The Pennsylvania State University The Graduate School Department of Engineering Science and Mechanics

BIOCHIPS EMPLOYING MICROELECTRODES AND MICROTRANSDUCERS FOR CHARACTERIZATION AND PERMEABILIZATION OF CELL MEMBRANE AT THE WHOLE–CELL AND CELLULAR LEVEL

A Dissertation in Engineering Science and Mechanics

by Myo Min Thein

© 2010 Myo Min Thein

Submitted in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

December 2010 The dissertation of Myo Min Thein was reviewed and approved* by the following:

Jian Xu Associate Professor of Engineering Science and Mechanics and Adjunct Professor of Electrical Engineering Thesis Advisor and Chair of Committee

Bernhard R. Tittmann Schell Professor of Engineering Science and Mechanics

Yinong Yang Associate Professor of Plant Pathology

Sulin Zhang Assistant Professor of Engineering Science and Mechanics

Judith A. Todd P. B. Breneman Department Head Chair Head of the Department of Engineering Science and Mechanics

* Signatures are on file in the Graduate School.

ii Abstract

In this study, two types of biochips, Integrated Microelectrode Array (IMA) chip and

Ultrasonic Micro–Transducer Array (UMTA) biochip, were designed and micro– fabricated. The IMA chip employs cell–size microelectrode array architecture in order to electrically characterize cellular components, especially the properties of cell membrane, at the single–cell level. The UMTA biochip consists of arrays of high–aspect–ratio piezoelectric micro–transducers that are employed to efficiently permeabilize the cell membrane using ultrasound (i.e., sonoporation) at cellular level with high spatial specificity and resolution. The device fabrication processes, which are tailored towards technology transfer to biochip industries, include laser lithography, proximity and contact photolithography, Physical Vapor Deposition (PVD), high–frequency pulsed electroplating, Chemical Vapor Deposition (CVD) and Plasma–Enhanced Chemical

Vapor Deposition (PECVD), furnace annealing and surface passivation, Reactive Ion

Etching (RIE) and Inductively–Coupled Plasma (ICP) etching, chemical and mechanical planarization/polishing (CMP), and bottom–up self–assembled monolayer (SAM)– based surface engineering approaches.

In demonstrating and testing the performance of these devices, firstly, the electrical characterizations of individual cells (mouse fibroblast (NIH3T3) cell line) were carried out using the IMA biochip integrated with phase–sensitive lock–in detection technique.

Bottom–up SAM–based surface engineering approaches were employed to immobilize individual NIH3T3 cells on the working microelectrodes of IMA biochip for the electrical characterizations. The application–specific models such as equivalent electrical circuits of cell–electrode hetero–structures were also developed and cellular properties such as cell membrane capacity, cell membrane resistivity, and average cell–to–substrate separation distance were estimated by numerically fitting the experimental data to the

iii developed models. The influence of cell adhesion and spreading, mediated by surface– bound cell adhesion molecules (e.g., fibronectin) or short peptides (e.g., lysine– arginine–glycine–aspartic acid), on the impedance characterizations and sensitivity of cell–based sensors were investigated. In addition, the significances and advantages of

single–cell–level impedance measurements in terms of sensitivity in drug screening and

toxicity assays were investigated by monitoring real–time cellular changes in individual

human glioblastoma (U87MG) cells, seeded on single–cell microelectrodes, upon

exposure to an ion channel inhibitor, chlorotoxin (CTX). Results show that

microelectrode array architecture combined with proper bioinstrumentations and

theoretical models are capable of estimating cellular properties, especially the properties

of cell membrane, at the single–cell level as well as monitoring cellular responses to

drugs or toxins with high sensitivity, rendering a huge potential of IMA chips in high–

throughput cell–based sensing and assays at the single–cell level.

Secondly, experiments were carried out to demonstrate the ultrasound–induced site–specific cell membrane permeabilization (or site–specific sonoporation) at the cellular level using the UMTA biochip. The site–specific sonoporation was achieved by applying ultrasonic waves, generated by driving the micro–transducer arrays with high– frequency RF signal generator, onto the targeted cancer cells (human melanoma

(LU1205) cell line) within the cell monolayer. Water–soluble Quantum Dots (QD)

(carboxylic–acid–surface–derivatized CdSe/ZnS core/shell QDs) were employed as optical probes in order to track the diffusion and transport of nanoparticles across the cell membrane after sonoporation and analyses were carried out by fluorescence microscopy and image processing algorithms. Results show that, above the threshold, moderate ultrasound radiation pressures generated by the micro–transducer arrays can transiently permeabilize the cell membrane at the cellular level with high spatial specificity and resolution. Furthermore, by comparing with the dynamics of

iv endocytosis–driven QD intake obtained via flow cytometry and time–lapse fluorescence microscopy, the resulting sonoporation showed significant enhancement in the transport and uptake of water–soluble QDs into the melanoma cells (by an QD–uptake enhancement factor of >80) due to passive diffusion of QDs through membrane wounds.

Models representing ultrasound–inflicted cell membrane wounds and mass transport of nanoparticles through wound openings were developed. Model–based analyses, supported with experimental findings, show that low and moderate acoustic radiation pressures generated from the micro–transducer stress the plasma membrane of LU1205 cells and, above the threshold radiation pressure (~0.12MPa for LU1205 cells), large transient wounds are created in the membrane and are actively resealed. The effective size of membrane wounds as well as the uptake of water–soluble QDs were found to be linearly dependent on the ultrasonic pressure applied. These results from site–specific sonoporation show that not only does the UMTA biochips have huge potentials in transfection and gene therapy assays at the cellular level, they also render a promising

future as sub–systems in enhancing the delivery efficiency of drug screening and characterization platforms.

v TABLE OF CONTENT

Abstract...... iii

List of Figures ...... xv

List of Tables ...... xxviii

Acknowledgement ...... xxix

Chapter 1 – Introduction...... 1

1.1. Motivation...... 1

1.2. Objectives and Goals...... 4

1.3. Significance of the Study...... 5

1.4. Overview of the Dissertation ...... 6

References ...... 8

Chapter 2 – Cell Membrane and Cell Junctions...... 10

2.1. General Structure and Functions of Cell Membrane ...... 10

2.2. Cell Membrane Dynamics...... 10

2.2.1. Diffusion across the Cell Membrane ...... 12

2.2.2. Protein–Mediated Transport and Cell Membrane Potential...... 13

2.2.2.1. The Resting Membrane Potential...... 16

2.2.3. Vesicular Transport ...... 16

2.2.4. Transepithelial Transport...... 17

vi 2.2.5. Osmosis and Tonicity ...... 18

2.3. Cell Junctions...... 19

2.3.1. Extracellular Matrix...... 20

2.3.2. Cell–Cell and Cell–Matrix Junctions and Adhesion...... 20

2.3.2.1. Cell–Cell Junctions...... 20

2.3.2.2. Cell–Matrix Junctions...... 23

2.4. Chapter Appendix ...... 25

2.4.1. Diffusion across the Cell Membrane ...... 25

2.4.1.1. Simple Diffusion ...... 25

2.4.1.2. Facilitated Diffusion ...... 26

2.4.2. Active Transport across the Cell Membrane ...... 27

2.4.3. Nernst Equation...... 28

2.4.4. Goldman–Hodgkin–Katz (GHK) Equation ...... 29

2.4.5. Osmotic Pressure ...... 30

References ...... 31

Chapter 3 – Biofunctionalized Interfaces...... 33

3.1. Self–Assembled Monolayers ...... 33

3.1.1. Gold Alkylthiolate Monolayers...... 36

3.1.2. Alkylsilane Monolayers...... 38

3.2. SAM–based Biofunctionalized Systems and Surfaces ...... 40

3.2.1. Common Methods for Preparing Mixed and Patterned SAMs...... 42

3.2.1.1. Coadsorption from Solutions of Mixed SAM Constituents ...... 42

3.2.1.2. MicroContact Printing...... 42

3.2.1.3. Substrates Having Surface Patterns

Made of Different Materials ...... 45

vii 3.3. Removal of Self–Assembled Monolayers...... 45

References ...... 47

Chapter 4 – Quantum Dots in Bio–Labeling and Bio–Imaging...... 52

4.1. Overview and Features of Quantum Dots ...... 52

4.2. QD Synthesis and Capping ...... 56

4.3. Surface Derivatizations...... 59

4.4. QD Bioconjugates for Bio–Labeling and Bio–Imaging ...... 60

4.5. Chapter Appendix ...... 67

4.5.1. Synthesis of CdSe/ZnS Core/Shell Quantum Dots ...... 67

4.5.1.1. Synthesis of Core CdSe Nanocrystals...... 67

4.5.1.2. Growth of ZnS Shell over CdSe Nanocrystals ...... 67

4.5.2. Synthesis of PbSe/PbS Core/Shell Quantum Dots ...... 68

4.5.2.1. Synthesis of Core PbSe nanocrystals...... 68

4.5.2.2. Growth of PbS Shells over PbSe Nanocrystals...... 69

4.5.3. Particle in a Three–Dimensional Quantum Box...... 70

4.5.4. Exciton Bohr Radius and the Energy Bandgap of QDs...... 73

References ...... 76

Chapter 5 – Cellular and Cell–Based Sensors ...... 82

5.1. Measurement of Cellular Activities and States ...... 82

5.2. Requirements of Biosensors...... 84

5.3. Cellular and Cell–Based Biosensors...... 85

5.3.1. Cellular Microorganism–Based Biosensors...... 85

5.3.2. Fluorescence Assays of Cellular Functions ...... 86

5.3.3. Cellular Biosensors Based on Cell Metabolism...... 88

viii 5.3.4. Cellular Biosensors Based on Electrical Impedance...... 90

5.3.5. Cellular Biosensors Based on Intracellular Potentials...... 93

5.3.6. Cellular Biosensors Based on Extracellular Potentials ...... 95

5.3.6.1. Analytical Methods ...... 97

5.3.7. Raman Spectroscopy Cell–Based Biosensors ...... 98

References ...... 102

Chapter 6 – Sonoporation: Ultrasound–Induced Permeabilization of

Cell Membrane...... 109

6.1. Basic Physics of Ultrasound...... 109

6.2. Ultrasonic Transducers...... 110

6.3. Acoustic Cavitations...... 112

6.3.1. Biological and Physical Consequences of Cavitations on Cells...... 115

6.4. Sonoporation...... 115

6.4.1. Mechanistic Studies of Plasma Membrane Sonoporation...... 116

6.4.2. Gene Therapy Prospects ...... 123

References ...... 126

Chapter 7 – Device Designs, Fabrications, and Characterizations...... 131

7.1. Integrated Microelectrode Array (IMA) Chips...... 131

7.1.1. Design and Architecture of IMA Chips ...... 131

7.1.2. Microfabrication of IMA Chips...... 133

7.1.2.1. Photomask Fabrication...... 133

7.1.2.2. Chip Fabrication ...... 134

7.1.2.3. Dicing ...... 136

7.1.3. Characterization of IMA Chips...... 138

ix 7.2. Ultrasonic Micro–Transducer Array (UMTA) Chips ...... 140

7.2.1. The Choice of Piezoelectric Material for UMTA Chips ...... 140

7.2.2. Design and Architecture of UMTA Chips...... 142

7.2.3. Process Development for UMTA Chip Fabrication...... 144

7.2.3.1. Inductively–Coupled Plasma Dry Etching of

PMN–PT Crystals ...... 145

7.2.3.1.1. Analysis of the Effect of Plasma Etching on

PMN–PT Crystals ...... 147

7.2.3.2. Etch–Mask Molding and Growth...... 149

7.2.3.2.1. Pulsed Electroplating ...... 149

7.2.3.2.2. Doubled–Exposure Laser Lithography...... 151

7.2.4. Microfabrication of UMTA Chips...... 154

7.2.4.1. PMN–PT Wafers...... 154

7.2.4.2. Photomask Fabrication...... 155

7.2.4.3. Biochip Fabrication ...... 155

7.2.4.3.1. Micro–Transducer Fabrication...... 155

7.2.4.3.2. Epoxy Filling, and Wafer Lapping, Polishing ...... 157

7.2.4.3.3. Top Electrodes and Interconnection...... 160

7.2.4.3.4. Device Encapsulation with Parylene C Coating ...... 160

7.2.4.3.5. Cell–Trapping Pads ...... 161

7.2.4.3.6. Wire–Bonding and Sealing ...... 163

7.2.5. Characterization of UMTA Chips ...... 163

7.2.5.1. Fundamentals of Ultrasounds Generated by

the Transducer...... 163

7.2.5.2. Micro–Transducer Performance Testing...... 165

x 7.2.5.2.1. Ultrasound Pressure Measurement

Using Hydrophone ...... 165

7.2.3.2. Electrical Impedance Spectrum ...... 168

7.3. Chapter Appendix ...... 170

7.3.1. List of Symbols ...... 170

7.3.2. List of Some Abbreviations ...... 170

7.3.3. List of Tools and Equipments Used in the Device Fabrication and

Process Characterizations ...... 171

References ...... 173

Chapter 8 – Cell Membrane Characterization at the Single–Cell Level ..... 176

8.1. Cell Membrane Characterization Using IMA Chip ...... 176

8.2. Materials and Methods ...... 177

8.2.1. Surface Modification of IMA Chip and

Single Cell Immobilization ...... 177

8.2.1.1. Protocols for Modification of IMA Chip’s

Surface Chemistries...... 180

8.2.1.2. Protocols for Single–Cell Immobilization on

Sensing/Detecting Microelectrodes ...... 182

8.2.2. Instrumentation and Impedance Spectroscopy...... 182

8.2.2.1. Phase–Sensitive Lock–in Detection Technique...... 182

8.2.2.2. Instrumentation and Impedance Spectroscopy...... 185

8.2.3. Experimental Results...... 187

8.2.4. Modeling and Data Fitting...... 190

8.2.4.1. Modeling the Cell–Electrode Heterojunctions with the Area–

Contact–Model–Based Distributive Circuit Network ...... 190

xi 8.2.4.1.1. Point–Contact Model with

Lumped Circuit Elements...... 192

8.2.4.1.2. Area–Contact Model with

Distributed Circuit Network...... 197

8.2.4.2. Numerical Data Fitting...... 203

8.2.5. Analysis and Discussions...... 205

8.2.5.1. Signal Enhancement via Cleft Thickness Control over

Surface–Modified Sensing Microelectrodes ...... 205

8.2.5.1.1. Influence of Cell Adhesion on the

Lock–in Measurement ...... 205

8.2.5.2. Frequency–Dependent Characteristics...... 209

8.3. Automation & Upgrade on IMA Chip Instrumention...... 215

8.3.1. Multiplexing and Data Acquisition ...... 215

8.3.2. Integration with Portable Tissue Culture Chamber and

Fluorescent Microscope...... 218

8.4. Chapter Appendix ...... 220

8.4.1. Single–Cell–Based Vs. Multi–Cell–Based Impedance

Sensing in Monitoring Cellular Response to Drug Treatment ...... 220

8.4.1.1. The Effect of Initial Cell Coverage on the Working

Electrode on the Transduced Electrical Signals...... 222

8.4.1.2. The Effect of Cell–to–Cell Junction on the Overall

Cell Shrinkage or Retraction ...... 226

8.4.2. Time Fractional Derivative...... 228

8.4.3. List of Symbols...... 229

8.4.4. List of Some Abbreviations...... 231

References ...... 233

xii

Chapter 9 – Site–Specific Cell Membrane Permeabilization ...... 236

9.1. Site–Specific Sonoporation at Cellular Level Using UMTA Chip...... 236

9.2. Materials and Methods ...... 240

9.2.1. Cell Culturing and Preparation ...... 240

9.2.1.1. Cell Thawing ...... 240

9.2.1.2. Cell Passing and Culturing ...... 241

9.2.2. Quantum Dots Tracking ...... 243

9.2.2.1. Characterization of Quantum Dots ...... 244

9.2.2.2. Preparation of QD–Suspended Tissue Culture Media ...... 245

9.2.3. Results from Control Experiments (QD – Uptake by

LU1205 Cells via Endocytosis Process ...... 246

9.2.3.1. Time–Lapsed Fluorescent Imaging ...... 246

9.2.3.2. Confocal Z–Stack Imaging ...... 248

9.2.3.3. Flow Cytometry...... 250

9.2.4. Site–Specific Cell Membrane Permeabilization...... 252

9.2.4.1. Instrumentations and Micro–Transducer Activation ...... 252

9.2.4.1.1. Peak Ultrasound Pressure on the Cell Membrane ...... 254

9.2.4.2. Cell Membrane Permeability Observation and Analysis ...... 255

9.2.4.3. Cell Viability Assay ...... 256

9.3. Experimental Results from Site–Specific Sonoporation ...... 256

9.4. Modeling and Analysis...... 260

9.4.1. Ultrasound–Induced Wound Model ...... 265

9.4.2. Calculation of Peak Acoustic Radiation Pressure ...... 268

9.4.3. Mass Transport through the Effective Wound and

xiii Intracellular Concentration of Quantum Dots...... 272

9.5. Analysis and Simulations...... 275

9.5.1. QD–Uptake and the Size of Membrane Wound ...... 275

9.5.1.1. Ultrasound–Pressure–Dependent QD–Uptake ...... 275

9.5.1.2. Ultrasound–Pressure–Dependent Wound Size...... 277

9.5.2. Model–Based Simulations and Calculations ...... 278

9.5.2.1. Analysis of the Influence of Transport–Model Parameters

on the Enhanced QD–Uptake by the Cells Sonoporated

with UMTAs ...... 278

9.5.2.2. Sonoporation–Enhanced Transport and QD–Uptake ...... 281

9.5.3. Effective Lateral Resolution of Sonoporation Using

UMTA Chips...... 282

9.6. Summary ...... 284

9.7. Chapter Appendix ...... 285

9.7.1. List of Symbols ...... 285

9.7.2. List of Some Abbreviations ...... 287

References ...... 288

Chapter 10 – Conclusions and Future Work ...... 292

Nontechnical Abstract ...... 300

xiv LIST OF FIGURES

2–1 The fluid mosaic model of a biological membrane...... 11

2–2 Transport across the pure phospholipid layer...... 12

2–3 Gated ion channels that respond to different types of stimuli...... 13

2–4 Schematics of a carrier protein mediating the passive transport...... 14

2–5 Different types of carrier proteins...... 15

2–6 Protein mediated active and passive transport...... 15

2–7 Endocytosis...... 17

2–8 Transepithelial transport...... 18

2–9 Cell swelling in hypotonic solution...... 19

2–10 Types of cell–cell junctions...... 22

2–11 (A) Diagram and (B) electron micrograph of a fibronectin molecule...... 24

2–12 Schematic of a hemidesmosome junction...... 24

2–13 The function of the rate of diffusion of small and non–polar molecules across the cell membrane...... 25

2–14 The function of rate of facilitated diffusion of molecules across the cell membrane...... 27

xv 3–1 (A) Typical SAM–forming molecule (decanthiol or CH3(CH2)9HS) with

the angular degrees of freedom for an all–trans chain, tilt angle (θt), tilt

direction (χt), and twist (ψ). (B) Schematic diagram of an ideal, single–

crystalline SAM of hexadecanethiolates (CH3(CH2)15SH) supported on a gold surface with a (111) texture...... 35

3–2 Chemisorption of n–Octadecyltrichlorosilane (OTS) on glass surface...... 39

3–3 Schematic illustration showing the reversible adsorption of a receptor to a mixed SAM presenting a ligand...... 41

3–4 Schematic illustration showing the physisorption of a protein to a patterned SAM formed by two different constituents...... 41

3–5 Procedure for patterning thiol–based SAMs using µCP...... 44

4–1 (A) Absorption (Abs) and emission (Em) of six different QD dispersions. (B) Photo demonstrating the size–tunable fluorescence properties and spectral range of the six QD dispersions plotted in (A) versus CdSe core size (r stands for the radius of CdSe core)...... 54

4–2 Transmission Electron Micrographs (TEM) of (A) one “bare” CdSe nanocrystallite and (B) one CdSe nanocrystallite with a 2.6 monolayer ZnS shell. (C) Artist’s rendering of CdSe quantum dot and CdSe/ZnS core–shell quantum dot...... 55

4–3 Absorbance and photoluminescence of the size–selected fractions of the thioglycolic acid–capped CdTe nanocrystals...... 57

4–4 Schematic of experimental setup for quantum dot synthesis...... 59

4–5 (A) Quantum dot resistance to photobleaching. (B) Quantitative analysis of changes in intensities of quantum dot 608–streptavidin and Alexa 488–streptavidin...... 61

xvi

4–6 Pseudocolored image depicting five–color QD staining of fixed human epithelial cells...... 62

4–7 A three–dimensional infinite potential energy box in which an electron is confined in three dimensions...... 71

4–8 Graphics representing the active region (top), allowed states in momentum space (middle), and density of states (bottom) for confinement in no dimensions (i.e., bulk material) (a), in one dimension (i.e., a quantum well) (b), in two dimensions (i.e., a quantum wire) (c), and in three dimensions (i.e., a quantum dot) (d)...... 72

4–9 Comparison of the electronic structure of the atomic orbitals in a silicon atom (left) to that of a silicon cluster molecule (middle) and to that of bulk silicon (right)...... 74

5–1 Classical electrical activity recording techniques...... 82

5–2 Patch–clamp recording to monitor ion channel activity...... 83

5–3 (A) Typical layout of a Fluorometric Imaging Plate Reader (FLIPRTM) (96–well mode). (B) A schematic representation of the general components of the FLIPRTM system...... 87

5–4 Light–addressable semiconductor sensors...... 89

5–5 A simplified schematic of the Electric Cell–Substrate Impedance Sensing (ECIS) instrumentation...... 91

5–6 Impedance measurements using ECIS technique showing: (A) an increase and fluctuations in effective resistance of the electrode due to cell spreading and cell micromotion respectively; (B) a sharp decrease and a transient rise in effective impedance of the electrode due to cell wounding after electroporation and cell healing/migration respectively...... 92

xvii 5–7 Neuroblastoma–glioma (NG108–15) cells as electrical detectors of chemical agents...... 94

5–8 Planar multi–electrode arrays permitting noninvasive, simultaneous recording from excitable tissue for measurement of action potential propagation...... 96

5–9 Culture of spinal cord neurons on MEA for toxicological evaluation...... 96

5–10 Experimental and instrumentation setup for Raman micro–spectroscopy measurements of cells...... 99

5–11 Raman spectra of viable (spectrum a) and dead (spectrum b) human lung derived (A549) cells...... 99

5–12 (a) Schematic of the final surface modification of gold-patterned silicon oxide substrate for subsequent single–cell adhesion. (b) Optical differential interference contrast (DIC) image of patterned DAOY cells on the surface–modified 20–um2 gold squares with 40–um spacing between squares...... 100

5–13 Raman spectra for two representative cells acquired prior to (0 h) and every 12 hours after etoposide exposure. (a) Raman spectra of a single cell that died after 36 hours. (b) Raman spectra of a single cell that showed resistance to etoposide and was viable over the course of the experiment...... 101

6–1 Acoustic cavitations...... 114

6–2 Intracellular delivery by ultrasound...... 116

6–3 Ultrastructural analysis of ultrasound’s effect on plasma membrane...... 117

6–4 Confocal fluorescence and brightfield microscopy of intracellular uptake and wound repair...... 118

xviii

6–5 Direct observation of ultrasound-induced pores by SEM...... 118

6–6 Intracellular concentration as a function of time after sonication for calcein (623 Da, diamond), bovine serum albumin (66 kDa, square), and dextrans (150 kDa, triangle; 500 kDa, circle; 2,000 kDa, star)...... 120

6–7 (A) Schematic diagram illustrating the effects of acoustic fields of identical frequency but differing intensity on microbubble (MCB) behavior. (B) Proposed model of the oscillating MCB–enforced pore formation in the cell membrane...... 122

6–8 A schematic representation of the transducer setup used for ultrasound application to cell suspension...... 123

6–9 An apparatus used to expose cells with ultrasound...... 124

6–10 (A) Transfection rate of living cells after ultrasound exposure as function of average peak pressure of the 20–s burst of 1.0–MHz ultrasound. (B) Transfection rate of living cells after exposure plotted against Albunex concentration at the time of exposure. (C) Transfection rate of live cells after repeated 1–s exposures to ultrasound at the indicated average peak pressure. (D) Transfection rate after 20–s exposure to 1.0–MHz ultrasound at indicated average peak pressure with a microbubble concentration of 50 × 106 bubbles/mL prior to exposure...... 125

7–1 Simplified cross–sectional schematic of integrated microelectrode array (IMA) biochip...... 132

7–2. The Integrated Microelectrode Array (IMA) chip layout showing sensing electrodes, counter electrodes, interconnection lines, and contact pads...... 133

7–3 Summary of fabrication process for Integrated Microelectrode Array (IMA) chip...... 136

xix

7–4 Integrated Microelectrode Array (IMA) Chip. (A) Size comparison with a penny. (B) The IMA chip with media reservoir well mounted. (C) Bright field image of microelectrode arrays (20–µm, 25–µm, 30–µm, and 250– µm in diameter). Bright field images of microelectrode: (D) 20–µm, (E) 25–µm, (F) 30–µm, and (G) 250–µm in diameter...... 137

7–5 Atomic Force Microscopy (AFM) scan image of 30–µm diameter Au microelectrode...... 139

7–6 Simplified cross–sectional schematic of ultrasonic micro–transducer array (UMTA) chip...... 143

7–7 SEM images of array structures produced by a 3–hour RIE with 10%

argon in SF6 showing (a) the array and (b) sidewall angle variation...... 145

7–8 X–Ray Diffraction (XRD) patterns of PMN–PT crystal’s surface. (a)

Before and (b) after Cl2–based plasma etching...... 148

7–9 Strain–electric field property of PMN–PT crystal after Cl2–based plasma etching...... 148

7–10 Cross–sectional FE–SEM image of two different Ni layers: the bottom layer was deposited by 2.5 MHz pulsed electroplating with current density of 10 mA/dm2; the top layer was deposited by DC electroplating with a current density of 35 mA/dm2...... 151

7–11 Schematics showing the difference between contact photolithography and doubled–exposure laser lithography...... 153

7–12 FE–SEM images of exposure profiles on SPR220 photoresist obtained by (A) contact photolithography and (B) doubled–exposure laser lithography...... 153

xx 7–13 PMN–PT (Lead Magnesium Niobate–Lead Titanate) single crystal piezoelectric disc embedded in a 2.5” PZT ceramic wafer...... 154

7–14 Ni etch–mask fabricated by doubled–exposure laser lithography and high–frequency pulsed electroplating. (A) Ni grown in photoresist mold. (B) Ni etch–mask after removing PR...... 155

7–15 FE–SEM images of: (A) 3 × 3 matrixes of transducers of 25 µm × 25 µm, 50 µm × 50 µm, 75 µm × 75 µm, and 100 µm × 100 µm cross–sectional area; (B) A 3 × 3 matrix of 25–µm transducers with etched depth of ~ 2 × 60 µm = 120 µm; (C) A 3 × 3 matrix of 75–µm transducers; (D) a single 75–µm transducer...... 156

7–16 Epoxy filling and curing process. (A) The wafer is placed at the bottom of the Teflon mold and Epoxy resin is poured to fill the space between the pillars. (B) After Epoxy is cured, the wafer is taken out of the mold (Etched–left substrate facing up in the picture). (C) Schematics showing the Epoxy filling and curing process...... 157

7–17 Flat Lapping and chemical mechanical planarization/polishing (CMP) process. (A) The etched–left substrate starts to be removed from the middle. (B) The end point for lapping at which the etched–left substrate is almost removed. (C) Schematics showing the flat lapping and CMP process...... 159

7–18 FE–SEM images of PMN–PT transducer top surface and Epoxy resin

surface conditions: (A) After 2 hours of rough polishing with 3–µm Al2O3 lapping powder; (B) After 1.5 hours of fine polishing with 0.1–µm 3M fine slurry...... 159

7–19 (A) Surface of the UMTA biochip showing 3 × 3 arrays of micro– transducers of four different size (25 µm, 50 µm, 75 µm, and 100 µm) [Courtesy of An Cheng]. (B) A bright field image of a 3 × 3 array of 25 µm × 25 µm transducer (top view) before Au electrodes and interconnections are formed. (C) An DIC image of a 3 × 3 array of 25 µm × 25 µm transducer (top view on the surface of finished UMTA biochip)...... 161

xxi

7–20 Schematic of UMTA biochip fabrication process...... 162

7–21 Conceptual illustration of the intensity (or pressure) distribution of continuous wave ultrasound: (a) the near–and far–field regions in relation to the transducer, (b) the field distribution of an ideal transducer source generating a continuous wave...... 164

7–22 Schematic of the experimental setup for measuring the acoustic pressure generated from the micro–transducers...... 166

7–23 Input electrical impedance spectra of micro–transducers: (A) 100–µm transducer; (B) 75–µm transducer; (C) 50–µm transducer; and (C) 25– µm transducer...... 169

8–1 Optical micrographs of Human Umbilical Vein Endothelial Cells (HUVEC) patterned on gold electrodes of silicon oxide substrates with gold electrodes coated with (a) fibronectin, (b) physically adsorbed REDVY (Ly –Arg– Glu–Asp–Val–Tyr), and (c) covalently bound KREDVY (Arg–Glu–Asp–Val–Tyr)...... 179

8–2 A schematic representation of surface molecular engineering of a gold– silicon dioxide substrate for guided (or surface–mediated) cell adhesion...... 180

8–3 A Schematics of a typical lock–in amplifier...... 183

8–4 Schematic of single–cell–electrode characterization experimental setup...... 186

8–5 Measured impedance–related RMS voltages (magnitude and phase) across sensing/detecting microelectrodes and large counter electrodes for different conditions at 6 operating frequencies...... 189

8–6 The equivalent Circuit Model (ECM) of the non–excitable cell membrane, which has Cl–, Na+, K+, and Ca2+ transmembrane ion channels...... 190

xxii

8–7 Schematic of a single cell immobilized over an electrode, illustrating transcellular (dashed) and paracellular current paths (solid), which are interlinked...... 192

8–8 Schematics of (A) point–contact (or) lumped circuit model, (B) area– contact (or) distributive circuit model of single–cell–covered surface– chemistry–modified microelectrode...... 196

8–9 Schematics of sensor circuitry with 5 weighted segments or domains in area–contact–model–based distributed circuit network...... 202

8–10 (A) Epi–DIC images of mouse fibroblast (NIH3T3) single cells patterned on 25µm and 30µm detecting electrodes and multiple cells patterned on a 110µm detecting electrode, which are modified with covalently bound KRGD peptide. (B) Fluorescent images of nuclei– (blue) and membrane– (green) stained NIH3T3 cells on electrodes of three different sizes and modified with physically absorbed (left, p–electrode) or covalently bound (right, c–electrode) KRGD...... 206

8–11 Voltage magnitudes of cell–free and mouse fibroblast (NIH3T3) cell– covered electrodes with surface areas of 625µm2 (25µm x 25µm), 900µm2 (30µm x 30µm), and 12,100µm2 (110µm x 110µm)...... 207

8–12 Simulated plot: normalized RMS voltage magnitudes, measured across the sensing microelectrode and the counter electrode, as a function of averaged cell–to–substrate separation distance...... 208

8–13 Simulated plot of frequency–dependent impedance difference, referenced to the cell–free baseline, after a single NIH3T3 cell was immobilized on the microelectrode (30µm–diameter) via different surface–chemistry– mediated cell adhesion processes...... 211

8–14 Overall impedance magnitude spectrum of cell–elcectrode hetero– structure (single–NIH3T3–covered covalently linked KRGD modified gold microelectrode, 30µm–diameter)...... 212

xxiii

8–15 Frequency–dependent response characteristics of Integrated Microelectrode Array (IMA) sensor configuration after single cell immobilization...... 214

8–16 (A) Schematics of automated excitation and data acquisition circuitry, which includes a lock–in amplifiers, a multiplexer, a function generator, a multi–channel pre–amplifiers module. (B) Circuit schematics of a unity– gain ultra–high input impedance amplifier in multi channel pre– amplifiers module...... 217

8–17 A complete multiplexing/addressing setup (a test setup with EG&G 5206 two–phase lock–in analyzer)...... 218

8–18 Experiment set–up to monitor real–time responses from individual cells and multiple–cell colonies upon exposure to toxins and drugs...... 219

8–19 (a) Schematic of IMA chip featuring optical epi–DIC micrographs of four single live U87MG cells on 30–µm working/detecting electrodes (right panel) and one 250–µm working/detecting electrode hosting a confluent monolayer of live U87MG cells (left panel) before and 16 hours post 5–µM CTX treatment. (b)–(e) Time–course normalized real impedance of the four U87MG single cells exposed to 5 µM CTX over a period of 16 hours...... 221

8–20 (a) Average normalized real impedance of multiple U87MG cells (blue) and single U87MG cells (green) in response to 5 µM CTX, as well as to single cells receiving no CTX treatment (brown). (b) Frequency spectroscopy plot of impedance magnitude |Z| at a frequency range from 500 to 20 kHz when cell–free electrodes of 30 µm and 250 µm diameters are seeded with U87MG cells...... 222

8–21 Modified point–contact–model–based equivalent circuit of single–cell– electrode heterostructure...... 223

xxiv 8–22 Influence of initial cell coverage (rcell,o/re) on the normalized real impedancse during the cell shrinkage or retraction...... 226

9–1 Schematic representation of various modes by which drug delivery can be enhanced by ultrasound...... 238

9–2 GFP–expressing Melanoma cancer cells (LU1205) after ~48 hours of incubation in the cell culture medium (DMEM + 10% FBS + 100 U/ml PS)

at 37 ˚C and 5% CO2 flow (~90–100 % confluent)...... 242

9–3 Melanoma LU1205 cancer cells grown on the surface of the UMTA chip. (A) Low magnification optical microscope image of the device surface with gold electrodes and connections lines. (B) High magnification showing melanoma cells growing on top of a 50 µm square electrode with Parylene C film between them...... 242

9–4 Emission (A) and absorption (B) spectrum of carboxylic–acid–surface– derivatized CdSe/ZnS core–shell quantum dots...... 245

9–5 The levels of QD uptake by LU1205 cells at QD–inoculation periods. (A) A fluorescent image before QD–suspended tissue culture medium was aspirated. QD uptake by LU1205 cells after 2 hours (B), 6 hours (C), and 12 hours (D) of incubation in the medium that contained 100 nM of carboxylic–acid–surface derivatized 615–nm QDs...... 247

9–6 Confocal microscopic images showing QD distribution inside LU1205 cells. Z–stack images from A to F: slices at (A), the bottom of LU1205 cells; (F), the top of LU1205 cells (A 2–µm increment in Z–axis between each image). (G) Reconstructed 3D image of QD–uptake distribution inside LU1205 cells...... 249

9–7 (A) Two–parameter histograms obtained by flow cytometry showing endocytosis–driven QD–uptake by LU1205 cells for different QD– inoculation periods. (B) Histogram of the percent increase of QD uptake by LU1205 cells at different incubation periods...... 251

xxv

9–8 Schematic of ultrasound–induced site–specific cell membrane permeabilization (sonoporation)...... 254

9–9 Micro–transducer arrays driven at three different RF powers at 30 MHz for 3 minutes. (A) GFP–expressing LU1205 cells seeded on the surface of the UMTA biochip. (B) LU1205 cells located above the micro– transducers (i.e., on the active areas) demonstrate CdSe/ZnS QD uptake by passive diffusion after 3–minute sonoporation. (C) Sum fluorescence– intensity spatial profile of QD transported into LU1205 cells located on the active areas of micro–transducers along the column 1, 2, and 3 of the 3 x 3 array structure...... 258

9–10 QD–uptake on active and non–active areas. (A) No QD–uptake was observed in the areas where micro–transducers are not activated (dashed lines show the location of micro–transducers). (B) QD–uptake by LU1205 cells after micro–transducer array (each micro–transducer in the array presents an area of 50μm × 50μm) was driven at 80mW, 30MHz for 3 minutes...... 259

9–11 Deformation responses of adherent living cells under mechanical stresses...... 261

9–12 (A) Cell membrane of adherent cell on the active area of UMTA biochip in normal steady state before being exerted by ultrasonic radiation pressure. (B) Cell membrane is stretched due to radiation pressure/force exerted by ultrasonic wave resulting in the increase in plasma membrane intermolecular distances. (C) Failure and rupture of the cell membrane above the threshold radiation pressure...... 262

9–13 QD nanoparticle endocytic pathway in human epidermal keratinocytes (HEKs)...... 263

9–14 Hypothesized model of enhanced QD transport (or transmembrane QD transport) induced by ultrasound generated from the micro–transducer of UMTA chip...... 265

xxvi 9–15 (A) Ultrasound–induced wound model for one wound. (B) Ultrasound– induced wound model for multiple wounds for mass transport calculations. (C) Geometrical model representing the membrane wound resealing overtime...... 266

9–16 (A) Peak ultrasonic radiation pressure (at the end of NFL) Vs frequency for different RF powers applied to the micro–transducer. (B) Normalized Peak Pressure Vs frequency plot showing the bandwidth of multi–mode micro–transducer (S = 25µm × 25µm)...... 271

9–17 Estimated intracellular concentrations of quantum dots, which were transported into the cytoplasm of the LU1205 cells seeded on the 25µm × 25µm active area presented by a micro–transducer, after 6–minute inoculation with QDs (post sonoporation) for three peak ultrasound radiation pressures applied to the cells...... 277

0 ( R ) 9–18 Calculated effective initial wound radius w,eff for three peak ultrasound pressures (0.2, 0.3, and 0.4 MPa)...... 278

9–19 Computed time–dependent intracellular concentrations of QDs transported into the cells for: (A) different effective initial wound size or

R0 radius ( w,eff ), i.e., ultrasound pressure dependence; (B) for different

mean effective wound lifetimes (τw,eff), i.e., wound healing dynamic

dependence; and (C) different lag times (to) for QD introduction following the sonoporation...... 279

9–20 Computed total concentrations of QDs transported into the cells in 6 minutes after the sonoporation (or until the wound is completely healed)

R0 for: (A) different effective initial wound size or radius ( w,eff ), i.e., ultrasound pressure dependence; (B) for different mean effective wound

lifetimes (τw,eff), i.e., wound healing dynamic dependence; and (C)

different lag times (to) for QD introduction following the sonoporation...... 280

xxvii LIST OF TABLES

2–1 The estimated ion concentrations and their respective equilibrium potentials in mammalian cells...... 29

4–1 Schematic of generic quantum dot (QD) solubilization and biofunctionalization (a–h)...... 64

7–1 Properties of a few important piezoelectric materials used in high frequency (HF) transducer designs...... 141

8–1 Formulas for weighted circuit elements of the area–contact–model–based distributed circuit network...... 201

8–2 List of estimated values of ECM model parameters that give the best fit to the experimental data...... 204

xxviii ACKNOWLEDGEMENTS

First, this work would not have been possible without funding provided by the

National Institute of Health (NIH/GMS) (Grant No. R01GM075095).

Second, I would like to express my greatest gratitude to my advisor, Prof. Jian Xu, for providing me the opportunity to complete my Ph.D. dissertation at the Pennsylvania

State University, for his kindly support and guidance on my dissertation work, for his expertise, motivation, enthusiasm, and immense knowledge on the research.

Third, I would like to give sincere gratitude to Prof. Bernhard Tittmann, Prof. Yinong

Yang, and Prof. Sulin Zhang for serving on my committee and their invaluable suggestions to improve my research work.

Fourth, I would like to thank my collaborators: Prof. Miqin Zhang from the department of materials science and engineering at University of Washington, Prof.

Cheng Dong and Late Prof. Nadine Barrie Smith from the department of bioengineering at the Pennsylvania State University.

Fifthly, I am also indebted to my fellow colleagues: Dr. An Cheng, Fareid Asphahani,

Ryan Buckmasters, Payal Khanna, Dr. Fan Zhang, Dr. Chunfeng Zhang, Daniel Ahmed, and Dr. Ting Zhu. I am especially grateful to the research group of Prof. Miqin Zhang, of

Prof. Cheng Dong, and of Late Prof. Nadine Barrie Smith.

Sixthly, I would also like to give my special gratitude to my family for being there for me. I sincerely thank to my father, Dr. Kyin Thein, for his constant love, support and encouragement. I also greatly appreciate the help from my aunts, Li Li and Pawt Pawt, and encouragement from my younger brother, Dr. Kyaw Min Thein. Without their supports and efforts, I could never go this far.

Seventhly, This acknowledgement will not be complete without mentioning of my friends, Tin Htun Oo and Tun Tun Win who helped and supported me in bad times.

Finally, I would like to thank to my late mother, Dr. Win Win Than, for her love, disciplines, and lifelong sacrifice that guided me to this accomplishment.

xxix Chapter 1 INTRODUCTION

1.1. Motivation

Cell–based sensors incorporate living biological cells as sensing or recognition elements that convert changes in an immediate environment to meaningful signals for processing and analysis [Asphahani et al., 2007; Bousse, 1996; Kovacs, 2003; Pancrazio et al., 1999]. They can be employed to detect the presence of chemicals or substances that cause physiological changes in the living cells being incorporated as well as monitoring the physiochemical effect of these chemicals on the living cells. Among various types of cell–based sensors, cellular impedance sensors, addressed electrically via impedance characterizations, are capable of determining the effects of internal and/or external stimuli on cells noninvasively and quantitatively [Curtis et al., 2009;

Ehret et al., 1997; Kovacs, 2003; Luong, 2003; Pancrazio et al., 1999]. These cellular

impedance sensors usually incorporate an array (or arrays) of planar microelectrodes

and offer high–throughput measurements and assays since each planar microelectrode in the array renders as a miniature test site.

Many current cell–based sensors usually measure or detect the averaged behavior or response of groups or populations of cells that often masks crucial extrinsic and intrinsic cell–to–cell differences [Chen et al., 2005; Levsky et al., 2003]. Studies have shown that the behavior and response of ‘‘outlier’’ cells, that are lost in averaged bulk measurements, are far from the averaged cell population response and can be of critical importance in the study of cellular biology and disease such as drug resistance in cancers and gene expression variations within clonal populations [Bahar et al., 2006; Levsky et al., 2003; Tolstonog et al., 2005]. Therefore, single–cell–level studies are required and

1 are as equally important as multi–cell– or tissue–level studies in order to understand the

fundamental nature and mechanism of a particular cell type. However, traditional

single–cell–level measurement techniques such as patch clamping are time–consuming and produce lower throughputs. Biological chips that can expedite single–cell–level measurements and studies could be beneficial for many scientists and researchers in cell biology and medicine. It is anticipated that by further reducing the dimension of planar

microelectrodes incorporated in cellular impedance sensors close to the size of single

cells of interest, such sensors could be employed in single–cell–level measurements and

studies as well as offering high–efficiency and high–throughput assays for drug screenings and characterizations.

Among many anticancer treatments, much attention has recently given to gene therapy in clinical trials for its low side effects compared to radiotherapy and chemotherapy [Verma and Somia, 1997]. Consequently, sonoporation (or cell membrane permeabilization using ultrasound) has been investigated by clinical

researchers as a viable method for delivering therapeutic materials (i.e., gene products)

into the cancer cells [Bao et al., 1997; Fechheimer et al., 1987; Siu et al., 2007; Song et

al., 2002; Tschoep et al., 2001] and microdevice–based ultrasonic transducers, which

consist of piezoelectric structures and can generate ultrasonic pressure waves to

sonoporate cells or tissues [Siu et al., 2007], become potential solutions for enhancing

drug/gene–product delivery into the cancer cells both in vitro and in vivo. Furthermore,

Sonoporation is used for cell transfection in vitro [Bao et al., 1997; Greenleaf et al.,

1998]. The method offers flexibility and improves transfection efficiency and cell

viability in some cell lines [Pepe et al., 2004].

Most studies in sonoporation–enhanced drug or gene product delivery into the cells

in vitro have been carried out on the population level (i.e., sonoporation of the cell

suspension) or the tissue level. Very few studies have been carried out on a single–cell

2 level or even on a small colony/cluster of multiple cells due to technical limitations such

as the size and dimension of piezoelectric transducers as well as time–consuming low–

throughput transducer setups and instrumentations. In addition, current sonoporation– enhanced drug delivery, gene product delivery, or DNA transfection completely lack the ability to target specific cells. These technical challenges limit the implementation of sonoporation–enhanced drug or gene–product delivery into the cells in biochip–based assays where miniature test sites are employed to achieve high–efficiency and high– throughput screenings.

In drug screenings and characterizations, the precise response of individual cells to drug should also be monitored and studied. As mentioned previously, in single–cell– level analyses, the influence of cell–to–cell interactions can also be ruled out, providing the direct and full response of the specific cell under study to drugs, which could be significantly different from average response of the cell population and could lead to the

better understanding of drug–cell interactions. Moreover, results from single–cell

analysis could be compared with those from multiple–cell analysis to optimize the drug

design. It is known that the implementation of biochip platforms in drug screenings and

characterizations provides high efficiency and throughput since biochips house many

miniature test sites or microenvironments. Therefore, by incorporating the array of

miniature ultrasonic transducers into the biochip architecture, not only would such

biochips further improve the efficiency and throughput of the assay by enhancing the

drug delivery into the cells via sonoporation, they would also enable studies of

sonoporation–enhanced drug or gene–product delivery into the cells at the single–cell or

cellular level with high efficiency and throughput.

It has been expected that single–cell level biochip architectures could overcome the

limitations inherent in multiple–cell platforms and studies using such biochips could

lead to a better understanding of the entire cellular population [Veiseh et al., 2007].

3 Furthermore, as previously mentioned, it is imperative that high–yield biological chips that deliver sensing and/or sonoporation capabilities at the single–cell level (as well as cellular level) could be advantageous to many researchers in biology and medicine. They are also versatile and can be implemented as sub–systems in sensing, screening, and characterization platforms. Therefore, it is imperative to develop biological chips that

could offer many advantageous such as high efficiency and throughput as well as

solutions to current technical limitations. Such biological chips will consist of cell–sized

micro–engineered structures such as micro–electrodes or/and micro–transducers.

Hence, the challenges in making these biochips lie in the fabrication of tiny or cell–sized features. Advances and breakthroughs in micro– and nano–fabrication, nano– technology, and biomaterials in recent years have made it possible to realize such biological chips. These micro–fabricated biological chips integrated with proper bioinstrumentations could deliver high efficiency and improve the quality of the research outcomes. Therefore, it is also foreseen that the micro–fabricated biological devices (or biochips) is a promising and growing research area and could become essential technology in the research and development of drugs and gene products for therapeutic purposes in the future.

1.2. Objectives and Goals

In this study, two prototype chips were designed and micro–fabricated. These chips include Integrated Microelectrode Array (IMA) chips and Ultrasonic Micro–Transducer

Array (UMTA) biochips. The IMA chip consists of arrays of cell–size microelectrodes, which are incorporated into the chip architecture to detect and characterize changes in cellular components, especially the cell membrane, at the single–cell level. The UMTA

biochip consists of arrays of high–aspect–ratio piezoelectric–based micro–transducers,

which are incorporated into the chip architecture to sonoporate cell membrane at the

4 cellular level with high spatial specificity or resolution. Hence, the main objectives of

this study were:

(1) To electrically characterize changes or variations in cellular components,

especially the cell membrane, at the single–cell level using the IMA chip that

employs microelectrode array architecture.

(2) To demonstrate the site–specific sonoporation (or the cell membrane

permeabilization) at the cellular level using the UMTA biochip that employs

Ultrasonic Micro–Transducer Array (UMTA) architecture.

(3) To develop models that represent the working mechanisms of IMA–based

sensor platforms and of UMTA–based site–specific sonoporation for enhanced

nanoparticle delivery into the cancer cells.

The primary and ultimate goal of this study is to successfully demonstrate the

concept and feasibility of microelectrode array and micro–transducer array

architectures in high–throughput cellular/cell–based impedance sensing and ultrasound–enhanced targeted nanoparticle delivery into the cancer cells respectively.

1.3. Significance of the Study

Firstly, the device fabrication processes and characterization methods developed and demonstrated in this study could be transferred to industries as simple and cost– effective processes for manufacturing miniature biological devices. Secondly, the novel device designs, architectures, and working concepts demonstrated in this study can be employed in the development of next–generation biochip platforms. Thirdly, bioinstrumentations and methods employed and demonstrated in this study could become part of the standard bioinstrumentations or techniques employed by researchers and scientists in cell biology and medicine. For instance, the methodologies, techniques,

5 and analytical methods demonstrated in this study can be employed in biosensing, drug efficacy study, and drug/gene product delivery study. Fourthly, the study addresses and

provides effective engineering solutions to many current technical limitations and difficulties in traditional cell–based screening and characterizations, which are being

encountered by researchers and scientists in cell biology and medicine. It is also

anticipated that high–yield biological chips will expedite current research in drug

screening assays and improve the quality of outcomes. Successful and promising results

from this study could lead to the development and implementation of commercial–level single–cell (as well as multi–cell) biosensing and assay chips. Fifthly, the results and findings from cellular characterizations at the single–cell level and site–specific sonoporation of cell membrane at the cellular level could lead to the further and deeper understanding of cell nature, cell response, and cellular mechanisms.

1.4. Overview of the Dissertation

Due to the multidisciplinary nature of this dissertation work, which includes acoustics, biochemistry, bioengineering, biomaterials, chemical engineering, electrical engineering, electrochemistry, materials science, micro–/nano–fabrication, molecular and cell biology, nanotechnology, and quantum optics, backgrounds, fundamentals, and literature reviews are divided into five chapters as follows:

(1) The fundamentals of cell membrane structures and their dynamics are

described and discussed in chapter 2.

(2) The fundamentals of self–assembled–monolayer–based biofunctionalization

techniques, which are widely employed in biochip technology, are introduced

and discussed in chapter 3.

(3) The fundamentals and synthesis of nanoparticle Quantum Dots and the brief

survey of their applications in bio–labeling and bio–imaging are presented in

chapter 4.

6 (4) Chapter 5 surveys the different types of cellular and cell–based sensors and

their applications.

(5) Sonoporation and bio–effects of ultrasound are introduced in chapter 6.

Applications of sonoporation in transfection are also presented in this chapter.

Nevertheless, it is almost impossible to include all the backgrounds and fundamentals in this dissertation. Therefore, some of the detail fundamentals related to ultrasounds, piezoelectric materials, transducers, electrochemistry, and electrical circuit analysis are not included since they can be found in many standard textbooks.

After the background information is provided from chapter 2 to chapter 6, the microfabrication of IMA chip and UMTA biochip are briefly described in chapter 7.

The underlying motivations, process developments, process integrations, and process characterizations are also discussed. However, the information regarding the device fabrication is so immense that only the brief summaries and important points are described in this chapter. Many detailed informations regarding the basic fundamentals of the fabrication processes employed in this study can easily be found in micro–/nano– fabrication handbooks.

As two types of biological chips were fabricated, the experimental study is divided into two major parts as mentioned in section 1.2 and is described in two separate chapters. The IMA chip was used in cell membrane characterization at single–cell level and the UMTA biochip was employed in site–specific cell membrane permeabilization at cellular level. Therefore, studies related to IMA chip and UMTA biochip are described in chapter 8 and chapter 9 respectively. Finally, conclusions and future works are discussed in chapter 10, which summarizes the whole study.

7 References

Asphahani, F., and Zhang, M., 2007. Cellular Impedance Biosensors for Drug Screening and Toxin Detection. Analyst. 132, 835–841.

Bao, S., Thrall, B., and Miller, D., 1997. Transfection of a Reporter Plasmid into Cultured Cells by Sonoporation in vitro. Ultrasound in Medicine & Biology. 23, 953–959.

Bousse, L., 1996. Whole Cell Biosensors. Sensors and Actuators B. 34, 270–275.

Chen C. S., Jiang, X., and Whitesides, G. M., 2005. Microengineering the Environment of Mammalian Cells in Culture. MRS Bulletin. 30, 194–201.

Curtis, T. M., Widder, M. W., Brennan, L. M., Schwager, S. J., van der Schalie, W. H., Fey, J., and Salazar, N., 2009. A Portable Cell – Based Impedance Sensor for Toxicity Testing of Drinking Water. Lab on a Chip. 9, 2176–2183.

Ehret, R., Baumann, W., Brischwein, M., Schwinde, A., Stegbauer, K. and Wolf, B., 1997. Monitoring of Cellular Behavior by Impedance Measurements on Interdigitated Electrode Structures. Biosensors and Bioelectronics. 12 (1) 29–41.

Fechheimer, M., Boylan, J., Parker, S., Sisken, J., Patel, G., and Zimmer, S., 1987. Transfection of Mammalian Cells with Plasmid DNA by Scrape Loading and Sonication Loading. Proceeding of National Academy of Science USA. 84, 8463–8467.

Greenleaf, W. J., Bolander, M. E., Sarkar, G., Goldring, M. B., and Greenleaf, J. F., 1998. Artificial Cavitation Nuclei Significantly Enhance Acoustically Induced Cell Transfection. Ultrasound in Medicine and Biology. 24 (4), 587–595.

Kovacs, G. T. A., 2003. Electronic Sensors with Living Cellular Components. Proceedings of the IEEE. 91 (6), 915–929.

Levsky, J. M., and Singer R. H., 2003. Gene Expression and the Myth of the Average Cell. Trends in Cell Biology. 13 (1), 4–6.

Luong, J. H. T., 2003. An Emerging Impedance Sensor Based on Cell–Protein Interactions: Applications in Cell Biology and Analytical Biochemistry. Analytical Letters. 36 (15), 3147– 3164.

Pancrazio, J. J., Whelan, J. P., Borkholder, D. A., Ma, W., and Stenger, D. A., 1999. Development and Application of Cell–Based Biosensors. Annals of Biomedical Engineering. 27 (6), 697– 711.

8 Pepe, J., Rincon, M., and Wu, J., 2004. Experimental Comparison of Sonoporation and Electroporation in Cell Transfection Applications. Acoustics Research Letters Online. 5 (2), 62–67.

Song, J., Tata, D., Li, L., Taylor, J., Bao, S., and Miller, D. L., 2002. Combined Shock-wave and Immunogene Therapy of Mouse Melanoma and Renal Carcinoma Tumors. Ultrasound in Medicine & Biology. 28, 957–964.

Sui, G., Orbulescu, J., Ji, X., Gattás–Asfura, K. M., Leblanc, R. M., and Micic, M., 2003. Surface Chemistry Studies of Quantum Dots (QDs) Modified with Surfactants. Journal of Cluster Science. 14 (2), 123–133.

Tolstonog, G. V., Belichenko–Weitzmann, I. V., Lu, J. –P., Hartig, R., Shoeman, R. L., Traub, U., and Traub, P., 2005. Spontaneously Immortalized Mouse Embryo Fibroblasts: Growth Behavior of Wild–Type and Vimentin–Deficient Cells in Relation to Mitochondrial Structure and Activity. DNA and Cell Biology. 24 (11), 680–709.

Tschoep, K., Hartmann, G., Jox, R., Thompson, S., Eigler, A., Krug, A., Erhardt, S., Adams, G., Endres, S., and Delius, M., 2001. Shock Waves: A Novel Method for Cytoplasmic Delivery of Antisense Oligonucleotides. Journal of Molecular Medicine. 79, 306–313.

Veiseh, M., Veiseh, O., Martin, M. C., Bertozzi, C., Zhang, M., 2007. Single–Cell–Based Sensors and Synchrotron FTIR Spectroscopy: A hybrid System towards Bacterial Detection. Biosensors and Bioelectronics. 23 (2), 253–260.

Verma, M. I., and Somia, N., 1997. Gene Therapy–Promises, Problems, and Prospects. Nature. 389, 239–242.

9 Chapter 2 CELL MEMBRANE AND CELL JUNCTIONS

2.1. General Structure and Functions of Cell Membrane

Cell Membrane, also known as plasma membrane, is a phospholipid bilayer that forms a two–dimensional fluid and separates the aqueous fluid of the cell’s interior and outside environment [Alberts et al., 2004]. Proteins of various kinds are inserted into or through the phospholipid bilayer, which are responsible for most of the functions. The

three–dimensional fluid mosaic model of a cell membrane, revealed by freeze–fracture electron micrographs, is illustrated in figure 2–1 [Silverthorn, 2007]. The general functions of the cell membrane are: (i) physical isolation or barrier: it separates intracellular fluid from the surrounding extracellular fluid; (ii) regulation of molecule exchange with the environment: it controls the entry of ions and nutrients into the cell, the elimination of cellular wastes, and release of products from the cell; (iii) communication between the cell and its environment: it contains proteins that enable the cell to recognize and respond to molecules or to changes in its external environment; and (iv) structural and mechanical support: it contains proteins that hold the cytoskeleton and cell shape.

2.2. Cell Membrane Dynamics

Cell membranes are known to be selectively permeable. The lipids and proteins embedded in the cell membrane usually determine which molecules will enter or leave the cell. Water, oxygen, carbon dioxide, and lipids, can move easily across most cell membranes. However, ions (K+, Na+, Cl–, Ca2+, etc.), most polar molecules, and large molecules (such as proteins), cannot enter the cell easily. As summarized in figure 2–2,

two properties of a molecule influence its movement across cell membranes: the size of

10 the molecule and its lipid solubility [Alberts et al., 2004; Silverthorn, 2007]. Very small molecules and those that are lipid soluble can cross directly through the phospholipid bilayer. But larger or less lipid–soluble molecules requires specific mechanisms for being transported and membrane proteins are the usual mediators in transporting these molecules. Different transport mechanisms across cell membranes are: (i) diffusion, (ii)

protein–mediated transport, (iii) vesicular transport, and (iv) transepithelial transport.

Figure 2–1. The fluid mosaic model of a biological membrane. The cell membrane is a double layer of phospholipid molecules. Proteins of various kinds are inserted into and through the phospholipid bilayer. After [Silverthorn, 2007].

11

Figure 2–2. Transport across the pure phospholipid layer. The rate at which a molecule diffuses across the cell membrane depends on its size and lipid solubility. The smaller the molecule and the less favorable interactions with water (i.e., the less polar it is), the more rapidly the molecule diffuses across the phospholipid bilayer. Many molecules that the cell uses as nutrients are also too large and polar to pass through the phospholipid bilayer. After [Albert et al., 2004].

2.2.1. Diffusion across the Cell Membrane Substances that are hydrophilic and can dissolve in water are usually lipophobic and cannot easily cross the cell membrane. Lipophilic molecules can pass through the phospholipid bilayer and move by diffusion, i.e., the passive movement of molecules down a chemical concentration gradient from an area of higher concentration to an area of lower concentration. According to the Fick’s law of diffusion, the rate of diffusion across the cell membrane is directly proportional to the membrane permeability (the ability of the diffusing molecule to dissolve in the lipid layer of the cell membrane), the magnitude of concentration gradient across the cell membrane, the surface area of the cell membrane, and inversely proportional and inversely related to the thickness of the cell membrane and the size of diffusing molecules respectively [Silverthorn, 2007].

12 2.2.2. Protein–Mediated Transport and Cell Membrane Potential

The vast majority of solutes, either lipophobic or electrically charged, cross cell membranes with the aid of membrane proteins, a process known as protein–mediated transport. Among various types of transport proteins, channel proteins form open, water–filled passage way and regulate the movement of substances through them.

Channel proteins can be classified into open channels (leak channels or pores) and gated channels. In gated channels, the opening or closing of the channel can be controlled by:

(i) intracellular messenger molecules or extracellular ligands (chemically gated channels), (ii) electrical state of the cell (voltage–gated channels), or (iii) a physical

change, such as increased in temperature or a force that put tension on the membrane

and pops the channel open. Once the channel is opened, molecules and solute can cross

the cell membrane down the concentration gradient. Most ion channels are gated and

the passage of ions is controlled by these mechanisms as illustrated in figure 2–3.

Figure 2–3. Gated ion channels that respond to different types of stimuli. Depending on the type of ion channel, the gates open in response to a change in the voltage difference across the membrane (A), to the binding of a chemical ligand to the channel, outside (B) or inside the cell (C), or the mechanical stimulation (D). After [Albert et al., 2004].

13 Carrier proteins, unlike channel proteins, do not form a continuous connection or passage between the intracellular and extracellular fluid as shown in figure 2–4. They bind to substrates and then change their conformations in order to transport the molecules across the cell membrane. As illustrated in figure 2–5, carrier proteins can be classified into: (1) uniport carriers: a single solute is transported across the membrane,

(2) symport carriers: one solute and another solute are transported simultaneously or

sequentially in the same direction, and (3) antiport carriers: one solute and another

solute are transported simultaneously or sequentially in the opposite direction.

If protein–mediated transport moves solutes and molecules down their concentration gradient, the process is known as passive transport. On the other hand, if protein–mediated transport requires energy from ATP or another outside sources (such as light) and moves a substance or molecule against its concentration gradient as illustrated in figure 2–6, the process is known as active transport. Only carrier proteins can carry out active transport, but both carrier proteins and channel proteins can carry out passive transport.

Figure 2–4. Schematics of a carrier protein mediating the passive transport. The carrier protein changes its conformation from state A to B upon binding the solute in order to transport it across the cell membrane down the concentration gradient. After [Albert et al., 2004].

14

Figure 2–5. Different types of carrier proteins (schematics). The carrier proteins transport solutes across the cell membrane in several different ways: a single solute (uniports), one solute and another solute simultaneously or sequentially in the same direction (symports), or in the opposite direction (antiports). After [Albert et al., 2004].

Figure 2–6. Protein mediated active and passive transport. The vast majority of solutes, either lipophobic or electrically charged, cross cell membranes by passive or active transport with the aid of membrane transport proteins. Passive transport, in the same direction as a concentration gradient, occurs spontaneously, whereas active transport (transport against a concentration gradient) requires an input of energy. After [Albert et al., 2004].

15 2.2.2.1. The Resting Membrane Potential

Most organic compounds and ions carry a net electrical charge. Potassium (K+) is a major cation within cells, and sodium (Na+) dominates the extracellular fluid. Chloride ions (Cl–) mostly remain with Na+ in the extracellular fluid, whereas phosphate ions

– 2– 3– (H2PO4 , HPO4 , and PO4 ) and negatively charged proteins are also the major anions

of the intracellular fluid. Protein–mediated diffusion and active transport of ions across the cell membrane create an electrical gradient, with the inside of cell being negative relative to the extracellular fluid. This electrical gradient between the extracellular fluid and the intracellular fluid is known as the resting membrane potential difference.

Moreover, the active transport of positive ions out of the cell also creates a concentration gradient. The combination of electrical and concentration gradients is called an electrochemical gradient and the movement of an ion across the cell membrane is influenced by the electrochemical gradient for that ion. The membrane potential that exactly opposes the concentration gradient of an ion is known as the equilibrium or

Nernst potential [Malmivuo 1995; Silverthorn, 2007]. In most excitable cells, potassium ion (K+) is the primary ion that determines the resting membrane potential and changes in membrane permeability to ions such as K+, Na+, Ca2+, or Cl– will alter the resting membrane potential and create electrical signals in excitable cells such as neurons and myocytes.

