determining the critical surface roughness that affects cells is difficult. Each cell type has unique behaviors that are dependent on its genetic code and physical environment. For example, micrometer scale ridges have been shown to cause polarization of porcine vascular endothelial cells (PVECs) [13, 76]. A surface with 5 m high, 5 m wide ridges separated by 5 m valleys has an R q = 2.5 m, which is 2 to 3 orders of magnitude higher than the roughness of these samples. Grooves as small as 100 nm in width and 30 nm deep direct cell growth and the contact guidance effect increases with the feature aspect ratio. The best results for directing cells without drastically changing the morphology occur at 500 nm depths [77]. Assuming equal groove width and spacing between grooves, this 500 nm depth results in a surface roughness of R q = 250 nm, which is still much larger than the inherent roughness of these PDMS and PS surfaces. Surface roughness from 20 to 70 nm increased rat neural cell adhesion to silicon wafers but caused no directed response [78]. Though different cell types are known to respond to surface topography differently, the low surface roughness of these materials is below the limit reported in the literature. Thus, it is concluded that the surface roughness, which is <10 nm, and the isotropic surface morphology of these samples will increase adhesion but will not elicit any directed cell response.

PAGE 82

64 10m 4.4m 0.0 2.0 4.0 6.0 8.0 10.0 12.0 0.0 2.0 4.0 6.0 X [ m ] Z [ n m ] C B A Figure 4-5. The 1 minute RIE plasma treated PDMSe had a surface morphology (A) that appeared similar in morphology to phase separated spinoidal decomposition seen in block copolymers. A close-up of this region (B) shows an ~4 nm height variation and the size of these ‘fingers’ can be quantified using (C) a cross-section plot. Force Curve Properties The as cast PDMSe force curves elicit a large adhesive jump-on and an even larger adhesive pull-off indicating that there are long-range attractive forces (figure 4-6). The rounding of the jump-on and pull-off minima, which are normally sharp, is due to the soft sample-tip interaction combined with the adhesion to the tip. The magnitude of the adhesion force is likely due to a contamination of the AFM tip with oligomeric PDMS. It

PAGE 83

65 is known that any PDMSe will have some unreacted PDMS oligomers left over that are not incorporated into the bulk. As the surface energy of the PDMS oligomers is lower than that of the bulk, they will wet the surface. The AFM tip is composed of silicon nitride and it too has a higher surface energy than the PDMS chains. When the AFM tip contacts the PDMSe surface, the oligomeric PDMS transfers to the tip and coats it. The resulting surface structure is illustrated in Figure 4-7. It is believed that the long-range interaction is due to the attractive force between the PDMS oil on the tip and sample surface. The PDMS oils do not interfere with the elasticity measurements because they are compressed during the linear contact region; however, they do have a measurable effect on adhesion. Polymer Elastic Modulus The PDMSe had a surface elastic modulus of E = 1.5 0.8 MPa, which is nearly identical to the bulk elastic modulus of E = 1.4 MPa. The similarity is likely due to the extremely low T g for PDMSe of -120C and the nearly ideal ‘rubber elasticity’ behavior exhibited by this polymer at room temperature. The plasma treated PDMSe did not have an associated change in the elastic modulus of the bulk because the plasma is only a surface treatment. The plasma treatment was applied to the standard Silastic T-2 so its bulk elastic modulus was 1.4 MPa and the surface elastic modulus when plasma treated was 3.0 0.9 MPa. The surface elastic modulus of the plasma treated PDMSe was significantly different than the untreated PDMSe (P < 0.05). There was a clear shift in the type of force curve interaction too with no jump-on adhesion and force curve shape that resembled that of a hard surface rather than a soft surface prior to the plasma treatment (4-6B).

PAGE 84

66 A B Figure 4-6. Representative force curves from (a) PDMSe and (b) argon plasma treated PDMSe. Force curves were recorded during force volume imaging. Figure 4-7. Drawing illustrating PDMS oligomers (oils) coating the surface of the bulk elastomer and the silicon nitride AFM tip. This is believed to occur during imaging of the PDMS samples and is responsible for the large jump-on and pull-off adhesive forces seen in the force curves.

PAGE 85

67 Table 4-2. Elastic modulus of polymers surfaces indented ~5 nm in to the substrate Material Elastic Modulus [MPa] Polystyrene 4.4 1.4 PDMSe 1.5 0.8 Plasma Treated PDMSe 3.0 0.9 The PS showed the highest elastic modulus of any of the polymers tested with E = 4.4 1.4 MPa. The modulus was statistically different than the four other materials (P < 0.05). This is a very low value compared to the bulk elastic modulus for standard PS, taken as E ~3 GPa for a thermoplastic below its T g [79]. The difference is theorized to be based on the difference in chain mobility between the bulk and surface material. PS is an amorphous thermoplastic with a T g of 100C. These experiments were conducted at 22C, therefore while the bulk material was in a glassy state, the exposed free surface had enhanced chain mobility and few physical entanglements to restrict it. As a result, the polymers radius of gyration should increase, resulting in a more “liquid-like” behavior and decreased elastic modulus. In addition, the plasma treatment oxidizes the surface, which increases the water content forming a hydrogel layer. Hydrogels such as polyHEMA have a bulk elastic modulus varying from 0.1 to 1 MPa depending on crosslink density and hydration state [80]. Thus, it is suggested that the low PS surface elastic modulus is due to the free, hydrogel-like chains. Change in elastic modulus with indentation depth Subtle difference in surface properties due to curing PDMSe against a silicon wafer or against air were examined along with the difference in a 10 second and 1 minute RIE plasma treatment. The PDMSe samples were indented at depths dictated by constant cantilever deflections of 34.5, 172.4, 345 and 700 nm. Figure 4-8 shows the elastic modulus as a function of cantilever deflection for the different sample treatments. The

PAGE 86

68 indentation depths depend on both the cantilever deflection and the elastic modulus of the material, but may be estimated as 9, 20, 50 and 70 nm for the 34.5, 172.4, 345 and 700 nm cantilever deflections respectively. A number of interesting results emerge from this data. The elastic modulus for PDMSe cured against air and cured against glass is ~225 kPa compared to a bulk elastic modulus of ~1.4 MPa. This result is consistent with the theory that the PDMS oligomers segregate to the surface. However, as indentation depth is increased the elastic modulus drops further to a value as low as ~20 kPa. The cause of this behavior is thought to be due to the fused silica filler used in the Silastic T-2 polymer as a reinforcement phase migrating to the surface, but this conclusion is unsubstantiated. The argon plasma treatment increased the surface elastic modulus by roughly 50% to ~330 kPa. There was no significant difference in the elastic modulus between the 10 sec and 1 min plasma treatment times indicating that the effects on the surface were equivalent at the RF power and times used and only was detectable within the first ~20 nm of the surface. The variation in elastic modulus of the PDMSe and plasma treated PDMSe compared to the values reported earlier demonstrate the main issue with AFM based elasticity measurements. Small variations in tip geometry and cantilever spring constants can easily skew results by 50% or more. Thus, this technique is best at reporting relative differences than absolute values. Even the best tip calibration methods are prone to 20% error and often require specially shaped cantilevers [81, 82].

PAGE 87

69 Elastic Modulus vs Indentation DepthDeflection [nm] 0200400600 Elastic Modulus [Pa] 01e+52e+53e+54e+5 Glass Cured Air Cured 10s RIE 1min RIE Figure 4-8. The elastic modulus of PDMS elastomer cured against glass, air cured, air cured with 10 sec argon plasma treatment and air cured with 1 min argon plasma treatment. All samples were examined under liquid buffer to minimize capillary effects. There was only a significant difference between the untreated and plasma treated samples (P<0.05). Force curve indentation were performed at 1 Hz for all samples. PDMSe viscoelastic properties The viscoelastic properties of the surface were examined at the maximum cantilever deflection of 700 nm (~70 nm indentation) at 0.50, 1.00, 3.49, and 7.00 Hz. With a force curve range of 1500 nm, the corresponding indentation speeds were 1.5, 3.0, 10.5 and 21.0 m/sec respectively. No statistical difference was found between the different indentation rates. However, there was a trend between the trace (indentation) and retrace (withdrawal) elastic moduli suggesting a small viscous component to the elastic modulus at these temperatures and indentation rates. In terms of establishing firm adhesion to the surface, most cells and microorganisms will see the PDMSe surface as

PAGE 88

70 completely elastic in character. The formation of adhesions occur at a frequency of 0.1 to 0.01 Hz, and the experiments (figure 4-8) show no viscous loss below 0.5 Hz. Rate Dependence of Elastic ModulusIndentation Rate 0.50 Hz1.00 Hz3.49 Hz7.00 Hz Elastic Modulus [Pa] 0500010000150002000025000 Trace Retrace Figure 4-9. Dependence of the elastic modulus on indentation speed for both trace (indentation into the sample) and retrace (withdrawal from the surface). There are clear trends of a higher elastic modulus for trace compare to retrace, but no significant difference between indentation rates. Force Volume Images In the Silastic T-2 system used here, there is a silica particulate phase that adds considerable strength to the bulk PDMSe. At the small indentation depths there is some evidence of a harder phase that would indicate the silica particles were near the surface. However, there is also no visible phase segregation in the force volume slice that is independent of the topography. Noise seen in the force volume slices is due to the high contrast and the surface shows no variation in elastic modulus with XY position. The exception to this is the plasma treated PDMSe as there are the occasional visible cracks in the plasma treated surface. These cracks are actually interruptions in the hardened

PAGE 89

71 surface layer that expose lower modulus PDMSe underneath. Close examination of the force volume slice shows a faint line that appears to correspond to the crack visible in the topographic image. Close up images of the crack do not reveal any change in elasticity though further examination is warranted. Conclusions The PDMS elastomer showed long-range attractive forces and a soft interface while the PS and plasma treated PDMS showed a stiffer interface with only a standard pull-off adhesion event. In terms of cell growth, the surfaces will appear quite different on the cellular level. The shallow probing with the AFM tip is on a similar scale to the surface seen by cells and reveals similar uniform elasticities. The real differences are the PDMS oil coating that is seen on the PDMS elastomers (not the plasma treated PDMS). These oils will definitely interact with any cell approaching the surface. Due to the inherent hydrophobicity of PDMS, the oil will not enter the aqueous solution of the cell environment and thus the proteoglycans produced by the cells will have to compete with the oil layer for anchoring on the bulk elastomer. It is therefore anticipated that the hydrophobic and oil covered PDMS elastomer will not be a good cell growth surface without further modification. In contrast, the oil free plasma treated PDMS and PS should be better. The increased hydrophilicity of the plasma treatment should also enhance cell adhesion and spreading. Unique variations in surface morphology of the RIE plasma treated PDMSe were observer, but are unlikely to influence cell adhesion given the small RMS roughness <10 nm. The lack of hysteresis between the trace and retrace curves on as cast PDMSe indicates that it is behaving as an ideal elastomer as ambient conditions.

PAGE 90

CHAPTER 5 ENDOTHELIAL CELLS CULTURED ON MICROENGINEERED SURFACES EVALUATED BY FLUORESCENT MICROSCOPY Introduction In these experiments, ECs are grown on various polymer surfaces that have been modified in order to improve cell adhesion, spreading and function. These surfaces are designed to elicit the formation of a normal EC monolayer for potential application in small diameter vascular grafts and stent coatings. The surfaces are microengineered with micrometer scale topographies and micrometer scale patterns of protein incorporated into the surface. This is achieved using a variety of techniques that combine microprocessing methods with microcontact printing (CP) of protein solutions. The goal is to direct the formation of focal contact adhesions between the ECs and the substrate with proteins and to direct cells through contact guidance cues with the topography. The cell growth is evaluated using immunofluorescent staining of the cell nuclei, actin cytoskeleton and focal contact adhesions. Qualitative assessment is based on the ability of the ECs to form a monolayer and on individual cell shape. Image processing is used to quantify the cell growth characteristics. Materials and Methods Substrate Production Microtopographical surface patterns Microtopographies were fabricated in polydimethylsiloxane elastomer (PDMSe) by curing prepolymer against micromachined silicon (Si) wafer masters following 72

PAGE 91

73 established protocols [39, 83]. Briefly described, the desired micropatterns were initially generated in one of the following computer aided design (CAD) programs: Macromedia Freehand 10, Autodesk Mechanical Desktop 5 or Conventor depending on the particular type of photomask to be used. These designs were then transferred to photomasks for use in photolithography. For designs with minimum feature sizes of 25 m or greater, photomasks (figure 5-1) were created on overhead transparency film using a Hercules linotype 5000 DPI image setter by Pageworks Inc. (Cambridge, MA). For designs with smaller feature spacing <25 m, a chrome on quartz glass photomask (figure 5-2 shows the CAD design) was fabricated by electron beam writing (Compugraphics, Inc.). Contact photolithography was used to transfer the patterns in the photomask into Si wafers as illustrated in figure 5-3. Polished Si wafers of 2”, 3” or 4” diameter were spin coated with positive photoresist (Shipley 1818) and exposed to UV light. Exposed photoresist coated Si wafers were developed with Microposit 321 and etched to various depths using deep reactive ion etching (DRIE). After etching, the remaining photoresist was removed with acetone and then the Si wafer was piranha etched to remove any remaining organic matter (photoresist and DRIE passivation layer residue). To ensure the PDMSe did not bond to the Si wafer, hexamethyldisilizane (HMDS) was used as a passivation layer. HMDS was removed from the bottle through a septum using 3 mL syringe with a needle tip and liberally applied to the surface of the wafer as an even coating. The HMDS was allowed to evaporate to a dry film and then the wafer was washed with distilled water to remove excess HMDS residue that was not adsorbed to the wafer. A single HMDS treatment was used for ten replications with PDMSe before it was reapplied.

PAGE 92

74 Figure 5-1. These are optical micrographs of photomasks printed on standard transparency using a high-resolution 5000 DPI image setter. The patterns are (A) 50 m wide hexagons and separated by 25m, (B) 100 m wide hexagons and separated by 25m, (C) 75x50 m wide hexagons and separated by 25m and (D) 100x50 m wide hexagons and separated by . CAD drawings of the patterns used for 25m. Scale bars represent 200 m. Figure 5-2the chrome on glass photomask. (A) 40 m wide stars spaced at every 60 m, alternate rows of stars rotated 45. o create micropatterned PDMSe films, Silastic T2 (Dow Corning) PDMSe was mixed B A (B) Hexagons 20 m from face to face with 2 m spacing between adjacent faces. T in 10:1 ratio by mass of base resin to curing agent and mixed thoroughly. To remove air bubbles, the PDMSe was degassed for 20 minutes under vacuum and then allowed to sit at atmospheric pressure for another 5 minutes for all bubbles to dissipate. A mold was used to ensure uniform film thickness. Two 8”x 8”x 3/8” glass plates were

PAGE 93

75 used as the top and bottom of the mold. Each glass plate was covered with a thin PET film held in place with double sided tape. A micropatterned Si wafer was placed in the center of the bottom plate and surrounded by three spacers, each made of two stacked microscope slides. The PDMSe prepolymer was poured on top of the wafer and allowed to spread to cover the entire wafer surface. Bubbles that formed and floated to the top of the liquid polymer were popped with a needle. The top plate was then placed with a 5 lbs weight on top to maintain the proper spacing. Samples were allowed to cure for 24 hours at room temperature and then removed and cut into small sections of patterned area ~1 cm 2 . Figure PhotomaskSilicon waferPhotoresistUV light Reactive Ion EtchingEpoxy PDMS elastomer A 5-3. This diagram outlines the steps to produce micrometer scale topographical features. (A) A Silicon wafer is coated with photoresist and exposed to UV y light through a photomask. (B) The exposed photoresist is removed and theremaining resist is baked to harden it. (C) Reactive ion etching is used to anisotropically etch the exposed silicon with micrometer scale patterns. (D) The remaining resist is removed from the etched silicon wafer and PDMS elastomer is then cured against the wafer to create a negative of the surface. (E) The PDMS elastomer negative is used as a mold to create an epoxy copthat is identical in surface topography to the original silicon wafer. (F) Additional PDMS elastomer samples are cast against the epoxy master to preserve the integrity of the original patterned silicon wafer. Photoresist Epoxy Silicon wafer PDMS elastomer B C D E F

PAGE 94

76 To cSe substrates were fixed to the bottom of 35 mm diameter tissue culture grade polystyrene Petri dishes (Corning) (Figure 5-4). To do this, the Petri dish was filled with 2 to 3 mL of PDMSe prepolymer and then the cured micropatterned PDMSe substrates were pushed down in the center of the prepolymer filled Petri dish. The PDMSe was allowed to cure for 24 hours at room temperature creating a uniform PDMSe surface with a textured region in the center of the Petri dish surrounded by flat PDMSe. Figure 5-4. Samprfaces are prepared by placing a cured meter Petri dish er. ECs in media The adhesion. The PDMSe films, fixed in the Petri dishes, are plasma treated with an Argon radio frequency glow discharge (RFGD) plasma. The plasma treatment system used is a gas flow system where the sample sits in a bell jar with a gas inlet column at the top and a vacuum outlet at the bottom. The gas inlet column has an RF coil surrounding it that ionizes the argon gas as it flows through and strikes the samples in the bell jar. Three 35 mm diameter Petri dishes were placed in the bottom half of a 9 cm diameter Petri dish and placed on a metal stand 2.5” below the RF coil in the bell jar. The pressure was reduced to 50 mTorr and then the chamber was purged 3 times with argon at a flow rate reate samples for cell culture experiments, the micropatterned PDM ECs and media Micropatterned area PDMSe film meter) Surrounding flat Petri dish (35 mm dia les for cell culture with micropatterned suicropatterned substrate in the center of a 35 mm diamand curing in place by surrounding it with PDMSe prepolym are poured into the remaining volume of the Petri dish and cultured for 4, 7 or 14 days. hydrophobic PDMSe is surface modified to increase the hydrophilicity for cell

PAGE 95

77 of 1000 sccm. PDMSe samples were exposed for 1 minute to plasma generated with an RF power of 50 W and an argon flow rate of 8 sccm at 50 mTorr. The samples were removed from the bell jar and covered with sterile Hanks balanced salt solution (HBSS) immediately after the plasma treatment. Some samples were also incubated with a solution of 50 g/mL fibronectin (FN) in HBSS for 1 hour. The FN treated samples were subsequently washed 3x with HBSS and then seeded with PVECs. Chemical surface micropatterns Chemical surface micropatter ning was performed using CP of FN on the 35 mm diame ter polystyrene Petri dishes and the plasma treated PDMSe. The CP process is well established as a method to pattern proteins and other molecules for directing cell adhesion and growth [45, 84-86]. In the experiments, the CP stamps were made of PDMSe and fabricated in the same manner as the PDMSe substrates with microtopographies just described. CP features 1 m deep were sufficient to effectively micropattern surfaces. Stamps were fabricated as 1 cm 2 PDMSe squares with the desired micropattern covering one entire side. The patterned sides of the stamps were Argon plasma treated using the previously described protocol for a period of 1 minute to enhance wetting by the FN solution. CP was performed using methods similar to those published and described in figure 5-5 [45, 84]. Before proceeding, both the CP stamp and substrate to be patterned were sterilized in 70% ethanol for 5 minutes, patted dry with Kimwipes and the allowed to air dry. Stamps were inked with a 50 g/mL solution of FN in phosphate buffered saline (PBS) by dripping on the solution using a pipette and allowing it to incubate at room temperature for 1 hour. Next, the stamps were rinsed thoroughly with HBSS to remove excess FN and then dried under nitrogen. To pattern

PAGE 96

78 flat surfaces, the stamps were placed patterned side down in the bottom of a 35 mm polystyrene Petri dish or in the bottom of a Petri dish coated with a thin layer of cured PDMSe. Care was taken to place the stamp down firmly the first time so that it did not shift and cause multiple stamping. Once positioned, the stamp was lightly pushed on the backside with a pair of tweezers to seal the stamp against the surface. To ensure a constant pressure, 15 mL centrifuge tubes were filled with 10 mL of distilled water and placed top down on the back of the stamp. Care was taken with balancing the centrifuge tube so it stood straight up and applied even pressure to the stamp. After 1 hour incubation, the stamps were removed and the FN patterned surface washed 3 times with sterile PBS. Figure 5-5. A AAddssoorrbb FFiibbrroonneeccttiinn oonnttoo SSttaammpp SSttaammpp FFiibbrroonneeccttiinn oonnttoo SSuubbssttrraattee SSttaammpp FFNN oonnttoo SSuubbssttrraattee PDMSeStamp Fibronectin Seed Endothelial Cells n illustration of the CP process where FN is adsorbed to a PDMSe stamp and transferred to another surface. Cell Culture porcine vascular endothelial cells (PVECs) were obtained from the researsuspended in additional media to reach the desired cell concentration. Cell seeding Primary ch group of Dr. Edward Block at the Malcolm Randall VA medical center (Gainesville, FL). PVECs were provided in suspension at an approximate concentration of 200,000 cells per mL and diluted to 50,000 cells per mL for seeding on microengineered surfaces. The cell were counted on a hemocytometer (Sigma) and

PAGE 97

79 Table 5-1. Different microengineered surfaces used as substrates for EC culture. Abbreviation Surface Type PS Standard tissue culture grade PS 35 mm diameter Petri dish PS_FN Standard tissue culture grade PS 35 mm diameter Petri dish coated with physisorbed FN PDMSe_plasma A 2 mm thick PDMplasma treated for Se film in the bottom of a PS Petri dish argon RFGD 1 minute PDM Se _FN A 2 mm thick PDMSe film in the bottom of a PS Petri dish a rgon RFGD plasma treated for 1 minute and coated with physisorbed FN PDMSe_hex_plas ma , 20 or 1 A 2 mm thick PDMSe film in the bottom of a PS Petri dish with l m deepm wide hexagons and separated by 2m; argon RFGD plasma treated f minute PDMSe_hex_ FN , 20 A 2 mm thick PDMSe film in the bottom of a PS Petri dish with l m deep m wide hexagons and separated by 2m; argon RFGD plasma treated for 1 minute and coated with physisorbed FN PDMSe_Hex_04_plasma A 2 mm thick PDMSe film in the bottom of a PS Petri dish with l m deep, 50 m wide hexagons and separated by 25m; argon RFGD plasma treated for 1 minute PDMSe_Hex_04_FN A 2 mm thick PDMSe film in the bottom of a PS Petri dish with l m deep, 50 m wide hexagons and separated by 25m; argon RFGD plasma treated for 1 minute and coated with physisorbed FN PDMSe_Hex_06_plasma A 2 mm thick PDMSe film in the bottom of a PS Petri dish with l m deep, 100 m wide hexagons and separated by 25m; argon RFGD plasma treated for 1 minute PDMSe_Hex_06_FN d A 2 mm thick PDMSe film in the bottom of a PS Petri dish with l m deep, 100 m wide hexagons and separated by 25m; argon RFGD plasma treatefor 1 minute and coated with physisorbed FN PDMSe_Hex_10_plasma A 2 mm thick PDMSe film in the bottom of a PS Petri dish with l m deep, 75x50 m wide hexagons and separated by 25m; argon RFGD plasma treated for 1 minute PDMSe_Hex_10_FN A 2 mm thick PDMSe film in the bottom of a PS Petri dish with l m deep, 75x 50 m wide hexagons and separated by 25m; argon RFGD plasma treated for 1 minute and coated with physisorbed FN PDMSe_Hex_13_plasma , A 2 mm thick PDMSe film in the bottom of a PS Petri dish with l m deep 100x50 m wide hexagons and separated by 25m; argon RFGD plasma treated for 1 minute PDMSe_Hex_13_FN , A 2 mm thick PDMSe film in the bottom of a PS Petri dish with l m deep 100x50 m wide hexagons and separated by 25m; argon RFGD plasma treated for 1 minute and coated with physisorbed FN PS_FN_star ars Standard tissue culture grade PS 35 mm diameter Petri dish with CP FN st 40 m tip to tip at 60 m intervals PDMSe_FN_hex A 2 mm thick PDMSe film in the bottom of a PS Petri dish argon RFGD plasma treated for 1 minute with CP FN hexagons 20 m face to face, separated by 2 m and hexagonally packed PDMSe_FN_ star A 2 mm thick PDMSe film in the bottom of a PS Petri dish argon RFGD plasma treated for 1 minute with CP FN stars 40 m tip to tip at 60 m intervals PDMSe_FN_circle A 2 mm thick PDMSe film in the bottom of a PS Petri dish argon RFGD plasma treated for 1 minute with CP FN circles 3 m diameter at 6 m intervals and square packing

PAGE 98

80 media consisted of RPMBS), 1% L-glutamine and 2% antibiotics (100 U/ml of penicillin, 100 g/ml of streptomycin, skeleton of the cells was stained using phalloidin-TRITC (Sigma). ns between the cells and the microengineered surfaces were visuantly stained to visualize the cells. PVECs were rinsed twice with phosphate buffe I 1640 media supplemented with 10% fetal bovine serum (F 20 g/ml of gentamicin, and 2 g/ml of Fungizone). Cells were incubated at 37C and 5% CO 2 for periods of up to 16 days. Media was replaced every 48 to 72 hours using a maintenance media of RPMI 1640 media supplemented with 4% FBS, 1% L-glutamine and 2% antibiotics. Fluorescent Staining The actin cyto Focal contact adhesio lized by staining vinculin with mouse anti-vinculin primary antibodies (Sigma). Goat anti-mouse secondary antibodies conjugated to Alexa Fluor 488 (Molecular Probes) were used to complete fluorescent staining of the focal contact adhesions. Micropatterns of FN on the substrate surfaces were stained with rabbit anti-FN primary antibodies (Sigma) followed by goat anti-rabbit conjugated to Alexa Fluor 488 secondary antibodies (Molecular Probes). Cell nuclei were stained using 4',6-Diamidino-2-phenylindole (DAPI). At the designated time points, PVECs were removed from the incubator and fluoresce red saline (PBS) and then fixed in 4% paraformaldehyde for 5 minutes. Following fixation, the PVECs were rinsed three times with PBS and then treated with 0.3% Triton X in PBS for 7.5 minutes to increase permeability. PVECs grown on PS, PS_FN, PDMSe_plasma, PDMSe_FN and topographically patterned PDMSe were stained for focal contact adhesions using mouse anti-vinculin primary antibody at a concentration of 400:1 in PBS for 1 hour at 37C. Next, the PVECs were rinsed 5 times in PBS and then

PAGE 99

81 incubated in goat anti-mouse Alexa Fluor 488 secondary antibody at a concentration of 400:1 in PBS for 1.5 hours at 37C. Following incubation with the secondary antibody, the PVECs were washed 5 times with PBS and then the cell actin was stained with phalloidin-TRITC at a concentration of 5 M for 12 hours at 37C. Before the PVECs with the phalloidin-TRITC were placed in the incubator, 4 L DAPI per mL of PBS was added. Following incubation, the PVECs were rinsed 5 times with PBS to remove non-specifically adsorbed phalloidin-TRITC and DAPI and then covered with PBS for imaging. For PVECs cultured on FN micropatterns, the FN was stained in order to visualize how the cells adapted to the underlying chemistry. FN was stained with rabbit anti-FN primary antibody at a concentration of 400:1 for 1 hour at 37C. Next, the PVECs were rinsed 5 times in PBS and then incubated in goat anti-rabbit Alexa Fluor 488 secondary antibody at a concentration of 400:1 in PBS for 1.5 hours at 37C. PVECs stained for FN were then subsequently stained with phalloidin-TRITC for actin and DAPI for cell nuclei as previously described. Imaging Samples were imaged on a Zeiss Axioplan 2 microscope equipped with cence and a digital capture system. Images were acquired at either 100x or 400x epifluores magnification with a resolution of 3900x3090 pixels. Digital images were acquired using an Axiocam digital camera (Zeiss) via the Axiovision 3.1 software and stored at full resolution in the TIFF image format to preserve maximum detail. Each TIFF file was ~35 MB in size requiring an external USB hard drive plugged into the microscope computer in order to remove the files to another PC for post processing. Attempts to store images as JPEG files resulted in loss of detail such as blurring of the actin stress fibers

PAGE 100

82 when using digital magnification. For each fluorescent stain, a separate image was captured and analyzed individually and subsequently combined into an overlaid image. Bright field was used to enhance imaging of the microtopographical features. The bright field images were incorporated into the post processed image to locate cells in relation to the surface microtopography. Image Processing The images were post p rocessed using ImageJ to quantify the stained cellular conversion from pixels to micrometers was accomplished using an opticause ECs have a well developed actin cytoskeleton, the entire components. The l reference standard (model P/N #P-000-0004-0, Pacific Nanotechnology) imaged at 100x and 400x with a maximum 3900x3090 resolution. The generated TIFFs were analyzed in ImageJ using the measure command. This conversion factor was entered into ImageJ and used in all subsequent quantitative calculations. Analysis was performed using the analyze particles command in ImageJ with slight variations to the options depending on the image type. The phalloidin-TRITC stain was used to quantify percent coverage of the ECs within the field of view. Beca cell is easily visible and enhancing the contrast and brightness of the image will produce signal from the entire cell. Figure 5-6 parts A, C and E illustrate the process used to enhance the signal from the actin stain. The TIFF is opened in ImageJ as seen in figure 5-6A and modified with a smooth filter if scan lines from the digital camera capture are present. Next, the contrast and brightness are increased in order make the cells bright red and the surrounding area dark black as shown in figure 5-6C. Then, the image is converted to grayscale by changing it to an 8-bit image. This is converted to a black and white binary image by selecting the “process binary threshold“

PAGE 101

83 command as shown in figure 5-6E. Once the binary image is generated, it can be analyzed using the “analyze analyze particles” command with the maximum particle size set to a value greater than the total image size of 11 mega pixels. The results generated were recorded in excel and the percent coverage was calculated as Equation 5-1 Percent coverage = (A covered by cells )/(A field of view )*100 where a confluent cell layer would have 100% coverage. The mouse anti-vinculin stain was used to quantify the number o f focal contact s illustrated in figure 5-6 parts B, D adhesions produced between the ECs and the substrate. A and F, a similar analysis procedure to that used for the actin stain was employed. Because of the small size of the focal contact adhesions, a smoothing filter was applied a maximum of one time to prevent loss of signal. As seen in figure 5-6B, there can be a lot of background signal that needs to be eliminated to see all the focal contact adhesions. This is done by increasing both the contrast and brightness to make the focal contact adhesions bright green and the surrounding cell body and uncovered areas dark black (figure 5-6D). The image is converted to 8-bit grayscale and then converted to a binary image using the “process binary threshold“ command (figure 5-6F). Focal contact adhesions were counted using the “analyze particles” command setting the minimum particles size as 1 and the maximum particle size as 100. Recorded in Excel are the total number of focal contact adhesions per cell and the average area and standard deviation of the focal contact adhesions.

