Biomolecular Films (Surfactant Science Series)

BIOMOLECULAR FILMS Design, Function, and Applications edited by James F. Rusling University of Connecticut Starrs, Connecticut, U.S.A.

MARCEL

MARCEL DEKKER, INC.

NEW YORK • BASEL

Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress. ISBN: 0-8247-0899-7 This book is printed on acid-free paper. Headquarters Marcel Dekker, Inc. 270 Madison Avenue, New York, NY 10016 tel: 212-696-9000; fax: 212-685-4540 Eastern Hemisphere Distribution Marcel Dekker AG Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland tel: 41-61-260-6300; fax: 41-61-260-6333 World Wide Web http://www.dekker.com The publisher offers discounts on this book when ordered in bulk quantities. For more information, write to Special Sales/Professional Marketing at the headquarters address above. Copyright # 2003 by Marcel Dekker, Inc. All Rights Reserved. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming, and recording, or by any information storage and retrieval system, without permission in writing from the publisher. Current printing (last digit): 10 9 8 7 6 5 4 3 2 1 PRINTED IN THE UNITED STATES OF AMERICA

SURFACTANT SCIENCE SERIES

FOUNDING EDITOR

MARTIN J. SCHICK 1918-1998 SERIES EDITOR

ARTHUR T. HUBBARD Santa Barbara Science Project Santa Barbara, California

ADVISORY BOARD

DANIEL BLANKSCHTEIN Department of Chemical Engineering Massachusetts Institute of Technology Cambridge, Massachusetts

ERICW KALER Department of Chemical Engineering University of Delaware Newark, Delaware

S. KARABORNI Shell International Petroleum Company Limited London, England

CLARENCE MILLER Department of Chemical Engineering Rice University Houston, Texas

LISAB QUENCER The Dow Chemical Company Midland, Michigan

DON RUBINGH The Procter & Gamble Company Cincinnati, Ohio

JOHN F SCAMEHORN Institute for Applied Surfactant Research University of Oklahoma Norman, Oklahoma

BEREND SMIT Shell International Oil Products B V Amsterdam, The Netherlands

P SOMASUNDARAN Henry Krumb School of Mines Columbia University New York, New York

JOHN TEXTER Strider Research Corporation Rochester, New York

1. Nonionic Surfactants, edited by Martin J. Schick (see also Volumes 19, 23, and 60) 2. Solvent Properties of Surfactant Solutions, edited by Kozo Shinoda (see Volume 55) 3. Surfactant Biodegradation, R. D. Sw/sAjer(see Volume 18) 4. Cationic Surfactants, edited by Eric Jungermann (see also Volumes 34, 37, and 53) 5. Detergency: Theory and Test Methods (in three parts), edited by W. G. Cutler and R. C. Davis (see also Volume 20) 6. Emulsions and Emulsion Technology (in three parts), edited by Kenneth J. Lissant 7. Anionic Surfactants (in two parts), edited by Warner M. Linfield (see Volume 56) 8. Anionic Surfactants: Chemical Analysis, edited by John Cross 9. Stabilization of Colloidal Dispersions by Polymer Adsorption, Tatsuo Sato and Richard Ruch 10. Anionic Surfactants: Biochemistry, Toxicology, Dermatology, edited by Christian Gloxhuber (see Volume 43) 11. Anionic Surfactants: Physical Chemistry of Surfactant Action, edited by E. H. Lucassen-Reynders 12. Amphoteric Surfactants, edited by B. R. Bluestein and Clifford L Hilton (see Volume 59) 13. Demulsification: Industrial Applications, Kenneth J. Lissant 14. Surfactants in Textile Processing, Arved Datyner 15. Electrical Phenomena at Interfaces: Fundamentals, Measurements, and Applications, edited byAyao Kitahara andAkira Watanabe 16. Surfactants in Cosmetics, edited by Martin M. Rieger (see Volume 68) 17. Interfacial Phenomena: Equilibrium and Dynamic Effects, Clarence A. Miller and P. Neogi 18. Surfactant Biodegradation: Second Edition, Revised and Expanded, R. D. Swisher 19. Nonionic Surfactants: Chemical Analysis, edited by John Cross 20. Detergency: Theory and Technology, edited by W. Gale Cutler and Erik Kissa 21. Interfacial Phenomena in Apolar Media, edited by Hans-Friedrich Eicke and Geoffrey D. Parfitt 22. Surfactant Solutions: New Methods of Investigation, edited by Raoul Zana 23. Nonionic Surfactants: Physical Chemistry, edited by Martin J. Schick 24. Microemulsion Systems, edited by Henri L Rosano and Marc Clausse 25. Biosurfactants and Biotechnology, edited by Nairn Kosaric, W. L Cairns, and Neil C. C. Gray 26. Surfactants in Emerging Technologies, edited by Milton J. Rosen 27. Reagents in Mineral Technology, edited by P. Somasundaran and Brij M. Moudgil 28. Surfactants in Chemical/Process Engineering, edited by Darsh T. Wasan, Martin E. Ginn, and Dinesh O. Shah 29. Thin Liquid Films, edited by I. B. Ivanov 30. Microemulsions and Related Systems: Formulation, Solvency, and Physical Properties, edited by Maurice Bourrel and Robert S. Schechter 31. Crystallization and Polymorphism of Fats and Fatty Acids, edited by Nissim Garti and Kiyotaka Sato

32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Interfacial Phenomena in Coal Technology, edited by Gregory D Botsaris and Yuli M Glazman Surfactant-Based Separation Processes, edited by John F Scamehorn and Jeffrey H Harwell Cationic Surfactants Organic Chemistry, edited by James M Richmond Alkylene Oxides and Their Polymers, F E Bailey, Jr, and Joseph V Koleske Interfacial Phenomena in Petroleum Recovery, edited by Norman R Morrow Cationic Surfactants Physical Chemistry, edited by Donn N Rubmgh and PaulM Holland Kinetics and Catalysis in Microheterogeneous Systems, edited by M Gratzel and K Kalyanasundaram Interfacial Phenomena in Biological Systems, edited by Max Bender Analysis of Surfactants, Thomas M Schmitt (see Volume 96) Light Scattering by Liquid Surfaces and Complementary Techniques, edited by Dominique Langevin Polymeric Surfactants, Irja Piirma Anionic Surfactants Biochemistry, Toxicology, Dermatology Second Edition, Revised and Expanded, edited by Christian Gloxhuberand Klaus Kunstler Organized Solutions Surfactants in Science and Technology, edited by Stig E Fnberg and Bjorn bndman Defoammg Theory and Industrial Applications, edited by P R Garrett Mixed Surfactant Systems, edited by Keizo Ogino and Masahiko Abe Coagulation and Flocculation Theory and Applications, edited by Bohuslav Dobias Biosurfactants Production • Properties • Applications, edited by Nairn Kosanc Wettabihty, edited by John C Berg Fluonnated Surfactants Synthesis • Properties • Applications, Erik Kissa Surface and Colloid Chemistry in Advanced Ceramics Processing, edited by Robert J Pugh and Lennart Bergstrom Technological Applications of Dispersions, edited by Robert B McKay Cationic Surfactants Analytical and Biological Evaluation, edited by John Cross and Edward J Singer Surfactants in Agrochemicals, Tharwat F Tadros Solubihzation in Surfactant Aggregates edited by Shernl D Christian and John F Scamehorn Anionic Surfactants Organic Chemistry, edited by Helmut W Stache Foams Theory, Measurements, and Applications, edited by Robert K Prud'homme and Saad A Khan The Preparation of Dispersions in Liquids, H N Stein Amphotenc Surfactants Second Edition, edited by Eric G Lomax Nonionic Surfactants Polyoxyalkylene Block Copolymers, edited by Vaughn M A/ace Emulsions and Emulsion Stability, edited by Johan Sjoblom Vesicles, edited by Morton Rosoff Applied Surface Thermodynamics, edited by A W Neumann and Jan K Spelt Surfactants in Solution, edited by Arun K Chattopadhyay and K L Mittal Detergents in the Environment, edited by Milan Johann Schwuger

66. Industrial Applications of Microemulsions, edited by Conxita Solans and Hironobu Kunieda 67. Liquid Detergents, edited by Kuo-Yann Lai 68. Surfactants in Cosmetics: Second Edition, Revised and Expanded, edited by Martin M. Rieger and Linda D. Rhein 69. Enzymes in Detergency, edited by Jan H. van Ee, Onno Misset, and Erik J. Baas 70. Structure-Performance Relationships in Surfactants, edited by Kunio Esumi and Minoru Ueno 71. Powdered Detergents, edited by Michael S. Showell 72. Nonionic Surfactants: Organic Chemistry, edited by Nico M. van Os 73. Anionic Surfactants: Analytical Chemistry, Second Edition, Revised and Expanded, edited by John Cross 74. Novel Surfactants: Preparation, Applications, and Biodegradability, edited by Krister Holmberg 75. Biopolymers at Interfaces, edited by Martin Malmsten 76. Electrical Phenomena at Interfaces: Fundamentals, Measurements, and Applications, Second Edition, Revised and Expanded, edited by Hiroyuki Ohshima and Kunio Furusawa 77. Polymer-Surfactant Systems, edited by Jan C. T. Kwak 78. Surfaces of Nanoparticles and Porous Materials, edited by James A. Schwarz and Cristian I. Contescu 79. Surface Chemistry and Electrochemistry of Membranes, edited by Torben Smith S0rensen 80. Interfacial Phenomena in Chromatography, edited by Emile Pefferkom 81. Solid-Liquid Dispersions, Bohuslav Dobias, Xueping Qiu, and Wolfgang von Rybinski 82. Handbook of Detergents, editor in chief: Uri Zoller Part A: Properties, edited by Guy Braze 83. Modern Characterization Methods of Surfactant Systems, edited by Bernard P. Binks 84. Dispersions: Characterization, Testing, and Measurement, Erik Kissa 85. Interfacial Forces and Fields: Theory and Applications, edited by Jyh-Ping Hsu 86. Silicone Surfactants, edited by Randal M. Hill 87. Surface Characterization Methods: Principles, Techniques, and Applications, edited by Andrew J. Milling 88. Interfacial Dynamics, edited by Nikola Kallay 89. Computational Methods in Surface and Colloid Science, edited by Matgorzata Borowko 90. Adsorption on Silica Surfaces, edited by Eugene Papirer 91. Nonionic Surfactants: Alkyl Polyglucosides, edited by Dieter Balzer and Harald Luders 92. Fine Particles: Synthesis, Characterization, and Mechanisms of Growth, edited by Tadao Sugimoto 93. Thermal Behavior of Dispersed Systems, edited by Nissim Garti 94. Surface Characteristics of Fibers and Textiles, edited by Christopher M. Pastore and Paul Kiekens 95. Liquid Interfaces in Chemical, Biological, and Pharmaceutical Applications, edited by Alexander G. Volkov

96 Analysis of Surfactants Second Edition, Revised and Expanded, Thomas M Schmitt 97 Fluormated Surfactants and Repellents Second Edition, Revised and Expanded, Erik Kissa 98 Detergency of Specialty Surfactants, edited by Floyd E Fnedli 99 Physical Chemistry of Polyelectrolytes, ecWed by Tsetska Radeva 00 Reactions and Synthesis in Surfactant Systems, edited by John Texter 01 Protein-Based Surfactants Synthesis, Physicochemical Properties, and Applications, edited by Ifendu A Nnanna and Jiding Xia 02 Chemical Properties of Material Surfaces, Marek Kosmulski 03 Oxide Surfaces, edited by James A Wingrave 104 Polymers in Particulate Systems Properties and Applications edited by Vincent A Hackley, P Somasundaran, and Jennifer A Lewis 105 Colloid and Surface Properties of Clays and Related Minerals, Rossman F Giese and Care/ J van Oss 106 Interfacial Electrokinetics and Electrophoresis, edited by Angel V Delgado 107 Adsorption Theory, Modeling, and Analysis, edited by Jozsef Toth 108 Interfacial Applications in Environmental Engineering, edited by Mark A Keane 109 Adsorption and Aggregation of Surfactants in Solution, eoMed by K L Mittal and Dinesh O Shah 110 Biopolymers at Interfaces Second Edition, Revised and Expanded, edited by Martin Malmsten 111 Biomolecular Films Design, Function and Applications, edited by James F Rusling 112 Structure-Performance Relationships in Surfactants Second Edition, Revised and Expanded, edited by Kunio Esumi and Mmoru Ueno

ADDITIONAL VOLUMES IN PREPARATION Liquid Interfacial Systems Oscillations and Instability, Rudolph V Binkh, Vladimir A Bnskman, Manuel G Velarde, and Jean-Claude Legros Novel Surfactants Preparation, Applications, and Biodegradabihty Second Edition, Revised and Expanded, edited by Krister Holmberg Colloidal Polymers Preparation and Biomedical Applications, edited by Abdelhamid Elaissan

Preface

Modifying solid interfaces with purposely designed supramolecular structures has become one of the most vibrant and dynamic areas in the physical sciences, engineering, and biology. Films of biomolecules can be used to tailor the properties of interfaces, providing solid surfaces that can catalyze enzyme reactions, serve in biosensors and as biorecognition elements, mediate nanoparticle formation, and provide a basis for fundamental biochemical and biophysical studies. Present and future applications in biomedicine and biomedical devices abound. This volume provides a modern collection of unique biomolecular film methodologies and presents techniques to investigate their properties. These activities go hand in hand. Traditional methods of chemical analysis such as various spectroscopies, nuclear magnetic resonance (NMR), and mass spectrometry need to be modified or redesigned altogether to address the special features of ultrathin biomolecular films. Reliable characterization methods are of critical importance to connect structure with the function of the films. The chapters in this book can be divided into four overlapping categories: 1) types and construction of biomolecular films, 2) methods of analyzing biomolecular films, 3) systems designed for chemical synthesis, and 4) design and function of biosensors. The first two categories are the most overlapped. Chapter 1 is from my lab and discusses stable multilayer protein and DNA films for electrochemical applications. Chapter 2, by iii

iv

Preface

Katsumi Niki and Brian Gregory, describes electrochemically active monolayer films of adsorbed proteins on self-assembled monolayers on electrode surfaces. In Chapter 3, Anne Plant and coworkers describe biomimetic lipid bilayer films on metals, often called supported bilayers, and many ways to characterize them. Paul Bohn discusses peptide- and protein-based biomolecular assemblies along with optical and other analysis methods in Chapter 4. Nongjian Tao and coworkers present a lucid treatise on the application of surface plasmon resonance spectroscopy to protein adsorption and electrochemistry in Chapter 5. Patrick Unwin’s group from the University of Warwick in England describes in Chapter 6 the use of the relatively new method of scanning electrochemical microscopy applied to samples ranging from monolayers to biological cell surfaces to tissues. In Chapter 7, Ernesto Calvo’s group from Argentina describes applications of quartz crystal microbalance frequency analysis to layered protein films. Katsuhiko Ariga and coworkers from Japan describe films designed for enzyme reactors in Chapter 8. Continuing with the synthetic theme, Janos Fendler of Clarkson University describes nanoparticle synthesis using biomimetic films in Chapter 9. The last three chapters cover applications to biosensors. Chapter 10, by Anthony Killard and Malcolm Smyth of Dublin City University, provides a broad overview of biosensors. In Chapter 11, Evelyne Simon and Philip Bartlett of the University of Southampton in England discuss films on electrodes designed for the important enzyme cofactor NADH, which can be employed for sensors designed with enzymes that use or generate NADH. Finally, Chapter 12, by Michael Tarlov and Adam Steel, describes the important and relatively new area of DNA sensors, which may become clinically important in the near future. We hope that these chapters together provide a broad, informative overview of the state of the art in research in biomolecular films.

ACKNOWLEDGMENTS I acknowledge the understanding and forbearance of my family, colleagues, and students during the editing and writing of this book. Special thanks are due to my wife, Penelope Williams, for her patience, kindness, and constant support. I am also indebted to the editorial staff of Marcel Dekker, Inc., for their excellent work on this project, and to Professor Arthur Hubbard for encouragement. The lion’s share of the book editing and the writing of Chapter 1 were completed while I was on sabbatical leave at the Irish National Center for Sensor Research (NCSR), School of Chemical Sciences, Dublin City

Preface

v

University (DCU). Special thanks go to Professor Malcolm Smyth, Dean and Chemical Sciences Head at DCU, for his stellar support and cheerful friendship during our stay in Dublin. Thanks also go to DCU faculty members Robert Forster and Han Vos for research collaboration and discussions that counterbalanced the sometimes tedious editing and writing duties, and to Dermot Diamond for excellent fiddle playing on many occasions. My wife and I also thank the entire staff and graduate/postdoctoral community of NCSR for making our stay in Dublin a very pleasant, friendly, and professionally satisfying experience. Kathleen Grennan deserves special mention for her friendship, and for organizing and participating in several weekend jaunts to her home county of Kilkenny, where I enjoyed playing traditional Irish tunes on an accordion with the friendly musicians at Mannion’s in New Ross. I would also be remiss not to mention the contributions of important local meeting places such as the Slipper and the Gravediggers near DCU, where the developing project and its intricacies were discussed with colleagues in a relaxed atmosphere. James F. Rusling

Contents

Preface Contributors

iii ix

1.

Designing Functional Biomolecular Films on Electrodes James F. Rusling and Zhe Zhang

2.

Electrochemistry of Redox-Active Protein Films Immobilized on Self-Assembled Monolayers of Organothiols Katsumi Niki and Brian W. Gregory

3.

Biomimetic Membranes on Metal Supports John T. Elliott, Curtis W. Meuse, Vitalii Silin, Susan Krueger, John T. Woodward, Teresa Petralli-Mallow, and Anne L. Plant

4.

PeptidePhysical Optimal Paul W.

and Protein-Based Biomolecular Assemblies: and Chemical Characterization for Function Bohn

1

65 99

163

vii

viii

5.

6.

7.

Contents

Surface Plasmon Resonance Spectroscopy: Applications in Protein Adsorption and Electrochemistry Shaopeng Wang, Salah Boussaad, and Nongjian J. Tao

213

Characterization of Biomolecular Interfaces with Scanning Electrochemical Microscopy: From Model Monolayers to Tissues and Cells Anna L. Barker, Catherine E. Gardner, Julie V. Macpherson, Patrick R. Unwin, and Jie Zhang

253

Layered Protein Films: Quartz Crystal Resonator Frequency and Admittance Analysis Ernesto J. Calvo, Claudia Danilowicz, Erica Forzani, Alejandro Wolosiuk, and Marcelo Otero

337

8.

Nano-Sized Thin Films for Enzyme Reactors Katsuhiko Ariga, Yoshihiro Sasaki, and Jun-ichi Kikuchi

381

9.

Biomimetic Nanoparticle Synthesis Janos H. Fendler

427

10.

Biosensors Anthony J. Killard and Malcolm R. Smyth

451

11.

Modified Electrodes for the Oxidation of NADH Evelyne Simon and Philip N. Bartlett

499

12.

DNA-Based Sensors Michael J. Tarlov and Adam B. Steel

545

Index

609

Contributors

Katsuhiko Ariga, Ph.D. ERATO Nanospace Project, Japan Science and Technology Corporation (JST), Tokyo, Japan Anna L. Barker, M.Chem., Ph.D. Department of Chemistry, University of Warwick, Coventry, United Kingdom Philip N. Bartlett, Ph.D. Department of Chemistry, University of Southampton, Southampton, United Kingdom Paul W. Bohn, Ph.D. Department of Chemistry and Beckman Institute for Advanced Science and Technology, University of Illinois at UrbanaChampaign, Urbana, Illinois, U.S.A. Salah Boussaad Department of Electrical Engineering, Arizona State University, Tempe, Arizona, U.S.A. Ernesto J. Calvo, Ph.D. INQUIMAE (Department of Inorganic, Analytical and Physical Chemistry), University of Buenos Aires, Buenos Aires, Argentina

ix




 


   
                !"  !"         !      "  #    $ % &"    '#''           *     + <* '#''                     !"  !"  
   
        = +*    >     
     #       '#'' ! "#  % "  #     #   "  #  $ #$    ?  $   " $% &   &     @' % +   #    # "&   

"&   ' "    " X     "  #    $ % &"    '#''  ( )  '         = +*    >
    !      "  #    $ % &"    '#''



   Z     [      &   " '#''

   Z     #    !   &   '#''

 




" *# 
     #       '#''  +                 !"  !"    !

 !      "  #    $ % &"    '#'' $ /   !      "  #    $ % &"    '#''   1         " # " '#'' 5  '#  % "  #     #   "  #  $ #$    ?  ( 6 '  !      "  # \    $ % &"    '#'' ;  '         #"   #"    >   1 '  '  '%<=& 6
6 1'


 #    # "&    "&   $  '  "    %] ' % \ &"    '#'' * >          ^ #    $  ^  '#''    ;    "     #  $ @ &     "  #    $ % &"    '#''



   Z      $ !*  "   "  $        '#''

xii

Contributors

Patrick R. Unwin, B.Sc., M.A., D.Phil. Department of Chemistry, University of Warwick, Coventry, United Kingdom Shaopeng Wang Department of Electrical Engineering, Arizona State University, Tempe, Arizona, U.S.A. Alejandro Wolosiuk INQUIMAE (Department of Inorganic, Analytical and Physical Chemistry), University of Buenos Aires, Buenos Aires, Argentina John T. Woodward, Ph.D. Biotechnology Division, National Institute of Standards and Technology, Gaithersburg, Maryland, U.S.A. Jie Zhang, B.Sc., Ph.D. Department of Chemistry, University of Warwick, Coventry, United Kingdom Zhe Zhang, Ph.D. Research and Development, Nature’s Sunshine Products, Spanish Fork, Utah, U.S.A.

1 Designing Functional Biomolecular Films on Electrodes James F. Rusling University of Connecticut, Storrs, Connecticut, U.S.A.