2.2.3. Vesicular Transport

Large macromolecules and particles are brought into cells by endocytosis: the process by which cells absorb particles or molecules from outside the cell by engulfing it with their cell membrane as shown in figure 2–7. The opposite of that process is excocytosis in which particles or molecules leave the cells. There are three main types of endocytosis: (1) phagocytosis: the process by which cells ingest large objects such as bacteria or viruses (In this process, the cell membrane folds inwards, enclosing the bacteria or virus in a pocket, and then engulfs the object by pinching it off); (2)

16 pinocytosis: the process by which solutes and proteins are up–taken; (3) receptor–

mediated endocytosis: the process by which proteins or solutes are locked onto

receptors/ligands in the cell membrane and then the particles are engulfed.

Figure 2–7. Endocytosis. Food particles are taken into the cell via endocytosis into a vacuole. Lysosomes attach to the vacuole and release digestive enzymes to extract nutrients. The leftover waste products of digestion are carried to the plasma membrane by the vacuole and eliminated through the process of exocytosis. After [Alberts et al., 2009].

2.2.4. Transepithelial Transport

Transporting solutes and molecules across epithelial cells, also known as

transepithelial transport, uses a combination of active and passive transport. Epithelia

in the intestine and kidneys have different membrane proteins on their apical and

basolateral surfaces. This polarized distribution of transporters results in one–way

movement of certain molecules across the epithelium. The movement of glucose from

the lumen of the kidney tubule or intestine to the extracellular fluid, as illustrated in

figure 2–8, is an example of directional movement across a transporting epithelium.

Moreover, larger molecules also cross epithelia by a process known as transcytosis,

which includes vesicular transport.

17

Figure 2–8. Transepithelial transport. Two types of glucose carriers enable gut epithelial cells to transfer glucose across the gut lining. Glucose is actively transported into the cell by Na+–driven glucose symports at the apical surface, and it is released from the cell down its concentration gradient by passive glucose uniports at the basal and lateral surfaces. These two types of glucose carriers are kept segregated in the plasma membrane by the tight junction. After [Alberts et al., 2004].

2.2.5. Osmosis and Tonicity

The concentration gradient of particles (ions or intact molecules) across the cell

membrane also induces the movement of water across the cell membrane. This

phenomenon is called osmosis. The cell volume changes when it is placed in the

solution that has different osmolarity (the number of particles per liter of solution) than

that of intracellular liquid. Tonicity of a solution is a measure of the osmotic pressure of

two solutions separated by a selectively permeable membrane. Tonicity of a solution is

also used to describe the change in cell volume, which occurs when the cell is placed in

18 that solution. Cells swell in hypotonic solutions as the water molecules move into the

cell (figure 2–9) and shrink in hypertonic solutions as the water molecules are drawn out

of the cell. If the cell does not change size at the equilibrium, the solution is isotonic

(both the solution and intracellular liquid have equal concentration of solutes). If the

concentration of solutes in intracellular fluid is higher than that of extracellular fluid,

cells use different tactics such as pumping out ions or periodically ejecting the water that

moves into the cell to avoid osmotic swelling [Alberts et al., 2004].

Figure 2–9. Cell swelling in hypotonic solution. Diffusion of water molecules across cell membrane is known as osmosis. If the concentration of solutes inside a cell is higher than that outside, water molecules will move in by osmosis, causing the cell to swell. If the difference in solute concentration is great enough, the cell will burst. After [Alberts et al., 2004].

2.3. Cell Junctions

Cells assemble into the larger units called tissues. The cells in any tissue are held or linked together by specialized connections called cell junctions and by other support structures. Types of tissues range from simple tissues containing only one cell type, such as the lining of blood vessels, to complex tissues containing many cell types and extensive extracellular material, such as connective tissue. The cells of most tissues work together to achieve a common purpose.

19 2.3.1. Extracellular Matrix

Extracellular matrix, also referred to as matrix, is extracellular material synthesized and secreted by the cells of a tissue. It plays a vital role in many physiological processes, ranging from cell growth and development to cell death. There are two basic components in the extracellular matrix: proteoglycans and insoluble protein fibers.

Proteoglycans are glycoproteins, which covalently bound to polysaccharide chains.

Insoluble protein fibers such as collagen, fibronectin, and laminin provide strength and anchor cells to the extracellular matrix. Attachments between the extracellular matrix

and proteins in the cell membrane or the cytoskeleton are one means of communication

between the cell and its external environment. The amount of extracellular matrix in a

tissue is highly variable (For instance, nerve and muscle tissue have very little matrix,

but the connective tissues, such as cartilage, bone, and blood, have extensive matrix).

2.3.2. Cell–Cell and Cell–Matrix Junctions and Adhesions

General functions of cell junctions are: communication, occlusion, and anchorage.

Cell–cell or cell–matrix adhesions are mediated by cell adhesion molecules (CAMs),

which are membrane–spanning proteins responsible both for cell junctions and for

transient cell adhesions. CAMs are essential for normal growth and development (e.g.,

growing nerve cells creep across the extracellular matrix with the help of nerve–cell

adhesion molecules, or NCAMs). Cell adhesion helps white blood cells escape from the

circulation and move into infected tissues, and it allows clumps of platelets to cling to

damaged blood vessels [Silverthorn, 2007].

2.3.2.1. Cell–Cell Junctions

There are three types of cell junctions: gap junctions, tight junctions, and anchoring junctions (as shown in figure 2–10). Gap junctions allow chemical and electrical signals

20 to pass directly from cell to cell. They create cytoplasmic communication bridges between adjoining cells. Cylindrical proteins, known as connexins, form passageways that are able to open and close regulating the movement of small molecules and ions through them.

Tight junctions are occluding junctions that prevent the movement of material between cells. They are usually formed with the help of proteins called claudins and occludins to fuse cell membranes of adjacent cells together. Tight junctions in the intestinal tract and kidney prevent most substances from moving freely between the external and internal environments [Albert et al., 2004; Silverthorn, 2007]. They also create the blood–brain barrier that prevents potentially harmful substances in the blood from reaching the extracellular fluid of the brain [Silverthorn, 2007].

Anchoring junctions hold cells to each other or to the extracellular matrix. CAMs are essential in cell adhesion and in anchoring junctions. Cell–cell anchoring junctions take the form of either adherens junctions or desmosomes. Adherins joins an actin bundle in one cell to a similar bundle in a neighboring cell. Desmosomes attach to intermediate filaments of the cytoskeleton and are the strongest cell–cell junctions. In vertebrates, desmosomes are created by CAMs called cadherins, which connect with one another across the intercellular space.

21

Figure 2–10. Types of cell–cell junctions. (a) A tight junction in intestinal epithelial cells, (b) an anchoring junction in intestinal epithelial cells, and (c) a gap junction in heart muscle. After [Silverthorn, 2007].

22 2.3.2.2. Cell–Matrix Junctions

There are two types of cell–matrix anchoring junctions: focal adhesion and hemidesmosomes. Cell adhesion molecules called integrins are used in cell–matrix junctions. In focal adhesions, the integrin ties intracellular actin fibers to different extracellular matrix proteins such as insoluble cellular fibronectin as shown in figure 2–

11(C). Fibronectins, synthesized and deposited by the cells, play a very important role in the adhesion of many cell types and the formation of extracellular matrix [Garcia et al.,

1997; Ruoslahti, 1984]. Cell adhesion strength also increases linearly with adsorbed fibronectin surface density [Garcia et al., 1997]. In hemidesmosomes, integrins anchors intermediate fibers (keratin filaments) of the cytoskeleton to fibrous matrix proteins such as laminin as shown in figure 2–12.

Integrins also play an important role in cell signaling. The external attachments the integral molecules make can activate intracellular signaling cascades [Albert et al., 2004;

Craig, 1996; Juliano, 1993; Silverthorn, 2007]. These cascades regulate many biological processes such as cell adhesion, cell migration, cell growth, cell differentiation, cellular aggregation, and gene expressions [Albert et al., 2004; Bernstein, 1994; Craig, 1996;

Juliano, 1993; LaFlamme, 2008; Ruoslahti, 1987].

23

Figure 2–11. (A) Diagram and (B) electron micrograph of a fibronectin molecule. It has a collagen–binding site and a cell–binding site (or an integrin–binding site). (C) Focal adhesion: integrin is anchored inside the cell to the cytoskeleton and externally via fibronectin to the extracellular matrix. After [Alberts et al., 2004].

Figure 2–12. Schematic of a hemidesmosome junction. Mediated by integrins, it anchors the keratin filaments in an intestinal epithelial cell to basal lamina. After [Alberts et al., 2004].

24 2.4. Chapter Appendix

2.4.1. Diffusion across the Cell Membrane

2.4.1.1. Simple Diffusion

For small and non–polar molecules, the rate of simple diffusion across the cell membrane can easily be expressed by a modified Fick’s Law of diffusion (equation 2–1) and is directly proportional to the concentration gradient of the diffusing molecules across the cell membrane [Silverthorn, 2007].

⎛ Membrane ⎞ ⎛ Surface ⎞ ⎛ Concentration ⎞ ⎜ ⎟ × ⎜ ⎟ × ⎜ ⎟ ⎝ permeability ⎠ ⎝ area ⎠ ⎝ gradient ⎠ Rate of Diffusion ∝ (2–1) ⎛ Membrane ⎞ ⎜ thickness ⎟ ⎝ ⎠

Figure 2–13 shows the linear relationship between the rate of diffusion and the concentration gradient for simple diffusion of small and non–polar molecules across the cell membrane. The slope depends on the membrane permeability of diffusing molecules, the thickness, and the surface area of cell membrane.

Figure 2–13. The function of the rate of diffusion of small and non–polar molecules across the cell membrane. The relationship between the rate of diffusion and the concentration gradient is linear.

25 2.4.1.2. Facilitated Diffusion

Facilitated diffusion involves a limited number of carrier proteins in the cell membrane. Therefore, there will be a maximum rate of diffusion when all the carrier proteins are saturated. The modified Fick’s law cannot be used to describe the rate of diffusion in this case. The rate of diffusion will increase with increasing solute concentration but asymptotically approach to the maximum saturation rate that is limited by the number of transport or carrier proteins in the cell membrane. The affinity of transport or carrier proteins to diffusing molecules determines how quickly they become saturated.

The kinetic of facilitated diffusion is similar to that of simple enzyme–catalyzed chemical reaction. For instance, the kinetic of unidirectional transport of glucose from the outside of a cell inward via the glucose transporter 1 (GLUT1) can be described as follows:

Km V Glucose + GLUT1 ⎯⎯→ Glucose−GLUT1 ⎯⎯max⎯→ Glucose + GLUT1 out ←⎯⎯ in

where “Glucose–GLUT1” represents the GLUT1 transport protein in the cell membrane with a bound glucose. By a similar derivation used to arrive at the Michaelis–Menten equation, the initial diffusion rate, ν, for glucose molecules into the cell, which is

facilitated by the GLUT1 transporters, can be approximated as follows [Lodish et al.,

2004]:

V max ν = K (2–2) 1+ m C

In equation 2–2, Vmax is the rate of transport or diffusion when all molecules of

GLUT1 in the cell membrane are saturated. The Km approximates the affinity of GLUT1 to the glucose. The lower the Km, the more tightly the glucose binds to the GLUT1, and

26 the greater the transport or diffusion rate at a fixed concentration of glucose outside the cell membrane. At low concentrations of glucose, the glucose molecules will pass through the carrier proteins in a way similar to that of simple diffusion. However, at high solute concentrations of glucose, all the transport proteins are occupied with the glucose molecules and further increasing the glucose concentration outside of the cell membrane will not change the rate of diffusion. Figure 2–14 shows the relationship between the rate of diffusion and the concentration gradient for facilitated diffusion.

Figure 2–14. The function of rate of facilitated diffusion of molecules across the cell membrane. The relationship between the rate of diffusion and the concentration gradient is not linear. The rate of diffusion asymptotically approaches to the maximum

diffusion rate limited by the number of transport proteins in the cell membrane. The Km determines how quickly the transport proteins become saturated.

2.4.2. Active Transport across the Cell Membrane

Active transport is the pumping of molecule against the concentration gradient across the cell membrane by a trans–membrane protein pumps. The proteins are highly specific for the molecule to be transported. They are also ATPase enzymes as they catalyze the splitting of ATP to ADP + phosphate (Pi) and use the energy released from the reaction to change their conformation as well as pumping the molecule. In general, the rate of active transport increases with increasing of ATP molecules and of the

27 number of protein pumps. Unlike simple and facilitated diffusion, the active transport might still have a high transport rate even when there is no concentration difference of molecules across the cell membrane (i.e, being independent of concentration gradient).

Active transport only stops when cellular respiration stops. The Na+–K+ pump and

H+/Sucrose pump are examples of trans–membrane protein pumps.

2.4.3. Nernst Equation

For the cell membrane that is permeable to only one ion, the equilibrium membrane

potential (Reversal Potential) that opposes the concentration gradient of the ion across

the cell membrane can be described by the Nernst Equation♠:

⎡ion⎤ RT ⎣⎢ ⎦⎥out E = ln (2–3) ion(in mV ) zF ⎡ion⎤ ⎣⎢ ⎦⎥in

where R, T, F, and z are the universal gas constant, absolute temperature in Kelvin, the

Faraday constant or number of coulombs per mole of electrons, and the electrical change on the ion respectively. [ion]out and [ion]in represent the concentrations of ions outside

and inside of the cell respectively. The Nernst Equation can be used to calculate the

equilibrium potentials for any biologically relevant ion. The estimated ion

concentrations and equilibrium potentials for Na+, K+, Cl–, and Ca2+ in mammalian cells are given in table 2–1.

♠ A complete derivation of Nernst equation can be found in [Malmivo, 1995].

28 Eion @ 37 °C Ion Extracellular Fluid Intracellular Fluid (Referenced from extracellular fluid)

K+ 5 mM (normal range: 3.5–5) 150 mM –90 mV

Na+ 145 mM (normal range: 135–145) 15 mM +60 mV

Cl– 108 mM (normal range: 100–108) 10 mM (range: 5–15) –63 mV

Ca2+ 1 mM 0.0001 mM +122 mV

Table 2–1. The estimated ion concentrations and their respective equilibrium potentials in mammalian cells. After [Silverthorn, 2007].

2.4.4. Goldman–Hodgkin–Katz (GHK) Equation

The Nernst equation may be used to determine the equilibrium potential for a single ion. However, The actual cell membrane is permeable to multiple ions. The equilibrium or resting membrane potential depends on the relative permeability of ions, and thus,

Goldman–Hodgkin–Katz (GHK) equation must be used [Malmivo, 1995]. For n

monovalent positive ionic species and m negative ionic species, the equilibrium potential

can be described as♥:

⎛ n + m − ⎞ ⎜ P ⎡C ⎤ + P ⎡ A ⎤ ⎟ ∑ i=1 C + i ∑ j=1 A− j ⎟ RT ⎜ i ⎣⎢ ⎦⎥out j ⎣⎢ ⎦⎥in V ln⎜ ⎟ m = ⎜ n m ⎟ (2–4) F ⎜ P ⎡C + ⎤ + P ⎡ A− ⎤ ⎟ ⎜∑ i=1 C + ⎢ i ⎥ ∑ j=1 A− ⎢ j ⎥ ⎟ ⎝ i ⎣ ⎦in j ⎣ ⎦out ⎠

The GHK equation includes respective membrane permeability values for monovalent ions since each ion’s contribution to the membrane potential is proportional to its membrane permeability. If the membrane is not permeable to an ion, the permeability term for that ion will be zero, and the related term drops out of the equation. Also, divalent ions are not included in the GHK equation since resting cells are

♥ A complete derivation of GHK equation can be found in [Malmivo, 1995].

29 not usually permeable to divalent ions. For mammalian cells at rest, it is usually assumed that Na+, K+, and Cl– are the three major ions that influence membrane potential. Based on the equation 2–4, the GHK Equation for cells that are permeable to

Na+, K+, and Cl– is:

⎛ ⎡ + ⎤ ⎡ + ⎤ ⎡ − ⎤ ⎞ ⎜P + K + P + Na + P − Cl ⎟ RT ⎜ K ⎣⎢ ⎦⎥out Na ⎣⎢ ⎦⎥out Cl ⎣⎢ ⎦⎥in ⎟ V = ln⎜ ⎟ (2–5) m F ⎜ ⎡ + ⎤ ⎡ + ⎤ ⎡ − ⎤ ⎟ ⎜ P + K + P + Na + P − Cl ⎟ ⎝ K ⎣⎢ ⎦⎥in Na ⎣⎢ ⎦⎥in Cl ⎣⎢ ⎦⎥out ⎠

If divalent ions (e.g., Ca2+, Mg2+, etc.) are to be considered, the GHK equation must

be modified [Chang, 1983; Egebjerg, 1993; Lewis, 1984; Mayer, 1987].

2.4.5. Osmotic Pressure

Osmosis is the phenomenon of solvent flowing through a semipermeable membrane

to equalize the solute concentrations on both sides of the membrane. Osmotic pressure,

in general, is a colligative property of a solution equal to the pressure that, when applied

to the solution, just stops osmosis [Gammon, 1999]. In other words, it is the pressure

that must be applied to a solution to prevent inward flow of solvent molecule across the

semipermeable membrane. The osmotic pressure of a solution can be estimated by:

Π = iMRT (2–6)

where i, M, R, and T are the dimensionless van ’t Hoff factor (for NaCl in water, i is 1 since it dissociates completely to Na+ and Cl–), the molarity of the solution, the universal

gas constant, and the absolute temperature in Kelvin.

30 References

Alberts M. B., Bray, D., Hopkin, K., Johnson, A., Lewis, J., Raff, M., Roberts, K., and Walter, P., 2004. Essential Cell Biology. 2nd Edition, Taylor & Francis Group, New York.

Alberts M. B., Stein, W. D., Laskey, R. A., Bernfield, M. R., Staehelin, L. A., Slack, J. M. W., 2009. Cell. Encyclopædia Britannica Online. 20 October 2009. .

Bernstein, L. R., and Loitta, L. A., 1994. Molecular Mediators of Interactions with Extracellular Matrix Components in Metastasis and Angiogenesis. Current Opinion in Oncology. 6, 106– 113.

Chang, D. C., 1983. Dependence of Cellular Potential on Ionic Concentrations: Data Supporting a Modification of the Constant Field Equation. Biophysical Journal. 43, 149–156.

Craig, S. W., and Johnson, R. P., 1996. Assembly of Focal Adhesions: Progress, Paradigms, and Portents. Current Opinion in Cell Biology. 8, 74–85.

Egebjerg, J., and Heinemann, S. F., 1993. Ca2+ Permeability of Unedited and Edited Versions of the Kainate Selective Glutamate Receptor GluR6. Neurobiology. 90, 755–759.

Gammon, E., 1999. General Chemistry, 6th Edition, Houghton Mifflin Company, Boston.

Garcia, A. J., Ducheyne, P., and Boettiger, D., 1997. Cell Adhesion Strength Increases Linearly with Adsorbed Fibronectin Surface Density. Tissue Engineering. 3 (2), 197–206.

Juliano, R. L., and Haskill, S., 1993. Signal Transduction from the Extracellular Matrix. The Journal of Cell Biology. 10 (3), 577–585.

LaFlamme, S. E. and Kowalczyk, A., 2008. Cell Junctions: Adhesion, Development, and Disease. 1st Edition, Wiley–VCH Verlag GmbH & Co. KGaA, Weinheim.

Lewis, C. A., 1984. Divalent Cation Effects on Acetylcholine – Activated Channels at the Frog Neuromuscular Junction. Cellular and Molecular Neurobiology. 4 (3), 273–284.

Lodish, H., Berk, A., Matsudaira, P., Kaiser, C. A., Krieger, M., Scott, M. P., Zipursky, S. L., Darnell, J., 2004. Molecular Cell Biology. 6th Edition, W. H. Freeman and Company, New York.

Malmivuo, J. and Plonsey, R., 1995. Bioelectromagnetism: Principles and Applications of Bioelectric and Biomagnetic Fields. Oxford University Press, Oxford.

31 Mayer, M. L., and WestBrook, G. L., 1987. Permeation and Block of N-Methyl-D-Aspartic Acid Receptor Channels by Divalent Cations in Mouse Cultured Central Neurones. 394, 501–527.

Ruoslahti, E., 1984. Fibronectin in Cell Adhesion and Invasion. Cancer Metastasis Reviews. 3, 43–51.

Ruoslahti, E., and Pierschbacher, M. D., 1987. New Perspectives in Cell Adhesion: RGD and Integrins. Science. 238 (4826), 491–497.

Silverthorn, D. U., 2007. Human Physiology: An Integrated Approach. 4th Edition, Pearson Education Inc., San Francisco.

32 Chapter 3 BIOFUNCTIONALIZED INTERFACES

3.1. Self–Assembled Monolayers

Biological chips and devices used in biosensing applications must possess some degree of biocompatibility that can be achieved by constructing the device with materials that are not harmful to the living cells and/or creating an interface between the living cells and the device itself. One of the widely used strategies is to coat the surfaces of the device with biocompatible thin films such as Self–Assembled Monolayers (SAMs).

Further surface derivatization of SAMs are sometimes needed in some biosensing applications. Not only do such coatings and functionalizations provide biocompatible interfaces, they can also improve the robustness of biological devices in harsh environments.

Self–assembled monolayers (SAMs) are organized layers of molecules that are formed spontaneously by the chemisorption of amphiphilic surfactants’ head groups to a substrate followed by a slow two–dimensional organization of alkane chains [Love et al.,

2005; Schreiber, 2004; Schwartz, 2001; Ulman, 1996]. The surfactant or the constituent of a SAM molecule consists of three significant parts: a head group, a spacer or alkane chain, and a terminal or functional group as shown in figure 3–1. Generally, the head group is hydrophilic and is designed to have a favorable and specific interaction with the substrate of interest [Schwartz, 2001]. Most commonly used head groups are thiols, disulphides, amines, organic acids, and silanes [Ulman, 1996]. Types of substrates range from planar surfaces (glass or silicon slabs supporting thin films of metal, metal foils, single crystals, oxides, etc.) to highly curved nanostructures (colloids, nanocrystals, nanorods, etc.) [Love et al., 2005]. Terminal groups are functionalized groups such as methyl groups (–CH3), carboxyl groups (–COOH), or hydroxyl groups (–OH) depending

33 on the applications. The terminal or functional group determines the surface properties of SAM–coated substrates. The alkane chain provides the thickness of SAM (typically 1–

3 nm) and acts as a physical barrier. Self–assembled monolayers are usually prepared

from solution or from vapor. The surfactants adsorb spontaneously as they lower the

surface energy of the substrate and the adsorption is stable due to the strong

chemisorption of the head groups [Love et al., 2005]. The chemisorbed bonds are

stronger than physisorbed bonds and, thus, resulting stable monolayers.

At least two major processes are involved in this adsorption process [Schwartz,

2001]: (i) the solution phase transport of surfactants molecules to the solid–liquid interface, which can involve some combination of diffusive (the random Brownian motion of individual surfactants in the fluid) and convective transport; (ii) The

adsorption of surfactants head groups on the substrate with some adsorption rate that is

related to a “sticking” probability. The overall adsorption dynamics of head groups may

be diffusion–controlled, adsorption rate controlled, or in an intermediate mixed–kinetic regime. After the chemisorption of the head groups, the monolayer packs tightly due to van der Waals interactions or forces between alkane chains, thereby reducing its own free energy [Love et al., 2005]. This two–dimensional molecular organization is a key ingredient for SAM stability and function. In the SAM formation, it is generally accepted that there is a stepwise evolution of the molecular order as adsorption progresses and the surface coverage increases. Possible states in the SAM formation include: (i) a low–density “vapor” phase in which isolated, mobile adsorbates or surfactant molecules are randomly deposited on the surface; (ii) an intermediate– density phase that could involve conformationally disordered molecules or ones lying flat on the surface; and (iii) a final, high–density “solid” phase in which the molecules are conformationally ordered, close–packed, and standing approximately normal to the surface plane with a uniform polar tilt angle of approximately ≤ 30˚ [Schwartz, 2001].

34

(A)

(B)

Figure 3–1. (A) Typical SAM–forming molecule (decanthiol or CH3(CH2)9HS) with the angular degrees of freedom for an all–trans chain, tilt angle (θt), tilt direction (χt), and twist (ψ). Adapted from [Schreiber, 2000; Schreiber, 2003]. (B) Schematic diagram of an ideal, single–crystalline SAM of hexadecanethiolates (CH3(CH2)15SH) supported on a gold surface with a (111) texture. Adapted from [Love et al., 2005].

35 Since self–assembled monolayer films are thin and homogeneous, they have been used as model surfaces [Schwartz, 2001]. Much of the interest in SAMs lies in their potential as inexpensive and versatile surface coatings for applications including control of wetting and adhesion, chemical and corrosion resistance, biocompatibility, molecular recognition for sensor applications, and etc. Although a very wide range of SAM systems is available, two of the most popular and widely used systems of SAMs are gold– alkylthiolate monolayers and alkylsilane monolayers.

3.1.1. Gold–Alkylthiolate Monolayers

Organosulfur compounds like dialkyl disulfides (X(CH2)mS–S(CH2)nX), or alkanethiols (HS(CH2)nX) form self–assembled monolayers (SAMs) on gold surface

[Bain et al., 1989; Biebuyck et al., 1994; Love et al., 2005; Nuzzo, 1983; Ulman, 1996], where n and m are the number of methylene units and X represents the end terminal group of the alkyl chain. Figure 3–1(B) shows schematic diagram of an ideal, single–

crystalline SAM of hexadecanethiolates (CH3(CH2)15SH) supported on a gold surface with a (111) texture.

The most common method for preparing gold–alkylthiolate monolayers on gold is immersion of a clean gold–film–coated substrate into a dilute (~1–10 mM) ethanolic solution of thiols for ~12–18 hours at room temperature [Bain et al., 1989; Love et al.,

2005]. Ethanol is widely used as a solvent for the following five reasons: (1) it solvates a variety of alkanethiols with varying degrees of polar character and chain length; (2) it is inexpensive; (3) it is available in high purity; (4) it has low toxicity; and (5) it has been

found to yield consistently highly ordered self–assembled monolayers on gold substrates

[Bain et al., 1989; Love et al., 2005]. Gold (Au) is widely used as a substrate for self–

assembled monolayers due to the following characteristics [Love et al., 2005]:

(1) It is easy to obtain, both as a thin film and as a colloid. Thin films of gold can be

prepared by sputtering, physical vapor deposition (PVD), or electro–deposition

and they can easily be patterned by standard microfabrication techniques.

36 (3) Gold is a reasonably inert metal: it does not oxidize at temperatures below its

melting point; it does not react with atmospheric O2; it does not react with most

chemicals. These properties make it possible to handle and manipulate

substrates under atmospheric conditions instead of under ultra high vacuum

(UHV) providing a great practical convenience for conducting experiments in

dirty conditions (for example, outside of a clean room environment).

(4) Gold binds thiols with a high affinity and it does not undergo any unusual

reactions with them [Nuzzo, 1983].

(5) Thin films of gold are common substrates used for a number of existing

spectroscopies and analytical techniques such as surface Plasmon resonance

(SPR) spectroscopy, quartz crystal microbalances (QCM), reflection absorption

infrared spectroscopy (RAIRS), and ellipsometry.

(6) Gold is biocompatible with cells, i.e., cells can adhere and function on gold

surfaces without evidence of toxicity.

The possible mechanism of the formation SAMs from disulfides is an oxidative addition of the S–S bond to the gold surface [Ulman, 1996]. This chemisorption reaction of dialkyl disulfides on gold can be describe as:

0 – + 0 RS–SR + Au n ⇒ RS Au ⋅ Au n

In the alkanethiol case, the reaction of gold and thiol group is considered to be an oxidative addition of the S–H bond to the gold surface, followed by a possible reductive

elimination of the hydrogen [Ulman, 1996]. The reaction can be describe as follows:

0 – + 0 R–S–H + Au n ⇒ R–S Au ⋅ Au n + ½ H2 (?)

The bonding of the thiolate group to the gold surface is very strong (homolytic bond strength is approximately 40 kcal mol-1 [Dubois, 1992]).

37 3.1.2. Alkylsilane Monolayer

Alkylsilane monolayers are the second most popular system. However, the relevant mechanisms differ substantially from those of thiols on gold. One of the typical examples is the irreversibly binding of a trichlorosilane or a similar headgroup (e.g., alkoxysilane or aminosilane) to a hydroxylated surface (for instance, the oxide surface of silicon) [Sagiv, 1980]. This different chemisorption process of n–Octadecyl– trichlorosilane (C18H37SiCl3) (or OTS) on the glass surface is shown in figure 3–2. In this

process, first, the side–groups (chlorines or other) are split off, and a strongly bound

network of the head groups is generated. Next, adsorbates are immobilized on the

surface due to the formation of hydrolytic bonds with hydroxyl (–OH) surface groups.

Finally, the self–assembly occurs due to the formation of polysiloxane that is connected

to surface silanol groups (–SiOH) via Si–O–Si bonds. The growth of alkylsilane–based

SAMs is very unique since it involves an irreversible covalent cross–linking step that is

essential for the desirable properties of this class of SAMs, including their chemical and

mechanical robustness and the stability of the final monolayer on a variety of substrates.

Alkylsilane monolayers can also be prepared from the solution. However, high–

quality SAMs are not simple to produce since the adsorption process is sensitive to

temperature, pH, and water content in the solution [Kessel, 1991; Schreiber, 2000;

Schwartz, 2001]. Carefully–controlled experimental conditions and protocols are also

required in the preparation of solution that contains active components [Huang et al.,

1994; Kessel, 1991; Schrieber, 2000; Xiao et al., 1995] and the surface of the substrate

requires pretreatment with either plasma [Parker et al., 1989] or chemicals [Carson,

1989].

38

Figure 3–2. Chemisorption of n–Octadecyltrichlorosilane (OTS) on glass surface. After [Sagiv, 1980].

Since substrates used in the formation of alkylsilane monolayers are amorphous, the packing and ordering of alkyl chains in the monolayer are determined by the underlying structure of the surface polysiloxane chain [Schreiber, 2001]. From a physical perspective, alkylsilane monolayers have less well–defined structures compared to gold– alkylthiolate monolayers (no long–range order in crystalline sense), and the comparatively low mobility of the monolayer molecules on the surface due to the strong and localized binding of the Si–O–Si network (This also implies that the structure cannot easily be changed by annealing.) [Schreiber, 2000; Ulman, 1996]. However, alkylsilane monolayers are considered to be more robust [Schreiber, 2001].

39 3.2. SAM–based Biofunctionalized Systems and Surfaces

One of the most exciting properties of self–assembled monolayers is that they can be modified in order to generate surfaces with biologically relevant functionalities. Several strategies have been developed to generate biofunctionalized surfaces. One of the most widely used strategies is to exploit the concept of lock–key recognition in protein binding. In this strategy, the mixed or patterned SAM, containing two or more constituent molecules, is terminated with specific functional groups or ligands to which receptors can be attached, serving as one part of a lock–key pair to specifically bind biomolecules of interest (as illustrated in figure 3–3). These types of SAMs are employed as: (i) model systems to study fundamental aspects of the interactions of surfaces with biomolecules [Jung et al., 2000; Lahiri et al., 1999], (ii) interfaces for molecular recognition in biosensing applications [Frederix et al., 2003; Subramanian et al., 2006; Jung et al., 1999], and (iii) model surfaces to study cell–surface interaction

[Houseman and Mrksich, 2001; Kato and Mrksich, 2004; Roberts et al., 1998].

Another strategy is to exploit the end group chemistry of SAM–forming molecules or constituents as shown in figure 3–4. In this strategy, the mixed or patterned SAM is formed by two or more SAM constituent molecules that are terminated with different functional or end groups. For instance, a mixed SAM formed by two SAM constituents, which presents structurally well–defined hydrophobic end groups at surfaces that are otherwise terminated with hydrophilic background group, is employed as a model surface to study physisorption of protein molecules onto the surface [Ostuni et al.,

2003]. In addition to mixed SAMs, substrates that are biofunctionalized with homogeneous SAM are also used as model surfaces to study the surface–mediated cellular activities as well as the biocompatibility of materials [Barbosa et al., 2003;

Barbosa et al., 2004].

40

Figure 3–3. Schematic illustration showing the reversible adsorption of a receptor to a mixed SAM presenting a ligand. The mixed SAM comprises of two different constituent molecules. One constituent is terminated with either a biological ligand (or a reactive site for linking to a biological ligand) and the other constituent is terminated with a functional group that resists the nonspecific adsorption of biomolecules and biological cells. The fraction of ligands on the surface is related to the mole fraction of the SAM constituents in the solution used to form the mixed SAM. Adapted from [Lahiri et al., 1999; Love et al., 2005; Mrksich et al., 1995].

Figure 3–4. Schematic illustration showing the physisorption of a protein to a patterned SAM formed by two different constituents. One constituent of the patterned SAM is terminated with a hydrophilic functional group and the other constituent is terminated with a hydrophobic functional group. Adapted from [Love et al., 2005; Ostuni et al., 2003].

41 3.2.1. Common Methods for Preparing Mixed and Patterned SAMs

3.2.1.1. Coadsorption from Solutions of Mixed SAM Constituents

Mixed self–assembled monolayer can be formed by either coadsorbing mixed alkanethiols from the solution directly onto the metal surface (especially gold) [Chapman et al., 2000; Heeg et al., 1999; Laibinis et al., 1991; Yang et al., 1997] or coadsorbing mixed alkylsilanes directly onto silicon surfaces [Heise and Menzel, 1997; Heise et al.,

1998]. In general, the solution that is used to form the mixed SAM usually comprises of two or more SAM constituents. One of the SAM constituents will be terminated with specific functional groups or ligands and the rest are terminated with end groups that present the background. The fraction of functional groups or ligands on the surface after the SAM formation is related to the mole fraction of the SAM constituents in the solution

[Love et al., 2005].

3.2.1.2. Microcontact Printing (µCP)

The Microcontact printing (µCP) is a method for patterning SAMs on surfaces that is operationally analogous to printing ink with a rubber stamp on the paper [Kumar et al.,

1994; Mrksich and Whitesides, 1995; Love et al., 2005; Smith et al., 2004; Xia and

Whitesides, 1998]. SAMs form in the regions of contact between a topographically patterned stamp (usually made of polydimethylsiloxane (PDMS)), which is wetted with reactive chemical ‘ink’ consisting of molecules that form SAMs, and the bare surface of a metal, metal oxide, or semiconductor. The stamp is cast from the master template usually fabricated by standard microfabrication techniques that involves photolithography and reactive ion etching (RIE).

When forming patterned SAMs on the substrate, the inked or wetted stamp is brought in contact with the surface and, then, left for a few seconds (5–10 sec) before it is removed. The molecules are transferred to the surface only at those regions where the

42 stamp contacts the surface. Due to the autophobicity of the SAM to its ink liquid, the lateral spreading across the surface is prevented [Biebuyck and Whitesides, 1994;

Mrksich and Whitesides, 1995; Xia and Whitesides, 1998; Xia et al., 2001]. The lateral dimensions of the SAMs formed depend on the dimensions of relief features on the stamp (the lateral dimension as small as ~50 nm can be generated). After the stamp is removed, another SAM can be formed in the remaining bare regions of the surface by either immersion of the substrate into a different solution for a few minutes (~1–10 min) or application of a second stamp wetted with a different “ink”. The latter approach requires a good alignment between the second stamp and the substrate, which can easily

be achieved via µCP instruments. Multiple copies of stamps can be cast from the master template and each stamp can be used hundreds of times without any loss of quality of the printed patterns. Hence, the µCP can produce patterned SAMs at relatively low cost.

The general procedure for patterning SAM using µCP is summarized in figure 3–5.

43

Figure 3–5. Procedure for patterning thiol-based SAMs using µCP. Photolithography or other standard microfabrication techniques generates a master template containing features of the pattern to be reproduced (a). A polydimethylsiloxane (PDMS) prepolymer is poured onto the master pattern, allowed to cure (b), and then peeled away from the master (c). The stamp is wetted with alkanethiol solution or “ink” (d), and is used to transfer the alkanethiol to the surface (e); this transfer forms a patterned SAM (f). Exposing the gold substrate to a solution that contains different alkanethiol derivatizes the bare regions (g). Adapted from [Mrksich and Whitesides, 1995; Xia and Whitesides, 1998].

44 3.2.1.3. Substrates Having Surface Patterns Made of Different Materials

Patterned SAMs can be formed on a substrate that presents two–dimensional surface

patterns formed by different materials. This type of approach is usually employed to

exploit the existing two–dimensional surface patterns in microfabricated devices for

biosensing applications [Arya et al., 2009; Asphahani et al., 2008; Veiseh et al., 2001;

Veiseh et al., 2002]. As existing surface patterns are formed by different materials,

different types SAM systems can be formed successively by directly adsorbing the

different SAM constituents onto the surface in different solutions.

3.3. Removal of Self–Assembled Monolayers

SAM–coated devices and substrates in biosensing applications are usually recycled

and, thus, the removal of self–assembled monolayers from surfaces is usually required.

Most common methods for SAM removal are:

(1) Wet chemical etching of SAM with piranha solution (concentrated H2SO4 + 30%

H2O2) [Guo et al., 1994] or Nano–strip™ solution.

(2) Desorption of SAM by electrochemical potential cycling in aqueous solution

[Canaria et al., 2006; Guo et al., 1994; Jiang et al., 2003]. This method is

usually employed for the removal of thiolate–based SAM formed on gold or

metal surfaces.

(3) Photolysis or photo–oxidation with UV light [Huang and Hemminger, 1993;

Hutt and Leggett, 1996; Lewis and Tarlov, 1995; Tarlov et al., 1993]: Oxidation

of an alkanethiolate layer by UV light results in sulfite and sulfate species (RSOx)

that are weakly adsorbed and can easily be rinsed from the surface with polar

solvents.

(4) Ozonolysis [Norrod and Rowlen, 1998; Zhang et al., 1998; Zhang et al., 1999]: O3

(ozone) oxidizes the sulfur head–groups of alkanethiol SAMs, similar to

oxidation process in photolysis, to generate polar solvent strippable products.

45 (5) Thermal desorption [Delamarche et al., 1994; Liu et al., 2002]: Thiolate–based

SAMs usually desorb from the surface at temperatures above 100 ˚C (Note:

Long hours of thermal treatment are usually required).

(6) Plasma cleaning or stripping [Raiber et al., 2005]: Oxygen or hydrogen plasma

is usually used. During the cleaning process, chemically reactive plasma gases

react with the SAM and form volatile products that are stripped and swept away

by the continuous gas flow inside the chamber of plasma cleaning equipments.

No additional rinsing with polar solvents is required.

Of these methods, the first two methods are wet chemical processes, which could

contaminate the substrate by the reagents used and, thus, are very unfavorable in some

applications. The most undesirable disadvantage of the piranha treatment or Nano–

strip™ treatment is that it induces recrystallization of the gold layer [Twardowski and

Nuzzo, 2002] and may even cause delamination of the gold layer by etching away the

frequently used adhesion layers. Although the electrochemical potential cycling is a wet

chemical process, it has some advantages in cellular biosensing where immobilized cells

on the SAM–based bio–functionalized electrodes can be released noninvasively [Jiang et

al., 2003]. It can be noticed that the UV/O3 oxidation often requires an additional wet cleaning step with polar solvents and thermal desorption is not suitable for devices and substrates that cannot be exposed to high temperature environments for long period of time. Therefore, cleaning with hydrogen or oxygen plasma is more favorable in some cases as it does not require additional wet rinsing step and is very simple as well as less harmful to carry out in significantly shorter time period compared to other methods.

46 References

Arya, S. K., Solanki, P. R., Datta, M., and Malhotra, B. D., 2009. Recent Advances in Self– Assembled Monolayers Based Biomolecular Electronic Devices. Biosensors and Bioelectronics. 24, 2810–2817.

Asphahani, F., Thein, M., Veiseh, O., Edmondson, D., Kosai, R., Veiseh, M., Xu, J., and Zhang, M., 2008. Influence of Cell Adhesion and Spreading on Impedance Characteristics of Cell–Based Sensors. Biosensors and Bioelectronics. 23, 1307–1313.

Bain, C. D., Troughton, E. B., Tao, Y., Evall, J., Whitesides, G. M., and Nuzzo, R. G., 1989. Formation of Monolayer Films by the Spontaneous Assembly of Organic Thiols from Solution onto Gold. Journal of the American Chemical Society. 111, 321–335.

Barbosa, J. N., Barbosa, M. A., and Aguas, A. P., 2003. Adhesion of Human Leukocytes to Biomaterials: An in vitro Study Using Alkanethiolate Monolayers with Different Chemically Functionalized Surfaces. Journal of Biomedical Materials Research. 65A (4), 429–434.

Barbosa, J. N., Barbosa, M. A., and Aguas, A. P., 2004. Inflammatory Responses and Cell Adhesion to Self–Assembled Monolayers of Alkanethiolates on Gold. Biomaterials. 25, 2557– 2563.

Biebuyck, H. A., and Whitesides, G. M., 1994. Autophobic Pinning of Drops of Alkanethiols on Gold. Langmuir. 10, 4581–4587.

Biebuyck, H. A., Bain, C. D., and Whitesides, G. M., 1994. Comparison of Organic Monolayers on Polycrystalline Gold Spontaneously Assembled from Solutions Containing Dialkyl Difulfides or Alkanethiols. Langmuir. 10, 1825–1831.

Canaria, C. A., So, J., Maloney, J. R., Yu, C. J., Smith, J. O., Roukes, M. L., Fraser, S. E., and Lansford, R., 2006. Formation and Removal of Alkylthiolate Self–Assembled Monolayers on Gold in Aqueous Solutions. Lab on a Chip. 6, 289–295.

Carson, G., and Granick, S., 1989. Self–Assembly of Octadecyltrichlorosilane Films on Mica. Journal of Applied Polymer Science. 37, 2767–2772.

Chapman, R. G., Ostuni, E., Yan, L., and Whitesides, G. M., 2000. Preparation of Mixed Self– Assembled Monolayers (SAMs) That Resist Adsorption of Proteins Using the Reaction of Amines with a SAM That Presents Interchain Carboxylic Anhydride Groups. Langmuir. 16, 6927–6936.

Delamarche, E., Michel, B., Kang, H., and Gerber Ch., 1994. Thermal Stability of Self–Assembled Monolayers. Langmuir. 10, 4103–4108.

47 Dubois, L. H., and Nuzzo, R. G., 1992. Synthesis, Structure, and Properties of Model Organic Surfaces. Annual Review of Physical Chemistry. 43, 437–463.

Frederiz, F., Bonroy, K., Laureyn, W., reekmans, G., Campitelli, A., Dehaen, W., and Maes, G., 2003. Enhanced Performance of an Affinity Biosensor Interface Based on Mixed Self– Assembled Monolayers of Thiols on Gold. Langmuir. 19, 4351–4357.

Guo, L.-H., Facci, J. S., McLendon, G., and Mosher, R., 1994. Effect of Gold Topography and Surface Pretreatment on the Self–Assembly of Alkanethiol Monolayers. Langmuir. 10, 4588– 4593.

Heeg, J., Schubert, U., and Küchenmeister, F., 1999. Mixed Self–Assembled Monolayers of Terminally Functionalized Thiols at Gold Surfaces Characterized by Angle Resolved X–ray Photoelectron Spectrocopy (ARXPS) Studies. Fresenius’ Journal of Analytical Chemistry. 365, 272–276.

Heise, A., and Menzel, H., 1997. Grafting of Polypeptides on Solid Substrates by Initiation of N– Carboxyanhydride Polymerization by Amino–Terminated Self–Assembled Monolayers. Langmuir. 13, 723–728.

Heise, A., Stamm, M., Rauscher, M., Duschner, H., and Menzel, H., 1998. Mixed Silane Self– Assembled Monolayers and Their in situ Modification. Thin Solid Films. 327–329, 199–203.

Houseman, B. T., and Mrksich, M., 2001. The Microenvironment of Immobilized Arg-Gly-Asp Peptides Is an Important Determinant of Cell Adhesion. Biomaterials. 22, 943–955.

Haung J., and Hemminger, J. C., 1993. Photooxidation of Thiols in Self–Assembled Monolayers on Gold. Journal of the American Chemical Society. 115, 3342–3343.

Huang, M., Schneiderman, E. D., Novotny, M. V., Fatunmbi, H. O., and Wirth, M. J., 1994. Self– Assembled Alkylsilane Monolayers for the Preparation of Stable and Efficient Coatings in Capillary Electrophoresis. Journal of Microcolumn Separations. 6, 571–576.

Hutt D. A., and Leggett, G. J., 1996. Influence of Adsorbate Ordering on Rates of UV Photooxidation of Self–Assembled Monolayers. Journal of Physical Chemistry. 100, 6657– 6662.

Jiang, X., Ferrigni, R., Mrksich, M., and Whitesides, G. M., 2003. Electrochemical Desorption of Self–Assembled Monolayers Noninvasively Releases Patterned Cells from Geometrical Confinements. Journal of the American Chemical Society. 125, 2366–2367.

Jung, L. S., Nelson, K. E., Campbell, C. T., Stayton, P. S., Yee, S. S., Luna, V. P., and Lopez, G. P., 1999. Surface Plasmon Resonance Measurement of Binding and Dissociation of Wild–Type and Mutant Streptavidin on Mixed Biotin–Containing Alkylthiolate Monolayers. Sensors and Actuators B. 54, 137–144.

48

Jung, L. S., Nelson, K. E., Stayton, P. S., and Campbell, C. T., 2000. Binding and Dissociation Kinetics of Wild–Type and Mutant Streptavidins on Mixed Biotin–Containing Alkylthiolate Monolayers. Langmuir. 16, 9421–9432.

Kato, M., and Mrksich M., 2004. Using Model Substrates to Study the Dependence of Focal Adhesion Formation on the Affinity of Integrin–Ligand Complexes. Biochemistry. 43, 2699– 2707.

Kessel, C. R., and Granick, S., 1991. Formation and Characterization of a Highly Ordered and Well–Anchored Alkylsilane Monolayer on Mica by Self – Assembly. Langmuir. 7, 532–538.

Kumar, A., Biebuyck, H. A., and Whitesides, G. M., 1994. Patterning Self–Assembled Monolayers: Applications in Materials Science. Langmuir. 10, 1498–1511.

Lahiri, J., Issacs, L., Crzybowski, B., Carbeck, J. D., and Whitesides, G. M., 1999. Biospecific Binding of Carbonic Anhydrase to Mixed SAMs Presenting Benzenesulfonamide Ligands: A Model System for Studying Lateral Steric Effects. Langmuir. 15, 7186–7198.

Laibinis, P. E., Fox, M. A., Folkers, J. P., and Whitesides, G. M., 1991. Comparisons of Self–

Assembled Monolayers on Silver and Gold: Mixed Monolayers Derived from HS(CH2)21X and

HS(CH2)10Y (X, Y = CH3, CH2OH) Have Similar Properties. Langmuir. 7, 3167–3173.

Lewis, M., and Tarlov, M., 1995. Study of the Photooxidation Process of Self–Assembled Alkanethiol Monolayers. Jounal of the American Chemical Society. 117, 9574–9575.

Liu, G., Rodriguez, J. A., Dvorak, J., Hrbek, J., and Jirsak, T., 2002. Chemistry of Sulfur– Containing Molecules on Au(111): Thiophene, Sulfur Dioxide, and Methanethiol Adsorption. Surface Science. 505, 295–307.

Love, J. C., Estroff, L. A., Kriebel, J. K., Nuzzo, R. G., and Whitesides, G. M., 2005. Self– Assembled Monolayers of Thiolates on Metals as a Form of Nanotechnology. Chemical Reviews. 105 (4), 1103–1169.

Mrksich, M., Grunwell, R. J., and Whitesides, G. M., 1995. Biospecific Adsorption of Carbonic Anhydrase to Self–Assembled Monolayers of Alkanethiolates That Present Benzenesulfonamide Groups on Gold. Journal of the American Chemical Society. 117, 12009–12010.

Mrksich, M., and Whitesides, G. M., 1995. Patterning Self–Assembled Monolayers Using Microcontact Printing: A New Technology for Biosensors? Trends in Biotechnology. 13 (6), 228–235.

Norrod, K. L., and Rowlen, K. L., 1998. Removal of Carbonaceous Contamination from SERS– Active Silver by Self–Assembly of Decanethiol. Analytical Chemistry. 70, 4218–4221.

49

Nuzzo, R. G., and Allara, D. L., 1983. Adsorption of Bifunctional Organic Disulfides on Gold Surfaces. Journal of the American Chemical Society. 105, 4481–4483.

Ostuni, E., Grzybowski, B. A., Mrksich, M., Roberts, C. S., and Whitesides, G. M., 2003. Adsorption of Proteins to Hydrophobic Sites on Mixed Self–Assembled Monolayers. Langmuir. 19, 1861–1872.

Parker, J. L., Cho, D. L., and Claesson, P. M., 1989. Plasma Modification of Mica: Forces between Fluorocarbon Surfaces in Water and a Nonpolar Liquid. Journal of Physical Chemistry. 93, 6121–6125.

Raiber, K., Terfort, A., Benndorf, C., Krings, N., and Strehblow, H. –H., 2005. Removal of Self– Assembled Monolayers of Alkanethiolates on Gold by Plasma Cleaning. Surface Science. 595, 56–63.

Roberts, C., Chen, C. S., Mrksich, M., Martichonok, V., Ingber, D. E., and Whitesides, G. M.,

1998. Using Mixed Self–Assembled Monolayers Presenting RGD and (EG)3OH Groups to Characterize Long–Term Attachment of Bovine Capillary Endothelial Cells to Surfaces. Journal of American Chemical Society. 120, 6548–6555.

Sagiv, J., 1980. Organized Monolayers by Adsorption. 1. Formation and Structure of Oleophobic Mixed Monolayers on Solid Surfaces. Journal of the American Chemical Society. 102 (1), 92– 98.

Schreiber, F., 2000. Structure and Growth of Self–Assembling Monolayers. Progress in Surface Science. 65, 151–256.

Schrieber, F., 2004. Self–Assembled Monolayers: from ‘Simple’ Model Systems to Biofunctionalized Inferfaces. Journal of Physics: Condensed Matter. 16, R881–R900.

Schwartz, D. K., 2001. Mechanisms and Kinetics of Self–Assembled Monolayer Formation. Annual Review of Physical Chemistry. 52, 107–137.

Smith, R. K., Lewis, P. A., Weiss, P. S., 2004. Patterning Self–Assembled Monolayers. Progress in Surface Science. 75, 1–68.

Subramanian, A., Irudayaraj, J., and Ryan, T., 2006. A Mixed Self–Assembled Monolayer–Based Surface Plasmon Immunosensor for Detection of E. coli O157:H7. Biosensors and Bioelectronics. 21, 998–1006.

Tarlov, M. J., Burgess, Jr., D. R. F., and Gillen, G., 1993. UV Photopatterning of Alkanethiolate Monolayers Self–Assembled on Gold and Silver. Journal of American Chemical Society. 115, 5305–5306.

50 Twardowski, M., and Nuzzo, R. G., 2002. Chemically Mediated Grain Growth in Nanotextured Au, Au/Cu Thin Films: Novel Substrates for the Formation of Self–Assembled Monolayers. Langmuir. 18, 5529–5538.

Ulman, A., 1996. Formation and Structure of Self–Assembled Monolayers. Chemical Reviews. 96, 1533–1554.

Veiseh, M., Zhang, Y., Hinkley, K., and Zhang, M., 2001. Two–Dimensional Protein Micropatterning for Sensor Applications Through Chemical Selectivity Technique. Biomedical Microdevices. 3 (1), 45–51.

Veiseh, M., Zareie, M. H., and Zhang M., 2002. Highly Selective Protein Patterning on Gold – Silicon Substrates for Biosensor Applications. Langmuir. 18, 6671–6678.

Veiseh, M., Wickes, B. T., Castner, D. G., and Zhang, M., 2004. Guided Cell Patterning on Gold– Silicon Dioxide Substrates by Surface Molecular Engineering. Biomaterials. 25, 3315–3324.

Veiseh, M., Veiseh, O., Martin, M. C., Asphahani, F., and Zhang, M., 2007. Short Peptides Enhance Single Cell Adhesion and Viability on Microarrays. Langmuir. 23, 4472–4479.

Xiao, X., Liu, G., Charych, D. H., and Salmeron, M., 1995. Preparation, Structure, and Mechanical Stability of Alkylsilane Monolayers on Mica. Langmuir. 11, 1600–1604.

Xia, Y., and Whitesides, G. M., 1998. Soft Lithography. Annual Review of Materials Science. 28, 153–184.

Xia, Y., Qun, D., and Yin, Y., 2001. Surface Patterning and Its Application in Wetting/Dewetting Studies. Current Opinion in Colloid & Interface Science. 6, 54–64.

Yang, Z., Engquist, I., Wirde, M., Kauffmann, J. M., Gelius, U., and Liedberg, B., 1997. Preparation and Characterization of Mixed Monolayer Assemblies Composed of Thiol Analogues of Cholestrol and Fatty Acid. Langmuir. 13, 3210–3218.

Zhang, Y., Terrill, R. H., Tanzer, T. A., and Bohn, P. W., 1998. Ozonolysis Is the Primary Cause of UV Photooxidation of Alkanethiolate Monolayers at Low Irradiance. Journal of the American Chemical Society. 120, 2654–2655.

Zhang, Y., Terrill, R. H., and Bohn, P. W., 1999. Ultraviolet Photochemistry and ex Situ Ozonolysis of Alkanethiol Self–Assembled Monolayers on Gold. Chemistry of Materials. 11, 2191–2198.

51 Chapter 4 QUANTUM DOTS IN BIO–IMAGING AND BIO–LABELING

4.1. Overview and Features of Quantum Dots

Quantum dots (QDs) are semiconductor nanocrystals with typical sizes ranging from

2 to 10 nm. They are composed of periodic groups of II–VI, III–V, or IV–VI materials.

A quantum dot represents the three–dimensional quantum box within which its excitons

are confined in all three spatial dimensions. The electrons in this type of confinement

are described as a zero–dimensional electron gas when they are present in the

conduction band and, consequently, the density of states (DOS) for the electron gas is a

discrete function [Prasad, 2004]. Quantum dots are often described in terms of the

degree of quantum confinement. A strong quantum confinement regime represents the

case when the size of the quantum dot is smaller than the exciton Bohr radius of the

semiconductor material being used. In this case, the energy separation between the

sub–bands is much larger than the exciton binding energy and the electrons and holes

are largely represented by the energy states of their respective sub–bands. Also, the

effective energy bandgap of semiconductor quantum dots in the strong quantum

confinement regime is significantly larger than that in the bulk semiconductor [Prasad,

2004]. On the other hand, in a weak quantum confinement regime, the size of the

quantum dot is much larger than the exciton Bohr radius of the semiconductor material

being used and the energy separation between the sub–bands is comparable to or less

than the exciton binding energy [Prasad, 2004].

The absorption spectrum of quantum dots is continuous and broad, as shown in figure 4–1, due to the overlapping of a series of absorption peaks that get larger at shorter wavelengths. The quantum dots will not absorb light that has a wavelength

52 longer than that of the first exciton peak, also referred to as the absorption onset.

Material compositions and the size of the quantum dot determine the wavelength of the first exciton peak and those of all other subsequent peaks. The emission spectrum has a

Gaussian shape and the peak occurs at a slightly longer wavelength than the absorption onset as shown in figure 4–1. This energy separation phenomenon is known as the

Stoke’s shift. The size of the energy bandgap governs the emission frequency of a quantum dot and can be controlled by adjusting the size of the quantum dot [Prasad,

2004]. It is, therefore, possible to precisely tune the emission of monochromic light from a quantum dot. The bandwidth of the emission spectra, usually quantified by the full width at half maximum (FWHM), stems from the temperature (thermal broadening), the natural spectral line width of the quantum dots (natural broadening), and the size distribution of the population of quantum dots within a solution or matrix material

(inhomogeneous broadening).

The efficiency of quantum dots is determined by a measure known as Quantum Yield

(QY), the ratio of photons emitted to photons absorbed. QY is mainly controlled by the nonradiative transition of electrons or holes between energy levels and nonradiative recombination of electrons and holes. It is established that capping a core quantum dot with a shell (several atomic layers of an inorganic semiconductor with a wider bandgap than the core semiconductor) improves photoluminescence quantum yield by passivating surface nonradiative recombination sites or surface defects [Dabbousi et al.,

1997; Hines and Guyot–Sionnest, 1996]. In addition, quantum dots passivated with inorganic shell structures are more robust. One of the examples is CdSe cores overcoated with a layer of ZnS (shown in figure 4–2).

53

(A)

(B)

Figure 4–1. (A) Absorption (Abs) and emission (Em) of six different QD dispersions. The black line shows the absorption of the 510–nm–emitting QDs. The wavelength of lowest absorption (or the absorption onset) for the 510–nm QD is ~450 nm. (B) Photo demonstrating the size–tunable fluorescence properties and spectral range of the six QD dispersions plotted in (A) versus CdSe core size (r stands for the radius of CdSe core). All samples were excited at 365 nm with a UV source. For the 610–nm–emitting QDs, this translates into a Stokes shift of ~250 nm. After [Medintz et al., 2005].

54

Figure 4–2. Transmission Electron Micrographs (TEM) of (A) one “bare” CdSe nanocrystallite and (B) one CdSe nanocrystallite with a 2.6 monolayer ZnS shell. After [Babbousi et al., 1997]. (C) Artist’s rendering of CdSe quantum dot and CdSe/ZnS core– shell quantum dot.

The surface chemistry of colloidal quantum dots can be modified with a variety of

molecules via cap–exchange/ligand–exchange [Gill et al., 2006; Hwang and Cho, 2005;

Pinaud et al., 2004; Sui et al., 2003; Wang et al., 2007 (a); Xu et al., 2006], cross–

linking chemistry [Bruchez et al., 1998; Chan and Nie, 1998; Wang et al., 2007 (b)], in

situ functionalization [Wang et al., 2008], or receptor modification [Palaniappan et al.,

2006]. By modifying the surface with appropriate molecules, quantum dots can be

dispersed or dissolved in nearly any solvent, solution, or matrix. This molecular

coupling ability greatly increases the flexibility of quantum dots for different types of

environments and applications.

55 4.2. QD Synthesis and Capping

The widely used approach in synthesizing high–quality nearly monodisperse colloidal quantum dots is based on the pyrolysis of organometallic precursors by injection into a hot coordinating solvent [Murray et al., 1993]. The synthesis begins with the rapid injection of organometallic precursors into a hot coordinating solvent (e.g., mixture of trioctylphosphine/trioctyl phosphine oxide, TOP/TOPO) to produce a temporally discrete homogeneous nucleation. Slow growth and annealing are then continued in the coordinating solvent resulting in uniform surface derivatization and regularity in core structure. The prepared nanocrystals have a broad size distritution and optimization approaches such as size–selective precipitation are usually employed to obtain nearly monodisperse colloidal nanocrystals [Bangal et al., 2005; Chemseddine &

Weller, 1993; Dabbousi et al., 1997; Guzelian et al., 1996; Gaponik & Rogach, 2008;

Weller, 2003]. The size–selective precipitation method exploits the difference in solubility of smaller and larger particles. A typical procedure is as follows: a sample of prepared nanocrystals with a broad size distribution is dispersed in a solvent and a non– solvent is added dropwise under stirring or sonication until the initially optically clear solution becomes slightly turbid. As the largest nanocrystals in the sample exhibit the greatest attractive van der Waals forces, they tend to aggregate before the smaller particles resulting in flocculation. Separation of supernatant and flocculate by centrifugation produces a precipitate enriched with the largest nanocrystals. This precipitate can be re–dispersed in any appropriate solvent. The next portion of non– solvent is then added to the supernatant to isolate the second size–selected fraction, and so on. The procedure can be repeated several times and allows for obtaining up to 20 or more size–selected fractions from the initial crude solution. Moreover, each size– selected fraction can be subjected again to the size selection to further narrow the size distribution. Size–selective fractions of quantum dot nanocrystals can be characterized via absorbance & photoluminescence spectra shown in figure 4–3.

56

Figure 4–3. Absorbance and photoluminescence of the size–selected fractions of the thioglycolic acid–capped CdTe nanocrystals. The spectra of initial crude solution are highlighted in bold line. All size–selected fractions possess sharp excitonic transitions in the absorption spectra which is a direct evidence of their narrow particle size distributions. After [Gaponik and Rogach, 2008].

Only a few compounds can act as the coordinating solvents and this limits the

synthesis of high–quality nanocrystals in most cases [Qu et al., 2001]. It has been

demonstrated that high–quality monodisperse colloidal nanocrystals can be synthesized by rapidly injecting reagents into a hot non–coordinating solvent [Yu et al., 2002; Yu et al., 2003 (a); Yu et al., 2003 (b); Yu et al., 2004; Xu et al., 2006]. In this alternate approach, the size of nanocrystals can be tuned over a wide range under controlled conditions such as the injection temperature, the initial molar ratio of chemicals in a non–coordinating solvent, and the growth time. The growth is monitored with visible/Near Infrared (NIR) absorption spectroscopy in order to obtain the desired size [Xu et al., 2006]. This method allows producing highly monodisperse nanocrystals without any further size–selective precipitation process.

57 Purified core nanocrystals can be further overcoated with a layer of wider–bandgap semiconducting material for improved quantum yields. In a traditional approach, appropriate organometallic precursors are injected dropwise into the vigorously stirring heated coordinating solution with core nanocrystals [Dabbousi et al., 1997; Hines and

Guyot–Sionnest, 1996; Peng et al., 1997]. The precursors dosage must be pre– determined based on the initial size of core nanocrystals and the desired final thickness of the shell structure. On the other hand, for the core nanocrystals prepared through the routes of non–coordinating solvent reaction systems, another widely used approach has been developed based upon the Successive Ion Layer Adsorption and Reaction (SILAR)

[Li et al., 2003; Xu et al., 2006]. In this second approach, the shell is designed to grow epitaxially, one monolayer at a time, by alternate syringe–injection of predetermined dosages of cationic and anionic precursors into the heated reaction mixture with core nanocrystals. The alternate injection of precursors eliminates the homogeneous nucleation of shell materials in the solution and ensures the homogeneous monolayer growth of shell precursors onto core nanocrystals. The number of syringe injections depends on the final desired thickness of the shell (i.e., the number of monolayers of shell structure). As the core–shell structure grows upon each set of two successive syringe–injections, the amount of precursors required for the subsequent shell growth increases with the size of the core–shell structure and, thus, the first and subsequent dosages of cationic and anionic precursors are pre–determined, respectively, based on the initial size of core nanocrystals and subsequent sizes of core–shell structures. In monitoring the shell growth, absorption and photoluminescence measurements are carried out on aliquots taken after every two successive injections. The schematic of general experimental setup for quantum dots synthesis is illustrated in figure 4–4.