PAGE 102

84 Figure 5-6. An illustration of the image processing performed in ImageJ in order to quantify the cell properties for the actin (A, C and E) and vinculin (B, D and F) stains. The TIFFs were opened as captured (A and B) and then brightness/contrast enhanced (C and D) to highlight the desired features. For the actin stain, contrast and brightness were increased to get signal from the entire cell (B). For the vinculin stain, the brightness was decreased and the contrast was increased to get signal only from the focal contact adhesions. (E and F) TIFFs were converted to a binary image using the threshold command and then quantified using the analyze particles command. F E D C B A

PAGE 103

85 The DAPI stain was used to visualize the cell nuclei and count the number of cells visible per image. The analysis is similar to that performed for the actin and vinculin stains, with a step added to separate apart adjacent and overlapping nuclei (figure 5-7). The TIFF images were opened in ImageJ (figure 5-7A) and contrast/brightness enhanced (figure 5-7B) to make the nuclei bright blue and the background dark black. TIFFs were then converted to 8-bit grayscale and then to a binary image using the “process binary threshold“ command (figure 5-7C). As seen in figure 5-7C, the close proximity of some cell nuclei results in overlapping of the fluorescent signal and merging of multiple nuclei. To obtain accurate cell counts using the “analyze particles” command these nuclei need to be separated. This is done automatically by using the “process binary watershed“ command which analyzes the shape of round particles and determines where the junctions are. The results of the watershed process is a thin, one pixel wide line that separates the nuclei as shown in figure 5-7D. The cell nuclei are then counted using the “analyze particles” command with a minimum particle size set at 10 and a maximum size set at 1000. Recorded in Excel are the total number of nuclei and the average area and standard deviation of the nuclei. Additional information is calculated by combining data from the individual stained images. The average cell area is found by dividing the area of cell coverage by the number of cell nuclei within an image. Equation 5-2 nucleiofnumberAAimagecell The average number of focal contact adhesions per cell is found by dividing the area of cell coverage by the total number of focal contact adhesions.

PAGE 104

86 Equation 5-3 numberoffocalcontactadhesionsfocalcontactadhesionspercellnumberofnuclei Figure 5-7. An illustration of the image processing used to count cell nuclei and nuclear area. The (A) TIFF in opened in ImageJ and (B) contras/brightness enhanced to maximize signal from all nuclei. The TIFF is then (C) converted to a binary image and merged nuclei are separated (D) using the watershed function. (D) Arrows indicate where the watershed function separates merged nuclei by inserting a thin, one pixel wide line between them. D C B A Statistical Analysis The data for cell density, percent cell coverage, cell spreading and the number of focal contact adhesions per cell was analyzed using SigmaStat 3.0 (SPSS Inc.). The affect of substrate type and culture time on these results were compared using 2-way

PAGE 105

87 ANOVA followed by Tukey’s pairwise comparison between the different means. Statistical significance was based on a P < 0.05. Results and Discussion Surface Analysis of Cell Culture Substrates Topography of flat and micropatterned PS and PDMSe All substrates used for cell culture were analyzed to determine the surface topography and chemistry. AFM was used to characterize the flat and microtopographically modified PS and PDMSe substrates. Analysis showed that there is an inherent surface roughness associated with the flat sections of the different substrates. As seen in figure 5-8A, the PS surface looks like a fibrous mat of crisscrossing strands. This surface structure is thought to be the result of the plasma treatment used by the manufacturer to increase the hydrophilicity of the PS to enhance cell adhesion. Even though the surface is quite irregular, the RMS roughness is still under 10 nm (table 5-2). The PDMSe surfaces are different with changes observed depending on how the samples were cast and the effects of the plasma treatment. Samples cured against the Si wafer had a surface structure that was a negative of the etched wafer. Figure 5-8B shows that this PDMSe surface has no evidence of any fiber-like structures as seen with the PS. Examination of the Si wafer revealed a very similar surface structure (figure 5-9) for the example of the hexagon pattern. In contrast to casting against the Si wafer, air cast PDMSe had a smoother surface characterized by an RMS roughness less than 10 nm (figure 5-8C and table 5-2). Plasma treatment is known to alter the PDMSe surface chemistry by changing it to a silica-like material [74, 75]. This occurs because the plasma ablates the surface and causes chain scission of the PDMSe polymer chains. The result is the formation of a hard

PAGE 106

88 silica-layer with a low surface RMS roughness less than 10 nm (figure 5-8D and table 5-2). Plasma treating the air cured PDMSe does not change the RMS surface roughness. Plasma treating the PDMSe cured against the etched Si wafer decreases the surface roughness to <10 nm, which is approximately the same as PS and air cured PDMSe. The cross-sectional profiles in figure 5-9 show that the roughness of the flat areas is many orders of magnitude below that of the engineered topographies and that they are well shaped with close to 90 angles between the vertical and horizontal surfaces. In addition, all the microfeatures have nearly identical heights of 1.2 m for the large hexagons, 0.5 m for the small hexagons and 2 m for the stars. As a result, all the cell culture substrates have essentially the same surface roughness for the completely flat surfaces and surfaces with flat areas between engineered microtopographies. Figure 5-8. (A) Fluid tapping mode image of PS (625 m 2 area). (B) PDMS elastomer imaged with tapping mode in fluid (100 m 2 area). This sample was the standard Silastic T-2 B A 26.89 nm0.00 n m 21.94 nm0.00 nm D C

PAGE 107

89 Rq= 13 nm 200nm Rq= 13 nm 200nm C B A Figure 5-9. AFM tapping mode topographic images of the hexagonally patterned silicon wafer master. (A) A large 100 m scan size of the hexagonally patterned area. (B) A 5 m scan size enlargement of the area within a single etched hexagon. (C) A 1 m scan size top down view of another etched area. Table 5-2. Variation in RMS roughness of the different materials examined. Material RMS Roughness, R q [nm] Polystyrene 9.1 8.0 Etched silicon wafer 22.5 6.4 PDMSe cured against silicon wafer 26.9 5.6 PDMSe cured against air 3.1 1.30 PDMSe Argon plasma treated 6.8 4.5 PDMSe Argon plasma treated and coated w/ FN 5.4 3.3 PDMSe replicated the microtopographies with high fidelity and the plasma treatment did not degrade these features. Figure 5-10 illustrates the topographies in PDMSe as analyzed by AFM. The important elements to note are the imperfections in the large hexagon sidewalls compared to the nearly flawless sidewalls of the stars (compare figure 5-10 parts A, B and C to part D). This loss of fidelity is not due to the

PAGE 108

90 PDMSe replication but rather the transparency based photomask used to generate the patterns. The toner used to print the transparency has a finite particle size that prevents straight lines to be produced on a scale less than 10 m. In contrast, the photomask generated with the electron beam has no noticeable roughness. As a result, the 20 m hexagons, 40 m stars and 3 m circles in PDMSe all had the high fidelity sidewalls. The topography of the Si wafer was also analyzed by AFM directly. This helped ensure that the fidelity of the PDMSe replication was good. It also determined the nanometer scale roughness of the Si wafer that was replicated by the PDMSe. This was useful because the PDMSe is hard to analyze at the nanometer scale due to adhesion to the AFM tip. Figure 5-9 shows a scan of hexagonal patterned PDMSe wafer at a 2 m depth with high resolution scans showing the flat etched areas with a root mean square (RMS) roughness of 13 nm. This roughness is a product of the DRIE process, which ablates the Si wafer during etching and does not remove material as atomic monolayers. Immunoflourescent staining of FN micropatterns on PS and PDMSe Immunoflourescent staining was used to visualize the FN patterns that were CP onto the surface of the PDMSe. For flat PDMSe and PS surface incubated with FN solutions, the entire stained surface glowed green (figure 5-11D). This is to be expected since the entire surface is coated and control staining PS and plasma treated PDMSe showed only low levels of non-specific staining. All FN micropatterns replicated well on both PS and plasma treated PDMSe. However, the increased hydrophilicity of PS produced more uniform FN coverage across the individual patterns compared to PDMSe where some FN accumulated at the pattern edge. With plasma treated PDMSe, there were always some regions where the pattern transfer was poor or inverted. The

PAGE 109

91 0 20 40 60 80 100 0 0.2 0.4 0.6 A 0.8 1 1.2 X[m] Z[m ] 0 20 40 60 80 10 0 0 0.2 0.4 0.6 0.8 1 1.2 X[m] Z[m] 0 20 40 60 80 100 0 0.2 0.4 0.6 0.8 1 1.2 X[m] Z[m] 0 20 40 60 80 100 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 X[m] Z[m] D C B Figure 5-10. AFM tapping mode images showing the topography of the PDMSe substrates.

PAGE 110

92 inverted pattern seems to result when some hydrophobic recovery of the PDMSe occurs and FN collects between the microfeatures on the stamp and is transferred this way to the substrate. The 3 m diameter circles at 6 m intervals (figure 5-11A) were actually the result of a star patterned stamp an order of magnitude smaller than in figure 5-11C. Due to the small feature size of 4 m stars at 6 m intervals, tailoring the photolithography process allowed these features to be morphed into 3 m diameter circles and used accordingly. Thus, it is valid to call these FN circles, as visualized in the stained micrograph (figure 5-11A). In addition to the intended patterns, there were occasional small spots of highly concentrated FN due to imperfection in the CP process. Figure 5-11. Fluorescent images of CP FN patterns. The patterns are (A) 3 m diameter circles separated by 3 m, (B) hexagons 20 m wide face-to-face, (C) 40 m wide stars and (D) flat continuously coated PDMSe. D C B A

PAGE 111

93 Immunoflourescent Imaging of Cell Structure Qualitative summary of EC response to microengineered surfaces PVECs cultured on all the surfaces were analyzed using immunofluorescent staining and quantified using image processing. Figure 5-12 is a collection of composite images combining a blue DAPI stain for the cell nucleus, a red phalloidin-TRITC stain for the actin cytoskeleton and a mouse anti-vinculin antibody followed by a green secondary antibody stain for focal contact adhesions. For cells grown on FN micropatterns, a rabbit anti-FN antibody followed by a green secondary antibody stain was used to visualize the micropatterns instead of using the mouse anti-vinculin stain. Cells were imaged after culture periods of 4, 7 and 14 days because it provided the best indication of which cells would reach confluence and become quiescent. A couple of cultured substrates were contaminated by bacteria and those squares are filled with a grey square in figure 5-12. The control surface was culture grade PS and cells reached confluence by 1 week. The addition of FN to the surface increased the rate of cell spreading on most substrates. Flat plasma treated PDMSe demonstrated very poor growth with cell shape typically highly elongated and spindle-like. FN coated plasma treated PDMSe had slightly better growth. But the cell morphology still exhibits poor spreading characteristics and low density. The surfaces with the hexagonally patterned topography inhibited cell adhesion. The only areas with cell growth were the flat areas between the 1 m microtopography steps consisting of the raised region surrounding the hexagons. Every sample tested with topography drastically reduced cell density and spreading. Cell apoptosis was complete after 14 days. FN coating did not improve cell growth across the microtopography, but it did increase the number of cells spreading onto the continuous areas. The end result is

PAGE 112

94 that for growing an EC monolayer, topographical steps of 1 m or more disrupt cell spreading and adhesion. Thus the microtopography evaluated is more significant than either the FN chemistry or the plasma generated surface chemistry. This response, however, can not be generalized to all cells because both the topography and protein reaction is specific to the cell type. PVECs grew better on flat surfaces patterned with the FN chemistry and will be examined in detail. EC response to microtopographies and the disruption of cell spreading ECs grown on the microtopographically modified PDMSe substrates did not produce a viable cell monolayer. Cell adhesion and spreading on the surface was disrupted by the 1 m topographical step height. None of the patterns tested were able to support desirable EC growth with measured cell densities far less than that achieved with flat PDMSe_plasma and PDMSe_FN at all culture time points. The addition of FN did increase cell density and spreading, but not enough to improve the poor performance of the surface. Figure 5-13 demonstrates the EC response to 4 different surfaces, where the topography directs cell growth, but not towards a monolayer. As seen in figure 5-13A the ECs group together into multicellular bundles, yet still send out long cytoplasmic protrusions. The extensions are likely probing for more favorable surface conditions that would promote adhesion. They may also be used to help anchor the cell to the surface, as the extreme ends of these extensions stain strongly for vinculin contained in focal contact adhesions. The multicellular bundles occur because the ECs prefer to be in contact with each other rather than the substrate. The addition of FN to the surface (figure 5-13B) did not vastly improve adhesion and spreading on the surface. However, what is clearly

PAGE 113

95 Figure 5-12 continued.

PAGE 114

96 Figure 5-12 continued.

PAGE 115

97 Figure 5-12. Fluorescent images taken at 4, 7 and 14 days for PVECs cultured on all the experimental surface conditions used. noticeably is the surface directed guidance of the multicellular EC bundle around the lowered hexagons in order to stay on the continuous raised area. This is interesting because it demonstrates a strong contact guidance affect of topographical features that are depressed into the surface as compared to those rising up from the surface. While numerous studies have demonstrated contact guidance by raised features, this is the only known study to show this phenomena for depressed features [22, 32, 34, 38, 41, 87-95]. It was initially theorized that the ECs would individually settle into each lowered

PAGE 116

98 hexagon, but this did not happen and very few ECs were ever found there. Lowered hexagons with over twice the area (figure 5-13C) and with FN added (figure 5-13D) did not change this result. With these longer hexagons the ECs still preferred adhering to the continuous raised area with the same contact guidance affect directing the ECs around the hexagons. In most of these cells there was significant colocalization of the cell nucleus with the actin cytoskeleton and vinculin. In fact, the actin cytoskeleton and vinculin are both diffusely stained, indicating that neither actin filaments/stress fibers nor focal contact adhesions are well developed. In addition to the large hexagons lowered into the PDMSe surface, another substrate was fabricated with hexagons 20 m wide face-to-face, raised 1 m from the surface and separated by 2 m face-to-face. An AFM topographic image of this surface is presented in figure 5-14. This same substrate was used as the stamp to CP FN hexagons of the same dimension on PS and PDMSe (figure 5-11B). ECs were cultured on the PDMSe_hex_plasma and PDMSe_hex_FN for periods of 4, 7 and 14 days with extremely poor results. There were sparse cells in completely rounded morphology after 4 days and no evidence of any ECs at 7 and 14 days. Adhered cells were not accidentally removed during the fixation and staining process either because prior observation using an inverted light microscope showed no cells as well. Interestingly, on these hexagon surfaces, ECs were visible as well adhered cells on the flat PDMSe surrounding the patterned area. This provides strong evidence to the conclusion that ECs are extremely sensitive to topography even at the scale of 1 m in the z-axis. The fact that this PDMSe surface with raised hexagons prevented any endothelial adhesion may prove useful for other applications. While there is no apparent need to

PAGE 117

99 prevent EC adhesion, preventing adhesion of other epithelial tissues may be desirable in specific situations. For example, stopping corneal epithelium from binding to ocular implants, stopping the urethral epithelium from binding to urinary catheters and preventing the formation of post surgical adhesion using microtopographically patterned and degradable barrier films. D C B A Figure 5-13. ECs cultured on microtopographically patterned PDMSe for 14 days. Substrates pictured are (A) PDMSe_Hex_00_plasma, (B) PDMSe_Hex_00_FN, (C) PDMSe_Hex_00_plasma and PDSMe_Hex_00_FN.

PAGE 118

100 Figure 5-14. AFM topographic image of PDMSe patterned with 20 m wide raised hexagons. This surface prevented adhesion of ECs with cell growth restricted to the flat surface surrounding the patterned area. Looking at EC adaptation to flat substrates with and without FN PVECs were grown on PS as a control surface because it is widely used to culture and experiment on nearly all cell types. Figure 5-15 provides close-up images of PVECs grown on PS and PS_FN at 4 and 14 days culture. These samples were stained for the nuclei, actin and focal contact adhesions; no staining of the FN was performed because it was a continuous coating. PS_FN demonstrated the most rapid cell spreading of any substrate tested reaching confluence prior to 4 days culture. The PS had reached 90% of confluence by 4 days and looked qualitatively the same after 14 days of culture. There was no obvious change in the number or organization of focal contact adhesions between the culture time points. The results for the FN coated PS were slightly different as the enhanced adhesion of the cell to the surface increased cell spreading. The FN coating decreased cell density while maintaining confluence by increasing individual cell spreading. An interesting result for ECs on PS_FN is that cell morphology changes

PAGE 119

101 between 4 and 14 days culture even though the EC are in a continuous monolayer the entire time. Comparing figure 5-15C to 5-15D, EC shape changes from essentially round to a combination of elongated and round cells. Qualitatively, it appears that the ECs try to continue growing by snaking around other cells as far as possible. It should be noted that even at 14 days of culture; the ECs remained in a monolayer. Cells grown on plasma PDMSe and FN-PDMSe did not adhere in high density or spread well. Figure 5-16 shows a close-up of cells cultured for a period of 14 days and stained for the nuclei, actin and focal contact adhesions; no staining of the FN was performed because it was a continuous coating. The ECs tended to cluster into multicellular bundles and did not form any kind of monolayer or multicellular layers. The individual cells were highly elongated and spindle-like with a cell body aspect ratio of at least 20:1. These cells had very thin sub micron wide pseudopods extending upwards of 25 m from the cell body. Cell density appeared slightly higher on the FN-PDMSe, however, the qualitative results were the same and the cells were abnormal in appearance and likely not functioning properly. There were some general observations of ECs on micropatterned FN surfaces. The cells always appeared to adapt to the underlying chemistry in some manner. Figure 5-17 shows the edge of a growing EC monolayer of cells on a substrate patterned with FN hexagons. In this image, the green stain is for vinculin; however a low level of cross-reactivity with the FN patterns allowed both to be visualized with some additional image processing in Photoshop. The bright yellow color of the advancing front indicates the overlay of actin and focal contact adhesion suggesting the cells are crawling in that direction. Looking at the magnified inset shows the ECs with many of the nuclei located

PAGE 120

102 over a FN hexagon. Closer examination of EC response to each type of FN micropattern on each type of substrate provides more detail. D C B A Figure 5-15. (A) PS 4 day. (B) PS 4 day. (C) PS with FN 4 day. (D) PS with FN 14 day. FN micropatterns on plasma PDMSe and PS had similar results. Given the poor cell adhesion and spreading on PDMSe, cells grown on the FN patterned PDMSe showed an extreme preference for the protein covered areas compared to the bare plasma PDMSe

PAGE 121

103 area. Figure 5-19 provides some close-up images of PVECs on these surfaces and the general results are well illustrated. On the FN stars and FN hexagons, the PVECs adhered preferentially to the protein and the cell body shape contorted to the pattern. Obviously, any cell growing in a shape defined by a simple geometric primitive is not going to support normal cell growth and function. In fact, often cells would become stuck on a single protein pad as in figure 5-19D, although at times they would have the energy to spread on the same pattern as in figure 5-19C. The best pattern proved to be one that had very small protein areas compared to the size of a cell. Figure 5-19A shows PVECs grown for 4 days on 3 m diameter FN circles and the cells spread well. By 7 days cultured cells formed a monolayer on these FN circles using them as anchor points at which to lay down focal contact adhesions. This is the first known example of getting ECs to form a monolayer on PDMSe using chemical micropatterning and is important because it demonstrates this morphology on an elastomer. B A Figure 5-16. (A) PDMSe plasma 14 day. (B) PDMSe plasma with FN 14 day

PAGE 122

104 Figure 5-17. PVECs cultured 2 days on plasma treated PDMSe patterned with 20 m FN hexagons. Hexagons visible from small cross reaction with primary or secondary antibody. This is the edge of the patterned region as the growing monolayer approaches the non-patterned area in black.

PAGE 123

105 Figure 5-18 shows PVECs growing on PS with FN stars at 4 and 14 days of culture. The cell actually tries to adapt its shape to that of the FN micropattern, and be seen clearly in figure 5-18A. Cells still reach confluence as seen in figure 5-18B, however, the shape is still influenced by the underlying pattern and cell density is lower than that found on PS and FN_PS. While the ECs can adapt to the pattern, it does disrupt cell adhesion and spreading reducing the utility as an EC tissue engineering material. B A Figure 5-18. PVECs grown on PS patterned with FN stars for periods of (A) 4 days and (B) 14 days in culture. Quantifying The Cell Density, Percent Coverage And Cell Area Cell density as function of time indicates how cells are proliferating on the different substrate types. From the qualitative results shown in figure 5-12, it is obvious that some surfaces promote excellent cell growth while others results in minimal growth. The surfaces with 1 m high microtopographies with and without FN coatings were not analyzed further because the ECs did not grow on them. Therefore, the surfaces of interest were the flat PS and PDMSe substrates and the FN micropatterns generated on

PAGE 124

106 these flat surfaces. Figure 5-20 shows the change in cell density with culture time on different substrates. See Appendix A for a complete statistical analysis of cell density as a function of substrate type and culture time. D C B A Figure 5-19. PVECs are grown on FN micropatterned plasma PDMSe. The micropatterns are (A) 3 m diameter circles, (B) 40 m stars and (C and D) 20 m hexagons. The TIFFs analyzed were all captured at 100x with a field of view of 1,527,295 m 2 . These results only indicate the cell density on a surface and do not completely

PAGE 125

107 describe how the cells are spreading. ECs on PS and PS_FN were consistently in the range of 500 to 700 cells/mm 2 at all time points with no statistical difference between them. Cell density on PS and PS_FN was significantly different than every other surface treatment except for PDMSe_FN_star. Similar to the PS, the PDMSe_plasma and PDMSe_FN were not significantly different even though there was a significant difference between these two surfaces and the FN patterned PDMSe substrates. ECs on PDMSe_FN_hex and PDMSe_FN_star formed confluent patches in small areas never larger than half a field of view. Thus, cell densities were low and the inconsistent growth suggests that these surfaces were not suitable for the entire EC population or for robust cell division and growth. The highest cell density was seen for PDMSe_FN_circle, which continued to increase in number at each culture time point reaching 1536 247 cells/mm 2 by 14 days. The cell growth on PDMSe_FN_circle was significantly higher than all other substrates including the PS and PS_FN controls. The continued increase in cell density for ECs cultured on PDMSe_FN_circle, even though confluence is reached between 4 and 7 days, demonstrates the proliferative capacity of this surface. This is not expected because only 20% of the surface is coated with FN, while the rest is plasma treated PDMSe which demonstrated poor cell adhesion, spreading and growth. Apparently, the absence or presence of an adhesion ECM protein is not the only factor needed to adhere cells. It seems that the distribution of that protein, FN in this case, also plays an important role as the PDMSe_FN_hex and PDMSe_FN_star surfaces failed.