Zhe Zhang Nature’s Sunshine Products, Spanish Fork, Utah, U.S.A.

1

INTRODUCTION

One of the major functions of highly developed biomolecular superstructures in living systems is to shuttle electrons to sites where they are required to support life processes. Nature has learned how to do this very e⁄ciently, often by designing supramolecular structures uniquely suited to the task. One example is the reaction center (RC) of the purple bacteria Rhodobacter sphaeroides, an integral membrane protein complex comprised of three protein subunits with multiple bound redox cofactors that are arranged to convert light into a transmembrane electrical potential [13]. This takes place via multiple electron transfers to drive photosynthesis and provide us with food and oxygen. Other examples include the membranebound liver cytochrome P450 (cyt P450) enzymes. One or more reductase enzymes bind to cyt P450s on membrane surfaces to deliver electrons from NAD(P)H for oxidative metabolism of lipophilic pollutans and drugs [4,5]. These examples illustrate the integral use of membranes by living systems for organizing supramolecular structures [6] that are utilized in electron transfer events. Such systems can be used to guide the design of ¢lms for arti¢cial

1

2

Rusling and Zhang

devices aimed at biomedical sensing as well as bioreactors for catalytic chemical synthesis. The combination of electrochemistry with ¢lms containing redox proteins and enzymes can facilitate driving biological electron transfer events with an electronic source or sink of electrons, by using a potentiostat to control the applied voltage. Indeed, direct exchange of electrons between native redox enzymes and electrodes has been a long-standing research goal [7]. Designing ¢lms that facilitate direct electron transfer is a maor focus of this chapter. Such ¢lms enable electrochemical studies of the redox chemistry of enzymes, may allow electrodes to replace natural redox partners, and simplify the design of biosensors. Over the past decade, a variety of ¢lm technologies have been developed to facilitate direct electron exchange between electrodes and proteins [7]. Yet there are many important enzymes, including those used in blood glucose sensors, for which direct electron transfer has not been achieved or is too slow. In such cases, small molecule or polymer mediators are used to deliver electrons between electrodes and enzymes [8]. This approach is of great practical importance, and will be treated in Chapter 10. The present chapter deals mainly with ¢lms that facilitate direct protein electron transfer without mediation. Genes are supramolecular aggregates of DNA and proteins. Hybridization of single strands of DNA with oligonucleotides of known sequences promises great utility in diagnosing genetic diseases [911], and is covered in Chapter 12. Applications of DNA ¢lms to damage analysis will be addressed in the present chapter. Electrochemical detection of damage using DNA ¢lms on electrodes could eventually be used as an in vitro measure of chemical toxicity, providing rapid toxicity screens for new chemicals [13]. Thus, this chapter highlights modern options for constructing biomolecular ¢lms for electrochemical biosensor and bioreactor development as well as for fundamental studies. It is not meant as a comprehensive review of biomolecular ¢lm electrochemistry, and so we have exercised our judgement in choosing methods and examples from the literature. We apologize in advance for the many ¢ne papers we have not been able to cite herein. To summarize what follows, the next section in this chapter presents capsule histories of protein and DNA electrochemistry, highlighting seminal discoveries that form the basis of modern progress in biomolecular ¢lm electrochemistry. Section 3 provides a brief overview of the various types of ¢lm methodologies available. Section 4 discusses the fundamentals of thin-¢lm voltammetry. Sections 58 describe various approaches to ¢lm construction using materials such as polyions, lipids, surfactants, and metal oxide nanoparticles. Section 9 describes progress in electrochemical DNA damage

Designing Functional Biomolecular Films

3

detection based on DNA ¢lms. Finally, in Sec.10 we attempt to predict future research directions for this area.

2 2.1

RECENT HISTORY OF BIOMOLECULE FILM ELECTROCHEMISTRY Proteins

Construction of protein ¢lms featuring direct electron transfer for electrochemical studies, biosensors, and bioreactors is a relatively recent development. Voltammetry of proteins prior to the mid-1970s was a di⁄cult endeavor, usually attempted with dissolved species, and often plagued by adsorption and denaturation of biomacromolecules on electrode surfaces [14,15] and irreversible electrode reactions. Beginning in the late 1970s, landmark papers began to appear suggesting that control of electrode surface chemistry and protein purity was the key to obtaining reversible protein electrochemistry. Yeh and Kuwana found chemically reversible (i.e., quasireversible) cyclic voltammetry for horse cytochrome c (cyt c) on tin-doped indium oxide electrodes [16]. Eddowes and Hill reported chemically reversible voltammetry on gold coated with a monolayer of 4,40 -bipyridyl [17]. Hawkridge found that cyt c puri¢ed immediately before experiment gave quasi-reversible cyclic voltammograms on several types of electrodes [18,19]. Armstrong, Hill, and Oliver used edge-plane pyrolytic graphite electrodes containing a high population of carboxylate groups to obtain quasi-reversible cyclic voltammograms for cyt c [20]. During this period, researchers had yet to achieve direct electron transfer between electrodes and native proteins in stable surface ¢lms. Concurrent work on electrochemical biosensors in the 1970s and 1980s was aimed at immobilizing enzymes on electrodes in polymeric and other sorts of ¢lms [2124]. Amperometric biosensors were being developed with mediators to deliver electrons between electrodes and enzymes.While direct electron transfer was not a general feature, this research demonstrated that enzymes could be immobilized in ¢lms and retain high catalytic activity. Protein ¢lm voltammetry really got its start in the late 1980s and early 1990s. Key progress in this period involved the demonstration by Yokota et al. of direct, reversible voltammetry for cyt c on a phosphatidylcholine Langmuir-Blodgett ¢lm transferred onto an indium tin oxide (ITO) electrode [25]. Salamon and Tollin used adsorbed phosphatidylcholine ¢lms to reduce cyt c quasi-reversibly [26,27]. Armstrong’s ¢lm voltammetry, co-adsorbing proteins with aminocyclitols and polymixins to give monolayers with highly reversible voltammetry on edge-plane pyrolytic graphite electrodes, began in this period [28]. Bowden showed that

4

Rusling and Zhang

self-assembled monolayers of alkylthiolcarboxylates on gold electrodes adsorb monolayers of positively charged cyt c and give reversible electrochemistry [29]. These research e¡orts represent the beginnings of direct thin ¢lm voltammetry techniques for proteins, which we feel have evolved into methods of choice for fundamental electrochemical studies of protein redox chemistry. The various ¢lm approaches that have been developed are described in the later sections of this chapter. 2.2

Nucleic Acids

Biosensors made with nucleic acids immobilized on electrodes constitute a very active modern research area, as illustrated by recent reviews [912] and selected examples [3033]. The intrinsic electroactivity of DNA bases, for example, the oxidation of guanine and adenine on polynucleotide chains, provides one basis for electroanalysis. Sections on DNA in this chapter focus on ¢lms for the analysis of DNA damage. DNA hybridiation biosensors are treated in Chapter 12. Damage to DNA by chemicals and their metabolites is a major toxicity pathway (3437). Cytochrome P450 enzymes in mammalian liver catalyze the oxidation of lipophilic pollutants and drugs [4,5] to metabolites which can often react with and damage DNA, frequently forming covalent adducts with guanine and adenine bases. Electrochemical methods based on DNA ¢lms are being explored to detect chemical DNA damage from these metabolites as a basis for in-vitro toxicity screening of new chemicals. Prior to the 1990s, electrochemical analysis of DNA most often employed adsorption of the nucleic acids from solutions onto mercury electrodes [38]. Single-stranded (ss) DNA was found to be much more readily oxidized at guanine and adenine base sites than double-stranded (ds)DNA [39]. The ds-DNA structure does not allow ready access of the bases to the electrode, but unwinding of the double helix provides better access and faster reaction rates. Adsorptive voltammetry on mercury electrodes has been applied to detecting DNA damage from strong acid [40], methylating agents [41], and hydroxyl radicals [42,43]. Damage to DNA from ionizing radiation was detected with mercury electrodes by adsorptive linear sweep [44] and AC voltammetry [45]. In these methods, the DNA is present in solution and accumulated on the electrodes by adsorption prior to analysis. Mercury drop electrodes are probably unsuitable choices for widespread DNA damage or toxicity screening because of the safety issues in handling this toxic liquid. However, guanine and adenine bases in nucleic acids are readily oxidized on solid electrodes [46]. Sections below deal with immobilization of DNA ¢lms on solid electrodes, and the use of these ¢lms for voltammetric detection of DNA damage directly in the ¢lm.

Designing Functional Biomolecular Films

3

5

SUMMARY OF MODERN APPROACHES TO FILM DESIGN

Sections 58 discuss various modern strategies for making biomolecular ¢lms on electrodes. Table 1 summarizes major examples of these techniques, including conceptual drawings and advantages and disadvantages of each approach. 4 4.1

FUNDAMENTALS OF THIN-FILM ELECTROCHEMISTRY Cyclic Voltammetry

Voltammetric methods provide important modern techniques for analyzing electroactive ¢lms on electrode surfaces, as well as for analytical applications. The most popular method for studies of ¢lm electrochemistry is cyclic voltammetry (CV). The electrochemical cell typically includes a working electrode, coated for example with a protein ¢lm, a reference electrode, and a counter electrode (Fig. 1). Under control of the potentiostat, which is in turn under control of the experimenter, the potential (E in volts,V) applied to the cell is ¢rst scanned in a forward direction, stopped at a desired potential, then scanned in a reverse direction,usually returning to its initial value at the end of the experiment.The potential range of the voltammogram is chosen to encompass the reduction and oxidation potentials of the electroactive species in the ¢lm, and the data are output in a graph of current (I) versus potential (E). A peak in the I versus E output denotes a £ow of electrons between the protein in the ¢lm and the underlying electrode, leading to a change in the oxidation state of the protein’s electroactive site.The scan rate n ¼ DE=Dt can be varied from a few tenths of a millivolt per second to a million or more volts per second, providing a wide time-scale range for the investigation of kinetic events in electrode processes. A reversible electrochemical reaction such as O þ ne ¼ R

ð1Þ

is one in which the conversion of oxidation (O) to reductant (R) and of R back to O are both fast on the time scale of the voltammogram, as controlled by the scan rate. A feeling for these time scales can be obtained from the time required for a scan of 1 V, which is the inverse of the scan rate. Thus, a scan rate of 0.1 V=s occurs on the scale of 10 s,while 1000 V=s re£ects a time scale of ca. 1 ms. The ideal, reversible voltammogram from a monolayer of electroactive protein about 510 nm thick on an electrode following Eq. (1) (Fig. 2) features symmetric oxidation and reduction peaks of equal heights with both peak potentials at the formal potential ðE 00 Þ of the surface redox reaction [47,48]. The theoretical separation between oxidation and reduction peaks

6

Rusling and Zhang

TABLE 1 Thumbnail Sketches of Some Types of Biomolecule Films Advantages: Simplicity, only tiny amounts of protein and perhaps coadsorbate needed. Rapid, reversible electron transfer. Excellent for fundamental studies of proteins. Adsorbed protein films on edgeplane pyrolytic graphite (Section 5.1)

Disadvantages: Often lack longterm stability, CVs require accurate background subtraction. Advantages: SAMs easily formed, tiny amount of protein adsorbed from solution. Nearly reversible electron transfer. Excellent for fundamental studies of proteins. Can be applied to thiolated oligonucleotides.

Protein adsorbed on alkythiol-R SAM on Au, where R is opposite charge of protein (Chapter 2)

Disadvantages: May lack long-term stability for enzyme biosensors.

Advantages: Simply mix components and cast. Nearly reversible electron transfer for redox proteins.

Cast films of polyions-protein or DNA-protein on electrode (Section 6.1)

Disadvantages: Low fraction of electroactive protein, sometimes ‘‘leaks’’ proteins.

Advantages: Ease of preparation, nearly reversible electron transfer for redox proteins.

BLM on cleaved metal wire (Section 7.1).

Disadvantages: Proteins in solution may not enter BLM, low singal to background ratio. Limited stability.

Designing Functional Biomolecular Films

7

Advantages: Simply mix components and cast. Nearly reversible electron transfer for redox proteins in ordered mutibilayer films, stable for a month in quiet solutions. Excellent for fundamental studies of proteins.

Cast films of proteins and lipids or surfactants (Section 7.2)

Disadvantages: Slowly removed in hydrodynamic solutions (days).

Advantages: Nearly reversible electron transfer for redox proteins in ordered multibilayer films, improved stability over simple surfactant films.

Disadvantages: Must prepare polyion-surfactant complex before casting. Composite polyion-surfactantprotein films (Section 7.2) Advantages: Nearly reversible electron transfer for redox proteins, nm control of thickness, extended order, versatile architecture, good stability in hydrodynamic fluids. Disadvantages: Many layers require long prep. time (  20 min=layer). Mass transport limitations for catalysis with thicker films. Polyion-protein films grown layerby-layer (Section 8) Advantages: Vectorial electron transfer chains, versatile architecture, good stability. Disadvantages: Multi-step synthesis; may require synthesis of linkers. Covalent bonding to gold (Section 4.2)

8

FIG. 1

Rusling and Zhang

Experimental setup for thin-film cyclic voltammetry.

ðDEp Þ is zero, and the peak width at half peak height is 90.6=n mV at 25 C. The integral under each peak is the charge Q, in coulombs, given by a form of Faraday’s law: ð2Þ Q ¼ nFAGT where GT is the total surface concentration of electroactive protein (mol=cm2), A is electrode area ðcm2 Þ, F is Faraday’s constant (96, 487 C=mol e’s), and n is the numer of electrons transferred in the electrochemical reaction. Thus, measurement of Q by CV provides a direct determination of the amount of electroactive protein in the ¢lm. The formal potential E 00 of the ¢lm redox process is simply the peak potential, or in practice the midpoint potential between the two peaks if there is a small separation between the reduction and oxidation peaks. Figure 2 also illustrates the graphic convention for voltammograms used in this chapter. This is the ‘‘American convention,’’ with reduction peaks positive on the y axis, and more negative potential presented to the right on the x axis. The IUPAC convention used extensively in Europe and

Designing Functional Biomolecular Films

9

FIG. 2 Ideal thin-film cyclic voltammogram of a monolayer of protein on an electrode.

elsewhere has oxidation peaks positive on the y axis, and more positive potential going to the right on the x axis. The ideal, reversible peak current ðIp Þ is n2 F 2 AGT n ð3Þ 4RT where R is the gas constant and T is temperature in Kelvin, showing that Ip increases linearly as scan rate ðnÞ is increased. Equations (2) and (3) are similar to those in thin-layer electrochemistry, in which the electrochemical cell con¢nes a thin layer of solution containing the dissolved electroactive species next to the electrode [49]. Speci¢c interactions between the redox sites are ignored. Thus, this ideal case of ¢lm voltammetry is often called the ‘‘ideal thin-layer electrochemistry’’ model. Additional materials, such as polymers, surfactants, or other coadsorbed molecules are often used to immobilize proteins on electrodes.The e⁄ciency of protein redox conversion then depends on (1) the thermodynamics of redox chemistry in the ¢lm controlled by E 00 ; (2) the kinetics of electron transfer at the ¢lmelectrode interface; (3) the rate of charge transport within the ¢lm, which may depend on counterion entry and exit rates, and electron self-exchange between redox sites [48]; and (4) structural transformations coupled to electron transfer, such as conformational changes. Ip ¼

10

Rusling and Zhang

Films thicker than a monolayer of electroactive protein are often employed to obtain a larger enzyme loading per unit electrode area and consequently larger peak currents. When complete electrolytic conversion throughout these ¢lms is achieved on the CV time scale, the thin-layer electrochemistry model may describe the CVs approximately. Charge transport through these thicker ¢lms may involve physical di¡usion of the proteins through the ¢lm, and=or ‘‘electron hopping’’ between redox centers, utilizing electron self-exchange reactions [50]. Ion exchange with solution and counterion transport within ¢lms is required to provide overall charge balance during redox reactions, and the kinetics of these processes may be important factors in determining the e⁄ciency of charge transport. It is a rare protein ¢lm system that exhibits all the ideal CV characteristics. Shapes of cyclic voltammograms of protein ¢lms are often very different [7,47,48] from ideal thin-layer model predictions (Fig. 2). For example, a nonideal reversible CV for the heme Fe III=FeII redox couple of the KatG catalase-proxidase enzyme from Mycobacterium tuberculosis in a thin ¢lm (ca. 0.5 mm) [51] (Fig. 3) of the lipid dimyristoylphosphatidylcholine (DMPC) shows roughly symmetrical peaks with a ¢nite peak separation of about 55 mV, much grater than the predicted value of zero. In such cases, E 00 is taken as the midpoint potential between the two peaks. These protein ¢lms gave linear plots of Ip versus n in the low scan rate range ð< 1 V=sÞ in accord with Eq. (3). The width of the peaks at half-height is nearly 200 mV, much larger than the predicted ideal value of 90=n mV. Even though background has been partly subtracted in Fig. 3, there remains signi¢cant background current on either side of the peaks. This current results from charging the electrical double layer with possible contributions from surface processes on the electrode surface [49]. This experimental CV is typical of thin-¢lm CVs of redox proteins, although speci¢c details will certainly di¡er and depend on protein and ¢lm properties [7]. Broadening or narrowing of CVpeaks with respect to the ideal 90=n mV width for surface-bound species suggests a breakdown of the model assumptions of no interactions between redox sites in the ¢lm that all have the same E 00. Protein ¢lms are often ‘‘diluted’’ with electrochemicaly inert materials, proteins are on average relatively far apart, and interprotein interactions may be minimized. Consequently, peak widths are usually larger, rather than smaller, than the theoretical 90=n mV. Voltammograms of protein ¢lms have been modeled utilizing the concept of distributions in E 00 and electron-transfer rate constants to account for the peak broadening [5254]. Other factors, including counterion transport e⁄ciency, could also in£uence peak widths, but have not been investigated in detail for protein ¢lms.

Designing Functional Biomolecular Films

11

FIG. 3 Cyclic voltammogram (CV) at 100 mV=s and 25 C of Mycobacterium tuberculosis KatG catalase-peroxidase in a thin dimyristoylphosphatidylcholine film on basal-plane PG electrode, in anaerobic pH 6.0 buffer. Background has been partially substracted. (Reprinted with permission from Ref. 51. Copyright 2001, Royal Society of Chemistry.)

An increasing DEp as the scan rate is increased for an electroactive thin ¢lm suggests kinetic limitations of the electrochemistry [47]. Possible causes include (1) slow electron transfer between the electrode and the protein redox sites, (2) slow transport of charge within the ¢lm limited by electron or counterion transport, (3) uncompensated voltage drop within the ¢lm, and (4) structural reorganization of the ¢lm or protein accompanying the redox reactions. These processes have been investigated in depth for electrodes coated with nonprotein redox polymers [48,50], but much less is known on this topic concerning protein ¢lms. The shapes of CVs will di¡er from the ideal thin-layer case when only partial electrolysis of the electroactive redox sites in the ¢lms occurs during a given linear branch of the scan. This occurs for thicker ¢lms at scan rates high enough to provide insu⁄cient time for complete electrolysis. In this case, only a fraction of the redox sites are electrolyzed during each linear scan, setting up concentraton gradients within the ¢lms. In the limit where only a small fraction of redox sites are electrolyzed, the peak shape is identical to that of a CV on an uncoated electrode for an ion or molecule in

12

Rusling and Zhang

solution under semi-in¢nite linear di¡usion conditions [49]. The resulting asymmetric shape is very di¡erent from that of the ideal CV for a monomolecular ¢lm, and DEp for a reversible di¡usion-controlled system at 25 C is predicted to be 59=n mV.When ¢lms show such‘‘di¡usion controlled’’ CVs, the integral under the peak is no longer proportional to the surface concentration of electroactive centers in the ¢lm, since only a fraction of the protein has been electrolyzed. A typical di¡usion-controlled protein ¢lm CV shape for the heme FeIII=FeII redox couple of myoglobin for a 40-mm-thick ¢lm of DNA (Fig. 4) shows an unsymmetric peak shape, and oxidation and reduction peaks of equal height. However, DEp is only about 40 mV, less than the predicted 59-mV value. The peak current for limiting di¡usion control for an n-electron reaction in a ¢lm is 1=2

Ip ¼ ð2:69  105 Þn3=2 ADct n1=2 Cf

ð4Þ

where the concentration of electroactive species Cf ¼ GT =d, d is ¢lm thickness, and Dct is the charge transport di¡usion coe⁄cient [47]. Equation (4) predicts a linear plot of Ip versus n1=2. This behavior is usually favored by

FIG. 4 Diffusion-controlled CV at 0.1 V=s for 40-mm-thick Mb-DNA film on a PG electrode in pH 5.5 buffer at 35 C. (Adapted with permission from Ref. 56. Copyright 1996, American Chemical Society.)