58

Figure 4–4. Schematic of experimental setup for quantum dot synthesis. Solvent is heated up and stirred in a three–neck flask. A thermocouple is inserted to precisely control the temperature of the solvent. When the desired injection temperature is reached, precursors are injected into the solvent with a syringe. The temperature is lowered to growth temperatures after the injection and the growth and annealing of nanocrystals is continued in the solvent.

4.3. Surface Derivatizations

In order to phase transfer quantum dots prepared using high–temperature routes, surface functionalization with ligands, either through cap exchange, a process mainly driven by mass action, or by encapsulating the original nanocrystals in a thick heterofunctional organic coating, driven mainly by hydrophobic absorption onto the

TOP/TOPO–capped quantum dots [Medintz et al., 2005]. These ligands mediate both the colloidal nanocrystal solubility in their respective solvents and serve as basis for further surface derivatizations and modifications. Additionally, capping ligands insulate, passivate, or protect the quantum dot surface from deterioration.

59 4.4. QD Bioconjugates for Bio–Imaging & Bio–Labeling

In order to study the complex spatio–temporal interplay of biomolecules from the cellular to integrative level, researchers traditionally use fluorescent labeling for both in vivo cellular imaging and in vitro assay detection [Miyawaki, 2003]. However, the intrinsic properties of organic and genetically encoded fluorophores, which generally have broad absorption/emission profiles [Miyawaki, 2003] and low photo–bleaching thresholds, have limited their effectiveness in long–term imaging and ‘multiplexing’

(simultaneous detection of multiple signals) without complex instrumentation and processing [Schröck et al., 1996].

The first biological uses of water–soluble biocompatible quantum–dot fluorophoes demonstrated that unique properties of quantum dots could overcome the limitations of traditional molecular dyes [Bruchez et al., 1998; Chen and Nie, 1998]. These properties includes: high quantum yield, high molar extinction coefficients [Dabbousi et al., 1997;

Leatherdale et al., 2002], broad absorption with narrow, symmetric photoluminescence

(PL) spectra spanning from UV to near–infrared, large effective Stokes shifts, high resistance to photobleaching (shown in figure 4–5), and exceptional resistance to photo– and chemical degradation [Alivisatos, 2004; Mattoussi et al., 2002; Murphy,

2002; Niemeyer, 2001; Parak et al., 2003]. For example, in comparison with organic dyes such as rhodamine, colloidal semiconductor quantum dots such as CdSe/ZnS core– shell QDs are 20 times as bright, 100 times as stable against photobleaching, and one– third as wide in spectral linewidth [Chen and Nie, 1998]. The ability to size–tune fluorescent emission as a function of core size and the broad excitation spectra allow multi–color labeling (shown in figure 4–6) by exciting mixed quantum dot populations at a single wavelength significantly far from their respective emissions. Additionally, the flexibility of QDs to surface modifications and further derivitization renders a wide range of bio–labeling and –imaging applications.

60

(A)

(B)

Figure 4–5. (A) Quantum dot resistance to photobleaching. Top row: nuclear antigens were labelled with QD 630–streptavidin (red), and microtubules were labeled with AlexaFluor 488 (green) simultaneously in a 3T3 cell. Bottom row: Microtubules were labeled with QD 630–streptavidin (red), and nuclear antigens were stained green with Alexa 488 (green). (B) Quantitative analysis of changes in intensities of quantum dot 608–streptavidin and Alexa 488–streptavidin. After [Wu et al., 2003].

61

Figure 4–6. Pseudocolored image depicting five–color QD staining of fixed human epithelial cells. Cyan corresponds to 655–nm Qdots labeling the nucleus, magenta 605– Qdots labeling Ki–67 protein, orange 525–Qdots labeling mitochondria, green 565– Qdots labeling microtubules and red 705–Qdots labeling actin filaments. After [Medintz et al., 2005].

It is suggested that quantum dots are cytocompatible and do not cause adverse effects on cell viability, morphology, function, or development over the duration of the experiments (from several hours to several days) at concentrations optimized for labeling efficiency [Michalet et al., 2005]. In addition, ligand–coated core–shell quantum dots (e.g., mercaptoacetic–acid–coated ZnS–capped CdSe QDs) show significant less interference with cell viability or function than core CdSe due to an improved protection of core quantum dot’s surface from oxidation [Derfus et al., 2004].

CdSe/ZnS core–shell quantum dots are the best available and widely used QD fluorophores for biological applications because the chemistry and synthesis are mostly refined and optimized [Medintz et al., 2005]. Furthermore, after surface modification/ functionalization with appropriate cap or ligand, CdSe/ZnS quantum dots can be conjugated to various types of biomolecules depending on labeling applications.

62 CdSe/ZnS core/shell QDs prepared using high–temperature routes have no intrinsic aqueous solubility, thus, phase–transfer to aqueous solution requires surface functionalization with hydrophilic ligands that can also serve as a basis for further derivatizations as well as conjugating biomolecules. A representative list of caps/ligands and the general QD–dispersal strategies are provided in Table 4–1. In general, there are three major strategies to functionalize the surface of quantum dots in bio–imaging and – labeling applications [Medintz et al., 2005]:

(1) The first strategy uses the cap exchange method and involves the substitution of

the native TOP/TOPO with bifunctional ligands, each presenting a surface

anchoring moiety to bind to the inorganic quantum dot surface (e.g., thiol

group) and an opposing hydrophilic end group (e.g., hydroxyl, carboxyl, etc.) to

achieve water–compatibility. These bifunctional ligands include an array of

thiol and phosphine mono– and multidentate ligands (Table 5–1 a, b) [Chan

and Nie, 1998; Goldman et al., 2002; Mattoussi et al., 2000; Mitchell et al.,

1999; Uyeda et al., 2005].

(2) The second strategy involves formation of polymerized silica shells

functionalized with polar groups, which insulate the hydrophilic quantum dot

(Table 5–1 c) [Bruchez et al., 1998; Gerion et al., 2001].

(3) The third method preserves the native TOP/TOPO on the quantum dots and

uses variants of amphiphilic ‘diblock’ and ‘triblock’ copolymers and

phospholipids to tightly interleave/interdigitate the alkylphosphine ligands

through hydrophobic attraction, whereas the hydrophilic outer block permits

aqueous dispersion and further derivatization (Table 5–1 d, f, g) [Ballou et al.,

2004; Dubertret et al., 2002; Gao et al., 2004; Mattheakis et al., 2004; Osaki et

al., 2004; Pellegrino et al., 2004; Wu et al., 2003].

63

Table 4–1. Schematic of generic quantum dot (QD) solubilization and biofunctionalization (a–h). Biofunctionalization (second panel from top) uses caps/ligands to provide three functions. Linkage to the QD (pink), water solubility (blue) and a biomolecule linking functionality (green). Examples of surface–capping strategies and the mechanism of interaction with the QD and the aqueous environment. For the cap exchange (top right) excess thiolated cap displaces the original TOP/TOPO organic coating by binding the ZnS surface with the thiol group and imparting hydrophilicity with the charged carboxyl (or other functionalities) yielding water–soluble colloidal QD dispersions. After [Medintz et al., 2005].

64 Biofunctionalization can be achieved by linking a biomolecule (e.g., protein, DNA, etc.) after appropriate ligands are attached to the quantum dot (the second panel from the top in Table 4–1). A wide range of biological applications of QD–bioconjugates has

been demonstrated. Some of the noteworthy applications are listed below:

• Labeling antidepressant–sensitive, human and Dorsophila serotonin

transporters (hSERT, dSERT) expressed in HeLa and HEK–293 cell membrane

with serotonin–labeled CdSe nanocrystals (SNACs) and serotonin–linker–arm–

conjugated CdSe/ZnS core–shell nanocrystals (LSNACs) [Rosenthal et al., 2002].

• Mapping the expression dynamics of the cytokine receptor [i.e., interleukin–2

receptor–α (IL–2Rα)] in Jurkat T cell by time–lapsed labeling with anti–IL2Rα–

conjugated quantum dots after activating the cell with Phorbol Myrastoyl Acetate

(PMA) and ionomycin [Warnement et al., 2006].

• Identifying the presence of respiratory syncytial virus (RSV) and monitoring the

progression of infection in HEp–2 cell monolayer cultures over time via labeling

F and G proteins, present at the surface of the virion particles in lipid bilayer

envelope, with antibody–conjugated QDs [Bentzen et al., 2005].

• Long–term multiple–color imaging of live mammalian (HeLa) cells or

Dictyostelium discoideum (AX2) cells via receptor–mediated endocytic uptake of

transferrin–conjugated merceptoacetic–acid–capped quantum dots and selective

labeling of cell surface proteins with antibodies–conjugated dihydrolipoic acid–

capped quantum dots [Jaiswal et al., 2003].

• Long–term labeling and tracking (up to 22 days) of human mesenchymal stem

cells (hMSCs) with RGD–conjugated CdSe/ZnS quantum dots during

proliferation and multilineage differentiations into osteogenic, chondrogenic, and

adipogenic cells [Shah et al., 2006; Shah et al., 2007].

65 • Strain– (i.e., gram–positive or gram–negative) and metabolism–specific

microbial labeling of a wide variety of bacteria (including pathogenic bacteria)

and fungi [Kloepfer et al., 2003]: for microorganism strain, specific components

of polysaccharide structures in the cell wall are targeted with lectin–conjugated

CdSe quantum dots; for pathogenic bacteria, transferrin receptors, present and

active on the microorganism’s surface, are targeted with transferrin–conjugated

CdSe quantum dots.

• Specific and simultaneous labeling of the cancer marker Her2, cytoskeleton

components, and nuclear antigens with QD–streptavidin bioconjugates [Wu et

al., 2003]:

− For the cancer marker Her2 over–expressed on the surface of fixed and

live human breast cancer cells (SK–BR–3), either QD–streptavidin probes,

coupled with a humanized anti–Her2 antiboidy and biotinylated goat anti–

human IgG (immunoglobulin G), or QD–IgG probes, coupled with

monoclonal anti–Her2 antibody, are used.

− In staining cytoskeleton fibers in mouse 3T3 fibroblast cells, QD–

streptavidin conjugates in combination with biotinylated phalloidin are

used for F–actin and, for microtubules, monoclonal anti–α–tubulin

antibody, biotinylated anti–mouse IgG, and QD–streptavidin are used.

− For nuclear antigens in human epithelial cells, QD–streptavidin

conjugates, coupled with human anti–nuclear antigen (ANA) antibodies

and biotinylated anti–human IgG, are used.

66 4.5. Chapter Appendix

4.5.1. Synthesis of CdSe/ZnS Core/Shell Quantum Dots

4.5.1.1. Synthesis of Core CdSe Nanocrystals

Core CdSe nanocrystals can be synthesized via the pyrolysis of the organometallic

precursors (dimethylcadmium [Me2Cd] and trioctylphosphine selenide [TOPSe]) in a coordinating solvent (trioctylphosphine oxide [TOPO]) [Murray et al., 1993]. The brief procedure is as follows: The precursors are rapidly injected into the heated coordinating solvent at temperatures ranging from 340 to 360 °C, and the initially formed small dots are grown at temperatures between 290 and 300 °C. The CdSe quantum dots are then collected as powders via size–selective precipitation [Murray et al., 1993] with methanol and then re–dispersed in hexane for further process.

4.5.1.2. Growth of ZnS Shell over CdSe Nanocrystals

In overcoating CdSe quantum dots with ZnS shell, diethylzinc [ZnEt2] and hexamethyldisilathiane [(TMS)2S] are used as the Zn and S precursors [Dabbousi et al.,

1997]. The amounts of precursors needed to grow a ZnS shell of desired thickness for

each CdSe sample is usually determined as follows: First, the average radius of the CdSe

quantum dots is estimated from TEM (Transmission Electron Microscope) or SAXS

(Small Angle X–ray Scattering) measurements. Next, the ratio of ZnS to CdSe necessary

to form a shell of desired thickness is calculated based on the ratio of the shell volume to

that of the core assuming the core/shell structure is spherical and taking into account of

the bulk lattice parameters of CdSe and ZnS. The actual amount of ZnS that grows onto

the CdSe cores is generally less than the amount of precursors added due to incomplete

reaction of the precursors and to the loss of some material on the walls of the reaction

flask during the addition.

67 The brief overcoating procedure is as follows: First, the CdSe quantum dots dispersed in hexane is transferred into the TOPO/TOP (trioctyl phosphine oxide/ trioctylphosphine) solution, and the solvent is pumped off. Next, the TOPO/TOP solution containing CdSe quantum dots is heated. The precursors (ZnEt2 & (TMS)2S) dissolved in TOP are then added dropwise to the vigorously stirring reaction mixture at temperatures ranging from 140 °C to 220 °C over a period of 5–10 min [Dabbousi et al.,

1997]. After the addition is complete, the reaction mixture is cooled to 90 °C and left stirring for several hours. A small aliquot of butanol is also added to the mixture to prevent the TOPO from solidifying upon cooling to room temperature.

The overcoated particles can be stored in the growth solution to ensure that the surface of the dots remains passivated with TOPO/TOP. Prepared quantum dots can be recovered in powder form later by precipitating with methanol in order to re–dispersed into a variety of solvents. The prepared quantum dots are characterized by optical characterizations (absorption & photoluminescence spectra), wavelength dispersive X– ray spectroscopy, X–ray photoelectron spectroscopy, transmission electron microscopy, and/or small angle X–ray scattering [Dabbousi et al., 1997].

4.5.2. Synthesis of PbSe/PbS Core/Shell Quantum Dots

4.5.2.1. Synthesis of Core PbSe Nanocrystals

The synthesis of core PbSe nanocrystals process involves the preparation of lead

oleate solution (a non–coordinating solvent) from PbO (lead II oxide) and oleic acid and

subsequent nucleation of PbSe by rapidly injecting selenium–trioctylphosphine (Se–

TOP) reagents into the reaction solution at an elevated temperature (~160 °C) [Xu et al.,

2006]. The reaction temperature must be lowered to ~135 °C following the injection to

allow the nuclei to grow into highly crystalline nanoparticles.

68 The growth is usually monitored with visible/NIR absorption spectroscopy in order to reach the desired size of nanocrystals. Under favorable conditions of injection temperature and growth time, highly monodisperse PbSe nanocrystals can be produced without any further size–selective precipitation process [Xu et al., 2006]. The resulting

PbSe nanocrystals are stabilized with a capping layer of oleate molecules coordinated to the Pb atoms. The size of the nanocrystals can be tunable over a wide range (4–11 nm) simply by varying the growth time, which is inherent to the noncoordinating solvent technique [Yu et al., 2002; Yu et al., 2003 (a); Yu et al., 2003 (b); Yu et al., 2004; Xu et al., 2006]. After the synthesis, PbSe core quantum dots are stored in chloroform for the subsequent growth of PbS shells.

4.5.2.2. Growth of PbS Shells over PbSe Nanocrystals

The brief procedure is as follows: PbSe nanocrystals are first purified by mixing with solvent (e.g., acetone), centrifuge–precipitating, and decantation. The purified PbSe nanocrystals are then diluted in octadecene (ODE), and heated to ~140 °C. Next, injection solutions are prepared: the lead–injection solution is prepared by mixing PbO, oleic acid, and ODE and the solution is heated to ~157 °C under an argon flow to obtain a colorless solution; the sulfur–injection solution is prepared by dissolving sulfur in trioctylphosphine (TOP) and ODE and then sonicating the mixture in an ultrasonic bath in order to get a clear solution. The temperature of the lead–injection solution is retained at ~80 °C, whereas the sulfur–injection solution is allowed to cool down to room temperature. The PbS shells are finally grown epitaxially, one monolayer at a time, by the consecutive syringe–injection of the predetermined dosages of lead– and sulfur– solutions into the preheated ODE solution that contains PbSe cores.

The amounts of Pb and S precursors required for the first shell–layer growth can be determined by the size and the lattice structure of the PbSe plain cores. For the

69 subsequent shell layers, the amounts of precursors required increase with the size of the core–shell structure. The temperature of the reaction mixture must be kept at 140 °C that is well below that required temperature for the nucleation of individual PbS nanocrystals. The reaction is usually monitored by taking absorption and photoluminescence measurements on aliquots taken 1–3 min after every two successive injections of lead– and sulfur–solution. The time between every two successive injections must be increased slightly with the layer number in order to account for the growing shell surface area. When the desired number of shell layers is reached, the reaction mixture is cooled to terminate the reaction. The final shell thickness must be thinner than the critical thickness of a PbS epitaxial layer grown on PbSe in order to avoid strain–related dislocations due to lattice mismatch. The final core–shell product is diluted with chloroform followed by a methanol extraction. Excess ligands are removed by repeatedly precipitating quantum dots with acetone, and the purified quantum dots are dispersed in chloroform for storage.

4.5.3. Particle in a Three–Dimensional Quantum Box

A three–dimensional potential energy well in which an electron is confined is

conceptually illustrated in figure 4–7. The volume of the potential energy well is marked by Lx, Ly, and Lz along the x, y, and z axes respectively. The potential energy is zero

inside the well (V = 0 in 0 < x < Lx, o < y < Ly, and 0 < z < Lz) and is infinite elsewhere.

Therefore, it is assumed that the electron cannot escape from the three–dimensional potential energy well.

70

Figure 4–7. A three–dimensional infinite potential energy box in which an electron is confined in three dimensions. Everywhere inside the box V = 0, but outside V is infinity. The electron cannot escape from the box. Adapted from [Kasap, 2002].

In order to analyze the behavior of the electron of mass me confined in the three– dimensional Potential Energy (PE) well, the three–dimensional version of time– independent Schrödinger equation must be employed as follows:

∂2ψ ∂2ψ ∂2ψ 2m + + + e (E −V )ψ = 0 (4–1) x2 y2 z2 2 ∂ ∂ ∂ 

where E and ψ are the energy and wave function of the electron in the infinite PE well respectively. By separation of variables [Kasap, 2002], the wave function of a confined electron in the infinite PE well, which is a solution to the equation (4–1), can be obtained as follows:

ψ x, y, z = A sin k x sin k y sin k z ( ) ( x ) ( y ) ( z ) (4–2) n π n π n π where k = x ; k = y ; k = z x L y L z L x y z

The constants, kx, ky, and kz (wave numbers), are determined from the boundary

conditions at x = Lx, y = Ly, and z = Lz respectively and are quantized. Each quantum number, nx, ny, or nz, can be any integer except zero.

71 It can be noticed that the equation (4–2) consists of the products of infinite one– dimensional potential well–type wave functions, one for each dimension, and each has its own quantum number n. Each possible eigen–function can be regarded as a state for the electron (for instance, ψ111 and ψ121 are two possible states). The comparison of the

nature of one–, two–, and three–dimensional potential well–type wave functions are

shown in figure 4–8.

Figure 4–8. Graphics representing the active region (top), allowed states in momentum space (middle), and density of states (bottom) for confinement in no dimensions (i.e., bulk material) (a), in one dimension (i.e., a quantum well) (b), in two dimensions (i.e., a quantum wire) (c), and in three dimensions (i.e., a quantum dot) (d). (d, k = p/  , QW, QR, and QD stands for dimension(s) of quantum structure, momentum, quantum well, quantum wire, and quantum dot respectively). After [Hoogland, 2008].

By substituting the wave function into the Schrödinger equation, the energy of the

electron can be obtained in terms of nx, ny, and nz as follows:

72 2 ⎛ 2 2 2 ⎞ h ⎜n ny n ⎟ E = ⎜ x + + z ⎟ n n n ⎜ 2 2 2 ⎟ (4–3) x y z 8m ⎜L L L ⎟ e ⎝ x y z ⎠

For a square box for which Lx = Ly = Lz = L, the energy of the electron is:

h2 n2 n2 n2 ( x + y + z ) En n n = (4–4) x y z 8m L2 e

The energy of an electron confined in the three–dimensional potential well depends on three quantum numbers, each one arising from boundary conditions along one of the coordinates. The lowest energy for the electron is E111 that is not zero. The next energy level corresponds to E211. In a square box, E211 is the same as E121 and E112. The number of states that have the same energy is termed the degeneracy of that energy level

(Hence, the second energy level E211 is three–fold degenerate).

4.5.4. Exciton Bohr Radius and the Energy Bandgap of QDs

The electron and hole are considered bound to each other via Coulomb attraction,

and this quasiparticle is known as an “exciton” [Murphy, 2002]. The exciton can be

considered a hydrogen–like system, and a Bohr approximation of the atom can be used

to calculate the spatial separation of the electron–hole pair of the exciton by:

εh2 r = (4–5) πm e2 r

where r is the radius of the sphere (defined by the three–dimensional separation of the electron–hole pair), ε is the dielectric constant of the semiconductor, mr is the reduced

mass of the electron–hole pair, h is Planck’s constant, and e is the charge on the electron.

Typically, the calculation suggests that the electron–hole pair spatial separation is ~1–10

73 nm for most semiconductors [Gaponenko, 1998]. Since the physical dimensions of a quantum dot can be smaller than the exciton diameter, the quantum dot is a good example of the “particle–in–a–box”. The electronic structure of semiconductor quantum dots, then, becomes intermediate between localized bonds and delocalized bands as shown in figure 4–9.

Figure 4–9. Comparison of the electronic structure of the atomic orbitals in a silicon atom (left) to that of a silicon cluster molecule (middle) and to that of bulk silicon (right). Atomic orbitals in the atom give rise to bonding and antibonding molecular orbitals in the cluster molecule, which give rise to the filled valence band and (mostly) empty conduction band in the bulk semiconductor. After [Gaponenko, 1998; Murphy, 2002].

The energies of the particle in the box depend on the size of the box (equation 4–3

and equation 4–4). Additionally, in the quantum dot, the bandgap energy is size–

dependent [Alivisatos, 1996 (a); Alivisatos, 1996 (b); Gaponenko, 1998; Murphy and

Coffer, 2002; Weller, 1993; Zhang, 1997]. The absorption onset from the QD absorption

spectrum verifies this phenomenon. As the particle size decreases, the absorption onset

shifts to higher energy (blue shifts), indicating an increase in bandgap energy [Alivisatos,

74 1996 (a); Alivisatos, 1996 (b); Brus, 1984; Murphy and Coffer, 2002; Rama Krishna and

Freisner, 1991; Weller, 1993; Zhang, 1997]. According to an early effective mass model calculation [Brus, 1984], The band gap energy of QD in the “strong confinement” regime can be estimated as:

2 ⎛ ⎞ 2 h ⎜ 1 1 ⎟ 1.8e E = E + ⎜ + ⎟− (4–6) g(QD) g(bulk) 8R2 ⎜m m ⎟ 4πε εR ⎝ e h ⎠ 0

where Eg is the bandgap energy of a quantum dot (or) bulk solid, R is the quantum dot

radius, me is the effective mass of the electron in the solid, mh is the effective mass of the hole in the solid, ε is the dielectric constant of the solid, and ε0 is the permittivity of a vacuum. The second term on the right–hand side of the equation is a particle–in–a– box–like term for the exciton, and the third term represents the electron–hole pair

Coulombic attraction, which is mediated by the solid (Note: Implicit in equation 4–6 is the assumption that the quantum dots are spherical and that the effective masses of the charge carriers and the dielectric constant of the solid are constant as a function of size).

However, the Brus model [equation (4–6)] does not always match experimental data very well for very small particle sizes. In this case, other calculations such as empirical pseudopotential method are performed to better match the experimental data [Rama

Krishna and Freisner, 1991].

75 References

Alivisatos, A. P., 1996 (a). Perspectives on the Physical Chemistry of Semiconductor Nanocrystals. Journal of Physical Chemistry. 100, 13226–13239.

Alivisatos, A. P., 1996 (b). Semiconductor Clusters, Nanocrystals, and Quantum Dots. Science. 271 (5251), 933–937.

Alivisatos, P., 2004. The Use of Nanocrystals in Biological Detection. Nature Biotechnology. 22 (1), 47–52.

Ballou, B., Lagerholm, B. C., Ernst, L. A., Bruchez, M. P., and Waggoner, A. S., 2004. Noninvasive Imaging of Quantum Dots in Mice. Bioconjugate Chemistry. 15, 79–86.

Bangal, M., Ashtaputre, S., Marathe, S., Ethiraj, A., Hebalkar, N., Gosavi, S. W., Urban, J., and Kulkarni, S. K., 2005. Semiconductor Nanoparticles. Hyperfine Interactions. 160, 81–94.

Bentzen, E. L., House, F., Utley, T. J., Crowe, J. E., and Wright, D. W., 2005. Progression of Respiratory Syncytial Virus Infection Monitored by Fluorescent Quantum Dot Probes. Nano Letters. 5 (4), 591–595.

Bruchez Jr., M., Moronne, M., Gin, P., Weiss, S., and Alivisatos, A. P., 1998. Semiconductor Nanocrystals as Fluorescent Biological Labels. Science. 281 (5385), 2013–2016.

Brus, L. E., 1984. Electron–Electron and Electron–Hole Interactions in Small Semiconductor Crystallites: The Size Dependence of the Lowest Excited Electronic State. Journal of Chemical Physics. 80 (9), 4403–4409.

Chan, W. C. W., and Nie, S., 1998. Quantum Dot Bioconjugates for Ultrasensitive Nonisotopic Detection. Science. 281 (5385), 2016–2018.

Chemseddine, A., and Weller, H., 1993. Highly Monodisperse Quantum Sized CdS Particles by Size–Selective Precipitation. Berichte der Bunsen–Gesellschaft Physical Chemistry. 97 (4). 636–637.

Dabbousi, B. O., Rodriguez–Viejo, J., Mikulec, F. V., Heine, J. R., Mattoussi, H., Ober, R., Jensen, K. F., and Bawendi, M. G., 1997. (CdSe)ZnS Core–Shell Quantum Dots: Synthesis and Characterization of a Size Series of Highly Luminescent Nanocrystallites. Journal of Physical Chemistry B. 101, 9463–9475.

Derfus, A. M., Chan, W. C. W., and Bhatia, S. N., 2004. Probing the Cytotoxicity of Semiconductor Quantum Dots. Nano Letters. 4 (1), 11–18.

76 Dubertret, B., Skourides, P., Norris, D. J., Noireaux, V., Brivanlou, A. H., and Libchaber, A., 2002. In vivo Imaging of Quantum Dots Encapsulated in Phospholipid Micelles. Science. 298 (5599), 1759–1762.

Gaponenko, S. V., 1998. Optical Properties of Semiconductor Nanocrystals. Cambridge University Press, New York.

Gaponik, N., and Rogach, A. L., 2008. Semiconductor Nanocrystal Quantum Dots: Synthesis, Assembly, Spectroscopy and Applications (editor: Rogach, A. L). 1st Edition, Springer, Wien, Austria. 73–99.

Gao, X., Cui, Y., Levenson R. M., Chung, L. W. K., and Nie, S., 2004. In vivo Cancer Targeting and Imaging with Semiconductor Quantum Dots. Nature Biotechnology. 22 (8), 969–976.

Gerion, D., Pinaud, F., Williams, S. C., Parak, W. J., Zanchet, D., Weiss, S., and Alivisatos, A. P., 2001. Synthesis and Properties of Biocompatible Water–Soluble Silica–Coated CdSe/ZnS Semiconductor Quantum Dots. Journal of Physical Chemistry B. 105, 8861–8871.

Gill R., Freeman, R., Xu, J. –P., Willner, I., Winograd, S., Shweky, I., and Banin, U., 2006. Probing Biocatalytic Transformations with CdSe–ZnS QDs. Journal of American Chemical Society. 128, 15376–15377.

Goldman, E. R., Balighian, E. D., Mattoussi, H., Kuno, M. K., Mauro, J. M., Tran, P. T., and Anderson, G. P., 2002. Avidin: A Natural Bridge for Quantum Dot–Antibody Conjugates. Journal of American Chemical Society. 124, 6378–6382.

Guzelian, A. A., Katari, J. E. B., Kadavanich, A. V., Banin, U., Hamad, K., Juban, E., and Alivisatos, A. P., 1996. Synthesis of Size–Selected, Surface–Passivated InP Nanocrystals. Journal of Physical Chemistry. 100, 7212–7219.

Hines, M. A. and Guyot–Sionnest, P., 1996. Synthesis and Characterization of Strongly Luminescing ZnS–Capped CdSe Nanocrystals. Journal of Physical Chemistry. 100, 468–471.

Hoogland, S., 2008. The Fuss about Quantum Dots. Photonics Spectra. 42 (1), 80.

Hwang, C. –S. and Cho, I. –H., 2005. Characterization of the ZnSe/ZnS Core Shell Quantum Dots Synthesized at Various Temperature Conditions and the Water Soluble ZnSe/ZnS Quantum Dot. Bulletin of the Korean Chemical Society. 26 (11), 1776–1782.

Jaiswal, J. K., Mattoussi, H., Mauro, J. M., and Simon, S. M., 2003. Long–term Multiple Color Imaging of Live Cells Using Quantum Dot Bioconjugates. Nature Biotechnology. 21, 47–51.

Kasap, S. O., 2002. Principles of Electronic Materials and Devices. 2nd Edition, McGraw–Hill, Boston.

77 Kloepfer, J. A., Mielke, R. E., Wong, M. S., Nealson, K. H., Stucky, G., and Nadeau, J. L., 2003. Quantum Dots as Strain– and Metabolism–Specific Microbiological Labels. Applied and Environmental Microbiology. 69 (7), 4205–4213.

Leatherdale, C. A., Woo, W. –K., Mikulec, F. V., and Bawendi, M. G., 2002. On the Absorption Cross Section of CdSe Nanocrystal Quantum Dots. Journal of Physical Chemistry B. 106, 7619–7622.

Li, J. J., Wang, Y. A., Guo, W., Keay, J. C., Mishima, T. D., Johnson, M. B., and Peng, X., 2003. Large–Scale Synthesis of Nearly Monodisperse CdSe/CdS Core/Shell Nanocrystals Using Air–Stable Reagents via Successive Ion Layer Adsorption and Reaction. Journal of American Chemical Society. 125, 12567–12575.

Mattheakis, L. C., Dias, J. M., Choi, Y. –J., Gong, J., Bruchez, M. P., Liu, J., and Wang, E., 2004. Optical Coding of Mammalian Cells Using Semiconductor Quantum Dots. Analytical Biochemistry. 327, 200–208.

Mattoussi, H., Mauro, M. J., Goldman, E. R., Anderson, G. P., Sundar, V. C., Mikulec, F. V., and Bawendi, M. G., 2000. Self–Assembly of CdSe–ZnS Quantum Dot Bioconjugates Using an Engineered Recombinant Protein. Journal of American Chemical Society. 122, 12142–12150.

Mattoussi, H., Kuno, M. K., Goldman, E. R., Anderson, G. P., and Mauro, J. M., 2002. Optical Biosensors: Present and Future (editors: Ligler, F. S. and Rowe C. A.). Elsevier, Amsterdam Netherlands. 537–569.

Michalet, X., Pinaud, F. F., Bentolila, L. A., Tsay, J. M., Doose, S., Li, J. J., Sundaresan, G., Wu, A. M., Gambhir, S. S., and Weiss, S., 2005. Quantum Dots for Live Cells, in vivo Imaging, and Diagnostics. Science. 307 (5709), 538–544.

Mitchell, G. P., Mirkin, C. A., and Letsinger, R. L., 1999. Programmed Assembly of DNA Functionalized Quantum Dots. Journal of American Chemical Society. 121, 8122–8123.

Miyawaki, A., 2003. Visualization of the Spatial and Temporal Dynamics of Intracellular Signaling. Developmental Cell. 4, 295–305.

Medintz, I. L., Uyeda, H. T., Goldman, E. R., and Mattoussi, H., 2005. Quantum Dot Bioconjugates for Imaging, Labelling, and Sensing. Nature Materials. 4, 435–446.

Murphy, C. J., 2002. Optical Sensing with Quantum Dots. Analytical Chemistry. 74, 520A– 526A.

Murphy, C. J., and Coffer J. L., 2002. Quantum Dots: A Primer. Applied Spectroscopy. 56 (1), 16A–27A.

78 Murray, C. B., Norris, D. J., and Bawendi M. G., 1993. Synthesis and Characterization of Nearly Monodisperse CdE (E = S, Se, Te) Semiconductor Nanocrystallites. Journal of American Chemical Society. 115, 8706–8715.

Niemeyer, C. M., 2001. Nanoparticles, Proteins, and Nucleic Acids: Biotechnology Meets Materials Science. Angewandte Chemie International Edition. 40, 4128–4158.

Osaki, F., Kanamori, T., Sando, S., Sera, T., and Aoyama, Y., 2004. A Quantum Dot Conjugated Sugar Ball and Its Cellular Uptake. On the Size Effects of Endocytosis in the Subviral Region. Journal of American Chemical Society. 126, 6520–6521.

Palaniappan, K., Xue, C., Arumugam, G., Hackney, S. A., and Liu, J., 2006. Water–Soluble, Cyctodextrin–Modified CdSe–CdS Core–Shell Structured Quantum Dots. Chemistry of Materials. 18, 1275–1280.

Parak, W. J., Gerion, D., Pellegrino, T., Zanchet, D., Micheel, C., Williams, S. C., boudreau, R., Le Gros, M. A., Larabell, C. A., and Alivisatos, A. P., 2003. Biological Applications of Colloidal Nanocrystals. Nanotechnology. 14, R15–R27.

Pellegrino, T., Manna, L., Kudera, S., Liedl, T., Koktysh, D., Rogach, A. L., Keller, S., Rädler, J., Natile, G., and Parak, W. J., 2004. Hydrophobic Nanocrystals Coated with an Amphiphilic Polymer Shell: A General Route to Water Soluble Nanocrystals. Nano Letters. 4 (4), 703–707.

Peng, X., Schlamp, M. C., Kadavanich, A. V., and Alivisatos, A. P., 1997. Epitaxial Growth of Highly Luminescent CdSe/CdS Core/Shell Nanocrystals with Photostability and Electronic Accessibility. Journal of American Chemical Society. 119 (30), 7019–7029.

Pinaud, F., King, D., Moore, H. –P., and Weiss, S., 2004. Bioactivation and Cell Targeting of Semiconductor CdSe/ZnS Nanocrystals with Phytochelatin–Related Peptides. Journal of American Chemical Society. 126, 6115–6123.

Prasad, P. N., 2004. Nanophotonics. John Wiley & Sons, Inc., Hoboken, New Jersey.

Qu, L., Peng, Z. A., and Peng X., 2001. Alternative Routes toward High Quality CdSe Nanocrystals. Nano Letters. 1 (6), 333–337.

Rama Krishna, M. V., and Fresner, R. A., 1991. Quantum Confinement Effects in Semiconductor Clusters. Journal of Chemical Physics. 95 (11), 8309–8322.

Rosenthal, S. J., Tomlinson, I., Adkins, E. M., Schroeter, S., Adams. S., Swafford, L., McBride, J., Wang, Y., DeFelice, L. J., and Blakely, R. D. 2002. Targeting Cell Surface Receptors with Ligand–Conjugated Nanocrystals. Journal of American Chemical Society. 124, 4586–4594.

79 Schröck, E., du Manoir, S., Veldman, T., Schoell, B., Wienberg, J., Ferguson–Smith, M. A., Ning, Y., Ledbetter, D. H., Bar–Am, I., Soenken, D., Garini, Y., Ried, T., 1996. Multicolor Spectral Karyotyping of Human Chromosomes. Science. 273 (5274), 494–497.

Shah, B., Clark, P., Stroscio, M., and, Mao, J., 2006. Labeling and Imaging of Human Mesenchymal Stem Cells with Quantum Dot Bioconjugates during Proliferation and Osteogenic Differentiation in Long Term. Proceedings of the 28th IEEE. 1470–1473.

Shah, B. S., Clark, P. A., Moioli, E. K., Stroscio, M. A., and Mao, J. J. 2007. Labeling of Mesenchymal Stem Cells by Bioconjugated Quantum Dots. Nano Letters. 7 (10), 3071–3079.

Sui, G., Orbulescu, J., Ji, X., Gattás–Asfura, K. M., Leblanc, R. M., and Micic, M., 2003. Surface Chemistry Studies of Quantum Dots (QDs) Modified with Surfactants. Journal of Cluster Science. 14 (2), 123–133.

Uyeda, H. T., Medintz, I. L., Jaiswal, J. K., Simon, S. M., and Mattoussi, H., 2005. Synthesis of Compact Multidentate Ligands to Prepare Stable Hydrophilic Quantum Dot Fluorophores. Journal of American Chemical Society. 127, 3870–3878.

Wang, S., Jarrett, B. R., Kauzlarich, S. M., and Louie, A. Y., 2007 (a). Core/Shell Quantum Dots with High Relaxivity and Photoluminescence for Multimodality Imaging. Journal of American Chemical Society. 129 (13), 3848–3856.

Wang, Q., Xu, Y., Zhao, X., Chang, Y., Liu, Y., Jiang, L., Sharma, J., Seo, D. –K., and Yan, H., 2007 (b). A Facile One–Step in situ Functionalization of Quantum Dots with Preserved Photoluminescence for Bioconjugation. Journal of American Chemical Society. 129, 6380– 6381.

Wang, Q., Liu, Y., Ke, Y., and Yan, H., 2008. Quantum Dot Bioconjugation During Core–Shell Synthesis. Angewandte Chemie International Edition. 47, 316–319.

Warnement, M. R., Faley, S. L., Wikswo, J. P., Rosenthal, S. J., 2006. Quantum Dots Probes for Monitoring Dynamic Cellular Response: Reporters of T Cell Activation. IEEE Transactions on Nanobioscience. 5 (4), 268–272.

Weller, H., 1993. Colloidal Semiconductor Q–Particles: Chemistry in the Transition Region Between Solid State and Molecules. Angewandte Chemie International Edition. 32, 41–53.

Weller, H., 2003. Synthesis and Self–Assembly of Colloidal Nanoparticles. Philosophical Transactions of the Royal Society A. 361, 229–240.

Wu, X., Liu, H., Liu, J., Haley, K. N., Treadway, J. A., Larson, J. P., Ge, N., Peale, F., and Bruchez, M. P., 2003. Immunofluorescent Labeling of Cancer Marker Her2 and other Cellular Targets with Semiconductor Quantum Dots. Nature Biotechnology. 21, 41–46.

80 Xu, J., Cui, D., Zhu, T., Paradee, G., Liang, Z., Wang, Q., Xu, S., and Wang, A. Y., 2006. Synthesis and Surface Modification of PbSe/PbS Core–Shell Nanocrystals for Potential Device Applications. Nanotechnology. 17, 5428–5434.

Yu, W. W., and Peng, X., 2002. Formation of High–Quality CdS and Other II–VI Semiconductor Nanocrystals in Noncoordinating Solvents: Tunable Reactivity of Monomers. Angewandte Chemie International Edition. 41 (13), 2368–2371.

Yu, W. W., Qu, L., Guo, W., and Peng, X., 2003 (a). Experimental Determination of the Extinction Coefficient of CdTe, CdSe, and CdS Nanocrystals. Chemical Materials. 15, 2854– 2860.

Yu, W. W., Wang, Y. A., and Peng, X., 2003 (b). Formation and Stability of Size–, Shape–, and Structure–Controlled CdTe Nanocrystals: Ligand Effects on Monomers and Nanocrystals. Chemical Materials. 15, 4300–4308.

Yu, W. W., Falkner, J. C., Yavuz, C. T., and Colvin, V. L., 2004. Synthesis of Monodisperse Iron Oxide Nanocrystals by Thermal Decomposition of Iron Carboxylate Salts. Chemical Communications. 20, 2306–2307.

Zhang, J. Z., 1997. Ultrafast Studies of Electron Dynamics in Semiconductor and Metal Colloidal Nanoparticles: Effects of Size and Surface. Accounts of Chemical Research. 30, 423–429.

81 Chapter 5 CELLULAR AND CELL–BASED BIOSENSORS

5.1. Measurement of Cellular Activities and States

Traditionally, electrical activity of cultured cells has been studied by employing micropipette or micro–wire electrodes. In these techniques, the cells to be monitored are seeded on a dish placed on a microscope stage, and micromanipulators are used to position the recording electrode. Several different techniques (figure 5–1) such as whole–cell patch recording, intracellular recording, and extracellular recording have been used depending on the type of electrical signal to be measured.

Figure 5–1. Classical electrical activity recording techniques: for intracellular recordings and monitoring of membrane impedance characteristics, a micropipette is brought into contact with a cell and a light suction is applied forming a tight seal. This technique is known as the “whole–cell patch” technique. The micropipette can be used to puncture the cell membrane and access the intracellular fluid. This method allows direct measurement of intracellular voltage as well as “intracellular” recordings. For “extracellular” recordings, a micropipette or micro wire is positioned in close proximity to the cell. Adapted from [Borkholder, 1998].

In whole–cell patch recording, the micropipette is brought into contact with the cell membrane with a light suction applied to form a tight seal (figure 5–2). The interior of the pipette is usually filled with saline solution. A micro–wire is placed in contact with this solution and conducts electrical current to the high input–impedance amplifier.

82 This setup allows for the observation of currents flowing through a small patch of membrane, of activities of ion channels, and the determination of the impedance characteristics of the cell membrane. However, an instable contact between the micro– pipette tip and the cell membrane limits the duration and efficiency of the experiment.

Figure 5–2. Patch–clamp recording is used to monitor ion channel activity. Because of the extremely tight seal between the mouth of the micropipette and the membrane, current can enter or leave the micropipette only by passing through the channels in the patch of cell membrane covering its tip. Recordings of the current through these channels can be made with the patch still attached to the rest of the cell, as in (A), or detached, as in (B) (the advantage of the detached patch is that it is easy to alter the composition of the solution on either side of the membrane to test the effect of various solutes on channel behavior). The micrograph (C) shows a nerve cell from the eye held in a suction pipette (the tip of which is shown on the left) while a micropipette (upper right) is being used for patch–clamp recording. (D) The circuitry for patch–clamp recording. After [Albert et al., 2004].

83 In order to directly measure the transmembrane potential, an intracellular recording

(relative to a distant reference electrode) is made by carefully inserting a micropipette electrode through the membrane (figure 5–1b). Although this method allows for a large signal (~100 mV) to be recorded as well as recordings of action potentials (AP), it also suffers from a mechanically fragile connection or weak sealing, which makes long–term recordings difficult. Additionally, in some cases, the damage caused by impalement of the cell compromises the intracellular ionic composition, which in turn can affect the intrinsic action potential firing rate [Breckenridge, et al., 1995].

The micropipette or micro–wire can be positioned in close proximity to the cellular membrane to record the extracellular Action Potentials (AP) (figure 5–1c). However, the measured signal is on the order of tens to hundreds of micro–volts (µV) relative to a

distant reference electrode immersed in the culture media. Although this passive and

non–invasive monitoring technique does not adversely affect cellular functions, it is still

limited due to the significantly smaller AP signal amplitudes as well as the recorded

signal shape that is vastly different from the actual transmembrane potential.

All of the classical techniques described above rely on the optical observation of the

cells for positioning of a recording electrode and significant human skills/interactions,

and, thus, usually result in fragile interfaces that limit the duration and efficiency of the

recordings. The experiment can be time–consuming and throughputs is significantly

low since recordings can only be taken from one or a few cells at a time. In addition, it is

almost impossible to study the network activity of excitable cells. Therefore, biological–

device platforms that provide high–yield and long–term recordings are required.

5.2. Requirements of Biosensors

When developing biological devices and instrumentations for monitoring the cellular

activities and states, it is necessary to design with many criteria in mind:

biocompatibility, maintenance of the physiochemical environment (temperature, pH,

84 etc.), maintenance of sterility during cell growth, methods of sample introduction, a transducer for monitoring the desired signal, electronics for extraction and acquisition of the electrical signal, and packaging, which facilitates insertion of the cell culture system into the measurement electronics while protecting the living system from the external environment [Borkholder, 1998]. These requirements often trade off against each other and require compromise for the best overall solution. Biocompatibility is usually the most important consideration in developing biological devices.

5.3. Cellular and Cell–Based Biosensors

Cellular and cell–based biosensors have been implemented using microorganisms and mammalian cells as sensing or recognition elements. Biosensors incorporating

microorganisms have been used particularly for environmental monitoring of pollutants

and those incorporating mammalian cells offer insight into the physiological effect of an

analyte on the cells [Pancrazio et al., 1999]. There are several approaches and methods

for transduction of cellular and cell–based sensor signals. Some of these approaches

include measures of fluorescence, metabolism, impedance, intracellular potentials,

extracellular potentials, and Raman spectra.

5.3.1. Cellular Microorganism–Based Biosensors

Some analytes, such as toxic pollutants, can activate microorganism pathways

involved in metabolism or nonspecific cell stress, resulting in the expression of one or

more genes [Belkin et al., 1997] or causing changes in metabolic activities. Therefore,

biological sensors, where microorganisms are incorporated as sensing elements, can be

used to detect such activities and changes, which are indicative of the presence of toxic

analytes. In one of the approaches, changes in extracellular pH due to effluents from the

microorganisms have been used as means of analyte detection [Rainina et al., 1996]. For

instance, detection of formaldehyde has been performed using immobilized yeast

85 coupled with pH–sensitive field effect transistors, where changes in metabolism of yeast due to the present of formaldehyde were detected via extracellular acidification rates

[Korpan et al., 1993]. Furthermore, a wide range of organophosphorus neurotoxins such as pesticides and chemical warfare agents can be detected using recombinant E. coli cells, which have been engineered to express the enzyme organophosphate hydrolase

[Rainina et al., 1996]. In this approach, resulting changes in extracellular pH due to protons being generated during the enzymatic hydrolysis of organophosphorus neurotoxins by organophosphate hydrolase are measured with pH electrodes.

The use of bacteria as sensor elements has a unique advantage as bacteria are amenable to genetic manipulations and can be engineered to generate an easily assayed signal in response to specific toxicants. The observed response will indicate the existence of a toxic compound, as well as providing some insightful information on its nature. For instance, bacteria can be genetically engineered such that a bioluminescent reporter gene is fused to the promoter sequence of a gene of the relevant metabolic pathway [Belkin et al., 1997]. The promoter reacts to the presence of toxicants and turns on the expression

of the bioluminescent reporter gene, which can be detected optically. These modified

bacteria have served as cell–based sensor elements for the detection of napthalene and

its metabolic product (salicylate) [Heitzer et al., 1994; King et al., 1990], benzene

[Applegate et al., 1998], toluene [Applegate et al., 1998; Burlage et al., 1994], mercury

[Selifonova et al., 1993], and middle chain alkanes such as octane [Sticher et al., 1997].

5.3.2. Fluorescence Assays of Cellular Function

Fluorescence imaging has been an invaluable tool for monitoring changes in the

concentrations of ions and protein expression related to cellular signaling [Dunn et al.,

1994; Tsien, 1988]. Most fluorescent reagents being developed for cellular and cell–

based assays are based on the combination of molecular biology, fluorescent probe

chemistry, and protein chemistry. For instance, reporter gene constructs, such as Green

86 Fluorescent Protein (GFP), have been implemented in genetically engineered mammalian and non–mammalian cell types [Giuliano et al., 1997; Zysk et al., 1998] to achieve measures of cell functions. Furthermore, the technology also offers High–

Throughput Screening (HTS) [Fernandes, 1998]. For example, an HTS system such as

Fluorometric Imaging Plate Reader (FLIPRTM), commercially available from Molecular

Devices [Groebe et al., 1999; figure 5–3], enables high–throughput fluorometric assays of membrane potential and intracellular calcium mobilization [Kuntzweiler et al., 1998;

Sullivan et al., 1999].

Figure 5–3. (A) Typical layout of a Fluorometic Imaging Plate Reader (FLIPRTM) (96–well mode). It generally consists of an incubated cabinet with integrated 96–channel pipettor and fluorometer. An argon laser is usually used to excite fluorophores in a 96–well microtiter plate and the emitted fluorescence is imaged by a cooled CCD camera. The image data from each well of the microtiter plate for each time point is processed and presented in real time on the computer screen (Note: the enclosure on the right of the tabletop houses the optics, the cell incubation chamber, and the multi–well pipettor. The water–cooled argon laser is seen toward the rear). After [Groebe et al., 1999]. (B) A schematic representation of the general components of the FLIPRTM system. Depicted are pipettor head for disposable tip loading and multi–well fluid addition, assay and compound addition plates, water–cooled argon laser for fluorescent excitation/illumination, and a cooled CCD camera for fluorescence data collection.

87 Despite the obvious advantages of fluorescent techniques, loading fluorescent dyes into the cells is usually considered a potentially invasive treatment. In addition, analytes of interest must be examined for autofluorescence to determine the feasibility of cellular fluorescent assays for resolution of small effects. Therefore, other cell–based approaches are more preferable in some cases.

5.3.3. Cellular Biosensors Based on Cell Metabolism

One category of cellular biosensors relies on the measurement of cell metabolism.

For instance, the Cytosensor® microphysiometer, developed by Molecular Devices [Parce

et al., 1989], as illustrated in figure 5–4, makes use of a light–addressable

potentiometric sensor [Hafeman et al., 1988] to measure extracellular pH, which is

correlated with cellular metabolic activity [Owicki et al., 1992]. This microphysiometer

has been used with a wide range of eukaryotic cells including primary central nervous

system neurons [Brown et al., 1997; Raley–Susman et al., 1992], hepatocytes [Cao et al.,

1998], neutrophils and endothelial cells [Gronert et al., 1998], and cells expressing

transfected receptors [Baxter et al., 1994]. However, the microphysiometer could not

discriminate different activations of a variety of receptors in cells, regardless of the signal

transduction method [Owicki et al., 1990]. For instance, measurements of metabolism

from primary neurons revealed that gamma–aminobutyric acid (GABA), an inhibitory

neurotransmitter, and kainic acid, an excitatory neurotransmitter, both cause an

elevation in cell metabolism and a subsequent elevation in extracellular acidosis [Brown

et al., 1997; Raley–Susman et al., 1992]. Therefore, this type of sensing approach usually

requires proper interpretation of data and parallel/control experiments in the presence

of known receptor antagonists that eliminate specific receptor responses.

88

Figure 5–4. Light–addressable semiconductor sensors. (a) A silicon plate with a surface insulator of oxynitride (shaded with diagonal lines) in contact with an electrolyte is photo–responsive to the light emitting diodes A, B, C, and D. The resulting alternating photocurrent I in the external circuit depends on the applied bias potential Ψ and the surface potential (electrochemical potential at electrolyte–oxynitride interface). Different chemistries located on different regions of the insulating surface produce variations in the local surface potential that can be determined by selective illumination with one or another of the light–emitting diodes. This configuration can be employed as microphysiometer cell–based biosensor that relies on metabolic rate for analyte detection. The electrolyte–oxynitride interface provides a Nernstian response to pH or cell membrane potential changes. The n– or p– type silicon insulated with oxynitride, which is pH sensitive, will be photo–responsive to light produced by one or more light– emitting diodes, resulting in different photocurrents depending on pH level. (b) For high–sensitivity measurements of enzyme activity, it is advantageous to localize the enzyme molecules in a small volume so that they are present at a high effective concentration. Volumes as small as a nano–liter can be achieved with a configuration such as the one illustrated. The controlling electrode acts as a piston. It can be raised to allow fluid to be passed over the sensor surface and then lowered to create the small reaction volume. After [Hafeman et al., 1988].

89 5.3.4. Cellular Biosensors Based on Electrical Impedance

The membranes of biological materials including cells exhibit dielectric properties

[Gordon et al., 1989] (i.e., it exhibits capacitive and resistive properties due to

phospholipid bilayer and transmembrane ion channels respectively). Hence, by

culturing cells over one or more planar electrodes, the effective electrode impedance

changes, permitting a noninvasive assay of cell adhesion, spreading, and motility

[Giaever and Keese, 1986; Mitra et al., 1991]. The impedance spectroscopy on cell–

covered electrodes has been used as a mean to address and monitor the behaviors of

non–excitable cell types such as macrophages [Kowolenko et al., 1990], endothelial cells

[Tiruppathi et al., 1992], fibroblasts [Giaever and Keese, 1993], and etc. One of the

impedance measurement techniques known as Electric Cell–Substrate Impedance

Sensing (ECIS) can reveal the dynamic of cellular morphological changes and

movements in response to hormones, enzymes, or chemical compounds (Figure 5–5 and

5–6) [Arndt et al., 2004; Giaever and Keese, 1991; Giaever and Keese, 1993; Noiri et al.,

1997; Smith et al., 1994; Wegner et al., 2000]. Furthermore, impedance measurements

from sensors that employ Inter–Digitated Electrode Structure (IDES) can also reveal

growth and morphological behavior of growing cells [Ehret et al., 1997; Wolf et al.,

1998]. From the perspective of cell–based assays and biosensing, the emphasis of these

impedance spectroscopy is on monitoring of time and concentration–dependent

variations in the impedance of cultured cells, which are incorporated as a sensor element, in the presence of chemical compounds or toxins [Curtis et al., 2009; Tlili, et al., 2003]. Not only does the use of living cells as sensor elements provide the high sensitivity for a broad range of biologically active substances that affect the response or behavior of cells, the approach can also offer insight into the physiological effect of these substances on the cell metabolism.

90

Figure 5–5. A simplified schematic of the Electric Cell–Substrate Impedance Sensing (ECIS) instrumentation. Impedance of the small electrode is measured with a lock–in amplifier in series with a 1–MΩ current limiting resistor. The measured fluctuations in the real and imaginary voltage are displayed for a 1–hr experiment with confluent WI– 38 VA13 fibroblasts using a 4–kHz, 1–V source. The in–phase and out–of–phase voltage values are approximately proportional to the respective changes in resistance and capacitive reactance of a RC circuit that represents the cell–covered electrode. After [Giaever and Keese, 1991].

91

Figure 5–6. Impedance measurements using ECIS technique showing: (A) an increase and fluctuations in effective resistance of the electrode due to cell spreading and cell micromotion respectively; (B) a sharp decrease and a transient rise in effective impedance of the electrode due to cell wounding after electroporation and cell healing/migration respectively. After [Applied BioPhysics, Inc., 2009].

92 5.3.5. Cellular Biosensor Based on Intracellular Potential

It is known that membrane excitability determines the control of secretion in neurons and of contraction in myocardiocytes. Thus, analytes that affect membrane excitability in excitable cells are expected to have profound effects on an organism.

Furthermore, the nature of the changes in excitability can yield physiologic implications for the organism’s response to analytes. Classically, direct monitoring of cell membrane potential can be achieved through patch–clamping techniques. For instance, repetitively firing neurons from the visceral ganglia of the pond snail has been used to quantitatively assess the concentration of a model analyte, serotonin [Skeen et al., 1990].

A neuroblastoma glioma cell line, NG108–15, can be used as sensing elements to a variety of toxins. Under serum–free media conditions, NG108–15 cells usually express a neuronal phenotype (figure 5–7) [Ma et al., 1998; Nirenberg et al., 1983] and the capability of spontaneous firing [Kowtha et al., 1993]. In bullfrog sympathetic ganglion neurons, both VX (O–ethyl s–2–NN–diisoprophlamin ethyl methylphosphono– fluoridate) and GD (soman; O–pincolyl methylphosphonofluoridate) have been shown to increase membrane excitability in a manner consistent with voltage–gated Ca2+ channel

blockade [Heppner et al., 1991; Heppner et al., 1992]. Therefore, these cells can be

employed as sensing elements to rapidly detect chemical warfare agents. As shown in

figure 5–7, spontaneous firing in NG108–15 cells was significantly affected by exposure

to chemical warfare agents: GD and VX. The effects of these two organophosphate

agents were distinctive; whereas GD irreversibly depolarized the resting membrane

potential (RMP), whereas, VX reversibly hyperpolarized the resting potential, resulting

in a loss of excitability. These records illustrate the utility of excitable cells as sensor

elements with sensitivity to chemical warfare agents.

However, the invasive nature of intracellular recording significantly limits the

robustness of this approach for biosensor applications. High throughput screening is

93 almost impossible for most techniques used can only access one or a couple of cells at a time. Furthermore, excitable cells assemble into coupled networks rather than acting as isolated elements. Neurons propagate information via synapses and myocardiocytes form a syncitium via gap junctions. As a result, the ability to simultaneously monitor two or more cells are imperative and sensors platforms that can provide such functionalities are required in order to make measurements of membrane excitability and cell coupling.

Figure 5–7. Neuroblastoma–glioma (NG108–15) cells as electrical detectors of chemical agents. Left panel: typical NG108–15 cells cultured under serum–free conditions after 21 days in vitro exhibiting neuronal phenotype. Right panel: differential effects of the chemical warfare agents VX (O–ethyl s-2-NN-diisopropylamin ethyl methylphosphonofluoridate) and GD (soman; O-pincolyl methylphosphonofluoridate) on spontaneous, rhythmic firing measured using standard glass microelectrodes filled with 3 M KCl as described by [Kowtha et al., 1993]. In contrast to GD, VX was completely reversible. Dotted line indicates the zero membrane potential level. Adapted from [Pancrazio et al., 1999].

94 5.3.6. Cellular Biosensors Based on Extracellular Potential

Couplings of neurons and cardiac myocytes are inaccessible via intracellular

recording that relies on glass micropipettes or patch clamp techniques. Furthermore,

due to the invasive nature and low throughputs, traditional intracellular recording

techniques are limited to a few applications. As a result, planar microelectrode arrays

have emerged as a powerful device for long–term recording of network dynamics.

Extracellular microelectrode arrays have been used as a noninvasive, long–term, and

high–yield approach to monitor electrical activity in excitable cells such as neurons and

myocytes [Connolly et al., 1990; Gross et al., 1995; Thomas et al., 1972]. A system of

microelectrode arrays presents a grid for data acquisition from networks of electrically

active cells [Egert et al., 1998].

A system of microelectrode arrays incorporated with amplifier/stimulator CMOS

chips can accomplish both extracellular recordings and stimulations [Pancrazio et al.,

1998 (a)]. Extracellular recording potentials from the microelectrodes are usually the

second or third derivative of the action potential [Connolly et al., 1990; Pancrazio et al.,

1999]. Not only can a Microelectrode Array (MEA) system record the action potentials,

but it can also allow the quantification of action potential propagation velocity by

revealing the spike delay between proximal microelectrode sites (Figure 5–8) [Pancrazio

et al., 1998 (b); Pancrazio et al., 1999]. Furthermore, planar microelectrode arrays have

been used to record extracellular potentials from tissue slices [Meister et al., 1994;

Stoppini et al., 1997]. The technique also allows monitoring the effect of chemical

compounds on the electrical activity of neurons (Figure 5–9) [Pancrazio et al., 1999].

95

Figure 5–8. Planar multi–electrode arrays permitting noninvasive, simultaneous recordings from excitable tissue for measurement of action potential propagation. Left panel: Microelectrode Array (MEA) with 36 microelectrode sites that are 10 µm in diameter. Right panel: Simultaneous recording from a monolayer of embryonic day 11 chick myocardiocytes after 2–3 days in vitro. The monolayer exhibited regular beating at a rate of 1–2 Hz, where spike delays between proximal microelectrode sites allow quantification of cardiac propagation velocity. After [Pancrazio et al., 1999].

Figure 5–9. Culture of spinal cord neurons on MEA for toxicological evaluation. Left panel: Phase contrast image of embryonic day–15 rat spinal cord neurons cultured for 18 days on a microelectrode array. Cells were cultured under serum–free defined media conditions on artificial self–assembled monolayer substrates of aminosilanes. Right panel: Addition of glutamate (50 µM) greatly augmented spike activity. Administration of an organophosphate, diisopropylfluorophosphate (DFP; 25 µM), revealed a marked ablation of spontaneous firing, illustrating the utility of neurons cultured on microelectrode arrays for detection of toxic compounds. After [Pancrazio et al., 1999].

96 5.3.6.1. Analytical Methods

In order to analyze cultured neuronal networks from MEA recordings, cross

correlations between multi–electrode channels and interspike interval (ISI) variances

have been usually evaluated. Both simulation and experimental findings suggested that

the interspike interval variance discriminated periodic bursting from asynchronous

firing, whereas the cross correlation between multi–electrode channels is indicative of

the synchronization among the neurons in the network [Bove et al., 1997; Canepari et al.,

1997]. Other statistical methods used are multivariate analysis, which includes linear

discriminant analysis [Gochin et al., 1994], and principal component analysis (PCA)

[Nicolelis and Chapin, 1994], which can take into account the activity of all the recorded

neurons simultaneously and incorporate spatio–temporal features into the analysis. For

instance, PCA allows construction of the activity of neuron populations by the weighted

addition of the integrated activity of the population, resulting in the structure underlying

the collective neuronal behavior [Takahashi, 1996]. Furthermore, reconstruction of the

first principal component from recordings of cortical, thalamic, and brain stem activity

was used to more reliably track oscillatory episodes than recordings from any single

neuron in the population [Nicoleis et al., 1995]. For analyte detection, minimal sets of

parameters, such as burst amplitude, duration, and frequency, must be determined to

perform risk/threat assessment. Multichannel analysis is also benefited from advances

in nonlinear dynamics, which can potentially distinguish deterministic behavior from

random fluctuations. For example, nonlinear time series analysis of the correlation

dimension from multielectrode electroencephalograph (EEG) records has been shown to

yield a prognostic indicator for seizure activity [Elger and Lehnertz, 1998]. Hence, these

analysis methods are not just limited to recordings from cultured neuronal networks.

97 5.3.7. Raman Spectroscopy Cell–Based Biosensors

The Raman spectrum of a cell represents an information–rich “fingerprint” of the overall biochemical composition of the cell, thus different toxic agents that initiate different cellular responses and biochemical changes produce distinct changes in the

Raman spectra [Notingher, 2007]. Furthermore, time–course Raman spectra can be acquired from live cells maintained in physiological conditions without the need of invasive procedures. One of the variations of Raman spectroscopy, known as Surface

Enhanced Raman Spectroscopy (SERS), renders huge potentials in biosensing as it enhance signals by orders of magnitudes and can detect some molecule vibration modes, which cannot be detected by traditional Raman Spectroscopy [Culha et al., 2003; Isola et al., 1998; Kneipp et al., 2006].

In Raman micro–spectroscopy (traditional or SERS), a spectrometer is coupled to an optical microscope to enable both excitation and collection of Raman spectra (Figure 5–

10). Additionally, the high–quality of optical microscopes makes it possible to obtain measurements with diffraction–limited spatial resolution (approximately half wavelength of the excitation laser). Raman micro–spectroscopy can be used to monitor the molecular changes in a single cell after the cell is exposed to a toxic chemical or a drug (Figure 5–11) [Notingher et al., 2004; Owen et al., 2006; Verrier et al., 2004; Yao et al., 2009]. Raman spectra or second derivative averaged Raman spectra can present important biochemical changes taking place during cell death, such as the degradation of proteins, DNA breakdown, and the formation of lipid vesicles. Furthermore, Raman spectroscopy can be used as a noninvasive method to distinguish cells at different stages of the cell cycle by the spectral differences between cells in different stages of the cell cycle, which are caused by variations in DNA vibrations [Notingher et al., 2003].