PAGE 126

108 Figure 5-20. A plot of cell density on different engineered surfaces at 4, 7 and 14 days culture time. Represented are cell numbers per 1,527,295 m 2 , the field of view at 100x, error bars represent one standard deviation. Cell Density as a Function of Surface and Culture Time02004006008001000120014001600180020004714Cutlure Time [days]Cell Density [cells/mm^2] PS PS_FN PS_FN_star PDMSe_plasma PDMSe_FN PDMSe_FN_star PDMSe_FN_hex PDMSe_FN_circle The cell density of ECs on PDMSe_FN_circle was plotted against a linear scale of culture time to look at the proliferation rates (figure 5-21). The initial number of cells at zero days was calculated by taking the seeding density of 50,000 cells/mL, multiplying by the seeding volume of 1.5 mL, dividing by total area of the Petri dish of 962,112,750 m 2 and then multiplying by the field of view of 1,527,295 m 2 . This results in an average of 119 cells settling out of solution within the 100x field of view at 2.5 hours of the seeding. The experimental data can be interpreted by comparing it to fermentation growth rates. In binary growth, there is an initial lag as sufficient cell numbers are built up. After this point the growth rate becomes exponential and the number of cells increases rapidly. This is followed by a rate limiting step that slows the growth rate and

PAGE 127

109 causes the number of cells to stabilize. In the case of bacteria, the rate limiting step is usually nutrient supply, however for properly functioning ECs the rate limiting step is contact inhibition. Applying this model to the data (figure 5-21), to lag and subsequent exponential growth occur in the first 4 to 5 days. After this point, as the ECs reach confluence, the cell growth rate drops off. The complex growth characteristics prevents an exact fit to the data, but qualitative interpretation indicates that the cell number is unlikely to exceed 3000 cells within the 1,527,295 m 2 field of view (a cell density of 1964 cells/mm 2 ). Figure 5-21. Plotted is the number of cells on PDMSe_FN_circle as a function of culture time. The dashed line represents a standard binary growth curve adapted from bacteria fermentation. Error bars represent one standard deviation and cell number is given per 100x field of view (1,527,295 m 2 ). Cells on PMDSe FN Circles with Time050010001500200025003000024681012141Culture Time [days]Cell Number 6 1. Initial growth lag 2. Exponential growth phase 3. Growth decay rate limited bcontact inhibition y 3 1 2

PAGE 128

110 While the increase in cell density with culture time is important, it needs to be compared to the rate at which the ECs form a confluent monolayer. This can be visualized by plotting the percent coverage of the substrate as function of cell culture time (figure 5-22). See Appendix A for a complete statistical analysis of percent coverage as a function of substrate type and culture time. The time to reach confluence depends on both the proliferation rate of the cells and the area each cell occupies. From figure 5-22 it becomes evident that there are two general responses seen. Both the PDMSe_plasma and PDMSe_FN fail to reach confluence achieving ~40% coverage at best with no significant difference between them. The PS surfaces and the FN patterned PDMSe surfaces are covered more rapidly with all having significantly higher percent coverage at early time points. The cells achieve confluence on the PS_FN substrate by 4 days culture proving that this surface has excellent growth characteristics, thought it is not significantly different than PS and PDMSe_FN_circle. The remaining FN patterned PS and PDMSe substrates stimulated similar initial cell growth rates. Confluence was only reached on the PS_FN_star surfaces. Apoptosis occurred after 7 days and 14 days on the PDMSe_FN_hex and PDMSe_FN_star surfaces, respectively. The fact that apoptosis occurred for ECs on the PDMSe_FN_star and PDMSe_FN_hex indicates that the area, and possibly shape of the FN pattern is vitally important. In terms of cell spreading and adhesion to a surface, one of the most important factors is cell area. This is calculated by dividing the area of substrate coverage by the number of cells to determine the average area occupied per cell, per substrate per culture time point (figure 5-23). See Appendix A for a complete statistical analysis of cell area as a function of substrate type and culture time. A cell area that is too low would suggest

PAGE 129

111 Figure 5-22. A plot of percent coverage of ECs on the different engineered substrates as a function of culture time. Error bars have been omitted to prevent clutter. There are no data points for PDMSe_FN_hex at 7 and 14 days or for PDMSe_FN_star at 14 days because cell cultures had died. Cell Coverage with Culture Time0204060801001200246810121416Culture Time [days]Persent Coverage PS PS_FN PS_FN_star PDMSe_plasma PDMSe_FN PDMSe_FN_star PDMSe_FN_hex PDMSe_FN_circles that the cell is poorly adhered to the surface while a high cell area would suggest that the cell is pinned down to the surface. Analysis of the fluorescently stained, fixed sections of porcine pulmonary artery reveals an average EC area of 316 35 m 2 . Thus, the ideal synthetic surface will direct ECs to adhere and grow at similar cell areas and densities. On PS and PS_FN, the cell area remained the same throughout culture at ~1500 m 2 with no significant difference between the surfaces. ECs on PDMSe_plasma had a significantly lower cell area than on PDMSe_FN at all time points yet both demonstrated a peak in cell area at 7 days that is significantly higher than cell area 4 and 14 days. This is thought to represent a maximum effort by the cells to spread onto to PDMSe before

PAGE 130

112 failing, and then retracting. One substrate, PS_FN_star had continued cell spreading throughout all time points and was significantly greater than all other surfaces. This is likely the result of the ECs being able to reach across to neighboring FN stars, spreading fully, and then extending to another star. This is actually an unwanted result because the ECs spread far beyond the normal in vivo size. ECs on both PDMSe_FN_star and PDMSe_FN_circle decreased in cell area with culture time with no significant difference between them. However, while the number of cells on PDMSe_FN_star decreased as well, the number of cells on PDMSe_FN_circle increased dramatically over the same time period (figure 5-20). Evaluating EC response to the different engineered surfaces shows that a number of properties are important in evaluating the relative performance. Comparing cell density, percent coverage and cell area demonstrates that the PDMSe_FN_circle substrate produces an EC monolayer that most closely resembles the ECs on the arterial lumen. At 14 days, this surface is confluent with an average cell area of 617 39 m 2 , or roughly twice the 316 35 m 2 found for the porcine artery. This is the best result achieved and suggests that small FN pads designed to direct the formation of adhesions to the surface are the best designs. The importance of the FN micropatterns for directing adhesion, and specifically the formation of focal contact adhesion to the substrate is examined further to answer this question. Quantifying Focal Contact Adhesions To The Substrate Analyzing how focal contact adhesions form between ECs and the engineered substrates is necessary to understand the resulting cell behavior. The first variable looked

PAGE 131

113 at was the variation in the size of focal contact adhesions as a function of the surface on which they were formed. Figure 5-24 looks at the variation in focal contact adhesion diameter between ECs and PS, PS_FN, PDMSe_plasma and PDMSe_FN. There were no statistical differences in the average diameter, i.e., 1.86 0.35 m 2 of the focal contact adhesions and the variation in cell spreading is due to the formation of additional adhesions. This suggests that any variation in size was less than the detectable limit and less than a 20% variation of average diameter of the focal adhesion.. Figure 5-23. A plot of the average cell area as a function of the engineered substrate ECs were cultured on and the length of time they were cultured. Cell areas need to be compared with cell density and cell morphology to understand if the response is favorable or not. Error bars represent one standard deviation. Change in Cell Area with Culture Time0100020003000400050004714Culture Time [days]Cell Area [um^2] PS PS_FN PS_FN_star PDMSe_plasma PDMSe_FN PDMSe_FN_circles PDMSe_FN_star

PAGE 132

114 Figure 5-24. A plot of the variation in the diameter of focal contact adhesions between ECs and engineered substrates. For each substrate type, values are pooled from 4, 7 and 14 days culture. Error bars represent one standard deviation. Focal Contact Adhesion Area on Different Substrates00.511.522.5PDMSe_FNPDMSe_plasmaPSPS_FNSubstrateFocal Contact Adhesion Area [um^2] Figure 5-25 shows the number of focal adhesions per cell that formed between the ECs and the PS, PS_FN, PDMSe_plasma, PDMSe_FN and PDMSe_FN_circle substrates with a clear difference depending on surface properties. Because FN contains the –RGDamino acid sequence known to bind integrin receptors responsible for forming focal contact adhesions, it is reasonable that the FN coating increased the number of contacts formed for PS and PDMSe. PS_FN showed a significant increase in focal contacts at an average of 90 8 per cell compared to the 44 10 per cell for PS at 4 days. Interestingly, the number of focal contact adhesions on PS_FN dropped to 28 6 and 35 10 per cell at 7 and 14 days respectively. This is likely due to a decrease in the number of adhesions needed to stabilize the cell since the EC monolayer is well formed and the

PAGE 133

115 cells have stopped proliferating and spreading. Cells on PDMSe_FN showed a trend towards twice the number of focal contact adhesions as compared to cells on PDMSe_plasma though there was no statistical difference at any time point. The low number of focal contact adhesions for PDMSe_plasma and PDMSe_FN is likely due to the overall poor adhesion and spindle-like shape that prevented binding to the surface. The PDMSe_FN_circle substrate elicited nearly the same number of contact adhesions as the PS at 4 days and the PS_FN at 14 days. Even though the number of cells increased and the average cell area decreased, the number of focal contact adhesions per cell remained relatively constant at ~40 per cell with no statistical difference between time points. Focal adhesions were not examined for PDMSe_FN_star and PDMSe_FN_hex due to apoptosis by 14 days. Figure 5-25. A plot of number of focal contact adhesions per cell as a function of the engineered substrate ECs were cultured on and the length of time they were cultured. Error bars represent one standard deviation. Focal Contact Adhesions per Cell0204060801001204714Culture Time [days]Focal Contact Adhesions PDMSe_FN PDMSe_plasma PS PS_FN PDMSe_FN_circles

PAGE 134

116 Looking at the arrangement of focal contact adhesions within the cells reveals some additional information about cell adhesion. Figure 5-26 shows a close-up centered on an EC from each surface analyzed for adhesions in figure 5-25. For surfaces treated with FN as a continuous coating or a patterned area, focal contact adhesions appear throughout the cell, and most notably underneath the nucleus (figure 5-26 A, C and E). In contrast, the PS and PDMSe_plasma samples had focal contact adhesions organized mostly at the periphery of the cell (figure 5-26 B and D). It should be noted that cells pictured are from 4 days culture time and that changes in cell area and focal contact adhesions per cell change with time as described above. Furthermore, only well spread cells were examined representing a minority of the EC population on PDMSe_plasma and PDMSe_FN where the cells tended to become spindle-like and bundle together into multicellular structures. By examining cells at only 4 days culture, this grouping behavior of ECs on PDMSe was minimized since it became more prevalent with longer culture times. Looking at the images directly, the PS (figure 5-26D) focal contact adhesions are arranged mostly at the periphery of the cells with some additional adhesions dispersed underneath the cell body. Furthermore, the actin cytoskeleton appears well developed with robust staining throughout most of the cytoplasm. When the PS is pretreated with FN, the PS_FN substrate elicits a much stronger adhesions with a visible increase in the number of focal contact adhesions throughout the cell with an apparent uniform distribution (figure 5-26E). Adhesion, spreading and proliferation were so rapid on the PS_FN that a complete monolayer was formed at 4 days culture and the EC had to be examined in contact with neighboring cells. In addition, some of the focal contact adhesions on the PS_FN appear as larger green spots. This is due to the brightness/contrast enhancement used to prepare

PAGE 135

117 the image for presentation causing the merger of the fluorescent signal from adjacent adhesion locations. This suggests that more adhesions formed in close proximity than were needed to adhere the cell to the substrate. Likely, this is the reason the number of focal contact adhesions decreased with culture time for ECs on PS_FN. EC adhesion and spreading on the PDMSe_plasma was the lowest seen for any of the flat surfaces tested. Focal contact adhesions were organized almost entirely near the periphery of the cell and the actin cytoskeleton was diffuse with very few brightly stained lines indicative of stress fibers (5-26B). This low number of adhesions coupled with a poorly developed cytoskeleton suggests that on the PDMSe_plasma the ECs have poor mechanical stiffness. Also visible in figure 5-26B is a cytoplasmic extension running out of the image area. These pseudopods were a common occurrence on the PDMSe_plasma and PDMSe_FN and could run many 10’s of micrometers in length. ECs on the PDMSe_FN had more focal contact adhesions than plasma treatment alone, however adhesion was still poor (figure 5-26A). Additional focal contact adhesions were evident under the cell body, usually organized directly under the cell nucleus as seen in figure 5-26A. The cells on PDMSe_FN appeared to have more brightly stained stress fibers in the cytoskeleton. The larger cell area and greater number of focal contact adhesions of ECs on PDMSe_FN versus PDMSe_plasma suggests better adhesion to the surface, though at all culture time points there was no statistical difference in the percent coverage of the substrate at ~40%. This is because the cells preferred to bundle into multilayers rather than spread out into a monolayer. The PDMSe_FN_circle had a unique result compared to the other engineered surfaces (figure 5-26C). The number of focal contact adhesions remained constant throughout the culture time even though the cell area decreased

PAGE 136

118 throughout that same period. The organization of the focal contact adhesions through out the cell were relatively uniform, likely due to integrin binding to the regularly spaced underlying FN micropattern. The actin cytoskeleton was also highly developed with many of the stress fibers appearing to run in between nearby focal contact adhesions. Also, many of the actin stress fibers appeared to run in two, over laid 2-fold symmetry axes. To clarify, the stress fibers appeared to align to the underlying FN micropattern either aligning to the sides the diagonals of the square lattice. The PDMSe_FN_circle surface appears to be the best option because it results in the highest cell density with a cell area close to that seen for ECs in natural artery and a well developed actin cytoskeleton. However, to understand why this is happening requires a more detailed analysis of ECs on the PDMSe_FN_circle substrate. While the focal contact adhesions and underlying FN micropattern could not be stained simultaneously due to a lack of the appropriate fluorescent filter (figure 5-27A), it is still possible to determine if the focal contact adhesions are co-localized with the micropattern. This is done by overlaying a grid with 6 m line spacing on top of the fluorescent image. From figure 5-27B, it is apparent that the focal contact adhesion do line-up along lines that are spaced at 6 m intervals. The alignment is not perfect due to the fact that each FN circle is 3 m in diameter and the overlaid line is less than 1 m wide at the image scale.

PAGE 137

119 E D C B A Figure 5-26. A close-up of an EC cultured on different surfaces showing the variation in the number and arrangement of focal contact adhesion (green immunofluorescent stain for vinculin) and organization of the actin cytoskeleton (red phalloidin-TRITC fluorescent stain for actin). ECs are cultured for 4 days on (A) PDMSe_FN, (B) PDMSe_plasma, (C) PDMSe_FN_circle, (D) PS and (E) PS_FN.

PAGE 138

120 Figure 5-27. Examining the alignment of focal contact adhesions of ECs on PDMSe_FN_circle to the underlying FN micropattern. The images are In addition to organization of the focal contact adhesions to the underlying FN micropattern, the actin filaments and stress fibers appear to arrange in a similar manner. Figure 5-28 shows a close-up of an EC on the PDMSe_FN_circle surface where the FN patterned has been stained instead of the focal contact adhesions. Yellow lines have been overlaid on the image and aligned to some of the actin fibers. The actin fibers appeared to stretch from one focal contact adhesion to another as seen in figure 5-27. Since these focal contact adhesions appear to be well aligned to the FN circle micropattern, it is reasonable that the actin fibers have the same arrangement. Actin fibers at the periphery seem to be the most well aligned with the actin fibers around the nucleus less clear in their alignment. In terms of cell function, causing the actin cytoskeleton and focal contact adhesions to align to such a regular square lattice array is undesireable. However, incorporation of some patterns variability, possibly generated through a noise factor in A B

PAGE 139

121 generating a new micropattern design, may provide enough discontinuity while still engineering the adhesion to the surface. Figure 5-28. Lines overlaid on top of an image of an EC on the PDMSe_FN_circle surface to show alignment to the underlying FN micropattern. Conclusions These results demonstrate that micropatterns of both topography and chemistry can change the way cells grow on a surface. While the goal of this research is to enhance EC monolayer growth, it was found that the topography tested actually disrupts EC growth. This may indicate why traditional small diameter vascular grafts fail due to failure of ECs to grow on the inner woven surface with topographical height changes in the range 10 to 100 m. The chemical micropatterning proved to be effective in the absence of topography for directing cell growth and allowing ECs to form a monolayer

PAGE 140

122 on the PDMSe. Clearly, the size and shape of the micropattern is critical in controlling this adhesion. For example, the 40 m stars and 20 m hexagons altered the entire cell shape into an unnatural morphology. However, the small 3 m diameter circles actually enhanced normal spreading by providing a regular location for focal contact adhesion formation to the substrate.

PAGE 141

CHAPTER 6 CELL RESPONSE TO ENGINEERED SURFACES ANALYZED BY ATOMIC FORCE MICROSCOPY Introduction The ability to image live cells using the AFM provides a powerful technique for quantifying a variety of biological properties. By physically probing the surface, tip/sample interactions can be analyzed in order to determine morphology, mechanical properties, cytoskeletal structure, cell surface markers and other properties. In addition to these capabilities, the AFM can operate in a liquid environment leading to a number of advantages. In liquid, the medium surrounding the tip and sample is well controlled eliminating the capillary forces that occur in air, which would rip apart a cell. By controlling the force between the AFM tip and sample to a few nanoNewtons, damage to cells can be avoided. More importantly, operating in liquid allows measurements on intact cell to be performed. This improves biological relevance because the cells have not been altered physically or chemical by exposure to stains or fixatives as used for optical and electron microscopy. In this chapter AFM is used to quantify the morphology, cytoskeletal structure and mechanical properties of the ECs on PS and PDMSe surfaces. While ECs on these surfaces have been evaluated with fluorescent microscopy, AFM provides valuable information to expand ones understanding of how the cells interact with the substrate. Measuring the cell morphology adds a critical 3 rd dimension to the data allowing calculation of cell volume and its relation to cell adhesion and normal cell function. 123

PAGE 142

124 Coupled to the morphology is the organization of the actin cytoskeleton as visualized through deformation of the cell membrane during contact mode imaging. Mechanical properties are determined by analyzing force curves generated from force volume imaging. These force curves provide data on the contact between the tip and sample as well as any adhesions or repulsions during the approach and retract to the surface. Fitting the data in these curves to mathematical models results in the ability to calculate the elastic modulus of the substrate. In addition, the shape of the force curves can reveal information about the viscoelastic properties of the material. By using the AFM in simulated physiologic fluid, the mechanical properties of ECs can be determined as a function of the underlying substrate and the length of culture. Materials and Methods Substrate Production Silastic T2 (Dow Corning) PDMSe was mixed in 10:1 ratio by mass of base resin to curing agent and mixed thoroughly. To remove air bubbles, the PDMSe was degassed for 20 minutes under vacuum and then allowed to sit at atmospheric pressure for another 5 minutes for all bubbles to dissipate. PDMSe samples for cell culture were created by curing 2 to 3 mL of PDMSe prepolymer in 35 mm diameter Petri dishes. The PDMSe was allowed to cure for 24 hours at room temperature creating a uniform and flat PDMSe surface.

PAGE 143

125 ECs and media Flat PDMSe film Petri dish (35 mm diameter) Figure 5-4. Samples for cell culture are prepared by placing a cured micropatterned substrate in the center of a 35 mm diameter Petri dish and curing in place by surrounding it with PDMSe prepolymer. ECs in media are poured into the remaining volume of the Petri dish and cultured for 4, 7 or 14 days. The hydrophobic PDMSe was surface modified to increase the hydrophilicity for cell adhesion. The PDMSe films, fixed in the Petri dishes, are plasma treated with an Argon radio frequency glow discharge (RFGD) plasma. The plasma treatment system used is a gas flow system where the sample sits in a bell jar with a gas inlet column at the top and a vacuum outlet at the bottom. The gas inlet column has an RF coil surrounding it that ionizes the argon gas as it flows through and strikes the samples in the bell jar. Three 35 mm diameter Petri dishes were placed in the bottom half of a 9 cm diameter Petri dish and placed on a metal stand 2.5” below the RF coil in the bell jar. The pressure was reduced to 50 mTorr and then the chamber was purged 3 times with argon at a flow rate of 1000 sccm. PDMSe samples were exposed for 1 minute to plasma generated with an RF power of 50 W and an argon flow rate of 8 sccm at 50 mTorr. The samples were removed from the bell jar and covered with sterile Hanks balanced salt solution (HBSS) immediately after the plasma treatment. Samples were also incubated with a solution of 50 g/mL fibronectin (FN) in HBSS for 1 hour. The FN treated samples were subsequently washed 3x with HBSS and then seeded with PVECs.

PAGE 144

126 Cell Culture Primary porcine vascular endothelial cells (PVECs) were obtained from the research group of Dr. Edward Block at the Malcolm Randall VA medical center (Gainesville, FL). PVECs were provided in suspension at an approximate concentration of 200,000 cells per mL and diluted to 50,000 cells per mL for seeding on microengineered surfaces. The cell were counted on a hemocytometer (Sigma) and suspended in additional media to reach the desired cell concentration. Cell seeding media consisted of RPMI 1640 media supplemented with 10% fetal bovine serum (FBS), 1% L-glutamine and 2% antibiotics (100 U/ml of penicillin, 100 g/ml of streptomycin, 20 g/ml of gentamicin, and 2 g/ml of Fungizone). Cells were incubated at 37C and 5% CO 2 for periods of up to 16 days. Media was replaced every 48 to 72 hours using a maintenance media of RPMI 1640 media supplemented with 4% FBS, 1% L-glutamine and 2% antibiotics. AFM Experimental Setup and Procedure AFM setup The AFM analysis was performed using standard imaging techniques. All imaging was performed on a Dimension 3100 AFM (Veeco Metrology) equipped with a Nanoscope IIIa controller and version 5.12r3 software. Imaging was performed in Hanks balanced salt solution (HBSS) using a liquid tip holder. For all imaging modes, 200 m long and thin silicon nitride cantilevers (Veeco Metrology) with 0.06 N/m spring constant were used. Contact mode was used to determine surface topography and force volume mode was used to map the mechanical properties. For scanning, the AFM scanner head was lowered directly into a 35 mm diameter Petri dish with ECs growing on the bottom.

PAGE 145

127 Figure 6-1A shows a schematic of the engaged scanner and figure 6-1B is a digital picture of the AFM tip over a monolayer of ECs as viewed by the AFM video camera. Figure 6-1. The AFM tip is lowered directly into a Petri dish with cultured ECs on the bottom. (A) A schematic showing how the AFM scanner is dunked into to Petri dish until contact with the substrate or cells is achieved. (B) Frame capture from the video camera on the AFM showing the top of the AFM cantilever positioned over a confluent EC monolayer. B A 80 m AFM operation To obtain the force curves used to measure the elastic modulus the AFM was operated in a force volume mode. This allowed for both the topographic image and the force data to be obtained in a single scan. The Z range ramp size varied from 2 to 3 m depending on the height of the cells on the substrate. The frequency of the z-axis was kept to 2 Hz, which maximized imaging speed while minimizing hydrodynamic drag on the cantilever . The image scan size was varied from 1 m 2 to 10,000 m 2 . The image resolution was either 64x64 or 128x128 pixels. Force curves were taken at a resolution of 256 data points per curve. File size limitations restrict the maximum force volume file to 1 MB, so increasing the image resolution requires a decrease in the force curve resolution. Due to the individual force curves taken for each individual image pixel of force volume image, the imaging times for the two resolutions were 2 hours and 6 hours respectively. This prolonged imaging time is the reason force volume was not used in

PAGE 146

128 liquid where concern for evaporation, moisture build-up and corrosion existed. The final data type returned was the actual force volume data set, which was a 2-dimensional slice of the force data. AFM data analysis Contact mode images were post processed using WSxM software (Nanotec Inc.). This freeware application opens a variety of SPM data formats and has enhanced 3-dimensional rendering capability as compared to the Nanoscope v5r12 software supplied with the Dimension 3100 AFM. All the contact mode topography images were opened in WSxM and then flattened as needed. A 0 th order flatten was always performed normalizing the relative position of each scan line to its neighbors so that a continuous surface is formed. This is a standard practice as an arbitrary Z-axis displacement exists for each scan line and this must be normalized to recreate the 3-dimensional surface. In addition to the 0 th order flatten, a 1 st order flatten was applied to those images where an obvious tilt to the surface existed. Force volume images were analyzed using two different processes. The topographic image produced by the force volume mode was analyzed using the WSxM software by the same procedure just described for the contact mode topographic images. Analysis of the individual force curves of the force volume image were performed using a combination of software. The Nanoscope v5r12 software (Veeco Metrology) was used to open the force volume dataset and save individual force curves as ASCII text files. These text files were then opened in the WSxM software as force curves and the slope of the contact region, non-contact deflection, maximum deflection and etc were recorded in an excel spreadsheet. From these values, the elastic modulus of the cell was calculated using the appropriate model in Excel as described in Appendix A.

PAGE 147

129 Results and Discussion EC topography on PS and PDMSe In these studies, PVECs were imaged in HBSS to maintain the cells, but also to eliminate any proteins in solution that would foul the AFM tips. Cells were also imaged at ambient temperatures of 22C in order to slow down the cell metabolism and increase cell lifespan while being imaged. Contact mode topographic imaging proved most effective for imaging cell morphology and the cytoskeletal architecture. Figure 6-2 shows topographic images of PVECs grown on PS and PDMSe_FN. Unfortunately, due to the poor adhesion of PVECs to the plasma PDMSe, cells could not be imaged on that surface. Comparing figure 6-2A and 6-2C shows a change in EC shape from round to highly elongated. This is an identical response as that seen with fluorescent microscopy for PVECs on PS at 4 and 14 days culture as shown in figure 6-2 B and D respectively. Cells grown on the PDMSe_FN (figure 6-2E) also have a morphology that was quite similar to that seen with fluorescent microscope (figure 6-2F). The cells were grouped into mutlicellular bundles and were highly elongated and spindle-like. These cells were difficult to image because they adhered poorly to the PDMSe_FN. Also, the mutlicellular bundles that formed would sometimes exceed the 6 m z range of the AFM prohibiting imaging of that area. As a result, the image for PVECs on PDMSe_FN represent the subpopulation of well adhered cells in bundles < 6 m high.

PAGE 148

130 Figure 6-2. AFM contact mode topographic images of PVECs imaged while living in HBSS with corresponding fluorescent optical micrographs of PVECs cultured under the same conditions. PVECs cultured of PS for 4 days and imaged by (A) AFM and (B) fluorescence. PVECs cultured on PS for 14 days and imaged by (C) AFM and (D) fluorescence. PVECs cultured on FN coated PDMSe for 4 days and imaged by (E) AFM and (F) fluorescence. F D B E C A

PAGE 149

131 Mechanical Properties of ECs In order to determine elastic modulus of the PVECs, force volume mode imaging was used and each force curve was individually analyzed. Typically, an area of interest was first imaged in contact mode for topography and then subsequently imaged in force volume mode. A force volume topography based as an iso-force surface through the force volume served as a comparison to ensure no cell degradation occurred (figure 6-3B). A relative measure of cell elasticity at a given height through the sample is provided by slices through the force volume. Demonstrated in figure 6-3A, it shows difference in slope of the force curves increasing from dark to light as slope increase from low to high. From the figure it is clear that cells are stiffer at the raised nucleus and the nearby cell body, likely due to the thick stress fibers visible in fluorescent microscopy and the structure of the nucleus. Elastic modulus drops towards the cell periphery. No PS substrate is exposed because it would be the brightest thing in the image since its elastic modulus is at least 3 orders of magnitude larger than any part of the cell. This discrepancy between substrate and cell elastic modulus can be seen in figure 6-2C where the force curve for bare PS has a sharp and constant change in slope as the AFM tip contacts the substrate. In contrast, the force curves that contact regions of the cell have a gradual change in slope due to the viscoelastic properties. It should be noted that the force curves for PS and cells are scaled differently with a deflection axis of 75 nm/div and 30 nm/div respectively. The difference between contact mode and force volume mode is well illustrated by comparing 3-dimensional topographic images produced by both methods (Figure 6-4). The most obvious difference is the resolution, which is 2 to 4 times higher for contact mode. As result of this difference in resolution, the 3-dimensional renderings of the force

PAGE 150

132 volume image are smoothed using the built in functions in the WSxM software. Comparing figure 6-4 parts A and C to parts B and D, it becomes apparent that detail of the underlying cytoskeleton is lost. It is lost, because it does not contribute because the difference in elastic modulus between the actin fibers and surrounding cell body is less than the difference in height. Figure 6-3. A comparison between contact mode and force volume images. A) An example of a topography image taken as a iso-force surface and a force image taken as a z-axis slice. B) An example of topography of the same region of cells imaged in contact mode and force volume. C) Example force curves comparing PS to PVECs grown on PS. By analyzing the force curves, it is possible to determine the elastic modulus as well as the viscoelastic response for different regions of the cell and substrate. In figure 6-4B, three regions are identified highlighting 1, a cell nucleus, 2, the cell periphery and 3, the substrate. Force curves for each region are plotted together in figure 6-4E with a different color for each type. Qualitatively, there are differences between the curves with

PAGE 151

133 region 1 force curves demonstrating a smaller slope with a large historesis between the approach and retract portions of the force curve. This historesis is an indicator that there is energy lost from viscous flow. Looking at the area of the historesis, it is apparent that the nucleus of region 1 has the largest viscous component with less for the cell periphery in region 2 and none for the substrate in region3. Analyzing the force curves using the modified Hertz model described in Appendix A leads to quantification of the elastic modulus for each region. As reported in table 6-1, there are distinct differences in the mechanical properties of each region. The cell nucleus of region 1 has the lowest elastic modulus, which is significantly different than the other regions (P<0.05). The elastic modulus increases for the cell periphery in region 2 and the substrate in region 3, however the difference between them is not significant. Given the prevalence of visible actin fibers in the contact and force volume topographic images, an attempt was made to see if there were measurable differences in the mechanical properties. In figure 6-4D, two regions are identified highlighting region 1, an actin stress fiber and 2, an adjacent area of the cell membrane. Force curves from each region are plotted together in figure 6-4F with no apparent difference. The shape of the curves as well as the historesis are the same and there is no statistical different between the two. It can be concluded that over short distances of 5 to 10 m, the cytoskeleton produces an overall cell mechanical stiffness that is homogeneous.