Designing Functional Biomolecular Films

13

higher scan rates and thicker ¢lms. The movement of charge (electrons and counterions) through the ¢lm is characterized by Dct, which can be obtained from the slope of the linear Ip versus n1=2 plot. The slope of this plot is 1=2 1=2 ð2:69  105 Þn3=2 ADct Cf , and Dct can be estimated when all other parameters are known. Other transient electrochemical techniques can also be employed to estimate Dct [55]. Analysis of linear Ip versus n1=2 plots for the DNAMb ¢lm [56] in bu¡er gave Dct ¼ 1:2  107 cm2 =s. This value may re£ect physical di¡usion of the protein in the ¢lm, or ‘‘electron hopping’’ by electron self-exchange reactions between Mb redox sites in the ¢lm [50]. When a protein in solution can be taken up by a protein-free ¢lm, a breakthrough experiment can be done to give information about physical di¡usion of the protein in the ¢lm. A relatively thick protein-free ¢lm is placed into a protein solution, and the time of ¢rst appearance of the voltammetric peak is measured. The Einstein equation for molecular displacement D at time t is [49] D2 ¼ 2Dt

ð5Þ

where D is ¢lm thickness, and D is the protein di¡usion coe⁄cient in the ¢lm. Using Eq. (5) with the Dct found from voltammetry predicted a breakthrough time of 9 s for a 15-mm adsorbed DNA ¢lm. The measured value was 10 ms, suggesting that Mb di¡uses physically through the DNA ¢lms. Standard electrochemical rate constants can also be used to characterize electroactive ¢lms. They are denoted ks ðs1 Þ when determined under thin-¢lm conditions or k 00 (cm/s) when estimated under di¡usioncontrolled conditions. Electrochemical rate constants depend on applied potential, and ks and k 00 are de¢ned as the apparent values of the standard electrochemical rate constants at E ¼ E 00 . Also, k 00  dks , where d is ¢lm thickness in centimeters. Methods to determine ks and k 00 in polymer ¢lms [55] involve measurements on time scales short enough to observe kinetic limitations in transient electrochemical signals. As mentioned, peak widths of thin protein ¢lms under nonideal thinlayer conditions often exceed the theoretical 90=n mV, a consequence of the failure of the ideal model. In such cases, employment of simple thin-layer models to obtain ks from CV data can lead to considerable bias. However, models invoking a dispersion of E 00 and=or rate constant values can be used to obtain an average ks [5254].We feel that ks for proteins in ¢lms of ¢nite thickness are best interpreted as measures of the relative e⁄ciency of the electrochemical performance of the ¢lm. Clearly, ks may contain contributions from factors which may in£uence electrochemical reversibility other than electron transfer at the electrode¢lm interface. Kinetics of electron

14

Rusling and Zhang

transfer at interfaces between electroactive polymer ¢lms and electrodes are usually not rate-limiting [50]. 4.2

Square-Wave Voltammetry

The family of pulsed voltammetric methods can provide large improvements of resolution, sensitivity and signal=noise over CV. Square-wave voltammetry (SWV) is one of the most sensitive and rapid of these techniques. It gives peak-shaped currentpotential curves and sensitivities several orders of magnitude better than CV. In SWV, the input potential wavefrom consists of symmetric forward and backward square potential pulses superimposed on a potentialtime staircase [57]. Pulse frequencies range from about 1 Hz to several thousands of hertz, and control the experimental time window. Current is sampled at the end of the each forward and backward pulse, and plotted as forward, reverse, and di¡erence current versus step potential. Di¡erence square-wave voltammograms typically lead to symmetric peaks resulting from real-time subtraction of currents measured at the end of each forward and reverse pulse. An example of a reversible square-wave voltammogram is shown for a thin ¢lm of the electroactive polymer bis-bipyridylruthenium(II)-poly (vinylpyridine) [Ru(bpy)2 -PVP]2þ (Fig. 5), depicting the forward, reverse, and di¡erence current curves [58]. Typically, the forwardreverse traces are analogous to CV and valuable for diagnostic purposes, while the di¡erence current is more useful for analytical applications. For thin electroactive ¢lms, the SWV di¡erence peak current is directly proportional to the frequency. However, the integral under this curve has no direct relation to electroactive surface concentration GT , and this quantity is best determined by CV [cf. Eq. (2)]. A model for SWV of thin protein ¢lms has been developed invoking a dispersion of E 00 to explain peak broadness [53,54]. The method employs nonlinear regression analysis to ¢t the model simultaneously to forward and reverse SWV curves. The model combines E 00 dispersion with the SWV model for a single surface-bound electroactive species [59]. The SWV current (I) is p X Ij ð6Þ I¼ j¼1

where Ij is the contribution of the jth of p classes of redox centers with formal potentials Ej00 to the total current, given by Cj ð7Þ Ij ¼ ðnFAGj  Þ tp

Designing Functional Biomolecular Films

15

FIG. 5 Square-wave voltammogram showing forward, reverse, and difference current curves at 5 Hz of ½RuðbpyÞ2þ 2 PVP film on PG electrode in 20 mM acetate buffer, pH 5.5. (Adapted with permission from Ref. 58. Copyright 2001 Elsevier.)

where n is the number of electrons transferred per redox center, F is Faraday’s constant, A is electrode area ðcm2 Þ; Gj  is the total surface concentration (mol=cm2) of the jth class, tp is the pulse width, and Cj is the dimensionless current: h i
 Cj ¼ ðkf ; j þ kb; j ÞG0o;j  kb; j exp ðkf ; j þ kb; j Þ

ð8Þ

and G00o; j ¼

Cj þ kb; j ðkf ; j þ kb; j Þ

ð9Þ

where G0o;j is the jth surface concentration ratio Go;j =Gj  at time t ¼ 0; G00o;j is the ratio Go;j =Gj  at time t ¼ tp ; kf ;j ¼ kf ;j tp and kb;j ¼ kb;j tp . Forward and reverse electron-transfer rate constants kf and kb , respectively, are de¢ned as in the Butler-Volmer theory [49]:

16

Rusling and Zhang

FIG. 6 Square-wave voltammograms (4-mV step, 75-mV pulse) showing forward and reverse currents for cyt P450cam-DMPC films in pH 7.0 buffer. Points represent background-subtracted experimental data, solid lines are the best fits by 0 nonlinear regression onto the E 0 -dispersion thin-layer model described in the text. Average parameter values obtained from the regression analysis are 0 E 0 ¼ 121 2 mV versus NHE;ks ¼ 25 5 s1 ; and a ¼ 0:53 0:03. (Adapted with permission from Ref. 60. Copyright 1997 Royal Society of Chemistry.)

    nF 0 kf ;j ¼ ks tp exp a ðE  Ej0 Þ RT and kb;j

    nF 00 ¼ ks tp exp ð1  aÞ ðE  Ej Þ RT

ð10Þ

ð11Þ

where ks is the apparent standard surface rate constant ðs1 Þ; a is the electrochemical transfer coe⁄cient, E is applied potential, R is the gas constant, and T is temperature in kelvin. Parameters used in the ¢tting are the average 0 ks , average a, the pEj0 values, and the pGj values. Use of this model to ¢t SWV data at several frequencies for the FeIII=FeII redox couple of cytochrome P450cam in lipid ¢lms [60] is depicted in Fig. 6.The goodness of ¢t is illustrated by the good agreement between 0 experimental and computed data. The average E 0 is within 8 mV of the CV midpoint potential, and good parameter reproducibility is illustrated by the

Designing Functional Biomolecular Films

17

standard deviations. This method has been successfully applied to SWV of thin ¢lms of various polyions and lipids containing a number of di¡erent metalloproteins [7]. The distribution that falls out of these analyses is typically Gaussian. Note that the average ks is an apparent rate constant. The model neglects rate-limiting counterion entry or ejection, electron self-exchange, and molecular interactions, and assumes thin-layer SWV conditions. Thus, ks obtained by this method can be interpreted as a relative measure of the rate of the overall electron transfer process dependent on ¢lm and electrode properties. It is suitable for between-¢lm comparisons, but it is not an absolute electron transfer rate constant. 4.3

Marcus Electron Transfer Theory

Historically, theoretical descriptions of voltammetry have relied on ButlerVolmer electron transfer theory [49]. Butler-Volmer kinetics predict that the rate constants for heterogeneous electron exchange between a redox site and an electrode increase exponentially with increasing positive applied potential for oxidations and negative applied potential for reductions. This agrees with experiment at potentials near the formal potential, but predicts an in¢nite rate constant at extreme potentials, which is unrealistic. The Marcus theory of electron transfer correctly predicts the increase in the electrontransfer rate constant with increasing applied potential, with eventual attainment of an upper limit [61]. According to Marcus theory, the rate constant ks for outer-sphere electron transfer depends on activation free energy DG  :   DG ks ¼ kA exp ð12Þ RT where k is the electronic transmission coe⁄cient, A is collision frequency, R is the gas constant, and T is temperature in kelvin [61]. DG  is related to the standard free energy DG0 and the reorganization energy l for the electron transfer: ðDG 0 þ lÞ2 4l The potential dependence of ks arises from the relation DG  ¼

DG 0  nF Z

ð13Þ

ð14Þ 0

where the overpotential Z ¼ ðE  E 0 Þ and E is the applied potential. From Eqs. (12) to (14), we can show that Marcus theory correctly predicts a sigmoidal increase in ks with increasing Z, with a plateau limiting value at high Z.Voltammetric models based on Butler-Volmer kinetics in thin

18

Rusling and Zhang

¢lms, while useful for comparisons between systems, may lead to biased estimates of kinetics at high Z [28]. Several research groups have applied Marcus kinetics to ¢lms of redox proteins. Bowden and co-workers used electrochemical methods to estimate a lower limit value of l of 0.28 eV for cyt c bound to carboxylalkythiol SAMs on Au electrodes [62]. This model was extended to include Gaussian distributions of all parameters and applied to SWVof myoglobin in thin ¢lms of didodecyldimethylammonium bromide on pyrolytic graphite electrodes [63]. Nonlinear regression analysis of SWV allowed direct estimation of electron-transfer rate constants and reorganization energies. Relatively large pulse heights, e.g., 100200 mV, were needed to extract reliably selfconsistent values of reorganization energies. An example of a ¢t is shown in Fig. 7. The mean values of reorganization energies for reduced and oxidized forms of Mb, respectively, found by this method were lRED ¼ 0:41 eV and lOX ¼ 0:21 eV.

FIG. 7 Experimental current (s)with model background current subtracted resulting from regression analysis of SWV data at 200-Hz frequency, 160-mV pulse height shown with the best-fit line from the Marcus model. The Marcus model was fitted to raw SWV data including background, and the model background was then subtracted. Average parameter values from the analysis of 22 voltammograms were standard rate constants log (knullRED,s1) ¼ 3.3 0.8, log(KnullOX,s1) of 3.0 0.6, reorganization energy lRED ¼ 0.41 0.02 eV, and lOX ¼ 0.21 0.01 eV. (Adapted with permission from Ref. 63. Copyright 2001, American Chemical Society.)

Designing Functional Biomolecular Films

19

SCHEME 1 Electron-hopping charge transport in a redox protein film.

Marcus theory also relates kinetics to the distance of electron transfer. The relation between ks and distance, d, of electron transfer from the electrode surface to a ¢xed redox site relative to rate constant k0 at the distance of closest contact d0 is [61] ks ¼ k0 exp½bðd  d0 Þ b is typically in the range 8.511.5 nm1, indicating a rapid decrease of ks with distance. This means that the farther a protein redox site is from the electrode, the slower is its rate of direct electron exchange with the electrode. However, ¢xed redox sites su⁄ciently far away from the electrode may participate in electron self-exchange by ‘‘electron hopping’’ involving neighboring sites. For example, a multiplayer ¢lm containing iron heme proteins in the PFeIII form at ¢xed sites could propagate charge from electrode-¢lm to solution-¢lm interfaces as shown in Scheme 1. 4.4

Enzyme Catalysis on Electrodes

Designers of catalysts have long been envious of enzymes with high selectivities and speci¢cities.We now possess the methodology to drive catalytic chemical reactions by redox conversion of enzymes (or proteins) on electrodes. Since many of the di¡erent types of ¢lms described in this chapter can support similar enzyme-like reactions [7],we provide here only a general summary of reactions that can be done in this way. Films on electrodes featuring direct electron transfer provide the means to do enzyme chemistry under the driving force of the applied electrical potential (Fig. 8). The process is similar to electrochemical catalysis using nonbiological redox centers, a simple example of which is shown in Scheme 2. Here, redox catalyst P in a ¢lm is reversibly reduced by the electrode to the activated catalyst Q, which reduces the reactant A to product B [49,64]. Typically, A is directly (noncatalytically) reduced irreversibly at an electrode 0 potential more negative than the E 0 of the P/Q catalyst couple. In the absence of A, the P/Q coupled in the ¢lm gives a reversible voltammogram,

20

Rusling and Zhang

FIG. 8 Conceptual illustration of electrochemical catalysis with an enzyme film on an electrode.

SCHEME 2

Simple pathway for electrochemical catalysis.

such as in Fig. 3. Addition of the reactant A to the solution allows reaction between Q and A, changing the shape of the voltammogram. If the reaction of Q þ A is fast enough with respect to scan rate in CV, the reduction current for P increases and the oxidation current for Q vanishes. This is because Q reacts with A, producing P, which continues the catalytic cycle. An analogous scheme pertains for catalytic electrochemical oxidations. While the general concept of Scheme 2 holds, most enzyme or proteincatalyzed catalytic reactions on electrodes are more complex. They usually involve more than one electron, and may feature an enzymesubstrate complex. An example of a two-electron protein-catalyzed reduction is the dechlorination of trichloroacetic acid by iron heme protein in thin ¢lms [65]. The pathway involves sequential two-electron reductions of C ^ Cl bonds. Films of Mb and the lipid DMPC are used to illustrate CV behavior for the catalytic reactions. In the absence of reactant trichloroacetic acid (TCA), a reversible thin-¢lm CV is found for the heme FeIII /FeII couple of myoglobin

Designing Functional Biomolecular Films

21

FIG. 9 Voltammograms showing effect of catalytic reduction of trichloroacetic acid (TCA) by Mb in DMPC films at 0.1 V=s and pH 5.5. Numbers on curves show millimolar TCA concentration in the buffer; dashed curve to the far right for TCA on a DMPC-coated electrode without Mb. (Adapted with permission from Ref. 53. Copyright 1997, Elsevier.)

(Fig. 9). Additons of TCA cause increases in the FeIII reduction peak with increasing concentration and decreases in the FeII oxidation peak. The general reaction is given in Scheme 3. At higher concentrations of TCA, the MbFeII oxidation peak disappears because it is completely used up in the second step of Scheme 3. Direct, uncatalyzed reduction of TCA on a DMPC-coated electrode begins to show a current at 1 V, whereas the catalytic reduction occurs at about 0.3 V (Fig. 6). This positive shift of 0.7 V in the reduction potential re£ects the decrease in activation free energy for the catalytic reduction.The increase in current with concentration of reactant is a general feature of electrochemical catalysis, and is the basis of applications to biosensors.

SCHEME 3 Electrochemical catalytic reduction of TCA with Mb.

22

Rusling and Zhang

Rotating-disk electrode voltammetry (RDV) employs an electrode which rotates at ¢xed speed to control mass transport to and from the electrode. In the absence of reactive substrate in solution, rotating the electrode has no e¡ect on the shape of a reversible thin protein ¢lm voltammogram because the electrochemistry is insensitive to mass transport external to the ¢lm. Addition of substrate to the solution provides catalytic steady-state voltammograms by RDV. The currentpotential curve now assumes a sigmoid shape (Fig. 10) with a steady-state plateau current. The Michaelis-Menton formulation of enzyme kinetics postulates an enzymesubstrate (reactant) complex with dissociation constant KM and turnover rate constant kcat for the conversion of this enzymesubstrate complex to product. If the electrochemistry of the enzyme in the ¢lm is reversible and mass transport of substrate in the solution is optimized using RDV, the steady-state catalytic current Icat is given approximately by a version of the Michaelis-Menton equation [66]: Icat ¼

nFAGkcat Cs Cs þ K M

ð16Þ

where G is the amount (or surface concentration) of enzyme in the ¢lm (mol=cm2), and Cs is the concentration of substrate. KM and kcat can be obtained by ¢tting Eq. (16) to Icat versus Cs data [67]. An important example of electrochemical catalytic oxidations employing iron heme enzymes can be achieved using oxygen. Films of myoglobin (Mb) and cyt P450cam on electrodes catalyze reduction of oxygen to hydrogen peroxide [6870]. Hydrogen peroxide then reacts with the FeIII form of the protein to yield the active oxidant. The process is thought to occur as shown in Scheme 4. Despite a more complex pathway, voltammograms for this catalytic reduction of oxygen are similar to that for reactions following Schemes1and 2. In the presence of oxygen, the reversible reduction peak for MbFeIII in ¢lms is converted to a reduction peak much larger than that in anaerobic solutions, while the FeII oxidation peak disappears (Fig. 11). The uncatalyzed reduction of oxygen occurs at 0.8 V versus SCE, but is shifted 0.55 V positive to about 0.25 V when the Mbpolyion ¢lm is used. Scheme 4 can be utilized for enzyme-like oxidative catalysis involving oxygen transfer from the active oxidant to a substrate, e.g., conversion of styrene to styrene oxide [6870]. Catalytic amperometric and voltammetric responses of peroxidase ¢lms to hydrogen peroxide are likely to be related to a pathway similar to Scheme 4 [71]. Stable ¢lms of DMPC containing M. tuberculosis catalaseperoxidase (KatG), several peroxidases, myoglobin, or catalase gave reversible FeIII=FeII voltammetry on pyrolytic graphite electrodes, and catalytic

Designing Functional Biomolecular Films

23

FIG. 10 Idealized voltammograms computed from theory for a rotating disk electrode: (a) reversible thin enzyme film; (b) catalytic current for a thin enzyme film and substrate (reactant) in solution featuring a steady-state current beyond the protein reduction potential.

SCHEME 4 Pathway for electrochemical catalytic reduction of oxygen to hydrogen peroxide, leading to an oxyferryl radical that can oxidize organic substrates.

current for hydrogen peroxide and oxygen. The similarity of catalytic voltammograms and amperometric responses for these ¢lms to H 2O2 at 0 V suggests signi¢cant contributions from catalytic reduction of oxygen produced during the catalytic cycles. Oxygen is produced by catalase-like reaction of H2O2 with the oxyferryl radical PFeIV ¼ O, also giving water as a product and regenerating the PFeIII form of the enzyme. The presence of oxygen sets up the catalytic cycle in Scheme 4. Relative apparent turnover rates at pH 6 obtained from steady-state currents at 0 V versus SCE in *

24

Rusling and Zhang

FIG. 11 Cyclic voltammograms at scan rate 0.3 V=s in pH 5.5 buffer illustrating protein-catalyzed reduction of oxygen: (a) reversible CV of a multilayer film on rough PG consisting of six alternating layers each of polystyrene sulfonate (PSS) and myoglobin (Mb) with no oxygen present; (b) catalytic CV of {PSS=Mb}6 film on PG after oxygen was injected into buffer; (c) PSS film without Mb after oxygen was injected into buffer, showing noncatalytic reduction of oxygen. (Adapted with permission from Ref. 182. Copyright 2000, American Chemical Society.)

response to H2O2 were in the order horseradish peroxidase > cytochrome c peroxidase > soybean peroxidase > myoglobin > KatG > catalase [71]. Lower currents for the very e⁄cient peroxide scavengers KatG and catalase may be related to the instability of their oxyferryl radicals in the presence of H2O2.

4.5

Oxidations of Nucleic Acid Bases

The most easily oxidized base in DNA is guanine, with a formal potential at pH 7 of about 1.3 V versus NHE (1.06 V versus SCE), while the other DNA bases have formal potentials up to 0.5 V more positive [72]. Detection of nucleic acid oxidation by direct electrochemistry depends on structure,with double-stranded (ds) DNA giving only minimal detectable oxidation, and single-stranded (ss) DNA giving oxidation peaks at about 1.01.2 V versus SCE on carbon. The key point seems to be access of the bases to the electrode. This access is poor in ds-DNA, but much better in ss-DNA and partly unraveled, damaged DNA. The direct electrochemical oxidations of guanines in DNA are chemically irreversible [12,38], and CV of ss-DNA ¢lms typically gives oxidation peaks with no accompanying reduction peaks.

Designing Functional Biomolecular Films

25

Catalytic electrochemical oxidation of DNA using transition metal complexes provides enhanced electrochemical signals [10]. Elegant work by Thorp et al. showed that Ru(bpy)2þ 3 is a very e⁄cient electrochemical catalyst that oxidizes only guanine bases in DNA and oligonucleotides [73,74]. The reaction follows the typical catalytic pathway in Scheme 5, and 3þ  RuðbpyÞ2þ 3 ¼ RuðbpyÞ3 þ e 2þ þ RuðbpyÞ3þ 3 þ DNAðguanineÞ ! RuðbpyÞ3 þ DNAðguanine Þ

ð17Þ ð18Þ

SCHEME 5

DNA(guanineþ ) is subsequently further oxidized [75]. Ru(bpy)33þ is cycled back to Ru(bpy)2þ 3 by the fast chemical step in Eq. (18), which provides voltammetric current that is greatly enhanced over that of Ru(bpy)2þ 3 or DNA alone. The peak current depends on the rate of the chemical step. The same accessibility issues pertain as in direct electrochemical oxidation, i.e., guanine in ss-DNA reacts more rapidly with than ds-DNA, and the reaction rate is also sequence-dependent.

5 5.1

MONOMOLECULAR AND SUBMONOMOLECULAR PROTEIN FILMS Adsorbed Films

Fraser Armstrong and coworkers showed that polyamines such as aminocyclitols and polymixins give stable co-adsorbed ¢lms with a wide variety of ferredoxins and other redox proteins at the negative surface of edge-plane pyrolytic graphite electrodes [66]. Presumably, the stability of these ¢lms is controlled by interactions of carboxyl groups on the electrode surface and negative sites on the protein with ammonium groups on the co-adsorbate. Co-adsorption of proteins with aminocyclitols and polymixins often gives a monolayer or submonolayer of noninteracting protein molecules on edge-plane pyrolytic graphite surfaces. This provides a high concentration of electroactive centers close to the electrode, eliminates di¡usion from the voltammetry, often avoids surface denaturation, and inhibits adsorption of impurities which might otherwise block electron transfer. A key feature of these ¢lms is that electron transfer kinetics between electrode and protein are often quite fast, and for some metalloproteins reversibility persists to scan rates of 3000 V=s and higher [76]. This allows estimation of standard electron transfer rate constants up to 5000 s1, as well as studies of coupled chemical processes on the millisecond time scale. Detailed models have been presented for interpreting catalytic voltammetry of monolayer ¢lms of adsorbed electroactive enzymes [77].