98

Figure 5–10. Experimental and instrumentation setup for Raman micro– spectroscopy measurements of cells. Adapted from [Notingher, 2007].

Figure 5–11. Raman spectra of viable (spectrum a) and dead (spectrum b) human lung derived (A549) cells. The arrows indicate the positions where the most significant changes, related to the degradation of proteins, DNA breakdown, and the formation of lipid vesicles, occur. After [Notingher et al., 2003].

99 Raman spectroscopy cell–based biosensor platforms have promising future in High

Throughput Screening (HTS) of drugs. For instance, Raman spectra from individual cancer cells such as human medulloblastoma (DAOY) cells are used as sensing elements to access the chemotherapeutic–agent–induced cellular changes [Buckmaster et al.,

2009]. DAOY cells were immobilized on the cell trapping pads presented on the surface of patterned single–cell biosensor platform (figure 5–12b). The cell immobilization was mediated by surface–chemistry–mediated adhesion and spreading (figure 5–12a).

Common chemotherapeutic agent (e.g., etoposide) is administered at different doses and confocal Raman micro–spectroscopy was taken from individual cells over the time course of 12 hours. Resulting spectra shows drug–induced changes at the single–cell level such as reduction in the concentrations of DNA and protein contents. The time– course spectra can also reveal cells that exhibit resistance to drug (figure 5–13), which is of special interest in the development of higher efficacy chemotherapeutic drugs.

Figure 5–12. (a) Schematic of the final surface modification of gold–patterned silicon oxide substrate for subsequent single–cell adhesion. (b) Optical differential interference

contrast (DIC) image of patterned DAOY cells on the surface–modified 20µm × 20µm gold squares with 40–µm spacing between squares. After [Buckmaster et al., 2009].

100

Figure 5–13. Raman spectra for two representative cells acquired prior to (0 h) and every 12 hours after etoposide exposure. (a) Raman spectra of a single cell that died after 36 hours. (b) Raman spectra of a single cell that showed resistance to etoposide and was viable over the course of the experiment. After [Buckmaster et al., 2009].

101 References

Alberts M. B., Bray, D., Hopkin, K., Johnson, A., Lewis, J., Raff, M., Roberts, K., and Walter, P., 2004. Essential Cell Biology. 2nd Edition, Taylor & Francis Group, New York.

Applegate, B. M., Kehrmeyer, S. R., and Sayler, G. S., 1998. A Chromosomally Based tod– luxCDABE Whole–Cell Reporter for Benzene, Toluene, Ethylbenzene, and Xylene (BTEX) Sensing. Applied and Environmental Microbiology. 64 (7), 2730–2735.

Arndt, S., Seebach, J., Psathaki, K., Galla, H. –J., and Wegener, J., 2004. Bioelectrical Impedance Assay to Monitor Changes in Cell Shape During Apoptosis. Biosensors and Bioelectronics. 19 (6), 583–594.

Baxter, G. T., Young, M. –L., Miller, D. L., and Owicki, J. C., Using Microphysiometry to Study the

Pharmacology of Exogenously Expressed m1 and m3 Muscarinic Receptors. Life Sciences. 55 (8), 573–583.

Belkin, S., Smulski, D. R., Dadon, S., Vollmer, A. C., van Kyk, T. K., and Larossa, R. A., 1997. A Panel of Stress–Responsive Luminous Bacteria for the Detection of Selected Classes of Toxicants. Water Research. 31 (12), 3009–3016.

Borkholder, D. A., 1998. Cell Based Biosensors Using Microelectrode. Ph.D. Dissertation. Stanford University, Stanford, CA.

Bove, M., Grattarola, M., and Verreschi, G., 1997. In Vitro 2-D Networks of Neurons Characterized by Processing the Signals Recorded with a Planar Microtransducer Array. IEEE Transactions on Biomedical Engineering. 44, 964–977.

Breckenridge, L. J., Wilson, R. J. A., Connolly, P., Curtis, A. S. G., Dow, J. A. T., Blackshaw, S. E., and Wilkinson, C. D. W., 1995. Advantages of Using Microfabricated Extracellular Electrodes for In Vitro Neuronal Recording. Journal of Neuroscience Research. 42, 266–276.

Brown, M. J., Wood, M. D., Coldwell, M. C., and Bristow D. R., 1997. Measurement of GABAA Receptor Function in Rat Cultured Cerebellar Granule Cells by the Cytosensor Microphysiometer. British Journal of Pharmacology. 121, 71–76.

Buckmaster, R., Asphahani, F., Thein, M., Xu, J., and Zhang, M., 2009. Detection of Drug– Induced Cellular Changes Using Confocal Raman Spectroscopy on Patterned Single–Cell Biosensors. Analyst. 134, 1440–1446.

Burlage, R. S., Palumbo, A. V., Heitzer, A., and Sayler, G., 1994. Bioluminescent Reporter Bacteria Detect Contaminants in Soil Samples. Applied Biochemistry and Biotechnology. 45, 731–740.

102 Canepari, M., Bove, M., Maeda, E., Cappello, M., and Kawana, A., 1997. Experimental Analysis of Neuronal Dynamics in Cultured Cortical Networks and Transitions between Different Patterns of Activity. Biological Cybernetics. 77, 153–162.

Cao, C. J.,Eldefrawi, M. E., Eldefrawi, A. T., Burnett, J. W., Mioduszewski, R. J., Menking, D. E., and Valdes, J. J., 1998. Toxicity of Sea Nettle Toxin to Human Hepatocytes and the Protective Effects of Phosphorylating and Alkylating Agents. Toxicon. 36 (2), 269–281.

Connolly, P., Clark, P., Curtis, S. G., Dow, J. A. T., and Wilkinson, C. D. W., 1990. An Extracellular Microelectrode Arrays for Monitoring Electrogenic Cells in Culture. Biosensors and Bioelectronics. 5 (3), 223–234.

Culha, M., Stokes, D., Allain, L. R., and Vo–Dinh, T., 2003. Surface–Enhanced Raman Scattering Substrate Based on a Self–Assembled Monolayer for Use in Gene Diagnostics. Analytical Chemistry. 75 (22), 6196–6201.

Curtis, T. M., Widder, M. W., Brennan, L. M., Schwager, S. J., van der Schalie, W. H., Fey, J., and Salazar, N., 2009. A Portable Cell–Based Impedance Sensor for Toxicity Testing of Drinking Water. Lab on a Chip. 9, 2176–2183.

Dunn, K. W., Mayor, S., Myers, J. N., and Maxfield, F. R., 1994. Applications of Ratio Fluorescence Microscopy in the Study of Cell Physiology. The FASEB Journal. 8, 573–582.

Egert, U., Schlosshauer, B., Fennrich, S., Nisch, W., Fejtl, M., Knott, T., Muller, T., and Hammerle, H., 1998. A Novel Organotypic long–term Culture of the Rat Hippocampus on Substrate–Integrated Multielectrode Arrays. Brain Research Protocols. 2 (4), 229–242.

Ehret, R., Baumann, W., Brischwein, M., Schwinde, A., Stegbauer, K. and Wolf, B., 1997. Monitoring of Cellular Behavior by Impedance Measurements on Interdigitated Electrode Structures. Biosensors and Bioelectronics. 12 (1) 29–41.

Elger, C. E., and Lehnertz, K., 1998. Seizure Prediction by Nonlinear Time Series Analysis of Brain Electrical Activity. European Journal of Neuroscience. 10, 786–789.

Fernandes, P. B., 1998. Technological advances in high-throughput screening. Current Opinion in Chemical Biology. 2 (5), 597–603.

Giaever, I., and Keese, C. R., 1986. Use of Electric Fields to Monitor the Dynamical Aspect of Cell Behavior in Tissue Culture. IEEE Transactions on Biomedical Engineering. 33, (2), 242–247.

Giaever, I., and Keese, C. R., 1991. Micromotion of Mammalian Cells Measured Electrically. Proceedings of the National Academy of Sciences USA. 88 (17), 7896–7900.

Giaever, I., and Keese, C. R., 1993. A Morphological Biosensor for Mammalian Cells. Nature. 366 (6455), 591–592.

103

Giuliano, K. A., DeBiasio, R. L., Dunlay, R. T., Gough, A., Volosky, J. M., Zock, J., Pavlakis, G. N., and Taylor, D. L., 1997. High–Content Screening: A New Approach to Easing Key Bottlenecks in the Drug Discovery Process. Journal of Biomolecular Screening. 2 (4), 249– 259.

Gochin, P. M., Colombo, M., Dorfman, G. A., Gerstein, G. L., and Gross, C. G., 1994. Neural Ensemble Coding in Inferior Temporal Cortex. Journal of Neurophysiology. 71 (6), 2325– 2337.

Gordon, L. G. M., Kottra, G., and Frömter, E., 1989. Electrical Impedance Analysis of Leaky Epithelia: Theory, Techniques, and Leak Artifact Problems. Methods in Enzymology. 171, 642–663.

Groebe, D. R., Gopalakrishnan, S., Hahn, H., Warrior, U., Traphagen, L., and Burns, D. J., 1999. Use of a Fluorometric Imaging Plate Reader in High–Throughput Screening. Proceedings of SPIE. 3603, 297–306.

Gronert, K., Colgan, S. P., and Serhan, C. N., 1998. Characterization of Human Neutrophil and Endothelial Cell Ligand-Operated Extracellular Acidification Rate by Microphysiometry: Impact of Reoxygenation. The Journal of Pharmacology and Experimental Therapeutics. 285, 252–261.

Gross, G. W., Rhoades, B. K., Azzazy, H. M. E., Wu, M. –C., 1995. The Use of Neuronal Networks on Multielectrode Arrays as Biosensors. Biosensors and Bioelectronics. 10 (6–7), 553–567.

Hafeman, D. G., Parce, J. W., and McConnell, H. M., 1988. Light Addressable Potentiometric Sensor for Biochemical Systems. Science. 240, 1182–1185.

Heitzer, A., Malachowsky, K., Thonnard, J. E., Bienkowski P. R., White, D. C., and Sayler, G. S., 1994. Optical Biosensor for Environmental on–line Monitoring of Napthalene and Salicylate Bioavailability with an Immobilized Bioluminescent Catabolic Reporter Bacterium. Applied and Environmental Microbiology. 60 (5), 1487–1494.

Heppner, T. J., and Fiekers, J. F., 1991. Soman Reversibly Decreases the duration of Ca2+ and Ba2+ Action Potentials in Bullfrog Sympathetic Neurons. Brain Research. 563, 303–305.

Heppner, T. J., and Fiekers, J. F., 1992. VX Enhances Neuronal Excitability and Alters Membrane Properties of Rana Catesbeiana Sympathetic Ganglion Neurons. Comparative Biochemistry and Physiology Part C: Comparative Pharmacology. 102 (2), 335–338.

Isola, N. R., Stokes, D. L., and Vo–Dinh, T., 1998. Surface–Enhanced Raman gene Probe for HIV Detection. Analytical Chemistry. 70 (7), 1352–1356.

104 King, J. M. H., DiGrazia, P. M., Applegate, B., Burlage, R., Sanseverino, J., Dunbar, P., Larimer, F., and Sayler, G. S., 1990. Rapid Sensitive Bioluminescent Reporter Technology for Napthalene Exposure and Biodegradation. Science. 249, 778–781.

Kneipp, K., Kneipp, H., and Kneipp, J., 2006. Surface–Enhanced Raman Scattering in Local Optical Fields of Silver and Gold Nanoaggregates: From Single–Molecule Raman Spectroscopy to Ultrasensitive Probing in Live Cells. Accounts of Chemical Research. 39 (7), 443–450.

Korpan, Y. I., Gonchar, M. V., Starodub, N. F., Shul’ga, A. A., Sibirny, A. A., and El’skaya A. V., 1993. A Cell Biosensor Specific for Formaldehyde Based on pH–Sensitive Transistors Coupled to Methylotrophic Yeast Cells with Genetically Adjusted Metabolism. Analytical Biochemistry. 215, 216–222.

Kowtha, V. C., Bryant, H. J., Krauthamer, V., Iwasa, K. H., and Stenger, D. A., 1996. Spontaneous Firing of NG108–15 Cells Induced by Transient Exposure to Ammonium Chloride. Cellular and Molecular Neurobiology. 16 (1), 1–9, 1993.

Kowolenko, M., Keese, C. R., Lawrence, D. A., and Giaever, I., 1990. Measurement of Macrophage Adherence and Spreading with Weak Electric Fields. Journal of Immunological Methods. 127 (1), 71–77.

Kuntzweiler, T. A., Arneric, S. P., and Donnelly–Roberts, D. L., 1998. Rapid Assessment of Ligand Actions with Nicotinic Acetylcholine Receptors Using Calcium Dynamics and FLIPR. Drug Development Research. 44, 14–20.

Ma, W., Pancrazio, J. J., Coulombe, M., Dumm, J., Sathanoori, R. R., Barker, J. L., Kowtha, V. C., Stenger, D. A., and Hickman, J. J., 1998. Neuronal and Glial Epitopes and Transmitter– Synthesizing Enzymes Appear in Parallel with Membrane Excitability during Neuroblastoma X Glioma Hybrid Differentiation. Developmental Brain Research. 106, 155–163.

Meister, M., Pine, J., and Baylor, D. A., 1994. Multi–neuronal Signals from the Retina: Acquisition and Analysis. Journal of Neuroscience Methods. 51 (1), 95–106.

Mitra, P., Keese, C. R., and Giaever, I., 1991. Electric Measurements Can Be Used to Monitor the Attachment and Spreading of Cells in Tissue Culture. Biotechniques. 11, 504–510.

Nicolelis, M. A. L., and Chapin, J. K., 1994. Spatiotemporal Structure of Somatosensory Responses of Many–Neuron Ensembles in the Rat Ventral Posterior Medial Nucleus of the Thalamus. The Journal of Neuroscience. 14 (6), 3511–3532.

Nicolelis, M. A. L., Baccala, L. A., Lin, R. C. S., and Chapin, J. K., 1995. Sensorimotor Encoding by Synchronous Neural Ensemble Activity at Multiple Levels of the Somatosensory System. Science. 268, 1353–1358.

105 Nirenberg, M., Wilson, S. P., Higashida, H., Rotter, A., Kreuger, K., Busis, N., Ray, R., Kenimer, J., Adler, M., and Fukui, H., 1983. Synapse Formation by Neuroblastoma Hybrid Cells. Cold Spring Harbor Symposia on Quantitative Biology. 48, 707–715.

Noiri, E., Hu, Y., Bahou, W. F., Keese, C. R., Giaever, I., Goligorsky, M. S., 1997. Permissive Role of Nitric Oxide in Endothelin–Induced Migration of Endothelial Cells. The Journal of Biological Chemistry. 272 (3), 1747–1752.

Notingher, I., Verrier, S., Haque, S., Polak, J. M., and Hench, L. L., 2003. Spectroscopic Study of Human Lung Epithelial Cells (A549) in Culture: Living Cells Versus Dead Cells. Biopolymers (biospectroscopy), 72 (4), 230–240.

Notingher, I., Selvakumaran, J., Hench, L. L., 2004. New Detection System for Toxic Agents Based on Continuous Spectroscopic Monitoring of Living Cells. Biosensors and Bioelectronics. 20 (4), 780–789.

Notingher, I., 2007. Raman Specroscopy Cell–Based Biosensors. Sensors. 7, 1343–1358.

Owen, C. A., Selvakumaran, J., Notingher, I., Jell, G., Hench, L. L., and Stevens, M. M., 2006. In Vitro Toxicology Evaluation of Pharmaceuticals Using Raman Micro–Spectroscopy. Journal of Cellular Biochemistry. 99 (1), 178–186.

Owicki, J. C., Parce, J. W., Kercso, K. M., Sigal, G. B., Muir, V. C., Venter, J. C., Fraser, C. M., and McConnell, H. M., 1990. Continuous Monitoring of Receptor–Mediated Changes in the Metabolic Rates of Living Cells. Proceedings of National Academy of Sciences USA. 87, 4007–4011.

Owicki, J. C., and Parce, J. W., 1992. Biosensors Based on the Energy Metabolism of Living Cells: The Physical Chemistry and Cell Biology of Extracellular Acidification. Biosensors and Bioelectronics. 7 (4), 255–272.

Pancrazio, J. J., Bey, Jr., P. P., Loloee, A., Manne, S., Chao, H. –C., Howard, L. L., Gosney, W. M., Borkholder, D. A., Kovacs, G. T. A., Manos, P., Cuttino, D. S., and Stenger, D. A., 1998 (a). Description and Demonstration of a CMOS Amplifier–Based–System with Measurement and Stimulation Capability for Bioelectrical Signal Transduction. Biosensors and Bioelectronics. 13 (9), 971–979.

Pancrazio, J. J., Bey, Jr., P. P., Cuttino, D. S., Kusel, J. K., Borkholder, D. A., Shaffer, K. M., Kovacs, G. T. A., and Stenger, D. A., 1998 (b). Portable Cell–Based Biosensor System for Toxin Detection. Sensors and Actuators B. 53 (3), 179–185.

Pancrazio, J. J., Whelan, J. P., Borkholder, D. A., Ma, W., and Stenger, D. A., 1999. Development and Application of Cell–Based Biosensors. Annals of Biomedical Engineering. 27, 697–711.

106

Parce, J. W., Owicki, J. C., Kercso, K. M., Sigal, G. B., Wada, H. G., Muir, V. C., Bousse, L. J., Ross, K. L., Sikic, B. I., and McConnell, H. M., 1989. Detection of Cell–Affecting Agents with a Silicon Biosensor. Science. 246, 243–247. 1989.

Rainina E. I., Efremenco, E. N., Varfolomeyev, S. D., Simonian, A. L., Wild, J. R., 1996. The Development of a New Biosensor Based on Recombinant E. coli for the Direct Detection of Organophosphorus Neurotoxins. Biosensors and Bioelectronics. 11 (10), 991–1000.

Raley-Susman, K. M., Miller, K. R., Owicki, J. C., and Sapolsky. R. M., 1992. Effects of Excitotoxin Exposure on Metabolic Rate of Primary Hippocampal Cultures: Application of Silicon Microphysiometry to Neurobiology. The Journal of Neuroscience. 12 (3), 773–780.

Selifonova, O., R. Burlage, and Barkay, T., 1993. Bioluminescent Sensors for Detection of Bioavailable Hg(II) in the Environment. Applied and Environmental Microbiology. 59 (9), 3083–3090.

Skeen, R. S., Kisaalita, W. S., and Van Wie, B. J., 1990. Evaluation of Neuron–Based Sensing with the Neurotransmitter Serotonin. Biosensors and Bioelectronics. 5, 491–510.

Smith, T. J., Wang, H. –S., Hogg, M. G., Henrikson, R. C., and Keese, C. R., 1994. Prostaglandin

E2 Elicits a Morphological Change in Cultured Orbital Fibroblasts from Patients with Graves Ophthalmopathy. Proceedings of the National Academy of Sciences USA. 91 (11), 5094– 5098.

Sticher, P., Jaspers, M. C. M., Stemmler, K., Harms, H., Zehnder, A. J. B., and van der Meer, J. R., 1997. Development and Characterization of a Whole–Cell Bioluminescent Sensor for Bioavailable Middle–Chain Alkanes in Contaminated Groundwater Samples. Applied and Environmental Microbiology. 63 (10), 4053–4060.

Stoppini, L., Duport, S., and Correges, P., 1997. A New Extracellular Multirecording System for Electrophysiological Studies: Application to Hippocampal Organotypic Cultures. Journal of Neuroscience Methods. 72 (1), 23–33.

Sullivan, E., Tucker, E. M., and Dale, I. L., 1999. Measurement of [Ca2+] using the Fluorometric Imaging Plate Reader (FLIPR). Methods in Molecular Biology. 114, 125–133.

Takahashi, T., 1996. Two Statistical Methods for Analyzing Multiple Neuronal Data. Internatinal Journal of Neuroscience. 88, 11–26.

Tiruppathi, C., Malik, A. B., Del Vecchio, P. J., Keese, C. R., and Giaever, I., 1992. Electrical Method for Detection of Endothelial Cell Shape Change in Real Time: Assessment of Endothelial Barrier Function. Proceedings of the National Academy of Sciences USA. 89 (17), 7919–7923.

107 Tlili, C., Reybier, K., Géloën, A., Ponsonnet, L., Martelet, C., Ouada, H. B., Lagarde, M., and Jaffrezic–Renault, N., 2003. Fibroblast Cells: A Sensing Bioelement for Glucose Detection by Impedance Spectroscopy. Analytical Chemistry. 75 (14), 3340–3344.

Tsien, R. Y., 1988. Fluorescence Measurement and Photochemical Manipulation of Cytosolic Free Calcium. Trends in Neurosciences. 11, 419–424.

Verrier, S., Notingher, I., Polak, J. M., and Hench, L. L., 2004. In Situ Monitoring of Cell Death Using Raman Microspectroscopy. Biopolymers. 74 (1–2), 157–162.

Wegener, J., Keese, C. R., and Giaever, I., 2000. Electric Cell–Substrate Impedance Sensing (ECIS) as a Noninvasive Means to Monitor the Kinetics of Cell Spreading to Artificial Surfaces. Experimental Cell Research. 259 (1), 158–166.

Wolf, B., Brischwein, M., Baumann, W., Ehret, R., and Kraus, M., 1998. Monitoring of Cellular Signalling and Metabolism with Modular Sensor–Technique: The PhysioControl– Microsystem (PCM®). Biosensors and Bioelectronics. 13 (5), 501–509.

Yao, H., Tao, Z., Ai, M., Peng, L., Wang, G., He, B., and Li, Y. –Q., 2009. Raman Spectroscopic Analysis of Apoptosis of Single Human Gastric Cancer Cells. Vibrational Spectroscopy. 50 (2), 193–197.

Zysk, J. R., and Baumbach, W. R., 1998. Homogeneous Pharmacologic and Cell–Based Screens Provide Diverse Strategies in Drug Discovery: Somatostatin Antagonists as A Case Study. Combinatorial Chemistry and High Throughput Screening. 1 (4), 171–183.

108 Chapter 6 SONOPORATION: ULTRASOUND–INDUCED PERMEABILIZATION OF CELL MEMBRANE

6.1. Basic Physics of Ultrasound

Ultrasound is cyclic sound pressure with a frequency above 20 kHz, i.e., the upper limit for human hearing [Shutilov, 1988]. It consists of a mechanical vibration of the particles or molecules of a material. Within the wave, regular pressure variations occur with alternating areas of compression, which correspond to areas of high pressure, and with areas of rarefaction or low pressure zones where widening of particles occurs.

Although each particle moves small distances from its rest position, the vibrational energy is propagated as a wave traveling from particle to particle through the material.

Sound waves are expressed as sine waves and classified as being longitudinal or transverse, depending on whether the vibration of each particle is parallel or transverse to the direction of wave propagation. Although all materials can support the propagation of longitudinal sound waves, only solids can support transverse waves. Parameters of an ultrasound wave include frequency, pressure, wavelength, velocity, speed, power, and intensity. In general, the higher the frequency of the ultrasound, the smaller is the beam divergence and the narrower is the beam width. Ultrasound pressure, which can be measured using a microphone in air or a hydrophone in water, is the local pressure deviation from the ambient equilibrium pressure caused by the ultrasonic wave.

Sound velocity, which is a vector quantity, is the rate at which the vibratory energy is transmitted in a specified direction of propagation. Acoustic speed, a scalar quantity, is greater in materials that are more rigid or less compressible. Power and intensity are measures of the strength of an ultrasound wave. Power is the total amount of energy

109 passing through a surface per unit time. Intensity is the concentration of energy per unit area (i.e., the energy pass through per unit area per unit time). Therefore, intensity changes depending on the width of the ultrasound beam. In focused ultrasound beams, the intensity is greatest at the focus where the beam width is the narrowest. In pulsed ultrasound beams, the greatest intensity value occurs during the pulse and is zero between pulses.

How ultrasound traverses a medium is measured by characteristic acoustic impedance of a medium, which is a material property and depends on the density of the medium and the propagation velocity of the ultrasound through the medium (i.e., Z0 =

ρ⋅c). Ultrasound is attenuated as it travels through the medium due to beam divergence

(diffraction), absorption, and deflection of acoustic energy out of the beam. The

deflection includes the processes of reflection, refraction, and scattering. An echo is a reflected wave and its magnitude depends on: (i) the orientation of the reflecting surface with respect to the sound beam, and (ii) the difference in acoustic impedance between media on either side of the reflecting surface (i.e., acoustic impedance mismatch). When an ultrasound beam encounters media of different velocities, the proportion of the beam that is not reflected but is transmitted undergoes refraction or bending (the change in the direction of transmitted beam is governed by the Snell’s law). Ultrasound is also nonionizing, i.e., it does not carry enough energy to completely remove an electron from an atom or a molecule. However, at sufficiently high intensities, ultrasound can produce temperature elevation, mechanical effects, cavitation, and chemical effects.

6.2. Ultrasonic Transducers

Ultrasound can be produced and detected using a transducer that employs piezoelectric materials. In order to generate an ultrasound with a transducer, an alternating voltage is applied across the piezoelectric structure. The structure will

110 lengthens or shortens depending on the polarity of applied voltage and in proportion to the strength of the applied electric field. Consequently, this mechanical vibration in the range of ultrasonic frequencies generates an ultrasound. The reverse effect occurs when the ultrasound hits the piezoelectric structure, i.e., the mechanical vibrating energy induced by the ultrasound on the piezoelectric structure is converted into the electrical energy. The beam pattern of ultrasonic energy emanating from the transducer can be determined by the dimensions of the transducer, the way it is constructed, the way it is energized, and the driving frequency [Kossoff, 2000].

The piezoelectric material used in the ultrasonic transducer is a piece of perovskite crystal (a polarized material). When an electric field is applied across the material, the polarized molecules will be oriented and align themselves with the electric field, resulting in induced dipoles within the molecular or crystal structure of the material.

Consequently, this alignment of molecules causes the material to change dimensions

(phenomenon known as electrostriction). Also, the piezoelectric material produces an electric field when the material changes dimensions as a result of an imposed mechanical force (phenomenon known as the piezoelectric effect). Tension and compression generate voltages of opposite polarity in proportion to the applied force. However, magnitudes of piezoelectric voltages and movements are usually small. Therefore, mechanical or electrical amplification mechanism is usually integrated to the transducer system depending on applications. Among various type piezoelectric materials, Lead–

Zirconate–Titanate–based (PZT–based) compounds are widely used in ultrasonic

transducers as it exhibits greater sensitivity and higher operating temperatures.

111 6.3. Acoustic Cavitation

Acoustic cavitation is the formation and/or activity of gas– or vapor–filled bubbles in a medium exposed to an ultrasound field [Apfel, 1982; Barnett, et al., 1994; Dalecki,

2004; Ter Haar et al., 1986]. According to the crevice model, it is suggested that gas nuclei, stabilized in crevices of hydrophobic impurities in the liquid, expand and separate from the impurities to form micro–bubbles as the pressure in the liquid decreases

[Flynn, 1964; Young, 1989]. When the ultrasonic wave passes through the liquid, gas bubbles of any size will expand at low pressure and contract at high pressure. The bubble radius varies about an equilibrium value [Dalecki, 2004], and the pulsating cavity exists for a number of cycles [Flynn, 1982]. If the resulting oscillation in the bubble is stable (i.e., repeatable over many acoustic cycles) and the amplitude changes in the bubble size during each acoustic cycle, as with rectified diffusion (i.e., the slow growth of an oscillating bubble due to a net flow of gas into the bubble over many acoustic cycles) and resonant bubble motion, is relatively moderate or does not exceed twice the equilibrium radius, the cavitation is classified as stable or non–inertial cavitation

[Apfel, 1982; Dalecki, 2004]. Such oscillation in response to the ultrasound creates time–independent second–order quantities [Nyborg, 1982]; these include (i) radiation pressure: an elevation or a decrease of the steady pressure near the oscillating bubble relative to that existing in the absence of oscillation; (ii) radiation force: a steady force on a body near the vibrating bubble, tending to deform the body or make it move; (iii) radiation torque: a steady twisting action on a particle near the oscillating bubble, tending to make the article rotate; and micro–streaming: a circulatory motion set up in the liquid near a vibrating bubble. The velocities and shear rates associated with the micro–streaming around the bubble are also proportional to the amplitude of the oscillation [Elder, 1958; Marmottant & Hilgenfeldt, 2003; Nyborg, 1982]. These bubble–associated small–scale events can occur at ultrasonic frequencies in the lower megahertz range (1–10 MHz) and even low levels of intensity or pressure [Nyborg, 1982].

112 As the ultrasonic intensity increases, the amplitude of oscillation also increases to a point at which the inward moving wall of fluid has sufficient inertia that it cannot reverse the direction when the acoustic pressure reverses, but continues to compress the gas in the bubble to a very small volume, creating extremely high pressures and temperatures

[Barnett et al., 1994; Brennen, 1995; May et al., 2002]. Such bubbles oscillate about

their equilibrium size, grow until the outward excursion of the surface exceeds a limiting

value (approximately 2 times the initial radius [Flynn, 1982]), and collapse violently.

This type of cavitation is classified as transient, inertial or collapse cavitation. A small increase in acoustic pressure amplitude can change the bubble response from a non–inertial cavity to an inertial cavity. The acoustic pressure, at which the transition occurs, is often called the threshold for inertial cavitation [Dalecki, 2004]. Furthermore, the collapse cavitation can also occur during the period of a single cycle [Barnett et al.,

1994; Flynn, 1982; Fowlkes & Crum, 1988] or a few cycles, depending on the applied ultrasonic pressure and the properties of the propagating medium. In general, the likelihood and intensity of collapse cavitation increases at higher ultrasound intensities and lower frequency [Apfel & Holland, 1991; Brennen, 1995; Urick, 1983]. Additionally, for appropriate exposure conditions, a single ultrasound pulse on the order of one microsecond in duration can also cause a bubble to expand rapidly and collapse violently

[Flynn, 1982]. The size of the bubble and its physical properties, i.e., gas species, interfacial tension, surface rigidity, etc., also affect this cavitation process [Allen et al.,

2003; Kvikliene et al., 2004; Sboros et al., 2003; Soetanto & Chan, 2000]. The motion of the bubble in inertial cavitation is highly nonlinear [Flynn, 1982] and various parameters, including acoustic frequency, pressure amplitude, and initial bubble radius, determine the violence of the collapse of the inertial cavity.

The collapsed bubble often fragments into smaller bubbles that serve as cavitation nuclei, grow in size, and eventually collapse again [Brennen, 1995; May et al., 2002].

113 High temperatures, high shear forces, and shock waves are generated by the sudden collapse of the cavitation bubble. If a sufficiently high temperature is generated (>1500

˚K), the released energy may even cause the emission of light and the formation of

reactive chemical species [Barnett et al., 1994]. Although the amount of energy released

is approximately 100 MeV [Apfel, 1986], the effects of collapse cavitation are

instantaneous and highly localized [Barnett et al., 1994]. Hence, this type of cavitation

concentrates the energy from ultrasound into a small volume [Apfel, 1982]. If the

collapse is near a relatively rigid surface, an asymmetrical collapse will occur, which

ejects a liquid jet at sonic speed toward the surface [Brennen, 1995; Smith, 2007].

Acoustic cavitation phenomena are summarized in figure 6–1.

Figure 6–1. Acoustic cavitation. (a) Acoustic streaming or micro–streaming: cavitation bubbles can oscillate around their resonant/equilibrium size and generate radiation pressures and forces. Bubble oscillations can be damped through viscous dissipation, sound radiation, and thermal conduction [Flynn, 1964; Leighton, 1994; Young, 1989]. (b) Sonochemistry: sudden collapse of bubbles generates momentary high temperatures in the bubble core. The hot bubble can induce chemical changes in the surrounding medium, including free–radical generation. (c) Shock waves: sudden collapse of cavitation bubbles leads to the formation of shock waves. (d) Liquid microjets: collapsing bubbles near a surface experience non–uniformities in their surroundings resulting high–velocity microjets at sonic speed. After [Mitragotri, 2005].

114 6.3.1. Biological and Physical Consequences of Cavitation on Cells

The liquid medium where bubbles are undergoing collapse cavitation is not a healthy environment for cells as: (i) shock waves and shear forces will be shearing the cell membrane [Guzman et al., 2003]; (ii) liquid microjets at sonic velocities may be piercing or lysing the cells; (iii) the free radicals may be interfering with essential biochemical processes. On the other side, bubbles experiencing mild–stable cavitation have no negative biological consequences to most cells. Fortunately, somewhere between the harsh and mild–state cavitation lies the desired realm of cavitational level that produces bubble–associated activities, which are sufficient to permeabilize plasma membranes without killing the cells [Pitt et al., 2004]. At this cavitational level, the transport or uptake of gene products or therapeutic molecules into the cells can be enhanced.

6.4. Sonoporation

Sonoporation is a technique that enhances cell membrane permeability via the application of ultrasound [Mehier–Humbert & Guy, 2005]. The consensus is that acoustic cavitation in response of ultrasound exposure is the most probable mechanism involved in sonoporation. Ultrasound by itself, in the absence of cavitation, has been

thought to have little effect on cells and tissues apart from some heating that may occur

at higher frequencies and intensities [Barnett et al., 1994; Barnett et al., 1997; Nyborg,

2001; Ziskin & Barnett, 2001]. Cells in an environment of cavitation events are

subjected to shear forces from micro–streaming and shock waves. Furthermore, a large semi rigid cell adjacent to a small cavitating bubble could also induce an asymmetric bubble collapse, by which a small jet of liquid would impinge directly into the cell, rupturing the cell membrane.

115 6.4.1. Mechanistic Studies of Plasma Membrane Sonoporation

It has been demonstrated that cavitation activities generated by ultrasound facilitate cellular incorporation of macromolecules up to 28 nm in radius through repairable micron–scale disruptions/holes in the plasma membrane with lifetimes of >1 min, which is a period similar to the kinetics of membrane repair after mechanical wounding

[Schlicher et al., 2006]. Figure 6–2 shows confocal micrographs of prostate cancer cells

(DU 145) exhibiting uptake of calcein, bovine serum albumin, and dextrans of various atomic weights into the cytosol after ultrasound exposure (20 acoustic pulses each at 24 kHz, 7 atm, and 0.1 s pulse length at 10% duty cycle. The cells were sonicated in buffer and then incubated in growth media that contained uptake marker molecules). Scanning

Electron Microscope (SEM) and Transmission Electron Microscope (TEM) images

(figure 6–3) confirm the evidence of structural changes in the plasma membrane of prostate cancer cells (DU 145) due to ultrasound as well as the evidence of healing or resealing of holes/disruptions in the plasma membrane.

Figure 6–2. Intracellular delivery induced by ultrasound. Confocal micrographs showing a nonsonicated DU 145 cell exposed to calcein (1) and sonicated cells exhibiting uptake of calcein (2), bovine serum albumin (3) and 150 (4), 500 (5) and 2,000 kDa (6) dextrans. Scale bars are 1µm. After [Schlicher et al., 2006].

116

Figure 6–3. Ultrastructural analysis of ultrasound’s effect on plasma membrane. (a) Evidence of structural changes in plasma membrane due to ultrasound: SEM of (a1) nonsonicated DU 145 cell and (a2, a3) sonicated cell, shown at two levels of magnification, with a region lacking characteristic membrane surface topography and revealing exposed cytoskeleton; and (b, c) evidence of repairing wounds: SEM of (b1) nonsonicated cell and (b2, b3) sonicated cell, and TEM of (c1) nonsonicated cell and (c2, c3) sonicated cell, each with membrane disruption associated with numerous vesicles believed to be of intracellular origin and that may facilitate active membrane resealing. Cells were fixed 2 seconds after sonication using EM–grade glutaraldehyde [Castejon et al. 2001]. Scale bars are 1 µm. After [Schlicher et al., 2006].

Furthermore, confocal fluorescent microscopy conducted after labeling plasma membrane with red–fluorescent (TRITC) wheat–germ lectin (figure 6–4a) and labeling intracellular vesicles with green fluorescent FM 1–43 (figure 6–4b) have led to the conclusion that the DU 145 cells reseal holes/disruptions in the membrane using a native healing response involving endogenous vesicle–based membrane resealing, which is a energy–dependent process [Schlicher et al., 2006]. Observation of rat mammary carcinoma cells (MAT B III) and red blood cells under scanning electron Microscope after insonification with a focused transducer in the presence of ultrasound contrast agent also shows ultrasound–induced pores in the cell membrane (figure 6–5).

117

Figure 6–4. Confocal fluorescence and brightfield microscopy of intracellular uptake and wound repair. (a) Labeling plasma membrane with red–fluorescent (TRITC) wheat–germ lectin [Sharon and Lis, 1989] and sonicating in presence of calcein, a green–fluorescent transport marker, yields (a1, a2) nonsonicated DU 145 cell with intact membrane labeling and absence of intracellular calcein, which is expected in a normal, viable cell; (a3, a4) sonicated cell with a region of cell membrane removed and uptake of calcein imaged in cells fixed 2 seconds after sonication; and (a5, a6) sonicated cell with intact membrane labeling and uptake of calcein imaged in cells fixed 5 minutes after sonication; and (b) labeling intracellular vesicles with green fluorescent FM 1–43 [Cochilla et al., 1999] and sonicating in the presence of propidium iodide, a red–fluorescent transport marker that stains nuclei, yields (b1, b2) nonsonicated cell containing intracellular vesicles with intact plasma membrane; and (b3, b4) sonicated cell with partially depleted intracellular vesicles and plasma membrane that was breached, as shown by intracellular labeling with propidium iodide. Scale bars shown are 1 µm. After [Schlicher et al., 2006].

Figure 6–5. Direct observation of ultrasound–induced pores by SEM. (A) Rat mammary carcinoma cells (MAT B III), at a concentration of 1 × 106 cells/ml, and (B) red blood cells, at a concentration of 12.4 × 106 cells/ml, were insonified with a focused transducer (2.25 MHz) at a peak negative pressure of 570 kPa, in the presence of UCA (25 particles/cell for MAT B III and 1.2 particles/cell for red blood cells). Cells were observed with a scanning electron microscope after gold sputtering at a magnification of 10,000. After [Mehier–Humbert et al., 2005 (b)].

118 The kinetics of cell membrane permeabilization induced by the sonoporation is a transient phenomenon that can be studied using the loading of fluorescent molecules of various sizes [Hallow et al., 2006; Mehier–Humbert et al., 2005 (a); Mehier–Humber et

al., 2005 (b); Juffermans et al., 2006; Korosoglou et al., 2006; Pan et al., 2005; Schlicher

et al., 2006; Sundaram et al., 2003; Zarnitsyn and Prausnitz et al., 2004; van Wamel et

al., 2004; van Wamel et al., 2006] or by measuring changes in ionic conductivity of the

plasma membrane [Deng et al., 2004]. These studies concluded that estimates of the cell

membrane recovery time range from a few seconds to at most a few minutes, with

differing kinetics for small and larger molecules, and some evidence of separate pore

populations that close at fundamentally different rates. Also, the degree of sonoporation

can be varied through changes in ultrasound conditions such as pulse repetition

frequency and ultrasound intensity. Estimates of pore size in the cell membrane, based

on the physical diameter of uptake marker molecules or compounds, are commonly in

the range of 30–100 nm.

It is also suggested that ultrasound exposure induces up to micron diameter

disruptions/wounds in the plasma membrane, but these wounds have a sieve–like

character that only allows passage of molecules considerably smaller than the diameter

of the disruptions/wounds [Schlicher et al., 2006]. Theoretical modeling, supported

with experimental studies, also predicted that membrane wounds would have a 300–nm

radius initially and then would shrink, with a half–life of 20 to 50 seconds [Zarnitsyn et al., 2008]. The uptake of molecules was shown to occur predominantly by diffusion and the increasing levels of uptake with decreasing molecular size (figure 6–6) was explained primarily by differences in molecular diffusivity and, for the larger molecules,

geometrical hindrance within the sieve–like wound [Zarnitsyn et al., 2008].

Furthermore, as shown in figure 6–6, during the cell membrane recovery, intracellular transport decreased over time after sonication due to resealing of the membrane (Note:

119 This was characterized by post–sonication (5–10 minutes after sonication) intracellular concentrations of uptake marker molecules that were added into the solution at known increments of time from 0–240 seconds in 15 seconds intervals after the cells were sonicated at time = 0 seconds. Experiments for measuring each post–sonication intracellular concentration were conducted separately) and this decay (or decrease in intracellular transport) was more rapid for larger molecules. Furthermore, it is also suggested that the transport of molecules through such porous wounds occurs largely at the edge of the wounds [Zarnitsyn et al., 2008].

Figure 6–6. Intracellular concentration as a function of time after sonication for calcein (623 Da, diamond), bovine serum albumin (66 kDa, square), and dextrans (150 kDa, triangle; 500 kDa, circle; 2,000 kDa, star). Cells were either sonicated with an uptake marker compound (at time = 0 s) or in buffer alone, after which a marker compound was added at different times after sonication. Intracellular solute concentration, determined by calibrated flow cytometry [Prausnitz et al., 1993], is normalized relative to its extracellular concentration and reported as the average value among viable cells affected by ultrasound. Inset shows expanded view of dextran uptake at short times. After [Schlicher et al., 2006].

120 Ultrasound contrast agents (e.g., Albunex, Optison, Sonovue, etc.), which consist of elastic and compressible gas–filled microbubbles and serve as artificial cavitation nuclei, can be used to promote ultrasound–mediated cavitation [Bao et al., 1997; Greenleaf et al., 1998; Mehier–Humbert & Guy, 2005; Miller and Quddus, 2001]. Bubble implosion from the contrast–agent–assisted cavitation in response to ultrasound may form localized liquid jets that would function like a random array of micro–needles, opening pathways through the cell membrane [Miller, 2000].

Studies on the microbubble–induced cell deformation and enhanced cell membrane permeability suggest that vibrating microbubbles, which are in the close vicinity to the cell, poke the cell membrane, forming hydrophobic or hydrophilic pores (as illustrated in figure 6–7b) [Wamel et al., 2004; Wamel et al., 2006]. In such cases, studies indicated that non–inertial cavitation–induced mechanisms are strong enough to cause the transient pore formation in the membrane and this permeabilization lasted for a short period without affecting cells viability. It is also suggested that pushing and pulling by microbubble in the close vicinity to the cell, which correlates to the rarefraction and compression phase of ultrasound, are suggested to be major contributing mechanism in the formation of pores [Wamel et al., 2006]. It may not be an absolute requirement to induce transient cavitation to achieve sonoporation in such cases. Furthermore, it was demonstrated that microbubbles are highly effective at concentrating and focusing low– intensity long–wavelength ultrasound so that it has profound effects at the micron or submicron level [Marmottant and Hilgenfeldt, 2003]. Stable and gentle oscillations of the microbubble without causing the bubble collapse can generate microstreamings and shear forces that are able to induce disruption of lipid bilayers. These studies indicate that close proximity between microbubbles and cell membranes facilitates sonoporation, suggesting that bringing microbubbles nearer to cells using acoustic radiation pressure and/or by ligand–targeting could improve the efficiency of membrane permeabilization and/or decrease the required acoustic power and the dependence on transient cavitation.

121

Figure 6–7. (A) Schematic diagram illustrating the effects of acoustic fields of identical frequency but differing intensity on microbubble (MCB) behavior. Low–intensity ultrasound induces oscillation of pre–existing MCB, with a gradual increase in its diameter until a resonant diameter is reached, when stable oscillation occurs (filled circles). (Note: MCB initially grows in size when being insonified primarily because the surface area for dissolved gas to enter the MCB during the expansion (rarefaction) phase is greater than that available for gas to diffuse out of the MCB during the compression phase, a process known as rectified diffusion). At higher intensities, the MCB grows rapidly for a few cycles. Very soon, however, the inertial energy of the fluid surrounding the MCB during the compression half cycle becomes so great that it cannot reverse direction when the next rarefaction half cycle arrives. It continues to rush in and forcibly collapses the MCB, generating highly localized extremes of temperature and pressure. This generates shock waves, free radical production and local heat. After [Newman and Bettinger, 2007]. (B) Proposed model of the oscillating MCB–enforced pore formation in the cell membrane. The pushing and pulling behavior of the microbubble causes rupture of the cell membrane creating a hydrophilic pore allowing trans– membrane flux of fluid and macromolecules. After [Wamel et al., 2006].

122 6.4.2. Gene Therapy Prospects

Sonoporation is used for cell transfection in vitro [Bao et al., 1997; Greenleaf et al.,

1998]. The method offers flexibility and improves transfection efficiency and cell viability in some cell lines [Pepe et al., 2004]. Although ultrasound covers a broad range of frequencies and waveforms, continuous sinusoidal or pulsed ultrasound probes at megahertz frequencies are usually used for sonoporation and lower frequencies (e.g., 20 kHz) are mainly used for cell lysis and disruption [Mehier–Humbert and Guy, 2005].

Piezoelectric–based transducers (Figure 6–8 and 6–9), lithotriper [Bao et al., 1998;

Miller et al., 1998], or clinical imaging system can be used to generate ultrasound [Miller and Quddus, 2001] for sonoporation of large population of cell suspensions.

Figure 6–8. A schematic representation of the transducer setup used for ultrasound application to cell suspension. The transducers in this setup were designed by sandwiching ceramic crystals between two metal resonators of appropriate lengths. A signal generator along with an amplifier was used to drive the transducers. The electric power applied to the transducer was measured using a sampling wattmeter. Transducers can be calibrated using laser interferometry and hydrophone measurements. The transducers were directly immersed in the cell suspension as shown and the tip of the transducer was located at the center of the well. After [Sundaram et al., 2003].

123

Figure 6–9. An apparatus used to expose cells with ultrasound. A 35–mm–diameter 1.0–MHz air–backed transducer was placed 2 mm below each of two wells of a six–well culture plate. The experiment was conducted within a distilled and degassed water bath maintained at 37 degree Celsius. Water completed the acoustic coupling with the bottom of the culture plate. After [Greenleat et al., 1998].

Studies also show that the presence of ultrasound contrast agents in the medium

(i.e., the cell suspension) during the sonication significantly enhances the ultrasound– mediated gene transfection [Greenleaf et al., 1998; Guzmán et al., 2001]. Furthermore,

it was also found that the degree of cell permeability and the level of gene transfection

increase (in a somewhat linear fashion) with ultrasound pressure, exposure time, and

numbers of pulses in the expense of cell viability above the threshold pressure (Figure

6–10) [Greenleaf at al., 1998; Guzmán et al., 2001].

124

Figure 6–10. (A) Transfection rate of living cells after ultrasound exposure as function of average peak pressure of the 20–s burst of 1.0–MHz ultrasound. The DNA concentration was 40 µg/mL and the Albunex concentration was 50 × 106 bubbles/mL prior to exposure. Continuous line is best linear fit. (B) Transfection rate of living cells after exposure plotted against Albunex concentration at the time of exposure. Exposure time is 20 seconds at the indicated average peak pressure. Continuous lines are hand drawn. (C) Transfection rate of live cells after repeated 1–s exposures to ultrasound at the indicated average peak pressure. Prior to each exposure, 50 × 106 microbubbles were added to the 1 mL of media in each well. (D) Transfection rate after 20–s exposure to 1.0–MHz ultrasound at indicated average peak pressure with a microbubble concentration of 50 × 106 bubbles/mL prior to exposure. After [Greenleaf et al., 1998].

Although early transfection results using sonoporation have achieved 10– to 30–fold increases in transfection efficiency in vitro, technical refinements on important parameters such as ultrasound intensity, pulse repetition frequency, duration of exposure (i.e., ultrasound pulse duration), microbubble concentrations, and etc. have shown enhancements of up to several thousand fold in vitro [Akowuah et al., 2005; Fischer et al., 2006; Guo et al., 2006; Liang et al., 2004; Mehier–Humbert et al., 2005 (a); Michel et al., 2004; Rahim et al., 2006 (a); Rahim et al., 2006 (b); Zhou et al., 2005]. Not only are these findings sufficient to encourage studies of ultrasound enhanced gene transfer in vivo, but also render promising prospects for ultrasound enhanced gene therapy in the future.

125 References

Akowuah, E. F., Gray, C., Lawrie, A., Sheridan, P. J., Su, C. –H., Bettinger, T., Brisken, A. F., Gunn, J., Crossman, D. C., Francis, S. E., Baker, A. H., and Newman, C. M., 2005. Ultrasound–Mediated Delivery of TIMP–3 Plasmid DNA into Saphenous Vein Leads to Increased Lumen Size in a Porcine Interposition Graft Model. Gene Therapy. 12, 1154–1157.

Allen, J. S., Kruse, D. E., Dayton, P. A., and Ferrara, K. W., 2003. Effect of Coupled Oscillations on Microbubble Behavior. The Journal of the Acoustical Society of America. 114 (3), 1678– 1690.

Apfel, R. E., 1982. Acoustic Cavitation: A Possible Consquence of Biomedical Uses of Ultrasound. British Journal of Cancer. 45 (Suppl. V), 140–146.

Apfel, R. E., 1986. Possibility of Microcavitation from Diagnostic Ultrasound. IEEE Transactions on Ultrasonic, Ferroelectrics, and Frequency Control. 33 (2), 139–142.

Apfel R. E., and Holland C. K., 1991. Gauging the Likelihood of Cavitation from Short–Pulse, Low–Duty Cycle Diagnostic Ultrasound. Ultrasound in Medicine and Biology. 17 (2), 179– 185.

Bao, S., Thrall, B. D., and Miller, D. L., 1997. Transfection of a Reporter Plasmid into Cultured Cells by Sonoporation in vitro. Ultrasound in Medicine and Biology. 23 (6), 953–959.

Bao, S., Thrall, B. D., Gies, R. A., and Miller, D. L., 1998. In Vivo Transfection of Melanoma Cells by Lithotripter Shock Waves. Cancer Research. 58, 219–221.

Barnett, S. B., Ter Haar, G. R., Ziskin, M. C., Nyborg, W. L., Maeda, K., and Bang, J., 1994. Current Status of Research on Biophysical Effects of Ultrasound. Ultrasound in Medicine and Biology. 20 (3), 205–218.

Barnett, S. B., Rott, H. –D., Ter Haar, G. R., Ziskin, M. C., and Maeda, K., 1997. The Sensitivity of Biological Tissue to Ultrasound. Ultrasound in Medicine and Biology. 23 (6), 805–812.

Brennen, C. E., 1995. Cavitation and Bubble Dynamics. Oxford University Press, New York.

Dalecki, D., 2004. Mechanical Bioeffects of Ultrasound. Annual Review of Biomedical Engineering. 6, 229–248.

Deng, C. X., Sieling, F., Pan, H., and Cui, J., 2004. Ultrasound–Induced Cell Membrane Porosity. Ultrasound in Medicine and Biology. 30 (4), 519–526.

Elder, S. A., 1958. Cavitation Microstreaming. The Journal of the Acoustical Society of America. 31 (1), 54–64.

126 Fischer, A. J., Stanke, J. J., Omar, G., Askwith, C. C., and Burry, R. W., 2006. Ultrasound– Mediated Gene Transfer into Neuronal Cells. Journal of Biotechnology. 122, 393–411.

Flynn, H. G., 1964. Physics of Acoustic Cavitation in Liquids. Physical Acoustics (editor: Mason, W. P.), vol. I, part B. Academic Press, New York.

Flynn, H. G., 1982. Generation of Transient Cavities in Liquids by Microsecond Pulses of Ultrasound. The Journal of the Acoustical Society of America. 72 (6), 1926–1932.

Fowlkes, J. B., and Crum, L. A., 1988. Cavitation Threshold Measurements for Microsecond Length Pulses of Ultrasound. The Journal of the Acoustical Society of America. 83 (6), 2190– 2201.

Greenleaf, W. J., Bolander, M. E., Sarkar, G., Goldring, M. B., and Greenleaf, J. F., 1998. Artificial Cavitation Nuclei Significantly Enhance Acoustically Induced Cell Transfection. Ultrasound in Medicine and Biology. 24 (4), 587–595.

Guo, D. –P., Li, X. –Y., Sun, P., Tang, Y. –B., Chen, X. –Y., Chen, Q., Fan, L. –M., Zang, B., Shao, L. –Z., Li, X. –R., 2006. Ultrasound–Targeted Microbubble Destruction Improves the Low

Density Lipoprotein Receptor Gene Expression in HepG2 Cells. Biochemical and Biophysical Research Communications. 343, 470–474.

Guzman, H. R., Nguyen, D. X., Khan, S., and Prausnitz, M. R., 2001. Ultrasound–Mediated Disruption of Cell Membrane. II. Heterogeneous Effects on Cells. The Journal of the Acoustical Society of America. 110 (1), 597–606.

Guzman, H. R., McNamara, A. J., Nguyen, D. X., and Prausnitz, M. R., 2003. Bioeffects Caused by Changes in Acoustic Cavitation Bubble Density and Cell Concentration: A Unified Explanation Based on Cell–to–Bubble ratio and Blast Radius. Ultrasound in Medicine and Biology. 29 (8), 1211–1222.

Hallow, D. M., Mahajan, A. D., McCutchen, T. E., and Prausnitz, M. R., 2006. Measurement and Correlation of Acoustic Cavitation with Cellular Bioeffects. Ultrasound in Medicine and Biology. 32 (7), 1111–1122.

Juffermans, L. J. M., Dijkmans, P. A., Musters, R. J. P., Visser, C. A., and Kamp, O., 2006. Transient Permeabilization of Cell Membranes by Ultrasound–Exposed Microbubbles is related to formation of Hydrogen Peroxide. American Journal of Physiology: Heart and Circulatory Physiology. 291, H1595–H1601.

Korosoglou, G., Hardt, S. E., Bekeredjian, R., Jenne, J., Konstantin, M., Hagenmuller, M., Katus, H. A., and Kuecherer, H., 2006. Ultrasound Exposure Can Increase the Membrane Permeability of Human Neutrophil Granulocytes Containing Microbubbles without Causing Complete Cell Destruction. Ultrasound in Medicine and Biology. 32 (2), 297–303.

127 Kossoff, G., 2000. Basic Physics and Imaging Characteristics of Ultrasound. World Journal of Surgery. 24, 134–142.

Kvikliene, A., Jurkonis, R., Ressner, M., Hoff, L., Jansson, T., Janerot–Sjöbert, B., Lukosevicius, A., and Ask, P., 2004. Modelling of Nonlinear Effects and the Response of Ultrasound Contrast Micro Bubbles: Simulation and Experiment. Ultrasonics. 42, 301–307.

Liang, H. –D., Lu, Q. L., Xue, S. –A., Halliwell, M., Kodama, T., Cosgrove, D. O., Strauss, H. J., Partridge, T. A., and Blomley, M. J. K., 2004. Optimisation of Ultrasound–Mediated Gene Transfer (Sonoporation) in Skeletal Muscle Cells. Ultrasound in Medicine and Biology. 30 (11), 1523–1529.

Marmottant, P., and Hilgenfeldt, S., 2003. Controlled Vesicle Deformation and Lysis by Single Oscillating Bubbles. Nature. 423 (6936), 153–156.

May, D. J., Allen, J. S., and Ferrara, K. W., 2002. Dynamics of Fragmentation of Thick–Shelled Microbubbles. IEEE Transactions on Ultrasonics Ferroelectrics and Frequency Control. 49 (10), 1400–1410.

Mehier–Humbert, S., and Guy, R. H., 2005. Physical Methods for Gene Transfer: Improving the Kinetics of Gene Delivery into Cells. Advanced Drug Delivery Reviews. 57, 733–753.

Mehier–Humbert, S., Bettinger, T., Yan, F., and Guy, R. H., 2005 (a). Ultrasound–Mediated Gene Delivery: Kinetics of Plasmid Internalization and Gene Expression. Journal of Controlled Release. 104, 203–211.

Mehier–Humbert, S., Bettinger, T., Yan, F., and Guy, R. H., 2005 (b). Plasma Membrane Poration Induced by Ultrasound Exposure: Implication for Drug Delivery. Journal of Controlled Release. 104, 213–222.

Michel, M. S., Erben, P., Trojan, L., Schaaf, A., Kiknavelidze, K., Knoll, T., and Alken, P., 2004. Acoustic Energy: A New Transfection Method for Cancer of the Prostate, Cancer of the Bladder and Benign Kidney Cells. Anticancer Research. 24 (4), 2303–2308.

Miller, D. L., Williams, A. R., Morris, J. E., and Chrisler, W. B., 1998. Sonoporation of Erythrocytes by Lithotripter Shockwaves in vitro. Ultrasonics. 36, 947–952.

Miller, M. W., 2000. Gene Transfection and Drug Delivery. Ultrasound in Medicine and Biology. 26 (Suppl. 1), S59–S62.

Miller, D. L., and Quddus, J., 2001. Lysis and Sonoporation of Epidermoid and Phagocytic Monolayer Cells by Diagnostic Ultrasound Activation of Contrast Agent Gas Bodies. Ultrasound in Medicine and Biology. 27 (8), 1107–1113.

128 Mitragotri, S., 2005. Healing Sound: The Use of Ultrasound in Drug Delivery and Other Therapeutic Applications. Nature Reviews Drug Discovery. 4, 255–260.

Newman, C. M. H., and Bettinger, T., 2007. Gene therapy Progress and Prospects: Ultrasound for Gene Transfer. Gene Therapy. 14, 465–475.

Nyborg, W. L., 1982. Ultrasonic Microstreaming and Related Phenomena. British Journal of Cancer. 45 (Suppl. V),156–160.

Nyborg, W. L., 2001. Biological Effects of Ultrasound: Development of Safety Guidelines. Part II: General Review. Ultrasound in Medicine and Biology. 27 (3), 301–333.

Pan, H., Zhou, Y., Izadnegahdar, O., Cui, J., and Deng, C. X., 2005. Study of Sonoporation Dynamics Affected by Ultrasound Duty Cycle. Ultrasound in Medicine and Biology. 31 (6), 849–856.

Pepe, J., Rincon, M., and Wu, J., 2004. Experimental Comparison of Sonoporation and Electroporation in Cell Transfection Applications. Acoustics Research Letters Online. 5 (2), 62–67.

Pitt, W. G., Husseini, G. A., and Staples, B. J., 2004. Ultrasonic Drug Delivery–A General Review. Expert Opinion on Drug Delivery. 1 (1), 37–56.

Prausnitz, M. R., Lau, B. S., Milano, C. D., Conner, S., Langer, R., and Weaver, J. C., 1993. A Quantitative Study of Electroporation Showing a Plateau in Net Molecular Transport. Biophysical Journal. 65, 414–422.

Rahim, A. A., Taylor, S. L., Bush, N. L., ter Haar, G. R., Bamber, J. C., and Porter, C. D., 2006 (a) (a). Physical Parameters Affecting Ultrasound/Microbubble–Mediated Gene Delivery Efficiency in vitro. Ultrasound in Medicine and Biology. 32 (8), 1269–1279.

Rahim, A. A., Taylor, S. L., Bush, N. L., ter Haar, G. R., Bamber, J. C., and Porter, C. D., 2006 (b). Spatial and Acoustic Pressure Dependence of Microbubble–Mediated Gene Delivery Targeted Using Focused Ultrasound. The Journal of Gene Medicine. 8, 1347–1357.

Sboros, V., Moran, C. M., Pye, S. D., and McDicken, W. N., 2003. The Behavior of Individual Contrast Agent Microbubbles. Ultrasound in Medicine and Biology. 29 (5), 687–694.

Schlicher, R. K., Radhakrishna, H., Tolentino, T. P., Apkarian, R. P., Zarnitsyn, V., and Prausnitz, M. R., 2006. Mechanism of Intracellular Delivery by Acoustic Cavitation. Ultrasound in Medicine and Biology. 32 (6), 915–924.

Smith, N. B., 2007. Perspectives on Transdermal Ultrasound Mediated Drug Delivery. International Journal of Nanomedicine. 2 (4), 585–594.

129 Soetanto, K., and Chan, M., 2000. Study on the Lifetime and Attenuation Properties of Microbubbles Coated with Carboxylic Acid Salts. Ultrasonics. 38, 969–977.

Sundaram, J., Mellein, B. R., and Mitragotri, S., 2003. An Experimental and Theoretical Analysis of Ultrasound–Induced Permeabilization of Cell Membranes. Biophysical Journal. 84, 3087– 3101.

Shutilov, V. A., 1988. Fundamental Physics of Ultrasound (translated by Alferieff, M. E.). 1st Edition, CRC Press, New York.

Ter Haar, G. R., Daniels, S., and Morton, K., 1986. Evidence for Acoustic Cavitation In Vivo: Theresholds for Bubble Formation with 0.75–MHz Continuous Wave and Pulsed Beams. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control. 33 (2), 162–164.

Urick, R. J., 1983. Principles of Underwater Sound. 3rd Edition, McGraw–Hill Book Company, San Francisco.

Van Wamel, A., Bouakaz, A., Versluis, M., and de Jong, N., 2004. Micromanipulation of Endothelial Cells: Ultrasound–Microbubble–Cell Interaction. Ultraound in Medicine and Biology. 30 (9), 1255–1258.

Van Wamel, A., Kooiman, K., Harteveld, M., Emmer, M., ten Cate F. J., Versluis, M., de Jong, N., 2006. Vibrating Microbubbles Poking Individual Cells: Drug Transfer into Cells via Sonoporation. Journal of Controlled Release. 112, 149–155.

Young, F. R., 1989. Cavitation. McGraw–Hill Companies, New York.

Zarnitsyn, V., and Prausnitz, M. R., 2004. Physical Parameters Influencing Optimization of Ultrasound – Mediated DNA Transfection. Ultrasound in Medicine and Biology. 30 (4), 527– 538.

Zarnitsyn, V., Rostad, C. A., and Prausnitz, M. R., 2008. Modeling Transmembrane Transport through Cell Membrane Wounds Created by Acoustic Cavitation. Biophysical Journal. 95, 4124–4138.

Zhou, Q. –H., Miller, D. L., Carlisle, R. C., Seymour, L. W., and Oupicky, D., 2005. Ultrasound– Enhanced Transfection Activity of HPMA–Stabilized DNA Polyplexes with Prolonged Plasma Circulation. Journal of Controlled Release. 106, 416–427.

Ziskin, M. C., and Barnett, S. B., 2001. Ultrasound and the Developing Central Nervous System. Ultrasound in Medicine and Biology. 27 (7), 875–876.

130 Chapter 7 DEVICE DESIGNS, FABRICATIONS, AND CHARACTERIZATIONS

This chapter describes chip architectures, designs, fabrications, and characterizations of Integrated Microelectrode Array (IMA) chips and Ultrasonic Micro–Transducer Array

(UMTA) biochips. IMA chips employ planar microelectrode array architecture and are used to electrically characterize (by means of electrical impedance) cellular properties, especially properties of cell membrane, at cellular and single–cell level. Ultrasonic

Micro–transducer array biochips employ piezoelectric–based thickness–mode micro–

transducer arrays, and are used for site–specific permeabilization of cell membrane (or sonoporation) at cellular level.