PAGE 152

134 3 2 1 5 4 D C B A Force Curves Comparing Cell to the Substrate-345-295-245-195-145-95-4551315429543657771885910001141128214221563170418451986212722682409255026912832297331143255Separation [nm]Deflection [n m Force Curve Variation at Small Distances-345-295-245-195-145-95-4551318134951768485210201188135615241692186020272195236325312699286730353202Separation [nm]Deflection [nm] 5 4 1 2 3 E F Figure 6-4. Comparison of the same region of ECs on PS imaged in both contact mode and force volume mode. A large scan showing the topography of ECs and the underlying PS (4 days culture) in both (A) contact and (B) force volume mode. A small scan looking at stress fibers on a single EC imaged in both (C) contact and (D) force volume mode. The force curves from the regions highlighted in parts (B) and (D) are plotted in parts (E) and (F) respectively.

PAGE 153

135 Table 6-1. Elastic modulus for the different regions identified in figure 6-4. Region Elastic Modulus [kPa] 1 5.2 0.4 2 22.3 6.0 3 60.7 2.5 4 67.1 1.3 5 73.8 0.9 The real power of the AFM techniques comes at looking at difference in cell properties as a function of the substrate on which the cells are grown. Fluorescent microscopy revealed differences in cell shape and cytoskeletal structure. Contact mode AFM topography images confirmed the cell shape cytoskeletal information while adding important data on cell morphology in the 3 rd (z-axis) dimension. From these experiments cell volume and the 3-dimensional path of actin stress fibers can be calculated. Mechanical properties also vary depending on the substrate upon which PVECs are cultured. Figure 6-5 provides clear evidence supporting this observation. In figure 6-5A, the topography reveals PVECs on PDMSe_FN with spindle-like morphology and multilayering as visible on the right side of the image. Representative force curves were selected at a region directly above the cell nucleus, at the periphery of the cell between adjacent cells and over the cytoplasm between the nucleus and periphery. For comparison, PVECs growing on PS at a similar time point were examined as seen in figure 6-5B. These PVECs have a round morphology typical of PVECs on PS at short culture times, some of these cell will begin to grow further and significantly elongate by 14 days in culture. Select force curves were examined for this sample at the same representative locations done fore figure 6-5A. A comparison of the force curves for the PVECs grown on the different surfaces is plotted in figure 6-5C. The changes in

PAGE 154

136 the cell cytoskeleton are quite apparent due to substrate properties. PVECs on PS similar elastic modulus throughout the cell regardless of the whether the cell nucleus or periphery were examined. In theory, this could be explained as a substrate phenomenon, where we are actually measuring the elastic modulus of the PS, which is masking variations in the cell elastic modulus. While this makes sense initially, comparison to the PVECs on PDMSe_FN shows that this is not the case. The PDMSe_FN has a low elastic modulus, about 10 times higher than that of a cell, yet the force curves have a wide range in elastic modulus depending on the region of the cell that is examined. The force curve with the highest slope is from the region above the cell nucleus while the force curve with the lowest slope is from the cell periphery. This difference demonstrates the difference in the stiffness, and hence the density of actin cytoskeleton, with location on the cell. Looking back at the PVECs on PS shows that in this case the stiffness and density of the cytoskeleton minimally changes across the cell independent of subcellular organelles such as the nucleus.

PAGE 155

137 Figure 6-5. Comparison of the mechanical properties of PVECs grown on PS or FN coated PDMSe. A) A topographic image of PVECs grown on fibronectin coated PDMSe for 4 days. (B) A topographic image of PVECs grown on PS for 4 days. (C) Plot of the force curves taken from the regions indicated by the stars in parts (A) and (B). A detailed analysis of ECs grown on both PS and PDMSe_FN supports the conclusion that substrate influences the cell mechanical properties. Figure 6-6, examines ECs grown on PS for 4 days and compares the mechanical properties of 7 different regions. Regions 1, 2 and 3 are cell nuclei, regions 4, 5 and 6 are the cell bodies and region 7 is the cell periphery. The force volume image (figure 6-6B) shows the nuclei are bright and thus have the lowest stiffness, increasing for the cell bodies and then the cell periphery. This same trend is seen for the force curves plotted in figure 6-6C. (where an example force curve from each region is shown). As described for figure 6-4, the

PAGE 156

138 historesis is greatest for the cell nuclei and decreases moving towards the cell periphery. The average elastic modulus of each region is presented in table 6-2. As seen previously, the cell nuclei (regions 1, 2 and 3) have the lowest elastic modulus with no significant difference between them. The elastic modulus of the cell bodies (regions 4, 5 and 6) are higher, but not significantly different than the cell nuclei. The cell periphery in region 7 has the highest elastic modulus at 25.2 3.3 kPa. This is not significantly different than the elastic modulus for the cell bodies, however it is significantly different than the elastic modulus for the cell nuclei (P<0.05). These results demonstrate that the mechanical properties of the ECs on PS vary by approximately 60% in the range of 7 to 25 kPa. Table 6-2. Elastic modulus for the different regions identified in figure 6-6. Region Elastic Modulus [kPa] 1 7.4 1.1 2 7.4 2.0 3 6.2 0.8 4 13.1 2.2 5 12.5 1.6 6 20.1 8.0 7 25.2 3.3

PAGE 157

139 1 2 3 4 5 6 7 Force Curve Variation for Multiple Cellls-350-300-250-200-150-100-500132114096078051003120214001598179619942192239025882786298431823380Separation [nm]Deflection [nm ] region-01 region-02 region-03 region-04 region-05 region-06 region-07 C A B Figure 6-6. An analysis of the topography and mechanical properties of ECs grown on PS for 4 days. (A) A topographic image of the ECs on PS, the highlighted regions were analyzed for the average elastic modulus. (B) A force volume image indicating the relative stiffness of the topographic image (bright yellow = soft, dark brown = stiff). (C) A plot of representative force curves from each region identified in part (A) to demonstrate the change in mechanical properties with region.

PAGE 158

140 ECs on PDMSe_FN showed a variation in mechanical properties with location on the cell that was wider than that seen for ECs on PS. As noted in figure 6-5, the ECs became stiffer as the cell was probed from the nucleus out to the periphery. In figure 6-7, ECs were examined more closely to determine the magnitude and variation in the mechanical properties for 5 different regions. As highlighted in figure 6-7A, regions 1 and 3 are cell nuclei, regions 2 and 5 are the cell bodies and region 4 is the cell periphery. The force curves from these different areas varied in terms of slope and historesis. An example force curve from each region is plotted in figure 6-7B. The average elastic modulus for each region is listed in table 6-3. The nuclei from regions 1 and 3 had the same low elastic modulus and viscoelastic response seen for EC nuclei on PS. There was no statistical difference between regions 1 and 3 demonstrating the uniformity in stiffness of the cell nuclei. Regions 2 and 5 were both from the cell body, however the force curve looked quite different. Region 2 was from an area consisting of stress fibers running down the length of the cell from the nucleus in region 1. The force curves between regions 1 and 2 look the same, and even though the elastic modulus of region 2 is slightly higher than region, it is significantly different (P<0.05) verifying the higher compliance of the nuclei compared to the rest of the cell body. Looking at region 5, the elastic modulus is ~6 times larger than region 2, however the large standard deviations results in no significant difference between the regions. Region 4 representing the cell periphery (based on its low height) shows no statistical difference with region 5 though it is significantly different than region 4 (P<0.05).

PAGE 159

141 A 5 4 2 1 3 Force Curves of ECs on PDMSe_FN-595-495-395-295-195-9551013526038450963475888310081132125713821506163117561880200521302254237925042628Separation [nm]Deflection [nm] region-01 region-02 region-03 region-04 . region-05 B Figure 6-7. An analysis of the topography and mechanical properties of ECs grown on PDMSe_FN for 4 days. (A) A topographic image of the ECs on PDMSe_FN, the highlighted regions were analyzed for the average elastic modulus. (B) A plot of representative force curves from each region identified in part (A) to demonstrate the change in mechanical properties with region.

PAGE 160

142 Table 6-3. Elastic modulus for the different regions identified in figure 6-7. Region Elastic Modulus [kPa] 1 6.3 0.7 2 10.1 1.7 3 7.2 0.7 4 56.7 15.8 5 66.1 25.8 Looking at the EC response to PS and PDMSe_FN shows some interesting properties. Even though the PS is glassy at room temperature and has an elastic modulus of ~3 GPa, there is little evidence of this contribution to cell elastic modulus as measured. In contrast, the ECs on PDMSe_FN have a larger elastic modulus at the periphery of the cell even though the underlying plasma treated PDMSe has an elastic modulus of ~1 MPa. Since the ECs on PS did not transfer the underlying substrate modulus to the force curve measurement, it is unlikely that the PDMSe modulus contributed to the measured force curves either. Making this assumption, it appears that the PDMSe_FN surface causes the ECs to form a denser cytoskeleton at the periphery than ECs on PS. Examining the fluorescent actin staining in figure 6-2 parts B and F, ECs on both surfaces stain strongly for actin at the periphery. Considering that the ECs on PS are more stable and close to confluence, it is unlikely they are growing or crawling at a significant rate. In contrast, the ECs on the PDMSe_FN are actively looking for a better place to adhere as evident by the long cytoplasmic extensions. This would indicate a higher rate of actin based motility at the cell periphery corresponding to a denser actin network and elastic modulus. Conclusions In his chapter the AFM is used to demonstrate differences in EC properties as a function of the cell culture substrate. Morphology changes between PVECs on PS or

PAGE 161

143 PDMSe_FN seen with fluorescent microscopy were confirmed by AFM. Furthermore, the topographic images revealed the location and organization of larger actin filaments and stress fibers. Force volume scans of the PVECs demonstrated that the elastic modulus of the cells is a function of the substrate. Some surfaces such as PS cause a well developed cytoskeleton to form with mechanical properties are nearly uniform across the cell surface. In contrast, cells on PDMSe_FN showed a wide variation of mechanical properties with location changing from low at the cell periphery to high at the cell nucleus. From this study, the AFM is verified as a useful tool to analyze the properties of live cells and can be used to understand differences in cells that can not be measured or quantified with optical techniques such as fluorescent microscopy.

PAGE 162

CHAPTER 7 SIMULTANEOUS QUANTIFICATION OF RECEPTOR-LIGAND LOCATION, BINDING FORCE AND INTERACTION DISTANCE ON LIVE CELLS Introduction Advances in tissue engineering and medical device coatings are dependent on controlling the interactions between biological systems and synthetic materials. Many experimental techniques allow analysis of cell and material properties, but very few can quantify both the physical and chemical attributes of living systems. Optical and electron microscopy can resolve fine details of morphology and structure, but the staining and fixation procedures often kill or alter cells. In addition, these techniques require immunogold or immunoflourescent stains to measure the short range receptor-ligand binding and long-range non-specific forces that dictate cell adhesion, chemotaxis and other processes. While techniques such as the surface forces apparatus (SFA) can measure specific and non-specific forces, the experimental setup does not allow measurements on living cells or provide morphological data [96]. Similarly, adaptations of the optical trap method have been used to measure weak biological forces, but can not accurately provide force-distance profiles or cell structure information [97, 98]. Micropipette aspiration methods, also referred to as the biomembrane force probe, can accurately measure receptor-ligand binding forces, but once again can not determine morphology [99]. Recently, there has been a rapid increase in the application of AFM to the study of biological systems because of the ability to operate in a liquid environment with 144

PAGE 163

145 simulated physiologic fluids and has been used to investigate the topographical and mechanical properties of living cells [63, 100, 101]. Chemical functionalization of the tip and surface has allowed the binding between molecules to be examined for a variety of receptor-ligand pairs. Examples from the literature provide the basic methodology for interpreting and analyzing the adhesion events seen in the retrace portion of the force curve and has been demonstrated for DNA strands, avidin-biotin, antigen-antibody and PSGL-1 to P-Selectin interactions [48, 49, 52, 53, 55, 57, 58, 60]. Within the last few years studies have emerged that measure the binding force between ligand coated tips and cultured cells. Lehenkari et al found a 32 2 pN adhesion force between integrin 5 1 molecules in osteoblasts and a GRDGDSP peptide coated tip, but the procedure requires fixing the cells in glutaraldehyde to resolve the interaction [102]. In a subsequent study, Hyonchol et al probed integrins in living 3T3 fibroblasts using a fibronectin coated microsphere tip but measured the interaction in terms of work of adhesion rather than single molecule binding force [103]. Currently, there has been no reported experiment that combines these techniques both to quantify and map receptor-ligand binding on living cells. Presented here is an evaluation of the biointerface that simultaneously examines the topographic location, specific binding force and interaction distance of receptors expressed on the free membrane surface of adhered cells (Figure 7-1A). In addition, the experimental setup is designed to approximate the actual interaction geometry by simulating the physiological process of leukocyte rolling on the endothelium in the vascular system (Figure 7-1B). The binding between P-selectin and sialyl Lewis X (sLe X ) was identified as a model system because the interaction has been well characterized by AFM and other

PAGE 164

146 techniques [60, 104]. The recruitment of white blood cells begins when leukocytes leave the blood flow by loosely adhering and rolling on the endothelial cells through carbohydrate mediated receptor-ligand interactions between selectins on the surface of the endothelial cells and complimentary ligands expressed by the leukocytes. Expressed by activated platelets and endothelial cells, P-selectin is the longest with a molecular weight of 140 kg/mol and extending nearly 40 nm from the membrane surface. The primary ligand for P-selectin on leukocytes is P-selectin glycoprotein ligand-1 (PSGL-1) and the functional component that elicits specific binding to P-selectin is a tetrasaccharide known as sLe X . Rodgers et al demonstrated that sLe X can support rolling on P-selectin without requiring the PSGL-1 ligand [105]. In these experiments, a sLe X functionalized AFM microsphere tip brought in contact with an endothelial cell during a force curve measurement is used to simulate a rolling leukocyte using a technique similar to that developed by Fritz et al (Figure 7-1B) [60]. By using the AFM microsphere tip, the similar interaction geometry and cell-tip separation velocity of the P-selectin-sLe X response will approximate that seen in vivo. Using the AFM to study the topography and mechanical properties of living cells has been demonstrated for endothelial, epithelial and other cell types [63, 65, 100, 106]. These studies used standard AFM cantilevers in contact and force curve modes and examined both living and fixed cells. Surface structures such as the cell membrane and cytoskeleton were resolved, with results similar to the contact mode images of the PVEC topography (Fig. 2). These images verified the need for a low tip/sample contact pressure and established a defined period of cell viability for subsequent force volume imaging. Single molecule AFM studies have examined both receptor-ligand binding and protein

PAGE 165

147 unfolding and were adapted to probe living cells in these experiments. Willemsen et al were able to map the location of specific antibody-antigen interactions, but not quantify the binding force [107]. Benoit et al examined cell-cell interactions by attaching a cell to the AFM cantilever to probe the surface of another cell [57]. This application of single-molecule force spectroscopy proved successful, but the variable surface chemistry of a live cell made detection of a single receptor-ligand pair difficult and the large cell-cell contact area prevented mapping of the interaction on the cell surface. The work presented in this chapter is an extension of the AFM technique I developed for probing receptors on live cells in my master’s thesis [1]. The expansion of this work is based on modifications of the analysis procedure used to interpret the data and no new experiments were actually performed. Images that have been adapted from the master’s thesis are referenced accordingly. The materials and methods are essentially the same since the experiments were not repeated. The presented analysis includes consideration of pulling rate kinetics and a more in depth modeling of the molecular interactions.

PAGE 166

148 Figure 7-1. Illustrations of the interaction between the sLe X coated AFM tip and the PVEC surface. (A) A 3-dimensional schematic illustrating the approach of a chemically functionalized AFM tip towards a monolayer of living endothelial cells (PVECs). The cell image is a topographic image of living endothelial cells imaged in Hank’s buffered salt solution at 22C using a silicon nitride contact mode tip rendered in 3D. (B) (top) As the leukocyte flows through the blood stream it rolls on the endothelial cells at the site of inflammation through receptor-ligand interactions. This rolling phenomenon can be simulated in vitro (bottom) using a biochemical functionalized AFM probe to measure the specific interaction force and distance (adapted from [1]). Materials and Methods Cantilever Preparation Topographic images were performed using standard, V-shaped, contact mode silicon nitride cantilevers with an integrated pyramidal tip (Veeco Metrology Group,

PAGE 167

149 Santa Barbara, CA). The cantilever with a reported 0.06 N/m spring constant was selected and cleaned with ethanol and UV exposure prior to imaging. Force volume images were performed using a chemically modified microsphere attached to the AFM tip with the ligand sialyl Lewis X (sLe X ) as the functional moiety. Specifically, tipless, V-shaped, silicon nitride cantilevers with a 0.06 N/m spring constant and with an aminopropyl triethoxysilane (APTES) coated silica 5.9 m diameter microsphere already mounted to the end (Novascan Inc., Ames, IA) were biochemically functionalized. A custom built, silicone rubber reaction vessel was used to modify the AFM microsphere tip by treating with 8% aqueous glutaraldehyde overnight at 37C and then washing 3 times with Hanks balanced salt solution (HBSS). SLe X linked to bovine serum albumin (Oxford Glycosciences, Abingdon Oxon, UK) was then reacted to the microsphere by placing a 2-3L droplet directly on the microsphere and incubating at 37C for 2 hours. The AFM probe was washed 3 times with HBSS and transferred in a hydrated state to the AFM. Once functionalized, AFM probes were used within 24 hours and stored in HBSS at ambient conditions when not mounted on the AFM scanner. Functionality of the sLe X ligand on the AFM tip was verified using FITC-tagged antibody fluorescent staining on identically treated samples. The spring constant of the AFM tip was calibrated using a reference cantilever (Park Scientific) following published methods [108]. As controls, AFM probes functionalized with APTES or APTES + glutaraldehyde were also prepared and used to image the Petri dishes with and without a layer of endothelial cells. This was done to determine if the specific binding was due solely to the receptor-ligand interaction.

PAGE 168

150 Cell Culture and Instrument Setup Primary porcine vascular endothelial cells (PVECs) harvested from fresh pulmonary artery were obtained from the research group of Dr. Edward Block (VAMC hospital, Gainesville, FL) at passage 2 to 6 using previously published methods [109]. Cells were maintained in RPMI 1640 culture medium (Life Technologies, Carlsbad, CA) containing antibiotics (100 U/ml of penicillin, 100 g/ml of streptomycin, 20 g/ml of gentamicin, and 2 g/ml of Fungizone) and supplemented with 10% fetal bovine serum. The cells were seeded onto sterile 35 mm diameter polystyrene culture grade Petri dishes (Corning Inc., Corning, NY) and grown to confluence in an incubator at 37C and 5% CO 2 for 3 to 5 days. Upon removal from the incubator, the PVECs were rinsed 3 times with HBSS to remove physisorbed extracellular matrix and serum proteins. The PVECs were then immediately transferred to the AFM for analysis in a hydrated state. The PVECs and AFM tip were mounted on the AFM chuck and scanner respectively, and then the volume of HBSS covering the PVECs was reduced to ~2.5 mL to minimize potential damage to the scanner piezoelectrics. The AFM probe was then lowered into the HBSS to a distance 1 mm above the PVEC monolayer and held for 1.5 to 2 hours to allow the cantilever to thermally equilibrate and stabilize. PVECs were then imaged at ambient conditions of ~22C in force volume mode using a Dimension 3100 AFM with Nanoscope IIIa controller (Veeco Metrology Group, Santa Barbara, CA). The force curve scan rate was set to 2 Hz to minimize hydrodynamic drag on the cantilever and maximize imaging speed. Scan size was set to 50 m, which in conjunction with the 32x32 force curve array and large tip sample contact area, resulted in a near continuous surface profile. Force volume scans required approximately 2 hours to complete with 1

PAGE 169

151 to 2 areas examined per Petri dish within the 6 hour cell viability period. Each functionalized tip was used to scan up to 4 PVEC areas, with loss of specific interactions and suspected tip fouling occurring beyond this point. Standard contact mode topographic images of PVECs were obtained using the same protocol outlined above, except once the cantilever was allowed to equilibrate the AFM was operated in contact mode rather than force volume mode. The deflection setpoint was set approximately equal to the non-contact tip deflection minimizing tip/sample interaction forces. Scan rate was adjusted in conjunction with scan size to optimize scan quality. The same PVEC sample was never used for both contact mode and force volume imaging because the scan time and cantilever equilibration time for both modalities would not fit within the 6 hour cell viability period. Once the AFM scans were completed, the data was analyzed offline using a number of software packages. Data Analysis The force curves were analyzed using single-molecule force spectroscopy techniques developed for receptor-ligand binding and macromolecule unfolding [55, 57, 58, 110-112]. Each individual force curve of the force volume array was analyzed offline to determine the peak height and distance of any specific adhesion events that corresponded to receptor-ligand binding. The curves were normalized in the force axis by setting the maximum deflection point at zero force. For each force curve, the interaction forces were recorded as the difference between the non-contact force at full retraction and the adhesion peak forces. The interaction distance was defined as the difference between subsequent adhesion events within each force curve. Interaction force and interaction distance were sorted using histograms with 20 pN and 5 nm sized bins respectively. The force histogram upper limit was set to 3.5 nN due to inconsistent data

PAGE 170

152 above this value. A 6 th order binomial filter was used to smooth the interaction force histogram [113]. To determine the periodicity in the binomial filtered histogram that would resolve the single interaction force, an autocorrelation function was applied using methods established by Florin et al [55]. For the interaction distance histogram, the upper limit was set to 150 nm, approximately three times the estimated length of the stretched selectin-sLe X -BSA-APTES chemistry. The interaction distance was calculated as the mean of the interaction distance histogram. Contact mode images were performed using a silicon nitride tip with an integral pyramidal tip with a 40 nm radius of curvature. The force volume measurements were performed with a microsphere tip with a 2.45 m radius of curvature. The resulting difference in contact area between the two tips meant a higher contact force was used with the microsphere tip while still keeping the applied contact stress below the critical value needed to perturb submembrane structures. Mapping the location of specific interactions on the PVEC surface was performed by reconstructing the topography from the force volume image. The surface was defined as the point of maximum tip deflection in each force curve. The resulting topographic image was then correlated with the force curves that demonstrated specific interactions in the retrace portion of the force curve. For visualization, the specific interactions were manually marked with neon green spots using photo editing software. Rendering top-down and 3-dimensional images of the mapped PVEC surface allowed the X, Y and Z coordinates of the P-selectin on the cell surface to be calculated.

PAGE 171

153 Results and Discussion Cell Response to AFM Imaging It was important that the cell and its environment were minimally perturbed during the scanning process to ensure the measurements were relevant to in vitro and in vivo cell physiology. If the AFM tip punctured the cells, applied excessive mechanical stress or ripped off the membrane on retraction, severe cell trauma or lysis may have occurred. To verify cell viability, PVECs were scanned in standard contact mode using a 0.06 N/m cantilever in physiologic buffer for up to 8 hours. By adjusting the contact force between the tip and cell, it was possible to image the delicate morphology of the cell membrane or the deeper features of the cytoskeleton (Figure 7-2A). Imaging of submembrane structures such as stress fibers required ~5 nN of contact force resulting in loss of cell structure after repeated scanning. A lower contact force of 0 to 0.5 nN allowed the membrane to be imaged for longer periods with cell structure well maintained for up to 4 hours of continuous scanning (Figure 7-2B). Quantifying Interaction Force To map and quantify the receptor-ligand interactions, an array of force curves was performed across a 50 m by 50 m square area. This 2-dimensional array of force curves is referred to as a force volume dataset and combines force and topographical data in one experimental procedure. Based on the instrument setup (see Materials and Methods), the area was divided into a 32x32 array resulting in a force curve measurement every 1.56 m. Each individual force curve of the force volume array was analyzed offline to determine the peak force and distance of any specific adhesion events in the retrace that correspond to receptor-ligand binding.

PAGE 172

154 Figure 7-2. Topographic AFM contact mode images of PVECs in HBSS revealing the effects of contact force and imaging time on the cell membrane and cytoskeleton. (A) High contact force allows cytoskeleton to be imaged and a Low contact force allows membrane to be imaged. (B) Sequential AFM deflection images of living PVECs imaged in HBSS showing (chronological clockwise from upper left) the appearance of a sub-membrane structure appearing at a time between image 1 and 2 and tracking its movement through time with images 3 and 4. The structure is thought to be an organelle, possibly with associated membrane specific proteins. Image 1 was taken after the cantilever was allowed to thermally equilibrate for 2 hours to minimize laser drift and image 4 was completed 2.5 hours later (adapted from [1]).

PAGE 173

155 Appropriate controls were maintained ensuring the measured specific interactions arose from receptor-ligand binding. As described, the functionalization of the AFM tip was verified using immunoflourescent staining. This technique showed that the sLe X bound to the AFM tip was in an accessible and bioactive state. By having the sLe X coupled to BSA, other specific interactions from the underlying chemistry were minimized because BSA demonstrates only non-specific interaction with other proteins. The control experiments were performed using a microsphere AFM tip with an APTES or ATPES + glutaraldehyde coating to probe a sterile Petri dish or a cultured monolayer of PVECs. The force curves produced with the control tips showed only non-specific interactions between the AFM tip and the Petri dish or PVECs. Furthermore, the interaction between the control tips and the endothelial cells were typical of the force curve profile for an AFM tip compressing a soft substrate (indistinguishable from Figure 7-3A). The sLe X functionalized AFM tips showed both specific and non-specific interactions between the tip and PVECs depending on location. The majority of the force curves on the PVEC surface showed non-specific interactions (Figure 7-3A) and were indistinguishable from the force curves of the control experiments. Force curves with specific interactions in the retrace (retraction) force curve (Figure 7-3D) were analyzed to determine the interaction force and distance of the discrete adhesion peaks, which represented receptor-ligand bond events. Each adhesion peak in the force curve exhibited a constant slope that corresponded to stretching of the polymer backbone, followed by a sudden drop to a lower force value as the receptor-ligand complex ruptured. AFM studies of receptor-ligand binding have found bond rupture forces ranging from 30 to 250 pN for single molecules. The force curves in these

PAGE 174

156 experiments showed specific interactions with forces ranging from 150 pN to almost 5 nN (Figure 7-3B). These large forces are the results of multiple, single receptor-ligand bonds being stressed at one time. This is a typical occurrence when a relatively large chemically functionalized microsphere is used as the AFM tip. This multiplexing of single receptor-ligand interactions requires further analysis to elucidate the single receptor-ligand interaction (see Materials and Methods). The sLe X AFM probe was used to image 5 areas from 3 distinct PVEC cultures resulting in over 5000 force curves. Of those force curves, 125 demonstrated specific interactions with more than 1000 individual force measurements obtained for the selectin-sLe X interaction. As a control, 125 force curves that showed only non-specific interactions were also analyzed using the same process. The force values for the specific and non-specific interactions were organized into histograms quantized in 20 pN bins (Figure 7-3 B and E respectively). The noise in the histogram was smoothed with a binomial filter and the periodicity was found using an autocorrelation function. Applying the autocorrelation function to the non-specific interactions produced a curve with no regular periodicity indicating that the adhesion was not from the selectin-sLe X binding (Figure 7-3C). Similar application of the autocorrelation function to the specific interactions revealed a periodic interaction force of 183 40 pN (Figure 7-3F).