26

Rusling and Zhang

Monolayer ¢lm voltammetry has been used for seminal studies of electron transfer properties, coupled chemical processes, and enzyme catalysis for a variety of proteins. For details,we refer the reader to recent reviews [66,78], and to the original literature cited therein. Some exciting examples of this work include demonstrations that transition metal ions can enter vacant subsites of iron clusters in [3Fe4S]0 ferredoxins directly in these monolayers to give [M3Fe4S]2þ clusters [79,80], mechanistic elucidation of proton-coupled electron transfer and long-range proton transfer involving ironsulfur cluster proteins [8183], and discovery of a tunnel diode e¡ect for monolayers of succinate dehydrogenase on electrodes [66,84] involving enzyme activity gated by electrode potential which controls the redox state of the enzyme.The method has been used for many kinetic, mechanistic, and H=D isotope studies of catalysis by native and mutant enzymes [66,8587]. The most recent studies of thin protein monolayers by voltammetry demonstrate increasingly sophisticated mechanistic resolving power. Examples include low-temperature protein ¢lm cryovoltammetry to resolve complex transformations in ferredoxin electron transfer [88], and elucidation of an atomically resolved long-range electron-coupled proton transfer in a buried protein redox center [89]. Catalytic electrochemical currents can sometimes be measured with enzyme-coated electrodes even when no voltammetric peaks are detected in the absence of the substrate [66,90]. Thus may suggest submonolayer coverage or partial blockage of the electrode.While lack of detection of the direct enzyme voltammetry limits kinetic analyses, catalytic responses may still ¢nd analytical or diagnostic uses. Many studies document that protein adsorption without co-adsorbates on solid metal, mercury, and carbon electrodes often results in denaturation that blocks electron transfer. However, there appear to be some interesting exceptions as summarized below. The use of antibody probes revealed three conformations of cyt c3 adsorbed on Hg that were all denatured, but the protein ¢lm remains electroactive [91]. The NiFeS enzyme hydrogenase from Desulfovibrio dusulfuricans Norway was adsorbed directly onto basal-plane pyrolytic graphite, allowing studies of its reaction with cyt c3 with a lifetime of the hydrogenase ¢lm of several hours [92]. Megasphaera elsdenii Fe hydrogenase was adsorbed from solution at submonolayer coverage onto rough glassy carbon electrodes [93]. Although no direct electron transfer peaks were found, electron transfer involving this enzyme was con¢rmed by catalytic currents for reduction of Hþ and oxidation of H2. In this case, addition of polyamine co-adsorbates did not help to form a stable ¢lm. Tao et al. adsorbed ¢lms of cyt c and myoglobin (Mb) on freshly cleaved highly ordered basal-plane pyrolytic graphite (HOPG) [94,95]. Atomic force microscopy (AFM) revealed a randomly distributed, adsorbed mono-

Designing Functional Biomolecular Films

27

layer of cyt c after a few minutes, but complete formation of chainlike aggregates of adsorbed Mb on HOPG took 40 min. AFM pictures were consistent with native adsorbed protein in both cases. Mb ¢lms gave direct, reversible electron exchange between HOPG and the iron heme center. Cyt c gave reversible iron heme electron transfer in phosphate bu¡ers only after the electrode potential was ¢rst taken to 0.75 V. At this potential, phosphate adsorbs to HOPG and may facilitate the reversible electrochemistry [94]. An AFM-electrochemical study of Cys3Ser azurin on gold showed an electroinactive ¢lm of protein also featuring two-dimensional aggregates [96]. Cys3Ser azurin ¢lms gave an unstable, decaying redox peak on edge-plane pyrolytic graphite. Thus, direct adsorption of proteins on conductive (nonoxide) electrode surfaces can give electroactive ¢lms, but may result in problems with protein aggregation and limited stability. An alternative to constructing monolayers of proteins on electrodes is to employ coatings of charged, self-assembled monolayers (SAMs) of alkanethiols on gold electrodes. For example, carboxy-alkanethiols on gold adsorbed positively charged cyt c and facilitated direct, reversible protein electrochemistry [97]. These ¢lms remain reasonably stable if extremes of applied voltage and pH are avoided. Chapter 2 reviews protein electrochemistry using SAMs.

5.2

Covalent Bonding of Proteins to Electrodes

Before the 1990s, methods involving covalent bonding of enzymes to electrodes did not generally result in direct electron exchange between the protein redox sites and electrodes [2123]. More recently, direct electron transfer to cyt c and other redox proteins has been achieved in ¢lms made by covalent bonding the proteins to functionalized SAMs as reviewed elsewhere [7], and in Chapter 2 in this book. A variation of the SAM approach involves attachment of thiol groups on the protein to gold. Thus, the redox protein azurin was mutated by replacing the serine with cysteine in position 118 [98]. The mutated azurin was chemisorbed to gold electrodes via the cysteine sulfurgold bond, and gave quasi-reversible voltammetry. Willner and co-workers at the Hebrew University of Jerusalem have devised elegantly constructed covalently bound protein layers on electrodes to obtain catalytic properties. In some cases, the coated electrode devices are photoswitchable [99]. A major strategic feature of their approach is to create covalent linkages that facilitate electron exchange from electrodes to redox sites in the protein.We discuss several of these systems below. One example involved design of a biocatalytic surface to produce maleic acid stereoselectively. Microperoxidase 11 (MP-11), an iron

28

Rusling and Zhang

FIG. 12 Stepwise construction of MP-11=MbCoII monolayer film on a propylamino-SAM on a gold electrode. Electrons can be transferred from electrode to MP-11 to MbCoII to a substrate in a contacting solution.

hemepeptide fragment featuring the active site of cyt c, was linked with amide bonds to mercaptoethylamine SAMs using a water-soluble carbodiimide (EDC) promotor. The MP-11 ¢lms bind the heme protein myoglobin (Mb), which was subsequently cross-linked with glutaric aldehyde [100] (Fig.12).Vectorial electron transfer from electrodes to MP-11 iron heme gave the FeII MP-11that donated the electron to the CoIIMb to obtain CoIMb.The latter species reduced the triple bond in acetylene dicarboxylic acid to yield maleic acid. Reactions of a large number of redox enzymes depend on the cofactor nicotinamide adenine dinucleotide (NAD), which associates near the active site of the enzymes. In another example, NADþ ¢lms were constructed on mercaptoethylamine SAMs and the NADþ was bound to cross-linked enzymes. These ¢lms provided biocatalytic oxidation of substrates [101]. The approach involved the initial binding of carboxylate groups of pyrroloquinoline quinone (PQQ), a catalyst for the oxidation of NADþ , to the SAM by EDC-assisted amide formation. EDC was then used to attach N6-(2-aminoethyl)-NADþ to unreacted PQQ carboxylates. This

Designing Functional Biomolecular Films

29

gave a Au-PQQ-NADþ electrode to which lactate dehydrogenase or alcohol dehydrogenase could be associated and then cross-linked. Substrate conversion from lactate to pyruvate and ethanol to acetic acid were accomplished. Catalytic currents obtained from these substrates increased linearly with concentration of the substrate, suggesting possible applications to biosensors. The above covalent bonding approach has been extended to sophisticated multibiomolecule devices. In one example, a four-helix bundle de novo-synthesized protein was attached as a monolayer to a gold electrode. Then, histidine residues on the helices were used to bind iron heme units that exchanged electrons with the electrode [102]. This monolayer coupled with a glutaric aldehyde-¢xed layer of nitrate reductase was used for catalytic reduction of nitrate ion. Biosynthetic applications were demonstrated using a coupled layer of CoII-reconstituted myoglobin. In other examples, a sequential Au-horseradish peroxidase (HRP) glucose oxidase electrode was designed to catalyzed oxidation of glucose by O2, producing H2O2. Catalytic oxidation of 1-chloro-1-napthol by the H2O2 HRP gave an insoluble product that precipitated on the electrode and was detected by Faradaic impedance changes, cyclic voltammetry, and quartz crystal microbalance frequency changes [103]. A similar strategy was used to detect acetylcholine with a three-enzyme layered electrode [104]. Covalent binding of proteins to electrodes that exhibit hinge-bending motions has allowed construction of nonenzymatic biosensors [105]. Genetically engineered periplasmic binding proteins tagged with an electroactive RuII complex at one end were attached at the other end to a functionalized SAM. Binding of speci¢c substrates (e.g., maltose, Zn2þ , glutamine) caused a bending of the protein such that the RuII complex was moved farther away from the electrode and gave smaller voltammetric oxidation peaks. Films of this type were used to successfully determine maltose in beer and glucose in rat blood serum.

6 6.1

CAST BIOMOLECULEPOLYION FILMS ProteinPolyion Films

Polymerprotein ¢lms have been utilized frequently for mediated enzyme electrochemistry [2123], but polymer ¢lms for direct electron transfer between proteins and electrodes have developed more slowly. The ¢rst report of direct, reversible electron transfer involving proteins in a polymer coating on an electrode seems to have been from Murray’s lab at the University of North Carolina, and involved cyt c in gel coatings of poly-

30

Rusling and Zhang

acrylamide and polyethylene oxide on a planar cell featuring an edge-plane pyrolytic graphite working electrode [106]. These gel coatings had a high water content, with salt added for ionic conduction. They gave reversible CVs for cyt c indistinguishable from those of the protein in aqueous solutions. Films of polyelectrolytes can support high water content and serve as suitable hosts for proteins. Extensive work with small electroactive ions has been done with ¢lms of Na¢on, a per£uosulfonate ionomer with less than 15% ionizable sulfonate groups per monomer. Na¢on’s structure results in a partly hydrophobic character leading to a high a⁄nity for hydrophobic cations [107109].

Structure of Na¢on Hill et al. studied ¢lms of proteins and Na¢on cast from aqueous mixtures onto basal-plane pyrolytic graphite [110], then coated after drying with a second layer of Na¢on. While cyt c gave no CV peaks, reversible voltammetry was observed for cyt c551, cyt b5, and azurin in these ¢lms. The charge on cyt c under the conditions used is about þ8, and it would be expected to interact electrostatically with negative charges on Na¢on. In contrast, charges for proteins that have good CV responses were 2 [cyt c551], 8 [cyt b5], and þ2 [azurin]. These results suggested that hydrophobic interactions may stabilize protein binding within the Na¢on ¢lms. In agreement with this view, Bianco and Haladjian found that electrodes coated with Na¢on did not incorporate positively charged protein cyt c3 [111]. If this protein was ¢rst cast onto the electrode, then coated with Na¢on, voltammetry could be observed but peaks decayed rapidly upon multiple scans. Negatively charged spinach ferredoxin deposited onto a PG electrode with the cationic polyion polylysine, dried, and then coated with Na¢on gave more stable, reversible CVs [111]. Polylysine has protonated amine groups at neutral pH. This soluble polycation can be added to solutions to facilitate co-adsorption of negatively charged proteins on electrodes. For example, proton reduction catalyzed by the negative hydrogenase was studied in this way at a pyroly¢c graphite (PG) electrode [112,113]. Catalytic peaks for the reduction of protons to hydrogen were found. Carboxyalkylthiol SAMs on Au were used to ¢rst adsorb poly(lysine), with subsequent adsorption of the negatively charged cytochrome b5 to obtain reversible voltammetry [114].

Designing Functional Biomolecular Films

31

Poly(lysine) was added to solutions to achieve direct electron transfer from indium tin oxide (ITO) electrodes to spinach ferredoxin, which transferred this electron to cyt P450cam in solution [115]. This system of poly(lysine), spinach ferredoxin, and cyt P450cam was used to dehalogenate toxic organohalides, mimicking bacterial bioremediation. Eastman AQ ionomers,which are poly(ester sulfonic acids), can also be used for making protein ¢lms. AQ ionomers have sulfonate groups at 11% of monomer sites and a backbone with carbomethoxyphenyl groups. Positive proteins cyt c and cyt c3 could be taken up from their solutions into AQ ¢lms on electrodes [116], giving reversible voltammetry. In contrast, no uptake of these proteins by Na¢on ¢lms was observed. A big di¡erence between these two polymers is that AQ ¢lms in bu¡er contain about 90% water and are hydrogels, while Na¢on ¢lms contain only about 30% water [117].

Structure of Eastman AQ38 and AQ29 Ionomers Thin ¢lms cast from mixtures of ionomer Eastman AQ38 and positively charged myoglobin (Mb) formed stable hydrogel ¢lms on PG electrodes after immersion in water [117]. Coulombic interactions between Mb and the ionomer control retention of the protein, and Soret absorbance band positions suggested that Mb in the ¢lms has a conformation similar to native Mb. Reversible cyclic voltammograms were reproducible for several months when Mb-AQ38 ¢lms were stored dry or immersed in aqueous bu¡er. Oxygen and trichloroacetic acid were catalytically reduced by Mb in AQ ¢lms with signi¢cant decreases in the electrode potentials required. Stable ¢lms on PG electrodes were also made from Eastman AQ55 and hemoglobin (Hb) featuring reversible voltammetry for the four iron heme groups of the protein [118]. Soret absorption bands were consistent with a near-native conformation for Hb in AQ ¢lms in the medium pH range. Square-wave voltammograms of Hb-AQ ¢lms were successfully ¢tted by nonlinear regression analysis using the model featuring dispersion of formal potentials. An average ks of 62 s1 was obtained. Hemoglobin in these ¢lms catalyzed the reduction of trichloroacetic acid. AQ ¢lms containing c-type cytochromes were used to study reactions of soluble FeIII ammonium citrate [119]. c-Cytochromes from sulfate or sulfur-reducing bacteria acted as FeIII reductases, but mitochondrial

32

Rusling and Zhang

c-cytochromes did not.Films of AQ and Mb-poly(ethylene oxide) were stable in ethanolic solutions, and were used to reduce hexachloroethane [120]. A great advantage of cast polymerprotein ¢lms is ease of preparation, involving only mixing of protein and polymer, and spreading the mixture on an electrode surface. However, a disadvantage is that the protein may not be well retained in ¢lms in contact with solutions, causing electrochemical signals to decay with time. 6.2

DNAIonomer Films

Films of double-stranded (ds) DNA and ionomers were stable enough to be used for detecting DNA damage from incubation of the ¢lm with styrene oxide, which reacts with guanine bases and disrupts the double helix. Films were prepared on PG electrodes by casting mixtures of ds-DNA with Eastman AQ38S or with Na¢on [13,121]. Damage to these DNA ¢lms was detected by oxidation peaks that developed during reaction with styrene oxide at 37 C for 10 min to several hours using derivative square-wave voltammetry. Single electrodes can be used to measure relative damage rates. Damage of DNA by reaction with styrene oxide under the electrode incubation conditions was con¢rmed by capillary electrophoresis. Total integrals of oxidation peaks increased with time of incubation with styrene oxide. Relative peak heights depended on the type of DNA in the order calf thymus ds-DNA > salmon sperm ds-DNA > supercoiled ds-DNA > highly polymerized calf thymus ds-DNA. Films of Mb=DNA also showed oxidation peaks after short incubations with styrene oxide that may be attributable to DNA damage.

7 7.1

FILMS WITH SURFACTANTS AND LIPIDS ON ELECTRODES Bilayer Lipid Membranes (BLMs)

Living biomembranes contain roughly equal amounts of phospholipid and protein. Many enzymes function while they are attached to membranes [122]. Enzymes can be bound on the surface, partly inserted, or extend fully across the biomembrane (Fig. 13). Biomembrane-like structures of lipids or insoluble synthetic surfactants containing proteins have been constructed on electrodes in a number of ways. To mimic biological membranes, lipids or surfactants should reside on the electrode in bilayer structures such as in Fig. 13. Below, we brie£y review structural characteristics that cause lipid and surfactant molecules to form bilayers.

Designing Functional Biomolecular Films

33

FIG. 13 Conceptual model of a lipid bilayer membrane showing three modes of protein binding.

Surfactant or lipid molecules contain charged or polar head groups and one or more long hydrocarbon tails (Fig. 14). The term surfactant (from surface-active agent) is more general, and the name lipid is usually reserved for biological surfactants that help to make up biomembranes. Molecular structure can be connected with supramolecular nanostructure by the surfactant packing parameter, v=ao lc , where v is the volume of the hydrocarbon tail region of the surfactant, ao is the optimal area per head group, and lc is the critical chain length [123,124]. Chain length and volume can be estimated from lc ¼ 1:5 þ 1:26ðn  1Þ ð—Þ v ¼ 27:4 þ 26:9ðn  1Þ ð—3 Þ

ð19Þ ð20Þ

where n is the number of carbon atoms in the chain. Note that v for a doublechain surfactant is twice that of a single-chain surfactant with the same chain length. Molecules with values of the packing parameter between 0.5 and 1 are predicted to form bilayer structures, are usually insoluble in water, and are good candidates for bilayer ¢lm formation. Surfactants with v=ao lc < 0:5 are water-soluble and form micelles. Molecules with packing parameters >1 tend to form reversed micelles in organic solvents. Lipids and surfactants in Fig. 14 and are examples of molecules that have been used to make proteinsurfactant ¢lms. All have v=ao lc between 0.5 and 1. Such molecules tend to have two or three hydrocarbon tails. This increases the value of v compared to a single-chain surfactant, while not changing ao lc very much.

34

FIG. 14

Rusling and Zhang

Water-insoluble double-chain surfactants that form bilayer structures.

One approach to lipid bilayer structures on electrodes is LangmuirBlodgett (LB) membrane transfer. Here an LB lipid ¢lm is formed at the airwater interface in an LB ¢lm balance, and transferred to an electrode by controlled dipping and extraction. Yokoto et al. reported direct, chemically reversible voltammetry for cyt c on a phosphatidylcholine layer on an ITO electrode [125]. Multilayer ¢lms can also be constructed on electrodes by LB transfer methods. Reversible CV for Mb was found in multilayer ¢lms of a variety of synthetic insoluble lipids made in this way [126]. However, LB ¢lm transfer is tedious and impractical for biosensor or bioreactor development. Simpler, more general methods of making lipid membranes on electrodes have evolved. Tien et al. reported [127,128] adsorbing lipids in decane or squalane=butanol onto freshly cleaved active noble metal wires. This wire is transferred to an aqueous solution and the adsorbed ¢lm thins to a lipid bilayer that is stable for 36 h. Films are equivalent to bilayer lipid membranes (BLM) formed across a small hole in a barrier between two solutions, a technique developed 30 years earlier.

Designing Functional Biomolecular Films

35

Dipping a basal plane pyrolytic graphite electrode into a detergent was also found to provide coatings on electrodes which facilitated direct electron transfer [129]. Reversible CVof negatively charged cyt b5 was obtained on an electrode coated with dodecylamine, but positively charged proteins gave no electrochemical response. Reversible cyt c voltammetry was achieved with a coating of lauric acid (dodecylcarboxylic acid). The BLM electrode method was used with lecithin (mixed phosphatidylcholines) ¢lms to reduce cyt c [130]. Results suggested that binding of the protein to the BLM is strongly facilitated by electrostatic interactions between the cyt c and the BLM [131]. Studies of thioredoxins (MW ca. 12,000) featuring disul¢de=dithiol redox sites at BLM electrodes indicated voltammetry was controlled by di¡usion of the proteins [132]. Direct electrochemistry of [Fe2S2 ] ferredoxins was observed at BLMs formed on gold from neutral phosphatidylcholines (2.5 mg=mL) or phosphatidylcholine (50 mg=mL) including dioctadecyldimethylammonium bromide (DODAB) [133]. Linear di¡usion control of the electrochemistry was found, and binding of the negative ferredoxins to the positive BLMs was largely electrostatic. Thus, while binding of the protein to the BLM is important in this method, electroactive proteins do not seem to have residence times of more than a few seconds on the charged membranes. Binding and good CV responses are favored by low concentrations of neutral lipid in the adsorbate solution, inclusion of a surfactant of complimentary charge of the protein, and low ionic strength in the protein solution. Both hydrophobic and electrostatic interactions between the BLM and the protein may be important for binding. On ITO electrodes, BLMs with low concentrations of neutral lipid (0.5 mg=mL) in the adsorbate solutions gave adsorption-controlled voltammetry of cyt c [134]. At higher lipid concentrations (5 mg=mL), with or without addition of anionic surfactant, di¡usion-controlled cyt c voltammetry was found. These results were similar to those for BLMs on gold electrodes. A later study on ITO reported conditions under which voltammetry suggested various degrees of insertion of cyt c into the BLM, involving the relative balance of electrostatic and hydrophobic interactions [135]. BLMs were also used to compare reactions of cyt c bound to membranes and in solution with £avodoxin semiquinone and reduced spinach ferredoxin [136]. BLMs on ITO were used to obtain CV of integral membrane proteins cyt f and cyt c oxidase [137]. BLMs on gold were used to study redox chemistry of spinach thioredoxins f and m and ferredoxin:thioredoxin reductase [138]. An approach to making bilayers containing membrane proteins on electrodes involves chemisorption of a submonolayer of alkane thiol onto a

36

Rusling and Zhang

gold or silver electrode in the presence of a bilayer-forming surfactant. Alkanethiols and detergent-solubilized enzyme in bu¡er are put in contact with a gold electrode while being dialyzed against bu¡er without additives [139]. Dialysis dilutes the solubilizing agents in the enzyme solution, helping to co-deposit enzyme with the alkanethiol on the electrode. Escherichia coli fumarate reductase and fructose dehydrogenase [140] were immobilized for biosensor applications by this method. The large transmembrane protein cyt c oxidase was immobilized in this way [141144], and was shown to mediate electron transfer with its natural redox partner reduced cyt c.