7.1. Integrated Microelectrode Array (IMA) Chip

7.1.1. Design and Architecture of IMA Chip

The IMA chip consists of an array of sensing/working planar Au microelectrodes

(sizes: 20 µm, 25 µm, 30 µm, and 250 µm in diameter) and very large–dimension

counter–electrodes deposited on a quartz substrate. Quartz substrate is used since it is a

good insulator and mechanically robust. Au is employed as electrode material for three

reasons: (i) it is chemically inert and cannot be easily oxidized below its melting point,

(ii) it is biocompatible [Schreiber, 2004], and (iii) it presents optimal surface chemistry

for further self–assembled monolayer–based bio–functionalizations [Love et al., 2005;

Schreiber, 2004]. The sensing microelectrodes and counter electrodes are electrically

linked to contact/measurement pads located at the chip perimeter by interconnections.

The interconnection metal is insulated from the tissue culture media with SiO2 over–

131 coating. This over–coating provide two functions: (i) it insulates the interconnections

from the media, and (ii) its surface chemistry is employed as a basis for silane–based self–assembled monolayer functionalization and further surface derivatization with bio– molecules [Sagiv, 1980; Ulman, 1996]. The planar electrodes can be exposed to tissue

culture media through via holes. A cell culture well/reservoir, usually made of

polypropylene (PP), can be mounted on the chip using biocompatible glue (such as

Sylgard® or epoxy) to enclose the electrodes. The simplified cross–sectional schematic of the IMA chip and the electrode layout are illustrated in figure 7–1 and 7–2 respectively. The sensing/working electrode and very large counter–electrode can be access electrically via contact pads, which are located at the chip perimeter (tungsten probes or wire–bonding methods are usually employed). After electrodes are immersed in tissue culture media, the effective impedance across the sensing/working and counter electrode can be measured once the equilibrium electrochemical potential is reached at the electrode–electrolyte interface. Since counter–electrodes have very large surface area compared to sensing electrodes and are connected to each others, their effective impedance can be neglected in the overall measurement [Borkholder, 1998].

Figure 7–1. Simplified cross–sectional schematic of integrated microelectrode array (IMA) chip. The surface area of counter–electrodes is very large compared to those of sensing electrodes. Individual electrodes can be access electrically via contact pads. After [Thein et al., 2010].

132

Figure 7–2. The Integrated Microelectrode Array (IMA) chip layout showing microelectrode array area, counter electrodes, interconnection lines, and contact pads.

7.1.2. Microfabrication of IMA Chip

Integrated microelectrode array (IMA) chips were fabricated on a 4–inch–diameter quartz wafer using top–down microfabrication techniques that include laser lithography, contact photolithography, Physical Vapor Metal Deposition (PVD), lift–off process,

Plasma Enhanced Chemical Vapor Deposition (PECVD), O2 passivation, thermal annealing, Reactive Ion Etching (RIE), and wet chemical etching.

7.1.2.1. Photomask Fabrication

Three dark–field photomasks, corresponding to three layers of IMA chip, were fabricated as follows. First, mask design layouts were drawn in L–Edit software. Next, positive photoresist spun on the chrome–coated 5–inch Soda Lime blank masks were exposed under Heidelberg DWL66 laser writer and developed (DWL66 employs laser lithography. It uses 440nm HeCd laser source). Then, transparent master patterns (or

133 clear images with an opaque background) were then defined by wet–chemical etching of exposed chrome followed by photoresist stripping and cleaning. These photomasks are repeatedly used hundreds of time in the IMA chip fabrication without losing the quality of transferred patterns.

7.1.2.2. Chip Fabrication

Before any chip fabrication process, the quartz wafer was cleaned with NanoStripTM

solution at 90 ˚C for 10 min, rinsed with DI water thoroughly, and dried using N2 spray gun. The chip fabrication processes are as follows: Firstly, microelectrodes and

interconnections are created on the wafer. In order to form these structures,

photoresists LOR5A/SPR3012 were spun on the cleaned wafer, UV–exposed (i–line 356

nm) under the photomask consisting of the defined patterns (Karl Suss MA/BA6 contact

photolithography aligner and exposure system was used), and developed. After

removing the residual resists in the exposed regions by an oxygen descum process, a Ti

(125 Å)/Au (4000 Å) twin layer was deposited onto the wafer at a rate of 1.0 Å/s by

electron–beam evaporations (Semicore e–gun evaporator was used. E–gun evaporator is

preferred over thermal evaporator as the way the e–gun evaporators are constructed

usually produces stable deposition rate and pure thin films). The Ti layer was used to

enhance the adhesion between the Au and the quartz substrate. Microelectrodes and

interconnections were then defined by a photoresist lift–off process.

Secondly, interconnections were over–coated with a conformal and electrically– insulating layer. To create such over–coating, a layer of SiO2 (~1 µm thick) was

deposited by Plasma Enhanced Chemical Vapor Deposition (PECVD). Silane (SiH4) and

Nitrous Oxide (N2O) gas mixture was used. The resulting SiO2 layer was subsequently annealed/passivated in the furnace at ~250 ˚C for at least 2 hours under the constant flow of ultra–pure oxygen gas. After the annealing/passivation process, photoresist

134 SPR220 was spun on the wafer, exposed using a second photomask, and developed in order to form a photoresist etch–mask that covers and protects everywhere except for the regions above the electrodes. Cell–sized circular openings (of 20–µm, 25–µm, 30–

µm, and 250–µm diameter), known as via holes over sensing microelectrodes, openings for two large counter electrodes and contact pads through the SiO2 layer were created by

Reactive Ion Etching (RIE) of SiO2 followed by the photomask removal in resist stripper

and subsequent oxygen plasma cleaning. Carbon Tetrafluoride (CF4) and Oxygen (O2) gas mixture was used to form fluorine–based chemistries in RIE chamber (PlasmaTherm

720) in order to etch SiO2.

Thirdly, exposed electrodes were coated with a thin Au layer. To achieve such coatings, photoresists LOR5A/SPR3012 were spun on the wafer, exposed using a third photomask, and developed in order to form opened patterns on top of exposed

electrodes. After the residual resists in the exposed regions were removed by oxygen

descum process, an Au layer (~1000 Å thick) was deposited onto the wafer at a rate of

0.8 Å/s by electron–beam evaporation. The Au coatings on top of the electrodes (or the

second Au layer) were then defined by photoresist lift–off (this Au coatings on top of the

exposed electrodes provide a micro–rough, amorphous Au surface that was free of

plasma–etch–induced surface impurities and defects and was found to lower the

baseline impedance of Au microelectrodes [Thein et al., 2010; figure 7–5]). The chip

fabrication process is summarized in figure 7–3.

135

Figure 7–3. Summary of fabrication process for Integrated Microelectrode Array (IMA) chip. Top–down microfabrication techniques such as contact photolithography, Physical Vapor Metal Deposition (PVD), lift–off process, Plasma Enhanced Chemical Vapor Deposition (PECVD), and Reactive Ion Etching (RIE) are employed.

7.1.2.3. Dicing

Photoresist SP1827 (a 2.7–µm–thick) was spun on the wafer (front side) in order to form a protective layer that limits scratching and contamination during dicing process. A disco saw was used to cut the wafers in order to obtain 12.5 mm × 12.5 mm IMA chips.

After the dicing, photoresist layer was removed using standard resist stripper and

individual chips were cleaned with Nano–StripTM solution at room temperature. Figure

7–4 shows the fabricated IMA chip after being diced.

136

Figure 7–4. Integrated Microelectrode Array (IMA) Chip. (A) Size comparison with a penny. Each IMA chip has a dimension of 12.5 mm × 12.5 mm. (B) The IMA chip with media reservoir well mounted. The polypropylene (PP) well is mounted on the IMA chip using biocompatible glue such as Sylgard®. (C) Bright field image of microelectrode arrays (20–µm, 25–µm, 30–µm, and 250–µm in diameter). Bright field images of microelectrode: (D) 20–µm, (E) 25–µm, (F) 30–µm, and (G) 250–µm in diameter.

137 7.1.3. Characterization of IMA Chip

Two properties of microelectrode are characterized: 1), surface roughness, which is an important property for self–assembled–monolayer–based biofunctionalization for

single–cell immobilization (see section 8.2.1) 2), baseline impedance spectrum, which

is an important property used in transductions of electrical signals from IMA–based sensors in monitoring cellular changes.

First, TappingModeTM Atomic Force Microscopy (AFM) with OTESPA probe tip was

used to characterize the surface roughness of the planar Au microelectrodes. Not only

can AFM with OTESPA probe tip provide the surface topology of the Au microelectrode,

but it can also perform Electric Force Microscopy (EFM) (Note: EFM measurements can

also provide information on whether Au microelectrodes are fully exposed through the

via holes and no residual SiO2 is left on the surface of the electrode after RIE).

In brief, TappingModeTM AFM, a patented technique by Veeco Instruments, is the most commonly used of all AFM modes that maps surface topography by lightly tapping the surface with an oscillating AFM probe tip. The cantilever’s oscillation amplitude changes with sample surface topography, and the topography image is obtained by monitoring these changes and closing the z feedback loop to minimize them.

TappingModeTM advantage is that it eliminates shear forces usually present in the traditional contact mode.

However, EFM is a secondary imaging mode derived from TappingModeTM that

measures electric field gradient distribution above the sample surface. EFM is

performed through LiftModeTM, where two–pass/scan technique that separately

measures topography and another selected property (magnetic force, electric force, etc.) using topographic information to track the probe tip at a constant distance above the surface. In EFM, a voltage may be applied between the probe tip and the sample. The cantilever's resonance frequency and phase change with the strength of the electric field gradient are used to construct the EFM images.

138 The AFM scan result (using an OTESPA probe tip) across 30–µm diameter Au microelectrode is shown in figure 7–5 (the AFM measurement showed that Root Mean

Square (RMS) surface roughness of Au microelectrode is approximately 112 nm).

Figure 7–5. AFM scan image of 30–um diameter Au microelectrode. Root Mean Square (RMS) surface roughness of Au microelectrode is ~112 nm (data shown in green). (Micro–roughness increases the effective surface area that in turn decreases the baseline impedance. However, if the surface is very rough, self–assembled monolayer–based biofunctionalized interface formed on the Au microelectrode for cell adhesion/immobilization will not be homogeneous. Therefore, somewhere between these two extremes, the optimal range of surface roughness exists.)

Second, baseline impedance spectrum for each microelectrode dimension is also

measured in tissue culture media using a Phase sensitive Lock–in amplifier. The baseline impedance depends on the type of tissue culture media, the contents in the tissue culture media, and the diameter of the Au microelectrode (or the surface area of the Au microelectrode). The detailed of baseline impedance spectroscopy including instrumentation setup is discussed in Chapter 8.

139 7.2. Ultrasonic Micro–Transducer Array (UMTA) Biochips

7.2.1. The Choice of Piezoelectric Material for UMTA biochips

Since the discovery of the ferroelectricity of polycrystalline ceramic (BaTiO3) during the early 1940’s, a succession of new piezoelectric materials has been discovered, which includes Pb(Zr1–xTix)O3 (PZT) [Jaffe, 1955] and Pb(B’B”)O3, (B’ = Zn, Mg, In, Yb and Sc

etc., and B” = Nb, Mo, Ta and W) in the form of ceramics and single crystal [Yokomizo et

al., 1970]. Furthermore, after the discovery of PZT, numerous new materials such as

Pb(Zn1/3Nb2/3)O3–PbTiO3 (PZN–PT) and Pb(Mg1/3Nb2/3)O3–PbTiO3 (PMN–PT), and etc. have been also found and investigated [Yamashita, 2001]. Most piezoelectric materials

available in the market have two forms: ceramics and single crystal materials. Single–

crystal piezoelectric materials have considerably higher piezoelectric coefficients and

electromechanical coupling factors than PZT ceramics, and as a result they are being

used to fabricate ultrasound transducers with unprecedented bandwidth (>40%) and

sensitivity, especially for High Frequency (HF) transducer fabrications. Table 7–1 shows

properties of a few important piezoelectric materials used in HF transducer designs.

High–quality single–crystal piezoelectric materials such as lithium niobate (LiNbO3),

Pb(Zn1/3Nb2/3)O3–PbTiO3 (PZNT), Pb(Mg1/3Nb2/3)O3–PbTiO3 (PMNT), polycrystalline lead titanate (PbTiO3) have been investigated since 1950’s [Yokomizo et al., 1970].

Piezoelectric strain of these crystals remains nearly hysteresis free up to levels of ~0.5%

to 0.6% depending on the crystal compositions. This hysteresis–free property of single– crystal piezoelectric materials is desired for many piezoelectric actuations. In addition, it has been found that single crystals such as Pb(Mg1/3Nb2/3)O3–PbTiO3 (PMN–PT), exhibit large increases in piezoelectric strain constant over conventional piezoelectric ceramics due to the ability to orient the crystals along a preferred high strain crystallographic direction [Park and Shrout, 1997]. PMN–PT single crystal also possesses finer grain size compared to its counterparts and, thus, it is suitable for fabrication of micro–sized

140 transducers. Furthermore, piezoelectric materials with finer grain size have proven to be able to retain their bulk material properties better at higher frequencies than their larger

grain counterparts [Shung et al., 2007]. Therefore, PMN–PT (Lead Magnesium

Niobate–Lead Titanate) crystals are selected for ultrasonic micro–transducers due to its

finer grain size, hysteresis free property under larger strains, and relatively higher

electrochemical coupling factors.

Table 7–1. Properties of a few important piezoelectric materials used in high frequency (HF) transducer designs. After [Shung et al., 2007].

The piezoelectric coefficient and the electromechanical coupling coefficient are denoted as d and k. A "33" subscript indicates that the electric field and the mechanical stress are both along the polarization axis and a “t” subscript represents the thickness mode.

141 7.2.2. Design and Architecture of UMTA Biochip

The UMTA biochip consists of arrays of high–aspect–ratio PMN–PT (Lead

Magnesium Niobate–Lead Titanate, d33) micro–pillars (3 × 3 arrays of 25 µm × 25 µm, 50

µm × 50 µm, 75 µm × 75 µm, and 100 µm × 100 µm square pillars) that can extend or shorten its length along their axial axes (thickness mode oscillations) when the electric field is applied across the axial direction. These micro–transducer pillars are laid down on a thin film of silver epoxy coated on the glass substrate. The silver epoxy layer serves as a common bottom electrode for driving micro–transducers. Epoxy resin is filled in the space between individual micro–transducers in order to dampen lateral–mode vibrations and prevent cross–talk between individual micro–transducers [Smith, 1986].

Top surfaces of the micro–transducers are coated with thin Au film, which serve as top electrodes. These top Au electrodes are individually connected to their respective bonding/contact pads through Au interconnections. The micro–transducers can be driven individually by connecting a RF function/signal generator to the top and bottom electrode in order to produce thickness–mode oscillations. Ultrasounds can be

generated if the micro–transducers are driven by the RF function generator with a sinusoidal wave of frequencies above 20 kHz.

The micro–transducers are also encapsulated under Parylene C over–coating.

Parylene–C films are pinhole free and posses a relatively high chemical resistance to most harsh chemicals [Feili et al., 2006]. Furthermore, it is biocompatible [Schmidt et al., 1988] and also a very good insulator [Loeb et al., 1977]. Parylene–C encapsulation serves as: (i) an acoustic impedance matching layer between the micro–transducer and tissue culture media [Hadimioglu and Khuri–Yakub, 1990] (Note: The characteristic acoustic impedance of Parylene C is ~2.7 MPa.s/m3, which is closer to that of water (~1.5

MPa.s/m3) than those of PMN–PT materials.), (ii) an insulating barrier between individual top Au electrodes and between Au interconnections, (iii) an insulating layer

142 between tissue culture media and top Au electrodes as well as their interconnections, and

(iv) a physical protection layer for micro–transducer arrays from harsh chemicals as well as mechanical stress. The surface of the chip (or the surface of Parylene–C coating) also presents thin Au cell trapping pads, which are located directly above each individual micro–transducers (Au is biocompatible and cells tend to attach more preferably on Au surface [Schreiber, 2004]). The bonding pads located at the chip perimeter can be accessed through via holes. Wires are soldiered onto bonding pads and the common

bottom electrode using silver epoxy and they are sealed under epoxy resin in order to

prevent electrical shorting through the tissue culture media during the experiment. The

whole UMTA biochip can be submerged in tissue culture media and cells can be seeded

and grown on the surface of the UMTA biochip under appropriate cell culture conditions.

The simplified cross–sectional schematic of the UMTA biochip is shown in figure 7–6.

Figure 7–6. Simplified cross–sectional schematic of ultrasonic micro–transducer array (UMTA) biochip. Micro–transducers can be driven individually to generate ultrasounds. Wires are soldiered onto the bonding pads and the common bottom electrode using silver epoxy and they are sealed under epoxy resin in order to prevent electrical shorting during the experiment. The whole UMTA biochip can be submerged in tissue culture media and cells can be seeded and grown on the surface of the UMTA biochip under appropriate cell culture conditions.

143 7.2.3. Process Development for UMTA Chip Fabrication

The novel design and architecture of UMTA biochip posed many scientific and

technology challenges in the fabrication. Traditional methods for fabricating small–size

transducer include dicing, molding, and chemical/mechanical polishing etc. [Safari,

1994; Wang et al., 1999 (a)]. The ultimate goal of UMTA biochips is to conduct study on

individual cells immobilized on the cell–trapping pads. This requires the cross sectional

dimension of micro–transducers pillars as small as 10 µm wide. Furthermore, PMN–PT crystals have finer gain size compared to its counterparts. These requirements and

properties make the designed transducer pillars extremely fragile for traditional dicing

techniques, which can easily cause pillar breakage and chippings. For single crystal

piezoelectric materials, both wet etching and plasma dry etching could be used to

fabricate transducers. Wet etching (using HF and HCl) [Wang et al., 2000] is usually a

cheaper process. However, the most important requirement for ultrasound transducers

in UMTA biochip is the aspect ratio that is greater than 2:1. (The aspect ratio needed for

thickness mode vibrations must be larger than 2:1. This is required so that the

transducer resonates in pure thickness modes, yielding the maximum electromechanical

coupling. High aspect ratio also dampens lateral vibration modes. Both effects are

needed to achieve broad bandwidth and high sensitivity). Therefore, wet chemical

etching of PMN–PT is not suitable due to isotropic nature of etching mechanism.

Therefore, plasma dry etching, which can yield anisotropic etching mechanism and

imposes low mechanical stresses in making fine microstructures, is chosen to fabricate the micro–transducer pillars. However, a number of major challenges arises in the fabrication of UMTA biochip, which includes the choice of plasma dry etching technique and associated preparation processes such as etch–mask deposition and lithography.

144 7.2.3.1. Inductively–Coupled Plasma Dry Etching of PMN–PT Crystals

Plasma dry etching has been shown to be an effective method for machining high–

aspect–ratio anisotropic features in many materials. It is also a relatively gentle

approach compared to mechanical micro–machining methods. Deep Reactive Ion

Etching (DRIE) of PZT ceramic has been investigated by a number of researchers and, in these studies, either photoresist or metal etch–masks have been used to define the etching patterns. The chemistries of etchants being used in these studies include chlorine– and fluorine–based chemistries such as SF6, SF6/N2/Ar, HBr/Ar, CF4, Cl2/CF4,

and etc. [Bale and Palmer, 2001; Chung et al., 2002; Jung and Lee, 2001; Wang et al.,

1999 (b)]. Furthermore, etching depths over 100 microns has been achieved in these

studies [Bale and Palmer, 2001]. Unfortunately, for most of the etch results, the sidewall

angles of the etch profile are less than 80˚ (figure 7–7). In order to achieve pure thickness vibration modes and dampen lateral vibration modes, the micro–transducers must posses both high aspect–ratio and nearly vertical sidewall profile (i.e., nearly 90˚ sidewall angles). Traditional RIE systems cannot produce nearly vertical sidewall profiles in the pattern transfer of high aspect–ratio structures into the wafer.

Figure 7–7. SEM images of array structures produced by a 3–hour RIE with 10% argon in SF6 showing (a) the array and (b) sidewall angle variation. The structure height is 26 µm including the remaining Nickel etch–mask. The dotted white line in (b) shows the nickel/PZT boundary. The RF power was 200 W, process pressure 2 mTorr, and the total gas flow rate was 20 sccm. After [Bale and Palmer, 2001].

145 Because of limitations of traditional RIE systems, new plasma etching techniques

have been developed, such as Reactive Ion Beam Etching (RIBE), Magnetically

Enhanced Reactive Ion Etching (MERIE), Electron Cyclotron Resonance (ECR), and

Inductively Coupled Plasma (ICP). Due to the ability to offer higher ion density and

independent control of ion energy and ion density, ECR and ICP systems can produce

nearly vertical sidewall profiles for high aspect–ratio structures. ICP and ECR systems

operate as simple RIE reactors if the respective high–density plasma source is not

applied. The major difference between ECR and ICP is the way high–density plasma is

generated. In ECR chamber, microwaves are introduced into the chamber resonance cavity through a wave–guide producing a dynamic electric field to dissociate the gases and generate the electrons and ions of the plasma that diffuses to wafer surface. Because of this unique design of the tool, high–density plasmas can be formed at low gas pressure with low plasma potentials and ion energies. In ICP systems, plasma is formed in a dielectric vessel encircled by an inductive coil into which RF power is applied. A strong magnetic field is induced in the center of the chamber, which generates high–intensity plasma due to the circular region of the electric field that exists concentric to the coil. At low pressure, the plasma diffuses from the generation region and drifts to the substrate at relatively low energy [Lee at al., 1998]. Thus, ICP is expected to produce low damage while achieving high etch rate. Anisotropic profile is obtained by superimposing a second RF bias on the sample to independently control the ion energy. Therefore, in

UMTA biochip fabrication, ICP plasma etching tool (Applied Material ICP–RIE plasma etcher at Penn State NanoFab), which is the best suitable technology available to achieve vertical sidewall and provide a relatively much higher etch rate, is chosen.

A number of preliminary etch experiment on ICP–RIE tool showed that, Chlorine– based chemistries has higher etch rate on PMN–PT crystal than Fluorine–based chemistries. However, instead of using a pure Cl2, Cl2/Ar gas mixture was used to

conduct the dry etching in ICP–RIE chamber. The motivation of choosing argon (Ar) as

146 the additive gas is to help stabilize the plasma and enhance the physical and chemical etching mechanism [Lee at al., 1996]. As a high mass atom, Ar can help the sputtering

process and reduce the etch resistance of the surface caused by etch–product re-

deposition. Ar+ ions play a key role in both the initial bond–breaking in the materials and etch–product desorption by ion enhancements and bombardments. Furthermore, it helps surface physical bombardment to obtain a smooth sidewall [Khan et al., 1999].

In addition to etching systems and etch chemistries, one of the important factors in plasma etching is the type of material used for the etch–mask that defines the etching patterns and protects the non–etching area. Nickel (Ni) is a favorable plasma etch– mask material and being widely used in ICP plasma etching due to its melting point at

1455 ˚C, boiling point at 2730 ˚C, and strong resistance to corrosion [Chang et al., 2001;

Takahashi et al., 2000]. Nickel chloride, the etch produce in chlorine–based plasma etch, has a high melting point at 1001 ˚C, making it hard to be removed, and thus, prevent further Ni etching during the process and improve the selectivity of the etched materials. Therefore, Ni is chosen as the material for etch–mask in ICP–RIE etch due to all of these characteristics.

7.2.3.1.1. Analysis of the Effect of Plasma Etching on PMN–PT Crystals

It is very important to evaluate the degree of damage caused by plasma etch chemistries and the degree of stress built up by the etching mechanisms on the PMN–PT single crystal piezoelectric material. The X–Ray Diffraction (XRD) scanning was carried out on both plasma–dry–etched surface and non–plasma–dry–etched surface of PMN–

PT material. The XRD patterns of the PMN–PT crystal’s surface before and after Cl2–

based ICP plasma etching are shown in figure 7–8a and 7–8b, respectively. The prior–

and post–etching XRD patterns matched with each other very closely, indicating that no

observable or no significant surface damage was induced by the ICP plasma etching.

Furthermore, piezoelectric properties of PMN–PT crystal remained the same after ICP

147 plasma etching. Figure 7–9 shows the strain vs. electric field property of PMN–PT crystal after ICP plasma etching, which closely matches with that of PMN–PT crystal before etching. These findings cleared the concerns that possible etching–induced damages would affect the performance of micro–machined PMN–PT single crystal piezoelectric transducers.

Figure 7–8. X–Ray Diffraction (XRD) patterns of PMN–PT crystal’s surface. (a) Before and

(b) after Cl2–based plasma etching [Courtesy of An Cheng].

Figure 7–9. Strain–electric field property of PMN–PT crystal after Cl2–based plasma etching [Courtesy of An Cheng].

148 7.2.3.2. Etch–Mask Molding and Growth

The etch–mask grown on the PMN–PT wafer for ICP–RIE etch must have three important properties: (i) it must have low etch selectivity over the PMN–PT crystals by

the etchant chemistries (i.e., the etch rate of the mask has to be much lower than that of

PMN–PT crystals), (ii) it must have enough thickness to withstand up to the point at

which the desired etch profile height is achieved in the PMN–PT wafer, and (iii) its defining patterns must have nearly vertical sidewall profiles. All these requirements

impose major challenges in growing and patterning of Ni etch–mask for ICP plasma

etching of PMN–PT crystal.

7.2.3.2.1. Pulsed Electroplating

In depositing the Ni etch–mask, e–gun evaporation technique is usually used due to its advantage in yielding pure and high quality metal films. However, PVD–deposited

Nickel film showed significant tensile stress [Madou, 2002] and Ni film starts to peel off once the thickness exceeds 2500 Å with smallest feature size of 25 µm. The height of

micro–transducers in UMTA chip must be at least 100 µm for 50 µm pillars in order to achieve 2:1 aspect ratio. Preliminary ICP plasma etch experiments (Designs of

Experiments (DoE)) and calculations showed that the etch selectivity of PMN–PT over

Ni is less than 10:1 at optimum process conditions and, thus, the thickness of Ni etch– mask must be at least 10 µm, which is thicker than e–gun deposition (or PVD) can achieve without having significant tensile stress. Hence, electroplating, which is a widely used technique for depositing metal film, becomes a reasonable approach for growing very thick Ni etch–mask. However, electroplated metal film quality varies with the electroplating process conditions such as pH value of electroplating solution, current density, solution bath temperature, deposition mode (AC, DC, or pulsed), and etc. It has been demonstrated that the Nickel layer’s internal stress can be minimized by carefully

149 controlling these process parameters [Maner et al., 1998; Marques et al., 1997]. Another

important Ni film property for ICP plasma etching is the film density. It has been shown

that denser film possesses stronger plasma etch resistance [Maner et al., 1998] and, thus,

increase the etch selectivity. Research showed that pulsed plating leads to smaller metal

grain size and smaller porosity due to a higher deposition potential [Madou, 2002]. It is

observed that the higher the frequency of the pulse, the smaller is the internal stress

induced in the deposited films. Furthermore, the deposited metal film hardness (or

density) decreases with the increasing current density [Ohno, 1988].

Preliminary Design of Experiments (DoE) coupled with ICP–RIE etch on the Ni films electroplated under different process conditions showed that Ni etch–mask deposited under high–frequency pulsed (2.5 MHz) with a small current density (10 mA/dm2) yields the best plasma etching resistance (etch rate as low as ~15–16 nm/min). Figure 7–10 shows the FE–SEM (Field Emission–Scanning Electron Microscope) image of cross–

sections of two Ni films grown using DC and high–frequency pulsed electroplating. Ni

layer grown using high–frequency (2.5 MHz) pulsed electroplating with a small current density (at 10 mA/dm2) possesses smaller grain size compared to that of Ni layer grown

using the DC electroplating with a larger current density (at 35 mA/dm2). These findings led to the conclusion that high–frequency pulsed electroplating with a small current is the most suitable approach for growing Ni etch–mask for ICP–RIE plasma etch.

150

Figure 7–10. Cross–sectional FE–SEM image of two different Ni layers: the bottom layer was deposited by 2.5 MHz pulsed electroplating with current density of 10 mA/dm2; the top layer was deposited by DC electroplating with a current density of 35 mA/dm2. The Ni layer deposited by high–frequency pulsed electroplating possesses smaller grain size and, thus, resulting in much denser film. Denser Ni films with smaller grain size are more resistant to plasma dry etching using Cl–based etching chemistries [Courtesy of An Cheng].

7.2.3.2.2. Doubled–Exposure Laser Lithography

Positive photoresist was chosen as the mold for growing Ni etch–mask by high– frequency pulsed electroplating. The desired thickness of Ni etch–mask based on the preliminary experiments and calculations is at least 10 µm. The required thickness of photoresist patterns in order to mold the desired thickness of Ni etch–mask must be at least 14 µm (at least 1.35 times). SPR220, i–line positive photoresist from MegapositTM,

was chosen in order to meet this requirement since SPR220 is designed to achieve >10

µm film thickness in a single coat with good uniformity. Another important factor in patterning photoresist mold is to have straight exposure profile since Ni etch–mask grown for ICP–RIE etch must also possess nearly–vertical–sidewall–profile defining

151 patterns. Nonetheless, using traditional contact photolithography to pattern such a thick

film of photoresist and to maintain the straight exposure profile impose another technical challenge due to the decreasing energy dose from top to bottom. This non–

uniformity in energy dose from top to bottom in contact photolithography usually results

in a trapezoid exposure profile in i–line positive photoresist [Madou, 2002; Todd et al.,

1999]. In order to overcome this problem, the exposure process was carried out by a

laser lithography tool (Heidelberg DWL66 laser writer). The focused 442 nm HeCd laser

in DWL66 laser writer has depth of focus of up to 8 µm. However, this depth of focus

still cannot satisfy the 14–µm exposure depth of photoresist mold. Therefore, in order to solve this issue, a doubled–exposure laser lithography is employed.

In this approach, the photoresist was exposed two times with smaller doses (65% of dose compared to that of single laser exposure to avoid over–exposure patterns). Each exposure has its respective focus point locating at different depth in the 14–µm thick

photoresist, thus, compensating the limitation in depth of focus during each individual

exposure. Furthermore, this approach also yields vertical exposure profile after

developing in standard Tetramethyl ammonium hydroxide (TMAH) developer (e.g., MF

CD–26). Figure 7–11 illustrates the difference in exposure profiles between the traditional contact photolithography and the doubled–exposure laser lithography.

Figure 7–12 shows the FE–SEM images of exposure profiles on SPR220 photoresist

obtained by two different photolithography techniques.

152

Figure 7–11. Schematics showing the difference between contact photolithography and doubled–exposure laser lithography.

Figure 7–12. FE–SEM images of exposure profiles on i–line SPR220 positive photoresist obtained by (A) contact photolithography, where the resulting exposure profiles are trapezoidal and (B) doubled–exposure laser lithography, where the resulting exposure profiles possess nearly vertical side wall (Note: The width and the height of test exposure profiles shown are ~9µm and ~14µm respectively). [Courtesy of An Cheng]

153 7.2.4. Microfabrication of UMTA Biochip

Microfabrication techniques employed in the fabrication of Ultrasonic Micro–

Transducer Array (UMTA) biochips include laser lithography and contact photolithography, lift–off process, Physical Vapor Metal Deposition (PVD), Chemical

Vapor Deposition (CVD), high–frequency pulsed electroplating, epoxy molding, chemical and mechanical planarization/polishing, plasma dry etching (inductively– coupled & conventional RIE), wet chemical etching, and wire bonding.

7.2.4.1. PMN–PT Wafers

The PMN–PT single crystal plates were prepared as wafers and lapped on both sides and polished on one, then embedded in a 2.5” diameter PZT wafer to avoid the difficulties in microfabrication process for small samples. Figure 7–13 shows 1.5 cm diameter PMN–PT crystal embedded in a 2.5” PZT ceramic wafer.

Figure 7–13. PMN–PT (Lead Magnesium Niobate–Lead Titanate) single crystal piezoelectric disc embedded in a 2.5” PZT ceramic wafer [Courtesy of TRS Ceramics Inc].

154 7.2.4.2. Photomask Fabrication

Three dark–field photomasks (one for top Au electrodes and interconnection, one for

Al etch–mask patterning for via holes, and one for cell–trapping pads) were fabricated.

Photomask fabrication procedures are the same as those described in section 7.1.2.1.

7.2.4.3. Biochip Fabrication

7.2.4.3.1. Micro–Transducer Fabrication

Fabrication of UMTA biochip starts from e–beam deposition of thin Ni seed layer

(~200 nm) film on the polished PNM–PT wafer. To define micro–transducer areas, positive photoresist (PR) SPR220 was spun on the wafer, exposed using Heidelberg

DWL66 laser lithography tool (doubled–exposure laser lithography), and developed. Ni etch–mask was grown inside the PR mold by HF pulsed electroplating. After desired thickness (~12.5 um) is reached, the PR mold was removed by bathing the wafer in standard resist stripper (EKC 830). The Ni etch–mask grown inside the PR mold and Ni etch–mask after removal of PR mold are shown in figure 7–14a and 7–14b respectively.

The Ni etch–mask protects the PMN–PT crystal underneath during ICP–RIE etching.

Figure 7–14. Ni etch–mask fabricated by doubled–exposure laser lithography and high– frequency pulsed electroplating. (A) Ni grown by HF pulsed electroplating in the photoresist mold. (B) Ni etch–mask after removing PR (The sample is tilted by 30˚. The actual height is approximately two times FE–SEM measured height, i.e., ~120µm) [Courtesy of An Cheng].

155 Nearly vertical (97–98˚ etch profile) high–aspect–ratio micro–transducer pillars are created by ICP–RIE etch. The optimized values for bias voltage, ICP coil power, chamber pressure, and balanced Ar/Cl2 gas mixture are –160 V bias voltage, 1200 W coil power, 10 mTorr chamber pressure with a mixed gas flow of 25 sccm of Cl2 and 7 sccm of Ar, respectively. These optimized process parameters gave an etch rate of ~150nm/min on PMN–PT with ~10:1 etch selectivity over Ni etch–mask. The transducer height is

approximately 110–120 µm. The roughness of the facet as inspected in FE–SEM image was < 40 nm. After ICP etching, the remaining Ni etch–mask are removed by wet chemical etching in Ni etchant solution at 40˚C (Type TFG). The resulting micro– transducer pillars are shown in figure 7–15. The peripheral PZT material was removed (by chipping off) from the wafer after Ni wet etching.

Figure 7–15. FE–SEM images of: (A) 3 × 3 matrixes of transducers of 25 µm × 25 µm, 50 µm × 50 µm, 75 µm × 75 µm, and 100 µm × 100 µm cross–sectional area; (B) A 3 × 3 matrix of 25–µm transducers with etched depth of ~ 2 × 60 µm = 120 µm; (C) A 3 × 3 matrix of 75–µm transducers; (D) a single 75–µm transducer. The samples are tilted by 30 degrees. The etched depth reached is ~ 2 × 64 µm = 128 µm. [Courtesy of An Cheng].

156 7.2.4.3.2. Epoxy Filling, Wafer Lapping, and Polishing

After micro–transducer pillars are created, PMN–PT disc was placed in the bottom of a Teflon mold and “EPO–Tech 301” epoxy, which has high elastic modulus and resistance to high temperature after it was fully cured, was poured into the mold as shown in figure 7–16A. Epoxy was fully cured overnight (after 24 hours) and the whole epoxy wafer was taken out of the Teflon mold as shown in figure 7–16B.

Figure 7–16. Epoxy filling and curing process. (A) The PMN–PT disc is placed at the bottom of the Teflon mold and Epoxy resin is poured to fill the space between the pillars [Courtesy of An Cheng]. (B) After Epoxy is fully cured, the PMN–PT–disc–embedded Epoxy wafer is taken out of the mold (Etched–left substrate facing up in the picture shown). (C) Schematics showing the Epoxy filling and curing process.

157 The remaining etched–left PMN–PT substrate was removed by flat lapping. Figure

7–17A shows the etched–left substrate started being lapped away from the center of the

wafer and figure 7–17B shows the end point at which the etched–left substrate was

almost removed. Once the whole etched–left substrate was removed, all the transducer

pillars will no longer be connected and become individual elements embedded in the

Epoxy wafer. Then, thin Ni film (~200 nm) was deposited on the lapped surface of the

transducer–embedded–Epoxy wafer by e–beam evaporation (PVD) and then the wafer

was mounted on the precision glass substrate using silver epoxy (lapped side facing down). Once the silver epoxy is cured the wafer will be glued to the precision glass substrate. Furthermore, silver epoxy layer also serve as a common bottom electrode for micro–transducers. The remaining epoxy on the top is then lapped away until the tips of individual pillars were exposed.

In order to improve the quality of the transducer–electrode contact, the device’s top surface topography, and the flatness and topology of transducers’ top surface, chemical mechanical planarization/polishing (CMP) followed after lapping process. A two–step surface CMP was employed. First, the 3–µm Al2O3 lapping powder was used for rough surface polishing and planarization. Then, 0.1–µm 3M fine slurry was used for fine surface polishing and the final finishing. Figure 7–17C shows the schematics of entire flat lapping and chemical mechanical planarization/polishing process on the PMN–PT– disc–embedded Epoxy wafer. Figure 7–18 compares the surface condition of the micro–

transducer top surface after Al2O3 power planarization/polishing and after 3M fine slurry polishing (images obtained by FE–SEM). Optical profilometer measurements (with

Wyko NT1100 in VSI (Vertical Scanning Interferometry) mode) on the micro–transducer top surface showed that RMS surface roughness is less than 1500 Å.

158

Figure 7–17. Flat Lapping and chemical mechanical planarization/polishing (CMP) process. (A) The etched–left substrate started being removed from the center [Courtesy of An Cheng]. (B) The end point for flat lapping at which the etched–left substrate is almost removed. (C) Schematics showing the flat lapping and CMP process.

Figure 7–18. FE–SEM images of PMN–PT transducer top surface and Epoxy resin surface

conditions: (A) After 2 hours of rough polishing with 3–µm Al2O3 lapping powder; (B) After 1.5 hours of fine polishing with 0.1–µm 3M fine slurry. [Courtesy of An Cheng].

159 7.2.4.3.3. Top Electrodes and Interconnection

After lapping and CMP process, the device is ready for photolithography and metal

deposition for top electrodes and interconnection lines. Photoresist SPR3012 were spun

on the cleaned wafer, UV–exposed (i–line 356 nm) under the photomask consisting of

the defined patterns for electrodes and interconnections (Karl Suss MA/BA6 contact

photolithography aligner and exposure system was used), and developed. After

removing the residual resists in the exposed regions by an oxygen descum process, a Ni

(50 Å)/Ti (50 Å)/Au (2000 Å) triplex layer was deposited onto the wafer at a rate of 0.5

Å/s by electron–beam evaporations (Semicore e–beam evaporator was used. The Ni/Ti

twin adhesion layer was used to enhance the adhesion between the Au and the wafer

substrate. Microelectrodes and interconnections were then defined by the photoresist

lift–off process.

7.2.4.3.4. Device Encapsulation with Parylene C Coating

Parylene C film was deposited conformally onto the surface of the wafer by Chemical

Vapor Deposition (Model PDS 2010 LABCOTER® deposition system, Specialty Coating

Systems). The thickness of the film coating is approximately 15 µm. Next, photoresist

SPR3012 was spun on the cleaned wafer, UV–exposed (i–line 356 nm) under the photomask consisting of via–hole patterns for contact pads (Karl Suss MA/BA6 contact photolithography aligner and exposure system was used), and developed. After removing the residual resists in the exposed regions by an oxygen descum process, an aluminum (2000 Å) layer was deposited onto the wafer at a rate of 0.5 Å/s by electron– beam evaporation (Semicore e–beam evaporator was used). Al etch–mask patterns were, then, defined by a photoresist lift–off process. After that, via holes are created on the contact pads through the Parylene C coating by RIE etch (O2 plasma was used).

Finally, the Al etch–mask was removed by wet chemical etching with standard Al etchant

in room temperature.

160 7.2.4.2.5. Cell–Trapping Pads

In order to create cell–trapping pads on the surface of UMTA biochip, photoresist

SPR3012 was spun on the wafer, UV–exposed (i–line 356 nm) under the photomask consisting of the defined patterns for cell trapping pads (Karl Suss MA/BA6 contact photolithography aligner and exposure system was used), and developed. After removing the residual resists in the exposed regions by an oxygen descum process, a Ti

(50 Å)/Au (500 Å) twin layer was deposited onto the wafer at a rate of 0.5 Å/s by electron–beam evaporations (Semicore e–beam evaporator was used. The thin Ti layer

was used to enhance the adhesion between the Au and the Parylene C). Cell trapping

pads were, then, defined by the photoresist lift–off process. The surface of the UMTA

biochip is shown in figure 7–19 and the entire device fabrication process is summarized

in figure 7–20.

A B

C

Figure 7–19. (A) Surface of the UMTA biochip showing 3 × 3 arrays of micro–transducers of four different size (25 µm, 50 µm, 75 µm, and 100 µm) [Courtesy of An Cheng]. (B) A bright field image of a 3 × 3 array of 25 µm × 25 µm transducer (top view) before Au electrodes and interconnections are formed. (C) An DIC image of a 3 × 3 array of 25 µm × 25 µm transducer (top view on the surface of finished UMTA biochip).

161

Figure 7–20. Schematic of UMTA biochip fabrication process. Process flow is as follows: 1) Start from the polished PNM–PT wafer; 2) Thin (200 nm) Ni film is deposited on the PMN–PT wafer surface as a seed layer by e–beam Physical Vapor Deposition (PVD). Transducer areas are defined on the seed layer using photoresist mold (double–exposure laser lithography is employed). An Ni etch hard–mask is grown on the seed layer via HF pulsed electroplating and photoresist mold is removed; 3) Vertical high–aspect–ratio micro–transducer pillars are created by plasma etching (ICP–RIE); 4) The Ni etch hard–mask is removed. Epoxy resin is filled between the etching pillars and cured; 5) The remaining PMN–PT substrate is lapped down until all the substrate material is removed (i.e., the bottom surface of the etched pillars are exposed and the pillars are disconnected); 6) The sample is then mounted on the precision glass substrate using silver epoxy; 7) The remaining epoxy resin on the top is lapped down until the top surface of the pillars are exposed. This is followed by chemical mechanical planarization/polishing (CMP); 8) Au top electrodes and interconnection lines are deposited using contact photolithography, PVD and lift–off process; 9) Parylene C film is deposited by Chemical Vapor Deposition (CVD) for acoustic impedance matching and device encapsulation. After Parylene C depostion, an Ni (or) Al etch mask is created on the surface using contact photolithography, PVD, and lift–off process; 10) the bonding pads are opened through via holes using RIE etching and metal etch–mask is removed; 11) cell–trapping pads are deposited on the surface of Parylene C film, which are located above each micro–transducer, using contact photolithography, PVD, and lift–off process.

162 7.2.4.2.6. Wire–Bonding and Sealing.

After cell trapping pads are deposited on the Parylene C film, UMTA biochip is ready

for wire bonding. The biochip is mounted on the bottom of the 4” glass petri dish. Wires

were, then, soldered onto contact pads and common bottom electrode with silver Epoxy

and, once the silver Epoxy was fully cured, the solder contacts were sealed under Epoxy

resin in order to insulate them from the tissue culture media.

7.2.5. Characterization of UMTA Biochip

In order to evaluate the performance of UMTA biochip, two properties of micro– transducers are characterized: 1), ultrasound pressure generated by the micro– transducer, which is an important property in cell membrane permeabilization at the cellular level using UMTAs; and 2), resonance frequency of the micro–transducer, which is an important characteristic in determining and maximizing the power transfer from transducer driving circuits and instruments to the transducer.

7.2.5.1. Fundamentals of Ultrasounds Generated by the Transducer

In brief, the intensity of ultrasound trespassing through a medium can be estimated using equation (7–1) if the ultrasound pressure is known (the ultrasound pressure can be obtained experimentally via a calibrated microphone in air or a hydrophone in water)

[Zagzebski, 1996].

P2 I = (7–1) 2 c ρ

ρ : Density of the medium (103 kg/m3 for water) c : Speed of sound (1500 m/s in water) P : Ultrasound pressure amplitude [Pa].

I : Ultrasound intensity [Watt/m2].

163 Ultrasounds generated from the ideal transducer, which is oscillating in one of its

thickness modes, have two regions: (i) near field region (Fresnel zone) and (ii) far field region (Fraunhofer zone). Figure 7–21 illustrates the two field regions in the ultrasonic

beam from a single–element, unfocused transducer. The beam of ultrasound remains

well collimated in the near field. However, the beam diverges in the far field region. The

highest ultrasound intensity and pressure occur at the end of the near field region as

illustrated in figure 7–21b.

Figure 7–21. Conceptual illustration of the intensity (or pressure) distribution of continuous wave (CW) ultrasound: (a) the near– and far–field regions in relation to the transducer, (b) the field distribution of an ideal transducer source generating a continuous wave. After [Wells, 1977].

164 The near field length (NFL) and far–field beam divergence angle (θ) of an ultrasound can be estimated by equation 7–2 and 7–3 respectively [Zagzebski, 1996]. According to these equations, the NFL and beam divergence angle (θ) mainly depend on the diameter

of the transducer and the wavelength (or frequency) of the ultrasound being generated.

The higher is the frequency (i.e., the smaller is the wavelength), the longer is the NFL

and the smaller is the beam divergence angle (θ). Therefore, in generating the focused ultrasound with ideal transducers, the two most important parameters are the driving frequency and transducer’s diameter.

2 2 ⎛ ⎞ d d ⎜ f ⎟ NFL ≈ = ⎜ ⎟ (7–2) 4λ 4 ⎜ c ⎟ ⎝ ⎠ ⎛ ⎞ 1.22λ 1.22⎜ c ⎟ sinθ ≈ = ⎜ ⎟ (7–3) d d ⎜ f ⎟ ⎝ ⎠

d : Diameter of transducer. c : Speed of sound (in water, it is 1500 m/s) f : Frequency of ultrasound.

λ : Wavelength of ultrasound. NFL : Near field length.

θ : Beam divergence angle.

7.2.5.2. Micro–Transducer Performance Testing

7.2.5.2.1. Ultrasound Pressure Measurement Using Hydrophone

A hydrophone was used to measure the acoustic pressure exerted by the ultrasound generated by individual micro–transducers from the UMTA biochip. Figure 7–22 shows

the simplified schematic of the experimental setup. The UMTA biochip was submerged

into the water tank, which was designed and built for ultrasound exposimetry, and a

hydrophone (75–µm diameter) was mounted to a three–axis–positioning system. The

165 hydrophone was brought near to the surface of the UMTA biochip by the stepper– motor–controlled positioning system and the surface of the UMTA biochip was scanned with the hydrophone (the hydrophone was kept at a constant distance above the surface of biochip) while micro–transducers were being driven by the RF signal generator. The hydrophone’s signals (voltages) were recorded at various positions across the UMTAs.

The pressure amplitude is calculated by multiplying the recorded signals (voltages) with

the calibrated amplitude coefficient of the hydrophone being used.

Figure 7–22. Simplified Schematic of the experimental setup for measuring the acoustic pressure generated from the micro–transducers. The water tank was designed and built for ultrasound exposimetry. Note: the schematic is not drawn to the actual scale.

166 Since biological cells will be seeded on the chip surface instead of being suspended in the tissue culture media, measuring the acoustic pressure at the surface of the UMTA biochip must have been more relevant in evaluating the impact of ultrasound generated by the micro–transducer on the cells. Therefore, the hydrophone should be placed as close to the top surface of the UMTA biochip as possible. However, the three–axis

motion system used in the setup lacks the fine movements (i.e., micrometer resolution)

due to low stepper–motor resolution and large screw pitch (i.e., coarse threads) on the

positioning shafts. In addition, the extreme small size (75–µm diameter) and the structure of hydrophone make it very fragile and easy to be broken if its tip crashes on the surface of the UMTA biochip. Due to these technological issues, the nearest distance achieved between the hydrophone and UMTA biochip surface was around 5 mm (roughly estimated by visual inspection through the water tank).

Micro–transducers of three different sizes, i.e., 25–, 50–, 75–µm, are tested. All transducers were driven by the same signals, i.e., a 30–MHz continuous sinusoidal signal with 50 mVp–p generated from the RF signal generator. The signal from the RF signal generator is amplified by the power amplifier (10 dB amplification) and then is fed to the UMTA biochip in order to drive the micro–transducers. The 30–MHz driving frequency was chosen in order to increase the Near Field Length (NFL) of the acoustic field radiated from micro–transducer and also to reduce the beam divergence angle

(Furthermore, the 30–MHz driving frequency also make the NFL of ultrasonic beam

generated from 25–µm transducer approximately equal to the transducer to cells

distance during the site–specific sonoporation. See Chapter 9). For instance, for 30–

MHz ultrasound in water with c = 1500 m/s, λ = 50 µm, and the diameter of transducer being set at d = 70 µm, the estimated NFL and beam divergence angle are approximately

13 µm and 60°, respectively (using equation 7–2 and 7–3) (Note: equation 7–3 is not

valid and cannot be used for d < 62 µm). Therefore, as the thickness of Parylene C over–

167 coating is approximately 15 µm, the 30–MHz frequency is the optimum frequency that

can bring relatively confined acoustic field at the surface of the UMTA biochip, where the

cells will be seeded and grown for site–specific sonoporation.

During the ultrasound pressure measurement with the ultrasound exposimetry setup, the hydrophone scanned the 1 cm × 1 cm square area centered at the micro– transducer being tested. By using the measured results (after voltage data is converted to pressure) from the hydrophone scanning and using equation 7–1, the acoustic intensity of the ultrasound across the 1 cm × 1 cm scanning field can be estimated. The estimated ultrasound intensity at 5 mm distance away from the UMTA biochip surface when a transducer is activated with the 10dB–amplified sinusoidal signal at 30 MHz

(Note: the signal has 50–mVp–p amplitude before 10dB amplification) is approximately

5.0 ± 0.02 mW/m2 for 25–, 50–, 75–µm transducers. These experimental data are only used to qualitatively determine whether the UMTA biochip can generate ultrasound. For precise quantitative measurements, the whole experimental setup needs further improvements (for instance, precise positioning of the hydrophone with high resolution positioning system).

7.2.3.2. Electrical Impedance spectrum

To further characterize the micro–transducer performance characteristics, the input electrical impedance spectra of micro–transducers were obtained using a network analyzer (Agilent E5100A Network Analyzer). Again, UMTA biochip was also submerged in water during the measurement. Figure 7–23 shows the input electrical impedance spectra of 25–, 50–, 75–µm, and 100–µm transducers. The resonance frequency was

found to be 61, 57, 54, and 50 MHz for 100–, 75–, 50– and 25–µm transducers,

respectively. The general concensus is that the higher is the aspect ratio (i.e., diameter

or width/thickness) of the thickness mode transducer, the lower is resonance frequency

168 [Iula et al., 2003]. As all micro–transducers have almost the same thickness and, thus,

the smaller is the micro–transducer’s cross–sectional area, the higher is its aspect ratio

and the lower is its resonance frequency. These data can also be useful in determining

the power transfer from the transducer driving circuit to the transducer. In other words,

electrical impedance mismatch can be calculated and impedance matching circuit can be

built and employed as an interface between the driving circuit and the transducer in

order to maximize the power transfer.

Figure 7–23. Input electrical impedance spectra of micro–transducers: (A) 100–µm transducer; (B) 75–µm transducer; (C) 50–µm transducer; and (C) 25–µm transducer. It can be noticed that the resonance frequency decreases with increasing aspect–ratio of the micro–transducer. [Courtesy of An Cheng] 100 um square area 169 7.3. Chapter Appendix

7.3.1. List of Symbols

c [m/s] Speed of sound in water

d [µm] Diameter or width of the transducer

f [Hz] Frequency of ultrasound

I [Watts/m2] Ultrasound intensity

λ [µm] Wavelength of ultrasound

P [Pa] Ultrasound pressure amplitude

3 ρ [kg/m ] Density of the medium trespassed by the ultrasound

θ [degree] Beam divergence angle

7.3.2. List of Some Abbreviations

AFM Atomic Force Microscopy

CMP Chemical Mechanical Planarization/Polishing

CVD Chemical Vapor Deposition

DI De–Ionized

DIC Differential Interference Contrast

DoE Design of Experiments

DRIE Deep Reactive Ion Etching

ECR Electron Cyclotron Resonance

EFM Electric Force Microscopy

FE–SEM Field Emission–Scanning Electron Microscope

HF High Frequency

ICP Inductively Coupled Plasma

IMA Integrated Microelectrode Array

170 MERIE Magnetically–Enhanced Reactive Ion Etching

NFL Near Field Length

PECVD Plasma–Enhanced Chemical Vapor Deposition

PMN–PT Lead Magnesium Niobate–Lead Titanate

PR PhotoResist

PVD Physical Vapor Deposition

PZT Lead Zirconate Titanate

RIBE Reactive Ion Beam Etching

RIE Reactive Ion Etching

RMS Root Mean Square

TMAH Tetra Methyl Ammonium Hydroxide

UMTA Ultrasonic Micro–Transducer Array

VSI Vertical Scanning Interferometry

XRD X–Ray Diffraction

7.3.3. List of Tools and Equipments Used in the Device Fabrication and

Process Characterizations

The following is the list of equipments at Penn State Nanofabrication Facility, which were used in the device fabrication and process characterizations, except for the chemical mechanical planarization/polishing (CMP) and dicing, which were carried out using the instruments and tools available at Penn State Materials Characterization Lab.

1) Photolithography Equipments

− Resist processing stations

− Karl Suss MA/BA6 exposure/contact aligner system (contact photolithography)

− Heidelberg DWL66 Laser Writer (laser lithography)

171 2) Plasma Etching Tools

− M4L plasma surface modification tool

− PlasmaTherm 720 RIE (Reactive Ion Etching)

− Applied Materials DPS (Inductively Coupled Plasma Etching)

3) Thin–Film–Deposition Equipments

− Semicore electron–beam evaporator (Physical Vapor Deposition)

− Applied Materials P–5000 PECVD cluster tool (Plasma–Enhanced

Chemical Vapor Deposition for dielectric materials)

− In–house electroplating system at Penn State nanofabrication facility

− Model PDS 2010 LABCOTER® deposition system (Parylene C deposition)

4) Chemical Microfabrication and Nanofabrication

− Wet chemical processing stations

5) Thermal Processing Equipment

− MRL (Black Max) annealing furnace

6) Process Characterization Equipments

− LeO 1530 FE–SEM (Field Emission–Scanning Electron Microscope)

− Micromanipulator 6000 probe station and Agilent C–V/I–V test equipments

− Gaertner® ellipsometer

− Nanometrics NanoSPEC 210 ellipsometer

− KLA–Tencor Alpha–Step 500 stylus profilometer

− Wyko NT110 optical profilometer

172 References

Bale, M., and Palmer, R. E., 2001. Deep Plasma Etching of Piezoelectric PZT with SF6. Journal of Vacuum Science and Technology B. 19 (6), 2020–2025.

Borkholder, D. A., 1998. Cell Based Biosensors Using Microelectrode. Ph.D. Dissertation. Stanford University, Stanford, CA.

Chang, L. –B., Liu, S. –S., Jeng, M. –J., 2001. Etching Selectivity and Surface Profile of GaN in

the Ni, SiO2 and Photoresist Masks Using an Inductively Coupled Plasma. Japanese Journal of Applied Physics. 40, 1242–1243.

Chung, C. W., Byun, Y. H., and Kim, H. I., 2002. Inductively Coupled Plasma Etching of a

Pb(ZrxTi1-x)O3 Thin Film in a HBr/Ar Plasma. Microelectronic Engineering. 63, 353–361.

Feili, D., Schuettler, M., Doerge, T., Kammer, S., Hoffmann, K. P., and Stieglitz, T., 2006. Flexible Organic Field Effect Transistors for Biomedical Microimplants Using Polyimide and Parylene C as Substrate and Insulator Layers. Journal of Micromechanics and Microengineering. 16, 1555–1561.

Hadimioglu, B., and Khuri – Yakub, B. T., 1990. Polymer Films as Acoustic Matching Layers. IEEE 1990 Ultrasonics Symposium Proceedings. 3, 1337–1340.

Iula, A., Cerro, D., and Pappalardo, M., 2003. 3D Finite Element Analysis of the Langevin Transducer. 2003 IEEE Symposium on Ultrasonics. 2, 1663–1667.

Jaffe, B., 1955. Properties of Piezoelectric Ceramics in the Solid Solution Series PT, PZ, PSn, and PHF. Journal of Research of the National Bureau of Standards. 55, 239–248.

Jung J. –K., and Lee, W. –J., 2001. Dry Etching Characteristics of Pb(Zr,Ti)O3 Films in CF4 and

Cl2/CF4 Inductively Coupled Plasmas. Japanese Journal of Applied Physics. 40, 1408–1419.

Khan, F. A., Zhou, L., Ping, A. T., and Adesida, I., 1999. Inductively Coupled Plasma Reactive Ion

Etching of AlxGa1-xN for Application in Laser Facet Formation. Journal of Vacuum Science and Technology B. 17 (6), 2750–2754.

Lee, J. W., Hong, J., and Pearton, S. J., 1996. Etching of InP at >1 µm/min in Cl2/Ar Plasma Chemistries. Applied Physics Letters. 68 (6), 847–849.

Lee, J., Cho, H., Hays, D. C., Abernathy, C. R., Pearton, S. J., 1998. Dry Etching of GaN and Related Materials: Comparison of Techniques. IEEE Journal of Selected Topics in Quantum Electronics. 4 (3), 557–562.

173 Loeb, G. E., Bak, M. J., Salcman, M., Schmidt, E. M., 1977. Parylene as a Chronically Stable, Reproducible Microelectrode Insulator. IEEE Transactions on Biomedical Engineering. BME–24 (2), 121–128.

Love, J. C., Estroff, L. A., Kriebel, J. K., Nuzzo, R. G., and Whitesides, G. M., 2005. Self– Assembled Monolayers of Thiolates on Metals as a Form of Nanotechnology. Chemical Reviews. 105 (4), 1103–1169.

Madou, M. J., 2002. Fundamentals of Microfabrication: The Science of Minimization. 2nd Edition, CRC Press, Boca Raton, Florida.

Maner, A., S. Harsh, S., and W. Ehrfeld, W., 1988. Mass Production of Microdevices with Extreme Aspect Ratios by Electroforming. Plating and Surface Finishing. 75 (3), 60–65.

Marques, C., Desta, Y. M., Rogers, J., Murphy, M. C., and Kelly, K., 1997. Fabrication of High– Aspect–Ratio Microstructures on Planar and Nonplanar Surfaces Using a Modified LIGA Process. Journal of Microelectromechanical Systems. 6 (4), 329–336.

Ohno, I., 1988. Methods for Determination of Electroless Deposition Rate. Proceedings of the Symposium on Electroless Deposition of Metals and Alloys. Honolulu, Hawaii, 129–141.

Park, S. –E., and Shrout, T. R., 1997. Relaxor Based Ferroelectric Single Crystals for Electro– Mechanical Actuators. Materials Research Innovations. 1, 20–25.

Sagiv, J., 1980. Organized Monolayers by Adsorption. 1. Formation and Structure of Oleophobic Mixed Monolayers on Solid Surfaces. Journal of the American Chemical Society. 102 (1), 92– 98.

Safari, A., 1994. Development of Piezoelectric Composite for Transducers. Journal de Physique III. 4, 1129–1149.

Schmidt, E. M., McIntosh, J. S., and Bak, M. J., 1988. Long–Term Implants of Parylene–C Coated Microelectrodes. Medical and Biological Engineering and Computing. 26, 96–101.

Schrieber, F., 2004. Self–Assembled Monolayers: from ‘Simple’ Model Systems to Biofunctionalized Inferfaces. Journal of Physics: Condensed Matter. 16, R881–R900.

Shung, K. K., Cannata, J. M., and Zhou, Q. F., 2007. Piezoelectric Materials for High Frequency Medical Imaging Applications: A Review. Journal of Electroceremics. 19, 139–145.

Smith, W. A., 1986. Composite Piezoelectric Materials for Medical Ultrasonic Imaging Transducers – A Review. 1986 Sixth IEEE International Symposium on Applications of Ferroelectrics. 249–256.

174 Takahashi, C., Takahashi, J., Notomi, M., and Yokohama, I., 2000. Accurate Dry Etching with Fluorinated Gas for Two–Dimensional Si Photonic Crystals. Materials Research Society Symposium Proceedings. 637, E1.8.1–E1.8.6.

Thein, M., Asphahani, F., Cheng, A., Buckmaster, R., Zhang, M., and Xu, J., 2010. Response Characteristics of Single–Cell Impedance Sensors Employed with Surface–Modified Microelectrode. Biosensors and Bioelectronics. 25, 1963–1969.

Todd, B., Flack, W. W., and White, S., 1999. Thick Photoresist Imaging Using a Three Wavelength Exposure Stepper. Micromachining and Microfabrication Process Proceedings, SPIE. 3874 (40), 1–15.

Ulman, A., 1996. Formation and Structure of Self–Assembled Monolayers. Chemical Reviews. 96, 1533–1554.

Wang, S., Li, J. F., Watanabe, R., and Esashi, M., 1999 (a). Fabrication of Lead Zirconate Titanate Microrods for 1–3 Piezocomposites Using Isostatic Pressing with Silicon Molds. Journal of American Ceramic Society. 82 (1), 213–215.

Wang, S., Li, X., Wakabayashi, K., and Esashi, M., 1999 (b). Deep Reactive Ion Etching of Lead Zirconate Titanate Using Sulfur Hexafluoride Gas. Journal of American Ceramic Society. 82 (5), 1339–1341.

Wang, L. P., Wolf, R., Zhou, Q., Trolier–McKinstry, S., and Davis, R. J., 2000. Wet–Etch Patterning of Lead Zirconate Titanate (PZT) Thick Films for Microelectromechanical systems (MEMS) Applications. Materials Research Society Symposium Proceedings. 657, EE5.39.1– EE5.39.6.

Wells, P. N. T., 1977. Biomedical Ultrasonics. Academic Press, London.

Yamashita, Y., 2001. Piezoelectric Materials in the 21st Century. Proceedings of the 2000 12th IEEE International Symposium on Applications of Ferroelectrics. 1, 139–144.

Yokomizo, Y., Takahashi, T., and Nomura, S., 1970. Ferroelectric Properties of Pb(Zn1/3Nb2/3)O3. Journal of the Physical Society of Japan. 28 (5), 1278–1284.

Zagzebski, J. A., 1996. Essentials of Ultrasound Physics. 1st Edition, Mosby, Inc., St. Louis.