PAGE 175

157 157

PAGE 176

158 Figure 7-3. The force curves were analyzed for adhesion peak forces, the values were sorted into histograms and then an autocorrelation function was applied. Representative examples of AFM force curves of (A) non-specific and (B) specific interactions. The adhesion peaks from the force curves were analyzed to produce histograms of (C) non-specific interactions (n = 125) and (D) specific interactions (n = 1000). An autocorrelation function was used to analyze the data for (E) non-specific and (F) specific interactions (part (D) adapted from [1]).

PAGE 177

159 This value is in close agreement for the bond rupture force of P-selectin and PSGL-1 reported in the literature. Fritz et al., demonstrated that single interactions between P-selectin grafted to an AFM tip and PSGL-1 grafted to a glass substrate show a bond rupture force that has a logarithmic dependence on pulling velocities from 0.2 to 4 m/s [60]. At a pulling rate of 2.8 m/s the bond rupture force was 159 30 pN and extrapolating the logarithmic dependence to the experimental pulling rate of 12 m/s predicts a bond rupture force between 170 and 200 pN [60]. A similar value was obtained by Smith et al that demonstrated an estimated maximum binding force of ~200 pN for the interaction between living leukocytes and endothelial cells using high-speed video microscopy [114]. When using single-molecule force spectroscopy, proper interpretation of the measured rupture forces is necessary to determine the correlation to molecular structure. A number of events may lead to an adhesion peak similar to those in the force curves with specific interactions (Fig. 3 D). Single molecule AFM studies of titin have shown similar 150 to 300 pN rupture forces due to unfolding of the immunoglobulin like domains in the protein backbone, so it is possible a similar effect was occurring in the 9 consensus repeat units of the P-selectin backbone [115]. However, this is unlikely because P-selectin appears to have a linear, rod-like backbone, as imaged by TEM, and was shown to have only one rupture force event in single receptor-ligand pair AFM studies [60, 116]. SFA experiments of other receptor-ligand pairs have shown that rupture forces can also occur due to pullout of the ligand or receptor from the respective lipid membranes [111]. The sLe X ligand was covalently linked to the AFM probe and could not pullout at the low forces measured in these experiments. Similarly, the

PAGE 178

160 cytoplasmic tail of L-selectin is attached to the actin cytoskeleton of leukocytes. A nearly identical structure of the membrane and cytoplasmic regions of P-selectin suggest a similar anchoring system that prevents membrane pullout [117, 118]. Recent studies have demonstrated that the bond rupture forces vary for AFM and BFP techniques depending on the bond dissociation constant and loading rate (linearly equivalent to pulling velocity) [119]. While this dependence is acknowledged, these experiments are designed to simulate leukocyte rolling on endothelial cells and thus are performed at a single pulling velocity similar to that experienced by selectin-sLe X interactions in vivo. Ley and Gaehtgens showed that leukocyte velocity was 20 to 40 m/s at physiologic shear rates in rat mesenteric venules, leading to an experimental pulling rate of 12 m/s to maximize speed and minimize hydrodynamic drag on the AFM cantilever [120]. The relatively large nanoNewton adhesion peaks seen in the force curve is the result of multiplexing of the single receptor-ligand bonds. This is similar to that found by researchers investigating avidin-biotin and other receptor-ligand bonds with AFM [55, 110, 111]. Since a relatively large microsphere AFM tip is used, multiple selectin-sLe X bonds frequently form and then rupture during retraction of the AFM tip from the cell membrane (Figure 7-3A). When the functionalized microsphere extends to the cell surface the sLe X ligand couples to selectins within the contact area. On retraction, a critical stress is reached when the first receptor ligand bond breaks after which the remaining bonds dissociate in rapid succession. The AFM tip separates from the cell surface with the dissociation of the last receptor-ligand bond.

PAGE 179

161 Quantifying Interaction Distance The length of the stretched receptor-ligand complex for these experiments was calculated differently than that typically found by AFM. Single molecule receptor-ligand studies, such as for PSGL-1 to P-selectin, measure the distance of the stressed bond and compare that to the contour length of the molecular complex [60]. For most single molecule studies with non-living systems, the molecules are grafted to rigid materials such as glass or silicon nitride that provide a well identified contact point between the sharp AFM tip and planar substrate [48, 52, 57, 58]. Therefore, the length of the stressed molecules is simply the distance between the AFM tip and substrate upon bond rupture. In contrast, the system used here employs a large microsphere indenter on a soft substrate. This interaction results in no readily identifiable contact point between the AFM tip and substrate characterized by a gradual transition from the noncontact to contact portions of the force curve (Figure 7-3 A and D). Researchers that have studied the binding between a living cell adhered to the AFM tip and a living cell substrate have reported similar force curve behavior [57]. For the selectin-sLe X binding in this study, the characteristic interaction distance was defined as the distance between consecutive adhesion peaks in each force curve. Calculating the length in this manner assumes that when a group of receptor-ligand bonds rupture, another group immediately begins to be stressed. By analyzing a large number of adhesion events, variation in receptor density on the cell surface is averaged out. To prevent interactions from widely spaced distances being included in the analysis, an upper limit on the histogram was set at three times the calculated maximum length of the stretched receptor-ligand molecule.

PAGE 180

162 The experimental interaction distance can be compared to a theoretical estimate of the length of the bound receptor-ligand molecules. The bound receptor-ligand complex is composed of P-selectin on the cell and sLe X , BSA, glutaraldehyde and APTES on the microsphere. SEM data indicates that the extracellular component of P-selectin will unravel into a linear structure with the amino acid sequence giving a length of 25 nm [116]. The combined length of the smaller glutaraldehyde, APTES and sLe X molecules is 3.2 nm. From crystallographic data, BSA is a single polypeptide chain of 583 amino acid residues composed of 67% -helical structure and 17 disulfide bonds with a 14 nm maximum length [121]. In physiologic solution, the length of BSA can be estimated from the radius of gyration (r g ) at 2.7 nm and a corresponding end-to-end distance of 6.5 nm [122]. However, BSA is known to have a number of conformational states induced by pH changes such as the series of balls and strings of the F-E transitions with a 25 nm length [122]. Using the normal and F-E transition conformations of BSA as limits, the length of the fully stressed receptor-ligand complex can be estimated to range from 34.7 to 53.2 nm. Using a histogram (Figure 7-4) to analyze the experimental data reveals an average characteristic interaction distance of 64 31 nm. The discrepancy between the measured and predicted interaction distance is thought to be due to a combination of effects. Some deformation of the cytoskeleton and membrane may occur on AFM retraction as a result of elastic recovery from the approach portion of the force curve measurement. Also on retraction, the AFM microsphere tip may pull the cytoskeleton and cell membrane with it due to non-specific Van der Waals and multiple receptor-ligand adhesive interactions. For BSA, elongation beyond the F-E transition to ~100 nm is possible by retention of the

PAGE 181

163 disulfide bonds but loss of the -helical structure. Therefore, the increased length of the experimental interaction distance is thought to be due to a combination of cell deformation and elongation of BSA beyond the F-E transition conformation. To improve the agreement between the experimental and theoretical interaction distances, future experiments will look to replace BSA with a short protein or polymer linker with known unfolding characteristics. Figure 7-4. A histogram of the specific interaction distance from 0 to 150 nm split into 12 nm bins. Mapping Receptor Location on the Cell Surface Each force curve in the 32x32 force volume array was analyzed for specific interactions and its location recorded. The topographical image was formed by taking a constant force slice through the force volume set at the maximum tip deflection (maximum force). The force volume topographic images (Figure 7-5 B and D) were similar to that produced in contact mode. The pixels in the force volume topographic images that corresponded to force curves containing specific interactions were marked

PAGE 182

164 with neon green dots (Figure 7-5 A and C). These neon green spots indicate where the P-selectins were located on the cell surface, thus providing a map of the receptors. The specific interactions occurred near the periphery of the cell away from the cell nucleus. In fact, only 1 out of 125 force curves showed specific interaction when above a cell nucleus. The selectins also appeared to cluster in groups on the surface with interactions occurring in similar areas on subsequent scan lines. The cluster size varied from 1.5 to 4.5 m in diameter with the resolution limited to 1.56 m by the spacing between scan lines. The observed selectin clustering was not attributed to lateral diffusion of the same selectins in the membrane because clusters were present in both fast and slow scan directions, which differ temporally by three orders of magnitude. L-selectin mobility, as we believe is true of P-selectins, is limited by anchoring to the cytoskeleton [117]. Both L-selectins and P-selectins organize into clusters that have been visualized with SEM and confocal microscopy. L-selectin clusters form on microvilli of leukocytes, whereas P-selectins clusters are associated with the clathrin-coated pits of endothelial cells [114, 118].

PAGE 183

165 Figure 7-5. Force volume images of PVECs grown on polystyrene imaged with the sLe X modified AFM probe. Images are 32x32 arrays of force curves with a 500 nm deflection setpoint. Specific interactions as identified by adhesion peaks in the retrace are shown by neon green dots. The selectins appear to form clusters located at the periphery of the cell. The images are 50x50 m scan areas and 6 m z axis scaling (adapted from [1]).

PAGE 184

CHAPTER 8 CONCLUSIONS AND FUTURE RESEARCH Conclusions A number of important conclusions may be drawn from the research presented in this dissertation. For tissue engineering EC monolayers, it can be determined that at least 1 m or greater topographic steps will totally disrupt EC adhesion and spreading. Previous studies demonstrated that 5 m high, 5 m wide ridges can be used to direct EC growth on PDMSe. However, this contact guidance is quantified in terms of nuclear alignment to the ridge pattern and not cell spreading. The critical point is that a surface that works well in directing ECs to elongate is not necessarily a surface that causes ECs to function properly. The differences in the metabolic state of the ECs is obvious from the morphology alone, as highly spindle-like, elongated cells have a different need for cytoskeleton and focal contact adhesion proteins than normal, rounded cells. This difference between contact guidance and directed function is important because it necessitates consideration of the operative outcome when designing tissue engineering scaffolds. For the case of ECs, certain design criteria can be determined from examination of the ECs in fresh artery. The combined AFM, SEM and fluorescent microscopy analysis shows that the ECs need to form a smooth and continuous monolayer. Furthermore, the actin cytoskeleton needs to be robust with no decrease in actin density between adjacent cells. Cell densities must also be high with ~3,100 cells/mm 2 and a corresponding cell area of xxx m 2 /cell for the section of porcine 166

PAGE 185

167 pulmonary artery examined. Any surface with micron scale features will cause ECs to elongate and decrease the packing density well below those found in vivo. AFM reveals that the smooth transition between cells is marked by ‘hills’ corresponding to the raised cell nuclei. The heights of these nuclei are ~2 m as measured from the tops of the nuclei to the lowest region between two adjacent nuclei. If the diameter of the EC nucleus is assumed to be relatively constant, then AFM topographic scans of ECs on PS shows that the nuclei are ~3.5 m in diameter. Combining the AFM data from ECs on PS and in fresh artery suggests that the ECs in the fresh artery are ~5.5 m diameter at the nucleus and 3.5 m at the periphery. From this data, it is apparent that any engineered surface with topographic features taller than ~5.5 m will extend above the EC monolayer and into the blood flow. This would be a problem because the ECs could not climb over it and it would likely serve as a point of clot formation and emboli in vivo. The topographies that were examined did not result in good EC monolayer growth leading to the analysis of flat surfaces with chemical micropatterns. Results from EC growth on FN micropatterns show that the size of the individual patterns is critical in how the cells develop. On PDMSe, FN micropatterns approximately the same size as the cells results in ECs getting stuck to the pattern. These patterns support initial proliferation, but all the cells die by 14 days in culture. More relevant to coatings that could be applied to a small diameter vascular graft are the 3 m diameter circles on PDMSe that can be used to direct formation of focal contact adhesions and grow an EC monolayer on PDMSe, a novel result.

PAGE 186

168 This result, that 3 m diameter FN micropatterns will support EC monolayer growth on PDMSe, is important in a variety of areas. First, it demonstrates proper EC growth on an elastomer surface in vitro. As such, an FDA approved PDMSe could be fabricated using this surface treatment and integrated into an “off the shelf” small diameter vascular graft for acute application. This type of graft would be valuable in emergency situations where autologous saphenous vein or mammary artery is not available due to time constraints or patient vascular pathology. The 3 m FN circle design is also interesting in the way it organizes the formation of focal contact adhesions to the surface and alignment of the actin stress fibers. This design can be improved by controlling the density of these FN circles to induce an as yet undetermined optimal number of focal contacts to the surface. In addition, a gradient design may be incorporated that changes the density of the circles as a functional of the radial distance from the center of the cell. This level of control should allow surfaces with micropatterned FN to be specifically tailored to different cell types. As a surface technique, this is most beneficial to cells that form monolayers such as corneal and intestinal epithelium. The mechanical properties of ECs on engineered surfaces are a valuable indicator of cell function. The measurements made on live EC in the fresh artery showed that the elastic modulus and mechanical properties were relatively homogeneous within the area of single AFM scans. The total variation in elastic modulus across the arterial surface was 10 to 20 kPa. For the ECs on synthetic surfaces, the variation in elastic modulus with position on the cell revealed that the cell periphery was stiffer than the nucleus. Quantitatively, the elastic modulus of ECs on PS ranged from approximately 6 to 25 kPa

PAGE 187

169 and on PDMSe_FN from 6 to 60 kPa moving from the nucleus out to the periphery. What this demonstrates is a distinct difference in the organization and stiffness in the cytoskeleton as a function of position on the cell and culture substrate. Future Work It is possible that designing a synthetic surface to support normal ECs maybe enough to form a long-term blood contacting surface. An example of how a microengineered surface can be incorporated into a complete small diameter vascular graft is presented in figure 6-1. The goal for this device would be an off-the-shelf product for bypass surgery, just like the Dacron and Teflon grafts used for abdominal aortic aneurysm and other large diameter vessels. As such, the graft is designed for in vivo tissue regeneration so that it may be used in acute situations rather than requiring weeks of in vitro cell expansion in a bioreactor. In this example, the macroscopic mechanical properties of the blood vessel are designed into a polymer composite consisting of an elastomer with reinforcing fibers mimicking the way elastin, collagen and smooth muscle cells of the tunic media maintain physiologic compliance. This will minimize the stress concentrations at the anastomoses that leads to smooth muscle cell proliferation and intimal hyperplasia. The lumen of the artificial graft will be modified to control endothelial cell adhesion and growth to create a cell based blood contacting surface. Circulating endothelial cells will be removed from circulation using a specific antibodies bound to the graft surface. A combination of micron scale topography and patterned fibronectin will direct integrin receptors and formation of focal contact adhesions. Vascular endothelial growth factor in low levels will help to stimulate proper endothelial cell function and minimize SMC growth. A similar coating could be applied

PAGE 188

170 to vascular stents to promote EC monolayer growth and suppress excessive growth of smooth muscle cells. Figure 8-1. A design for a prototype small diameter vascular graft incorporating a microengineered surface layer and polymer composite artificial media that is compliance matched to the natural blood vessel it would replace. Endothelial Cells, Circulating Endothelial Cells and Progenitor Endothelial Cells A main question is how to recruit ECs to adhere to the surface of artificial grafts to act as blood-tissue interface. ECs are the ideal anti-thrombogenic blood contacting surface because they produce a host of biochemicals that prevent clotting, control vessel dilation, recruit leukocytes and more. In canine, porcine and bovine models PET and ePTFE vascular grafts re-endothelialize, but not in humans. Even after long-term implantation in humans, ECs extend at most ~1 cm into the graft from the anastomosis at

PAGE 189

171 the proximal end. Attempts to pre-seed grafts with ECs have met with poor success and loss of ECs as the lumen is remodeled over time. The reasons for poor EC adhesion to vascular grafts in humans is unknown, however there new methodologies to improve EC adhesion to surfaces. Rather then rely on EC ingrowth from the anastomoses into the graft, an alternative is to adhere circulating endothelial cells (CECs). CEC origin is unclear, whether they are ECs that have sloughed off the vascular system or matured in the bone marrow [124-127]. The number of CECs in the blood is an indicator of a number of disease states such as diabetes and arteriosclerosis [128]. In theory, these CECs can be used to re-endothelialize a surface through coatings with specific antibodies. This method is used to separate CECs from whole blood using antibody coated magnetic beads. In addition to the CECs are progenitor endothelial cells (PECs) that have not yet differentiated into CECs and have an increased capability to multiply and divide. The PECs originate in the bone marrow, but depending on developmental stage may be indistinguishable from CECs except for their increased proliferative capacity. This difference in proliferation and both CECs and PECs would be advantageous to adhere and re-grow a viable EC layer on a graft [125]. The ability to capture CECs and PECs to re-endothelialize artificial graft surfaces has been realized by others and explored in recent experiments. One example is the concept where a polymer scaffold is preseeded with CECs and then implanted as a complete artificial graft [129, 130]. This method utilizes a decellularized porcine artery leaving behind a collagen and elastin ECM. The CECs are extracted from the host’s peripheral blood, expanded in in vitro culture and then are used solely to repopulate the graft lumen. The grafts were placed in the carotid artery of sheep and demonstrated

PAGE 190

172 patency to 130 days. The TEBVs had a functional endothelium as measured by production of vWF and NO and infiltrated SMCs responded to vasoconstrictors. This study is important because it demonstrates the ability of CECs to repopulate and form a functional EC layer. Furthermore, it shows that a lumen consisting solely of EC can maintain patency when backed by an appropriate compliance matched substrate (the decellularized collagen and elastin porcine artery). However, this approach still suffers from the need for in vitro expansion of cells in order to have enough cells to preseed the graft lumen. Control studies that used the decellularized porcine artery without preseeding had no EC formation and occluded within 15 days, thus this method is not suitable for acute applications. Endothelial Cell Adhesion Molecules Specific molecules can be grafted to the PDMSe surface to enhance EC adhesion and growth. There are a large number proteins that have been identified in EC adhesion to ECM and other cells [131]. The cell surface receptors involved in firm adhesion to surfaces thorough focal contact adhesions are integrins. Integrins bind to the arginine – glycine – aspartic acid (RGD) sequence found in ECM proteins [131]. Fibronectin is a widely used ECM protein with this sequence and has been used to enhance the adhesion of ECs and other cells to polymer, metal and ceramic surfaces [13, 61, 132]. Coating a surface with fibronectin can serve to approximate ECM, especially when the spatially organization is controlled. A number of different growth factors have been identified for different cell types, for ECs the relevant growth factor for angiogenesis and re-endothelialization is vascular endothelial growth factor (VEGF) [133]. VEGF can induce ECs to form capillary-like tubes in polymeric scaffolding and elicit EC function from PECs. However, VEGF has negative impacts as well causing an increase in cardiac

PAGE 191

173 allograft arteriosclerosis and SMC based intimal thickening [134]. This makes concentrations of VEGF critical and it bioavailability in vivo will affects its bioactivity. Serum albumin is also important because it has poor adhesion characteristics, making it useful for spatially patterning cells [131].

PAGE 192

APPENDIX A ANALYSIS OF FORCE CURVES FOR ELASTIC MODULUS Determining the sample elastic modulus from the force curves required modeling the tip and sample interaction. A commonly used model for this interaction is the Hertz model. The Hertz model predicts the elastic deformation when two spheres are brought together under load [72]. Sneddon extended this relation to include the case of an infinitely stiff tip with a conical shape indenting a soft, planar sample [73]. In reality, the tip is not a cone, but rather a four-sided pyramid as seen in Figure A-1. The approximation of a cone is used to simplify the model and due to the small indentation depth, the error is assumed to be negligible. The force is related to the indentation depth by Equation A-1 )tan()1(222 EF where F is the loading force, E is the elastic modulus, is Poisson’s ratio, is the opening cone angle of the AFM tip and is the indentation depth. A B Figure A-1. 3-dimensional drawings representing (a) the conical tip used in the Sneddon model and (b) the actual tip shape, which is a four-sided pyramid (adapted from [1]). 174

PAGE 193

175 Unfortunately, the above model is flawed because during the z-axis extension there is both deflection of the cantilever and deformation of the soft polymer sample. This was addressed by Radmacher et al who defined the deflection of the cantilever d as equal to the movement of the z piezoelectric, z, minus the sample deformation, [134]. The cantilever deflection and samEquation A-2 cantilever force can be described as a spring using Hooke’s law and can include bothple deformation. Taking both into account yields, )( zkkdF whereEquation A-3 k is defined as the spring constant of the AFM cantilever. If the Hertz model is applied to the linear contact region of the force as curve seen in Figure A-2, then equations A-1 and A-2 can be combined yielding, )tan()1(2)(20Eddkwhere the value d is the zero deflection point and is equivalent to the deflection 00ddzz 0 in the noncontact region of the force curve. The values for lving simultaneous equations. For the V-shaantilever used, k = 0.06 N/m and = 35.Determining the AFM cantilever spring constant requires careful calibration. There z and d can be read directly from the contact region of the curve leaving z 0 and E as the only two unknowns. By selecting two points on the linear region of the force curves (z 1 , d 1 ) and (z 2 , d 2 ), the elastic modulus and z-axis initial contact point can be calculated by so ped silicon nitride c are a number of ways to do this that depend on the type of cantilever and whether or not the calibration is destructive [81, 82]. For the low spring constant cantilevers needed to examine biological materials, the V-shape requires calibration against a reference

PAGE 194

176 cantilever [81]. This method results in a 30% error, not much better than the 50% error associated with the manufacturer reported spring constant. As such, most cantilevers were not calibrated for force curve measurements assuming the ma nufacturer reported valuecurve in order to calculate the elastic modulus of the sample. The points (z1, from Radmacher et al) [134] as a bulk property. The problem is that e increased de material type and consider any surface treatments that may have been applied. This is s. There is a brand new ‘diving board’ style cantilever manufactured by Olympus and sold through Asylum Research that should allow calibration with only 10% error using non-destructive methods based on the thermal resonance [82]. Future studies should consider these cantilevers as an option. Figure A-2. Sketch illustrating the method for analyzing the linear portion of the force d 1 ) and (z 2 , d 2 ) are taken from the linear contact region of the curve. (Adapted Selecting a value for the Poisson’s ratio to plug into equation A-3 can be problematic depending on the sample. For PDMSe at ambient conditions the Poisson’s ratio was assumed to be 0.5 for an ideal elastomer. PS was assumed to have a Poisson’s ratio of 0.33 as the typical value reported for PS d 2 d 1 d 0 E, z 0 range of ana l y s i s Deflection Loading Force Z 2 Z 1 Sample height many materials, especially polymers, may be in a different state at the surface due to thegree of freedom. Thus, the Poisson’s ratio needs to be carefully selected for th

PAGE 195

177 comp licated even further in the case of cells and biological materials. The Poisson’s ratio is typically assumed to be 0.5; however the cells mechanical properties are far from that of an ideal elastomer. Considering cells as a fluid filled membrane sack with an actin cytoskeleton means that there will be viscous drag associated with the movement of cytoplasm through the actin network during mechanical loading. Proper interpretation of the data requires consideration of this assumption until a better model has been developed.