7.2

Cast Surfactant Films Incorporating Proteins

Spreading organic solutions or aqueous vesicle dispersions of waterinsoluble surfactants onto solid surfaces is a simple way to make ordered, stable multilayer ¢lms [145]. In a procedure called casting, the solution or dispersion is spread evenly onto the surface, and the solvent is allowed to evaporate. The surfactants self-assemble into ordered stacks of bilayers (Fig. 15).When these ¢lms are annealed in aqueous bu¡ers, large amounts of water become associated with the head groups between bilayers. Di¡erential scanning calorimetry can be used to observe gel-to-liquid crystal phase transitions of these ¢lms. This features a transition from the solid-like gel phase in which hydrocarbon tails are extended in all-trans conformations to a more £uid liquid crystal state featuring dynamic gauche bonds in the tails [6].Gel-to-liquid crystal transitions are characteristic of surfactant and lipid bilayer structures, and transition temperatures in the ¢lms are similar to those of vesicle dispersions of the same surfactant. Double- and triple-chain surfactants, by virtue of the large volume of their tail regions, have surfactant packing parameters between 0.5 and 1 and can be used to form these multibilayer ¢lms. The casting technique provides a viable alternative to LB ¢lm transfer for easily preparing multiple bilayer ¢lms of surfactants on surfaces. Kunitake et al. ¢rst inserted a protein, myoglobin (Mb), in an oriented fashion between bilayers of dialkylphosphate surfactants [146]. This work pre¢gured the use of metalloproteinsurfactant ¢lms on electrodes for voltammetry. Polymer-stabilized surfactant ¢lms can be made by replacing the counterions of the surfactants with ionic polymers in the casting solutions [147149]. For example, bromide in DDAB (Fig. 14) can be replaced with polystyrene sulfonate, and the ¢lm can be cast from the resulting surfactantpolymer conjugate. Films are structured in multiple bilayers (cf. Fig. 15), with polyions associated with head groups.

Designing Functional Biomolecular Films

FIG. 15

37

Idealized model of a cast multibilayer surfactant film on a surface.

Proteins incorporated into stacked lipid bilayers can lead to larger concentrations of redox sites per unit electrode area, resulting in larger voltammetric peaks. Self-assembled ¢lms are made by casting from aqueous vesicle dispersions containing the protein. Vesicles are closed globular structures. They are essentially water pools surrounded by a bilayer of lipid. When these structures dry after being cast onto a solid surface, they £atten, resulting in stacks of bilayers. Metalloproteins in such ¢lms have given reversible cyclic voltammetry, as illustrated by a CV of the bacterial iron heme enzyme cyt P450cam in ¢lms (ca. 0.5 mm) of dimyrystoylphosphatidylcholine (DMPC) (Fig. 16) [60]. This CV clearly shows nearly symmetric peaks for FeIII reduction on the forward scan and FeII oxidation on the reverse scan. A Soret absorption band at 447 nm for the FeIIcarbon monoxide (CO) complex of cyt P450cam in these ¢lms showed that the enzyme was in its native state. CV midpoint potentials shifted by about þ60 mV when the cyt P450FeII-CO complex was formed during CV scans with CO in the solution. Bare PG electrodes in cyt P450cam solutions or DMPC ¢lms in bu¡er gave no CV peaks (Fig. 16, b and c). This ¢gure helps to illustrate general

38

Rusling and Zhang

FIG. 16 Cyclic voltammograms on basal-plane PG electrodes at 0.1 V=s in pH 7 buffer: (a) cyt P450cam-DMPC film in oxygen-free buffer containing no enzyme; (b) bare electrode in oxygen-free buffer containing 40 mM cyt P450cam; (c) DMPC film in oxygen-free buffer containing no enzyme. (Adapted with permission from Ref. 60. Copyright 1997, Royal Society of Chemistry).

features of voltammetry of metalloproteins in lipid ¢lms and many of the other types of ¢lms discussed in this chapter, i.e., such ¢lms ‘‘turn on’’direct protein voltammetry. Many metalloproteins do not show direct electrochemical reductionoxidation peaks on bare solid electrodes. For example, quasireversible CVs of cyt P450cam in solution were found only with highly pure enzyme at temperatures near 0 C [150]. In contrast, cast surfactant ¢lms containing cyt P450cam give reversible CVs and were stable for over a month when stored in bu¡er at 5 C. Further, in solution at these temperatures the enzyme would have fully denatured in several days. Solution pH controls the E00 values of cyt P450cam in these ¢lms [60,64], which were considerably positive of solution values [151]. Soluble proteins can also be taken up from bu¡ers into cast liquid crystal ¢lms of surfactants on electrodes [152154]. For example, Mb in pH 5.5 bu¡er was taken up into DDAB ¢lms, giving chemically reversible voltammetry for the heme FeIII=FeII redox couple of Mb [152]. DDAB and Mb are both positive at pH 5.5, so hydrophobic interactions are probably important for ¢lm stability. However, ¢lms made by mixing the protein

Designing Functional Biomolecular Films

39

solution with a vesicle dispersion and casting are preferable because it is easier and faster, provides a known amount of protein in the ¢lm, and gives ¢lms that are slightly better ordered [155]. Both techniques lead to comparable voltammetry. CVs of thin proteinsurfactant ¢lms (i.e. 1mm) at scan rates <0.5 V=s typically show nonideal thin layer behavior, with peak current proportional to scan rate. However, CV peaks for 20 mm Mb-DDAB ¢lms (Fig. 17) have a characteristic ‘‘di¡usional tail’’ consistent with the in£uence of di¡usion of charge within the ¢lm. Reductionoxidation peak separations were about 100 mV at 0.1 V=s, and peak current was proportional to the square root of scan rate consistent with di¡ustion-controlled ¢lm voltammetry. The slope of the linear Ip versus u1=2 plot gave a charge transport di¡usion coe⁄cient (Dct) of 4  107 cm2=s [152]. This may re£ect physical di¡usion of the protein in the ¢lm, or ‘‘electron hopping’’ by electron self-exchange reactions between Mb redox sites [50]. When 20-mm DDAB ¢lms containing no protein were placed into a Mb solution, periodic voltammetric scans showed that the protein arrived at the electrode in under 10 s. Using the Dct of

FIG. 17 Cyclic voltammograms at 100 mV=s and 25 C: (a)pH 5.5 buffer containing no protein on a bare PG electrode; (b) 25 mM Mb purified by ultrafiltration in buffer on bare PG; (c) 20 mm Mb-DDAB film on PG electrode in buffer, no Mb in solution. (Adapted from Ref. 153 with permission. Copyright 1995, American Chemical Society.)

40

Rusling and Zhang

4  107 cm2=s, the relation between molecular displacement D, D, and di¡usion time t in Eq. (5) predicts t ¼ 5 s, consistent with the arrival time of Mb at the electrode.This suggests that physical movement of Mb through the DDAB ¢lms transport the charge. Mass and charge transport within lipid ¢lms depends on their phase. For example, transport studies like those described above showed that, at 25 C, Mb di¡uses rapidly through lamellar liquid crystal ¢lms of DDAB and phosphatidylcholines, but not through ¢lms of solid-like gel-state DHP (cf. Fig. 14) [53]. DDAB ¢lms are in the lamellar liquid crystal phase at 25 C. The £uidity of this phase facilitates the movement of the protein during voltammetry. In thick ¢lms, normal pulse voltammetry (NPV) limiting currents are a direct measure of Dct1=2 [49]. NPV limiting currents for gel-state Mb-DDAB ¢lms at temperatures below the gel-to-liquid crystal phase transition temperature (Tc) were small. As the temperature was increased, the limiting current increased in a sigmoid shape (Fig. 18) consistent with a transition from the gel to liquid crystal phase.The in£ection point of this Ilim versus T curve occurs near the Tc of about 12 C measured by di¡erntial scanning calorimetry. Sigmoid-shaped Ilim versus T curves are not observed for very thin ¢lms under conditions where di¡usion does not in£uence the voltammogram. Atomic force microscopy images also suggested that Mb inserts into DDAB bilayers [94,95]. A potential-driven phase transition in DDAB ¢lms was uncovered by AFM which may cause movement of DDAB microstructures and possibly aid in-¢lm transport of Mb during CV scans. Unlike Mb-DDAB, ¢lms of phosphatidylcholines thin to 0.5 mm after soaking in bu¡er for about 1 h, but then remain stable for weeks. This may occur because phosphatidylcholines are slightly more water-soluble that the very insoluble DDAB. Midpoint potentials of the CVs of Mblipid and cyt P450camlipid ¢lms shifted negative with increasing solution pH. The properties of the proteins in these ¢lms are controlled by the pH and salt content of the external solution. Mb gave direct, reversible heme FeIII=FeII electron transfer in a variety of ¢lms of double-chain or triple-chain surfactants [64,156] on PG, Au, ITO, or Pt electrodes. None of these electrodes gave signi¢cant voltammetric peaks in nominally pure Mb solutions. However, quasi-reversible peaks developed after several hours of repetitive scanning of PG electrodes in Mb solutions [157]. Adsorption of proteins and macromolecular impurities on bare electrodes can inhibit electron exchange between the protein of interest and the electrode. X-ray photoelectron spectroscopy, surface infrared spectroscopy, and voltammetric studies detected the presence of such macromolecular

Designing Functional Biomolecular Films

41

FIG. 18 Influence of temperature on limiting current from normal pulse voltammetry (pulse width 10 ms) for Mb-DDAB films on PG in pH 5.5 buffer. (Adapted from Ref. 153 with permission. Copyright 1994, American Chemical Society.)

adsorbates on electrodes in Mb solutions [154]. However, competitive adsorption on an electrode surface between surfactant, protein, and macromolecular impurities in the protein solution is always won by the surfactants when present in large enough amounts. This is true in cast ¢lms of insoluble surfactants or in a 0.1 M solution of a soluble ionic surfactant. The adsorbed layer of ordered surfactant or lipid inhibits macromolecule adsorption, and maintains a viable path for electron exchange between the electrode and the protein. The type of surfactant greatly in£uences E00 of the protein in these ¢lms. Values in DDAB ¢lms are the most positive and closest to reported solution E00 s for Mb [7], although still about 50 mV negative of solution values. The most negative E 00 s for Mb were found in ¢lms of the zwitterionic phosphatidylcholines (DMPC and DLPC).Values in DDAB ¢lms are about

42

Rusling and Zhang

100 mV positive of those in DMPC ¢lms. Similar trends were observed for hemoglobin (Hb). Thus, surfactant head group charge may be important in determining E00. Electrode material also had a strong in£uence on E00. For Mb in DDAB ¢lms, E00 values ranged from 50 to þ 120 mV versus NHE at pH 5.5 in the order ITO < Au < PG < Pt [7]. The apparent rate constant k s depended weakly on electrode material. The in£uence of surfactant type and electrode material on E00 s suggests a possible electrical double-layer e¡ect at the electrode¢lm interface on the potential actually felt by the protein. Surfactantprotein interactions may also be important. Re£ectance-absorbance FT-IR (RAIR) and visible spectroscopy showed that Mb in the DDAB and lipid ¢lms had a nearly native conformation in the medium pH range, and that partial unfolding occurred at pH 5. Low-angle X-ray di¡raction con¢rmed that the sufactant ¢lms are arranged in bilayers [7,64]. The Bragg di¡ractions for Mbsurfactant ¢lms were generally consistent with the width of a bilayer with tilted hydrocarbon tails. Gel-to-liquid crystal phase transition temperatures (Tc) within several degrees of values for bilayer vesicle dispersions of the same surfactant also suggest that surfactants in the ¢lms are arranged in bilayers. Electron spin resonance (ESR) anisotropy and linear dichroism of Mb-DDAB ¢lms showed that the iron heme group was oriented in surfactant ¢lms, but with relatively broad angular distributions [155]. Mb hemes were oriented at angles about 60 to ¢lm normal, regardless of head group charge or surfactant type, suggesting that orientation does not depend on coulombic interactions. The protein may be partly imbedded in the hydrocarbon tail regions of the bilayers and depend on hydrophobic interactions for orientation and retention. NMR showed that anions such as chloride and acetate bind strongly to Mb within the ¢lms [158], neutralizing some of the surface charge and possibly allowing the protein to reside in hydrophobic regions more easily. A simpli¢ed model of Mbsurfactant ¢lms based on characterization studies is shown in the ¢fth entry in Table 1. The cast surfactant ¢lm technique appears general, and a variety of other metalloproteins have given chemically reversible voltammetry in these ¢lms. They include hemoglobin [159161], cytochromes c, c3, and c553 [162164], ferredoxins (ironsulfur proteins) [165169], M. tuberculosis catalase-peroxidase, and other peroxidases [51]. The most complex protein studied in lipid ¢lms has probably been the very large reaction center protein (RC) from the purple bacterium Rhodobacter sphaeroides, which contains multiple bound redox cofactors. Cyclic voltammetry of thin ¢lms made from RC and dimyristoylphosphatidyl

Designing Functional Biomolecular Films

43

choline on PG electrodes in bromide-free bu¡ers at 4 C revealed a reproducible, chemically irreversible oxidation peak at 0.98 V and a reduction peak at 0.17 V versus NHE [170]. The reduction peak disappeared for RCs with the quinones removed, suggesting that this peak represents reduction of quinone cofactors. The oxidation peak showed a catalytic current increase in the presence of ferrous cytochrome c, and decreased by 85% when illuminated by visible light, suggesting assignment to the primary electron donor cofactor which should display these features. On gold electrodes, RClipid ¢lm voltammetry was complicated by halide ion adsorption [171]. Polyionsurfactant ¢lms have also been used to incorporate proteins on electrodes. The polyionsurfactant complexes are dispersed in water by sonication, and protein is added to this solution before casting. Myoglobin was incorporated into cast and LB ¢lms of several polyionsurfactant complexes on PG electrodes. Reversible voltammetry in the liquid crystal phase, and characteristic gel-to-liquid crystal phase transitions were observed in all cases [126]. Similar to the simple surfactant ¢lms, formal potentials depended on surfactant structure and charge. Mb was also studied in composite ¢lms of DDAþ and polyacrylate  (PA ) [172]. Electrochemical results and a Tc of 18 C compared of 11C for DDABMb ¢lms suggest slightly less £uidity in DDAPAMb ¢lms compared to DDABMb. Electrochemical sensors based on antigenantibody interactions show great promise for biomedical and biohazard detection. In a study optimizing surfactantpolyion ¢lms for such uses, electrical impedance was used to show that molecular organization of ¢lms of dioctadecyldimethylammonium bromide with poly(styrenesulfonate) (PSS) or Na¢on on pyrolytic graphite electrodes was improved by annealing in water [173]. Films made by dipping negatively charged Na¢on-coated electrodes into DODAB dispersions had similar phase-transition temperatures to water-annealed cast DODABNa¢on ¢lms. Impedance changes upon binding of bovine serum albumin (BSA) in solution to anti-BSA in DODABNa¢on ¢lms suggested that the method of preparation and optimization of the amount of cholesterol in the ¢lm were important for applications of impedance-based detection of antigenantibody binding [173]. Exfoliated clay platelets can serve as inorganic counter anions for cationic surfactants such as DDAþ . Films were made by casting aqueous dispersions of Hb and DDAB-clay composite onto PG electrodes [174]. Cyclic voltammetry gave chemically reversible peaks at a midpoint potential of 0.06 V versus NHE at pH 7. Only the inner layers of the ¢lm were electrochemically active, and Hb was not completely inserted between clay platelets. These ¢lms are stable upon storage for over 2 months.

44

8

Rusling and Zhang

FILMS CONSTRUCTED LAYER BY LAYER

Preparation of biomolecule ¢lms by constructing a single layer at a time provides excellent control over thickness, and enables the design of ¢lm architecture according to the preconceived plans of the builder. Films containing one or two layers of enzyme that are less than 10 nm thick are easily made. Thicker ¢lms can be made by adding more layers. Also, several different enzymes can be incorporated and spatially separated on the normal axis of the ¢lm in this way. One modern approach utilizes the highly selective binding of antigens with antibodies. Enzymes linked to antibodies can be bound to antigen layers adsorbed onto electrodes [175,176]. Additional layers of enzyme can be added with similar strategies. An advantage is high enzyme activity and orientation. A disadvantage is that the enzyme conjugates need to be synthesized and puri¢ed. Alternate adsorption of monolayers of biomolecules and polyions is a more general method that has been developed over the past decade by Lvov, Decher, and others [177179]. This technique has been used to make ultrathin ¢lms of a wide variety of proteins and oppositely charged polyions [178181]. The procedure is illustrated for a construction of a ¢lm on a negatively charged solid (Fig. 19). The initial negative charge might be obtained on a metal oxide by treatment with base, or on gold or silver by treatment with an alkanethiol terminating in a sulfonate or carboxylic acid group. This negatively charged solid is then immersed into a 13-mg=mL solution of positively charged polyions. The polycations adsorb at monolayer coverage in about 1520 min, e¡ectively reversing the charge on the solid surface. The solid is rinsed in water, then immersed in a 13-mg=mL solution of negatively charged proteins. The negative surface charge on the protein is stabilized by dissolving it in a bu¡er of pH larger than the protein’s isoelectric point. A monolayer of protein is adsorbed from this solution, and the surface develops a negative charge. Both adsorption steps can be repeated many times to obtain the desired number of reproducible alternating layers in a multilayer assembly. Film growth can be monitored during or after each adsorption step with quartz crystal microbalance (QCM) weighing, surface plasmon resonance, spectroscopy, or voltammetry. Figure 19 also shows a conceptual picture of layer-by-layer ¢lms. Nearest-neighbor protein and polyion layers are somewhat intimately mixed, as documented for ¢lms of linear polycations and polyanions [177], but ¢rst and third (or fourth) layers are spatially separated. Most enzymes retain good activity in these ¢lms [178]. Some polyions that have been used to make such ¢lms are shown above. They include metal oxide nanoparticles and DNA.

Designing Functional Biomolecular Films

45

Alternate layer-by-layer adsorption has been used to provide ultrathin ¢lms of redox proteins myoglobin and cyt P450cam featuring direct electron exchange with electrodes [68]. Films made on gold electrodes required an undercoating of mercaptopropanesulfonate (MPS) to facilitate direct voltammetry of the proteins. MPS chemisorbs via AuS bonds, placing a layer of negative sulfonate groups at the electrodesolution interface to adsorb the ¢rst layer of polycation or cationic protein. Chemically reversible cyclic voltammograms were found for the FeIII =FeII redox couple of the proteins in ¢lms on smooth gold, as illustrated for a ¢lm containing a monolayer of cyt P450cam (Fig. 20). Films containing Mb or cyt P450 on smooth vapor-deposited gold electrodes had only about 1.31.5 electroactive layers. The number of electroactive layers was greatly increased by growing ¢lms on mechanically roughened PG electrodes, which may provide a disorder-inducing template

46

Rusling and Zhang

FIG. 19 Illustration of the method of alternate layer-by-layer adsorption for constructing polyionprotein films.

Designing Functional Biomolecular Films

47

for the ¢lm that could decrease average interprotein distance and enhance electron transport by ‘‘electon hopping’’ between redox sites in the ¢lms. Also, when polyions were deposited from solutions with relatively high salt concentrations, where they are coiled rather than linear, much more protein was adsorbed in the subsequent deposition step. Thus, ¢lms constructed on rough PG electrodes by adsorbing PSS from 0.5 M NaCl and Mb from dilute pH 5.5 bu¡er gave up to 7 electroactive Mb layers, as illustrated by CV (Fig. 21) [182]. Similar ¢lms on smoothly polished PG electrodes showed a saturation in electroactivity after three or four layers of Mb. Electrode roughness seems to be important in obtaining optimal electroactivity. Films were stable for several months upon storage dry or in bu¡er solutions. Peak currents of Mb and cyt P450cam in these ¢lms were proportional to scan rate at n <1V=s, but peak widths were much larger than 90 mV, suggesting nonideal thin-layer electrochemical behavior. Reversible voltammetry was also found for ¢lms of polycations and putidaredoxin, the natural bacterial ferredoxin redox partner of cyt P450cam [169]. SWVdata on {PSS (0.5 M NaCl)=Mb}6 ¢lms constructed on rough PG analyzed with the E0dispersion model gave an average ks of 53 s1, in the same range as Mbsurfactant and cast Mbpolyion ¢lms [182]. An average

FIG. 20 Cyclic voltammogram at scan rate 0.5 V=s of PEI=PSS=Cyt P450cam film on a AuMPS-coated QCM resonator in pH 7 buffer. (Adapted from Ref. 68 with permission. Copyright 1998, American Chemical Society.)

48

Rusling and Zhang

FIG. 21 Background-subtracted cyclic voltammograms in pH 5.5 buffer at 0.3 V=s of rough PG electrode coated with {PSS(0.5 M Nacl)=Mb}n films with n ¼ 1,4,6, and 7. (Adapted from Ref. 182 with permission. Copyright 2000, American Chemical Society.)