175 Chapter 8 CELL MEMBRANE CHARACTERIZATION AT THE SINGLE–CELL LEVEL

8.1. Cell Membrane Characterization Using IMA Chip

Characterizing the frequency responses of the cell–electrode heterojunction impedance has been the fundamental mechanism of signal detection in Electric Cell– substrate Impedance Sensing (ECIS) [Xiao et al., 2002] and Electrochemical Impedance

Spectroscopy (EIS) [Tlili et al., 2003] techniques. In addition, alterations in the impedance characteristics of cell–covered microelectrodes due to exposing living cells,

utilized as sensing bio–elements, to biologically active substances [Pancrazio et al., 1999;

Tlili et al., 2003] render potential applications of single–cell–based biosensors in

drug/bacteria detection and identification (chapter 5). In a traditional electrical–

impedance–based biosensing device, a large cell population is often cultured over one

large electrode because of the lack of control over cell isolation and patterning on the

sensor substrate. Although averages of cell properties, such as proliferation, motility,

and cell–cell separation, can be achieved over the cell population by traditional IMA

biosensing schemes, they have been almost impossible to examine individual cells and to

precisely monitor changes in the cell membrane properties of individual cells. In order

to obtain a more fundamental understanding of cell nature and cell metabolism, studies

should also be emphasized on the single–cell level. However, there exists a number of

challenges in single–cell–level studies including the time–consuming and low–yield

manipulation of individual cells, the transduction of weak biological signals through

sensor configurations, and the implementation of proper models on a single–cell level to

analyze and interpret the experimental data.

176 This chapter describes the underlying sensing mechanism of single–cell–based

Integrated Microelectrode Array (IMA) biosensors, which is investigated via experimental and modeling studies. IMA chips were microfabricated as described in

Chapter 7. The sensitivity and feasibility of IMA chips at single–cell–level sensing schemes were demonstrated by successful characterizations of 1), cell membrane properties of mouse fibroblast cells (NIH3T3); and 2), cell–electrode heterojunctional interface, which are influenced by two types of surface chemistries. Bioinstrumentations and methodologies employed in this demonstration render a foundation for cell–based and cellular impedance sensing applications. Not only do theoretical models developed

in this study provide insightful understandings of the working mechanism of cell–based

and cellular impedance sensing, they also reveal the influence of interfacial properties of

cell–electrode–heterojunction on the cellular and cell–based impedance sensing

schemes.

8.2. Materials and Methods

8.2.1. Surface Modification of IMA Chip and Single–Cell Immobilization

In a single–cell–level impedance biosensor, it is very important that individual cells should be isolated and selectively immobilized onto individual cell–sized electrodes in a repeatable, highly–efficient, and controllable manner. Biological signals transduced from impedance sensors are usually weak and occasionally unobservable (or indistinguishable to that of control or baseline data) even in multiple–cell level spectroscopy [Tlili et al., 2003]. A significant factor that accounts for the observed weak signals is poor cell adhesion on the electrode [Lo and Ferrier, 1998; Tlili et al., 2003].

Additionally, when a probing current is injected through the electrode, poor cell

adhesion also results in larger paracellular currents, i.e., currents that flow between the

bottom (the ventral part) of the cell and the top surface of the substrate (or electrode),

177 and consequently, the measured impedance–related signal does not contain much information on the cell itself, especially the information about the cell membrane

[Huang et al., 2004; Lo and Ferrier, 1998]. Many efforts have been made to achieve

tight cell–electrode adhesion by selectively and specifically modifying the surface

chemistry of specially designed microelectrodes with dimensions comparable to those of

single cells [Asphahani et al., 2008]. This surface–mediated approach renders

successful isolation and immobilization of individual cells onto the microelectrodes as

well as facilitating the precise control over cell adhesion, cell viability, and cell

morphology in single–cell biosensors [Nam et al., 2004; Veiseh et al., 2004, Veiseh et al.,

2007]. In this approach, the surface chemistries of microfabricated surfaces or biochips

are modified by SAM–based biofunctionalized interfaces and subsequent surface

derivitizations. For instance, the surface chemistry of the IMA chip can be modified in

such as way that gold electrode surface presents biocompatible short peptide (KRGD:

lysine–arginine–glycine–aspartic acid) or fibronectin (extracellular cell adhesion

molecule). On the other hand, surface modification can also make silicon dioxide surface

(i.e., the insulation layer) on the chip present carboxyl functional groups. Then, the

modified SiO2 surface becomes hydrophilic and eventually repels the cells onto the

biofunctionalized gold electrodes. After individual cells are trapped or anchored onto

the sensing or working electrodes, they tend to spread and cover over the sensing

microelectrodes. Furthermore, the modified (or biofunctionalized) surface of gold

microelectrodes also enhances the cell adhesion. Figure 8–1 shows Human Umbilical

Vein Endothelial Cells (HUVEC) immobilized on an Au–square array with SiO2

background. The schematic of surface chemistry modification methods is also illustrated

in figure 8–2.

178

Figure 8–1. Optical micrographs of Human Umbilical Vein Endothelial Cells (HUVEC) immobilized on gold patterns presented on silicon oxide substrates with gold patterns coated with (a) fibronectin, (b) physically adsorbed REDVY (Ly–Arg–Glu–Asp–Val–Tyr), and (c) covalently bound KREDVY (Arg–Glu–Asp–Val–Tyr). The insets show a magnified cell image for each case to reveal the cell morphology and cell spreading. After [Veiseh et al., 2007].

179

Figure 8–2. A schematic representation of surface molecular engineering of a gold– silicon dioxide substrate for guided (or surface–mediated) cell immobilization and adhesion. Adapted from [Veiseh et al., 2004].

8.2.1.1. Protocols for Modification of IMA Chip’s Surface Chemistries

Mouse fibroblast cells (NIH3T3) were immobilized onto microelectrodes through a

lysine–arginine–glycine–aspartic acid (KRGD) short peptide–mediated or fibronectin

extracellular–adhesion–molecule mediated natural cell adhesion process [Asphahani et

al., 2008]. The following materials and chemicals are used: Remover PG (MicroChem,

Newton, MA); Nano–Strip™ 2X (Cyantek, Fremont, CA); 11–mercaptoundecanoic acid

95% (11–MUA), N–hydroxysuccinimide 97% (NHS), 1–ethyl–3–(3–(dimethylamino)– propyl) carbodiimide (EDAC), trypsin–EDTA (ethylenediamine tetra–acetic acid), fibronectin (from bovine plasma), glutaraldehyde, and paraformaldehyde (Sigma–

Aldrich, Milwaukee, WI); 2–[methoxy–(polyethyleneoxy)–propyl] trimethoxysilane

(PEG) (Mw = 460–590 Da; Gelest, Morrisville, PA); Opti–MEM® I Reduced–Serum

180 Medium (O–MEM) liquid (w/o phenol red), 1X phosphate buffered saline solution

(PBS), fetal bovine serum (FBS), penicillin–streptomycin–neomycin (PSN) antibiotic

100X mixture (Invitrogen, Carlsbad, CA), and lysine–arginine–glycine–aspartic acid

(KRGD) peptide (SynBioSci, Livermore, CA). All solvents, including toluene and

triethylamine, were HPLC grade and were purchased from Aldrich (Milwaukee, WI).

Absolute ethanol was deoxygenated by dry nitrogen before use. Mouse fibroblast

(NIH3T3) cells were obtained from the American Type Culture Collection (Manassas,

VA).

The fabricated IMA chips were immersed in NanoStrip™ 2X piranha solution at

room temperature for 30 minutes, rinsed thoroughly with DI water, and dried under

nitrogen. The chips were subsequently placed in 20 mM solution of 11–MUA for 16

hours to form a self–assembled monolayer (SAM) on the gold electrodes. Following the

11–MUA reaction, the chips were sonicated in ethanol and then immersed in a solution

containing 3 mM PEG and 1% triethylamine in anhydrous toluene at 60 ˚C for 18 hrs.

The PEG–modified SiO2 surfaces were cleaned by sonicating in toluene, ethanol, and DI water before drying under nitrogen. For surfaces to be covalently linked with biological ligands (e.g., KRGD peptide or fibronectin), substrates were immersed in an aqueous solution of 150 mM EDAC and 30 mM N–hydroxysuccinimide (NHS) for 30 min to attach the NHS ester intermediate to activate carboxylate groups of the alkanethiol SAM to chemically bond primary amino groups of the ligand, where the lysine residues displace the NHS group during the reaction. The substrates were then sterilized with absolute ethanol for 15 min and exposed to the biological ligand at a concentration of 100

μg/mL in PBS of 8.2 pH at room temperature for 1 hour. To remove loosely bound moieties after each step of the surface modification, the chips were rinsed with its original buffer solution and DI water, respectively.

181 8.2.1.2. Protocols for Single–Cell Immobilization on Sensing/Working

Microelectrodes

NIH3T3 cell line was cultured in 75 cm2 flasks at 37 ˚C in a humidified atmosphere with 5% CO2. The cell medium contains 10% FBS and 1% PSN antibiotic in O–MEM.

The medium was changed every third day. For cell adhesion, 1.0 mL of NIH3T3 cells at a concentration of 100 × 103 cells/mL was plated onto the KRGD– (or) fibronectin– patterned IMA chips. The cells were allowed to adhere to the substrates for 24 hrs. under the standard culture conditions. Excess (or) wandering cells on PEG–modified

SiO2 surface were washed away with O–MEM solution after thorough microscopic

inspection showed that working microelectrodes were covered with individual cells.

8.2.2. Instrumentation and Impedance Spectroscopy

8.2.2.1. Phase–Sensitive Lock–in Detection Technique

Lock–in amplifiers use a technique known as phase–sensitive detection to single out

the component of the signal at a specific reference frequency and phase. Noise signals, at

frequencies other than the reference frequency, are rejected and do not affect the

measurement. Therefore, the heart of the lock–in amplifier is the phase–sensitive

detector (PSD), which is also known as a demodulator or mixer. The PSD detector

operates by multiplying two signals together. Typically, an experiment is excited at a

fixed frequency (from an oscillator or a function generator), and the lock–in amplifier

detects the response from the experiment at the reference frequency. Figure 8–3

illustrates the schematics of a typical lock–in amplifier.

182

Figure 8–3. A Schematics of a typical lock–in amplifier. Phase–Sensitive–Detector (PSD) and low pass filters are the heart of lock–in technique. After [PerkinElmerTM instruments, 2000].

Suppose that the response signal from the experiment is Vsig sin(ωr t + θsig) where Vsig

is the signal amplitude, ωr is the signal frequency and θsig is the signal phase. The lock–in internal reference signal is VL sin(ωL t + θref), where VL is the reference signal amplitude,

ωL is the reference signal frequency and θref is the reference signal phase. First, the lock– in amplifier amplifies the input signal and then multiplies it by the lock–in reference signal using a phase–sensitive detector or multiplier. Therefore, the output of the PSD is simply the product of the two sinusoidal waves as follows:

VPSD = VsigVL sin(ω rt +θsig )sin(ω Lt +θref ) 1 = V V cos ⎡ω −ω ⎤t +θ −θ −cos ⎡ω +ω ⎤t +θ +θ sig L { (⎣⎢ r L ⎦⎥ sig ref ) (⎣⎢ r L ⎦⎥ sig ref )} 2

Therefore, the PSD output is two AC signals, one at the difference frequency (ωr – ωL)

and the other at the sum frequency (ωr + ωL). If the PSD output is passed through a low

pass filter, the AC signals are removed. What will be left in the general case is nothing.

However, if ωr equals ωL, (or the frequencies of response signal and reference signal are

the same), the difference frequency component of PSD output will be a DC signal. In this

case, the filtered PSD output will be:

183 1 1 VPSD = VsigVL cos(θsig −θref )= VsigVL cos(Θ) 2 2

The output of a PSD will be a very nice DC signal if ωr equals ωL and is proportional to the

signal amplitude and the phase difference between the response signal and the reference

(i.e., Vsig cos(Θ), where Θ = θsig – θref). Since the reference phase is fixed throughout the

experiment any change in amplitude or phase of the signal will change the output of PSD

and will be detected.

Fourier's theorem basically states that any input signal can be represented as the sum of many sine waves of differing amplitudes, frequencies, and phases. This is generally considered as representing the signal in the "frequency domain". Normal oscilloscopes display the signal in the "time domain". Except in the case of clean sine waves, the time domain representation does not convey very much information about the various frequencies, which make up the signal. As mentioned previously, a lock–in multiplies the signal by a pure sine wave at the reference frequency. All components of the input signal are multiplied by the reference simultaneously. Mathematically speaking, sine waves of differing frequencies are orthogonal, i.e. the average of the product of two sine waves is zero unless the frequencies are exactly the same. Hence, the product of this multiplication yields a DC output signal proportional to the component of the signal whose frequency is exactly locked to the reference frequency. The low pass filter (which follows the multiplier) provides the averaging which removes the products of the reference with components at all other frequencies. Therefore, a lock–in amplifier, because it multiplies the signal with a pure sine wave, measures the single Fourier (sine) component of the signal at the reference frequency. For instance, suppose that the input signal is a simple square wave at frequency f. The square wave is actually composed of many sine waves at multiples of f with carefully related amplitudes and phases.

According to Fourier theorem, a 2.0 Vp–p square wave can be expressed as:

184 V t = 1.273sin ωt + 0.4244sin 3ωt + 0.2546sin 5ωt +... ( ) ( ) ( ) ( )

where ω = 2πf. The lock–in, locked to f, will single out the first component. The measured signal will be 1.273sin(ωt), not the 2.0 Vp–p. In the general case, the input consists of signal plus noise. Noise is represented as varying signals at all frequencies.

The ideal lock–in only responds to noise at the reference frequency. Noise at other frequencies is removed by the low pass filter following the multiplier. This "bandwidth narrowing" is the primary advantage that a lock–in amplifier provides. Only inputs with frequencies at the reference frequency result in an output. Lock–in amplifiers, as a general rule, display the input signal in volts rms. When a lock–in displays a magnitude of 1.0 Vrms, the component of the input signal (at the reference frequency) is a sine wave with an amplitude of 1.0 Vrms, or 2.8 Vp–p (the amplitude of the reference is usually set at

1 Vp–p). Thus, in the previous case with a 2.0 Vp–p square wave with the frequency locked to f, the lock–in would detect the first sine component, 1.273sin(ωt). The measured and displayed magnitude would be 0.90 Vrms (or 1.273/√2).

8.2.2.2. Instrumentation and Impedance Spectroscopy

In this study, the cell–electrode heterojunction impedance was characterized with voltage divider circuitry as shown in figure 8–4. Integrated Microelectrode Array (IMA) chip is used in the setup. The cell–size sensing microelectrode, on which a single

NIH3T3 cell was immobilized, was electrically connected to a function generator (Agilent

33220A) through a current–limiting 1–MΩ resistor and the large counter electrode was

grounded. All the contacts were made on Au bonding/measurement contact pads of the

IMA chip using low parasitic capacitance tungsten probes. A 100–mV peak–to–peak

sinusoidal voltage was applied from the function generator that injected a low–intensity

ac probe current signal, clamped by the limiting resistor, into the interface between the

185 sensing microelectrode and the testing cell. The resulting differential potential

(comprising both magnitude and phase components) across the sensing and grounded

counter electrodes of the IMA chip was lock–in detected with a phase sensitive digital lock–in amplifier (Stanford Research SR–810) for several frequencies (1, 2, 4, 6, 8, and

10 kHz), which provided an accurate measure of the NIH3T3’s membrane impedance and cell–electrode heterojunction impedance. The frequency band of interest (1–10

kHz) was determined based on the frequency–dependent dielectric properties of the

biological membrane ensuring that the cell membrane impedance was measured.

Figure 8–4. Schematic of single–cell–electrode characterization experimental setup. The impedance–related RMS voltages across the sensing electrode (covered with a single cell) and the large counter electrodes were recorded with a lock–in amplifier over a frequency range of 1 to 10 kHz. After [Thein et al., 2010].

186 The impedance characterization was performed with a four–step procedure as follows. First, the spreading resistance of the microelectrodes was measured in the tissue culture medium prior to cell seeding in order to obtain the baseline data. For the second and third steps, the impedance measurements were carried out, respectively, after the surfaces of the microelectrodes were coated with SAM and after subsequent biofunctionalization via covalently linking fibronectin proteins (or) KRGD peptides. The

final measurement was conducted when the surface–chemistry–modified

microelectrodes were covered with a single NIH3T3 cell after the cell seeding. The entire

set of measurements (cell–free baseline, cell–free SAM–coated, cell–free surface– chemistry–modified, and single–NIH3T3–covered modified microelectrode) was

recorded separately from a set of chips that consistently demonstrated approximately the

same cell–free baseline impedance characteristics.

8.2.3. Experimental Results

An optical photomicrograph of a single NIH3T3, immobilized on 30–µm diameter

covalently–linked–KRGD–modified microelectrode, is shown in figure 8–5 (inset). The image shows the high degree of cell spreading and coverage achieved on the modified microelectrode of IMA chip. In addition, the picture also suggests that the PEG– modified SiO2 surface (surrounding area) facilitated cell–to–cell isolation by repelling a

single NIH3T3 cell onto the modified microelectrode. The experimental measurements

were carried out based on the protocols described in Section 8.2.2. The recorded root–

mean–square (RMS) voltages across the 30–µm–diameter sensing microelectrode and large counter-electrode (of both magnitude and phase), which reflect impedances of different cell–sensing electrode heterostructures at six different operating frequencies (1,

2, 4, 6, 8, and 10 kHz), are plotted in figure 8–5. Each set of data is averaged over eight measurements taken from separate sensing microelectrodes. It was observed that the

187 reproducibility of the experimental data mainly depends on the degree of cell coverage and cell adhesion on the sensing microelectrode, which can be precisely controlled by the surface chemistry modification approaches [Asphahani et al., 2008; Veiseh et al., 2004].

As consistent and reproducible data are observed with higher degree of cell coverage, the

impedance related voltages were recorded from the electrodes onto which the cell covers

approximately >95% of the electrode area. No significant difference in the cell coverage

was observed before and after the recordings.

188

Figure 8–5. Measured impedance–related RMS voltages (magnitude and phase) across sensing/detecting microelectrodes and large counter electrodes for different cell–electrode heterostructures at 6 operating frequencies. Inset: a single NIH3T3 cell immobilized on the 30µm–diameter covalently–linked–KRGD–modified microelectrode (The scale bar is 10 µm). After [Thein et al., 2010].

189 8.2.4. Modeling and Data Fitting

8.2.4.1. Modeling the Cell–Electrode Heterojunctions with the Area–

Contact–Model–Based Distributed Circuit Network

It is known that the cell membrane exhibits dielectric and insulating properties

(capacitance and resistance) [Borkholder, 1998; Pancrazio et al., 1999]. The impedance

of the cell membrane can be modeled as a parallel combination of a capacitor that arises

from the phospholipid bilayer and a variable resistor that represents transmembrane ion

transport proteins [Asphahani and Zhang, 2007]. Figure 8–6 shows the Equivalent

Circuit Model (ECM) of non–excitable cell membrane, which has Cl–, Na+, K+, and Ca2+

transmembrane ion channels. In addition to the cell membrane impedance, the

cytoplasm, which contains ions in the intracellular fluid, is conductive and can be

modeled as a variable resistor [Mossop et al., 2004; Pilwat and Zimmermann, 1985].

Figure 8–6. The equivalent Circuit Model (ECM) of the non–excitable cell membrane, – + + 2+ which has Cl , Na , K , and Ca transmembrane ion channels. Cm represents the

capacitance of the cell membrane and RCl, RCa, and RNa,K represent the ion channels’ resistances (or channel protein resistances).

It is also important to take into account of the planar electrode impedance in modeling of the cell–electrode impedance. When an electrode is immersed in the

conductive electrolyte solution such as tissue culture media, according to the

190 electrochemical theory, a current flow is possible by charging and discharging the double layer at the interface or by oxidizing/reducing substances from the electrode or/and in the solution [Panke et al., 2008]. Subsequently, a reasonably approximated equivalent circuit model for the microelectrode–electrolyte interfacial impedance is a parallel combinatory circuit of a constant phase element (CPE) that represents the double–layer capacitance and a resistor that represents the charge transfer between the electrode and electrolyte solution [Franks et al., 2005; Lasseter et al., 2004; Panke et al., 2008].

The circular–shaped planar microelectrodes were used in developing of the equivalent circuit model (ECM) in this study. It has been reported that there exists a very thin layer of tissue culture media between the bottom (the ventral part) of the cell and the top surface of the electrode when a single cell covers over a microelectrode

[Braun and Fromherz, 2004; Huang et al., 2004; Iwanaga et al., 2001; Lo et al., 1995; Lo and Ferrier, 1998]. When an AC probe current is injected through the microelectrode,

only a portion of it will flow through the cell (transcellular current) while the rest of it

will shunt through the cleft region, i.e., thin tissue culture media layer between the bottom (the ventral part) of the cell and the substratum (figure 8–7). The ECM model developed employs the first–order approximate of the radial shunt (or paracellular) currents flowing outwards from the center of the microelectrode due to the centrosymmetry [Lo and Ferrier, 1998]. Constriction of shunt current path can be

achieved through tight cell adhesion for the resistance of the shunt current path, i.e., the

sealed resistance, is inversely proportional to the averaged cell–to–substrate separation

distance [Asphahani et al., 2008; Lo and Ferrier, 1998]. There are two different ECM

models that can represent these currents flowing through cell–electrode heterojunction:

(i) point–contact–model–based lumped circuit and (ii) area–contact–model–based

distributed circuit network. The former is the simplified version and the latter is a more

accurate model.

191

Figure 8–7. Schematic of a single cell immobilized over a planar electrode, illustrating transcellular (dashed) and paracellular current paths (solid), which are interlinked. After [Thein et al., 2010].

8.2.4.1.1. Point–Contact Model with Lumped Circuit Elements

In the point–contact model with lumped circuit elements as shown in figure 8–8a, the current flowing out of the microelectrode diverges into two paths: (i) one path through the cell (a lumped sum of transcellular currents) and the other travels radially outwards from the center of the microelectrode to the edge (a lumped sum of shunt or paracellular currents) [Borkholder, 1998]. It is assumed that a cell of radius (rcell) is centered over a surface–chemistry–modified microelectrode of radius (re) and the cell

completely covers the whole area of microelectrode. It is also assumed that cell–to– substrate distance (d) is uniform (averaged). A single resistor (Rseal) can be used to

represent the shunt current path (equation 8–1), which is directly proportional to the

radius of the cell (rcell) as well as the resistivity of tissue culture medium (ρsoln), but

inversely proportional to the averaged cell–to–substrate separation distance (or cleft

height) (d) [Borkholder, 1998]. The cross–sectional area of the shunt current path at any

location in the radial direction of the cleft equals the perimeter/circumference of the

circular ring at that location times the averaged cell–to–substrate separation distance.

192 r ⎛ ⎞ ρ L cell ρ ρ r soln soln soln ⎜ cell ⎟ Rseal = = lim dr = lim ln⎜ ⎟ (8–1) r →0 ∫r r →0 ⎜ ⎟ A o o 2πrd o 2πd ⎜ r ⎟ ⎝ o ⎠

The averaged cell–to–substrate distance (or the thickness of thin tissue culture media layer) reflects how strong the cell adheres to the substrate [Asphahani et al.,

2008; Thein et al., 2010; Borkholder, 1998]. According to the nature/mechanism of parallel combination of two impedance elements, Rseal must be on the same order or

much larger than the cell membrane impedance if changes in membrane properties are

to be observed in cell–based biosensors. In other words, the major portion of the

probing current must be transcellular. Therefore, stronger cell adhesion is very

important for single–cell–based sensors in order to have very large sealed resistance.

Cell membrane capacitances (Cm,b & Cm,t) are directly proportional to cell radius (or the surface area of the cell membrane) and can be estimated by equations 8–1a and 8–1b

(Note: they are averaged and, thus, cell membrane folding is neglected). On the other hand, ion–channel resistances (Rm,b & Rm,t) are inversely proportional to the cell radius,

which can be estimated by equations 8–2a and 8–2b. Here, the combined top and side

(the dorsal and lateral part) of the cell membrane area is approximated to be one third of

the surface of a sphere (Acell,b, and Acell,t are areas of bottom (ventral) cell membrane and top (dorsal & lateral) cell membrane respectively. (cm), and (ρm) are specific capacity and resistivity of cell membrane respectively.).

rcell 2 Cm,b = cm Acell,b = cm 2πr dr = cmπ (rcell ) (8–2a) ∫0

⎛1 ⎞ 2π π 2 ⎛4⎞ 2 ⎜ ⎟ ⎜ ⎟ Cm,t = cm Acell,t ≈ cm ⎜ ⎟ (rcell ) sinθ dθ dφ =⎜ ⎟cmπ (rcell ) (8–2b) ⎜3⎟∫0 ∫0 ⎜3⎟ ⎝ ⎠ ⎝ ⎠

ρ ρ ρ R m m m m,b = = r = 2 A cell r (8–3a) cell,b 2πr dr π cell ∫0

193 ρ ρ 3ρ R = m ≈ m = m m,t A 2π π 2 2 (8–3b) cell,t 1 r sin d d 4π r ( 3 ) ( cell ) θ θ φ ( cell ) ∫0 ∫0

Similarly, the circular–shaped surface–chemistry–modified microelectrode CPE

(CCPE,e) is directly proportional to the electrode radius (or the surface area of the

electrode, equation 8–4), and the charge transfer resistance (Re) is inversely proportional to the electrode radius (or the surface area of the electrode, equation 8–5).

(Ae, cCPE,e, and ρe are the surface area, specific CPE, and resistivity of surface–chemistry–

modified microelectrode respectively.)

re 2 CCPE,e = cCPE,e Ae = cCPE,e 2πr dr = cCPE,eπ (re ) (8–4) ∫0

ρ ρ ρ R e e e e = = r = 2 A e r (8–5) e 2πr dr π e ∫0

As illustrated in figure 8–8a, the total measured impedance across the

sensing/working microelectrode and large counter–electrode consists of the surface– chemistry–modified microelectrode impedance (CCPE,e || Re), the sealed/cleft resistance

(Rseal), the membrane capacitance and ion channel resistance over the microelectrode

(Cm,b and Rm,b), the membrane capacitance and ion–channel resistance of the top and sides (the dorsal & lateral part) of the cell (Cm,t and Rm,t), cytoplasmic resistance (Rcyto), the solution resistance (Rsoln) and the counter–electrode impedance (Zc). The large counter–electrode impedance (Zc) can be negligible (very large surface area compared to

that of sensing/working microelectrode) and the impedance of cell–electrode

heterostructure is mostly dominated by the sealed resistance and the cell impedance.

194 However, the lumped circuit model is a simplified model and does not really reflect the physical nature of cell–electrode heterojunction as shown in figure 8–7 since the paracellular currents pathways are normally meshed with currents coming out of the microelectrodes and those passing through the dorsal cell membrane (transcellular currents). Therefore, a more accurate area–contact–model–based distributed circuit network is developed (figure 8–8b) in order to analyze the experimental data.

195

(A)

(B)

Figure 8–8. Schematics of (A) point–contact (or) lumped circuit model, (B) area– contact (or) distributed circuit model of single–cell–covered surface–chemistry– modified microelectrode. After [Asphahani et al., 2008; Thein et al., 2010].

196 8.2.4.1.2. Area–Contact Model with Distributed Circuit Network

The current at any point in the 2D cylindrical cleft disc can be derived using

Kirchhoff’s Circuit Laws as shown in equation 8–6a, where ∇ is the spatial gradient

operator, ρseal is the resistivity of tissue culture media in the cleft region (ρseal/d is the

sheet resistivity of the cleft disc), cCPE is the specific CPE of the electrode, ρe is the charge transfer resistivity of the electrode, cm and ρm are specific capacity and resistivity of cell

membrane respectively, VJ, Ve, and Vm are the potentials (voltages) at the cell–electrode heterojunction at the point of interest, the electrode, and the intracellular side of the cell membrane respectively. Furthermore, β is the fractional coefficient (or phase coefficient)

of the electrode CPE and 0 < β < 1 [Bisquert et al., 2000; Biswas et al., 2006; Guyomar et al., 2008; Itagaki et al., 2002; Zoltowski, 1998]. This CPE phase coefficient takes into

account of the heterogeneity of the electrode surface and of the bulk material (e.g. water

content in the SAM) and resulting frequency dispersions. The mathematical model in

equation 8–6a and 8–6b represents the mashing nature of paracellular and transcellular

currents in 2D cylindrical cleft disc, i.e., the current along the cleft disc is balanced by the

displacement current through the electrode and by the ionic and displacement current

through the cell membrane.

⎛ d ⎞ ⎛ ∂ βV ∂ βV ⎞ 1 ⎛ ∂V ∂V ⎞ 1 −∇ ⋅ ∇V = c e − J + V − V + c m − J + (V − V ) ⎜ J ⎟ CPE ⎜ β β ⎟ ( e J ) m ⎜ ⎟ m J (8–6a) ⎝ ρ ⎠ ∂t ∂t ρ ⎝ ∂t ∂t ⎠ ρ seal ⎝ ⎠ e m

⎡ ⎤ ⎛ 1 ⎞ ∂ ⎛ d ∂VJ ⎞ ⎛ 1 ⎞ ∂ ⎛ d ∂VJ ⎞ − ⎢r ⎜ ⎟ ⎥ − ⎜ ⎟ ⎜ r ⎟ ∂r ρ ∂r ⎜ r ⎟ ∂φ rρ ∂φ ⎝ ⎠ ⎣⎢ ⎝ seal ⎠ ⎦⎥ ⎝ ⎠ ⎝ seal ⎠ (8–6b) ⎛ ∂ βV ∂ βV ⎞ 1 ⎛ ∂V ∂V ⎞ 1 = c e − J + V − V + c m − J + V − V CPE ⎜ β β ⎟ ( e J ) m ⎜ ⎟ ( m J ) ∂t ∂t ρ ⎝ ∂t ∂t ⎠ ρ ⎝ ⎠ e m

In this modeling analysis, VJ is assumed to vary in radial direction only and the current

flows outwards radially within the cleft disc. Then, the equation is simplified as follows:

197 ⎡ 2 ⎤ ⎛ d ⎞ ∂ VJ ⎛ 1 ⎞ ∂VJ ⎜ ⎟ ⎢− 2 − ⎥ ρ r ⎜ r ⎟ ∂r ⎝ seal ⎠ ⎣⎢ ∂ ⎝ ⎠ ⎦⎥ (8–6c) ⎛ ∂ βV ∂ βV ⎞ 1 ⎛ ∂V ∂V ⎞ 1 = c e − J + V − V + c m − J + V − V CPE ⎜ β β ⎟ ( e J ) m ⎜ ⎟ ( m J ) ∂t ∂t ρ ⎝ ∂t ∂t ⎠ ρ ⎝ ⎠ e m

Equation 8–6c is analogous to the following differential equation (equation 8–6d)

employed in Electric Cell–substrate Impedance Sensing (ECIS) model [Lo et al., 1995].

∂ 2V ⎛ 1 ⎞ ∂V J + J − γ 2V + α = 0 (8–6d) r2 ⎜ r ⎟ ∂r J ∂ ⎝ ⎠ where

ρ ⎛ 1 1 ⎞ γ 2 = seal + d ⎜ z z ⎟ ⎝ e m ⎠ ρ ⎛ V V ⎞ α = seal e + m d ⎜ z z ⎟ ⎝ e m ⎠ (8–6e) −1 β z = 1 / ρ + j2π f c e ( e ( ) CPE )

−1 z = 1 / ρ + j2π f c m ( m m )

here, ze and zm are specific impedance of the electrode and of the cell membrane

respectively. The general solution to the equation 8–6d is [Hildebrand, 1949]:

α V r C I r C K r J ( ) = 1 0 (γ ) + 2 0 (γ ) + 2 (8–6f) γ

where I γ r and K γ r are modified Bessel functions of the first and second kinds 0 ( ) 0 ( )

with complex arguments, respectively. However, K γ r goes to infinity as r goes to 0 ( ) zero, and rcell > r > 0; hence, C2 = 0. Thus, the solution of equation 8–6d or the voltage

at the cell–electrode heterojunction is:

198

α V r C I r J ( ) = 1 0 (γ ) + 2 (8–6g) γ

The constant C1 can be approximated by applying the following boundary value:

⎛ V − V ⎞ V r = r ≈ V = I R = in e R (8–6h) J ( cell ) m,t probe soln ⎜ R ⎟ soln ⎝ Limit ⎠

where Vm,t is the averaged potential at the extracellular side of the dorsal and lateral part

of the cell membrane, Iprobe is the total current, i.e., combined transcellular and

paracellular currents, injected through the limiting resistor (RLimit) and working

electrode (figure 8–4) (or the total current spreading through the bulk tissue culture

media), Vin is the applied potential from the functional generator (figure 8–4), and Rsoln

is the spreading resistance (figure 8–8a and 8–8b). (Note: the potential at the surface of the counter–electrode is assumed to be zero since its impedance is neglected).

Inspired by the equation 8–6c (or) 8–6d, the area–contact–model–based equivalent distributed circuit network can be developed. In this model, currents flowing through the cell–electrode heterostructure are described by weighted circuit elements. The lumped circuit elements of point–contact model are broken down into weighted circuit elements in a distributed network to reflect the true physical nature of the cell–electrode heterojunction. Therefore, the area–contact–model–based ECM for the single–cell– covered microelectrode takes into account of shunt/paracellular currents, transcellular currents, and the probe current from the microelectrode as shown in figure 8–8b

[Borkholder, 1998; Gordon et al., 1989; Weis et al., 1996; Weis and Fromherz, 1997].

As illustrated in figure 8–8b, the cell–electrode heterojunction in this model is partitioned into a series of peripheral domains or segments with microelectrode and cell radius increment (∆re and ∆rcell), which are denoted with the subscript number n. Each

199 domain or segment is represented by weighted equivalent–circuit components (Zn’s, Cn’s

and Rn’s). Like in the point–contact–model–based ECM, the counter–electrode

impedance (Zc) is negligible due to its very large surface area as compared to the

sensing/working microelectrode [Borkholder, 1998; Lo et al., 1995; Lo and Ferrier, 1998;

Xiao et al., 2002]. The following assumptions were made in the derivation of formulas

for weighted equivalent–circuit elements (summarized in table 8–1): (1) the cell is

centered over and completely covers the surface–chemistry–modified microelectrode,

and thus, the immobilized cell’s radius (rcell) is equal to the planar microelectrode’s

radius (re); (2) the current spreads through the cytoplasm of the immobilized cell.

Therefore, the total measured impedance across the sensing microelectrode and large counter electrode consists of the combination of all weighted circuit elements, including weighted surface–chemistry–modified microelectrode impedance (Ze,n), weighted sealed resistance (Rseal,n), weighted cell membrane impedance over the microelectrode (Zm,n), the total membrane impedance of the top and sides (the dorsal & lateral part) of the cell

(Cm,t||Rm,t), the cytoplasmic resistance (Rcyto), the solution (or spreading) resistance

(Rsoln), and the large counter–electrode impedance (Zc) (can be neglected).

200 Table 8–1. Formulas for weighted circuit elements of the area–contact–model–based distributed circuit network. Weighted circuit elements are functions of their respective model parameters employed in the data– fitting. These parameters include: resistivity of tissue culture media (ρs), averaged cell–to–substrate separation distance (d), charge transfer resistivity of surface–modified microelectrode (ρe), specific CPE of surface–modified microelectrode (cCPE), cell membrane resistivity (ρm), specific cell membrane capacity (cm), resistivity of cytoplasm (ρcyto), and CPE phase coefficient (β). Note: n is an integer. After [Thein et al., 2010].

Weighted Circuit Elements Formulas

Impedance of nth weighted −1 β segment of surface– Z ⎛1 R j2 f C ⎞ (8–7) e,n = e,n + ( π ) CPE,n modified microelectrode ⎝ ⎠

th Impedance of n weighted −1 segment of bottom (ventral) Z = 1 R + j2π f C (8–8) cell membrane m,n ( m,n m,n )

Impedance of dorsal and −1 Z = 1 R + j2π f C (8–9) lateral cell membranes m,t ( m,t m,t )

1 1 r ⎧ ( + 2 )Δ e ⎪ lim ρsoln d 1 2πr dr n = 1 r →0 ( ) ∫r ( ) ( ) ⎪ o o n 1 r Shunt/sealed resistance ⎪ ( + 2 )Δ e th R = ⎨ ρ d 1 2πr dr n ≥ 2 (8–10) (n weighted segment) seal,n ( soln ) ∫ n− 1 Δr ( ) ( ) ⎪ ( 2 ) e nΔr ⎪ e ρ d 1 2πr dr n = n ⎪ ( soln ) n 1 r ( ) ( ) ∫( − 2 )Δ e ⎩ Resistance of nth weighted nΔre segment of surface– R = ρ 2πr dr (8–11) e,n e ∫ n−1 Δr modified microelectrode ( ) e CPE of nth weighted segment nΔre of Surface–modified C = c 2πr dr (8–12) CPE,n CPE ∫ n−1 Δr microelectrode ( ) e Resistance of nth weighted nΔre segment of bottom (ventral) R = ρ 2πr dr (8–13) m,n m ∫ n−1 Δr cell membrane ( ) e Capacitance of nth weighted nΔre segment of bottom (ventral) C = c 2πr dr (8–14) m,n m ∫ n−1 Δr cell membrane ( ) e

Resistance of dorsal and 2π π 2 2 R ≈ ρ 1 r sinθ dθ dφ = 3ρ 4π r (8–15) lateral cell membranes m,t m ( 3 ) ∫0 ∫0 ( cell ) m ( cell )

Capacitance of dorsal and 2π π 2 4 2 C ≈ c 1 r sinθ dθ dφ = c π r (8–19) lateral cell membranes m,t m ( 3 ) ∫0 ∫0 ( cell ) ( 3 ) m ( cell )

Spreading/solution R = ρ 4r (8–20) Resistance soln s e

Resistance of cytoplasm R = ρ 4r (8–21) cyto cyto cell

201 According to the sensor circuitry in the experimental setup, the RMS voltage across the sensing microelectrode and large counter–electrode (Vout–RMS) can be derived by applying Kirchhoff’s voltage and current laws (KVL and KCL) after connecting the current limiting resistor (RLimit) and applying sinusoidal voltage source to the circuitry

with n = 5 as shown in figure 8–9. It can be noticed that (Vout–RMS) is simply the output of a voltage divider circuitry.

Figure 8–9. Schematics of sensor circuitry with 5 weighted segments or domains in the area– contact–model–based distributed circuit network.

202 The Vout–RMS derived is a function of applied frequency ( f ) as well as parameters (ρe,

cCPE, β, ρm, cm, ρcyto, and d) to be estimated. The parameters are estimated by numerically

fitting the function, Vout–RMS (ρe, cCPE, β, ρm, cm, ρcyto, d, f ), to the measured experimental data. Least Square fitting shown in equation 8–22 is used in the error minimization.

S(ρsoln , ρe , cCPE , d, ρm , cm , ρcyto )

k ⎛ ⎞ 2 ⎜ 1 ⎟ (8–22) = ⎜ ⎟ V 2π f −V 2π f ,ρ , ρ , c , d, ρ , c , ρ ∑⎜ 2 ⎟{ mea,i ( i ) out−RMS ( i soln e CPE m m cyto ) } i=1 ⎝⎜σ ⎟ i ⎠

In equation 8–22, (Vmea,i) is the complex voltage (magnitude & phase) measured at respective frequencies ( fi ), (σi) is the variance of measured (Vmea,i ’s), and ( k ) is the number of frequencies applied in each set of spectroscopy.

8.2.4.2. Numerical Data Fitting

At the given frequency, sinusoidal voltage input, and current clamping resistance, the

RMS voltage across the sensing microelectrode and counter–electrode, Vout–RMS, which

is also a function of the physical and electrical parameters (ρe, ce, β, ρm, cm, ρcyto, and d),

were derived from the area–contact–model–based equivalent circuit network shown in

figure 8–9. These model parameters were then numerically extracted by iteratively

least–square–fitting the function to the measured impedance–related RMS output

voltages shown in figure 8–5 [Bodmer et al., 2005]. Numerical data fitting was carried out in Mathematica® Software and methods employed for minimization include

Levenberg Marquardt, Conjugate Gradient, and Principal Axis. A multiple–step–fitting approach was employed in our multi–parameter extraction process in order to minimize the errors [Gordon et al., 1989; Lo et al., 1995]. For computing efficiency, the number of weighted segments (n) used in the area–contact model was 5 (figure 8–9). The estimated model parameters and their respective error values (summarized in table 8–2) give the best fit of the model to the experimental measured data since they were

203 estimated based on both the reproducibility of experimental measured data and the stability of the model parameters [Bodmer et al., 2005; Lo and Ferrier, 1998]. The cell membrane resistivity (ρm), cell membrane capacity (cm), and cell–to–substrate separation distance (d), are of particular importance among all the parameters as the accurate determination of their values is vital to the sensing function and mechanism of the single–cell IMA biosensors under study. In other words, as the cell membrane resistivity and capacity contain information related to cell membrane activities and dynamics, alterations in the cell membrane due to the external or internal stimuli will be accurately represented by the changes in the values of these two model parameters.

Table 8–2. List of estimated values of ECM model parameters that give the best fit to the experimental data. After [Thein et al., 2010].

Estimated Values

Single NIH3T3 Single NIH3T3 Immobilized on the ECM Model Parameters Immobilized on the Covalently–linked– Covalently–linked– fibronectin– KRGD–modified modified Microelectrode Microelectrode

Charge transfer resistivity of surface– 2 10.97 ± 0.55 11.83 ± 0.59 modified microelectrode (ρe) (kΩ-cm )

Specific CPE of surface–modified 2 186.4 ± 9.3 195.8 ± 9.8 microelectrode (cCPE) (µF/cm )

CPE phase coefficient ( β ) 0.867 ± 0.043 0.914 ± 0.046

Specific cell membrane capacity (cm) 97.9 ± 4.9 83.7 ± 4.2 (µF/cm2)

2 Cell membrane resistivity (ρm) (kΩ–cm ) 0.0312 ± 0.0016 0.0286 ± 0.0014

Resistivity of cytoplasm (ρcyto) (kΩ–cm) 0.606 ± 0.030 0.679 ± 0.034

Averaged cell–to–substrate separation 14.52 0.73 11.59 0.58 distance (d) (nm) ± ±

204 8.2.5. Analysis and discussions

8.2.5.1. Signal Enhancement via Cleft–Thickness Control over

Surface–Modified Sensing Microelectrodes

It is well known that the accuracy of cell–electrode impedance characterization will be enhanced effectively by reducing the “cleft” thickness at the cell–electrode

heterojunctional interface. When cells are immobilized on planar metal electrodes, cell

adhesion is mediated by protein molecules that protrude from the cell membrane

(integrins, glycocalix, etc.) and those that are deposited on the substrate (extracellular matrix proteins). These proteins keep the lipid core of the ventral part of the membrane at a certain distance from the substrate, stabilizing a cleft between the cell and the chip, which is filled with electrolyte solution [Fromherz, 2003]. The resulting shunt resistance that arises from the cleft may seriously degrade the actual impedance characterization of the cell membrane, and, in some cases, even dominate the measured data.

8.2.5.1.1. Influence of Cell Adhesion on the Lock–in Measurement

The estimated averaged cell–to–substrate separation distance for NIH3T3 cell on the surfaces of both covalently–linked–KRGD–modified & covalently–linked–fibronectin– modified microelectrodes is in the range of 11–16 nm, which is obtained via numerically data–fitting the ECM model to the experimentally recorded data in the present study

(figure 8–5) and suggests very tight adhesion between the single NIH3T3 cell and the surface–chemistry–modified Au microelectrode [Braun and Fromherz, 1998, 2004]. In comparison, the averaged cell–to–substrate separation distance for fibroblast cells and rat neural cells on the surfaces of substrates coated with physically adsorbed fibronectin or laminin was estimated to be approximately 50–100nm according to previous studies carried out using fluorescent interferometry and voltage–sensitive fluorescent dye

[Braun and Fromherz, 1998; Braun and Fromherz, 2004; Iwanaga et al., 2001; Garcia et

205 al., 1997]. These results, therefore, indicate that covalently linking biocompatible peptides or cell–adhesion molecules on the substrate surface improves the cell adhesion significantly. This observation is also consistent with the finding that covalently– linked–KRGD peptides on the Au surfaces mediates a higher degree of cell spreading than physically–adsorbed–KRGD peptides and enhances the signal transduction of impedance–based cellular sensors [Asphahani et al., 2008; Figure 8–10 and 8–11].

Figure 8–10. (A) Epi–DIC images of mouse fibroblast (NIH3T3) single cells patterned on 25µm– and 30µm–detecting electrodes and multiple cells patterned on a 110µm–detecting electrode, which are modified with covalently–bound KRGD peptide. (B) Fluorescent images of nuclei– (blue) and membrane– (green) stained NIH3T3 cells on electrodes of three different sizes and modified with physically absorbed (left, p–electrode) or covalently bound (right, c– electrode) KRGD. The scale bar is 25µm in (A), and 5µm for 25–µm and 30–µm electrodes and 20 µm for 110–µm electrodes in (B). After [Asphahani et al., 2008].

206

Figure 8–11. Voltage magnitudes, measured across the working electrode and the counter electrode, of cell–free and mouse fibroblast (NIH3T3) cell–covered electrodes with surface areas of 625µm2 (25µm x 25µm), 900µm2 (30µm x 30µm), and 12,100µm2 (110µm x 110µm). The electrodes were pre–coated with either covalently–bound or physically–absorbed KRGD peptides. After [Asphahani et al., 2008].

It can be concluded that stronger cell adhesion and higher degree of cell spreading on the electrode improve the transduction of signal from cell–based biosensor circuitry.

According to the ECM models, the shunt resistance increases with the cell radius (or cleft disc radius) and is inversely proportional to the averaged cell–to–substrate separation distance according to the area–contact model. Therefore, stronger cell adhesion on the surface of the electrode, provided that the cell spreads entirely on the electrode, results in smaller averaged cell–to–substrate separation distance that, in turn, yields larger values of shunt resistances. In order to understand and analyze the effect of cell adhesion on the single–cell–based sensor circuit response, the normalized RMS voltage magnitudes across the sensing microelectrode (20µm or 30µm in diameter) and the

counter electrode (i.e., v = |Vcell-covered/Vbaseline|), which represent the strength of the

207 single–cell IMA biosensor response, were calculated and plotted (figure 8–12) for different averaged cell–to–substrate separation distances (d), ranging from 10 to 120 nm, under the probing condition of 1 kHz frequency, 10MΩ current clamping resistance,

and an input impedance (Zin) of 20 GΩ || 25 pF for the lock–in detection (estimated

ECM model parameters from table 8–2 were used in the calculations). The computed plot shows that when the cell–to–substrate separation distance is reduced from 120 to 10 nm, the single–cell IMA biosensor response, represented by the normalized RMS voltage magnitude, is enhanced by the factors of approximately 1.8 and 2.4 for 20µm and 30µm electrodes, respectively.

Figure 8–12. Simulated plot: normalized RMS voltage magnitudes, measured across the sensing microelectrode and the counter electrode, as a function of averaged cell–to– substrate separation distance. After [Thein et al., 2010].

208 The fact that the calculated normalized RMS voltage magnitudes for the same type of cell is smaller in the 20µm–diameter electrode compared to that of the 30µm–diameter electrode indicates the strong cell adhesion, and hence the minimized cell–to–substrate separation distance is absolutely essential for smaller electrodes suitable for single–cell– based sensing schemes, which utilize small–sized biological cells. A plausible explanation for this phenomenon is that cell–to–substrate separation must be dominantly smaller in order to obtain a large shunt resistance due to shortening of paracellular or shunt current paths in smaller electrodes covered with a small–sized cell.

Otherwise, the injected ac current will shunt through the paracellular paths and the signal becomes hardly distinguishable from that of cell–free baseline.

8.2.5.2. Frequency–dependent characteristics

The ultimate goal of Integrated Microelectrode Array (IMA) chips is to be implemented in cell–based and cellular sensor configurations. The successful characterization of cell membrane at single–cell level demonstrated the feasibility of

IMA chips. However, other characteristics such as frequency response, sensitivity, and signal–to–noise ratio of the sensor configuration are also important. Hence, frequency– dependent characteristics of IMA chips employed with NIH3T3 cells are investigated.

These characteristics include: (i) frequency–dependent impedance difference between cell–covered and cell–free electrode, (ii) impedance spectrum of cell–electrode heterostructures, and (iii) frequency–dependent response characteristics of IMA sensor configuration (overall circuitry), which is employed with cell–electrode heterostructures.

To investigate the frequency–dependent characteristics of the Integrated–

Microelectrode–Array–based single–cell sensing scheme, firstly, the spectrum of impedance difference between the single–NIH3T3–covered electrode and the unmodified cell–free baseline electrode was calculated using the area–contact–model–

209 based circuit network over the frequency range between 100 Hz to 1MHz (the estimated values from table 8–2 are used). Figure 8–13 plots the spectrum of the computed impedance difference (ΔZ = Zcell–covered – Zcell–free baseline), which is associated with the

covalently–linked–KRGD–mediated or covalently–linked–fibronectin–mediated cell adhesion of single NIH3T3 cells on 30µm–diameter microelectrodes and provides a mean of measure of the cell–electrode–heterojunction interfacial impedance as well as the cell membrane impedance. The frequency–dependent characteristic of the

magnitudes of the impedance difference (|ΔZ|) approximates to a linear function in log–

scale up to the frequency of 50 kHz. This characteristic is governed by the resistive and

capacitive properties of the cell membrane, the binding–specific cell adhesion molecules

(KRGD or fibronectin) and their regulated cell adhesion, and the intracellular cytoplasm.

The cell membrane becomes more capacitive and leaky at higher frequencies (≥50 kHz)

and, thus, the capacitive elements of NIH3T3 are effectively short–circuited at higher

frequencies [Gordon et al., 1989]. The impedance change due to the present of a single

NIH3T3 on the microelectrode, therefore, becomes frequency independent at high

frequencies (≥50 kHz) as it is dominated by the resistance of cytoplasm (Rcyto). This resistive nature of the single–NIH3T3–covered microelectrode impedance at high frequencies is also revealed by the reduced phase angles of the impedance difference in the spectral diagram (figure 8–13). An important conclusion drawn from the spectrum analysis is that there exists a suitable frequency band within which the electrical properties of multiple cellular components become dominant in measured signals and variations or changes in these properties can be detected by the lock–in technique.

210

Figure 8–13. Simulated plot of frequency–dependent impedance difference, referenced to the cell–free baseline, after a single NIH3T3 cell was immobilized on the microelectrode (30µm–diameter) via different surface–chemistry–mediated cell adhesion processes. After [Thein et al., 2010].

Secondly, the impedance spectra of overall cell–electrode heterostructure is calculated. The computed spectrum of overall impedance magnitude of single–

NIH3T3–covered microelectrode in figure 8–14 is found to trace along the asymptotic approximation line of the NIH3T3’s membrane capacitance and the microelectrode constant phase element (CPE) within the frequency band of 100 Hz to 50 kHz. In addition, the phase shifts of the impedance due to the presence of a morphologically controlled single NIH3T3 on the microelectrode, represented by the phase angles of the

211 impedance difference in figure 8–13, are very prominent within the same frequency band. These characteristics suggest the constriction of shunt currents due to tight cell

adhesion (most of the AC probe current is flowing through the cell) and confirm that the measured signal in this frequency range is mostly dominated by NIH3T3’s cell membrane impedance as well as the surface–modified microelectrode’s CPE.

Figure 8–14. Overall impedance magnitude spectrum of cell–electrode heterostructure (single–NIH3T3–covered covalently–linked–KRGD–modified gold microelectrode, 30µm–diameter). Dashed lines are asymptotic approximation lines representing dominant electrical properties in the measured data. After [Thein et al., 2010].

212 Again, the most important aspect of these characteristics for single–cell IMA–based sensing schemes is that real–time changes and variations in NIH3T3’s cell membrane properties (such as membrane permittivity and conductivity) due to external or internal stimuli can be transduced and revealed in this frequency band. The uncovered frequency band also provides maximum sensitivity in the single–cell IMA–based sensing scheme and can be utilized to optimize the operation of the sensor circuit in real–time analysis so that the time needed for impedance recording and processing can be reduced.

Thirdly, the optimal frequency range for the maximum sensor response and sensitivity in the present study was also investigated by simulating the frequency– dependent sensor response characteristics of the IMA sensing circuitry. The normalized

RMS voltage magnitudes, v = |Vnon–baseline/Vbaseline|, were calculated (with 1 MΩ current

limiting resistor) over the frequency range between 100 Hz to 1 MHz and plotted for

NIH3T3 cells immobilized on covalently–linked–KRGD– or fibronectin–modified microelectrodes, respectively, as shown in figure 8–15. A small impedance difference between the two cell–electrode configurations, resolved by the difference in normalized

RMS voltage magnitude plots, is observed due to two different surface chemistries. This difference is unambiguously resolved over the frequency range of 1–100 kHz. This suggests the frequency–dependent impedance–specific sensitivity of the single–cell– based IMA impedance sensor circuitry. The reduced impedance–specific resolution and sensitivity at lower (<1 kHz) frequencies, where resistive properties dominate, also suggests that modifying the surface chemistry with KRGD peptides or fibronectin proteins alters significantly on the double–layered capacitance of the electrode and cell membrane capacitance of NIH3T3 (see table 8–2). A plausible explanations are: (i) different proteins could influence the cell membrane folding and number of focal adhesion points. Cell membrane folding would increase the cell membrane capacity per unit area; (ii) different number of charges (or polar molecules) in the protein could affect

213 the equilibrium charge distribution of at the electrode–electrolyte interface; (iii) it is also known that cell adhesion proteins mediate or trigger excretion of cell metabolites through cell signaling pathways, which are deposited on the substrate or matrix [Albert et al., 2004]. Hence, these metabolites dumped into the cleft are expected to affect the cell–electrode interfacial properties, causing small difference in overall impedances that can be detected by the IMA sensor configuration.

Figure 8–15. Frequency–dependent response characteristics of Integrated Microelectrode Array (IMA) sensor configuration after single cell immobilization. After [Thein et al., 2010].

214 The response of IMA sensor configuration remains constant or frequency independent at higher frequencies (>100 kHz) as shown in figure 8–15 for the impedances are mainly contributed by the resistance of the cytoplasm at higher frequencies. No difference in sensor responses between two different types of modified electrodes is observed at higher frequencies above 100 kHz. This could also suggest that the resistance of NIH3T3 cytoplasm remains unaltered in substituting from fibronectin– mediated to KRGD–mediated cell adhesion. Furthermore, the response characteristics

in figure 8–15 also suggest that the probing frequency must be within 1–100 kHz if the real–time variations in NIH3T3 cell membrane due to external or internal stimuli are to be monitored. It is also expected that utilizing different types of cells as a sensing bio– element and different surface chemistries could shift the frequency band and optimum operating/probing frequencies. Therefore, a proper equivalent circuit model is necessary and prior experimental measurement and analysis must be carried out to obtain the insight knowledge of the contribution of each property of a single cell to the overall impedance, as well as determining optimum sensor circuit operating/probing frequencies for single–cell–based biosensing applications (e.g., real–time constant– frequency analysis of the single–cell response to a given dose of a pharmaceutical agent).

8.3 Automation & Upgrade on IMA Chip Instrumentations and

Sensor Platforms

8.3.1. Multiplexing and Data Acquisition

Microelectrode arrays are employed in the chip in order to improve the efficiency and yield of the experiment. Automatic data acquisition and synchronization between instruments will further improve the efficiency and yields of cellular assays. Therefore,

multiplexing and data acquisition circuitry was constructed and integrated to the IMA

chip platform as shown in figure 8–16. The integrated microelectrode array chip was

215 placed inside the portable chip carrier/holder (custom-built) and wires from the chip

carrier were connected to multiplexer relay switching system (2 Agilent 34901A 20

channel multiplexer (2/4–wire) modules installed in Agilent 34970A data

acquisition/control unit) through current limiting resistors (high tolerance). The

function generator (Agilent 33220A) and lock–in amplifier (Stanford Research System

SR–810) were also connected to the input and output of the multiplexer respectively so

that ac current injection from the function generator and data recording from the lock–

in amplifier can be synchronized. The lock–in amplifier is interfaced to the chip carrier

through unity–gain high–input–impedance pre–amplifiers (in–house built) in order to reduce the loading effect (Impedances from biological samples are usually high and, thus, instruments with high–input impedance are always required). Each pre–amplifier channel includes two–stage cascaded pre–amplifier setup where the signal passes the first stage of in–amp (differential) and then is fed through the second stage of op–amp

(buffer) to the output (figure 8–16b). Polytetrafluoroethylene (PTFE) coated wires were used in the multiplexing network in order to reduce parasitic coupling between wires

(PTFE has much lower dielectric constant and much higher bulk resistivity than PVC elastomer traditionally used in electrical wire insulation). Figure 8–17 shows the test setup with EG&G 5206 two–phase lock–in analyzer. Automated excitation and data acquisition is accomplished by connecting the function generator, lock–in amplifiers, and multiplexers to a PC through RS–232 and IEEE GPIB interface respectively. Data acquisition from the lock–in amplifier is controlled by LabVIEW (National

Instruments).

216

(A)

(B)

Figure 8–16. (A) Schematics of automated excitation and data acquisition circuitry, which includes a lock–in amplifiers, a multiplexer, a function generator, and a multi–channel pre– amplifier module. (B) Circuit schematics of a unity–gain ultra–high–input–impedance amplifier in each channel of the multi–channel pre–amplifier module.

217

Chip Carrier with Current Limiting Resistors

Function Generator

Unity–gain High–input–Z Multiplexer Pre–amplifiers Control Unit

EG&G 5206 Two Phase Lock–in Analyzer

Figure 8–17. A complete multiplexing/addressing setup (a test setup with EG&G 5206 two–phase lock–in analyzer).

8.3.2. Integration with Portable Tissue Culture Chamber and

Fluorescent Microscope

Biological experiments are usually coupled with microscopic observations (including

fluorescence spectroscopy). In addition, the environment in which the experiment is

carried out must be under desired controlled conditions. Therefore, Olympus upright microscope (capable of taking fluorescent images) and Nikkon portable tissue culture chamber are integrated to the IMA sensor platform. The chip carrier with the media

218 reservoir, inside which the IMA chip is mounted, can be placed inside the chamber. The

Nikkon tissue culture chamber has a temperature control unit and a gas–flow control unit. In addition, the environment inside the portable tissue culture chamber is sterilized. Therefore, this upgraded set up is very relevant for real–time long–term monitoring of the effects of chemicals (such as drugs or toxins) on the cells as well as cellular assays using the IMA chip platform.

Figure 8–18. Experiment set–up to monitor real–time responses from individual cells and multiple–cell colonies upon exposure to toxins and drugs. The inset shows the modified chip carrier with tissue culture media reservoir placed inside the portable tissue culture chamber.

219 8.4. Chapter Appendix

8.4.1. Single–Cell–Based Vs. Multi–Cell–Based Impedance Sensing in

Monitoring Cellular Response to Drug Treatment

In order to compare the single–cell–based impedance sensing to multi–cell–based impedance sensing in monitoring cellular response to drug treatment, human glioblastoma (U87MG) cells were immobilized (or seeded) on single– and multi–cell microelectrodes through ligand–mediated natural cell adhesion process (see section

8.2.1). These cancer cells seeded on both types of microelectrodes were then exposed to an ion channel inhibitor, chlorotoxin (CTX), for up to 16 hrs. The cancer cells exhibited shape, size, and morphological changes upon exposure to CTX, which can be detected by cellular impedance sensing technique using IMA chips.

Figure 8–19a show the epi–DIC images of single cells and a monolayer of multiple cells seeded on the respective electrodes before and after the 5–µM CTX treatment. The

single–cell images (right panels in figure 8–19a) reveal that single U87MG cells immobilized on the cell–size electrodes shrank or retracted after 16–hour inoculation with 5–µM CTX. Time–course normalized real impedance (measured at 2.0 kHz) of

single–cell–electrode heterostructures were also obtained for the period of 16–hour

inoculation with 5–µM CTX (figure 8–19b to 8–19e). It can be concluded from the

time–course plots, coupled with epi–DIC images, that the impedances of single–cell– electrode heterostructures significantly decrease over the period of 16–hour inoculation with 5–µM CTX due to cell shrinkage or retraction. On the other hand, no or little drop in the normalized real impedance (measured at 2.0 kHz) of multi–cell–electrode

heterostructure was observed for the same concentration of CTX over the same

inoculation period (figure 8–20a). Furthermore, changes in frequency spectrum of cell–

electrode heterostructure after cell seeding are more prominent in single–cell electrodes

than in multi–cell electrodes (figure 8–2ob).

220

Figure 8–19. (a) Schematic of IMA chip featuring optical epi–DIC micrographs of four single live U87MG cells on 30–µm working/detecting electrodes (right panel) and one 250–µm working/detecting electrode hosting a confluent monolayer of live U87MG cells (left panel) before and 16 hours post 5–µM CTX treatment. Both scale bars represent 30µm. (b)–(e) Time– course normalized real impedance of the four U87MG single cells exposed to 5 µM CTX over a period of 16 hours. Real impedance was normalized to the point just prior to CTX inoculation. [Courtesy of Fareid Asphahani]

221

Figure 8–20. (a) Average normalized real impedance of multiple U87MG cells (blue) and single U87MG cells (green) in response to 5 µM CTX, as well as to single cells receiving no CTX treatment (brown). Real impedance was normalized to the point just prior to CTX inoculation. (b) Frequency spectroscopy plot of impedance magnitude |Z| at a frequency range from 500 to 20 kHz when cell–free electrodes of 30 µm and 250 µm diameters are seeded with U87MG cells. [Courtesy of Fareid Asphahani]

These experimental findings suggest that there are two intrinsic phenomena that

degrade the signal–to–noise ratio (SNR) of multi–cell–based cellular impedance sensing

technique in monitoring the full toxicity or drug response of the cells. These two

phenomena are: 1), the initial cell coverage on the working electrode; and 2), the effect of

cell–to–cell junction on the cell shrinkage or retraction.

8.4.1.1. The Effect of Initial Cell Coverage on the Working Electrode on the

Transduced Electrical Signals

In order to analyze the effect of initial cell coverage prior to the CTX administration

on the transduced electrical signals, the point–contact–model–based equivalent electrical circuit of single–cell–electrode heterostructure is modified as illustrated in figure 8–21. In developing this application–specific equivalent circuit, the following

222 assumptions are made: 1), the cell covers almost 95–98% of the surface–chemistry– modified microelectrode just prior to the CTX administration and, thus, the electrode possesses cell–occupied area, A r , and cell–free area, A r = A r − A r ; cell ( cell ) free ( cell ) e ( e ) cell ( cell )

2), the cell adhesion is strong due to ligand–mediated cell adhesion, i.e., very large Rseal

[Asphahani et al., 2008; Thein et al., 2010]; 3), the cell impedance, Zcell, is much larger than the surface impedance of cell–covered portion of the electrode, Ze,cell, and the

surface impedance of unoccupied portion of the electrode, Ze,free, for the

probing/operating frequency used in this study (i.e., 2.0 kHz); and 4), the spreading

resistance (Rsoln) is much smaller than other electrical components in the ECM.