PAGE 196

APPENDIX B STATISTICAL ANALYSIS OF CELL RESPONSE TO ENGINEERED SURFACES Cell Density Two Way Analysis of Variance Data source: Data 1 in EC_on_surfaces_stats.SNB Balanced Design Dependent Variable: Cell Density Normality Test: Passed (P > 0.050) Equal Variance Test: Passed (P = 0.499) Source of Variation DF SS MS F P Surface Type 6 14321093.786 2386848.964 56.950 <0.001 Culture Time 2 379958.024 189979.012 4.533 0.014 Surface Type x Culture Time 12 6707594.143 558966.179 13.337 <0.001 Residual 63 2640413.000 41911.317 Total 83 24049058.952 289747.698 The difference in the mean values among the different levels of Surface Type is greater than would be expected by chance after allowing for effects of differences in Culture Time. There is a statistically significant difference (P = <0.001). To isolate which group(s) differ from the others use a multiple comparison procedure. The difference in the mean values among the different levels of Culture Time is greater than would be expected by chance after allowing for effects of differences in Surface Type. There is a statistically significant difference (P = 0.014). To isolate which group(s) differ from the others use a multiple comparison procedure. The effect of different levels of Surface Type depends on what level of Culture Time is present. There is a statistically significant interaction between Surface Type and Culture Time. (P = <0.001) Power of performed test with alpha = 0.0500: for Surface Type : 1.000 Power of performed test with alpha = 0.0500: for Culture Time : 0.635 Power of performed test with alpha = 0.0500: for Surface Type x Culture Time : 1.000 Least square means for Surface Type : Group Mean ps 962.750 ps_fn 935.750 pdmse_plasma 502.500 pdmse_FN 411.250 ps_FN_star 450.250 178

PAGE 197

179 pdmse_FN_star 855.250 pdmse_FN_circle 1688.583 Std Err of LS Mean = 59.098 Least square means for Culture Time : Group Mean 4.000 746.250 7.000 910.964 14.000 831.214 Std Err of LS Mean = 38.689 Least square means for Surface Type x Culture Time : Group Mean ps x 4.000 977.750 ps x 7.000 1008.500 ps x 14.000 902.000 ps_fn x 4.000 814.250 ps_fn x 7.000 1064.250 ps_fn x 14.000 928.750 pdmse_plasma x 4.000 527.250 pdmse_plasma x 7.000 409.000 pdmse_plasma x 14.000 571.250 pdmse_FN x 4.000 497.250 pdmse_FN x 7.000 250.000 pdmse_FN x 14.000 486.500 ps_FN_star x 4.000 531.000 ps_FN_star x 7.000 465.750 ps_FN_star x 14.000 354.000 pdmse_FN_star x 4.000 794.000 pdmse_FN_star x 7.000 1541.000 pdmse_FN_star x 14.000 230.750 pdmse_FN_circle x 4.000 1082.250 pdmse_FN_circle x 7.000 1638.250 pdmse_FN_circle x 14.000 2345.250 Std Err of LS Mean = 102.361 All Pairwise Multiple Comparison Procedures (Tukey Test): Comparisons for factor: Surface Type Comparison Diff of Means p q P P<0.050 pdmse_FN_circle vs. pdmse_FN 1277.333 7 21.614 <0.001 Yes pdmse_FN_circle vs. ps_FN_star 1238.333 7 20.954 <0.001 Yes pdmse_FN_cir vs. pdmse_plasma 1186.083 7 20.070 <0.001 Yes pdmse_FN_cir vs. pdmse_FN_sta 833.333 7 14.101 <0.001 Yes pdmse_FN_circle vs. ps_fn 752.833 7 12.739 <0.001 Yes pdmse_FN_circle vs. ps 725.833 7 12.282 <0.001 Yes ps vs. pdmse_FN 551.500 7 9.332 <0.001 Yes ps vs. ps_FN_star 512.500 7 8.672 <0.001 Yes ps vs. pdmse_plasma 460.250 7 7.788 <0.001 Yes ps vs. pdmse_FN_star 107.500 7 1.819 0.856 No ps vs. ps_fn 27.000 7 0.457 1.000 Do Not Test ps_fn vs. pdmse_FN 524.500 7 8.875 <0.001 Yes ps_fn vs. ps_FN_star 485.500 7 8.215 <0.001 Yes ps_fn vs. pdmse_plasma 433.250 7 7.331 <0.001 Yes ps_fn vs. pdmse_FN_star 80.500 7 1.362 0.960 Do Not Test

PAGE 198

180 pdmse_FN_star vs. pdmse_FN 444.000 7 7.513 <0.001 Yes pdmse_FN_star vs. ps_FN_star 405.000 7 6.853 <0.001 Yes pdmse_FN_star vs. pdmse_plasma 352.750 7 5.969 0.002 Yes pdmse_plasma vs. pdmse_FN 91.250 7 1.544 0.928 No pdmse_plasma vs. ps_FN_star 52.250 7 0.884 0.996 Do Not Test ps_FN_star vs. pdmse_FN 39.000 7 0.660 0.999 Do Not Test Comparisons for factor: Culture Time Comparison Diff of Means p q P P<0.050 7.000 vs. 4.000 164.714 3 4.257 0.010 Yes 7.000 vs. 14.000 79.750 3 2.061 0.318 No 14.000 vs. 4.000 84.964 3 2.196 0.274 No Comparisons for factor: Culture Time within ps Comparison Diff of Means p q P P<0.05 7.000 vs. 14.000 106.500 3 1.040 0.743 No 7.000 vs. 4.000 30.750 3 0.300 0.976 Do Not Test 4.000 vs. 14.000 75.750 3 0.740 0.860 Do Not Test Comparisons for factor: Culture Time within ps_fn Comparison Diff of Means p q P P<0.05 7.000 vs. 4.000 250.000 3 2.442 0.203 No 7.000 vs. 14.000 135.500 3 1.324 0.620 Do Not Test 14.000 vs. 4.000 114.500 3 1.119 0.710 Do Not Test Comparisons for factor: Culture Time within pdmse_plasma Comparison Diff of Means p q P P<0.05 14.000 vs. 7.000 162.250 3 1.585 0.505 No 14.000 vs. 4.000 44.000 3 0.430 0.950 Do Not Test 4.000 vs. 7.000 118.250 3 1.155 0.694 Do Not Test Comparisons for factor: Culture Time within pdmse_FN Comparison Diff of Means p q P P<0.05 4.000 vs. 7.000 247.250 3 2.415 0.210 No 4.000 vs. 14.000 10.750 3 0.105 0.997 Do Not Test 14.000 vs. 7.000 236.500 3 2.310 0.239 Do Not Test Comparisons for factor: Culture Time within ps_FN_star Comparison Diff of Means p q P P<0.05 4.000 vs. 14.000 177.000 3 1.729 0.444 No 4.000 vs. 7.000 65.250 3 0.637 0.894 Do Not Test 7.000 vs. 14.000 111.750 3 1.092 0.722 Do Not Test Comparisons for factor: Culture Time within pdmse_FN_star Comparison Diff of Means p q P P<0.05 7.000 vs. 14.000 1310.250 3 12.800 <0.001 Yes 7.000 vs. 4.000 747.000 3 7.298 <0.001 Yes 4.000 vs. 14.000 563.250 3 5.503 <0.001 Yes

PAGE 199

181 Comparisons for factor: Culture Time within pdmse_FN_circle Comparison Diff of Means p q P P<0.05 14.000 vs. 4.000 1263.000 3 12.339 <0.001 Yes 14.000 vs. 7.000 707.000 3 6.907 <0.001 Yes 7.000 vs. 4.000 556.000 3 5.432 <0.001 Yes Comparisons for factor: Surface Type within 4 Comparison Diff of Means p q P P<0.05 pdmse_FN_circle vs. pdmse_FN 585.000 7 5.715 0.003 Yes pdmse_FN_cir vs. pdmse_plasma 555.000 7 5.422 0.005 Yes pdmse_FN_circle vs. ps_FN_star 551.250 7 5.385 0.006 Yes pdmse_FN_cir vs. pdmse_FN_sta 288.250 7 2.816 0.430 No pdmse_FN_circle vs. ps_fn 268.000 7 2.618 0.520 Do Not Test pdmse_FN_circle vs. ps 104.500 7 1.021 0.991 Do Not Test ps vs. pdmse_FN 480.500 7 4.694 0.024 Yes ps vs. pdmse_plasma 450.500 7 4.401 0.042 Yes ps vs. ps_FN_star 446.750 7 4.364 0.045 Yes ps vs. pdmse_FN_star 183.750 7 1.795 0.863 Do Not Test ps vs. ps_fn 163.500 7 1.597 0.917 Do Not Test ps_fn vs. pdmse_FN 317.000 7 3.097 0.316 No ps_fn vs. pdmse_plasma 287.000 7 2.804 0.436 Do Not Test ps_fn vs. ps_FN_star 283.250 7 2.767 0.452 Do Not Test ps_fn vs. pdmse_FN_star 20.250 7 0.198 1.000 Do Not Test pdmse_FN_star vs. pdmse_FN 296.750 7 2.899 0.395 Do Not Test pdmse_FN_star vs. pdmse_plasma 266.750 7 2.606 0.525 Do Not Test pdmse_FN_star vs. ps_FN_star 263.000 7 2.569 0.542 Do Not Test ps_FN_star vs. pdmse_FN 33.750 7 0.330 1.000 Do Not Test ps_FN_star vs. pdmse_plasma 3.750 7 0.0366 1.000 Do Not Test pdmse_plasma vs. pdmse_FN 30.000 7 0.293 1.000 Do Not Test Comparisons for factor: Surface Type within 7 Comparison Diff of Means p q P P<0.05 pdmse_FN_circle vs. pdmse_FN 1388.250 7 13.562 <0.001 Yes pdmse_FN_cir vs. pdmse_plasma 1229.250 7 12.009 <0.001 Yes pdmse_FN_circle vs. ps_FN_star 1172.500 7 11.455 <0.001 Yes pdmse_FN_circle vs. ps 629.750 7 6.152 0.001 Yes pdmse_FN_circle vs. ps_fn 574.000 7 5.608 0.004 Yes pdmse_FN_cir vs. pdmse_FN_sta 97.250 7 0.950 0.994 No pdmse_FN_star vs. pdmse_FN 1291.000 7 12.612 <0.001 Yes pdmse_FN_star vs. pdmse_plasma 1132.000 7 11.059 <0.001 Yes pdmse_FN_star vs. ps_FN_star 1075.250 7 10.504 <0.001 Yes pdmse_FN_star vs. ps 532.500 7 5.202 0.008 Yes pdmse_FN_star vs. ps_fn 476.750 7 4.658 0.026 Yes ps_fn vs. pdmse_FN 814.250 7 7.955 <0.001 Yes ps_fn vs. pdmse_plasma 655.250 7 6.401 <0.001 Yes ps_fn vs. ps_FN_star 598.500 7 5.847 0.002 Yes ps_fn vs. ps 55.750 7 0.545 1.000 No ps vs. pdmse_FN 758.500 7 7.410 <0.001 Yes ps vs. pdmse_plasma 599.500 7 5.857 0.002 Yes ps vs. ps_FN_star 542.750 7 5.302 0.007 Yes ps_FN_star vs. pdmse_FN 215.750 7 2.108 0.749 No ps_FN_star vs. pdmse_plasma 56.750 7 0.554 1.000 Do Not Test pdmse_plasma vs. pdmse_FN 159.000 7 1.553 0.926 Do Not Test

PAGE 200

182 Comparisons for factor: Surface Type within 14 Comparison Diff of Means p q P P<0.05 pdmse_FN_cir vs. pdmse_FN_sta 2114.500 7 20.657 <0.001 Yes pdmse_FN_circle vs. ps_FN_star 1991.250 7 19.453 <0.001 Yes pdmse_FN_circle vs. pdmse_FN 1858.750 7 18.159 <0.001 Yes pdmse_FN_cir vs. pdmse_plasma 1774.000 7 17.331 <0.001 Yes pdmse_FN_circle vs. ps 1443.250 7 14.100 <0.001 Yes pdmse_FN_circle vs. ps_fn 1416.500 7 13.838 <0.001 Yes ps_fn vs. pdmse_FN_star 698.000 7 6.819 <0.001 Yes ps_fn vs. ps_FN_star 574.750 7 5.615 0.003 Yes ps_fn vs. pdmse_FN 442.250 7 4.320 0.049 Yes ps_fn vs. pdmse_plasma 357.500 7 3.493 0.188 No ps_fn vs. ps 26.750 7 0.261 1.000 Do Not Test ps vs. pdmse_FN_star 671.250 7 6.558 <0.001 Yes ps vs. ps_FN_star 548.000 7 5.354 0.006 Yes ps vs. pdmse_FN 415.500 7 4.059 0.078 No ps vs. pdmse_plasma 330.750 7 3.231 0.268 Do Not Test pdmse_plasma vs. pdmse_FN_star 340.500 7 3.326 0.237 No pdmse_plasma vs. ps_FN_star 217.250 7 2.122 0.743 Do Not Test pdmse_plasma vs. pdmse_FN 84.750 7 0.828 0.997 Do Not Test pdmse_FN vs. pdmse_FN_star 255.750 7 2.499 0.575 Do Not Test pdmse_FN vs. ps_FN_star 132.500 7 1.294 0.969 Do Not Test ps_FN_star vs. pdmse_FN_star 123.250 7 1.204 0.978 Do Not Test A result of "Do Not Test" occurs for a comparison when no significant difference is found between two means that enclose that comparison. For example, if you had four means sorted in order, and found no difference between means 4 vs. 2, then you would not test 4 vs. 3 and 3 vs. 2, but still test 4 vs. 1 and 3 vs. 1 (4 vs. 3 and 3 vs. 2 are enclosed by 4 vs. 2: 4 3 2 1). Note that not testing the enclosed means is a procedural rule, and a result of Do Not Test should be treated as if there is no significant difference between the means, even though one may appear to exist. Percent Coverage Two Way Analysis of Variance Data source: Data 1 in EC_on_surfaces_stats.SNB Balanced Design Dependent Variable: Percent Coverage Normality Test: Failed (P = <0.001) Equal Variance Test: Failed (P = <0.001) Source of Variation DF SS MS F P Surface Type 6 61693.749 10282.292 60.189 <0.001 Culture Time 2 2573.105 1286.553 7.531 0.001 Surface Type x Culture Time 12 15777.762 1314.813 7.697 <0.001 Residual 63 10762.440 170.832 Total 83 90807.056 1094.061

PAGE 201

183 The difference in the mean values among the different levels of Surface Type is greater than would be expected by chance after allowing for effects of differences in Culture Time. There is a statistically significant difference (P = <0.001). To isolate which group(s) differ from the others use a multiple comparison procedure. The difference in the mean values among the different levels of Culture Time is greater than would be expected by chance after allowing for effects of differences in Surface Type. There is a statistically significant difference (P = 0.001). To isolate which group(s) differ from the others use a multiple comparison procedure. The effect of different levels of Surface Type depends on what level of Culture Time is present. There is a statistically significant interaction between Surface Type and Culture Time. (P = <0.001) Power of performed test with alpha = 0.0500: for Surface Type : 1.000 Power of performed test with alpha = 0.0500: for Culture Time : 0.908 Power of performed test with alpha = 0.0500: for Surface Type x Culture Time : 1.000 Least square means for Surface Type : Group Mean ps 95.850 ps_fn 99.942 pdmse_plasma 29.533 pdmse_FN 38.258 ps_FN_star 82.742 pdmse_FN_star 54.725 pdmse_FN_circle 93.200 Std Err of LS Mean = 3.773 Least square means for Culture Time : Group Mean 4.000 66.368 7.000 78.425 14.000 67.029 Std Err of LS Mean = 2.470 Least square means for Surface Type x Culture Time : Group Mean ps x 4.000 87.575 ps x 7.000 99.975 ps x 14.000 100.000 ps_fn x 4.000 99.825 ps_fn x 7.000 100.000 ps_fn x 14.000 100.000 pdmse_plasma x 4.000 23.850 pdmse_plasma x 7.000 36.350 pdmse_plasma x 14.000 28.400 pdmse_FN x 4.000 37.650 pdmse_FN x 7.000 36.125 pdmse_FN x 14.000 41.000 ps_FN_star x 4.000 58.150 ps_FN_star x 7.000 92.375 ps_FN_star x 14.000 97.700 pdmse_FN_star x 4.000 71.450 pdmse_FN_star x 7.000 84.850 pdmse_FN_star x 14.000 7.875

PAGE 202

184 pdmse_FN_circle x 4.000 86.075 pdmse_FN_circle x 7.000 99.300 pdmse_FN_circle x 14.000 94.225 Std Err of LS Mean = 6.535 All Pairwise Multiple Comparison Procedures (Tukey Test): Comparisons for factor: Surface Type Comparison Diff of Means p q P P<0.050 ps_fn vs. pdmse_plasma 70.408 7 18.661 <0.001 Yes ps_fn vs. pdmse_FN 61.683 7 16.348 <0.001 Yes ps_fn vs. pdmse_FN_star 45.217 7 11.984 <0.001 Yes ps_fn vs. ps_FN_star 17.200 7 4.559 0.031 Yes ps_fn vs. pdmse_FN_circle 6.742 7 1.787 0.866 No ps_fn vs. ps 4.092 7 1.084 0.987 Do Not Test ps vs. pdmse_plasma 66.317 7 17.576 <0.001 Yes ps vs. pdmse_FN 57.592 7 15.264 <0.001 Yes ps vs. pdmse_FN_star 41.125 7 10.900 <0.001 Yes ps vs. ps_FN_star 13.108 7 3.474 0.193 No ps vs. pdmse_FN_circle 2.650 7 0.702 0.999 Do Not Test pdmse_FN_cir vs. pdmse_plasma 63.667 7 16.874 <0.001 Yes pdmse_FN_circle vs. pdmse_FN 54.942 7 14.562 <0.001 Yes pdmse_FN_cir vs. pdmse_FN_sta 38.475 7 10.197 <0.001 Yes pdmse_FN_circle vs. ps_FN_star 10.458 7 2.772 0.450 Do Not Test ps_FN_star vs. pdmse_plasma 53.208 7 14.102 <0.001 Yes ps_FN_star vs. pdmse_FN 44.483 7 11.790 <0.001 Yes ps_FN_star vs. pdmse_FN_star 28.017 7 7.425 <0.001 Yes pdmse_FN_star vs. pdmse_plasma 25.192 7 6.677 <0.001 Yes pdmse_FN_star vs. pdmse_FN 16.467 7 4.364 0.045 Yes pdmse_FN vs. pdmse_plasma 8.725 7 2.312 0.661 No Comparisons for factor: Culture Time Comparison Diff of Means p q P P<0.050 7.000 vs. 4.000 12.057 3 4.881 0.003 Yes 7.000 vs. 14.000 11.396 3 4.614 0.005 Yes 14.000 vs. 4.000 0.661 3 0.267 0.981 No Comparisons for factor: Culture Time within ps Comparison Diff of Means p q P P<0.05 14.000 vs. 4.000 12.425 3 1.901 0.376 No 14.000 vs. 7.000 0.0250 3 0.00383 1.000 Do Not Test 7.000 vs. 4.000 12.400 3 1.897 0.378 Do Not Test Comparisons for factor: Culture Time within ps_fn Comparison Diff of Means p q P P<0.05 7.000 vs. 4.000 0.175 3 0.0268 1.000 No 7.000 vs. 14.000 0.000 3 0.000 1.000 Do Not Test 14.000 vs. 4.000 0.175 3 0.0268 1.000 Do Not Test Comparisons for factor: Culture Time within pdmse_plasma Comparison Diff of Means p q P P<0.05

PAGE 203

185 7.000 vs. 4.000 12.500 3 1.913 0.372 No 7.000 vs. 14.000 7.950 3 1.216 0.667 Do Not Test 14.000 vs. 4.000 4.550 3 0.696 0.875 Do Not Test Comparisons for factor: Culture Time within pdmse_FN Comparison Diff of Means p q P P<0.05 14.000 vs. 7.000 4.875 3 0.746 0.858 No 14.000 vs. 4.000 3.350 3 0.513 0.930 Do Not Test 4.000 vs. 7.000 1.525 3 0.233 0.985 Do Not Test Comparisons for factor: Culture Time within ps_FN_star Comparison Diff of Means p q P P<0.05 14.000 vs. 4.000 39.550 3 6.052 <0.001 Yes 14.000 vs. 7.000 5.325 3 0.815 0.833 No 7.000 vs. 4.000 34.225 3 5.237 0.001 Yes Comparisons for factor: Culture Time within pdmse_FN_star Comparison Diff of Means p q P P<0.05 7.000 vs. 14.000 76.975 3 11.779 <0.001 Yes 7.000 vs. 4.000 13.400 3 2.050 0.322 No 4.000 vs. 14.000 63.575 3 9.728 <0.001 Yes Comparisons for factor: Culture Time within pdmse_FN_circle Comparison Diff of Means p q P P<0.05 7.000 vs. 4.000 13.225 3 2.024 0.331 No 7.000 vs. 14.000 5.075 3 0.777 0.847 Do Not Test 14.000 vs. 4.000 8.150 3 1.247 0.654 Do Not Test Comparisons for factor: Surface Type within 4 Comparison Diff of Means p q P P<0.05 ps_fn vs. pdmse_plasma 75.975 7 11.626 <0.001 Yes ps_fn vs. pdmse_FN 62.175 7 9.514 <0.001 Yes ps_fn vs. ps_FN_star 41.675 7 6.377 <0.001 Yes ps_fn vs. pdmse_FN_star 28.375 7 4.342 0.047 Yes ps_fn vs. pdmse_FN_circle 13.750 7 2.104 0.751 No ps_fn vs. ps 12.250 7 1.874 0.837 Do Not Test ps vs. pdmse_plasma 63.725 7 9.751 <0.001 Yes ps vs. pdmse_FN 49.925 7 7.639 <0.001 Yes ps vs. ps_FN_star 29.425 7 4.503 0.035 Yes ps vs. pdmse_FN_star 16.125 7 2.467 0.589 No ps vs. pdmse_FN_circle 1.500 7 0.230 1.000 Do Not Test pdmse_FN_cir vs. pdmse_plasma 62.225 7 9.522 <0.001 Yes pdmse_FN_circle vs. pdmse_FN 48.425 7 7.410 <0.001 Yes pdmse_FN_circle vs. ps_FN_star 27.925 7 4.273 0.053 No pdmse_FN_cir vs. pdmse_FN_sta 14.625 7 2.238 0.694 Do Not Test pdmse_FN_star vs. pdmse_plasma 47.600 7 7.284 <0.001 Yes pdmse_FN_star vs. pdmse_FN 33.800 7 5.172 0.009 Yes pdmse_FN_star vs. ps_FN_star 13.300 7 2.035 0.779 Do Not Test ps_FN_star vs. pdmse_plasma 34.300 7 5.249 0.008 Yes ps_FN_star vs. pdmse_FN 20.500 7 3.137 0.301 No pdmse_FN vs. pdmse_plasma 13.800 7 2.112 0.748 No

PAGE 204

186 Comparisons for factor: Surface Type within 7 Comparison Diff of Means p q P P<0.05 ps_fn vs. pdmse_FN 63.875 7 9.774 <0.001 Yes ps_fn vs. pdmse_plasma 63.650 7 9.740 <0.001 Yes ps_fn vs. pdmse_FN_star 15.150 7 2.318 0.658 No ps_fn vs. ps_FN_star 7.625 7 1.167 0.981 Do Not Test ps_fn vs. pdmse_FN_circle 0.700 7 0.107 1.000 Do Not Test ps_fn vs. ps 0.0250 7 0.00383 1.000 Do Not Test ps vs. pdmse_FN 63.850 7 9.770 <0.001 Yes ps vs. pdmse_plasma 63.625 7 9.736 <0.001 Yes ps vs. pdmse_FN_star 15.125 7 2.314 0.660 Do Not Test ps vs. ps_FN_star 7.600 7 1.163 0.982 Do Not Test ps vs. pdmse_FN_circle 0.675 7 0.103 1.000 Do Not Test pdmse_FN_circle vs. pdmse_FN 63.175 7 9.667 <0.001 Yes pdmse_FN_cir vs. pdmse_plasma 62.950 7 9.633 <0.001 Yes pdmse_FN_cir vs. pdmse_FN_sta 14.450 7 2.211 0.706 Do Not Test pdmse_FN_circle vs. ps_FN_star 6.925 7 1.060 0.989 Do Not Test ps_FN_star vs. pdmse_FN 56.250 7 8.607 <0.001 Yes ps_FN_star vs. pdmse_plasma 56.025 7 8.573 <0.001 Yes ps_FN_star vs. pdmse_FN_star 7.525 7 1.151 0.983 Do Not Test pdmse_FN_star vs. pdmse_FN 48.725 7 7.456 <0.001 Yes pdmse_FN_star vs. pdmse_plasma 48.500 7 7.421 <0.001 Yes pdmse_plasma vs. pdmse_FN 0.225 7 0.0344 1.000 No Comparisons for factor: Surface Type within 14 Comparison Diff of Means p q P P<0.05 ps vs. pdmse_FN_star 92.125 7 14.097 <0.001 Yes ps vs. pdmse_plasma 71.600 7 10.956 <0.001 Yes ps vs. pdmse_FN 59.000 7 9.028 <0.001 Yes ps vs. pdmse_FN_circle 5.775 7 0.884 0.996 No ps vs. ps_FN_star 2.300 7 0.352 1.000 Do Not Test ps vs. ps_fn 0.000 7 0.000 1.000 Do Not Test ps_fn vs. pdmse_FN_star 92.125 7 14.097 <0.001 Yes ps_fn vs. pdmse_plasma 71.600 7 10.956 <0.001 Yes ps_fn vs. pdmse_FN 59.000 7 9.028 <0.001 Yes ps_fn vs. pdmse_FN_circle 5.775 7 0.884 0.996 Do Not Test ps_fn vs. ps_FN_star 2.300 7 0.352 1.000 Do Not Test ps_FN_star vs. pdmse_FN_star 89.825 7 13.745 <0.001 Yes ps_FN_star vs. pdmse_plasma 69.300 7 10.604 <0.001 Yes ps_FN_star vs. pdmse_FN 56.700 7 8.676 <0.001 Yes ps_FN_star vs. pdmse_FN_circle 3.475 7 0.532 1.000 Do Not Test pdmse_FN_cir vs. pdmse_FN_sta 86.350 7 13.213 <0.001 Yes pdmse_FN_cir vs. pdmse_plasma 65.825 7 10.072 <0.001 Yes pdmse_FN_circle vs. pdmse_FN 53.225 7 8.144 <0.001 Yes pdmse_FN vs. pdmse_FN_star 33.125 7 5.069 0.011 Yes pdmse_FN vs. pdmse_plasma 12.600 7 1.928 0.819 No pdmse_plasma vs. pdmse_FN_star 20.525 7 3.141 0.299 No A result of "Do Not Test" occurs for a comparison when no significant difference is found between two means that enclose that comparison. For example, if you had four means sorted in order, and found no difference between means 4 vs. 2, then you would not test 4 vs. 3 and 3 vs. 2, but still test 4 vs. 1 and 3 vs. 1 (4 vs. 3 and 3 vs. 2 are enclosed by 4 vs. 2: 4 3 2 1). Note that not testing the enclosed means is a

PAGE 205

187 procedural rule, and a result of Do Not Test should be treated as if there is no significant difference between the means, even though one may appear to exist. Cell Area Two Way Analysis of Variance Data source: Data 1 in EC_on_surfaces_stats.SNB Balanced Design Dependent Variable: Cell Area Normality Test: Failed (P = 0.002) Equal Variance Test: Failed (P = <0.001) Source of Variation DF SS MS F P Surface Type 6 40503645.284 6750607.547 73.413 <0.001 Culture Time 2 1005253.363 502626.682 5.466 0.006 Surface Type x Culture Time 12 19335984.049 1611332.004 17.523 <0.001 Residual 63 5793120.320 91954.291 Total 83 66638003.017 802867.506 The difference in the mean values among the different levels of Surface Type is greater than would be expected by chance after allowing for effects of differences in Culture Time. There is a statistically significant difference (P = <0.001). To isolate which group(s) differ from the others use a multiple comparison procedure. The difference in the mean values among the different levels of Culture Time is greater than would be expected by chance after allowing for effects of differences in Surface Type. There is a statistically significant difference (P = 0.006). To isolate which group(s) differ from the others use a multiple comparison procedure. The effect of different levels of Surface Type depends on what level of Culture Time is present. There is a statistically significant interaction between Surface Type and Culture Time. (P = <0.001) Power of performed test with alpha = 0.0500: for Surface Type : 1.000 Power of performed test with alpha = 0.0500: for Culture Time : 0.751 Power of performed test with alpha = 0.0500: for Surface Type x Culture Time : 1.000 Least square means for Surface Type : Group Mean ps 1566.245 ps_fn 1673.548 pdmse_plasma 905.857 pdmse_FN 1465.829 ps_FN_star 3029.790 pdmse_FN_star 930.735 pdmse_FN_circle 932.281 Std Err of LS Mean = 87.538 Least square means for Culture Time :

PAGE 206

188 Group Mean 4.000 1349.975 7.000 1606.465 14.000 1545.397 Std Err of LS Mean = 57.307 Least square means for Surface Type x Culture Time : Group Mean ps x 4.000 1442.191 ps x 7.000 1549.714 ps x 14.000 1706.830 ps_fn x 4.000 1892.583 ps_fn x 7.000 1450.750 ps_fn x 14.000 1677.313 pdmse_plasma x 4.000 710.779 pdmse_plasma x 7.000 1303.778 pdmse_plasma x 14.000 703.014 pdmse_FN x 4.000 1103.940 pdmse_FN x 7.000 2099.105 pdmse_FN x 14.000 1194.443 ps_FN_star x 4.000 1681.186 ps_FN_star x 7.000 3054.603 ps_FN_star x 14.000 4353.582 pdmse_FN_star x 4.000 1382.989 pdmse_FN_star x 7.000 844.008 pdmse_FN_star x 14.000 565.208 pdmse_FN_circle x 4.000 1236.160 pdmse_FN_circle x 7.000 943.293 pdmse_FN_circle x 14.000 617.390 Std Err of LS Mean = 151.620 All Pairwise Multiple Comparison Procedures (Tukey Test): Comparisons for factor: Surface Type Comparison Diff of Means p q P P<0.050 ps_FN_star vs. pdmse_plasma 2123.933 7 24.263 <0.001 Yes ps_FN_star vs. pdmse_FN_star 2099.055 7 23.979 <0.001 Yes ps_FN_star vs. pdmse_FN_circle 2097.509 7 23.961 <0.001 Yes ps_FN_star vs. pdmse_FN 1563.961 7 17.866 <0.001 Yes ps_FN_star vs. ps 1463.545 7 16.719 <0.001 Yes ps_FN_star vs. ps_fn 1356.242 7 15.493 <0.001 Yes ps_fn vs. pdmse_plasma 767.691 7 8.770 <0.001 Yes ps_fn vs. pdmse_FN_star 742.813 7 8.486 <0.001 Yes ps_fn vs. pdmse_FN_circle 741.267 7 8.468 <0.001 Yes ps_fn vs. pdmse_FN 207.719 7 2.373 0.633 No ps_fn vs. ps 107.303 7 1.226 0.976 Do Not Test ps vs. pdmse_plasma 660.388 7 7.544 <0.001 Yes ps vs. pdmse_FN_star 635.510 7 7.260 <0.001 Yes ps vs. pdmse_FN_circle 633.964 7 7.242 <0.001 Yes ps vs. pdmse_FN 100.416 7 1.147 0.983 Do Not Test pdmse_FN vs. pdmse_plasma 559.972 7 6.397 <0.001 Yes pdmse_FN vs. pdmse_FN_star 535.094 7 6.113 0.001 Yes pdmse_FN vs. pdmse_FN_circle 533.548 7 6.095 0.001 Yes pdmse_FN_cir vs. pdmse_plasma 26.424 7 0.302 1.000 No pdmse_FN_cir vs. pdmse_FN_sta 1.546 7 0.0177 1.000 Do Not Test