E 00 of 0.085 V versus NHE was similar to values for Mb in cast phospholipid and Eastman AQ ionomer ¢lms. The mechanical stability of ¢lms constructed layer by layer in £owing or stirred solutions is much better than that of cast surfactant ¢lms, and so these layered ¢lms will probably be more useful for device applications. Layered polyionprotein ¢lms have storage stability of several months as long as the protein is table. A quartz crystal microbalance (QCM) can be used to monitor ¢lm formation [178,179]. The frequency of a QCM resonator decreases in direct proportion to the mass on its metal coating, provided the visoelasticity of the interface does not change [183]. QCM is typically sensitive to 1 ng. In-situ QCM measurements on MPS-coated gold=quartz resonators showed that adsorbed saturation for protein and polyion layers is reached in 1520 min. Estimates of the weight of each layer, and of the repeatability of the multiple adsorption steps, can be obtained from QCM of dry ¢lms. Drying minimizes bias from interfacial viscoelasticity changes. For 9-MHz quartz resonators, mass per unit M =Aðg=cm2 Þ of the ¢lm layer is related to the QCM frequency shift DF (Hz) by [180] M =A ¼ DF =ð1:83  108 Þ

ð21Þ

where the area is that of the metal disk on the quartz resonator in square centimeters. Direct scaling between DF and nominal ¢lm thickness ðdÞ is given by

Designing Functional Biomolecular Films

49

dðnmÞ  ð0:016 0:002ÞDF ðHzÞ

ð22Þ

This expression was con¢rmed by comparison with high-resolution crosssectional images from scanning electron microscopy of ¢lms made from a variety of proteins and polyions [180]. Measurements of QCM frequency changes after each layer of the ¢lm is adsorbed are illustrated for a multilayer assembly of Mb and DNA on a MPSAu surface (Fig. 22). The ¢lm was dried in a stream of nitrogen before each measurement. Frequency decreasing linearly and at regular intervals throughout the multiple adsorption steps demonstrates repeatable adsorption. From Eq. (22), the thickness of this ¢lm of 18 layers is 95 nm. The fraction of electroactive protein can be obtained by dividing the electroactive mass of protein obtained from integrating slow-scan CVs on the gold resonator by the total mass measured by QCM. Negatively charged metal oxide nanoparticles can also be used with positively charge proteins for layer-by-layer construction of ¢lms. For example, Mb was adsorbed from pH 5.2 bu¡er,where it is positively charged, and SiO2 nanoparticles of 45 nm diameter were adsorbed from weakly basic solutions onto a bed of PSS=PDDA adsorbed separately from 0.5 M NaCl

FIG. 22 QCM frequency changes for dry films at each step of assembly of a Mb=DNA film on a smooth AuMPS underlayer. DNA was double-stranded calf thymus (ds-CT). Measurements after adsorption cycles 911 were omitted. (Adapted from Ref. 68 with permission. Copyright 1998, American Chemical Society.)

50

Rusling and Zhang

FIG. 23 Cyclic voltammograms at 0.3 V=s in pH 5.5 buffer of PG=PSS= PDDA=SiO2(Mb=SiO2)n for films with n ¼ 2,5, and 9. (Adapted from Ref. 184 with permission. Copyright 2000, American Chemical Society.)

onto rough PG. CV peak currents increased with layer number for up to nine bilayers of protein and SiO2 nanoparticles (Fig. 23) [184]. Masses measured by QCM during formation of the SiO2Mb ¢lms showed that there was enough protein in the ¢lm to completely coat each nanoparticle. Similar ¢lms were made from Mb and 2050-nm MnO2 nanoparticles. Full protein coating of each particle may provide connecting pathways of closely packed Mb molecules which could facilitate propagation of charge within the ¢lm by a electron hopping mechanism. As shown in Fig. 22, DNA can be used as a polyanion in the layer-bylayer method. Films of DNA and Mb or cyt P450cam also gave chemically reversible voltammetry. CVs were similar to those in Figs. 20, 21, and 23 [121]. The amount of electroactive protein in layered ¢lms depends on the materials used. Figure 24 compares the number of moles of electroactive Mb per unit area on rough PG electrodes determined by integrating CVs for ¢lms made using nanoparticles, PSS adsorbed from 0.5 M NaCl, and ds-DNA. PSS and DNA gave larger amounts of Mb per unit area, but no further

Designing Functional Biomolecular Films

51

FIG. 24 Influence of number of bilayers deposited on the amount of electroactive myoglobin in film constructed on rough PG with various polyanions. PSS and dsCT DNA were adsorbed from 0.5 M NaCl solutions.

increase in electroactivity were found for more bilayers than 7. Increase in protein electroactivity extended to 10 bilayers with the MnO2 and SiO2 nanoparticles, but smaller total amounts of Mb were obtained in these ¢lms. These latter ¢lms have thicknesses of 350 nm for SiO2=Mb ¢lms and 30 nm for MnO2 =Mb ¢lms, and charge transfer through these ¢lms probably follows an ‘‘electron hopping’’ pathway. The MnO2=Mb ¢lms are thinner for nominally similar numbers of bilayers because of competitive adsorption during layer growth [184]. QCM and AFM were used to characterize layers of polyions and cyt P450s and their binding to protein redox partners during layer-by-layer ¢lm growth [185]. When applied as the ¢rst adsorbed layer on a silver-coated QCM resonator, both cationic PEI and anionic PSS formed polymer islands featuring globular structures of about 7 and 11 nm in diameter, respectively, as visualized by AFM. By the time the fourth alternate layer of PEI or PSS had been adsorbed, the polymer islands merged and protein adsorption as a ¢fth layer was visualized on a nearly continuous, relatively smooth ¢lm. Cyt P450cam and cyt P450 2B4 appeared as globules on the polymer underlayers

52

Rusling and Zhang

with diameters of about 10 nm. In general, cyt P450s have nonuniform charge distributions featuring negative and positive ends of the macromolecule, and could be oriented with either charged end up (i.e., positive or negative) by controlling the charge sign of the underlying polyion layer [185]. 9

DNA DAMAGE DETECTION WITH FILMS

Electrochemical sensors coated with stable DNA ¢lms could facilitate rapid in-vitro electrochemical detection of DNA damage by toxic metabolities.On solid electrodes, oxidation of guanine bases is a viable target as guanine is the most easily oxidized base. As mentioned above, cast ¢lms of ds-DNA and ionomers Eastman AQ and Na¢on on graphite electrodes were used to detect DNA damage by direct oxidation using derivative square-wave voltammetry [13,121]. Oxidation peaks developed as the ¢lm was incubated with styrene oxide,which forms known covalent adducts with guanine and adenine bases in DNA. The incubation method was shown by capillary electrophoresis to generate damaged DNA bases under the conditions used. In related work, damage from ionizing radiation to DNA adsorbed onto carbon electrodes was detected by chronopotentiometry [186]. Disadvantages of direct electrochemical oxidations include poor signal-to-noise ratios and data analysis which requires derivative or other background corrections. Catalytic methods of DNA oxidation mentioned in Sec. 4.5 have been applied to improve signal to noise in SWV detection of calf thymus (CT) dsDNA damage in ¢lms [187]. Layer-by-layer growth was used to make ¢lms of DNA and the polycation PDDA. The best sensitivity was found with dsDNA as the outer layer of the ¢lm and at low salt concentration.QCM on gold resonators showed a linear change in average frequency with addition of multiple layers of PDDA and CT ds-DNA (Fig. 25). Error bars illustrate the typical reproducibility obtainable in forming the layers. Films on oxidized PG electrodes of structure (PDDA=ds-DNA)2 were used of analytical purposes. From the QCM results and Eqs. (21) and (22), the average thickness was 6 nm with 0.23-mg ds-DNA per ¢lm. When (PDDA=ds-DNA)2 ¢lms were reacted with styrene oxide, the SWV oxidation peaks increased with the incubation time (Fig. 26). Average peak current for catalytic SWV oxidation of the ds-DNA ¢lms using 50-mM 0 Ru(bpy)2þ 3 (bpy=2,2 -bipyridine) as the catalyst in solution increased with incubation time for the ¢rst half-hour, then decreased slightly (Fig. 27). Error bars re£ect electrode-to-electorde variability. Incubation of ¢lms with toluene, for which no chemical reactions with DNA have been reported, gave catalytic oxidation peaks that remained within electrode-to-electrode variations of controls (Figs. 26 and 27). Catalytic SWV oxidation with Ru(bpy)2þ 3 provided more sensitive detection of DNA damage than direct

Designing Functional Biomolecular Films

53

FIG. 25 QCM frequency shifts for cycles of alternate PDDA=DNA adsorption on gold resonators coated with mixed monolayers of mercaptopropionic acid and mercaptopropanol. (Average values for eight resonators). (Adapted from Ref. 187 with permission. Copyright 2001, American Chemical Society.)

SWV oxidation. Studies of DNA and polynucleotides in solutions and ¢lms suggested that oxidation of guanine and chemically damaged adenine in partly unraveled, damaged DNA were the most likely contributors to the catalytic peaks [187]. An alternative method to monitor DNA damage in layered ¢lms involves using an electroactive cation probe that binds better to ds-DNA than to damaged DNA.This approach was used with Co(bpy)3þ 3 as the probe for ¢lms of (PDDA=ds-DNA)2 grown layer by layer on PG electrodes ¢rst coated with a layer of PSS [188]. After incubation of ¢lms of (PDDA=dsDNA)2 with saturated styrene oxide at 37 C in a separate reactor for various times, electrodes were rinsed, placed into 20 mM Co(bpy)3þ 3 , and analyzed by SWV. The peak for a ¢lm before incubation is shown in Fig 28a, labeled ‘‘0 min.’’ When the ¢lm was incubated with styrene oxide, peak current

54

Rusling and Zhang

FIG. 26 SWV of (PDDA=ds-DNA)2 films on oxidized PG in pH 5.5 acetate buffer containing 50 mM NaCl, with 50 mM RuðbpyÞ2þ 3 in buffer, and after incubations at 37 C with styrene oxide (SO) or toluene. Controls were incubated in buffer only. (Adapted from Ref. 187 with permission. Copyright 2001, American Chemical Society.)

decreased with increasing time because of the decreasing ability of the damaged outer layer of DNA to bind Co(bpy)3þ 3 . Control incubations with toluene or in bu¡er alone showed no trends in peak current with incubation time (Fig. 28b). An advantage of this method is that one electrode can be used to measure a single incubation time course. Denoting the initial SWV peak current as Ip;initial , and the peak current after incubation as Ip;final , plots of the peak current ratio, Ip;initial =Ip;final versus incubation time (Fig. 29) can be used to monitor rates of DNA damage. Use of this ratio helps correct for electrodeto-electrode variations in Ip;initial . Increase in Ip;initial =Ip;final were found for ds-DNA electrodes with increasing incubation time in styrene oxide solu-

Designing Functional Biomolecular Films

55

FIG. 27 The influence of incubation time with styrene oxide and toluene on the average catalytic peak current (corrected for controls) for 515 trials per data point for (PDDA=ds-DNA)2 films. Error bars represent standard deviations. (Adapted from Ref. 187 with permission. Copyright 2001, American Chemical Society.)

tions. No signi¢cant variations were found for controls with or without toluene. 10

PROSPECTS FOR THE FUTURE

One of the main themes of this chapter is the excellent recent progress that has been made in preparing stable electroactive ¢lms containing biomolecules. Several types of ¢lms can now facilitate direct electron exchange between native enzymes and electrodes, an achievement which in most cases was only a dream as little as 20 years ago. Films can also be designed to catalyze enzyme reactions with an electrochemical driving force, either for speci¢c reactions by covalent bonding strategies, or with more general approaches based on monolayer or multilayer co-adsorption with polyions or surfactants. Films of DNA can be designed for applications such as damage analysis or hybridization. Taken together, the ¢lms described here

56

Rusling and Zhang

FIG. 28 SWV difference currents in 20 mM CoðbpyÞ3þ 3 at pH 5.5 of (PDDA=dsDNA)2 films after: (a) incubations with syrene oxide; (b) control incubations with toluene. (Adapted from Ref. 188 with permission. Copyright 2002, Wiley-VCH.)

provide complementary alternatives for fundamental studies of biomolecules, and for device development. What might we expect in the future from this research area? We predict signi¢cant progress in applications of enzyme and DNA ¢lms to biosensors, bioreactors, biomedical devices, and fundamental biochemical

Designing Functional Biomolecular Films

57

FIG. 29 Influence of incubation time on average SWV peak current ratios for (PDDA=ds-DNA)2 films in 20 mM CoðbpyÞ3þ at pH 5.5 after incubations with 3 styrene oxide, toluene, or pure buffer. Ifinal corresponds to the peak current after each incubation; Iinitial corresponds to peak current before incubation. (Adapted from Ref. 188 with permission. Copyright 2002 Wiley-VCH.)

research. We could envision future biosensors that can detect physiological abnormalities from analysis of blood, urine, sweat, or even from noninvasive analyses that make measurement through the skin or in the saliva. Electrode arrays may be developed to provide many biomedical tests simultaneously. It is very likely that biosensors will be included in feedback loops to control implanted biomedical devices. Testing patients for genetic diseases via mutated DNA hybridization assays may be possible by using enzyme tags that develop signals via direct electron transfer and electrochemical catalysis. Future sensors may be constructed to detect toxicity of metabolites of organic pollutants and drugs. Simple biosensors could be developed to detect hazardous biological materials such as pathogenic bacteria.

58

Rusling and Zhang

In closing, the area of electroactive biomolecular ¢lms is a good example of how fundamental research leads to technological advances that bene¢t our society.What began purely as fundamental laboratory endeavors only 20 or so years ago has opened the door to a wide variety of useful devices that promise to come to fruition in the next few decades.The ¢rst success story in this ¢eld has been the advent of electrochemical enzyme-based glucose sensors for home use by diabetic patients [8].We expect and hope that this is just the tip of the iceberg. ACKNOWLEDGMENTS The author’s research described herein was supported by Grant ES03154 from National Institute of Environmental Health Sciences (NIEHS), National Institutes of Health (NIH), and the U.S. Department of Agriculture (USDA) through Grant 99-35306-7609. Contents are solely the responsibility of the author and do not necessarily represents o⁄cial views of NIEHS, NIH, or USDA.The author is also grateful to students and colleagues named in joint publications whose valuable contributions made the work possible. REFERENCES 1. 2. 3. 4. 5. 6. 7.

8. 9. 10. 11. 12. 13. 14.

J Deisenhofer, H Michel. Science 245:14631473, 1989. G Feher, JP Allen, MY Okamura, DC Rees. Nature 339:111116, 1989. RE Blankenship, MT Madigan, CE Bauer, eds. Anoxyygenic Photosynthetic Bacteria. Dordrecht: Kluwer, 1995. JB Schenkman, H Greim, eds. Cytochrome P450. Berlin: Springer-Verlag, 1993. PR Ortiz de Montellano, ed. Cytochrome P450. 2nd ed. New York: Plenum, 1995. A Kotyk, K Janacek, J Koryta. Biophysical Chemistry of Membrane Function. Chichester, UK: Wiley, 1988. JF Rusling, Z Zhang. In: HS Nalwa, ed. Handbook of Surfaces and Interfaces of Materials.Vol. 5. Biomolecules, Biointerfaces, and Applications. San Diego, CA, Academic: 2001, pp 3371. G Ramsay, ed. Commercial Biosensors. New York: Wiley, 1998. SR Mikklesen. Electroanalysis 8:1519, 1996. HH Thorp. Trends Biotechnol 16:117121, 1998. J Wang. Chem Eur J 5:16811685, 1999. E Palacek, M Fojta. Anal Chem 73:74A183A, 2001. J Mbindyo, L Zhou, Z Zhang, JD Stuart, JF Rusling. Anal Chem 72:20592065, 2000. EF Bowden, FM Hawkridge, HN Blount. In: S Srinivasan, YA Chizmadzhev, JO’M Bockris, BE Conway, E Yeager, eds. ComprehensiveTreatise of Electrochemistry.Vol. 10. NewYork: plenum, 1985, pp 297346.

Designing Functional Biomolecular Films 15. 16. 17. 18. 19. 20. 21.

22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.

59

FA Armstrong. In: Bioinorganic Chemistry. Structure and Bonding 72. Berlin: Springer-Verlag, 1990, pp 137221. P Yeh,T Kuwana. Chem Lett 11451148, 1977. MJ Eddowes, HAO Hill. Chem Commun 31543155, 1977. DE Reed, FM Hawkridge. Anal Chem 59:23342339, 1987. EF Bowden, FM Hawkridge, JC Chlebowski, EE Bancroft, C Thorpe, HN Blount. J Am Chem Soc 104:76417644, 1982. FA Armstrong, HAO Hill, BN Oliver. Chem Commun 976977, 1982. GG Guilbault, M Mascini. Analytical Uses of Immobilized Biological Compounds for Detection, Medical, and Industrial Uses. Holland: Reidel, 1988. JM Kau¡mann, GG Guilbault. In: Bioanalytical Applications of Enzymes. Methods Biochemical AnalysisVol. 36. New York: Wiley, 1992, pp 63113. GG Guilbault. Analytical Uses of Immobilized Enzymes. New York: Marcel Dekker, 1984. JD Czaban. Anal Chem. 57:345A356A, 1985. T Yokota, K Itoh, A Fujishima. J Electroanal Chem. 216:289292, 1987. Z Salamon, G Tollin. Bioelectrochem Bioenerg 25:447454, 1991. Z Salamon, G Tollin. Bioelectrochem Bioenerg 26:321334, 1991. FA Armstrong, HA Heering. J Hirst. Chem Soc Rev 26:169179, 1997. MJ Tarlov, EF Bowden. J Am Chem Soc 113:18471849, 1991. EM Boon, DM Ceres,TG Drummond, MG Hill, JK Barton. Nature Biotechol 18:10961100, 2000. F Azek, C Grossiord, M Joannes, B Limoges, P Brossier. Anal Biochem 282:107113, 2000. DJ Caruana, A Heller. J Am Chem Soc 121:769774, 1999. F Patolsky, E Katz, A Bardea, I Willner. Langmuir 15:37033706, 1999. B Singer, D Grunberger. Molecular Biology of Mutagens and Carcinogens. New York: Plenum, 1983. W. Pauwels, P Vodiceka, M. Servi, K. Plna, H Veulemans, K Hemminki. Carcinogenisis 17:26732680, 1996. EE McConnell, JA Swenberg. CRC Crit RevToxicol 24:S49S55, 1994. ER Nestmann, DW Bryant, CJ Carr, TT Fennell, NJ Gorelick, JE Gallagher, JA Swenberg, GM Williams. Regul Toxicol Pharmacol 24:918, 1996. E Palecek. Electroanalysis 8:714, 1996. E Palecek, P Boublikova, F Jelen. Anal Chim Acta 187:99107, 1986. F Jelen, M Fojta, E Palecek. Electroanal Chem 427:4956, 1997. F Jelen, M Tomschik, E Palecek. J Electroanal Chem 423:141148, 1997. M Fojta, E Palecek. Anal Chim Acta 342:112, 1997. M Fojta, V Stankova, E Palecek, P Koscielniak, J Mitas. Talanta 46:155161, 1998. J-M Sequaris, P Valenta, HW Nurnberg. Int J Radiat Res 42:407415, 1982. J-M Sequaris, P Valenta, J Electroanal Chem 227:1120, 1987. J Wang, X Cai, J Wang, C Johnsson, E Palecek. Anal Chem 67:40654075, 1995.

60 47. 48.

49. 50.

51. 52. 53. 54. 55.

56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71.

Rusling and Zhang RW Murray. In: AJ Bard,ed. Electroanalytical Chemistry. Vol. 13. New York: Marcel Dekker, 1984, pp 191368. RW Murray. In: RW Murray, ed. Molecular Design of Electrode Surfaces. Techniques of Chemistry Series.Vol. 22. New York: Wiley-Interscience, 1992, pp 148. AJ Bard, LR Faulkner. Electrochemical Methods. 2nd ed. Newyork: Wiley, 2001. M Madja. In: Murray, RW ed. Molecular Design of Electrode Surfaces. Techniques of Chemistry Series. Vol. 22. New York: Wiley-Interscience, 1992, pp 159206. Z Zhang, S Chouchane, RS Magliozzo, JF Rusling. Chem Commun 177178, 2001. TM Nahir, RA Clark, EF Bowden. Anal Chem 66:25952598, 1994. Z Zhang, JF Rusling. Biophys Chem 63:133146, 1997. A-EF Nassar, Z Zhang, N Hu, JF Rusling, TF Kumosinski. J Phys Chem B 101:22242231, 1997. N Oyama, T Ohsaka. In: RW Murray ed. Molecular Design of Electrode Surfaces. Techniques of Chemistry Series.Vol. 22. New York: Wiley-Interscience, 1992, pp 333402. A-EF Nassar, JF Rusling, N Nakashima. J Am Chem Soc 118:30433044, 1996. J Osteryoung, JJ O’Dea. In: AJ Bard, ed. Electroanalytical Chemistry.Vol. 14. New York: Marcel Dekker, 1986, pp 209308. A Mugweru, JF Rusling. Electrochem Commun 3:406409, 2001. JJ O’Dea, J Osteryoung. Anal Chem 65:30903097, 1993. Z Zhang, A-EF Nassar, Z Lu, JB Schenkman, JF Rusling. J Chem Soc Faraday Trans 93:1769-1774, 1997. RA Marcus, N Sutin. Biochem Biophys Acta 811:265322, 1985. TM Nahir, EF Bowden. J Electroanal Chem 410:913, 1996. TM Saccucci, JF Rusling. J Phys Chem B 105:61426147, 2001. JF Rusling. Acc Chem Res 31:363369, 1998. A-E Nassar, JM Bobbitt, JD Stuart, JF Rusling. J Am Chem Soc 117:1098610993, 1995. FA Armstrong. In: G Lenz, G Milazzo. eds. Bioelectrochemistry of Biomacromolecules. Basel, Switzerland: Birkhauser Verlag, 1997, pp 205255. JF Rusling, TF Kumosinski. Nonlinear Computer Modeling of Chemical and Biochemical Data. New York: Academic, 1996, pp 100102. YM Lvov, Z Lu, JB Schenkman, X Zu, JF Rusling. J Am Chem Soc 120:40734080, 1998. X Zu, Z Lu, Z Zhang, JB Schenkman, JF Rusling. Langmuir 15:73727377, 1999. J Kong, JN Mbindyo, X Wu, JX Zhou, JF Rusling. Biophys Chem 79:219229, 1999. Z Zhang, S Chouchane, RS Magliozzo, JF Rusling. Anal Chem 74:114119, 2002.