Figure 8–21. Modified point–contact–model–based equivalent circuit of single–cell–

electrode heterostructure. Rseal is very large due to very tight cell adhesion mediated by

ligands/peptides covalently linked to the SAM–functionalized Au electrode. Zcell is

comparably larger than Ze,ell and Ze,free at 2–kHz operating frequency.

Based on the circuit schematic shown in figure 8–21, the overall impedance of single–cell–electrode heterostructure can be described as follows:

223 −1 −1 ⎡⎛ Z j R ⎞ ⎤ cell ( ω ) seal 1 Z jω = ⎢⎜ Z jω + ⎟ + ⎥ + R (8–23) het ( ) ⎢⎜ e,cell ( ) Z j R ⎟ Z j ⎥ soln ⎝ cell ( ω ) + seal ⎠ e, free ( ω ) ⎣⎢ ⎦⎥

Furthermore, the surface impedance of cell–occupied portion of microelectrode is:

z jω Z j e ( ) (8–24) e,cell ( ω ) = 2 πr cell

The surface impedance of unoccupied portion of microelectrode is:

z jw e ( ) Z jw = e, free ( ) 2 2 (8–25) π ⎡r − r ⎤ ⎣ e cell ⎦

And the impedance of the cell is:

z jω Z j cell ( ) (8–26) cell ( ω ) = 2 πr cell

The impedance of sealed resistance (Rseal) and spreading resistance (Rsoln) can be described by equation 8–1 and 8–20 respectively.

For all operating frequencies, the overall impedance of single–cell–electrode

heterostructure can generally be described by equation 8–23. However, operating at 2– kHz frequency and having very tight cell adhesion between the cell and the electrode, the impedance of the single–cell–electrode heterostructure is mainly contributed by the

surface impedance of unoccupied portion (or cell–free region) of the microelectrode, i.e.,

⎡Z ⎤ ⎡Z ⎤ ⎡z ⎤ ⎡ A r A r ⎤ [Haung et al., 2004]. het ≈ e, free = e e ( e ) − cell ( cell ) ⎣ ⎦( f =2kHz) ⎣ ⎦( f =2kHz) ⎣ ⎦( f =2kHz) ⎣ ⎦

Therefore, the overall impedance Zhet at 2.0 kHz is inversely proportional to the cell–free

2 surface area of the electrode or, specifically, is proportional to 1 ⎡1 r r ⎤ . − ( cell e ) ⎣⎢ ⎦⎥

224 For multi–cell–electrode heterostructures, the equivalent impedance is simply the parallel combination of individual single–cell–electrode domains. Then, the total area

covered by each individual cells can be treated as the single large area covered by a very

n 2 large virtual cell, i.e., r ≈ r . Hence, the average percent change in the cell,eff ∑i=1 ( cell )

mean radius of individual cells due to the CTX toxicity effect is directly correlated to that

in the effective radius. However, the percentage of initial cell–free area just prior to the

CTX administration on the multi–cell electrode is relatively larger than that on single–

cell electrode due to cell–to–cell gap in the monolayer, i.e., r r < r r . cell,eff e,multi cell e,single

Therefore, the initial cell–occupied area on the multi–cell electrode is usually less than that on the single–cell electrode. The dependence of normalized real impedances, i.e.,

2 2 Re ⎡Z r ⎤ /Re ⎡Z r ⎤ ∝ ⎡1 − r r ⎤ ⎡1 − r r ⎤ , on the initial cell ⎣ het ( cell )⎦ het,o ( cell,o ) ⎢ ( cell,o e ) ⎥ ( cell e ) ⎣ ⎦ ⎣ ⎦ ⎣⎢ ⎦⎥

coverage or the ratio rcell,0/re at 2–kHz operating frequency during the cell shrinkage or retraction are shown in figure 8–22. The plots show that 15% reduction in cell radius from initial rcell,o yields different overall decreases in normalized real impedances (or overall decreases in real impedances, i.e., Re[Zhet,o]–Re[Zhet,ss]) for different initial cell coverage on the electrode. The trend shows that the higher is the initial cell coverage just prior to the CTX administration or rcell,0/re ratio, the larger is the decrease in normalized

real impedances for the same degree of cell shrinkages or retractions. Therefore, the

electrical signal transduced via multi–cell–based impedance sensing does not usually

reveal significant decrease in normalized real impedance for the same concentration of

Chlorotoxin administration due to the relatively larger initial cell–free area resulting

from cell–to–cell gaps.

225

Figure 8–22. Influence of initial cell coverage (rcell,o/re) on the normalized real impedancse during the cell shrinkage or retraction.

8.4.1.2. The Effect of Cell–to–Cell Junction on the Overall Cell Shrinkage or

Retraction

It is hypothesized that the overall cell shrinkage or retraction in the multiple–cell environment due to Chlorotoxin is relatively less than that in single–cell environment due to the linkage between adjacent cells in the monolayer by cell–to–cell junctions. In other words, the adjacent cells are linked or hold together in the monolayer resulting in relatively less shape, size, and morphological response to CTX. Therefore, it is conceived that the overall decrease in the normalized impedance of multi–cell–electrode heterostructure is relatively less than that in the normalized impedance of single–cell– electrode heterostructure due to lower degree of cell shrinkage or retraction and, thus,

226 the full toxicity effect by CTX may not be observed in the multiple–cell–level measurements.

It is concluded that both effects, the initial cell coverage on the electrode and cell– to–cell junctions, contribute to the apparently smaller decrease in the normalized real impedance observed in multiple–cell–level measurements. These highlight the importance of single–cell–level measurements for drug and toxicity studies. Although

the suppression of cell–to–cell junction linkages and higher cell confluency may improve

the multiple–cell–level measurement in this study, the single–cell–level measurement

provides much more easier and robust control over the cell manipulation such as initial

cell coverage and adhesion in analyzing the full toxicity effect by CTX on the cell.

227 8.4.2. Time Fractional Derivative

The mathematical model of current travelling through the cleft region (equation 8–

6a and 8–6b) includes fractional derivatives of V t and V t , which can be solved by e ( ) J ( ) applying the following definition.

The fractional derivative of a function f t is a convolution between f t and ( ) ( )

t β−1 Γ β where Γ β is the gamma function and β is the order of the fractional derivation ( ) ( )

(see equation 8–23). Generally, –1 < β < 1.

∂β f (t) 1 = ⎡ t β−1 ∗ f t ⎤ β ⎢( ) ( )⎥ ∂t Γ(β) ⎣ ⎦ (8–23) 1 t β−1 = (t −τ) f (τ)dτ Γ β ∫0 ( )

From the spectral point of view, the frequency spectrum of the fractional derivative of f t can be described as: ( )

⎡ β ⎤ ⎢∂ f (t)⎥ β F ⎢ ⎥ =( jω) F (ω) (8–24) ⎢ ∂t β ⎥ ⎣ ⎦

where F ω is Fourier transform of f t . ( ) ( )

228 8.4.3. List of Symbols

2 Acell [µm ] Surface Area of the microelectrode occupied by the cell/cells

2 Ae [µm ] Surface Area of the microelectrode

2 Afree [µm ] Surface Area of the microelectrode unoccupied by the cell/cells

β Phase coefficient of CPE.

2 cCPE [µF/cm ] Specific CPE of surface–modified microelectrode

2 cm [µF/cm ] Specific cell membrane capacity

th CCPE,n [µF] CPE of n weighted segment of surface–modified microelectrode

th Cm,n [µF] Capacitance of n weighted segment of bottom (ventral) cell membrane

Cm,t [µF] Capacitance of top (dorsal) and lateral cell membranes

d [nm] Averaged cell–to–substrate separation distance

Δre [µm] Electrode radius increment

ΔZ [Ω] Impedance difference

f [Hz] Applied or excitation frequency in Hertz

Iprobe [A] Total current (paracellular + transcellular) injected

n Number of domains or segments

ω [Radians] Applied or excitation frequency in radians

rcell [µm] Radius of cell

rcell,o [µm] Initial radius of the cell

rcell,avg [µm] Averaged radius of individual cells in the monolayer

rcell,eff [µm] Effective radius of combined area of individual cells

re [µm] Radius of microelectrode

Rcyto [Ω] Resistance of cytoplasm

229 th Re,n [Ω] Resistance of n weighted segment of surface–modified microelectrode

RLimit [Ω] Current limiting resistor

th Rm,n [Ω] Resistance of n weighted segment of bottom (ventral) cell membrane

Rm,t [Ω] Resistance of top (dorsal) and lateral cell membranes

th Rseal,n [Ω] Sealed/Shunt resistance (n weighted segment)

Rsoln [Ω] Spreading/solution resistance

ρcyto [Ω–cm] Resistivity of cytoplasm

2 ρe [Ω–cm ] Charge transfer resistivity of surface–modified microelectrode

2 ρm [Ω–cm ] Cell membrane resistivity

ρs [Ω–cm] Resistivity of tissue culture media

ρseal [Ω–cm] Resistivity of tissue culture media in the cleft region

σ Variance of experimental data

Ve [V] Potential at the electrode

Vin [V] Potential applied from the functional generator

VJ [V] Potential at the cell–electrode heterojunction

Vm [V] Potential at the intracellular side of the bottom (ventral) cell membrane

Vm,t [V] Averaged potential at the extracellular side of the top (dorsal) and lateral part of the cell membrane

Vmea,I [V] Potential across the working electrode and counter electrode (experimentally measured)

Vout–RMS [V] Potential across the working electrode and counter electrode (calculated)

v Normalized RMS voltage magnitude

230

2 zcell [Ω–cm ] Specific impedance of the cell/cells

2 ze [Ω–cm ] Specific surface impedance of the surface–modified microelectrode

2 zm [Ω–cm ] Specific impedance of the cell membrane

Zc [Ω] Impedance of large–dimension counter–electrode

Zcell [Ω] Impedance of the cell/cells

Ze,cell [Ω] Surface impedance of electrode occupied by the cell/cells

Ze,free [Ω] Surface impedance of the electrode unoccupied by the cell/cells

th Ze,n [Ω] Surface impedance of n weighted segment of surface–modified Microelectrode

Zhet [Ω] Overall cell–electrode heterostructure impedance

Zhet,0 [Ω] Initial overall cell–electrode heterostructure impedance

Zhet,ss [Ω] Steady–state overall cell–electrode heterostructure impedance

Zin [Ω] Input impedance

th Zm,n [Ω] Impedance of n weighted segment of bottom (ventral) cell membrane

Zm,t [Ω] Impedance of top (dorsal) and lateral cell membranes

8.4.4. List of Some Abbreviations

CPE Constant Phase Element

CTX Chlorotoxin

DI De–Ionized

DIC Differential Interference Contrast

ECM Equivalent Circuit Model

231 ECIS Electric Cell–substrate Impedance Sensing

EIS Electrochemical Impedance Spectroscopy

FBS Fetal Bovine Serum

IMA Integrated Microelectrode Array

KCL Kirchhoff’s Current Law

KRGD lysine–arginine–glycine–aspartic acid

KVL Kirchhoff’s Voltage Law

NHS N–Hydroxysuccinimide

PBS Phosphate Buffered Saline

PEG Polyethylene Glycol

PSD Phase Sensitive Detection

PSN Penicillin–Streptomycin–Neomycin

PTFE Polytetrafluoroethylene (Teflon)

RMS Root Mean Square

SAM Self–Assembled Monolayer

SNR Signal–to–Noise Ratio

11–MUA 11–mercaptoundecanoic acid

232 References

Alberts M. B., Bray, D., Hopkin, K., Johnson, A., Lewis, J., Raff, M., Roberts, K., and Walter, P., 2004. Essential Cell Biology. 2nd Edition, Taylor & Francis Group, New York.

Asphahani, F., and Zhang, M., 2007. Cellular Impedance Biosensors for Drug Screening and Toxin Detection. Analyst. 132, 835–841.

Asphahani, F., Thein, M., Veiseh, O., Edmondson, D., Kosai, R., Veiseh, M., Xu, J., and Zhang, M., 2008. Influence of Cell Adhesion and Spreading on Impedance Characteristics of Cell– Based Sensors. Biosensors and Bioelectronics. 23 (8), 1307–1313.

Bisquert, J., Garcia–Belmonte, G., Fabregat–Santiago, F., Ferriols, N. S., Bogdanoff, P., and Pereira, E. C., 2000. Doubling Exponent Models for the Analysis of Porous Film Electrodes

by Impedance. Relaxation of TiO2 Nanoporous in Aqueous Solution. Journal of Physical Chemistry B. 104, 2287–2298.

Biswas, K., Sen, S., and Dutta, P. K., 2006. Realization of a Constant Phase Element and Its Performance Study in a Differentiator Circuit. IEEE Transactions on Circuits and Systems II: Express Briefs. 53 (9), 802–806.

Bodmer, J. E., English, A., Brady, M., Blackwell, K., Haxhinasto, K., Fotedar, S., Borgman, K., Bai, E., and Moy, A. B., 2005. Modeling Error and Stability of Endothelial Cytoskeletal Membrane Parameters Based on Modeling Transendothelial Impedance as Resistor and Capacitor in Series. American Journal of Physiology – Cell Physiology. 289, C735–C747.

Borkholder, D. A., 1998. Cell Based Biosensor Using Microelectrode. Ph.D. Dissertation, Stanford University. Stanford, CA.

Braun, D., and Fromherz, P., 1998. Fluorescence Interferometry of Neuronal Cell Adhesion on Microstructured Silicon. Physical Review Letters. 81 (23), 5241–5244.

Braun, D., and P. Fromherz, P., 2004. Imaging Neuronal Seal Resistance on Silicon Chip Using Fluorescent Voltage–Sensitive Dye. Biophysical Journal. 87, 1351–1359.

Franks, W., Schenker, I., Schmutz, P., and Hierlemann, A., 2005. Impedance Characterization and Modeling of Electrodes for Biomedical Applications. IEEE Transactions on Biomedical Engineering. 52(7), 1295–1302.

Fromherz, P., 2003. Semiconductor Chips with Ion Chanels, Nerve Cells and Brain. Physica E, Low-Dimensional Systems & Nanostructures. 16, 24–34.

Garcia, A. J., Ducheyne, P., and Boettiger, D., 1997. Cell Adhesion Strength Increases Linearly with Adsorbed Fibronectin Surface Density. Tissue Engineering. 3 (2), 197–206.

233 Gordon, L. G. M., Kottra, G., and Frömter, E., 1989. Electrical Impedance Analysis of Leaky Epithelia: Theory, Techniques, and Leak Artifact Problems. Methods in Enzymology. 171, 642–663.

Guyomar, D., Ducharne, B., and Sebald, G., 2008. Time Fractional Derivatives for Voltage Creep in Ferroelectric Materials: Theory and Experiment. Journal of Physics D: Applied Physics. 41, 125410 (6pp).

Hildebrand, F. B., 1949. Advanced Calculus for Engineers. Prentice–Hall, Englewood Cliffs, New Jersey. 152–181.

Huang, X., Nguyen, D., Greve, D. W., and Domach, M. M., 2004. Simulation of Microelectrode Impedance Changes Due to Cell Growth. IEEE Sensors Journal. 4 (5), 576–583.

Itagaki, M., Taya, A., Watanabe, K., and Noda, K., 2002. Deviations of Capacitive and Inductive Loops in the Electrochemical Impedance of a Dissolving Iron Electrode. Analytical Sciences. 18, 641–644.

Iwanaga, Y., Braun, D., and Fromherz, P., 2001. No Correlation of Focal Contacts and Close Adhesion by Comparing GFP–vinculin and Fluorescence Interference of Dil. European Biophysical Journal. 30, 17–26.

Lasseter, T. L., Cai, W., and Hamers, R. J., 2004. Frequency–Dependent Electrical Detection of Protein Binding Events. Analyst. 129, 3–8.

Lo, C., Keese, C. R., and Giaever, I., 1995. Impedance Analysis of MDCK Cells Measured by Electric Cell–Substrate Impedance Sensing. Biophysical Journal. 69, 2800–2807.

Lo, C., and Ferrier, J., 1998. Impedance Analysis of Fibroblastic Cell Layers Measured by Electric Cell–Substrate Impedance Sensing. Physical Review E. 57 (6), 6982–6987.

Mossop, B. J., Barr, R. C., Zaharoff, D. A., and Yuan, F., 2004. Electric Fields Within Cells as a Function of Membrane Resistivity – A Model Study. IEEE Transaction on Nanobioscience. 3 (3), 225–231.

Nam, Y., Chang, J. C., Wheeler, B. C., and Brewer, G. J., 2004. Gold–Coated Microelectrode Array with Thiol Linked Self–Assembled Monolayers for Engineering Neuronal Cultures. IEEE Transactions on Biomedical Engineering. 51 (1), 158–165.

Pancrazio, J. J., Whelean, J. P., Brokholder, D. A., Ma, W., and Stenger, D. A., 1999. Development and Application of Cell–Based Biosensors. Annals of Biomedical Engineering. 27, 697–711.

Panke, O., Balkenhohl, T., Kafka, J., Schafer, D., and Lisdat, F., 2008. Impedance Spectroscopy and Biosensing. Advance Biochemical Engineering/Biotechnology. 109, 195–237.

234

PerkinElmerTM instruments, 2000. What is a Lock–in Amplifier? Technical Note TN 1000. PerkinElmerTM Inc. URL: http://www.signalrecovery.com/ApplicationsNotes.htm (accessed August 30, 2010).

Pilwat, G., and Zimmermann, U., 1985. Determination of Intracellular Conductivity from Electrical Breakdown Measurements. Biochimica et Biophysica Acta. 820, 305–314.

Thein, M., Asphahani, F., Cheng, A., Buckmaster, R., Zhang, M., and Xu, J., 2010. Response Characteristics of Single–Cell Impedance Sensors Employed with Surface–Modified Microelectrodes. Biosensors and Bioelectronics. 25, 1963–1969.

Tlili, C., Reybier, K., Géloën, A., Ponsonnet, L., Martelet, C., Ouada, H. B., Lagarde, M., Jaffrezic- Renault, N., 2003. Fibroblast Cells: A Sensing Bioelement for Glucose Detection by Impedance Spectroscopy. Analytical Chemistry. 75, 3340–3344.

Veiseh, M., Wickes, B. T., Castner, D. G., M. Zhang, M., 2004. Guided Cell Patterning on Gold – Silicon Dioxide Substrates by Surface Molecular Engineering. Biomaterials. 25 (16), 3315– 3324.

Veiseh, M., Veiseh, O., Martin, M. C., Asphahani, F., and Zhang, M., 2007. Short Peptides Enhance Single Cell Adhesion and Viability on Microarrays. Langmuir. 23 (8), 4472–4479.

Weis, R., Muller, B., and Fromherz, P., 1996. Neuron Adhesion on a Silicon Chip Probed by an Array of Field–Effect Transistors. Physical Review Letters. 76 (2), 327–330.

Weis, R., and Fromherz, P., 1997. Frequency Dependent Signal Transfer in Neuron Transistors. Physical Review E. 55 (1), 877–889.

Xiao, C., Lachance, B., Sunahara, G., and Luong, J. H. T., 2002. An In–Depth Analysis of Electric Cell–Substrate Impedance Sensing to Study the Attachment and Spreading of Mammalian Cells. Analytical Chemistry. 74, 1333–1339.

Zoltowski, P., 1998. On the Electrical Capacitance of Interfaces Exhibiting Constant Phase Element Behavior. Journal of Electroanalytical Chemistry. 443, 149–154.

235 Chapter 9 SITE–SPECIFIC CELL MEMBRANE PERMEABILIZATION (SONOPORATION)

9.1. Site–Specific Sonoporation at Cellular Level Using UMTA Chip

Ultrasound–induced cell membrane permeabilization (or sonoporation) has been used to transfect gene products in vitro (see Chapter 6). A number of researches have

led to the general consensus that sonoporation results from the cavitation activities

induced by the ultrasound. Furthermore, the efficiency of sonoporation can be enhanced by employing ultrasound contrast agent (UCA) or microbubbles (MCB), which serves as artificial cavitation nuclei (see Chapter 6). Studies show that the effects of therapeutic ultrasound on the living tissues is usually non–invasive and relatively mechanical [Pitt et al., 2004]. Mechanical stresses cause cell membrane stretching, which increases plasma membrane inter–molecular distances [Valahakis and Hubmayr, 2000]. The cell usually responds by intracellular lipid trafficking to the plasma membrane in order to prevent membrane rupture (i.e., cytoprotective mechanisms). When the mechanical stresses are large enough, the cytoprotective mechanisms fail to prevent the rupture and the plasma membrane breaks down. The plasma membrane breaks are usually non lethal and that they are spontaneously resealed by site–directed exocytosis (ATP dependent).

It has been reported that ultrasound–induced perturbations and cavitations can facilitate or enhance drug delivery processes through several mechanisms:

• The first mechanism is from the oscillatory motion of the fluid driven by the

ultrasound. The oscillating motion of the fluid increases the effective diffusivity

of drug molecules, whether free or bound to carriers [Starritt et al., 1989]. This

236 enhanced transport can occur within blood, cells, and extracellular fluids

[Nyborg, 1982; Rooney, 1970].

• Perturbation of the drug carrier is another impact of ultrasound toward drug

delivery. The shear force in the ultrasound field can help to shear the drug from

the carrier polymer backbone [Guzman et al., 2003; Sundaram et al., 2003], and

then release the drug molecules. For drugs carried inside the liposome vesicle,

when the shear stress generated in ultrasound field exceeds the strength of

vesicle membrane, it will rupture the membrane and release the interior content

[Sundaram et al., 2003].

• The third contribution of the ultrasound to drug delivery relates to the stresses

that pressure wave applied on cells and tissues that result in cell permeabilization

and capillary ruptures. Cells in an environment with cavitation events are subject

to shear forces due to microstreaming, sonic jets, and shock waves (also see

Chapter 6). It is possible that a small jet of liquid would shoot into the cells

directly at sonic speed if a large semi rigid cell were adjacent to a small cavitating

bubble, which undergoes an asymmetric bubble collapse. Such activities can

probably rupture the cell membrane. Likewise, a collapse of a microbubble near

a capillary or blood vessel wall will cause the liquid jet to shoot right into the cell

wall. Such a collapse may be the source of the large amount of extravasation

caused in tissue exposed to ultrasound in the presence of microbubble contrast

agents [Song et al., 2002].

Therefore, by harnessing these mechanisms, the transport of drugs or gene products into

the cells can be enhanced or facilitated. Figure 9–1 illustrates various modes by which

drug delivery can be enhanced using ultrasound.

237

Figure 9–1. Schematic representation of various modes by which drug delivery can be enhanced by ultrasound. A: therapeutic agent (triangles); B: gas bubble undergoing stable cavitation; C: microstreaming around the cavitating bubble; D: collapse cavitation emitting a shock wave; E: asymmetrical bubble collapse producing a liquid jet that pierces the endothelial lining; F: completely pierced and ruptured cell; G: non– ruptured cells with increased membrane permeability due to sonication; H: cell with damaged membrane from microstreaming or shock wave; I: extravascular tissue; J: thin–walled microbubble decorated with agents on the surface; K: thick–walled microbubble with agents in lipophilic phase; L: micelle with agents in lipophilic phase; M: liposome with agents in aqueous interior; N: vesicle decorated with targeting moieties attached to a specific target. After [Pitt et al., 2004].

In all these transfection and drug delivery applications, ultrasound is usually generated by devices or instruments based on piezoelectric–based transducers.

However, these devices currently lack the functionality of targeting specific locations with high degree of spatial resolution. Although, varieties of device designs have showed that ultrasound can be focused to a specific region in the cell suspension or in the tissue

[Rahim et al., 2006 (a); Rahim et al., 2006 (b)], the spatial resolution is still not high enough to target a small cluster of cells or individual cells (in those designs, cell samples are placed at the focal point of ultrasound beam generated by complex focused

238 ultrasound transducers). In order to achieve high degree of spatial resolution, a potential sonoporation device must possess three major functionalities:

(1) In most current applications, the sonoporation device is located relatively far

from the site for ultrasound applications, therefore, higher ultrasound pressure

has to be generated in order to achieve, at least, the threshold pressure at the

desired location in the cell suspension or tissue. Therefore, this compromises

cells or tissues in the surroundings.

(2) The dimensions of the transducers are still relatively large compare to the

dimension of individual cells so that focusing ability at the cellular level is very

poor. Therefore, transducers must be miniaturized.

(3) Current sonoporators are still relatively large for using as medical implants.

Therefore, the devices must also be miniaturized. In addition, the bio–

compatibility of the devices is also important.

Therefore, in this study, Ultrasonic Micro–Transducer Array (UMTA) biochip was

microfabricated (The detail of the device fabrication and characterization are described

in Chapter 7). The UMTA chip employs miniature transducers (cell–size), which are

designed to be located very close to the seeded cells. Due to the proximity of these

micro–transducers to the cells, the intensity of ultrasound, needed to be generated by

the transducers in order to permeabilize cell membrane, can be reduced. Furthermore,

due to its miniature size, the micro–transducer demonstrated renders huge potentials

for the future development of cellular/cell–based assay platforms and ultrasound–based

implants. The feasibility of the UMTA biochip prototype was demonstrated by applying ultrasound effectively at the specific locations within the cell monolayer of melanoma

cancer cells (LU1205) in order to induce sonoporation at the cellular level. Successful

sonoporation at the cellular level was also confirmed by the facilitated uptake of

nanoparticle quantum dots (QDs).

239 9.2. Materials and Methods

9.2.1. Cell Culturing and Preparation

In order to demonstrate the feasibility of the UMTA biochips, site–specific cell

membrane permeabilization (sonoporation) was carried out on the monolayer of Green

Fluorescent Protein (GFP)–expressing human melanoma (LU1205) cells. The cells will

be seeded and grown on the surface of the UMTA biochip until the monolayer is

achieved. LU1205 cell line was chosen for three reasons: (i) it is an aggressive skin

cancer cells with no current viable therapy and the ultimate goal is to develop UMTA

biochips for ultrasound–enhanced transfection/gene therapy assays of cancer cells at

single–cell/cellular level in vitro; (ii) melanoma is also one of the fastest growing cancers

in the developing world with the incidence having tripled in the last three decades.

Chemotherapy, immunotherapy, and vaccines have produced very limited benefits

especially since the response are typically short–lived, with no significant effect on

overall survivals [Ghosh et al., 2005]; (iii) its very fast growth rate, good morphology,

well–understood proliferation, and robustness to less friendly cell culture environment

[Dothager et al., 2005] make LU1205 melanoma cancer cell line easier to handle so that

the feasibility and working mechanisms of the UMTA biochip can be emphasized. In

addition, GFP–expressing cell line is chosen for fluorescent imaging and mapping.

9.2.1.1. Cell Thawing

A vial of GFP–expressing LU1205 melanoma cancer cell line was taken out of from

the freezer at –80 ˚C, then thawed quickly in a 37 ˚C water bath for 2 minutes until the

medium become liquid. Then cells were transferred into a 100 ml Petri dish with cell

media. The media contains 90% Dulbecco’s Modified Eagle Medium (DMEM), 20%

Fetal Bovine Serum (FBS), and 100 U/ml of penicillin–streptomycin (PS). The whole

transfer process takes place in the sterile hood with air curtain. The total solution

240 volume is 10 ml (9 ml medium + 1 ml cells from frozen vial). The Petri dish was then placed in the incubator with constant temperature of 37 ˚C and 5% CO2 environment for

6 hours for cells to recover and settle on the surface.

9.2.1.2. Cell Passing and Culturing

After 6 hours of staying in the incubator for cell thawing, the GFP–expressing

LU1205 are ready for passing and culturing on the surface of micro–sized transducer

chip. 30–ml cell culture medium (90% DMEM + 10% FBS + 100 U/ml PS) was warmed

up to 37 ˚C. After washing cells two times with 5 ml of Dulbecco’s Phosphate Buffered

Saline (DPBS) (without Ca and Mg), then 1 ml trypsin was placed into the Petri dish for

approximately 1 minute to detach the cell from the bottom of the Petri dish. After cells

begin to round–up and detach, cell culture media was put into the Petri dish. Before

passing the cell onto the surface of the UMTA biochip using a pipette, medium was

gently agitated to fully detach and break the clumps of the cells. Then, the cells were

transferred onto the surface of the UMTA biochip that is placed at the bottom of 6–in

glass Petri dish. Then, 100 ml of tissue culture media (90% DMEM + 10% FBS + 100

U/ml PS) was added to obtain the total tissue culture media volume of 110 ml in the glass

Petri dish with the UMTA biochip placed at the bottom (Note: the 6–in glass Petri dish, the UMTA biochip, and peripheral accessories are thoroughly cleaned with ethanol solution and sterilized under the UV light prior to the cell passing). Then, the glass Petri dish was put back into the incubator and GFP–expressing LU1205 melanoma cells were allowed to seed and grow for approximately 36 hours on the device surface until the cells were approximately 70% confluent. Figure 9–2 shows GFP–expressing LU1205

melanoma cancer cells in the Petri dish after 48 hours of culturing (the cells reached

about 90–100% confluent in this case). Figure 9–3 shows the optical microscope image

of melanoma (LU1205) cancer cells grown on the surface of UMTA biochip after 24

hours of incubation.

241

Figure 9–2. GFP–expressing Melanoma cancer cells (LU1205) after ~48 hours of incubation in the cell culture medium (90% DMEM + 10% FBS + 100 U/ml PS) at 37 ˚C

and 5% CO2 flow (~90–100 % confluent). The scale bar is 10 µm. The image was taken with inverted fluorescent microscope.

Figure 9–3. Melanoma LU1205 cancer cells grown on the surface of the UMTA biochip. (A) Low magnification optical microscope image of the device surface with gold electrodes and interconnections lines. (B) High magnification showing melanoma cells growing on top of a 25–µm square electrode with Parylene C film between them. What underneath the electrode is the micro–sized transducer, which will be driven to generate ultrasonic waves. The images are taken under an upright microscope. [Courtesy of An Cheng]

242 9.2.2. Quantum Dots Tracking

Successful sonoporation will create transient holes/disruptions in the cell membrane and facilitate/enhance the transport of some molecules into the cytoplasm by means of passive diffusion. Therefore, by tracking and monitoring the concentrations of these molecules inside the cells, the level of cell membrane permeabilization and the degree of delivery enhancement can be determined. In this study, Quantum Dots (QDs) were used as an optical probe to quantify the level of cell membrane permeabilization after LU1205 cells were exposed to the ultrasound generated by micro–transducers. There are six reasons for choosing QDs:

(1) QDs are biocompatible (non–toxic) and have been used in bio–labeling and bio–

imaging in recent years. They have been tested as “smart” targets for diagnostic

or therapeutic purpose in the areas such as cancer detection and drug delivery

[Rosenthal et al., 2002]. Furthermore, they are stable inside the cell without

causing noticeable change in cell characteristics.

(2) The size of QDs (5–20 nm) is comparable to those of most drug molecules and

gene products such as plasmid vectors [Muller–Borer et al., 2007; Papazoglou et

al., 2004].

(3) QDs have unique properties such as enhanced brightness, narrow emission

spectra, white–light excitable (i.e., multiplexing capability), long–term

photostability, and strong resistance to photobleaching in cell culture [Moioli et

al., 2006]. These properties make QDs far superior than traditional fluoropores.

(4) QDs are resistant to chemical and metabolic degradation especially those with

core–shell structures.

(5) Cells ingest water–soluble quantum dots by endocytocis (this makes QDs perfect

candidates for comparing sonoporation–induced nanoparticle delivery to

endocytosis–driven nanoparticle uptake).

243 (6) QDs are alternative to drug molecules or gene products in demonstrating the

feasibility of the UMTA biochip since using real drug molecules would bring a lot

of challenges from drug labeling, drug tracking, and maintaining cell viability

(the main emphasis of this study is to demonstrate the mechanism and

feasibility of UMTA biochips).

Studies suggest that carboxylic–acid–surface–derivatized CdSe/ZnS core/shell QDs are water–soluble and cytocompatible [Medintz et al., 2005; Michalet et al., 2005; Zhang and Monteiro–Riviere, 2009]. Core–shell QD are more biocompatible since the core is being protected inside the shell from oxidation in tissue culture media. Furthermore, synthesis and structure of CdSe/ZnS core–shell QDs has been thoroughly studied.

Therefore, carboxylic–acid–surface–derivatized CdSe/ZnS core/shell QDs (emission wavelength at 615nm, hydrodynamic diameter deff ≈ 20nm (core/shell + polymers), quantum yield: >50%, emission spectrum bandwidth (emission FWHM): < 25nm,

Emission tolerance: ±10nm, and pH stability: 4–10) are used in this study.

9.2.2.1. Characterization of Quantum Dots

The CdSe/ZnS QDs were coated/functionalized with carboxyl–terminated polymer

(or surface derivatized with carboxylic acid) to make the QDs’ surface hydrophilic and more soluble in water–based tissue culture media. Since QDs’ narrow emission spectrum and efficient energy conversion are critical to the accuracy of the experiment output, both absorption and emission spectrum of CdSe/ZnS QDs was characterized before being placed in the cell media. Figure 9–4A shows the photoluminescence (PL) emission at ~615 nm wave length with an emission FWHM (Full Width at Half

Maximum) of only 20 nm, which indicates a narrow size distribution inferring that the size of QDs suspended in the solution are uniform. The QDs transported into the cells will emit sharp red–orange light due to this single narrow emission spectrum and, thus,

244 the fluorescence intensity can be used to precisely estimate the quantity of QD uptake by

LU1205 cells. Figure 9–4B shows the QDs first absorption peak (absorption onset) at

608 nm. The small stokes shift (difference between the PL and the first absorption peak) shows a dominant band edge emission indicating the high–quality semiconductor structure of the CdSe/ZnS QDs.

Figure 9–4. Emission (A) and absorption (B) spectrum of carboxylic–acid–surface– derivatized CdSe/ZnS core–shell quantum dots. [Courtesy of Ocean NanoTech, LLC].

9.2.2.2. Preparation of QD–Suspended Tissue Culture Media

In order to prepare QD–suspended tissue culture media, 0.05 ml of carboxylic–acid– surface–derivatized CdSe/ZnS core–shell QDs in DI–water solution (concentration of 8

µM) were diluted in the tissue culture media (90% DMEM + 10% FBS only. No

penicillin–streptomycin (PS)) to obtain 100–nM QD–suspended tissue culture

medium. The solution is then filtered by an Acrodisc® syringe filter with 0.2–µm HT

Tuffryn® membrane (polysulfone). There are two purposes for filtering: (i) to sterilize

the QD–suspended medium and filter out most bacteria; and (ii) to prevent the aggregation of QDs particles before they are homogeneously suspended.

245 9.2.3. Results from Control Experiments (QD–Uptake by LU1205 Cells via

Endocytosis Process)

After adequate ultrasonic pressures are generated from micro–transducers, which

induce successful sonoporation, QDs will be transported or loaded into the LU1205 cell

and it is expected that the rate of ultrasound–facilitated QD–uptake should be much

faster compared to that of endocytosis–driven QD–uptake. Therefore, in order to

compare and contrast between the endocytosis–driven QD–uptake by LU1205 cells to

ultrasound–facilitated QD–uptake, three control experiments were also carried out: (1),

time–lapsed fluorescent imaging to qualitatively determine the rate of endocytosis–

driven QD–uptake by LU1205 cells; (2), confocal Z–stack imaging to confirm whether

QDs are ingested or transported into the LU1205 cells; and (3), flow cytometry

measurement to quantitatively determine the rate of endocytosis–driven QD–uptake by

LU1205 cells.

9.2.3.1. Time–Lapsed Fluorescent Imaging

First, LU1205 cells were cultured in three Petri dishes under normal tissue culture

conditions for 48 hours until 80 % of confluency was achieved. Then, the culture media

was replaced with fresh media that contains 100 nM of carboxylic–acid–surface–

derivatized 615–nm CdSe/ZnS core/shell QDs. Next, the three Petri dishes were put

back into the incubator for 2 hrs, 6 hrs, and 12 hrs, respectively. The Petri dishes were

taken out at the end of their respective incubation periods and the cells were washed

three times with DMEM (washing with DMEM removes most of the background QDs

that were still suspended in the cell culture medium as well as those QDs that were

possibility sticking or adsorbing on the surface of the cell membrane). Fluorescent

microscopic inspections were carried out on all three samples. Figure 9–5 shows the

time–lapsed fluorescent images that show the levels of endocytosis–driven QD–uptake

246 after different incubation periods. The images show that more QDs were ingested at longer incubation period (or longer QD–inoculation period) and significant QD–uptake is observed approximately at 2 hours after QD introduction. Figure 9–5A shows the fluorescence image without washing the media (The whole background is red, and no cell can be identified. Hence, washing and replacing with fresh media is a necessary step before fluorescence imaging).

Figure 9–5. The levels of QD uptake by LU1205 cells at different QD–inoculation periods. (A) A fluorescent image before QD–suspended tissue culture media was aspirated. QD uptake by LU1205 cells after 2 hours (B), 6 hours (C), and 12 hours (D) of incubation in the medium that contained 100 nM of carboxylic–acid–surface– derivatized 615–nm QDs. The images shown are generated by overlapping two types of images (bright field and fluorescence) taken at the same location. It clearly shows that QDs uptake at 12 hrs (D) > 6 hrs (C) > 2 hrs (B). [Courtesy of An Cheng]

247 9.2.3.2. Confocal Z–Stack Imaging

In order to find out whether QDs were just sticking or adsorbing on the melanoma cancer cell membrane due to hydrophilic property of phospholipids’ head groups, 3D Z– stack Confocal microscope images of the LU1205 cells, which was incubated in QD– suspended medium for ~24 hours, was taken under “Olympus FluoView 1000” triple–

laser Confocal microscope (figure 9–6) (again, the cells were washed three times before

Confocal microscope observation to remove any QD adsorbing on the surface of the cell

membrane). Images show that no red–orange fluorescence emission was noticed in the

background. Besides top–slice (figure 9–6a) and bottom–slice (figure 9–6b) of the cell, all other Z–levels have clear and sharp QD–emissions from different location inside the cells, confirming that the QDs were loaded into the cell after ~24 hours of incubation.

Furthermore, the majority of QD–emissions was located inside the LU1205 cell rather than at the edges (figure 9–6b to 9–6e). This also confirms the fact that after washing three times with tissue culture medium, most QDs sticking or adsorbing on the cell membrane were removed and possible remaining QDs can be negligible. Figure 9–6g shows the 3D reconstruction of Z–stack images.

248

Figure 9–6. Confocal microscopic images showing QD distribution inside LU1205 cells. Z–stack images from A to F: slices at (A), the bottom of LU1205 cells; (F), the top of LU1205 cells (A 2–um increment in Z–axis between each image). (G) Reconstructed 3D image of QD distribution inside LU1205 cells. The images were taken after ~24 hours of incubation in the culture medium that contained 100 nM of carboxylic–acid– surface–derivatized 615–nm QDs. [Courtesy of An Cheng]

249 9.2.3.3. Flow Cytometry

Flow cytometry was used to further quantify the QD–uptake of LU1205 cell as a function of QD–inoculation time. Six cell samples were prepared and, for all the

samples, the cell concentration was made to maintain at ~0.5 x 106 cells/ml. Then, QD–

suspended tissue culture medium was evenly introduced into the samples except for one

control sample. After 1 hr, 3 hrs, 6 hrs, 12 hrs and 24 hrs, respectively, samples were

washed, detached, agitated and suspended in the fresh tissue culture media (without

QDs), and then put into the flow cytometer “Guava PCA–96 System” for fluorescence

intensity analysis (five measurements at five different QD–inoculation time).

The PCA–96 system shine a 532–nm green laser into the cell suspending media and

use two detectors to detect excited emission from QDs transported into the cells (note: the green laser has shorter wavelength than the first absorption peak of the CdSe/ZnS core–shell QDs). The higher is the total concentration of QDs inside the cells, the higher is the fluorescent intensity reading from PCA–96 system. Figure 9–7 shows the histogram based on the flow–cytometer readings of fluorescence intensities from 5 samples (1 hr, 3 hrs, 6 hrs, 12 hrs, and 24 hrs. The % increase of QD intake represents the increase in the QDs’ fluorescent intensity (or the intracellular QD concentration), which is referenced from the baseline intensity taken from the control sample, due to endocytosis–driven QD uptake). The readings show that, for the same QD concentration of 100 nM in tissue culture media, the longer is the QD–inoculation time (or the longer is the incubation time with the QD–suspended tissue culture media), the higher is the level of QD–uptake by the LU1205 cells.

250

(A)

(B)

Figure 9–7. (A) Two–parameter histograms obtained by the flow cytometry showing endocytosis–driven QD–uptake by LU1205 cells for different QD–inoculation periods. The amount of QD–uptake increases with longer QD–incubation period. (B) Histogram of the percent increase of QD uptake by LU1205 cells at different incubation periods. The cells were incubated in the culture medium that contained 100 nM of carboxylic– acid–surface–derivatized 615–nm QDs. The histogram data is based on the average readings from the flow cytometry. [Courtesy of An Cheng]

251 9.2.4. Site–Specific Cell Membrane Permeabilization

In order to determine whether the UMTA biochip can permeabilize the cell membrane and facilitate/enhance the transport of molecules into the cells, GFP– expressing malenoma cancer cells (LU1205) were seeded and grown on the surface of the

UMTA biochip. Quantum Dots (QDs) are used as optical probes to determine whether successful sonoporation occurred after micro–transducers from the UMTA biochip generated ultrasounds. Furthermore, the fluorescent emission intensity of QDs transported into LU205 cells can also be used to estimate the amount of QD–uptake by the cells due to ultrasound–induced cell membrane permeabilization. Five major objectives in this experiment are:

(1) To determine whether micro–transducers can generate ultrasounds that can

permeabilize LU1205’s cell membrane (sonoporation).

(2) To determine the spatial specificity (i.e., the beam confinement or the lateral

resolution) of ultrasound generated from the micro–transducer array (site–

specific cell membrane permeabilization).

(3) To investigate and compare ultrasound–facilitated QD–uptake to endocytosis–

driven QD–uptake by LU1205 cells (sonoporation–induced delivery or

transmembrane transport enhancement).

(4) To determine the ultrasound–intensity–dependent QD–uptake if successful

sonoporation can be carried out.

(5) To determine the viability of LU1205 cells after ultrasound applications.

9.2.4.1. Instrumentations and Micro–Transducer Activation

Figure 9–8 shows the instrumentation and experimental setup for site–specific sonoporation at the cellular level. The UMTA biochip placed at the bottom of glass Petri dish (glued with Sylgard®) and submerged in tissue culture medium is connected to a RF

252 signal generator (Agilent E4422B ESG) through a power amplifier (ENI A500). Wires were soldiered to the contact pads of UMTA biochips with Silver epoxy and soldered

connections were sealed under Epoxy resin to insulate and prevent electrical shorting

through the tissue culture medium. Wires were then connected to the power amplifier

with Crimp–on BNC connectors. The RF signal generator produced sinusoidal RF wave

(continuous wave). This HF signal was amplified by the power amplifier and then

delivered to the individual micro–transducers of the UMTA biochip.

Site–specific cell membrane permeabilization (sonoporation) was carried out as follows. First, GFP–expressing human melanoma cells (LU1205) were cultured on the

Ultrasonic Micro–Transducer Array (UMTA) biochip until the cells achieved 50–70%

confluency (Section 9.2.1.2.). Second, the media was aspirated and fresh tissue culture

medium that contains 100 nM of carboxylic–acid–surface–derivatized 615–nm

CdSe/ZnS core/shell QDs was added. 30–MHz sinusoidal RF signals were applied to the high–aspect–ratio piezoelectric micro–transducer arrays at 4 different RF powers for approximately 3 minutes (80 mW, 2 mW, 1 mW, and 0.5 mW). The 30–MHz frequency was chosen to achieve a Near Field Length (NFL) approximately in the range of 13–15

µm (see Chapter 7, Section 7.2.5). Furthermore, the sonoporation time period and

RF powers chosen were much lower than those that cause ultrasound–induced

hyperthermia [Liu et al., 2001; Wang et al., 2010]. Four separate experiments were

conducted for four different RF powers (i.e., each RF power was applied to the micro– transducer arrays at each respective experiment. After fluorescent images were taken, the cells were lifted–off and the same experimental procedures followed except for the

RF power applied). The schematic of the site–specific sonoporation is illustrated in

figure 9–9.

253

Figure 9–8. Schematic of ultrasound–induced site–specific cell membrane permeabilization (sonoporation). A RF signal is applied across the PMN–PT pillar in order to generate ultrasonic waves directed towards the seeded cell(s) above it.

9.2.4.1.1. Peak Ultrasound Pressure on the Cell Membrane

As high aspect–ratio PMN–PT micro–transducers in the UMTA biochip are embedded in Epoxy resin and having the piezoelectric coefficients d33 >> d31, the flexural

vibration modes are dampened, leaving the longitudinal modes as the dominant

254 vibration pattern for the transducer operation. In the UMTA biochip, when the micro–

transducer is activated, the maximum peak ultrasound pressure occurs at the end of

Fresnel Zone (or near field). The Near–Field Length, NFL = w2 4λ = w2 f 4c , is d d estimated to be ~15µm, given the 30–MHz frequency, ~55µm averaged transducers diagonal width (wd), and the speed of sound (c) in water. Therefore, by the novel design of the UMTA biochip and tailoring the driving frequency, the peak ultrasonic pressure exerting on the cell membrane is approximately the same as the peak pressure at the end of Fresnel zone of micro–transducers in the UMTA biochip since the Au cell–trapping pads, on which the LU1205 cells are being seeded, are vertically separated from the underlying micro–transducers by a ~15µm–thick Parylene–C encapsulation layer for the maximum transducer efficiency (in other words, by tailoring the driving frequency, NFL is adjusted so that it is almost equal to the micro–transducer to cells distance, which is preset by the UMTA device architecture). The detailed calculations for ultrasound pressures exerting on the cell membrane are discussed in section 9.4.2.

9.2.4.2. Cell Membrane Permeability Observation and Analysis

After each ultrasound application for 3 minutes, the LU1205 cells were incubated in

QD–suspended media for ~6 minutes and then gently washed away with fresh tissue culture medium (contains no QDs) for 3 times. Then, the sonoporated cells were incubated for 30 minutes in normal tissue culture conditions for recovery. Next, fluorescence images on activated 3 × 3 array areas (both for Green Fluorescence Protein

(GFP) and CdSe/ZnS QDs) were taken under the water immersion objective lens (10X,

numerical aperture 0.3, working distance 3.3mm) (Olympus BX60) with white light

excitation (mercury lamp). Emission detections at 515 nm for GFP and at 615 nm for

QDs were achieved by setting appropriate filter combinations with a cooled CCD camera

(effective image pixel size 90 nm x 90 nm). Then, fluorescent intensities of CdSe/ZnS

255 QDs transported into the sonoporated cells were computed by processing with ImageJ software (normalized with the same background intensities). To estimate the intracellular concentration of QDs and to evaluate the ultrasound lateral resolution, the sum fluorescent intensity spatial profiles of QDs transported into the sonoporated cells were calculated by the following algorithm: the 3 × 3 micro–transducer array area were

further divided into a matrix of 450 rows and 450 columns, i.e., 450 × 450 pixel array

(each matrix element has a dimension of ~278nm × ~278nm). The background intensity

was subtracted (normalized) and the fluorescent intensity of QDs along each element

column was summed up and plotted in order to obtain the sum intensity spatial profile.

The profiles were fitted as Gaussian functions and numerically integrated to calculate,

respectively, the FWHM for lateral resolution of sonoporation (or beam confinement

analysis) and the total concentration of QDs transported into the sonoporated cells.

9.2.4.3. Cell Viability Assay

Trypan Blue Exclusion Assay, which stains non–viable cells that do not possess intact

cell membrane with trypan blue, was also performed 2 hours after ultrasound treatment

(on those cells treated with the highest power of ultrasound, i.e., 80 mW RF power). It is

found that approximately 80–90% of cells were viable after 2 hours of post 3–min

sonoporation suggesting that the LU1205 cells spontaneously repaired their plasma

membrane and targeted sonoporation did not cause much adverse effects on the cells.

9.3. Experimental Results from Site–Specific Sonoporation

Fluorescent microscopic images of GFP–expressing LU1205 cells on the surface of the UMTA biochip after ultrasound applications (three different RF powers: 0.5 mW, 1.0

mW, and 2.0 mW) are shown in figure 9–9. The Green Fluorescent Protein (GFP)

emission, which is used to map the confluency of cell monolayer on the surface of the

256 UMTA–biochip, shows that, after 2–day incubation, LU1205 cells grow homogeneously

(~70–80% averaged monolayer confluency) on the surface of the UMTA–biochip having

relatively more cells grown on the active areas of 25–µm transducer arrays due to the

present of Au pads (1A, 2A, and 3A of figure 9–9). The cells were healthy and viable

after the sonoporation since high level of GFP expressions (strong GFP emissions) were

observed from the ultrasound–treated cells. By comparing the GFP and QD emissions, it is evident that cellular–level site–specific sonoporation is achieved within the cell monolayer after UMTAs were activated (the micro–transducers, when they are activated, generate ultrasounds that are targeted towards the cells seeded above them). The red

QD emissions from active areas (1B, 2B, and 3B of figure 9–9) confirm that the ultrasonic pressures generated from the micro–transducers permeabilized the LU1205 cells located above them and enhanced the transport of QDs into the cells by passive diffusion within a short time period (< 6 minutes). Furthermore, highly–confined ultrasonic beams result in high lateral resolution for the site–specific sonoporation process, which is confirmed by the observation that no or negligible QD–uptake occurred for the cells located on the interspace and surrounding areas. The sum intensity spatial profiles of QD–emissions (1C, 2C, and 3C of figure 9–9) reveal that the degree of sonoporation and the amount of QD–uptake increase monotonically with the ultrasound

radiation pressure, which was correlated to the RF power in the present study. On the other hand, no or very little QD–emission was observed in the active areas where micro– transducers array was not activated (figure 9–10A).

257

Figure 9–9. Micro–transducer arrays driven at three different RF powers at 30 MHz for 3 minutes. Each micro–transducer in the array presents 25µm × 25µm active area. RF Powers applied: (1) 0.5 mW, (2) 1.0 mW, and (3) 2.0 mW. (A) GFP–expressing LU1205 cells seeded on the surface of the UMTA biochip. GFP–expressing cells are chosen in order to map the confluency of the cell monolayer grown on the surface of the biochip. (B) LU1205 cells located above the micro–transducers (i.e., on the active areas) demonstrate CdSe/ZnS QD–uptake by passive diffusion after 3–minute sonoporation. The cells located on the areas between individual micro–transducers and on the surrounding areas show little or no QD intake. (C) Sum fluorescence–intensity spatial profile of QD transported into LU1205 cells located on the active areas of micro–transducer array along the column 1, 2, and 3 of the 3 × 3 array structure (Note: In the image processing, the array area was further divided into 450 × 450 pixels. The background was subtracted and QD fluorescent intensity along the pixel elements of each column was summed up and plotted against the row pixels in order to obtain sum intensity spatial profile. See section 9.2.4.2.). The scale bar represents 50μm.

258

Figure 9–10. (A) No QD–uptake was observed in the active areas where micro– transducers are not activated (dashed lines show the location of micro–transducers). (B) QD–uptake by LU1205 cells after micro–transducer array (each micro–transducer in the array presents an area of 50μm × 50μm) was driven at 80mW, 30MHz for 3 minutes. The image shows significant amount of QD–uptake by LU1205 cells seeded above the micro–transducers’ active areas and interspaces. The higher ultrasound power and larger transducer cross–sectional area result in poor spatial targeting or lateral resolution due to prominent side lobes in ultrasonic beams. The scale bar represents 100μm. [Courtesy of An Cheng].

Nevertheless, the trend does not hold for very high RF power levels. The site– specific sonoporation result at an applied RF power of 80.0 mW with a frequency of 30

MHz for 3 minutes was shown in figure 9–10B. The image shows significant increase in

QD–uptake by LU1205 cells located on the micro–transducer array area. However, the effective lateral resolution of sonoporation becomes very poor when micro–transducer array is driven at high input RF power (80 mW), where pressure from the side lobes of the ultrasonic beam produced by the micro–transducer becomes sufficiently large to sonoporate cells located on the interspaces between micro–transducer active areas.

High RF power level results in poor spatial targeting ability or site–specificity of

Ultrasonic Micro–Transducer Arrays (UMTAs).

259 9.4. Modeling and Analysis

The purpose of modeling is to 1), determine the average size of ultrasound–inflicted membrane wound and the concentration of QDs transported into LU1205 cells as a function of radiation pressure applied by UMTAs; and 2), determine the threshold radiation pressure required for sonoporation of LU1205 cells, and 3) compute the

UTMA–based–sonoporation–induced QD–transport/–uptake enhancement factors.

Two models are developed, which includes the model for ultrasound–inflicted cell membrane wound and that for transmembrane transport of QDs through the transient wound after sonoporation.

Plane–progressive acoustic waves exert radiation forces on the incident object, which are always in the direction of wave propagation [Doinikov, 1994; Hasegawa et al., 1993;

Mitri, 2006; Nyborg, 2001]. Plasma membrane of living cells breaks temporarily when enough mechanical stresses (or forces) are imposed on the membrane [Valhakis and

Hubmayr, 2000]. It is hypothesized that when the adherent cell is strained under mechanical stress such as tension, its plasma membrane surface–to–cell volume ratio increases during the deformation. There are four hypothetical plasma membrane deformation responses, all of which have been observed experimentally (figure 9–11).

These responses are: (1), unfolding of excess plasma membrane; (2), elastic deformation

(i.e., stretching) of the plasma membrane; (3), translocation of lipids from intracellular stores to the plasma membrane; and (4), plasma membrane stress failure. Any or all of these responses might be active at any one time, and the relative contribution of each differs depending on the type and magnitude of deformation imposed and the cell system studied [Akinlaja and Sachs, 1998; Liu et al., 1999; McNeil and Steinhardt, 1997;

Tschumperlin et al., 2000].

260

Figure 9–11. Deformation responses of adherent living cells under mechanical stresses. Four responses: (A) cell surface unfolding; (B) increased plasma membrane intermolecular distances; (C) intracellular lipid trafficking to the plasma membrane in order to prevent cell membrane rupture (cytoprotective mechanisms); (D) plasma membrane stress failure (represents the short–term failure of cytoprotective mechanisms). There is evidence that plasma membrane breaks are nonlethal and that they are resealed by site–directed exocytosis. After [Valahakis and Hubmayr, 2000].

It is conceived that when an adherent cell on the surface of the UMTA biochip is subject to mechanical stress due to ultrasonic radiation pressures generated by the micro–transducer beneath the cell, the plasma membrane undergoes elastic deformation

(i.e., stretching). Above the threshold pressure, stress failure occurs and plasma membrane ruptures resulting in transient wounds in the cell membrane (figure 9–12)

[Schlicher et al., 2006].

261

Figure 9–12. (A) Cell membrane of adherent cell on the active area of UMTA biochip in normal steady state before being exerted by ultrasonic radiation pressure. (B) Cell membrane is stretched due to radiation pressure/force exerted by ultrasonic wave resulting in the increase in plasma membrane intermolecular distances. Lipid vesicles from intracellular origin travel to the cell membrane to prevent from membrane rupture. (C) Failure and rupture of the cell membrane above the threshold radiation pressure. Transient wound is created in the plasma membrane. Intracellular lipids travel to the wound for repair (i.e., membrane patching).

Experimental results (control experiment) show that the endocytosis–driven QD– uptake by LU1205 cells is a slow process (Note: It is also conceived that the smaller is the hydrodynamic diameter of the QDs being used, the slower is the endocytosis–driven QD uptake. This is because higher binding energy is required between the surface of QD and the cell membrane in engulfing QDs due to larger radius of curvature in smaller QD).

Time–lapsed fluorescent microscopic images (figure 9–5b) also confirm that significant

(or detectable) amount of QD uptake was observed approximately 2 hours after QDs

were introduced. Furthermore, according to the data from the flow cytometry (figure 9–

7), only 22 percents increase in QD–uptake (referenced from the baseline or control data

with no QD–inoculation) was observed after 24 hours of incubation in the QD–

suspended tissue culture medium. It is hypothesized that the endocytosis–driven QD–

uptake occurs via G–protein coupled receptor–mediated endocytic pathways (illustrated

in figure 9–13) [Zhang and Monteiro–Riviere, 2009].

262

Figure 9–13. QD nanoparticle endocytic pathway in human epidermal keratinocytes (HEKs). QD uptake mechanism and subcellular localization with 24 inhibitors with different inhibitory functions and protein markers for organelles. Lipid rafts, caveolin 1, clathrin, early endosome, late lysosomes are stained (teal). Inhibitors are located near the targets where they exert their functions. Inhibitory effects are labeled in green, whereas no inhibitory effects are labeled in black. Briefly, QDs were first recognized by lipid rafts (CTX as marker, which binds to ganglioside GM1) and get internalized into early endosome (EEA1 as the marker), then localized in late endosome (CD63) and remained in the lysosome (LAMP1). QD uptake was through the GPCR and the downstream proteins regulated by G–protein, PLC, and PKC. These inhibitors blocked QD uptake, indicating QD endocytosis may be recognized by specific receptors. The inhibitors for scavenger receptors (PolyI and FCD) greatly blocked QDs. LDL/AcLDL competed with QDs and reduced QD internalization suggesting that LDLR/SR–BI may be the most appropriate receptors of QD uptake. After [Zhang and Monteiro–Riviere, 2009].

On the other hand, after the UMTA biochip generated ultrasounds, QD–uptake by

the sonoporated LU1205 cells is very fast (observed after 3–minute sonoporation) and

the level of QD uptake is very significant since ultrasound–induced cell membrane

permeabilization is followed by the passive diffusion of QDs into the cytoplasm through

263 repairable holes/disruptions. The hydrodynamic size (core/shell + polymer) of QDs used in the experiment is approximately 15–20 nm in diameter [Dabbousi et al., 1997;

Pons et al., 2006; Smith and Nie, 2008; Tomczak et al., 2009; Ocean Nanotech, LLC]

and, thus, it is concluded that the size of transient pores/holes induced in the membrane

must be much larger than 20 nm in order to transport the significant amount of

CdSe/ZnS QDs into the cell by passive diffusion in 6 minutes of post sonoporation.

The consensus is that after ultrasound application, cell membrane permeability changes (or disruptions/holes are created in the cell membrane) due to radiation pressure, acoustic cavitation–induced events, shear stress by micro–streaming, and oscillating fluid. It was observed that at low or moderate input RF powers (0.5 mW, 1.0 mW, and 2.0 mW), the resulting sonoporation has high spatial specificity (or lateral resolution). At much higher input RF power (80 mW), the degree of spatial specificity was significantly reduced although there was a higher level of QD–uptake (figure 9–

10B). This is probably due to the fact that, at higher input RF powers, which correspond to higher ultrasound pressures, more powerful random cavitations are created and larger

mechanical stresses are also imposed on the cell membrane by larger radiation

pressures, more vigorous micro–streaming, and more powerful fluid oscillations.

Therefore, it is hypothesized that the increase in the number of random cavitation events

and much higher stresses induced from these events also permeabilize cells seeded on the interspaces between individual micro–transducers resulting in low spatial resolution

of sonoporation. Another reason for resulting low spatial resolution (or low site–

specificity) at high RF powers is that, at high RF powers, pressures from the side lobes of

ultrasonic beam due to radial vibrations of micro–transducers become large enough to sonoporate cells located on the interspaces. The hypothesized model of enhanced QD– uptake (or transmembrane QD transport) after sonoporation with UMTA biochip is

illustrated in figure 9–14.

264

Figure 9–14. Hypothesized model of enhanced QD transport (or transmembrane QD transport) induced by ultrasound generated from the micro–transducer of UMTA biochip. The size of the hole/disruption in the membrane is concluded to be much larger than hydrodynamic diameter of quantum dots.

9.4.1. Ultrasound–Induced Wound Model

Ultrasound–induced mechanical stresses produce transient wounds in the cell membrane [Schlicher et al., 2006]. Unlike electroporation, where the pores are small

(1–10 nm in radius) and short lived (a typical lifetime from milliseconds to seconds)

[Weaver and Chizmadzhev, 1996], the induced wound is large (> 150 nm in radius) and is resealed in a transient manner with lifetime τw > 1 minute via a spontaneous and active

(ATP– and Ca2+–dependent process) healing response involving endogenous vesicle– based membrane resealing [McNeil et al., 2000; Schlicher et al., 2006; Terasaki et al.,

1997; Togo et al., 1999; Zarnitsyn et al., 2008; Zhou et al., 2008]. In this process, intracellular vesicles are recruited to the site of the wound and their lipid material is used

265 to patch the membrane. Due to the presence of cytoskeletons (e.g., Spectrin spacing) and their associated proteins supporting the cell membrane, the wound cannot be treated as a complete open hole. Rather, the wound opening behaves like a sieve with nano–size pores as large as 50–70 nm in radius [Zarnitsyn et al., 2008]. The wound can be modeled as a circular opening of radius Rw with nano-size pores of radius dpore/2 within the wound (figure 9–15A) [Zarnitsyn et al., 2008].

Figure 9–15. (A) Ultrasound–induced wound model for one wound (Rw is wound radius; j is QD flux density; r is the distance from the center of the wound to the point of QD–flux measurement; Deff is effective diffusion coefficient of QDs; dpore is the averaged diameter of nanopores within the wound; h is cell membrane thickness; t is time; Cin and Cout are the intracellular and extracellular concentrations of QDs respectively). (B) Ultrasound–induced wound model for multiple wounds for mass transport calculations (Rw,eff is effective wound

th radius; Rw,i is the radius of i wound; N is the total number of wounds). The circumference of effective wound is equal to the total circumferences of individual wounds combined as the mass transport is largely along the wound edges. (C) Geometrical model representing the membrane wound resealing overtime.

266 For large wounds (i.e., Rw >> dpore/2), the flux density weakly depends on the number, size, and distribution of nano–size pores within the wound and the transport through the membrane wounds largely depends on the mass transfer along its edges

[Sneddon, 1966]. Research shows that the transient transport of nanoparticles during the ultrasound application is statistically insignificant in comparison to the post– sonoporation process [Zarnitsyn et al., 2008] since QDs are transported predominantly by passive diffusion in the absence of ultrasound–induced activities. It is hypothesize that at higher ultrasound pressure, larger wound is created (i.e., larger Rw) as experimental results show higher concentration of intracellular QDs after the membrane is completely resealed. For a monolayer of small cluster of cells (as well as more wounds in the single cell), the wound model is modified and induced wounds are estimated as a single large circular opening with the circumference equal to the sum of circumferences of individual wounds (i.e., total effective wound radius is equal to the sum of the radius of individual wounds, figure 9–15B) since the mass transfer is mostly along the edges.