PAGE 207

189 pdmse_FN_star vs. pdmse_plasma 24.878 7 0.284 1.000 Do Not Test Comparisons for factor: Culture Time Comparison Diff of Means p q P P<0.050 7.000 vs. 4.000 256.489 3 4.476 0.007 Yes 7.000 vs. 14.000 61.067 3 1.066 0.733 No 14.000 vs. 4.000 195.422 3 3.410 0.049 Yes Comparisons for factor: Culture Time within ps Comparison Diff of Means p q P P<0.05 14.000 vs. 4.000 264.640 3 1.745 0.438 No 14.000 vs. 7.000 157.116 3 1.036 0.745 Do Not Test 7.000 vs. 4.000 107.524 3 0.709 0.871 Do Not Test Comparisons for factor: Culture Time within ps_fn Comparison Diff of Means p q P P<0.05 4.000 vs. 7.000 441.833 3 2.914 0.107 No 4.000 vs. 14.000 215.270 3 1.420 0.577 Do Not Test 14.000 vs. 7.000 226.563 3 1.494 0.544 Do Not Test Comparisons for factor: Culture Time within pdmse_plasma Comparison Diff of Means p q P P<0.05 7.000 vs. 14.000 600.763 3 3.962 0.018 Yes 7.000 vs. 4.000 592.998 3 3.911 0.020 Yes 4.000 vs. 14.000 7.765 3 0.0512 0.999 No Comparisons for factor: Culture Time within pdmse_FN Comparison Diff of Means p q P P<0.05 7.000 vs. 4.000 995.166 3 6.564 <0.001 Yes 7.000 vs. 14.000 904.662 3 5.967 <0.001 Yes 14.000 vs. 4.000 90.504 3 0.597 0.907 No Comparisons for factor: Culture Time within ps_FN_star Comparison Diff of Means p q P P<0.05 14.000 vs. 4.000 2672.396 3 17.626 <0.001 Yes 14.000 vs. 7.000 1298.978 3 8.567 <0.001 Yes 7.000 vs. 4.000 1373.418 3 9.058 <0.001 Yes Comparisons for factor: Culture Time within pdmse_FN_star Comparison Diff of Means p q P P<0.05 4.000 vs. 14.000 817.781 3 5.394 0.001 Yes 4.000 vs. 7.000 538.981 3 3.555 0.038 Yes 7.000 vs. 14.000 278.800 3 1.839 0.400 No Comparisons for factor: Culture Time within pdmse_FN_circle Comparison Diff of Means p q P P<0.05 4.000 vs. 14.000 618.770 3 4.081 0.015 Yes 4.000 vs. 7.000 292.867 3 1.932 0.365 No

PAGE 208

190 7.000 vs. 14.000 325.904 3 2.149 0.289 No Comparisons for factor: Surface Type within 4 Comparison Diff of Means p q P P<0.05 ps_fn vs. pdmse_plasma 1181.804 7 7.795 <0.001 Yes ps_fn vs. pdmse_FN 788.643 7 5.201 0.008 Yes ps_fn vs. pdmse_FN_circle 656.422 7 4.329 0.048 Yes ps_fn vs. pdmse_FN_star 509.594 7 3.361 0.226 No ps_fn vs. ps 450.392 7 2.971 0.365 Do Not Test ps_fn vs. ps_FN_star 211.397 7 1.394 0.955 Do Not Test ps_FN_star vs. pdmse_plasma 970.407 7 6.400 <0.001 Yes ps_FN_star vs. pdmse_FN 577.246 7 3.807 0.117 No ps_FN_star vs. pdmse_FN_circle 445.026 7 2.935 0.380 Do Not Test ps_FN_star vs. pdmse_FN_star 298.197 7 1.967 0.805 Do Not Test ps_FN_star vs. ps 238.995 7 1.576 0.921 Do Not Test ps vs. pdmse_plasma 731.411 7 4.824 0.019 Yes ps vs. pdmse_FN 338.251 7 2.231 0.697 Do Not Test ps vs. pdmse_FN_circle 206.030 7 1.359 0.960 Do Not Test ps vs. pdmse_FN_star 59.201 7 0.390 1.000 Do Not Test pdmse_FN_star vs. pdmse_plasma 672.210 7 4.434 0.040 Yes pdmse_FN_star vs. pdmse_FN 279.049 7 1.840 0.849 Do Not Test pdmse_FN_sta vs. pdmse_FN_cir 146.829 7 0.968 0.993 Do Not Test pdmse_FN_cir vs. pdmse_plasma 525.381 7 3.465 0.196 No pdmse_FN_circle vs. pdmse_FN 132.221 7 0.872 0.996 Do Not Test pdmse_FN vs. pdmse_plasma 393.160 7 2.593 0.531 Do Not Test Comparisons for factor: Surface Type within 7 Comparison Diff of Means p q P P<0.05 ps_FN_star vs. pdmse_FN_star 2210.595 7 14.580 <0.001 Yes ps_FN_star vs. pdmse_FN_circle 2111.310 7 13.925 <0.001 Yes ps_FN_star vs. pdmse_plasma 1750.826 7 11.547 <0.001 Yes ps_FN_star vs. ps_fn 1603.853 7 10.578 <0.001 Yes ps_FN_star vs. ps 1504.889 7 9.925 <0.001 Yes ps_FN_star vs. pdmse_FN 955.498 7 6.302 <0.001 Yes pdmse_FN vs. pdmse_FN_star 1255.097 7 8.278 <0.001 Yes pdmse_FN vs. pdmse_FN_circle 1155.812 7 7.623 <0.001 Yes pdmse_FN vs. pdmse_plasma 795.328 7 5.246 0.008 Yes pdmse_FN vs. ps_fn 648.355 7 4.276 0.053 No pdmse_FN vs. ps 549.391 7 3.623 0.156 Do Not Test ps vs. pdmse_FN_star 705.706 7 4.654 0.026 Yes ps vs. pdmse_FN_circle 606.421 7 4.000 0.086 No ps vs. pdmse_plasma 245.937 7 1.622 0.911 Do Not Test ps vs. ps_fn 98.965 7 0.653 0.999 Do Not Test ps_fn vs. pdmse_FN_star 606.742 7 4.002 0.085 No ps_fn vs. pdmse_FN_circle 507.456 7 3.347 0.230 Do Not Test ps_fn vs. pdmse_plasma 146.972 7 0.969 0.993 Do Not Test pdmse_plasma vs. pdmse_FN_star 459.769 7 3.032 0.340 Do Not Test pdmse_plasma vs. pdmse_FN_cir 360.484 7 2.378 0.631 Do Not Test pdmse_FN_cir vs. pdmse_FN_sta 99.285 7 0.655 0.999 Do Not Test Comparisons for factor: Surface Type within 14 Comparison Diff of Means p q P P<0.05 ps_FN_star vs. pdmse_FN_star 3788.373 7 24.986 <0.001 Yes

PAGE 209

191 ps_FN_star vs. pdmse_FN_circle 3736.192 7 24.642 <0.001 Yes ps_FN_star vs. pdmse_plasma 3650.567 7 24.077 <0.001 Yes ps_FN_star vs. pdmse_FN 3159.138 7 20.836 <0.001 Yes ps_FN_star vs. ps_fn 2676.269 7 17.651 <0.001 Yes ps_FN_star vs. ps 2646.751 7 17.456 <0.001 Yes ps vs. pdmse_FN_star 1141.622 7 7.530 <0.001 Yes ps vs. pdmse_FN_circle 1089.440 7 7.185 <0.001 Yes ps vs. pdmse_plasma 1003.816 7 6.621 <0.001 Yes ps vs. pdmse_FN 512.387 7 3.379 0.220 No ps vs. ps_fn 29.518 7 0.195 1.000 Do Not Test ps_fn vs. pdmse_FN_star 1112.104 7 7.335 <0.001 Yes ps_fn vs. pdmse_FN_circle 1059.923 7 6.991 <0.001 Yes ps_fn vs. pdmse_plasma 974.298 7 6.426 <0.001 Yes ps_fn vs. pdmse_FN 482.869 7 3.185 0.284 Do Not Test pdmse_FN vs. pdmse_FN_star 629.235 7 4.150 0.066 No pdmse_FN vs. pdmse_FN_circle 577.053 7 3.806 0.118 Do Not Test pdmse_FN vs. pdmse_plasma 491.429 7 3.241 0.264 Do Not Test pdmse_plasma vs. pdmse_FN_star 137.806 7 0.909 0.995 Do Not Test pdmse_plasma vs. pdmse_FN_cir 85.624 7 0.565 1.000 Do Not Test pdmse_FN_cir vs. pdmse_FN_sta 52.182 7 0.344 1.000 Do Not Test A result of "Do Not Test" occurs for a comparison when no significant difference is found between two means that enclose that comparison. For example, if you had four means sorted in order, and found no difference between means 4 vs. 2, then you would not test 4 vs. 3 and 3 vs. 2, but still test 4 vs. 1 and 3 vs. 1 (4 vs. 3 and 3 vs. 2 are enclosed by 4 vs. 2: 4 3 2 1). Note that not testing the enclosed means is a procedural rule, and a result of Do Not Test should be treated as if there is no significant difference between the means, even though one may appear to exist. Size of Focal Contact Adhesions Two Way Analysis of Variance Data source: Data 1 in EC_on_surfaces_stats.SNB General Linear Model (No Interactions) Dependent Variable: Col 12 Normality Test: Passed (P > 0.050) Equal Variance Test: Passed (P = 0.250) Source of Variation DF SS MS F P surface type 3 4 1.695 0.424 7.427 <0.001 Culture time 3 2 2.475 1.238 21.696 <0.001 Residual 48 2.738 0.0570 Total 54 6.278 0.116 The difference in the mean values among the different levels of surface type 3 is greater than would be expected by chance after allowing for effects of differences in Culture time 3. There is a statistically significant difference (P = <0.001). To isolate which group(s) differ from the others use a multiple comparison procedure.

PAGE 210

192 The difference in the mean values among the different levels of Culture time 3 is greater than would be expected by chance after allowing for effects of differences in surface type 3. There is a statistically significant difference (P = <0.001). To isolate which group(s) differ from the others use a multiple comparison procedure. Power of performed test with alpha = 0.0500: for surface type 3 : 0.988 Power of performed test with alpha = 0.0500: for Culture time 3 : 1.000 Least square means for surface type 3 : Group Mean SEM ps 1.982 0.0690 ps_fn 1.733 0.0690 pdmse_plasma 1.831 0.0690 pdmse_FN 1.866 0.0722 pdmse_FN_circle 1.400 0.0881 Least square means for Culture time 3 : Group Mean SEM 4.000 2.023 0.0534 7.000 1.795 0.0534 14.000 1.470 0.0646 All Pairwise Multiple Comparison Procedures (Tukey Test): Comparisons for factor: surface type 3 Comparison Diff of Means p q P P<0.050 ps vs. pdmse_FN_circle 0.582 5 7.365 <0.001 Yes ps vs. ps_fn 0.249 5 3.614 0.095 No ps vs. pdmse_plasma 0.152 5 2.200 0.533 Do Not Test ps vs. pdmse_FN 0.116 5 1.649 0.771 Do Not Test pdmse_FN vs. pdmse_FN_circle 0.466 5 5.790 0.002 Yes pdmse_FN vs. ps_fn 0.133 5 1.882 0.674 Do Not Test pdmse_FN vs. pdmse_plasma 0.0353 5 0.500 0.997 Do Not Test pdmse_plasma vs. pdmse_FN_cir 0.431 5 5.447 0.003 Yes pdmse_plasma vs. ps_fn 0.0975 5 1.414 0.854 Do Not Test ps_fn vs. pdmse_FN_circle 0.333 5 4.214 0.035 Yes Comparisons for factor: Culture time 3 Comparison Diff of Means p q P P<0.050 4.000 vs. 14.000 0.552 3 9.315 <0.001 Yes 4.000 vs. 7.000 0.228 3 4.260 0.011 Yes 7.000 vs. 14.000 0.325 3 5.478 0.001 Yes A result of "Do Not Test" occurs for a comparison when no significant difference is found between two means that enclose that comparison. For example, if you had four means sorted in order, and found no difference between means 4 vs. 2, then you would not test 4 vs. 3 and 3 vs. 2, but still test 4 vs. 1 and 3 vs. 1 (4 vs. 3 and 3 vs. 2 are enclosed by 4 vs. 2: 4 3 2 1). Note that not testing the enclosed means is a procedural rule, and a result of Do Not Test should be treated as if there is no significant difference between the means, even though one may appear to exist.

PAGE 211

193 Focal Contact Adhesions per Cell Two Way Analysis of Variance Data source: Data 1 in EC_on_surfaces_stats.SNB Balanced Design Dependent Variable: Contacts per Cell Normality Test: Failed (P = 0.004) Equal Variance Test: Passed (P = 0.094) Source of Variation DF SS MS F P Surface Type 2 5 45992.288 9198.458 140.281 <0.001 Culture Time 2 2 248.911 124.456 1.898 0.160 Surface Type x Culture Time 10 14267.806 1426.781 21.759 <0.001 Residual 54 3540.863 65.572 Total 71 64049.868 902.111 The difference in the mean values among the different levels of Surface Type 2 is greater than would be expected by chance after allowing for effects of differences in Culture Time 2. There is a statistically significant difference (P = <0.001). To isolate which group(s) differ from the others use a multiple comparison procedure. The difference in the mean values among the different levels of Culture Time 2 is not great enough to exclude the possibility that the difference is just due to random sampling variability after allowing for the effects of differences in Surface Type 2. There is not a statistically significant difference (P = 0.160). The effect of different levels of Surface Type 2 depends on what level of Culture Time 2 is present. There is a statistically significant interaction between Surface Type 2 and Culture Time 2. (P = <0.001) Power of performed test with alpha = 0.0500: for Surface Type 2 : 1.000 Power of performed test with alpha = 0.0500: for Culture Time 2 : 0.189 Power of performed test with alpha = 0.0500: for Surface Type x Culture Time : 1.000 Least square means for Surface Type 2 : Group Mean ps 55.000 ps_fn 51.126 pdmse_plasma 7.557 pdmse_FN 15.239 ps_FN_star 82.742 pdmse_FN_circle 37.609 Std Err of LS Mean = 2.338 Least square means for Culture Time 2 : Group Mean 4.000 43.185 7.000 42.506 14.000 38.945 Std Err of LS Mean = 1.653 Least square means for Surface Type x Culture Time :

PAGE 212

194 Group Mean ps x 4.000 43.825 ps x 7.000 64.000 ps x 14.000 57.176 ps_fn x 4.000 90.053 ps_fn x 7.000 28.365 ps_fn x 14.000 34.961 pdmse_plasma x 4.000 8.448 pdmse_plasma x 7.000 10.708 pdmse_plasma x 14.000 3.516 pdmse_FN x 4.000 17.798 pdmse_FN x 7.000 22.561 pdmse_FN x 14.000 5.358 ps_FN_star x 4.000 58.150 ps_FN_star x 7.000 92.375 ps_FN_star x 14.000 97.700 pdmse_FN_circle x 4.000 40.837 pdmse_FN_circle x 7.000 37.028 pdmse_FN_circle x 14.000 34.961 Std Err of LS Mean = 4.049 All Pairwise Multiple Comparison Procedures (Tukey Test): Comparisons for factor: Surface Type 2 Comparison Diff of Means p q P P<0.050 ps_FN_star vs. pdmse_plasma 75.185 6 32.163 <0.001 Yes ps_FN_star vs. pdmse_FN 67.503 6 28.877 <0.001 Yes ps_FN_star vs. pdmse_FN_circle 45.133 6 19.308 <0.001 Yes ps_FN_star vs. ps_fn 31.615 6 13.525 <0.001 Yes ps_FN_star vs. ps 27.742 6 11.868 <0.001 Yes ps vs. pdmse_plasma 47.443 6 20.296 <0.001 Yes ps vs. pdmse_FN 39.761 6 17.010 <0.001 Yes ps vs. pdmse_FN_circle 17.391 6 7.440 <0.001 Yes ps vs. ps_fn 3.874 6 1.657 0.848 No ps_fn vs. pdmse_plasma 43.569 6 18.639 <0.001 Yes ps_fn vs. pdmse_FN 35.887 6 15.352 <0.001 Yes ps_fn vs. pdmse_FN_circle 13.518 6 5.783 0.002 Yes pdmse_FN_cir vs. pdmse_plasma 30.052 6 12.856 <0.001 Yes pdmse_FN_circle vs. pdmse_FN 22.370 6 9.570 <0.001 Yes pdmse_FN vs. pdmse_plasma 7.682 6 3.286 0.203 No Comparisons for factor: Culture Time 2 Comparison Diff of Means p q P P<0.050 4.000 vs. 14.000 4.240 3 2.565 0.175 No 4.000 vs. 7.000 0.679 3 0.411 0.955 Do Not Test 7.000 vs. 14.000 3.561 3 2.154 0.288 Do Not Test Comparisons for factor: Culture Time 2 within ps Comparison Diff of Means p q P P<0.05 7.000 vs. 4.000 20.175 3 4.983 0.003 Yes 7.000 vs. 14.000 6.824 3 1.685 0.463 No 14.000 vs. 4.000 13.351 3 3.298 0.060 No

PAGE 213

195 Comparisons for factor: Culture Time 2 within ps_fn Comparison Diff of Means p q P P<0.05 4.000 vs. 7.000 61.688 3 15.236 <0.001 Yes 4.000 vs. 14.000 55.092 3 13.607 <0.001 Yes 14.000 vs. 7.000 6.596 3 1.629 0.487 No Comparisons for factor: Culture Time 2 within pdmse_plasma Comparison Diff of Means p q P P<0.05 7.000 vs. 14.000 7.192 3 1.776 0.426 No 7.000 vs. 4.000 2.260 3 0.558 0.918 Do Not Test 4.000 vs. 14.000 4.932 3 1.218 0.667 Do Not Test Comparisons for factor: Culture Time 2 within pdmse_FN Comparison Diff of Means p q P P<0.05 7.000 vs. 14.000 17.202 3 4.249 0.011 Yes 7.000 vs. 4.000 4.763 3 1.176 0.685 No 4.000 vs. 14.000 12.439 3 3.072 0.085 No Comparisons for factor: Culture Time 2 within ps_FN_star Comparison Diff of Means p q P P<0.05 14.000 vs. 4.000 39.550 3 9.768 <0.001 Yes 14.000 vs. 7.000 5.325 3 1.315 0.624 No 7.000 vs. 4.000 34.225 3 8.453 <0.001 Yes Comparisons for factor: Culture Time 2 within pdmse_FN_circle Comparison Diff of Means p q P P<0.05 4.000 vs. 14.000 5.876 3 1.451 0.564 No 4.000 vs. 7.000 3.810 3 0.941 0.785 Do Not Test 7.000 vs. 14.000 2.067 3 0.510 0.931 Do Not Test Comparisons for factor: Surface Type 2 within 4 Comparison Diff of Means p q P P<0.05 ps_fn vs. pdmse_plasma 81.605 6 20.155 <0.001 Yes ps_fn vs. pdmse_FN 72.255 6 17.846 <0.001 Yes ps_fn vs. pdmse_FN_circle 49.215 6 12.156 <0.001 Yes ps_fn vs. ps 46.228 6 11.418 <0.001 Yes ps_fn vs. ps_FN_star 31.903 6 7.880 <0.001 Yes ps_FN_star vs. pdmse_plasma 49.702 6 12.276 <0.001 Yes ps_FN_star vs. pdmse_FN 40.352 6 9.966 <0.001 Yes ps_FN_star vs. pdmse_FN_circle 17.313 6 4.276 0.042 Yes ps_FN_star vs. ps 14.325 6 3.538 0.142 No ps vs. pdmse_plasma 35.377 6 8.738 <0.001 Yes ps vs. pdmse_FN 26.027 6 6.428 <0.001 Yes ps vs. pdmse_FN_circle 2.987 6 0.738 0.995 No pdmse_FN_cir vs. pdmse_plasma 32.390 6 8.000 <0.001 Yes pdmse_FN_circle vs. pdmse_FN 23.040 6 5.691 0.002 Yes pdmse_FN vs. pdmse_plasma 9.350 6 2.309 0.581 No Comparisons for factor: Surface Type 2 within 7

PAGE 214

196 Comparison Diff of Means p q P P<0.05 ps_FN_star vs. pdmse_plasma 81.667 6 20.171 <0.001 Yes ps_FN_star vs. pdmse_FN 69.814 6 17.243 <0.001 Yes ps_FN_star vs. ps_fn 64.010 6 15.810 <0.001 Yes ps_FN_star vs. pdmse_FN_circle 55.347 6 13.670 <0.001 Yes ps_FN_star vs. ps 28.375 6 7.008 <0.001 Yes ps vs. pdmse_plasma 53.292 6 13.162 <0.001 Yes ps vs. pdmse_FN 41.439 6 10.235 <0.001 Yes ps vs. ps_fn 35.634 6 8.801 <0.001 Yes ps vs. pdmse_FN_circle 26.972 6 6.662 <0.001 Yes pdmse_FN_cir vs. pdmse_plasma 26.320 6 6.501 <0.001 Yes pdmse_FN_circle vs. pdmse_FN 14.467 6 3.573 0.135 No pdmse_FN_circle vs. ps_fn 8.663 6 2.140 0.658 Do Not Test ps_fn vs. pdmse_plasma 17.657 6 4.361 0.036 Yes ps_fn vs. pdmse_FN 5.804 6 1.434 0.911 Do Not Test pdmse_FN vs. pdmse_plasma 11.853 6 2.927 0.318 No Comparisons for factor: Surface Type 2 within 14 Comparison Diff of Means p q P P<0.05 ps_FN_star vs. pdmse_plasma 94.184 6 23.262 <0.001 Yes ps_FN_star vs. pdmse_FN 92.342 6 22.807 <0.001 Yes ps_FN_star vs. pdmse_FN_circle 62.739 6 15.496 <0.001 Yes ps_FN_star vs. ps_fn 62.739 6 15.496 <0.001 Yes ps_FN_star vs. ps 40.524 6 10.009 <0.001 Yes ps vs. pdmse_plasma 53.660 6 13.253 <0.001 Yes ps vs. pdmse_FN 51.818 6 12.798 <0.001 Yes ps vs. pdmse_FN_circle 22.215 6 5.487 0.004 Yes ps vs. ps_fn 22.215 6 5.487 0.004 Yes ps_fn vs. pdmse_plasma 31.445 6 7.767 <0.001 Yes ps_fn vs. pdmse_FN 29.603 6 7.311 <0.001 Yes ps_fn vs. pdmse_FN_circle 0.000 6 0.000 1.000 No pdmse_FN_cir vs. pdmse_plasma 31.445 6 7.767 <0.001 Yes pdmse_FN_circle vs. pdmse_FN 29.603 6 7.311 <0.001 Yes pdmse_FN vs. pdmse_plasma 1.843 6 0.455 1.000 No A result of "Do Not Test" occurs for a comparison when no significant difference is found between two means that enclose that comparison. For example, if you had four means sorted in order, and found no difference between means 4 vs. 2, then you would not test 4 vs. 3 and 3 vs. 2, but still test 4 vs. 1 and 3 vs. 1 (4 vs. 3 and 3 vs. 2 are enclosed by 4 vs. 2: 4 3 2 1). Note that not testing the enclosed means is a procedural rule, and a result of Do Not Test should be treated as if there is no significant difference between the means, even though one may appear to exist.

PAGE 215

LIST OF REFERENCES 1. Feinberg, A.W., Quantifying Active Endothelial Cell Receptor-Ligand Binding as a Function of Engineered Surfaces, in Biomedical Engineering. 2002, University of Florida: Gainesville. p. 154. 2. Levy, B. and A. Tedgui, eds. Biology of the Arterial Wall. 1999, Kluwer Academic Press: Boston. 3. Rhoades, R. and R. Pflanzer, Human Physiology. Fourth ed. 2003, USA: Brooks/Cole. 999. 4. Teebken, O.E. and A. Haverich, Tissue engineering of small diameter vascular grafts. European Journal of Vascular and Endovascular Surgery, 2002. 23(6): p. 475-485. 5. Hoerstrup, S.P., et al., Tissue engineering of small caliber vascular grafts. European Journal of Cardio-Thoracic Surgery, 2001. 20(1): p. 164-169. 6. Rabkin, E. and F.J. Schoen, Cardiovascular tissue engineering. Cardiovascular Pathology, 2002. 11(6): p. 305-317. 7. Niklason, L.E., et al., Morphologic and mechanical characteristics of engineered bovine arteries. Journal of Vascular Surgery, 2001. 33(3): p. 628-638. 8. Niklason, L.E., et al., Functional arteries grown in vitro. Science, 1999. 284(5413): p. 489-493. 9. L'Heureux, N., et al., A completely biological tissue-engineered human blood vessel. Faseb Journal, 1998. 12(1): p. 47-56. 10. Ballyk, P.D., et al., Compliance mismatch may promote graft-artery intimal hyperplasia by altering suture-line stresses. Journal of Biomechanics, 1998. 31(3): p. 229-237. 11. Ulrich, M., et al., In vivo analysis of dynamic tensile stresses at arterial endto-end anastomoses. Influence of suture-line and graft on anastomotic biomechanics. European Journal of Vascular and Endovascular Surgery, 1999. 18(6): p. 515-522. 12. Feinberg, A.W., et al. Engineering Micrometer and Nanometer Scale Features in Polydimethylsiloxane Elastomers for Controlled Cell Function. in MRS Fall 2001. 2001. Boston, MA: MRS Publications. 197

PAGE 216

198 13. Gibson, A.L., et al. Characterization of Chemically and Topographically Modified Siloxane Elastomer for Controlled Cell Growth. in MRS Fall 2001. 2001. Boston, MA: MRS Publications. 14. Jeon, N.L., et al., Neutrophil chemotaxis in linear and complex gradients of interleukin-8 formed in a microfabricated device. Nature Biotechnology, 2002. 20(8): p. 826-830. 15. SHUICHI TAKAYAMA, J.C.M., EMANUELE OSTUNI, MICHAEL N. LIANG, PAUL J. A. KENIS, RUSTEM F. ISMAGILOV, AND GEORGE M. WHITESIDES, Patterning cells and their environments using multiple laminar fluid flows in capillary networks. Proc. Natl. Acad. Sci., 1999. 96: p. 5545. 16. Janelle R. Anderson, D.T.C., Rebecca J. Jackman, Oksana Cherniavskaya, J. Cooper McDonald, Hongkai Wu, Sue H. Whitesides, and George M. Whitesides, Fabrication of Topologically Complex Three-Dimensional Microfluidic Systems in PDMS by Rapid Prototyping. Anal. Chem., 2000. 72: p. 3158-3164. 17. Daniel T. Chiu, N.L.J., Sui Huang, Ravi S. Kane, Christopher J. Wargo, Insung S. Choi, Donald E. Ingber, and George M. Whitesides, Patterned deposition of cells and proteins onto surfaces by using three-dimensional microfluidic systems. PNAS, 2000. 97(6): p. 2408. 18. Vaidya, R., et al., Computer-controlled laser ablation: A convenient and versatile tool for micropatterning biofunctional synthetic surfaces for applications in biosensing and tissue engineering. Biotechnology Progress, 1998. 14(3): p. 371-377. 19. Kappelt, M. and D. Bimberg, Wet chemical etching of high quality V-grooves with {111} a sidewalls on (001) InP. Journal of the Electrochemical Society, 1996. 143(10): p. 3271-3273. 20. Thomas, C.H., et al., The role of vitronectin in the attachment and spatial distribution of bone-derived cells on materials with patterned surface chemistry. Journal of Biomedical Materials Research, 1997. 37(1): p. 81-93. 21. Tien, J., A. Terfort, and G.M. Whitesides, Microfabrication through electrostatic self-assembly. Langmuir, 1997. 13(20): p. 5349-5355. 22. Alaerts, J.A., et al., Surface characterization of poly(methyl methacrylate) microgrooved for contact guidance of mammalian cells. Biomaterials, 2001. 22(12): p. 1635-1642. 23. Curtis, A. and C. Wilkinson, Review: Topographical Control of Cells. Biomaterials, 1997. 18: p. 1573-1583.