Designing Functional Biomolecular Films

61

72. S Steenken, SV Jovanovic. J Am Chem Soc 119:617617, 1997. 73. (a) DH Johnston, KC Glasgow, HH Thorp. J Am Chem Soc 117:89338938, 1995. (b) ME Napier, HH Thorp. Langmuir 13:63426344, 1997. (c) BT Farrer, HH Thorp. Inorg Chem 39:4449, 2000. (d) IV Yang, HH Thorp. Inorg Chem 39:49694976, 2000. (e) PM Armistead, HH Thorp. Anal Chem 72:37643770, 2000. (f ) MF Sistare, SJ Codden, G Heimlich, HH Thorp. J Am Chem Soc 122:47424749, 2000. (g) VA Szalai, HH Thorp. J Phys Chem B 104:68516859, 2000. 74. AC Onkto, PM Armistead, SR Kircus, HH Thorp. Inorg Chem 38:18421846, 1999. 75. PM Armistead, HH Thorp. HH Anal Chem 73:558564, 2001. 76. J Hirst, FA Armstrong. Anal Chem 70:50625071, 1998. 77. HA Heering, J Hirst, FA Armstrong. J Phy Chem B 102:68896902, 1998. 78. FA Armstrong, GS Wilson. Electrochim Acta 45:26232645, 2000. 79. JN Butt, A Sucheta, FA Armstrong, J Breton, SJ George, AJ Thompson, EC Hatchikian. J Am Chem Soc 113:89488950, 1991. 80. JN Butt, FA Armstrong, J Breton, SJ George, AJ Thompson, EC Hatchikian. J Am Chem Soc 113:66636670, 1991. 81. J Hirst, JLC Du¡, GNL Jameson, MA Kemper, BK Burgess, FA Armstrong. J Am Chem Soc 120:70857094, 1998. 82. J Hirst, GNL Jameson, JWA Allen, FA Armstrong. J Am Chem Soc 120:1199411999, 1998. 83. JL Breton, JLC Du¡, JN Butt, FA Armstrong, SJ George,Y Petillot, E Forest, G Schafer, AJ Thompson. Eur J Biochem 233:937946, 1995. 84. A Sucheta, BAC Ackrell, B Cochrane, FA Armstrong. Nature 356:361362, 1992. 85. J Hirst, BAC Ackrell, FA Armstrong. J Am Chem Soc 119:74347439, 1997. 86. HA Heering, JH Weiner, FA Armstrong. J Am Chem Soc 119:1162811638, 1997. 87. MS Mondal, DG Goodin, FA Armstrong. J Am Chem Soc 120:62706276, 1998. 88. JP McEvoy, FA Armstrong. Chem Commun 16351636, 1999. 89. K Chen, J Hirst, R Camba, CA Bonagura, CD Stout, BK Burgess. FA Armstrong. Nature 405:814817, 2000. 90. T Ikeda, S Miyaoka, K Miki. J Electroanal Chem 352:267278, 1993. 91. KL Egodage, BS de Silva, GS Wilson. J Am Chem Soc, 119:52955301, 1997. 92. K Draoui, P Bianco, J Haladjian. F Guerlesquin. M Bruschi. J Electroanal Chem 313:201214, 1991. 93. JN Butt, M Filipiak,WR Hagen. Eur J Biochem 254:116122, 1997. 94. S Boussaad, NJ Tao, R Arechabaleta. Chem Phys Lett 280:397403, 1997. 95. S Boussaad, NJ Tao. J Am Chem Soc 121:45104515, 1999. 96. EP Friis, JET Andersen, LM Madsen, N Bonander, P Moller, J Ulstrup. Electrochim Acta 42:28892897, 1997. 97. EF Bowden, Interface 6:4044, 1997.

62

Rusling and Zhang

98. JJ Davis, CM Halliwell, HAO Hill, GW Canters, MC van Amsterdam,Verbeet, M.Ph. New J Chem 11191123, 1998. 99. AN shipway, I Willner. Accts Chem Res 34:421432, 2001. 100. V Heleg-Shabtai, E Katx I Willner. J Am Chem Soc 119:81218122, 1997. 101. A Bardea, E Katz, AF Bˇckmann, I Willner. J Am Chem Soc 119:91149199, 1997. 102. I Willner, V Hegel-Shabtai, E Katz, HK Rau, W Haehnel. J Am Chem Soc, 121:64556468, 1999. 103. F Patolsky, M Zayats, E Katz, I Willner. Anal Chem 71:31713180, 1999. 104. L Alfonta, E Katz, I Willner. Anal Chem 72:927935, 2000. 105. DE Benson, DW Conrad, RM de Lorimier, SA Trammell, HW Hellinga. Science 293:16411644, 2001. 106. BN Oliver, JO Egekeze, RW Murray. J Am Chem Soc 110:23212322, 1988. 107. CR Martin, H Freiser. Anal Chem 53:902904, 1981. 108. MN Szentirmay, CR Martin. Anal Chem 56:18981902, 1984. 109. CAC Sequeira. In: RG Linford, ed. Electrochemical Science and Technology of Polymers 2. London: Elsevier Applied Science, 1990, pp 319374. 110. CEW Hahn, HAO Hill, MD Ritchie, JW Sear. Chem Commun 125126, 1990. 111. P Bianco, J Haladjian. Electroanalysis 6:415422, 1994. 112. P Bianco, J Haladjian. J Electrochem Soc 139:24282432, 1992. 113. P Bianco, J Haladjian S Giannandrea-Derocles. Electroanalysis 6:6774, 1994. 114. JDH Glenn, EF Bowden, Chem Lett 399400, 1996. 115. M Wirz, J Klucik M Rivera. J Am Chem Soc 122:10471056, 2000. 116. P Bianco, A Taye, J Haladjian J Electroanal Chem 337:299303, 1994. 117. N Hu, JF Rusling. Langmuir 13:41194125, 1997. 118. J Yang, N Hu, JF Rusling. Electroanal Chem 463:5362, 1999. 119. E Lojou, P Bianco, M Bruschi. Electrochim Acta 43:20052013, 1998. 120. M Wright, MJ Honeychurch, HAO Hill. Electrochem Commun 1:609610, 1999. 121. JF Rusling, L Zhou, B Munge, J Yang, C Estavillo, JB Schenkman. Faraday Disc 116:7787, 2000. 122. A Kotyk, K Janacek, J Koryta. Biophysical Chemistry of Membrane Function Chichester, UK: Wiley, 1988. 123. J Issraelachvili. Intermolecular and Surface Forces. 2nd ed. San Diego, CA Academic, 1992. 124. DF Evans, DJ Mitchell, BW Ninham. J Phys Chem 90:28172825, 1986. 125. T Yokota, K Itoh, A Fujishima. J Electroanal Chem 216:289292, 1987. 126. A-EF Nassar, Y Narikiyo, T Sagara, N Nakashima, JF Rusling. J Chem Soc FaradayTrans 91:17751782, 1995. 127. HT Tien, Z Salamon. Bioelectrochem Bioenerg 22:211218, 1989. 128. T Martynski, HT Tien. Bioelectrochem Bioenerg 25:317342, 1991. 129. A Guerrieri,TR Cataldi, HAO Hill. J Electroanal Chem 297:541547, 1991. 130. Z Salamon, G Tollin. Bioelectrochem Bioenerg 25:447454, 1991. 131. Z Salamon, G Tollin. Bioelectrochem Bioenerg 26:321334, 1991.

Designing Functional Biomolecular Films 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163. 164. 165. 166.

63

Z Salamon, FK Gleason, G Tollin. Arch Biochem Biophys 299:193198, 1992. Z Salamon, G Tollin. Bioelectrochem Bioenerg 27:381391, 1992. Z Salamon, G Tollin. Photochem Photobiol 58:730736, 1993. Z Salamon, G Tollin. J Bioenerg Biomembranes 29:211221, 1997. G Cheddar, G Tollin. Arch Biochem Biophys 310:392396, 1994. Z Salamon, JT Hazzard, G Tollin. Proc Natl Acad Sci USA 90:64206423, 1993. Z Salamon, G Tollin, M Hirisawa, L Gardet-Salvi, A-L Stritt-Etter, DB Kna¡, P Schˇrmann. Biochim Biophys Acta 1230:114118, 1230. KT Kinnear, HG Monbouquette. Langmuir 9:22552257, 1993. KT Kinnear, HG Monbouquette. Anal Chem 69:17711775, 1997. JK Cullison, FM Hawkridge, N Nakashima, S Yoshikawa. Langmuir 10:877882, 1994. JD Burgess, FM Hawkridge. Langmuir 13:38713786, 1997. JD Burgess, MC Rhoten, FM Hawkridge. Langmuir 14:24672475, 1998. JD Burgess, MC Rhoten, FM Hawkridge. J Am Chem Soc 120:44884491, 1998. N Nakashima, R Ando,T Kunitake. Chem Lett 15771580, 1983. I Hamachi, S Noda,T Kunitake. J Am Chem Soc 113:96259630, 1991. T Kunitake, M Shimomura, T Kajiyama, A Harada, K Okuyama, M Takayanagi. Thin Solid Fims 121:L89L91, 1984. T Kunitake, A Tsuge, N Nakashima. Chem Lett 17831786, 1984. M Shimomura,T Kunitake. Polymer J 16:187190, 1984. J Kazlauskaite, ACG Westlake, L-L Wong, HAO Hill. J Chem Soc Chem Commun 21892190, 1996. SG Sligar, DL Cinti, GG Gibson, JB Schenkman. Biochem Biophys Res Commun 90:925932, 1979. JF Rusling, A-EF Nassar. J Am Chem Soc 115:1189111897, 1993. JF Rusling, A-EF Nassar. Langmuir 10:28002806, 1994. A-EF Nassar,WS Willis, JF Rusling. Anal Chem 67:23862392, 1995. A-EF Nassar, Z Zhang, JF Rusling, V Chynwat, HA Frank, K Suga. J Phys Chem B 99:1101311017, 1995. A-EF Nassar, Y Narikiyo, T Sagara, N Nakashima, JF Rusling. J Chem Soc FaradayTrans 91:17751782, 1995. A Onuoha, JF Rusling. Unpublished results. JF Rusling, A-EF Nassar,TF Kumosinski. Biophys Chem 67:107116, 1997. J Yang, N Hu. Bioelectrochem Bioenerg 48:117127, 1999. Z Lu, Q Huang, JF Rusling. J Electroanal Chem 423:5966, 1997. M Ciureanu, S Goldstein, MA Mateescu. J Electrochem Soc 145:533541, 1998. P Bianco, J Haladjian. J Electroanal Chem 367:7984, 1994. P Bianco, J Haladjian. Electochim Acta 39:911916, 1994. J Hanzlik, P Bianco, J Haladjian. J Electroanal Chem 380:287290, 1995. P Bianco, J Haladjian. Electroanalysis 7:442446, 1995. P Bianco, J Haladjian. Electrochim Acta 42:587594, 1997.

64 167.

Rusling and Zhang

M Tominaga, J Yanagimoto, JF Rusling, AEF Nassar, N Nakashima. Chem Lett 523524, 1996. 168. AEF Nassar, JF Rusling M Tominaga, J Yanagimoto N Nakashima. J Electronal Chem 416:183185, 1996. 169. Z Lu,YM Lvov, I Jansson, JB Schenkman, JF Rusling. J Colloid Interface Sci 224:162168, 2000. 170. B Munge, Z Pendon, HA Frank, JF Rusling. Bioelectrochemistry 54:145150, 2001. 171. Z Gao, HA Frank, YM Lvov, JF Rusling. Bioelectrochemistry 54:97100, 2001. 172. Y Hu, N HuY Zeng. Talanta 50:11831195, 2000. 173. K Abe, RE Schmukler, JF Rusling. Electroanalysis 10:948954, 1998. 174. X Chen, N Hu,Y Zheng, JF Rusling J Yang. Langmuir 15:70227030, 1999. 175. C Bourdillon, C Demaille, J Moiroux, JM Saveant. Accts Chem Res 29:529535, 1996. 176. C Demaille, J Moiroux, JM Saveant, C Bourdillon. In: Y Lvov H M˛hwald, H eds. Protein Architecture: Interfacing Molecular Assemblies and Immobilization Biotechnology. New York: Marcel Dekker, 2000, pp 311335. 177. G Decher. Science 277:12311237, 1997. 178. Y Lvov. In: Y Lvov, H M˛hwald, eds. Protein Architecture: Interfacing Molecular Assemblies and Immobilization Biotechnology. New York: Marcel Dekker, 2000, pp 125167. 179. Y Lvov. In: RW Nalwa ed. Handbook of Surfaces and Interfaces of Materials Vol. 3. Nanostructured Materials, Micelles and Colloids.Academic, CA: San Diego, 2001, pp 170189. 180. Y Lvov, K Ariga, I Ichinose,T Kunitake. J Am Chem Soc 17:61176123, 1995. 181. M Onda,Y Lvov, K Ariga,T Kunitake. J Ferment Bioeng 82:502506, 1996. 182. H Ma, N Hu, JF Rusling. Langmuir 16:49694975, 2000. 183. DA Buttry, MD Ward. Chem Rev 92:13551379, 1992. 184. Y Lvov, B Munge, O Giraldo, I Ichinose, SL Suib, JF Rusling. Langmuir 16:88508857, 2000. 185. JB Schenkman, I Jansson, Y Lvov, JF Rusling, S Boussaad, NJ Tao. Arch Bichem Biophs 385:7887, 2001. 186. J Wang, G Rivas, M Ozsoz, DH Grant, X cai. C Parrado. Anal Chem 69:14571460, 1997. 187. L Zhou, JF Rusling. Anal Chem 73:47804786, 2001. 188. J Yang, Z Zhang, JF Rusling. Electroanalysis, 2002, in press.

2 Electrochemistry of Redox-Active Protein Films Immobilized on Self-Assembled Monolayers of Organothiols Katsumi Niki* and Brian W. Gregoryy Illinois State University, Normal, Illinois, U.S.A.

1

INTRODUCTION

Electrode reactions of proteins and enzymes on bare electrodes usually do not take place at their formal potentials (reactions are irreversible). A new era of modern bioelectrochemistry began in 1977 with voltammetric studies of small protein molecules, and was pioneered by groups led by Kuwana [1], Hill [2], and Niki [3]. From these studies, it was found that modi¢cations of electrode surfaces are imperative to facilitate the direct electrochemistry of biological molecules. The direct electrochemistry of biological molecules not only provides important thermodynamic and mechanistic information from biological reactions, but also allows one to develop ‘‘molecular wiring’’ (electrical communication) between electrochemically inactive biological molecules and electrodes for various advanced applications. The use of surface-modi¢ed electrodes has since become well established in electrochemical studies of small protein molecules such as cytochromes, ferredoxins, and blue copper proteins. Taniguchi’s group ¢rst

*Present a⁄liation: The Beckman Institute, California Institute of Technology, Pasadena, California, U.S.A. y Present a⁄liation: Samford University, Birmingham, Alabama, U.S.A.

65

66

Niki and Gregory

proposed direct electrochemistry of cytochrome c (cyt c) on the pyridinethiol-modi¢ed electrodes,which was found later to form self-assembled monolayers (SAMs) on gold electrodes [4,5]. Since that time, numerous surface modi¢ers have been reported [6,7]. However, the application of chemically modi¢ed electrodes to direct electrochemical studies of biological molecules has still been limited. Among the various surface modi¢ers that have been employed, o-functional alkanethiol SAMs are considered the most attractive for mimicking physiological electron transfer (ET) partners with positively charged protein-binding sites [8]. Electrochemical approaches which utilize such SAMs for investigating ET mechanisms are advantageous in that the driving force of the ET reaction can be regulated through control of the electrode potential, and the ET path length can be varied by using alkanethiols of varying chain lengths. In this regard, the electrode reaction of cyt c through carboxylate-terminated alkanethiol SAMs is considered to be a simpli¢ed model system for biological ET processes. Electrostatic interactions between the positively charged lysine amino groups immediately surrounding the heme crevice of cyt c and the negatively charged carboxylate termini of the SAMs stabilize the binding of cyt c, analogous to its complex with negatively charged physiological redox partners such as the cyt c=cyt b5 complex. Although there are numerous surface modi¢ers (from well-de¢ned SAMs to polymer composites) which facilitate rapid ET reactions of biological molecules (from simple proteins to enzymes containing multiple redox active sites) without electron transfer mediators, this chapter focuses on the fundamental electrochemistry of SAMs and SAM=protein ¢lms. In particular, an emphasis is placed on those SAMs that can be formed on conductive substrates and that have been utilized as templates for protein adsorption. Due to the fact that the vast majority of work in this area has centered primarily on organothiol-based SAMs on metals, this chapter will focus on the use of these particular SAMs in studies of protein binding and adsorption. Therefore, the utilization of organosilane monolayers on conductive oxide substrates (e.g., In2O3) as templates for protein adsorption will only be mentioned here [9]. The ¢rst section of this review will summarize the physicochemical properties and structures of SAMs that have been utilized as surface modi¢ers for binding proteins to electrode surfaces, and will focus on alkanethiol monolayers. The second section will discuss results from some of the redox proteins that have been studied at these modi¢ed surfaces, although the electrochemistry of cyt c will be emphasized. Due to the meteoric rise in popularity of self-assembly over the last 10^20 years and the vast quantity of literature that has followed, this review is not intended to be comprehensive, but rather illustrative of the general, present understanding of both SAMs and the behavior of redox proteins at SAMs. For

Electrochemistry of Redox-Active Protein Films

67

more speci¢c information, the reader is encouraged to seek other excellent and more exhaustive reviews=monographs of SAM methodology and structure [10^15] and faradaic bioelectrochemistry [7]. 2

SELF-ASSEMBLED MONOLAYERS OF ORGANOTHIOLS

Molecular self-assembly has been described as ‘‘the spontaneous association of molecules under equilibrium conditions into stable, structurally wellde¢ned aggregates joined by noncovalent bonds’’ [16,17]. Such noncovalent synthetic routes are directed toward ‘‘designing’’ molecules that spontaneously self-organize into predetermined structures, and are inspired by the multitude of highly functional, complex biological structures found in nature (cell membranes, secondary protein structure, the a-helical structure of DNA, etc.). The rational design of molecular functionality by controlling both the self-organization process and resulting aggregate structure is therefore the ultimate goal of many self-assembly strategies. Due to the wide variety of noncovalent intermolecular interactions that may be utilized (dispersion, hydrogen bonding, dipole ^ dipole, ionic, etc.), self-assembly methods o¡er great potential as versatile alternatives to lithographic techniques. Furthermore, self-assembly methods are expected to be inherently less wasteful, since they allow the construction of organized assemblies from the ‘‘bottom up’’ rather than through the ‘‘top-down’’ approach of lithographic techniques. Self-assembled monolayers have been touted over the past two decades as useful model systems for investigations of the self-assembly process. SAMs usually consist of densely packed molecular units that are bound to the surface through a head group which has a speci¢c interaction with the substrate. The molecular precursors that constitute the ¢lm are typically surfactant-based in structure, and spontaneously assemble and selforganize on the substrate surface during deposition. Deposition is achieved either via vapor-phase dosing of the substrate or by exposure of the substrate to a solution containing the precursor molecule. The most widely investigated types of monolayer assemblies formed via this methodology have been alkane- and arylthiols (and their disul¢de derivatives) chemisorbed on noble metal surfaces [10^15], and organosilicon derivatives on hydroxylated surfaces (e.g., SiO2, Al2O3, SnO2, TiO2) [10]. The usefulness of SAMs in the investigation of a wide variety of important interfacial phenomena (e.g., electron transfer [18,19] and corrosion inhibition [20]) has also been widely exploited [10,11]. Clearly, the ability to modify the chemical and=or physical properties of a surface with molecular-level control has been a major driving force toward understanding both the spatial and electronic structure of these ¢lms. In this section of the chapter, we will brie£y review both the current

68

Niki and Gregory

understanding of the structure and physicochemical properties of these SAMs, including those that are critical in in£uencing protein adsorption. 2.1

Formation of SAMs

The speci¢c a⁄nity of the thiol head group for a wide variety of metal (and some nonmetal) surfaces makes these types of monolayer ¢lms appealing since they can be formed on a fairly wide range of materials. From a synthetic standpoint, this is attractive since a vast array of precursor molecules can be fashioned for potential use in SAM formation simply through chemical attachment of the ^ SH moeity. Since the formation of SAMs is dictated by both intermolecular interactions and those with the substrate, not all thiolderivatized molecules necessarily yield self-assembled ¢lms. In fact, the terms ‘‘self-assembled monolayer’’and ‘‘self-assembled ¢lm’’ have been used excessively for any number of adsorbed ¢lms, regardless of their structural organization. The vast majority of organothiols (and organodisul¢des) that have been studied and utilized for self-assembly work have been those based on alkyl-chain derivatives [e.g., R(CH2)nSH,where n  1 and R ¼ CH3, OH, COOH, NH2, or some other functional group, hereafter referred to as RCnSH] or simple aromatic functionalities (RSH, where R ¼ C6H5, C5H4N, etc.). The various types of o-functionalized alkanethiol precursors used for self-assembly are particularly vast; a more complete listing of these endgroup-modi¢ed alkanethiol precursors (and other thiol-derivatized molecules) can be found elsewhere [11,21]. Gold, in both polycrystalline and single-crystal form, has been the most extensively employed substrate material for SAM formation. This is due not only to its availability in high purity, but also to the observations that its surface supports little or no native surface oxide and is easily cleaned. Procedures for cleaning Au substrates are well established, and depend in large part on the form of the substrate: oriented and polished bulk Au single crystals versus single-crystalline evaporated Au ¢lms (on solid support) versus polycrystalline bulk Au [13]. Since the organization and structure of the ¢lm (both local and long-range) is often strongly in£uenced by substrate surface order, smooth well-de¢ned surfaces [such as evaporated Au(111) ¢lms on mica] are generally employed for investigating molecular-level structural details in these ¢lms. Despite the high degree of ¢lm crystallinity, alkanethiol SAMs on these surfaces have been shown electrochemically to be more permeable to solution species than those formed on polycrystalline, macroscopically rough Au substrates [22,23]. Such access to the substrate is primarily limited to defect sites and grain boundaries within the ¢lm, step edges, etc. The ability of the SAM to ‘‘block’’ the electrode is essential if one wishes to investigate ET kinetics at large overpotentials, where solvent