N R R (9–1) w,eff = ∑ w,i i=1

In this modified model, Rw,eff reflects two extreme regimes (i.e., a few numbers of

very large wounds or many relatively smaller wounds) and any moderate regime in

between. During the resealing process, the porosity and pore size within the wound area

decreases over time due to the fusion of lipid vesicles. The process is modeled as overall wound area (or Rw,eff) decreases over time with the same pore distribution of smaller

nanopores within the wound (figure 9–15C). As the pore size decreases overtime, the

characteristic diffusion time for QDs through the wound (or) the mean effective wound

lifetime (defined as τw,eff), which largely depends on the hydrodynamic diameter (deff) of the diffusing molecule, is relatively shorter than the actual lifetime of the wound due to the geometrical hindrance. Since the wound decays exponentially [Zarnitsyn et al.,

267 2008; Zhou et al., 2008], the mathematical model of time–dependent effective wound radius for the passive diffusion of QDs is described by equation 9–2, where τw,eff is

employed in lieu of the actual wound lifetime.

t − 0 τ w,eff (9–2) R t = R e w,eff ( ) w,eff

9.4.2. Calculation of Peak Acoustic Radiation Pressure

By design, the peak ultrasonic pressure exerting on the cell is estimated to be almost

equal to the peak pressure at the end of Fresnel zone (frequency is adjusted so that NFL

is equal to the transducer to cell distance of ~15µm). As high aspect–ratio PMN–PT

micro–transducers are embedded in the Epoxy resin and piezoelectric coefficient d33 >> d31, the flexural vibration modes are dampened, leaving the longitudinal modes as the dominant vibration pattern for the transducer operation. The longitudinal vibrations of the PMN–PT transducer pillar of length L can be described by the wave equation

(equation 9–3) that has the general solution described in equation 9–4.

2 ˆ 2 ˆ ∂ ξ(x,t) ρ ∂ ξ(x,t) = (9–3) x2 t2 ∂ γ ∂

j(ωt−kx) j(ωt+kx) ξˆ x,t = Aˆe + Bˆe (9–4) ( )

ˆ where ρ, γ, ξ , ω, and k are the density of PMN–PT, Young’s modulus of PMN–PT,

longitudinal displacement of the micro–transducer, frequency of vibration, and wave number respectively. The micro–transducer is fixed at one end (attached on the glass substrate with silver Epoxy) and is mass–loaded at the other end (fluid column above the transducer), and thus, two boundary conditions can be obtained (equation 9–5 and 9–6)

[Kinsler et al., 1999; Uchino and Giniewicz, 2003].

268

Fixed end (no displacement): ξˆ 0,t = 0 (9–5) ( )

∂ξˆ fˆ d Eˆ Mass support (strain): = − + 33 (9–6a) ∂x Sγ x=L

∂ξˆ fˆ = Z S (9–6b) M ∂t x=L

ˆ ˆ where f , E , ZM, and S are complex force exerted on/by the micro–transducer, complex electric field applied across the micro–transducer, characteristic acoustic impedance of water, and cross–sectional area of the micro–transducer respectively. By applying the boundary conditions to the general solution of the wave equation, the longitudinal displacement of the micro–transducer (equation 9–7) and peak acoustic radiation pressure exerted by the micro–transducers onto the fluid column (equation 9–10) can be obtained.

∞ d Eˆ sin k x 33 γ ( n ) ⎡ ⎛ π ⎞⎤ ξˆ x,t = Exp ⎢ j⎜ω t − ⎟⎥ ( ) ∑ ⎢ ⎜ n ⎟⎥ (9–7) n=1 ⎝⎜ 2⎠⎟ 2ω n Zm sin(knL)− j ργ cos(knL) ⎣⎢ ⎦⎥ ( )

Resonance frequencies are:

⎛2n−1⎞⎛ c ⎞ ⎜ ⎟⎜ ⎟ fr,n =⎜ ⎟⎜ ⎟; n = 1, 2, 3, ... (9–8) ⎜ 4 ⎟⎜L⎟ ⎝ ⎠⎝ ⎠

Anti–resonance frequencies are:

⎛m⎞⎛ c ⎞ ⎜ ⎟⎜ ⎟ fa,n =⎜ ⎟⎜ ⎟; m = 1, 2, 3, ... (9–9) ⎜ 2 ⎟⎜L⎟ ⎝ ⎠⎝ ⎠

269 The magnitude of the peak acoustic radiation pressure is:

1 − ⎪⎧ 2 2 ⎪⎫ 2 ⎪⎛ 1 ⎞ ⎡ 1 ⎛ω ⎞⎤ ⎪ ˆ ˆ ⎪⎜ ⎟ ⎢ ⎜ ⎟⎥ ⎪ (9–10) ppeak (ω) = d33 E ⎨⎜ ⎟ + cot⎜ L⎟ ⎬ ⎪⎜ ⎟ ⎢ Z c ⎜ c ⎟⎥ ⎪ ⎪⎝γ ⎠ ⎣⎢ M ⎝ ⎠⎦⎥ ⎪ ⎩⎪ ⎭⎪

Eˆ Vˆ / L where c is velocity of sound in PMN–PT crystal and = is an applied ac electric field. Computed peak ultrasonic radiation pressures exerted by the micro–transducers on the LU1205 cells at 0.5 mW (0.96 Vp–p), 1.0 mW (1.31 Vp–p), and 2.0mW (1.79 Vp–p) at

30 MHz are 0.21 MPa, 0.29 MPa, and 0.40 MPa respectively [note: the output electrical impedance of RF signal generator (50 Ω) and input electrical impedance of micro– transducer (figure 7–23) were taken into account in the calculation]. Equation 9–10 shows that the peak pressure is maximum at resonance frequencies and is minimum (or zero) at anti–resonance frequencies for a particular Eˆ or RF power (figure 9–16). The normalized peak pressure Vs. frequency plot in figure 9–16b also shows that the quality factor, Q = Δf −6dB f , is very high for the 25–µm micro–transducers (characterized ( ) 1,n by very sharp peaks at resonance frequencies).

270

(A)

(B)

Figure 9–16. (A) Peak ultrasonic radiation pressure (at the end of NFL) Vs frequency for different RF powers applied to the micro–transducer. (B) Normalized Peak Pressure Vs frequency plot showing the bandwidth of multi–mode micro–transducer (S = 25µm × 25µm).

271 9.4.3. Mass Transport through the Effective Wound and Intracellular

Concentration of Quantum Dots

According to volumetric mass–balance equations, the rate of change in concentration

(C) of nanoparticles inside a reservoir of volume V is:

dC V = Jin − Jout (9–11) dt

where Jin and Jout are time–dependent rate of transport of nanoparticles into and out of the reservoir respectively. For one single cell, V = Vcell and since there is no significant

QD transport out of the cell (i.e., Jout = 0), the transport of QDs into the cell through the

wound opening can be described as:

dC in QD (9–12) Vcell = Jin dt

where C is the intracellular concentration of QDs and J QD is the time–dependent rate in in of transport of QDs (or the time–dependent flux of QDs) into the cell.

According to Fick’s 2nd law of diffusion in the cylindrical coordinate system, the rate of change in concentration of nanoparticles at any point across the cell membrane can be described as follows:

C r, , z, t ∂ ( θ ) 2 = Deff ∇ C (r,θ, z, t) ∂t ⎡ ⎛ ⎞ 2 2 ⎤ (9–13) ⎢1 ∂ ⎜ ∂C (r,θ, z, t)⎟ 1 ∂ C (r,θ, z, t) ∂ C (r,θ, z, t)⎥ = D ⎜r ⎟+ + eff ⎢ ⎜ ⎟ 2 2 2 ⎥ ⎢r ∂r ⎝⎜ ∂r ⎠⎟ r ∂θ ∂z ⎥ ⎣ ⎦

where Deff is the effective diffusivity of nanoparticles. For the simplified case in which the concentration gradient of nanoparticles is only in the z direction, the equation 9–13

can be reduced to:

272

2 ∂C (z, t) ∂ C (z,t) ∂ ∂C (z,t) = D = D = ∇ i ⎡D ∇C z,t ⎤ (9–14) eff 2 eff ⎣⎢ eff ( )⎦⎥ ∂t ∂z ∂z ∂z

Furthermore, at steady–state, equation 9–14 is reduced to Fick’s 1st law of diffusion as

follows:

⎡ ⎤ ∇ i Deff ∇C (z,t) = 0 ⇒ Deff ∇C (z,t)= const = −J (9–15) ⎣⎢ ⎦⎥

where J is the flux of nanoparticles through the cell membrane wound. As the plasma membrane is very thin compared to the wound size, it is hypothesized that the diffusion of QDs through the cell membrane wound is steady–state (i.e., the flux of QDs into and out of the wound are equal at any time during the diffusion although these fluxes are changing with time). Then, the time–dependent the flux of QDs into the cell is equal to the time–dependent flux of QDs through the cell membrane wound. Therefore, the volumetric mass–balance (equation 9–12) and Fick’s 1st law (equation 9–15) can be

combined as follows:

dC V in = J QD = D ∇C z,t = D ⎡C t −C t ⎤ (9–16) cell in eff ( ) eff ⎣⎢ out ( ) in ( )⎦⎥ dt where

−1 ⎡ h 1 1 ⎤ D ⎢ ⎥ eff = ⎢ + + ⎥ ⎢ Dpore Spores 4Dout Rw 4DinRw ⎥ ⎣ ⎦ h : Cell membrane thickness

Rw : Radius of the wound

Dout : Diffusion coefficient of QDs in the extracellular region

Din : Diffusion coefficient of QDs in the intracellular region

Dpore : Diffusivity of QDs in the pore

Spores : Total area of nanopores within the wound

Cout : Extracellular concentration of QDs

273 The effective diffusivity of QDs (Deff) is equal to the inverse of the total diffusion

resistance that is comprised of three individual diffusion resistances connected in series

(one resistance from the system of pores in the membrane of thickness h, the others two are constriction resistances, which arise from the wound openings on the extracellular

and intracellular side of the membrane). From the Stoke–Einstein equation, diffusion

coefficient of quantum dots in the extracellular fluid (or tissue culture media) can be

described as:

k T D = B (9–17) out 3π ηd eff where η : Viscosity (0.00078 Pa.s for DMEM)

deff : Hydrodynamic diameter of QDs

kB : Boltzmann constant

Experimental result show that the ratio Din/Dout is approximately 0.3 [Seksek et al., 1997;

Verkman, 2002]. Combining equation 9–2 (time–dependent effective wound radius) and equation 9–16, the rate of change in intracellular concentration of QDs can be described by the following differential equation:

⎡ ⎤−1 ⎢ ⎛ ⎞⎥ dCin (t) h 1 1 1 ⎟ V ⎢ ⎜ ⎟⎥ ⎡C C t ⎤ (9–18a) cell = ⎢ + ⎜ + ⎟⎥ ⎢ out − in ( )⎥ dt D S 0 ⎡ t ⎤ ⎜D D ⎟ ⎣ ⎦ ⎢ pore pores 4R Exp − ⎝ out in ⎠⎥ ⎢ w,eff ⎢ τ ⎥ ⎥ ⎣ ⎣⎢ w,eff ⎦⎥ ⎦

Initial condition: C t = 0 (9–18b) in ( 0 )

where t0 is the time at which QDs are introduced into the extracellular liquid (for

instance, if QDs are introduced immediately after the sonoporation, t0 = 0).

274 By using separation of variables and applying the initial condition, the time– dependent intracellular concentration of quantum dot can be obtained (equation 9–19).

⎧ ⎡ −1 ⎤⎫ ⎪ ⎛ ⎞ ⎪ ⎪ ⎢ ⎜ ⎟ ⎥⎪ ⎪ ⎢ t ⎛ 1 ⎟⎞⎜ h 1 ⎛ 1 1 ⎟⎞⎟ ⎥⎪ C t = C ⎪1−Exp ⎢− ⎜ ⎟⎜ + ⎜ + ⎟⎟ dt⎥⎪ in ( ) out ⎨ ⎢ ∫ ⎜ ⎟⎜ ⎜ ⎟⎟ ⎥⎬ (9–19) ⎪ t0 ⎜V ⎟⎜D S 0 ⎡ t ⎤ ⎜D D ⎟⎟ ⎪ ⎪ ⎢ ⎝ cell ⎠⎜ pore pores 4R Exp − ⎝ out in ⎠⎟ ⎥⎪ ⎪ ⎜ w,eff ⎢ τ ⎥ ⎟ ⎪ ⎪ ⎢ ⎝ ⎣⎢ w,eff ⎦⎥ ⎠ ⎥⎪ ⎩⎪ ⎣ ⎦⎭⎪

For the wound opening larger than the thickness (i.e., Spores >> h) and for multiple cells

the equation 9–19 can be simplified as follows:

⎧ ⎡ −1 ⎤⎫ ⎪ ⎛ ⎞ ⎪ ⎪ ⎢ ⎜ ⎟ ⎥⎪ ⎪ ⎢ t ⎛ 1 ⎟⎞⎜ 1 ⎛ 1 1 ⎟⎞⎟ ⎥⎪ C t = C ⎪1−Exp ⎢− ⎜ ⎟⎜ ⎜ + ⎟⎟ dt⎥⎪ in ( ) out ⎨ ⎢ ∫ ⎜ ⎟⎜ ⎜ ⎟⎟ ⎥⎬ (9–20) ⎪ t0 ⎜V ⎟⎜ 0 ⎡ t ⎤ ⎜D D ⎟⎟ ⎪ ⎪ ⎢ ⎝ cells ⎠⎜4R Exp − ⎝ out in ⎠⎟ ⎥⎪ ⎪ ⎜ w,eff ⎢ τ ⎥ ⎟ ⎪ ⎪ ⎢ ⎝ ⎣⎢ w,eff ⎦⎥ ⎠ ⎥⎪ ⎩⎪ ⎣ ⎦⎭⎪

9.5. Analysis and Simulations

9.5.1. QD–Uptake and the Size of Membrane Wound

The input RF power applied correlates to the radiation pressure of the ultrasound generated by the piezoelectric micro–transducers. The higher is the input RF power (or the higher Eˆ ), the higher is the ultrasound pressure generated (equation 9–10 and figure

9–16). Furthermore, ultrasound–pressure dependent QD–uptake can be correlated to the ultrasound–pressure–dependent cell membrane wound size. The size of the cell membrane wound is a measure of cell membrane permeability after sonoporation.

9.5.1.1. Ultrasound–Pressure–Dependent QD–Uptake

The intracellular concentrations of quantum dots loaded into the LU1205 cells seeded on the micro–transducers’ active area can be estimated from the experimental data (figure 9–9) using equation 9–21, where the sum fluorescent intensity spatial

275 profiles (figure 9–9) are fitted as Gaussian function as follows (note: the reason for using

Gaussian function will be explained in section 9.5.4):

⎪⎧ ⎡ 2 ⎤ ⎪⎫ ⎪ ∞ I x − µ ⎪ ⎪ ( sum )max ⎢ ( ) ⎥ ⎪ Cin (t = 360s, t0 = 0)= ⎨ Exp ⎢− ⎥ dx⎬ × α (9–21) ⎪∫−∞ 3 ⎢ 2 2 ⎥ ⎪ ⎪ ⎢ σ ⎥ ⎪ ⎩⎪ ⎣ ⎦ ⎭⎪

−13 where α = 1.0413×10 [M/A.U.] is a fluorescence–to–intracellular–QD concentration conversion factor (an empirical conversion factor) and Isum is the sum fluorescent

intensity per element column in the pixel array (see section 9.2.4.2.). The α is

calculated as follows: it is estimated empirically from QD photoluminescence (PL)

spectra that a 2D quantum–dot layer in each pixel area (~278nm × ~278nm) emits

fluorescence intensity of approximately 1000 A.U. and contains ~196 quantum dots. The

averaged thickness of cell monolayer is estimated to be ~5µm (i.e., the volume occupied by a portion of cell above each pixel area is: Volume = ~278nm × ~278nm × ~5µm).

Then, the fluorescence–to–intracellular–QD–concentration conversion factor (α

[M/A.U.]) can be obtained by the equation 9–22.

# of quantum dots per 2D layer α = (9–22) 1000 Avogadro's # Volume × ×

Using equation 9–10 (peak ultrasound radiation pressure at the end of Fresnel zone),

equation 9–21, and equation 9–22, the estimated intracellular concentrations of QDs, which are transported into the cytoplasm of the LU1205 cells (seeded on the 25µm–by–

25µm active area presented by a micro–transducer), after post–sonoporation 6–minute

inoculation with QDs for three applied radiation pressures of 0.21 MPa (0.5 mW), 0.29

MPa (1.0 mW), and 0.40 MPa (2.0 mW) are 7.8 ± 2.3 nM, 22.8 ± 1.6 nM, and 29.9 ± 2.5

nM respectively (see figure 9–17).

276

Figure 9–17. Estimated intracellular concentrations of quantum dots, which were transported into the cytoplasm of the LU1205 cells seeded on the 25µm × 25µm active area presented by a micro–transducer, after 6–minute inoculation with QDs (post sonoporation) for three peak ultrasound radiation pressures applied to the cells.

9.5.1.2. Ultrasound–Pressure–Dependent Wound Size

From equation 9–20, ultrasound–induced initial effective wound radius can be calculated (using t = 0, = 40 s for d = 20 nm [Zarnitsyn et al., 2008], t = 6 min, o τw,eff eff

Cout = 100nM, and Cin’s from figure 9–17). The initial effective wound radius is a

measure of cell membrane permeability due to sonoporation. The estimated initial effective wound radius ( R0 ) are 150 ± 45 nm, 460 ± 50 nm and 650 ± 70 nm for w,eff applied ultrasound radiation pressures of 0.21 MPa, 0.29 MPa, and 0.40 MPa respectively. It is observed that the initial effective wound size increases in a linear fashion with ultrasonic pressure (figure 9–18). Extrapolated plot shows that the

pˆ 30MHz threshold ultrasound pressure ( peak ( )) required to induce enhanced QD th transport into LU1205 cells by sonoporation is ~0.12 MPa. The threshold pressure is the pressure at which cell membrane undergoes stress failure. Below this threshold, the cell membrane is still intact and may not be permeable enough for QDs to be transported via passive diffusion.

277

Figure 9–18. Calculated initial effective wound radius ( R0 ) for three peak w,eff ultrasound radiation pressures (0.2, 0.3, and 0.4 MPa). Extrapolated plot shows that the threshold ultrasound pressure required to induce enhanced QD transport into LU1205 cells by sonoporation is ~0.12 MPa.

9.5.2. Model–Based Simulations and Calculations

9.5.2.1. Analysis of the Influence of Transport–Model Parameters on the

Enhanced QD–Uptake by the Cells Sonoporated with UMTAs

Equation 9–20 can be used to analyze the influence of ultrasound radiation pressure and wound healing kinetics on the sonoporation–enhanced transport process of nanoparticles (e.g. QDs) into the cells and will allow us to precisely control and enhance the delivery of drugs or gene products into the cells using UMTA–biochip platforms.

This section describes the analysis of the influence of ultrasound radiation pressure

(represented by the initial effective wound radius ( R0 ) due to the observed linear w,eff relationship between the low– and medium–range pressure to the wound radius in

figure 9–18), mean effective wound lifetime (τw,eff), and the lag time of QD introduction following the sonoporation (to) on the sonoporation–enhanced QD–transport process across the cell membrane (using equation 9–20).

278 First, time–dependent intracellular concentrations of QDs transported into the cells were computed and for: 1), ultrasound–pressure–controlling regime (or the sonoporation–induced transmembrane permeability enhancement) (figure 9–19A); 2), wound–healing–limiting regime (figure 9–19B), where the enhanced transmembrane transport of QDs through the wound is limited by the wound healing dynamics

(represented by the characteristic QD diffusion time or mean effective wound lifetime

(τw,eff)); and 3), different lag times (figure 9–19C). These time–dependent plots of intracellular QD concentrations (figure 9–19A, 9–19B, 9–19C) show the dynamics and rate of QD–uptake by the cells after 3–minute sonoporation (or the rate of enhanced

QD–transport through membrane wounds) before the cell membrane is completely resealed. The time–dependent QD concentrations inside the cells will saturate due to the wound healing/sealing. The plots also show that the initial QD–transport rate is determined by the initial effective wound size (or the ultrasound radiation pressure) and the total QD–uptake is determined by both the wound healing dynamics, i.e., the mean effective wound lifetime (τw,eff), and the ultrasound radiation pressure.

Figure 9–19. Computed time–dependent intracellular concentrations of QDs transported into the cells for: (A) different initial effective wound size or radius ( R0 ), i.e., ultrasound pressure w,eff dependence; (B) for different mean effective wound lifetimes (τw,eff), i.e., wound healing dynamic dependence; and (C) different lag times (to) for QD introduction following the sonoporation. is fixed at 40s in (A) and (C), R0 is fixed at 500nm in (B) and (C). The τw,eff w,eff extracellular concentration of QDs is kept at 100 nM for all calculations.

279 Second, total concentrations of QDs transported into the cells in ~6 minutes after the

site–specific sonoporation were computed for varying R0 , , and t . The plots in w,eff τw,eff o

figure 9–20A and 9–20B show linear dependence of total QD–uptake, which is also

confirmed experimentally, on the applied ultrasonic radiation pressure onto the cell

membrane (represented by R0 ) as well as on the mean effective wound lifetime ( ). w,eff τw,eff

On the other hand, the total uptake of QDs by the cells in ~6 minutes after the site– specific sonoporation decreases exponentially with increasing lag time (to) (figure 9–

20C) mainly due to the fact that the membrane wound decays exponentially [Zarnitsyn et al., 2008; Zhou et al., 2008]. These analyses provide better understanding of sonoporation–enhanced transmembrane transport of nanoparticles and will allow us to precisely control and enhance the delivery of drugs or gene products into the cells by

UMTA–biochip platforms.

Figure 9–20. Computed total concentrations of QDs transported into the cells in 6 minutes after the sonoporation (or until the wound is completely healed) for: (A) different initial effective wound size or radius ( R0 ), i.e., ultrasound pressure dependence; (B) for different w,eff mean effective wound lifetimes (τw,eff), i.e., wound healing dynamic dependence; and (C)

different lag times (to) for QD introduction following the sonoporation. τw,eff is fixed at 40s in (A) and (C), R0 is fixed at 500nm in (B) and (C). The extracellular concentration of QDs is w,eff kept at 100 nM for all calculations.

280 9.5.2.2. Sonoporation–Enhanced Transport and QD–Uptake

Equation 9–20 describes the time–dependent intracellular concentration of QDs due to site–specific sonoporation by UMTAs. Therefore, to compute the sonoporation–

induced transmembrane permeability (or transport) enhancement factor (φE) and QD–

uptake enhancement factor (υE), a second time–dependent equation that describes the dynamic of endocytosis–driven QD–uptake is required and is obtained as follows: the flow cytometry data related to endocytosis–driven QD–uptake by LU1205 cells are converted into intracellular concentrations for respective QD–inoculation periods (the data are coupled with time–lapsed fluorescence microscopy and fluorescence–to– intracellular–QD–concentration conversion factor (α) is also used). The resulting values are numerically fitted (3rd–order polynomial fitting in Mathematica®) to obtain the

time–dependent intracellular QD concentrations due to endocytosis (equation 9–22).

C endo t, in hours ≈−0.0011t 3 + 0.1395t2 + 0.8021t +1.8669 (9–22) in ( )

The sonoporation–induced transmembrane permeability enhancement factors (φE), which is the ratio of sonoporation–induced rate of change of intracellular QD concentration (or sonoporation–induced rate of QD uptake) just after the sonoporation process (t = t0) to the rate of endocytosis–driven QD–uptake at the point just after the

introduction of QD–suspended media, is calculated using equation 9–23. The φE also represents the degree of sonoporation by UMTAs and largely depends on ultrasound pressure, the hydrodynamic diameter of QDs, endocytosis mechanism, and extracellular concentration of QDs. The averaged permeability enhancement factors via sonoporation for LU1205 cells are 712, 1894, and 2927 for applied ultrasonic radiation pressures of

0.21 MPa, 0.29 MPa, and 0.40 MPa respectively.

281 For sonoporation–enhanced QD–uptake (υE) by LU1205 cells, the estimated enhancement factor, which is the ratio of the diffusion time it takes to achieve specified intracellular concentration of QDs via endocytocis to that by sonoporation, is evaluated using equation 9–24. The υE takes into account of wound healing kinetics of a particular

cell line and also depends on the ultrasound pressure, extracellular concentration of

QDs, the hydrodynamic diameter of QDs, and endocytosis mechanism. From transfection and gene therapy perspective, υE is more of a practical evaluation. The averaged QD–uptake enhancement factors by sonoporation are 86, 200, and 241 for applied ultrasonic radiation pressures of 0.21 MPa, 0.29 MPa, and 0.40 MPa respectively

(Note: in calculation of υE, the specified intracellular concentrations are chosen at

respective ΔC = C sono t = 180 s for different ultrasound pressures, i.e., Δt = 180 s . The in in ( ) sono

reason is that cell membrane is almost resealed and QD diffusion through the membrane

wound is not significant after 3 minutes).

⎛ sono ⎞ ⎛ endo ⎞ ⎜dCin ⎟ ⎜dCin ⎟ φ ≈⎜ ⎟ ⎜ ⎟ (9–23) E ⎜ dt ⎟ ⎜ dt ⎟ ⎝ ⎠ ⎝ ⎠ t=t 0

⎛ ⎞ ⎛ ⎞ t ⎜ ΔCin ⎟ ⎜ ΔCin ⎟ Δ endo υ =⎜ ⎟ ⎜ ⎟= (9–24) E ⎜Δt ⎟ ⎜Δt ⎟ Δt ⎝ sono ⎠ ⎝ endo ⎠ sono

9.5.3. Effective Lateral Resolution of Sonoporation Using UMTA Biochips

For ultrasounds generated from the piston–type transducer, the diameter of ultrasound beam (or the lateral resolution) at the end of Fresnel zone (or at the end of

Near Field Length (NFL)) is 9.1 µm (calculated using equation 9–25 with normalized focal length SF = 1 and diagonal width or aperture width of wd = 35 µm for 25–µm unfocused transducer).

282 BD −6dB ≈ 0.26w S (9–25) ( ) d ( F )

w here

wd : the diagonal width of micro–transducer

SF : normalized focal length (1 for unfocused transducer)

For an unfocused transducer, the ultrasound pressure is peak at the end of NFL and the pressure distribution decreased rapidly with radial distance within the beam with the first diffraction minimum (much lower than the peak and can be negligible) near the

edge of the beam [Rahim et al., 2006(a); Lu et al., 1994]. Therefore, the sum intensity

spatial profiles (figure 9–9) are correlated to the ultrasound pressure distribution at the

end of NFL, which is estimated as Gaussian function. As the cells are seeded at the end

of the Fresnel zone, the lateral resolution of site–specific sonoporation by the micro–

transducer array can be approximated as BD(–6dB). However, in practice, the effective lateral resolution of sonoporation is lower due to side lobes of ultrasonic beam generated

from the transducer due to radial vibrations [Lu et al., 1994]. Therefore, the FWHM

(equation 9–26) of sum intensity spatial profiles from figure 9–9, which are correlated to the ultrasound pressure distributions, is used to estimate the approximate lateral resolution of site–specific sonoporation provided that sonoporated cells seeded above

the transducers regained their morphology and did not migrate within ~30 minutes

during the experiment. Using FWHM of sum intensity spatial profiles should take into

account of reduced lateral resolution due to cell morphology, cell spreading, and

intracellular QD diffusion. The estimated effective lateral resolution of site–specific

sonoporation at the end of Fresnel zone (Δlsono) for 25–µm transducers at 30–MHz

frequency is 18 ± 3µm (averaged for all three peak ultrasound pressures, which are 0.21

MPa, 0.29 MPa, and 0.40 MPa).

Δl = FWHM ≈ 2σ 2 ln2 (9–26) sono

283 It is, however, noteworthy, that this uncovered resolution of site–specific sonoporation with UMTAs only applies to low and medium RF power levels. The effective lateral resolution of site–specific sonoporation becomes very poor when micro– transducer array is driven at very high input RF power (80 mW) (figure 9–10), where pressures from side lobes of the ultrasonic beam become large enough to sonoporate cells located on the interspaces between transducer active areas.

9.6. Summary

The experimental results show that UMTA biochip can permeabilize the cell membrane with high degree of spatial resolution (or site specificity) and enhance the transport of nanoparticle quantum dots into the cells. The architecture and design of

UMTA biochip brings miniature high–aspect–ratio transducers close to the targeted cells. This enables us to tailor NFL equal to transducer–to–cell distance. Owing to this novel device design, direct mechanical stresses induced by ultrasounds can create transient wounds in the cell membrane. These stresses are more controllable by adjusting ultrasonic radiation pressures compared to random–cavitation–induced stresses generated indirectly by traditional sonoporation setups/techniques. The spatial resolution of site–specific sonoporation as well as ultrasound generated is also high enough for the implementation of UMTAs in cell–based assays. CdSe/ZnS QD–uptake is significantly enhanced (an enhancement factor of > 100 at 0.29 MPa for LU1205 cells) due to passive diffusion of QDs in comparison to much slower QD–uptake driven by endocytosis. Model–based calculation and analysis show that sonoporation level, determined by the wound size, is linearly dependent on the ultrasound pressure applied and the threshold pressure required to permeabilize LU1205 cells is ~0.12MPa. The work represents an important step towards the development of high–throughput cell– based multiplex drug screening, transfection assay, and sensing platforms.

284 9.7. Chapter Appendix

9.7.1. List of Symbols

α [M/A.U] Fluorescence–to–intracellular–QD–concentration conversion factor (an empirical conversion factor)

c [m/s] Speed of sound in water

Cin [nM] Intracellular concentration of quantum dots

Cout [nM] Extracellular concentration of quantum dots

d33 [C/N] Piezoelectric coefficient of PMN–PT (both polarization and stress in direction 3)

d31 [C/N] Piezoelectric coefficient of PMN–PT (polarization in direction 3 and stress in direction 1)

dpore [nm] Diameter of nanopores within the membrane wound

deff [nm] Hydrodynamic diameter of quantum dots

2 Dpore [cm /s] Diffusion coefficient of quantum dots within the membrane wound channel

2 Dout [cm /s] Diffusion coefficient of quantum dots (extracellular region)

2 Din [cm /s] Diffusion coefficient of quantum dots (intracellular region)

2 Deff [cm /s] Effective diffusion coefficient (or diffusivity) of quantum dots

Δlsono [µm] Effective lateral resolution of UMTAs for site–specific cellular sonoporation

Eˆ [N/C] Applied electric field across the micro–transducer

ˆ ξ [nm] Displacement of the micro–transducer

η [Pa.s] Viscosity of tissue culture media

f [Hz] Frequency

ˆ f [N] Radiation force

fr [Hz] Resonance frequency of micro–transducer

285 fa [Hz] Anti–resonance frequency of micro–transducer

γ [Pa] Young’s Modulus of PMN–PT

h [nm] Cell membrane thickness

Isum [A.U.] Sum fluorescence intensity of QD emissions per element column in the pixel array

k [m–1] Wave number

kBT [J/K] Boltzmann constant

L [µm] Length of micro–transducer

λ [m] Wavelength

µ Mean value

N Number of cell membrane wounds

n Longitudinal wave mode

υE Sonoporation–induced QD–uptake enhancement factor

ω [rad] Frequency

Pˆ [Pa] Peak ultrasound radiation pressure peak

φE Sonoporation–induced transmembrane permeability (or transport) enhancement factor

Rw [nm] Radius of the cell membrane wound

o Rw,eff [nm] Initial effective wound radius

Rw,eff [nm] Effective wound radius

r [nm] The distance from the center of the wound to the point of QD–flux measurement

ρ [kg/m3] Density of Lead Magnesium Niobate–Lead Titanate (PMN–PT)

S [µm2] Cross–sectional area of micro–transducer

286 SF [µm] Normalized focal length

2 Spores [nm ] Total area of nanopore openings within the wound

σ Standard deviation

to [s] Lag time

τ w [s] Characteristic wound healing time (or) wound lifetime

τ w,eff [s] Characteristic diffusion time for QDs through cell membrane wound (or) mean effective wound lifetime

3 Vcells [µm ] Total volume of cells occupied on the micro–transducer’s active area

wd [µm] Diagonal width of micro–transducer

ZM [Pa.s/m] Characteristic acoustic impedance of water

9.7.2. List of Some Abbreviations

BD Beam Diameter

DMEM Dulbecco's Modified Eagle's Medium

DPBS Dulbecco’s Phosphate Buffered Saline

FBS Fetal Bovine Serum

FWHM Full Width at Half Maximum

GFP Green Fluorescent Protein

MCB Microbubbles

NFL Near Field Length

PL Photoluminescence

PMN–PT Lead Magnesium Niobate–Lead Titanate

PS Penicillin–Streptomycin

QD Quantum Dot

RF Radio Frequency

UCA Ultrasound Contrast Agent

UMTA Ultrasonic Micro–Transducer Array

287 References

Akinlaja J., and Sachs, F., 1998. The Breakdown of Cell Membranes by Electrical and Mechanical Stress. Biophysical Journal. 75, 247–254.

Dabbousi, B. O., Rodriguez – Viejo, J., Mikulec, F. V., Heine, J. R., Mattoussi, H., Ober, R., Jensen, K. F., and Bawendi, M. G., 1997. (CdSe)ZnS Core–Shell Quantum Dots: Synthesis and Characterization of a Size Series of Highly Luminescent Nanocrystallites. Journal of Physical Chemistry B. 101, 9463–9475.

Doinikov, A. A., 1994. Acoustic Radiation Pressure on a Compressible Sphere in a Viscous Fluid. The Journal of Fluid Mechanics. 267, 1–21.

Dothager, R. S., Putt, K. S., Allen, B. J., Leslie, B. J., Nesterenko, V., and Hergenrother, P. J., 2005. Synthesis and Identification of Small Molecules that Potently Induce Apoptosis in Melanoma Cells through G1 Cell Cycle Arrest. Journal of American Chemical Society. 127, 8686–8696.

Ghosh, R., Nadiminty, N., Fitzpatrick, J. E., Alworth, W. L., Slaga, T. J., and Kumar, A. P., 2005. Eugenol Causes Melanoma Growth Suppression through Inhibition of E2F1 Transcriptional Activity. The Journal of Biological Chemistry. 280 (7), 5812–5819.

Guzman, H. R., McNamara, A. J., Nguyen, D. X., and Prausnitz, M. R., 2003. Bioeffects Caused by Changes in Acoustic Cavitation Bubble Density and Cell Concentration: A Unified Explanation Based on Cell–to–Bubble ratio and Blast Radius. Ultrasound in Medicine and Biology. 29 (8), 1211–1222.

Hasegawa, T., Hino, Y., Annou, A., Noda, H., Kato, M., and Inoue, N., 1993. Acoustic Radiation Pressure Acting on Spherical and Cylindrical Shells. Journal of the Acoustical Society of America. 93 (1), 154–161.

Kinsler, L. E., Frey, A. R., Coppens, A. B., and Sanders, J. V., 1999. Fundamentals of Acoustics. 4th Edition, John Willy & Sons, New York.

Liu, M., Tanswell, A. K., and Post, M., 1999. Mechanical Force–Induced Signal Transduction in Lung Cells. American Journal of Physiology: Lung Cellular and Molecular Physiology. 277, 667–683.

Liu, Y., Cho, C. –W., Yan, X., Henthorn, T. K., Lillehei, K. O., Cobb, W. N., and Ng, K. –Y., 2001. Ultrasound–Induced Hyperthermia Increases Cellular Uptake and Cytotoxicity of P– Glycoprotein Substrates in Multi–Drug Resistant Cells. Pharmaceutical Research. 18, 1255– 1261.

Lu, J. –Y., Zou, H., and Greenleaf, J. F., 1994. Biomedical Ultrasound Beam Forming. Ultrasound in Medicine and Biology. 20 (5), 403–428.

288

McNeil, P. L., and Steinhardt, R. A., 1997. Loss, Restoration, and Maintenance of Plasma Membrane Integrity. The Journal of Cell Biology. 137 (1), 1–4.

McNeil, P. L., Vogel, S. S., Miyake, K., and Terasaki, M., 2000. Patching Plasma Membrane Disruptions with Cytoplasmic Membrane. Journal of Cell Science. 113 (11), 1891–1902.

Medintz, I. L., Uyeda, H. T., Goldman, E. R., and Mattoussi, H., 2005. Quantum Dot Bioconjugates for Imaging, Labelling, and Sensing. Nature Materials. 4, 435–446.

Michalet, X., Pinaud, F. F., Bentolila, L. A., Tsay, J. M., Doose, S., Li, J. J., Sundaresan, G., Wu, A. M., Gambhir, S. S., and Weiss, S., 2005. Quantum Dots for Live Cells, in vivo Imaging, and Diagnostics. Science. 307 (5709), 538–544.

Mitri, F. G., 2006. Acoustic Radiation Force due to Incident Plane–Progressive Waves on Coated Cylindrical Shells Immersed in Ideal Compressible Fluids. Wave Motion. 43, 445–457.

Moioli, E. K., Shah, B., Clark, P. A., Stroscio, M., and Mao, J. J., 2006. Quantum Dot Labeling of Stem Cells during Proliferation and Differentiation. Molecular and Cellular Biomechanics. 3 (4), 153–156.

Muller–Borer, B. J., Collins, M. C., Gunst, P. R., Cascio, W. E., Kypson, A. P., 2007. Quantum Dot Labeling of Mesenchymal Stem Cells. Journal of Nanobiotechnology. 5 (9).

Nyborg, W. L., 1982. Ultrasonic Microstreaming and Related Phenomena. British Journal of Cancer. 45 (Suppl. V), 156–160.

Nyborg, W. L., 2001. Biological Effects of Ultrasound: Development of Safety Guidelines. Part II: General Review. Ultrasound in Medicine and Biology. 27 (3), 301–333.

Papazoglou, E., Edrissi, B., Rezvan, A., and Pourrezaei, K., 2004. Q–Dots as Probes for Dynamic Observation of Intracellular Pathways. Proceedings of the IEEE 30th Annual Northeast Bioengineering Conference. 231–232.

Pitt, W. G., Husseini, G. A., and Staples, B. J., 2004. Ultrasonic Drug Delivery – A General Review. Expert Opinions in Drug Delivery. 1 (1), 37–56.

Pons, T., Uyeda, H. T., Medintz, I. L., and Mattoussi, H., 2006. Hydrodynamic Dimensions, Electrophoretic Mobility, and Stability of Hydrophilic Quantum Dots. Journal of Physical Chemistry B. 110 (41).

Rahim, A. A., Taylor, S. L., Bush, N. L., ter Haar, G. R., Bamber, J. C., and Porter, C. D., 2006 (a). Spatial and Acoustic Pressure Dependence of Microbubble–Mediated Gene Delivery Targeted Using Focused Ultrasound. The Journal of Gene Medicine. 8, 1347–1357.

289 Rahim, A., Taylor, S. L., Bush, N. L., ter Haar, G. R., Bamber, J. C., and Porter, C. D., 2006 (b). Physical Parameters Affecting Ultrasound/Microbubble–Mediated Gene Delivery Efficiency in vitro. Ultrasound in Medicine and Biology. 32 (8), 1269–1279.

Rooney, J. A., 1970. Hemolysis Near an Ultrasonically Pulsating Gas Bubble. Science. 169, 869– 871.

Rosenthal, S. J., Tomlinson, I., Adkins, E. M., Schroeter, S., Adams, S., Swafford, L., McBride, J., Wang, Y., DeFelice, L. J., and Blakely, R. D., 2002. Targeting Cell Surface Receptors with Ligand–Conjugated Nanocrystals. Journal of American Chemical Society. 124, 4586–4594.

Schlicher, R. K., Radhakrishna, H., Tolentino, T. P., Apkarian, R. P., Zarnitsyn, V., and Prausnitz, M. R., 2006. Mechanism of Intracellular Delivery by Acoustic Cavitation. Ultrasound in Medicine and Biology. 32 (6), 915–924.

Seksek, O., Biwersi, J., and Verkman, A. S., 1997. Translational Diffusion of Macromolecule– Sized Solutes in Cytoplasm and Nucleus. The Journal of Cell Biology. 138, 131–142.

Smith, A. M., and Nie, S., 2008. Minimizing the Hydrodynamic Size of Quantum Dots with Multifunctional Multidentate Polymer Ligands. Journal of American Chemical Society. 130 (34), 11278–11279.

Sneddon, I. N., 1966. Mixed Boundary Value Problems in Potential Theory. John Willy & Sons, New York.

Song, J., Chappell, J. C., Qi, M., VanGieson, E. J., Kaul, S., and Price, R. J., 2002. Influence of Injection Site, Microvascular Pressure and Ultrasound Variables on Microbubble–Mediated Delivery of Microspheres to Muscle. Journal of the American College of Cardiology. 39 (4), 726–731.

Sundaram, J., Mellein, B. R., and Mitragotri, S., 2003. An Experimental and Theoretical Analysis of Ultrasound–Induced Permeabilization of Cell Membranes. Biophysical Journal. 84, 3087– 3101.

Starritt, H. C., Duck, F. A., and Humphrey, V. F., 1989. An Experimental Investigation of Streaming in Pulsed Diagnostic Ultrasound Beams. Ultrasound in Medicine and Biology. 15 (4), 363–373.

Terasaki, M., Miyake, K., and McNeil, P. L., 1997. Large Plasma Membrane Disruptions Are Rapidly Resealed by Ca2+–dependent Vesicle–Vesicle Fusion Events. The Journal of Cell Biology. 139 (1), 63–74.

Togo, T., Alderton, J. M., Bi, G. –Q., and Steinhardt, R. A., 1999. The Mechanism of Facilitated Cell Membrane Resealing. Journal of Cell Science. 112, 719–731.

290 Tomczak, N., Janczewski, D., Han, M., and Vancso, G. J., 2009. Designer Polymer–Quantum Dot Architectures. Progress in Polymer Science. 34, 393–430.

Tschumperlin D. J., Oswari J., and Margulies S. S., 2000. Deformation–Induced Injury of Alveolar Epithelial Cells. American Journal of Respiratory and Critical Care Medicine. 162, 357–362.

Uchino, K., and Giniewicz, J. R., 2003. Micromechatronics. Marcel Dekker, New York.

Valhakis, N. E., and Hubmayr, R. D., 2000. Cellular Responses to Mechanical Stress. Invited Review: Plasma Membrane Stress Failure in Alveolar Epithelial Cells. Journal of Applied Physiology. 89, 2490–2496.

Verkman, A. S., 2002. Solute and Macromolecule Diffusion in Cellular Aqueous Compartments. Trends in Biochemical Sciences. 27, 27–33.

Wang, Y. –X. J., Leung, K. C. –F., Cheung, W. –H., Wang, H. –H., Shi, L., Wang, D. –F., Qin, L., and Ahuja, A. T., 2010. Low–Intensity Pulsed Ultrasound Increases Cellular Uptake of Superparamagnetic Iron Oxide Nanomaterial: Results from Human Osteosarcoma Cell Line U2OS. Journal of Magnetic Resonance Imaging. 31, 1508–1513.

Weaver, J. C., and Chizmadzhev, Y., 1996. Theory of Electroporation: A Review. Bioelectrochemistry and Bienergetics. 41, 135–160.

Zarnitsyn, V., Rostad, C. A., and Prausnitz, M. R., 2008. Modeling Transmembrane Transport through Cell Membrane Wounds Created by Acoustic Cavitation. Biophysical Journal. 95, 4124–4138.

Zhang, L. W., and Monteiro–Riviere, N. A., 2009. Mechanisms of Quantum Dot Nanoparticle Cellular Uptake. Toxicological Sciences. 100 (1), 138–155.

Zhou, Y., Shi, J., Cui, J., and Deng, C. X., 2008. Effects of Extracellular Calcium on Cell Membrane Resealing in Sonoporation. Journal of Control Release. 126: 34–43.

291 Chapter 10 CONCLUSIONS AND FUTURE WORK

The work presented has shown and demonstrated the many capabilities and benefits of microfabrication technology (i.e., the science of miniaturization) in the development of biochips, which is a growing biotechnology industry. The integrated Microelectrode

Array (IMA) chip, which employs microelectrode array architecture, and the Ultrasonic

Micro–Transducer Array (UMTA) biochip, which employs piezoelectric–based micro– size transducers array architecture, were successfully micro–fabricated. The process modules, which were developed, characterized, and integrated for the fabrication of these chips, includes photolithography, laser lithography, high frequency pulsed electroplating, physical vapor deposition (PVD), plasma–based reactive ion etching (i.e.,

Reactive Ion Etching (RIE) and Inductively Coupled Plasma (ICP)), plasma enhanced chemical vapor deposition (PECVD), Chemical Vapor Deposition (CVD), furnace annealing/passivation, wet chemical etching, chemical mechanical planarization/ polishing (CMP), and wire bonding. However, two processes developed for fabrication of

UMTA biochip will require process optimization in order to improve yields (i.e., more biochips per processing) and efficiency (i.e., reduction in processing time). These two process modules are CMP and wire bonding. In this work, these two processes were carried out mostly by hands using in–house tools and fixtures. The reproducibility (or precision) was quite poor and the yield was low. Therefore, these two processes should be optimized for automated processing using industrial–standard precision tools.

It was demonstrated that by integrating with proper instrumentations (such as

phase–sensitive lock–in detection, fluorescent microscopy, etc.), the IMA chip and

UMTA biochips become powerful, robust, and versatile characterization/screening tools.

For instance, bottom–up surface engineering approaches were employed to IMA chips

292 and single–cell level manipulations such as cell separation and cell immobilization on the working microelectrodes were successfully achieved through surface–mediated guided cell adhesion. Then, these surface–modified IMA chips, integrated with the phase–sensitive lock–in amplifier, enable characterization of cell membrane properties of individual cells immobilized on the working microelectrodes (i.e., cell membrane characterization at the single–cell level). In these characterization studies, area– contract–model–based Equivalent Circuit Models (ECMs), which represent the nature of

cell–electrode heterostructure, were employed. These ECMs were used to accurately

interpret the electrical signals transduced from IMA sensor configurations and enabled

successful characterization of the cell membrane properties at the single–cell level.

In the UMTA–biochip–based configuration, an RF signal generator was used to

activate micro–transducers in order to generate ultrasound radiation pressures. These

ultrasound radiation pressures sonoporated (i.e., permeabilized the cell membrane) the

targeted human melanoma cells (LU1205) within the cell monolayer, which is grown on

the surface of the UMTA biochip. Furthermore, fluorescent microscopy and image

processing algorithms were employed to quantitatively estimate the degree of cell

membrane permeability and the level of enhanced nanoparticle (i.e., Quantum Dot) uptake by the cells due to sonoporation. In these analyses, theoretical models that

represent the nature of ultrasound–inflicted cell membrane wounds and the mechanism

of mass transport of nanoparticles through the wound were employed. These models

enable successful characterization and analysis of sonoporation–enhanced nanoparticle

delivery into the melanoma cells.

Although the working concept of UMTA–biochip–based configuration was

successfully demonstrated in this study, a few methods used to characterize the quality

and performance of micro–transducers require some modifications and upgrades. In

this work, the ultrasound pressures generated by the micro–transducers were estimated

based on model–based calculations and simulations. These calculated values must be

293 confirmed experimentally. In other words, the ultrasonic pressures and intensities generated by the micro–transducers must be measured experimentally using hydrophone and the results should be compared with model–based calculations and simulations. In order to carry out this characterization, the ultrasound exposimetry setup employed in this study require major modification since the system lacks high– resolution movement in scanning the hydrophone very close above the UMTA biochip’s surface. This problem can easily be resolved by installing very high–resolution stepper motors, high–resolution–pitch positioning shafts, and position sensors to precisely position and track the hydrophone very close to the top surface of the micro–transducer.

This setup will enable measurement of the actual ultrasound pressures and intensities generated from the micro–transducers, which are almost equal to the pressures and intensities at the end of Near Field. Furthermore, RLC networks must be designed and interfaced between the driving circuit and the micro–transducers in order to minimize the impedance mismatch so that the power transfer is maximized. This RLC network, integrated with the multiplexer or the relay switches, will improve the efficiency in multiplex–driving of individual micro–transducers.

Both experimental and computational results from the cell membrane characterizations at the single–cell level showed that the IMA–based sensor configuration provides high sensitivity and can even resolve the subtle differences in the cell membrane of fibroblast cells (NIH3T3) influenced by different surface chemistries

(i.e., KRGD peptide or fibronectin) being employed on the working microelectrodes.

Moreover, the significance of single–cell–level impedance characterizations in terms of sensitivity and Signal–to–Noise Ratio (SNR) were also highlighted by comparing the response signals from single–glioblastoma cell–electrode heterostructure to those of multi–glioblastoma cell–electrode heterostructure, which were transduced via IMA– based sensor configuration, upon exposure to toxins (e.g., chlorotoxin). All of these results render many potential applications of IMA chips in cellular/cell–based sensing.

294 Nevertheless, in the single–cell–level cell membrane characterization study using IMA chips, only one cell line, i.e., NIH3T3 (fibroblast cell lines), was used as a sensing element to demonstrate the feasibility of IMA chips in cellular/cell–based sensing. In order to obtain a complete evaluation on the sensitivity of IMA sensors, impedance spectroscopy measurements of various cell lines, which possess different densities of transmembrane ion channels, should be carried out. The impedance measurement

results should be compared with those obtained by traditional patch–clamp techniques.

Depending on the comparison result, the signal transduction and data acquisition

circuitry should be optimized in terms of designs, materials, and/or electronic

components so that such subtle difference in cell membrane properties can be

ambiguously resolved by IMA sensors, offering the highest sensitivity possible.

Furthermore, experimental parameters such as probing frequencies and the degree of

cell spreading and adhesion are also important in optimizing the sensitivity of IMA senor

(or maximizing SNR). Calculations based on the ECM model revealed that there was the

optimum probing frequency bandwidth of IMA sensor configuration where the highest

SNR could be achieved. However, this frequency bandwidth could be shifted if a

different cell line, which has a different density of ion channels in the membrane, was

employed. Moreover, degrees of cell adhesion on a particular type of surface chemistry

could be different for various cell lines. According to the studies on the influence of cell

adhesion and spreading on the IMA sensor’s SNR, differences in degrees of cell adhesion

of various cell lines could affect the sensitivity of IMA sensor configuration. Different

peptides or proteins had to be employed in surface modification methods if the adhesion

of the cell line of interest was poor on the peptides being used. Therefore, impedance

spectroscopy measurements of cell–electrode heterostructures formed with various cell

lines should be carried out in order to evaluate and optimize the IMA sensor response.

Furthermore, in order to evaluate the full potential of IMA chips, real time

monitoring of drug–induced changes such as cellular apoptosis, necrosis, and ion

295 conductivity variations in the cell membrane should be carried out at the single–cell level

using IMA sensing platforms. In these characterization experiments, theoretical models

with parameters, which represent the nature and mechanism of such changes, are

required and, thus, they should be developed for interpretation and analysis of IMA

sensor response. By employing the suitable models, important information regarding

drug–binding kinetics and dose–dependent cell responses to drugs (even at the single cell level) could be retrieved from the transduced signals. Such investigations will demonstrate the full potential of IMA chips and will convince many researchers and scientists in biopharmaceutical and biotechnology industries that IMA sensing platforms are extremely powerful, versatile, and robust technology for cellular/cell–based drug screening.

In the site–specific sonoporation experiment using the UMTA biochip, only one type of cell line (i.e., human melanoma cells (LU1205)) was used for demonstration and study of sonoporation–enhanced nanoparticle delivery into the cells. However, the threshold

ultrasound pressure required to permeabilize the cell membrane and the mean effective

lifetime of ultrasound–inflicted cell membrane wound could be different if a different

cancer cell line was used. Furthermore, cell viabilities of different cancer cell lines after

sonoporation at a specific ultrasound pressure could also be different. Such varying

properties could result in different level of enhanced nanoparticle uptake in various

cancer cell lines after sonoporation. Therefore, in order to complete the study on the

sonoporation–enhanced nanoparticle delivery into the cells, site–specific sonoporation

of various cancer cell lines should be carried out in the future studies. Additionally, only

one type of quantum dot (carboxylic–acid–derivatized CdSe/ZnS QD) was employed as

an optical probe in the characterization of the mass transfer of nanoparticle through the

ultrasound–inflicted cell membrane wound. In this characterization, the amount of QDs

(estimated from the fluorescence emissions of QDs) transported into the cells was used

296 to access the degree of cell membrane permeabilization after the site–specific

sonoporation. Since the QD diffusion through the wound is also dependent on the size of

QDs, characterizations with various sizes of QDs should also be carried out in the future

studies. In these characterizations, Design of Experiments (DoE) should be developed

and implemented to investigate the QD uptake of a particular cancer cell line for

different sizes of QDs as well as different post–sonoporation lag times after which QDs

are added into the tissue culture media. Not only these characterizations will provide the

information regarding the QD–size–dependent enhanced cellular uptake and mass transfer through the membrane wound, they also determine the largest size of nanoparticles that can be transported through the sieve–like wound by passive diffusion

after sonoporation. Moreover, adding QDs (of same size) after different post– sonoporation lag times (note: experiments must be conducted separately) and characterizing the resulting intracellular QD concentration using proper mass–transport

models can give information regarding the characteristic diffusion time of a particular

QD for a particular cancer cell line (or the mean effective wound lifetime of a particular

cancer cell line). Therefore, future studies should be emphasized on these

characterizations, which will provide further insight into the sonoporation–enhanced nanoparticle delivery into the cancer cells. Also, the dynamics of sonoporation– enhanced QD uptake to that of endocytosis–driven QD intake were contrasted in this study. It was found that the sonoporation enhanced the uptake of QDs by the cells significantly. However, it was speculated that the calculated enhancement factors could vary from one cell line to another cell line as well as from one type of nanoparticle to another since the rate and mechanism of endocytosis process could differ depending on the cell line and the type of nanoparticle being used. Therefore, again, a set of characterizations with the aid of careful Design of Experiments (DoE) should be conducted in the future studies in order to complete the comparison study.

297 In addition, the site–specific sonoporation study also led to the conclusion that the ultrasound radiation pressures stretch the cell membrane of adherent cells and when the exerting pressure is equal to or above the threshold pressure, the cell membrane is ruptured resulting membrane wounds or openings, which are transient and actively re– sealed. Owing to the device design and architecture as well as transducer’s miniature size, the resulting site–specific sonoporation showed high degree of spatial specificity and resolution (i.e., only targeted cells within the cell monolayer were sonoporated).

With the aid of theoretical models for ultrasound–inflicted cell membrane wound and for mass transport through the wound openings, ultrasound–pressure–dependent cell membrane wound size and QD–uptake were also investigated. The results showed that the higher the ultrasound pressure/intensity, the higher was the level of enhanced QD uptake due to larger degree of cell membrane permeability (or larger effective wound size). In this investigation, the effective wound sizes for different ultrasound radiation pressures were calculated from the experimental data using equations derived from the models developed. Therefore, in order to observe the true nature and dynamics of cell membrane wound, in situ monitoring of the cell membrane during and/or after the site– specific sonoporation should be carried out. Current techniques being used to observe

the cell membrane wounds, which include Scanning Electron Microscopy (SEM),

Transmission Electron Microscopy (TEM), and fluorescence microscopy, require post–

processing (such as sample washing and fixing). Furthermore, fluorescent dyes and

nanoparticles employed are usually invasive for the cells under study. The invasive

nature of these characterizations as well as sample preparation procedures could become

sources of experimental artifacts in characterizing the wound size and wound re–sealing

dynamics. Therefore, the techniques that are less invasive and require little or no post–

processing should be employed to investigate the dynamics of cell membrane during or

after the site–specific sonoporation. For instance, high resolution scanning acoustic

microscopy or atomic force microscopy (using biological AFM probes for liquid

environments) could be employed for imaging of the cell membrane.

298 For long term prospects and impacts, successful cell membrane characterization and permeabilization as well as successful monitoring of the changes in the cell–electrode heterostructure at the single–level upon exposure to a toxic chemical (i.e., chlorotoxin) using micro–fabricated biochips have rendered huge potentials in many applications such as biosensing, biochemical assay, and transfection assay at multi– or single–cell level screenings and characterizations. The novel design and architecture of these chips bring efficiency, improve throughputs, offer multiplexing capability, provide robust analysis, and overcome the technological challenges and limitations encountered by many scientists in biopharmaceutical and biotechnology research. For instance, IMA chips can be implemented in cell–based biosensing applications as efficient and sensitive biosensors and UMTA biochips can be implemented in drug or gene product assay platforms as delivery–enhancement sub–systems for improved screening efficiency.

Therefore, not only does the research work conducted in this study have a huge impact on the biochip industries, it also brings benefits to the biomedical research.

Furthermore, by integrating with state–of–the–art instrumentations, other micro–

/nano–devices such as Micro–Electro–Mechanical Systems (MEMS) and micro–/nano– fluidic components, and/or micro–array transfer printing systems, the biochips used in this study can be implemented as sub–systems in next–generation state–of–the–art cellular and cell–based assay platforms. These platforms will offer high efficiency with multiplexing capabilities and high screening throughputs for many biochemical assays and drug discovery research. With the aid of suitable theoretical, analytical, and computational analysis, not only do these assay platforms bring efficiency and throughputs, they also become very powerful, versatile, and robust characterization and screening tools for biotechnology and pharmaceutical industries. Moreover, the microfabrication processes and device characterization techniques developed in this study only require minor modifications/optimizations and the technology can easily be transferred to the industries for large–scale manufacturing. Therefore, it is foreseen that huge market exists in the near future for the biochips developed in this work.

299 NONTECHNICAL ABSTRACT

The plasma membrane of living cells imposes a physical barrier to most polar and large molecules. Permeability of the plasma membrane to polar molecules such as ions mainly depends on the number and the state (either open or closed) of ion channels present in the membrane. Large molecules such as food particles or drugs are usually absorbed into the cell by a slow process known as endocytosis. Varying the cell membrane permeability to ions alters the cellular state (either dead or alive) and activities. In addition, by temporarily breaking or tearing the cell membrane, the transport of large molecules into the cell can be enhanced. In assays of drugs that specifically affect the state of ion channels, the efficacy of drugs can be determined by monitoring the changes in permeability of the cell membrane to the particular ion of interest and subsequent cellular responses. In these assays, measurements are usually made on the cell population. However, monitoring the cellular responses at the single–cell level upon exposure to drugs offers unique advantage. It rules out the influence from cell–to–cell interactions and is suitable for stochastic model–based assays. Current single–cell–level techniques such as patch–clamping are inefficient and has very low throughputs. Successful non–viral gene therapy relies on efficient delivery techniques that transport gene products (they are large molecules) into the cells. In non–viral transfection and gene delivery, the cell membrane needs to be broken or torn temporarily to enhance the delivery of gene products. New enhanced–delivery approaches are being needed in order to improve the efficiency and throughput in the studies of the genetic events after gene products are transported into the living cells, i.e., transfection assays. Biochips, usually consist of arrays of isolated small environments for living cells, are important emerging tools and enable researchers in biology and medicine to carry out cost– effective and high–throughput experiments and assays. They are also portable for on–site experiments and are robust. In this study, two biochips, Integrated Microelectrode Array (IMA) biochip and Ultrasonic Micro–Transducer Array (UMTA) biochip, were designed and fabricated. The IMA biochip employs cell–size microelectrode array to measure the permeability of the cell membrane to ions at the single–cell level. Individual mouse fibroblasts were seeded on the microelectrodes and electrical impedance of the cell membrane, which represents the membrane permeability to ions were measured. The IMA biochip architecture has potentials in single–cell level high through–put cell–based sensing and assays. The UMTA biochip consists of arrays of piezoelectric micro–transducers employed to temporarily break or tear the cell membrane at the cellular level. Ultrasonic waves, generated from micro–transducers, temporarily tore the cell membrane of the human melanoama cells seeded very close to the transducers. The enhanced transport of large molecules into the cells after temporarily tearing the cell membrane was confirmed by tracking the transport of quantum dots, which were used as biocompatible optical probes. The UMTA biochips have high potentials in non–vial transfection and gene therapy assays at the cellular level as well as in drug screenings as sub–systems for enhancing the delivery efficiency.

300 MYO THEIN C URRICULUM V ITAE

Home Address [email protected] College Address 348 Blue Course Dr. 212 Earth–Engineering Sciences Bldg. Bldg. 10 Apt. 119 University Park, PA 16802 State College, PA 16803 Cellular: (814) 232–0234

EDUCATION

THE PENNSYLVANIA STATE UNIVERSITY, University Park, Pennsylvania, U.S.A. Doctor of Philosophy in Engineering Science and Mechanics (December 2010) Dissertation: Biochips Employing Microelectrodes and Microtransducers for Characterization and Permeabilization of Cell Membrane at the Whole–Cell and Cellular Level Cumulative GPA: 3.71/4.00

Bachelor of Science in Electrical Engineering (May 2005) Bachelor of Science in Engineering Science with Honors (May 2005) Honor Thesis: Modification of the Selectivity of Gold–Nanowire–Based Mercury Sensors by Self–Assembled– Monolayer Functionalization Cumulative GPA: 3.60/4.00

PUBLICATIONS

Thein, M., Asphahani, F., Cheng, A., Buckmaster, R., Zhang, M., and Xu, J., 2010. Response Characteristic of Single–Cell Impedance Sensors Employed with Surface–Modified Microelectrodes. Biosensors and Bioelectronics. 25 (8), 1963–1969. [doi:10.1016/j.bios.2010.01.023]

Tan, Z., Zhu, T., Thein, M., Gao, S., Cheng, A., Zhang, F., Zhang, C., Su, H., Wang, J., Henderson, R., Hahm, J. –I., Yang, Y., and Xu, J., 2009. Integration of Planar and Bulk Heterojunctions in Polymer/Nanocrystal Hybrid Photovoltaic Cells. Applied Physics Letters. 95, 063510-063513. [doi:10.1063/1.3189083 ]

Buckmaster, R., Asphahani, F., Thein, M., Xu, J., and Zhang, M., 2009. Detection of Drug-induced Cellular Changes Using Confocal Raman Spectroscopy on Patterned Single-cell Biosensors. Analyst. 134, 1440-1446. [doi:10.1039/b900420c]

Asphahani, F., Thein, M., Veiseh, O., Edmondson, D., Kosai, R., Veiseh, M., Xu, J., and Zhang, M., 2008. Influence of Cell Adhesion and Spreading on Impedance Characteristics of Cell-based Sensors. Biosensors and Bioelectronics. 23 (8), 1307- 1313. [doi:10.1016/j.bios.2007.11.021]

HONORS AND AWARDS

 Penn State College of Engineering Fellowship (2005/2006 academic year)  Joseph C. Conway Jr., Memorial Award for outstanding graduate teaching assistant

TEACHING EXPERIENCE

Graduate Teaching Assistant 2006 January – Present Department of Engineering Science and Mechanics The Pennsylvania State University, University Park, Pennsylvania

Graduate Teaching Assistant/Laboratory Demonstrator 2005 June – 2005 December PSU Center for Nanotechnology Education and Utilization The Pennsylvania State University, University Park, Pennsylvania