PAGE 217

199 24. Rahul Singhvi, A.K., Gabriel P. Lopez, Gregory N. Stephanopoulos, Daniel I. C. Wang, George M. Whitesides, Donald E. Ingber, Engineering Cell Shape and Function. SCIENCE, 1994. 264: p. 696-698. 25. Michel, H.S.a.B., Siloxane Polymers for High-Resolution, High-Accuracy Soft Lithography. Macromolecules, 2000. 33(8): p. 3042-3049. 26. Wu, H., T.W. Odom, and G.M. Whitesides, Generation of chrome masks with micrometer-scale features using microlens lithography. Advanced Materials, 2002. 14(17): p. 1213-+. 27. Goldmann, T. and J.S. Gonzalez, DNA-printing: utilization of a standard inkjet printer for the transfer of nucleic acids to solid supports. Journal of Biochemical and Biophysical Methods, 2000. 42(3): p. 105-110. 28. Allain, L.R., et al., Microarray sampling-platform fabrication using bubble-jet technology for a biochip system. Fresenius Journal of Analytical Chemistry, 2001. 371(2): p. 146-150. 29. Calvert, P., Inkjet printing for materials and devices. Chemistry of Materials, 2001. 13(10): p. 3299-3305. 30. den Braber, E.T., et al., Effect of parallel surface microgrooves and surface energy on cell growth. J Biomed Mater Res, 1995. 29(4): p. 511-8. 31. den Braber, E.T., et al., Quantitative analysis of fibroblast morphology on microgrooved surfaces with various groove and ridge dimensions. Biomaterials, 1996. 17(21): p. 2037-44. 32. Mudera, V.C., et al., Molecular responses of human dermal fibroblasts to dual cues: Contact guidance and mechanical load. Cell Motility and the Cytoskeleton, 2000. 45(1): p. 1-9. 33. van Kooten, T.G. and A.F. von Recum, Cell adhesion to textured silicone surfaces: The influence of time of adhesion and texture on focal contact and fibronectin fibril formation. Tissue Engineering, 1999. 5(3): p. 223-240. 34. Walboomers, X.F., et al., Contact guidance of rat fibroblasts on various implant materials. Journal of Biomedical Materials Research, 1999. 47(2): p. 204-212. 35. Matsuzaka, K., et al., The attachment and growth behavior of osteoblast-like cells on microtextured surfaces. Biomaterials, 2003. 24(16): p. 2711-2719. 36. Clark, P., et al., Topographical Control of Cell Behavior .1. Simple Step Cues. Development, 1987. 99(3): p. 439-448. 37. Clark, P., et al., Topographical Control of Cell Behavior .2. Multiple Grooved Substrata. Development, 1990. 108(4): p. 635-644.

PAGE 218

200 38. Turner, A.M.P., et al., Attachment of astroglial cells to microfabricated pillar arrays of different geometries. Journal of Biomedical Materials Research, 2000. 51(3): p. 430-441. 39. Feinberg, A.W., et al., Investigating the Energetics of Bioadhesion on Microengineered Siloxane Elastomers: Characterizing the Topography, Mechanical Properties, and Surface Energy and Their Effect on Cell Contact Guidance, in Synthesis and Properties of Silicones and Silicone-Modified Materials, Clarson, et al., Editors. 2003, ACS. p. 196-211. 40. Carter, S.B., Haptotaxis and Mechanism of Cell Motility. Nature, 1967. 213(5073): p. 256-&. 41. Britland, S., et al., Micropatterned Substratum Adhesiveness a Model for Morphogenetic Cues Controlling Cell Behavior. Experimental Cell Research, 1992. 198(1): p. 124-129. 42. McFarland, C.D., et al., Protein adsorption and cell attachment to patterned surfaces. Journal of Biomedical Materials Research, 2000. 49(2): p. 200-210. 43. Chandy, T., et al., Use of plasma glow for surface-engineering biomolecules to enhance bloodcompatibility of Dacron and PTFE vascular prosthesis. Biomaterials, 2000. 21(7): p. 699-712. 44. Patterson, R.B., A. Messier, and R.F. Valentini, Effects of Radiofrequency Glow Discharge and Oligopeptides on the Attachment of Human Endothelial Cells to Polyurethane. ASAIO Journal, 1995. 41(3): p. M625-M629. 45. Chen, C.S., et al., Micropatterned surfaces for control of cell shape, position, and function. Biotechnology Progress, 1998. 14(3): p. 356-363. 46. Chen, C.S., et al., Cell shape provides global control of focal adhesion assembly. Biochemical and Biophysical Research Communications, 2003. 307(2): p. 355-361. 47. Riviere, J.C. and S. Myhra, eds. Handbook of Surface and Interface Analysis: Methods for Problem-Solving. 4th ed. 1998, Marcel Dekker: New York. 48. Lee, G.U., L.A. Chrisey, and R.J. Colton, Direct Measurement of the Forces between Complementary Strands of DNA. Science, 1994. 266(5186): p. 771-773. 49. Dammer, U., et al., Binding Strength between Cell-Adhesion Proteoglycans Measured by Atomic-Force Microscopy. Science, 1995. 267(5201): p. 1173-1175. 50. Furuike, S., T. Ito, and M. Yamazaki, Mechanical unfolding of single filamin A (ABP-280) molecules detected by atomic force microscopy. Febs Letters, 2001. 498(1): p. 72-75.

PAGE 219

201 51. Valle, F., G. Dietler, and P. Londei, Single-molecule imaging by atomic force microscopy of the native chaperonin complex of the thermophilic archaeon Sulfolobus solfataricus. Biochemical and Biophysical Research Communications, 2001. 288(1): p. 258-262. 52. Tromas, C. and R. Garcia, Interaction forces with carbohydrates measured by atomic force microscopy. Host-Guest Chemistry, 2002. 218: p. 115-132. 53. Hinterdorfer, P., et al., Surface attachment of ligands and receptors for molecular recognition force microscopy. Colloids and Surfaces B-Biointerfaces, 2002. 23(2-3): p. 115-123. 54. Camesano, T.A. and K.J. Wilkinson, Single molecule study of xanthan conformation using atomic force microscopy. Biomacromolecules, 2001. 2(4): p. 1184-1191. 55. Florin, E.L., V.T. Moy, and H.E. Gaub, Adhesion Forces between Individual Ligand-Receptor Pairs. Science, 1994. 264(5157): p. 415-417. 56. Harada, Y., M. Kuroda, and A. Ishida, Specific and quantized antigen-antibody interaction measured by atomic force microscopy. Langmuir, 2000. 16(2): p. 708-715. 57. Benoit, M., et al., Discrete interactions in cell adhesion measured by singlemolecule force spectroscopy. Nature Cell Biology, 2000. 2(6): p. 313-317. 58. Clausen-Schaumann, H., et al., Force spectroscopy with single bio-molecules. Current Opinion in Chemical Biology, 2000. 4(5): p. 524-530. 59. Willemsen, O.H., et al., Simultaneous height and adhesion imaging of antibody-antigen interactions by atomic force microscopy. Biophysical Journal, 1998. 75(5): p. 2220-2228. 60. Fritz, J., et al., Force-mediated kinetics of single P-selectinyligand complexes observed by atomic force microscopy. Proc. Natl. Acad. Sci., 1998. 95: p. 12283-12288. 61. Zhao, L.C., Cell Adhesion: Characterization of Adhesive Forces and Effect of Topography, in Materials Science and Engineering. 2000, University of Florida: Gainesville. p. 88. 62. Marieb, E.N., Human Anatomy and Physiology. 4th ed. 1998, Menlo Park: Benjamin/Cummings Science Publishing. 1192. 63. Sato, M., et al., Local mechanical properties measured by atomic force microscopy for cultured bovine endothelial cells exposed to shear stress. Journal of Biomechanics, 2000. 33: p. 127-135.

PAGE 220

202 64. Miyazaki, H. and K. Hayashi, Atomic force microscopic measurement of the mechanical properties of intact endothelial cells in fresh arteries. Medical & Biological Engineering & Computing, 1999. 37(4): p. 530-536. 65. Mathur, A.B., et al., Endothelial, cardiac muscle and skeletal muscle exhibit different viscous and elastic properties as determined by atomic force microscopy. Journal of Biomechanics, 2001. 34(12): p. 1545-1553. 66. Ebenstein, D.M., Biomechanical Characterization of Atherosclerotic Plaques: A Combined Nanoindentation and FTIR Approach, in Bioengineering. 2002, University of California, Berkeley: Berkeley. p. 399. 67. Mathur, A.B., G.A. Truskey, and W.M. Reichert, Atomic force and total internal reflection fluorescence microscopy for the study of force transmission in endothelial cells. Biophysical Journal, 2000. 78(4): p. 1725-1735. 68. Sato, H., et al., Kinetic study on the elastic change of vascular endothelial cells on collagen matrices by atomic force microscopy. Colloids and Surfaces B-Biointerfaces, 2004. 34(2): p. 141-146. 69. Sato, H., et al., Estimation for the elasticity of vascular endothelial cells on the basis of atomic force microscopy and Young's modulus of gelatin gels. Polymer Bulletin, 2001. 47(3-4): p. 375-381. 70. Ohashi, T., et al., Experimental and numerical analyses of local mechanical properties measured by atomic force microscopy for sheared endothelial cells. Bio-Medical Materials and Engineering, 2002. 12(3): p. 319-327. 71. Corning, D., Silastic T-2 Translucent Base and Silastic T-2/T-2 Hgh Durometer Curing Agent, in Ref. no. 80-3143-01. 2002. p. 4. 72. Hertz, H.J., Reine Angew. Mathematik, 1882. 92: p. 156. 73. Sneddon, I.N., The Relaxation Between Load and Penetration in the Axisymmetric Boussinesq Problem for a Punch of Arbitrary Profile. Int. J. Engng. Sci., 1965. 3: p. 47-57. 74. Hillborg, H., M. Sandelin, and U.W. Gedde, Hydrophobic recovery of polydimethylsiloxane after exposure to partial discharges as a function of crosslink density. Polymer, 2001. 42(17): p. 7349-7362. 75. Hillborg, H. and U.W. Gedde, Hydrophobicity recovery of polydimethylsiloxane after exposure to corona discharges. Polymer, 1998. 39(10): p. 1991-1998. 76. Wilkerson, W.R., et al., Bioadhesion studies on microtextured siloxane elastomers. Abstracts of Papers of the American Chemical Society, 2001. 221: p. 253-POLY.

PAGE 221

203 77. Flemming, R.G., et al., Effects of synthetic microand nano-structured surfaces on cell behavior. Biomaterials, 1999. 20: p. 573-588. 78. Fan, Y.W., et al., Adhesion of neural cells on silicon wafer with nano-topographic surface. Applied Surface Science, 2002. 187(3-4): p. 313-318. 79. Gedde, U.W., Polymer Physics. 1999, Boston: Kluwer Academic Press. 80. Johnson, B., et al. Mechanical Properties of a pH Sensitive Hydrogel. in Society for Experimental Mechanics. 2002. Milwaukee, WI. 81. Tortonese, M., Cantilevers and tips for atomic force microscopy. Ieee Engineering in Medicine and Biology Magazine, 1997. 16(2): p. 28-33. 82. Sadler, J.E., J.W.M. Chon, and P. Mulvaney, Calibration of rectangular atomic force microscope cantilevers. Review of Scientific Instruments, 1999. 70(10): p. 3967-3969. 83. Callow, M.E., et al., Microtopographic cues for settlement of zoospores of the green fouling alga Enteromorpha. Biofouling, 2002. 18(3): p. 237-245. 84. Kane, R.S., et al., Patterning proteins and cells using soft lithography. Biomaterials, 1999. 20(23-24): p. 2363-2376. 85. Delamarche, E., et al., Microcontact printing using poly(dimethylsiloxane) stamps hydrophilized by poly(ethylene oxide) silanes. Langmuir, 2003. 19(21): p. 8749-8758. 86. Kam, L. and S.G. Boxer, Cell adhesion to protein-micropatterned-supported lipid bilayer membranes. Journal of Biomedical Materials Research, 2001. 55(4): p. 487-495. 87. Anselme, K., et al., Effect of grooved titanium substratum on human osteoblastic cell growth. Journal of Biomedical Materials Research, 2002. 60(4): p. 529-540. 88. Britland, S., et al., Synergistic and hierarchical adhesive and topographic guidance of BHK cells. Experimental Cell Research, 1996. 228(2): p. 313-325. 89. Chou, L.S., et al., Substratum Surface-Topography Alters Cell-Shape and Regulates Fibronectin Messenger-Rna Level, Messenger-Rna Stability, Secretion and Assembly in Human Fibroblasts. Journal of Cell Science, 1995. 108: p. 1563-1573. 90. Craighead, H.G., C.D. James, and A.M.P. Turner, Chemical and topographical patterning for directed cell attachment. Current Opinion in Solid State & Materials Science, 2001. 5(2-3): p. 177-184.

PAGE 222

204 91. Matsuzaka, K., et al., Effect of microgrooved poly-l-lactic (PLA) surfaces on proliferation, cytoskeletal organization, and mineralized matrix formation of rat bone marrow cells. Clinical Oral Implants Research, 2000. 11(4): p. 325-333. 92. Walboomers, X.F., et al., Growth behavior of fibroblasts on microgrooved polystyrene. Biomaterials, 1998. 19: p. 1861-1868. 93. Walboomers, X.F., et al., Attachment of fibroblasts on smooth and microgrooved polystyrene. Journal of Biomedical Materials Research, 1999. 46(2): p. 212-220. 94. Walboomers, X.F. and J.A. Jansen, Microgrooved silicone subcutaneous implants in guinea pigs. Biomaterials, 2000. 21(6): p. 629-636. 95. Wang, J.H.-C., et al., Alignment and proliferation of MC3T3-E1 osteoblasts in microgrooved silicone substrata subjected to cyclic stretching. Journal of Biomechanics, 2000. 33: p. 729-735. 96. Israelachvili, J.N., Adhesion forces between surfaces in liquids and condensable vapours. Surface Science Reports, 1992. 14(3): p. 109-159. 97. Leckband, D., Measuring the forces that control protein interactions. Annual Review of Biophysics and Biomolecular Structure, 2000. 29: p. 1-26. 98. Clapp, A.R. and R.B. Dickinson, Direct measurement of static and dynamic forces between a colloidal particle and a flat surface using a single-beam gradient optical trap and evanescent wave light scattering. Langmuir, 2001. 17(7): p. 2182-2191. 99. Evans, E., K. Ritchie, and R. Merkel, Sensitive force technique to probe molecular adhesion and structural linkages at biological interfaces. Biophys. J., 1995. 68(6): p. 2580-2587. 100. Tsilimbaris, M.K., et al., The Use of Atomic Force Microscopy for the Observation of Corneal Epithelium Surface. Investigative Opthamology and Visual Science, 2000. 41(3): p. 680-686. 101. Camesano, T.A., M.J. Natan, and B.E. Logan, Observation of Changes in Bacterial Cell Morphology Using Tapping Mode Atomic Force Microscopy. Langmuir, 2001. 16(10): p. 4563-4572. 102. Lehenkari, P.P. and M.A. Horton, Single integrin molecule adhesion forces in intact cells measured by atomic force microscopy. Biochemical and Biophysical Research Communications, 1999. 259(3): p. 645-650. 103. Hyonchol, K., et al., Quantification of fibronectin and cell surface interactions by AFM. Colloids and Surfaces B-Biointerfaces, 2002. 25(1): p. 33-43.

PAGE 223

205 104. Vestweber, D., ed. The selectins : initiators of leukocyte endothelial adhesion. Advances in vascular biology. Vol. 3. 1997, Harwood Academic Publishers: Australia. 225. 105. Rodgers, S.D., R.T. Camphausen, and D.A. Hammer, Sialyl Lewis(x)-Mediated, PSGL-1-independent rolling adhesion on P-selectin. Biophysical Journal, 2000. 79(2): p. 694-706. 106. Radmacher, M., Measuring the elastic properties of biological samples with the AFM. Ieee Engineering in Medicine and Biology Magazine, 1997. 16(2): p. 47-57. 107. Willemsen, O.H., M.M.E. Snel, and B.G.d.G. Kees O. van der Werf, Jan Greve, Peter Hinterdorfer, Hermann J. Gruber, Hansgeorg Schindler, Yvette van Kooyk, and Carl G. Figdor, Simultaneous Height and Adhesion Imaging of Antibody-Antigen Interactions by Atomic Force Microscopy. Biophysical Journal, 1998. 75: p. 2220-2228. 108. Tortonese, M. and M. Kirk. Characterization of application specific probes for SPMs. in SPIE. 1997. 109. Patel, J.M., et al., Effect of Nitrogen-Dioxide on Surface-Membrane Fluidity and Insulin-Receptor Binding of Pulmonary Endothelial-Cells. Biochemical Pharmacology, 1988. 37(8): p. 1497-1507. 110. Moy, V.T., E.L. Florin, and H.E. Gaub, Intermolecular Forces and Energies between Ligands and Receptors. Science, 1994. 266(5183): p. 257-259. 111. Wong, J., A. Chilkoti, and V.T. Moy, Direct force measurements of the streptavidin-biotin interaction. Biomolecular Engineering, 1999. 16(1-4): p. 45-55. 112. Best, R.B., et al., Force mode atomic force microscopy as a tool for protein folding studies. Analytica Chimica Acta, 2003. 479(1): p. 87-105. 113. Marchand, P. and L. Marmet, Binomial Smoothing Filter a Way to Avoid Some Pitfalls of Least-Squares Polynomial Smoothing. Review of Scientific Instruments, 1983. 54(8): p. 1034-1041. 114. Smith, M.J., E.L. Berg, and M.B. Lawrence, A direct comparison of selectin-mediated transient, adhesive events using high temporal resolution. Biophysical Journal, 1999. 77(6): p. 3371-3383. 115. Rief, M., et al., Reversible unfolding of individual titin immunoglobulin domains by AFM. Science, 1997. 276(5315): p. 1109-1112.

PAGE 224

206 116. Ushiyama, S., et al., Structural and Functional-Characterization of Monomeric Soluble P-Selectin and Comparison with Membrane P-Selectin. Journal of Biological Chemistry, 1993. 268(20): p. 15229-15237. 117. Snapp, K.R., C.E. Heitzig, and G.S. Kansas, Attachment of the PSGL-1 cytoplasmic domain to the actin cytoskeleton is essential for leukocyte rolling on P-selectin. Blood, 2002. 99(12): p. 4494-4502. 118. Setiadi, H., et al., Interactions of the Cytoplasmic Domain of P-Selectin with Clathrin-coated Pits Enhance Leukocyte Adhesion under Flow. J. Cell Biol., 1998. 142(3): p. 859-871. 119. Merkel, R., et al., Energy landscapes of receptor-ligand bonds explored with dynamic force spectroscopy. Nature, 1999. 397(6714): p. 50-53. 120. Ley, K. and P. Gaehtgens, Endothelial, Not Hemodynamic, Differences Are Responsible for Preferential Leukocyte Rolling in Rat Mesenteric Venules. Circulation Research, 1991. 69(4): p. 1034-1041. 121. Peters, T., Serum-Albumin. Advances in Protein Chemistry, 1985. 37: p. 161-245. 122. Carter, D.C. and J.X. Ho, Structure of Serum Albumin. Advances in Protein Chemistry, 1994. 45: p. 153-203. 123. Woywodt, A., et al., Circulating endothelial cells: life, death, detachment and repair of the endothelial cell layer. Nephrology Dialysis Transplantation, 2002. 17(10): p. 1728-1730. 124. Rafii, S., Circulating endothelial precursors: mystery, reality, and promise. Journal of Clinical Investigation, 2000. 105(1): p. 17-19. 125. Asahara, T., et al., Isolation of putative progenitor endothelial cells for angiogenesis. Science, 1997. 275(5302): p. 964-967. 126. Guleserian, K.J., et al., Human umbilical cord blood derived endothelial progenitor cells (HUCB-EPCs): A novel cell source for cardiovascular tissue engineering. Circulation, 2001. 104(17): p. 3585. 127. Segal, M.S., A. Bihorac, and M. Koc, Circulating endothelial cells: tea leaves for renal disease. American Journal of Physiology-Renal Physiology, 2002. 283(1): p. F11-F19. 128. Mooney, D., Nimble progenitors rescue vascular grafts. Nature Medicine, 2001. 7(9): p. 996-997. 129. Kaushal, S., et al., Functional small-diameter neovessels created using endothelial progenitor cells expanded ex vivo. Nature Medicine, 2001. 7(9): p. 1035-1040.

PAGE 225

207 130. Vale, T.K.a.R., ed. Guidebook to the extracellular matrix and adhesion proteins. 1993, Oxford University Press: Oxford. 176. 131. Wilkerson, W.R., Contribution of Modulus to the Contact Guidance of Endothelial Cells on Microtextured Siloxane Elastomers, in Biomedical Engineering. 2001, University of Florida: Gainesville. p. 111. 132. Hormbrey, E., et al., A critical review of vascular endothelial growth factor (VEGF) analysis in peripheral blood: Is the current literature meaningful? Clinical & Experimental Metastasis, 2002. 19(8): p. 651-663. 133. Lemstrom, K.B., et al., Vascular endothelial growth factor enhances cardiac allograft arteriosclerosis. Circulation, 2002. 105(21): p. 2524-2530. 134. Domke, J. and M. Radmacher, Measuring the Elastic Properties of Thin Polymer Films with the Atomic Force Microscope. Langmuir, 1998. 14(12): p. 3320-3325.

PAGE 226

BIOGRAPHICAL SKETCH I was born on June 2, 1977, to my loving parents Linda R. Feinberg and Edward B. Feinberg in Trenton, NJ. I grew up in the town of Lawrenceville, NJ, and had a great childhood playing with friends, video games and transformers, a true child of the 80s. By the time I was 10 I already had a pile of Legos, erector sets, RadioShack science kits and countless models piled in the corner. I loved to build and tinker, but I always became bored of my ‘typical’ toys, never having enough power or capability to satisfy my curiosity. Radio controlled cars became my first passion, much to the financial dismay of my parents, and I built and raced RC cars for the next 2 years. Well, as I grew into a young man my interests changed with me and I moved from the world of RC cars to bicycles. I began working in a local bicycle shop the day after I got my working papers at 14. I learned to be a bicycle mechanic that can take apart, adjust, and fix anything on a bike. Of course, most of my time was spent tinkering with my own bikes, and soon I began racing them. What started out as an occasional diversion soon turned into a passion and road racing became my life. I joined the United States Cycling Federation as a junior and began racing and training 4 to 5 hours a day, 6 days a week. This may seem like obsession, and to a degree it was, but I had success when I raced and this drove me on. I raced until I was 17, and in that time I was ranked third in my state and competed in the United States Junior Championships, and this experience showed me the rewards of hard work and determination. 208

PAGE 227

209 In high school I still had plenty of time to pursue other interests in academics, sports, drama and relationships. I did well in school and never had to work very hard even though I took the honors and advanced placement classes. I even had the opportunity to take physics at nearby Princeton University while still in high school. Since bicycling was an outdoor sport, the winter gave me time to participate on the varsity swim team, and though I was not a standout, I was consistent and always helped add points to the team score. In the winter I also had time to participate in drama, but rather than act I helped run the lights and sound for the student productions. I did this with my two best friends in high school for my junior and senior years and have more good memories than I can remember. Luckily I still had a personal life, discovering girls and all the excitement and confusion that did, and still does, go along with them. But like all good things, high school came to an end and I moved from NJ to the small mountains of Ithaca, NY, to attend college at Cornell University. Cornell was a unique experience, and the education I received in materials science and engineering was second to none. I am grateful for having the opportunity to learn at such an institution. I also made great friends both from my classes and my fraternity, Phi Gamma Delta or Fiji as we are commonly known. Unfortunately, the unrelated deaths of two of my friends, a concussion my senior year that temporarily damaged my short term memory and the total lack of sun for 8 months out of the year clouded my experience. I don’t want to sound bitter, but upon graduation from Cornell I was definitely ready to start the next phase of my life. Lessons, though tough learned, such as the true value of friendship, the importance of one’s health, the fragility of life and the strength to pursue the things that make you happy are truths that will not be forgotten.

PAGE 228

210 From Cornell in upstate NY I moved to sunny Gainesville, FL, to start graduate school in biomedical engineering. The University of Florida was a great fit for me and I found a great research program, good friends and beautiful weather. My education at UF has been first rate and the experiences I have had in the lab will prepare me well for my future in biomedical engineering. I look forward to applying my knowledge to tissue engineering and bio-artificial organs to improve the quality of and save lives. Unfortunately, I became afflicted with vertigo and dizziness towards the end of July 2001 that stretched all the way into December. My progress stopped, and my life changed as the doctors tried to tell me what was wrong and how to stop it. I learned that many medical questions have no answers and the cause of my ailment was ruled viral and untreatable. Luckily, the symptoms began to fade and I have slowly adapted. I now make it a priority to balance the stress in my life to help maintain my health and take the time to enjoy the simple things that make life worth living, from a good cup of coffee to talking with friends and family. I received my Masters of Science in biomedical engineering from UF in August of 2001. After considering a transfer to Duke University to complete my PhD studies, I decided to stay at UF and work on tissue engineering endothelial cells monolayers thanks to the combined financial support of Dr. Anthony Brennan and Dr. Winfred Phillips. With their guidance, I rapidly completed my PhD research and dissertation as I graduate with my doctorate in May 2004. Finally, no bio of my time at UF would be complete without inclusion of my fianc Ayelet Tal. She has helped my through sickness and health and been a constant source of encouragement.