Electrochemistry of Redox-Active Protein Films

69

decomposition or electrolyte redox processes might otherwise preclude their observation. The formation of thiol-based SAMs is typically achieved either through exposure of the gold to a solution containing the thiol or via vaporphase dosing of Au with the gas-phase adsorbate precursor. Awide variety of protocols for self-assembly from solution are available from the literature, and di¡er primarily in choice of solvent, thiol solution concentration, deposition time, and temperature [13]. Ethanolic deposition solutions generally tend to be preferred due to their availability in high purity, their low solvent toxicities, and the fact that long-chain alkanethiols exhibit reasonable solubilities in them ( 103 M range). Although other, less commonly used solvents (e.g., hexane, hexadecane) o¡er larger alkanethiol solubilities, their intercalation within the SAM has been suspected of increasing ¢lm disorder [24]. Furthermore, the composition of mixed SAMs may be highly dependent on the particular solvent chosen for the deposition (due to di¡ering solution solubilities). At the millimolar concentrations often employed for deposition, alkanethiol monolayer formation occurs on a very quick time scale (seconds), although the two-dimensional organizational process continues for a number of hours [11^15]. Consequently, SAM formation from millimolar ethanolic solutions is often allowed to proceed from times ranging from overnight up to a week past immersion. The ¢rst kinetic studies of alkanethiol adsorption from solution were initiated by Bain et al. [24]. For moderately concentrated solutions (1 mM), it was found that the self-assembly process on polycrystalline gold surfaces consisted of two distinct steps: an initial fast step which lasts at most a few minutes, during which time the contact angles and thicknesses of the ¢lms have reached 80^90% of their limiting values; and a second, slower step lasting several hours, during which the thickness (by ellipsometry) and contact angles slowly approach their ¢nal values. Consequently, the overall mechanism has been explained by the initial formation of a relatively amorphous monolayer (¢rst step), followed both by the slow expulsion of solvent and contaminants from the ¢lm and by the slow consolidation and ordering of the ¢lm through additional alkanethiol adsorption and its lateral di¡usion on the surface (second step). Thus, the second, slower step has been characterized as a two-dimensional (2-D) crystallization process within the ¢lm. A number of subsequent time-domain studies have demonstrated that alkanethiol monolayer formation on Au exhibits Langmuir adsorption kinetic behavior. This process has been investigated in some detail by secondharmonic generation (SHG) [25^27]. X-ray photoelectron spectroscopy (XPS) [25,27], near-edge X-ray absorption ¢ne structure (NEXAFS) [28], surface acoustic wave (SAW) experiments (via vapor-phase adsorption) [29], quartz crystal microbalances (QCM) [30,31], and infrared re£ection-absorption

70

Niki and Gregory

spectroscopy (IR-RAS) [32]. For deposition from dilute solution (108^107 M), it has been shown that adsorbed alkanethiols adopt a conformation and orientation at submonolayer coverages that allows them to maximize their interactions with the substrate (i.e., lie in contact with the substrate surface). As the coverage increases, nucleation and a gradual change in the average molecular tilt angle (toward the surface normal) occurs. The formation of such a ‘‘lying-down’’ phase for presaturation coverages of alkanethiols on Au(111) during the self-assembly process (both from solution and via gas-phase dosing) has recently been veri¢ed by IRRAS [32], ultrahigh-vacuum scanning tunneling microscopy (STM) [14,33^ 36], and by low-energy He atom di¡raction [37,38].Thus, many of the atomic and molecular resolution details inherent to the evolution of the ¢lm structure during ¢lm formation are now being explicitly delineated. Exchange kinetics between adsorbed and solution species for alkanethiol ¢lms on gold have been investigated using electroactive alkanethiols [39] and 35S-radiolabeled alkanethiols [40]. In both studies, signi¢cant exchange (>50%) was observed for concentrations typical of most solution deposition work (0.1^1 mM), while full exchange was observed only in neat alkanethiol solution [40]. Similar exchange reactions have also been observed recently for alkanethiol SAMs exposed to H2S vapor [41]. These studies, furthermore, have shown that the exhange kinetics have a pronounced dependence on alkanethiol chain length, with the shorter-chain homologues exhibiting faster displacement rates [39]. Such results have suggested that exchange occurs near defect sites and domain boundaries, where the alkanethiol surface mobility (which is dependent on alkyl chain length) is greatest.These observations have been corroborated by others who have monitored solution exchange reactions by IR-RAS [12] and by changes in contact angle [42,43]. Early X-ray photoelectron spectroscopic (XPS) studies concluded that the products of alkanethiol or dialkyl disul¢de chemisorption on such surfaces were indistinguishable and were consistent with the formation of a metal thiolate (e.g., Au(I) thiolate (RSAu þ Aun0): Sð2p3=2 Þ  162 eV; polymeric Au(I) hexadecylthiolate: Sð2p3=2 Þ162.5 eV [44]) [45^48]. DGads for alkanethiol adsorption on gold surfaces has been estimated by QCM [30,31] and from electrochemical data to be (17^21) kJ=mol, with a slight dependence on the chain length. The similarity in these values with estimates for aliphatic chain interaction energies has again suggested that the exchange dynamics between adsorbed and solution species are dictated by interchain dispersion interactions and therefore most likely occur near defects, step edges, and domain boundaries [30]. Furthermore, these results have argued in favor of the formation of dihydrogen gas [rather than H þ (aq)] as the primary mechanism for loss of hydrogen during the deposition process [30,40].

Electrochemistry of Redox-Active Protein Films

71

Schleno¡ et al. [40] have estimated (from prior electrodeposition data [49]) the homolytic RS ^ Au bond strength to be approximately 167 kJ=mol, which agreed well with bond energetics considerations ( 184 kJ=mol) [12]. 2.2

SAM Film Structure

The orientation, conformation, and two-dimensional structure of the polymethylene portion of alkanethiols SAMs at saturation coverage on Au(111) have been extensively investigated both qualitatively and quantitatively by a number of methods, including IR-RAS [50,51], Raman spectroscopy [52,53], grazing incidence X-ray di¡raction (GIXD) and He atom di¡raction [54,55], sum frequency generation [56,57], NEXAFS [58], and ultraviolet photoemission spectrscopy [59]. It is now well established that for long-chain alkanethiols (n  8) on these surfaces, the alkyl chains are fully extended in a nearly all-trans con¢guration, tilted 30 (from the surface normal) with a twist angle b about the molecular axis of 52 [50], and closely packed to yield a highly dense, crystalline two-dimensional arrangement at the surface. The all-trans conformation has been con¢rmed by the observation of CH2 twisting and wagging bands in the low-frequency infrared spectra of these ¢lms, implying that the local environment of the chains is similar to that of bulk crystalline n-alkanes [50]. In addition, the SAM ¢lm thickness has been shown to vary linearly with chain length for polymethylene chains composed of 9 to 21 carbons [24,51,60,61]. The molecular surface density on Au(111) ( 21.6 —2=molecule [12,14,33]) is slightly larger than that for bulk

FIG. 1 Schematic depicting the general organizational features of a CH3(CH2)9SH SAM at saturation coverage on an Au electrode surface.

72

Niki and Gregory

crystalline n-alkanes ( 18.4 —2=chain [62]), and is a consequence of chain tilting which facilitates interchain dispersion interactions. The atomic and molecular resolution details of the two-dimensional organization of alkanethiol SAMs on Au(111) at saturation coverages have been examined extensively by STM [14,63^70] and by GIXD and He atom di¡raction [54,55,71]. It is now generally agreed that at saturation pffiffiffi coverage, long-chain alkanethiols ðn  8Þ close-pack to form a ð3  2 3Þ unit cell on Au(111), are tilted toward next-nearest neighbors (NNN), and that four thiols are contained per unit cell in a pair-wise arrangement (which suggests that there are two crystallographically distinct thiols per unit cell) [14,72]. Although this notation signi¢es that two-dimensional structure is a pffiffiffithe p ffiffiffi superlattice based on an underlying ð 3  3ÞR30 periodicity, it denotes a rectangular unit cell on a hexagonal substrate mesh [54]. Thus, it is often supplanted by a more common structural designation which is based on the hexagonal mesh of the thiols: a c(2  4) superlattice. It has been suggested that the crystallographic di¡erence between the thiols that comprise the c(2  4) unit cell, and thus the origin of the superlattice structure, arises from twisting about the hydrocarbon chain axis in either a clockwise or counterclockwise fashion [72]; this would be consistent with previous IR-RAS results [73]. However, this likely cannot explain the di¡erences in tunneling current observed between the crystallographically distinct species within the unit cell. Based on analyses by both X-ray di¡raction [74] and X-ray standing waves [75], a more controversial explanation as to the origin of the observed superlattice has been put forth involving ‘‘dimerization’’of the sulfur head groups. Although this description has been met with some criticism [11,55,76], other studies have appeared which support its general conclusions [57,77,78]. STM studies of short-chain alkanethiols (n ¼ 4 or 6) at saturation coverage on Au(111) have evidenced the presence of a two-dimensional liquid phasepatffiffiffi room temperature.The slow formation of ordered domains having a ( p  3) ( p ¼ 8, 9, or 10) structure within this surface liquid phase was attributed either to the slow desorption of thiol or of entrained solvent within the ultrahigh vacuum chamber [64]. Subsequent studies of C4SH on Au(111) have demonstrated similar surface structures [70]. Alkanethiol SAMs at saturation coverage on Au(100) have also been investigated by electron di¡raction [79,80], He atom di¡raction [81,82], and by GIXD and X-ray re£ectivity (XR) [82]. Li et al. employed both He atom di¡raction and GIXD=XR to study the annealed, equilibrium structure of C18SH=Au(100) [82]. The proposed structure consistent with all diffraction=re£ectivity data consisted of: (1) rows of Au adatoms, forming a p(1 4) array on the unreconstructed Au(100) surface; and (2) alkanethiol molecules adsorbed onto both the adatom rows and terrace atoms in

Electrochemistry of Redox-Active Protein Films

73

between. The large c(2  8) unit cell observed by He atom di¡raction was attributed to the height modulation imposed on the adsorbed alkanethiol ¢lm by the adatom rows. The extracted alkanethiol orientation and surface density (i.e., yt ¼ 33.5 and 22.19 —2=molecule, respectively) were found to be similar to that on Au(111) (i.e., yt ¼ 30.3 and 21.62 —2=molecule) [82]. Recent STM studies of C4SH SAMs on Au(100) have evidenced surface structures which support this model,with the exception of the observed periodic height modulation being due rather to missing rows of Au atoms [83]. The self-assembly of alkanethiol monolayers on other metal (and some semiconductor) surfaces has also been fairly extensively investigated, although to a lesser degree than those on Au. These studies have included alkanethiol SAMS on polycrystalline Ag [52,53,84^91], Ag(111) [92^95], Cu [84,85,96,97], and GaAs [98]. It has been established that the SAM structures on the other coinage metals (Ag and Cu) are in some respects similar to those on Au; in all three cases, ¢lms created from the long-chain homologues (n10) are densely packed with fully extended, all-trans hydrocarbon chains. In other respects, however, they exhibit distinct di¡erences relative to the former. In particular, IR-RAS data has indicated that the alkanethiol molecular orientation is closer to the surface normal, ranging from 11 to 15 on both Ag and Cu (compared to 30 on Au), and that the twist angle is signi¢cantly less (b ¼ 41^48 on both Ag and Cu, compared to 52^53 on Au). 2.3

Electrochemistry of SAMs

The formation of an organized SAM on an electrode surface modi¢es the electrochemical properties of the electrode in various ways. The restricted approach of solution species to the electrode surface acts to decrease the electrical double-layer capacitance at the electrode=solution interface, and thus the observed charging current; this of course can be mediated somewhat by the presence of holes or permeable defect sites,which would consequently permit access of such species. For well-ordered and organized SAM systems (such as alkanethiols on Au), the kinetics of interfacial electron transfer can be easily regulated since the ET distance to suitable redox species (either in solution or tethered to the alkyl chain) can be accurately controlled via the length of the chain. Furthermore, the ability to utilize a wide variety of endgroup-functionalized SAMs, coupled with the ease with which multicomponent mixed ¢lms can be constructed, allows one to tailor the electrode surface in a purposeful fashion to develop electrochemical sensors. The electrochemical behavior of thiol-based SAMs has been extensively investigated, and an excellent, thorough monograph on this subject of recent vintage is available [13]. Therefore, only some of the results more relevant to the formation of stable SAM=protein ¢lms are brie£y mentioned here.

74

Niki and Gregory

An understanding of the intrinsic electrochemical characteristics of SAM-modi¢ed electrodes is essential if one intends to investigate the redox behavior of proteins adsorbed onto them. As a starting point, the electrochemical stability limits of the SAM must certainly be established, as these would result in physical removal of the SAM from the electrode surface. For n-alkanethiol monolayers on Au in basic solution (0.5 M KOH), the potential window over which no SAM-speci¢c faradaic processes occur is quite large (>1.5-V range), and extends from potentials corresponding to cathodic stripping of the ¢lm [99^105] to those for its oxidative stripping [99].While the former results in formation of a thiolate product (which often accumulates in a layer adjacent to the electrode surface due to chain-length-dependent solubility e¡ects), the latter transpires at potentials at which oxidation of adsorbed thiol occurs concomitantly with partial oxidation of the Au substrate. Highly basic solutions are required to observe voltammetric peaks corresponding to cathodic stripping of the SAM, since they are obscured by hydrogen evolution even in neutral pH solutions. Widrig et al. have demonstrated that in su⁄ciently basic solution (pH > 11), cathodic stripping of CnSH SAMs (n ¼ 3^18) from evaporated Au(111) ¢lms on mica occurs in the range 0.7 to 1.4 V range (versus Ag=AgCl), and that the peak potential shifts in a linear fashion to more negative values with increasing chain length [99]. The integrated charge (90 7 mC=cm2), however, was found to be independent of chain length. 10 2 Assuming an alkanethiol pffiffiffi pffiffiffi surface coverage of 7.6  10 mol=cm [calculated based on a ( 3  3) alkanethiol surface periodicity on Au(111)] and a surface roughness factor of 1.2, the authors concluded that the cathodic desorption process is consistent with a one-electron reduction of the adsorbed thiol to form a thiolate: Au  SCn þ e ! Auð0Þ þ Cn S

ð1Þ

Although changes in magnitude of the interfacial charging current during the stripping process have led to some controversy over the assignment of a unit faradaic charge for this process [106,107], many subsequent electrochemical investigations of alkanethiol SAMs have utilized the reductive desorption process as a means of measuring adsorbate surface coverage [100^105]. Weisshaar et al. have also demonstrated that alkanethiol surface coverages can be electrochemically controlled via oxidative deposition from alkanethiolate-containing basic solutions, or through oxidative redeposition of the reductively stripped monolayer [101]. Thus, electrochemical methods for estimating surface coverage are becoming increasingly important in mixed SAM systems formed via co-deposition procedures (see Sec. 2.5), since the ¢lm composition may di¡er signi¢cantly from that in solution. Finally, by comparing the cathodic stripping curves for C10SH and

Electrochemistry of Redox-Active Protein Films

75

HOOC ^(CH2)10SH SAMs on Au, Imabayashi et al. have shown that the nature of the end group can have a signi¢cant e¡ect on the reductive desorption peak potentials [104]. Oxidative stripping of short-chain alkanethiol SAMs (e.g, C 3SH, C4SH) has been observed to coincide with oxidation of the Au surface; oxidation of CnSH (n > 4) was not observed due to solvent oxidation. After correcting for the charge associated with Au oxidation, the remaining integrated charge (280 35 mC=cm2) led the authors to propose the formation of a sul¢nate upon alkanethiol oxidation in basic solution:  Au  SCn þ 4OH ! Auð0Þ þ Cn SO 2 þ 2H2 O þ 3e 0

ð2Þ 0

Although early studies using 4,4 -dipyridyl disul¢de (4,4 -PySSPy) as an electrode surface modi¢er found that it facilitated heterogeneous ET reactions of cyt c [4,5], only upon its more recent classi¢cation as a SAMforming material has its electrochemical behavior [and that of the corresponding thiol, 4-mercaptopyridine (4-PySH)] come under serious investigation [108^111]. Recently, Lamp et al. presented electrochemical and spectroscopic evidence indicating that SAMs formed from either 4-PySH or 4,40 -PySSPy decompose spontaneously into atomic and oligomeric sulfur upon prolonged exposure times to even dilute deposition solutions [108]. Samples prepared from 2-PySH, however, exhibited no evidence of such sulfurization, and consequently the authors were unable to explain the mechanism of decomposition. Using voltammetry and in-situ STM, Sawaguchi et al. investigated SAMs of 4-PySH, 2-PySH, and thiophenol (PhSH) on Au(111). Although surface coverages (via reductive desorption) for the three were similar (4-PySH: 4.6  1010 mol=cm2; 2-PySH: 4.7 1010 mol=cm2; PhSH: 4.4  1010 mol=cm2), only 4-PySH demonstrated an ability to facilitate facile electron transfer with cyt c, in agreement with previous reports [112^114]. In addition, STM images of the 2-PySH SAM indicated that the adsorbate coordinated to the surface through both the sulfur and the pyridine nitrogen, explaining its inability to facilitate the ET process [110]. The electrochemical behavior and structure of 4-PySH SAMs on unreconstructed Au(100) surfaces have also been investigated by voltammetry and STM [111]. It has been demonstrated that for relatively high electrolyte concentrations (>0.1 M), the total interfacial capacitance in these SAM systems is governed by that for the monolayer itself (CML). Using the simple Helmholtz model,CML can be related to ¢lm thickness: 1 d ¼ CML e0 eML

ð3Þ

76

Niki and Gregory

where e0 and eML are the permittivity of free space and the static dielectric constant of the ¢lm, respectively, and d is the ¢lm thickness [60,99]. In early SAM studies, Porter et al. showed that plots of CML1 versus the number of methylene units n for CnSH SAMs deposited on polycrystalline Au substrates exhibited linear behavior for n 10 [60]. However, signi¢cant nonlinear behavior was observed for n 8, the degree of which strongly depended on the nature of the electrolyte solution. The appearance of these two chain-length regimes suggested early on that long-chain CnSH monolayers on Au were highly impermeable to aqueous solution species. Subsequent interfacial capacitance investigations of CnSH SAMs on atomically £at Au(111)=mica substrates in 0.5 M KOH demonstrated a high degree of linearity extending over the whole chain-length regime n ¼ 3 to 18 [99]. However, a rather large value for eML was extracted from the slope of the plot (eML ’ 7, in comparison to the expected value eML  2:5); this was interpreted as arising from a higher defect density within the SAM [99]. 2.4

Endgroup-Modified SAMs

Fundamental interest in the acid ^ base properties of o-functionalized alkanethiol SAMs possessing carboxylate ^ or amino-terminal groups has largely been driven by their ability to interact=react with other functional compounds. Depending on solution pH, such terminal groups can electrostatically bind charged solution species (including charged proteins such as cyt c) at the SAM=solution interface. Furthermore, covalent attachment of appropriate precursors to these functionalities can also provide a convenient ¢rst step toward fabricating novel supramolecular architectures. Due to their well-ordered nature, alkanethiol SAMs have become useful model systems for both experimental and theoretical investigations of terminalgroup acid ^ base behavior in relatively well-de¢ned environments. In this regard, the surface pKa values of carboxylic acid-derivatized alkanethiol SAMs have been extensively investigated by contact angle titrations [115^118], voltammetry [119], interfacial capacitance measurements [120], QCM [121^123], atomic force microscopy (AFM) [124,125], and an indirect laser-induced temperature method [126]. The transition from fully protonated to a fully deprotonated state in these SAMshasbeen reported tobe considerably broader than that observed for aqueous soluble alkanoic acids [116].WhilesuchaqueoussystemsexhibitpKa’sof4^5,ithasbeenexpectedthat the corresponding surface acid dissociation constants might di¡er from these due to (1) the proximity of a low dielectric material (i.e., alkane layer) near the end-group region; (2) di¡erences in hydrogen-bonding characteristics relativetothatfor thefreespeciesinsolution(which likelychangewiththeextentof deprotonation); (3) changes in the dielectric constant in the solution region

Electrochemistry of Redox-Active Protein Films

77

adjacenttotheendgroupsastheybecomeprogessivelydeprotonated;(4)other interfacialelectric ¢eld e¡ects that arise fromthe increasing negativecharge at the SAM=solution interface. In some cases, the surface pKa’ s for these carboxylic acid-derivatized SAMs have been reported to be 3^5 pH units more alkaline than corresponding aqueous soluble systems (e.g., butyric acid). For example, a reversible voltammetric wave was observed at 0.6 V (versus Hg=Hg2SO4) byWhite et al. for a mixed SAMcontaining HOOC ^(CH2)10SH and C10SH, which they attributed to an electric ¢eld-driven protonation=deprotonation process of the carboxylic acid moeities at the interface [119]. These waves exhibited maximum peak currents for solution pH’ s near 8.5. Kakiuchi et al. have measured changes in the double-layer capacitance of HOOC ^(CH2)nSH SAMs as a means of estimating apparent surface pKa’ s [120]. In these systems, apparent surface pKa’ s of 7.7, 8.7, 9.2, and 10.3 were found for n ¼ 2, 4,6, and10, respectively, suggesting a signi¢cant chain-length dependence.Other studies, however, have indicated that the surface pKa’s for carboxylic acid-derivatized SAMs are only slightly higher than the analogous aqueous systems [122,123,127].Using QCM, Sugihara et al.estimated surface pKa0 s of 5.8, 6.0, 6.2, and 6.5 for chain lengths n ¼ 2, 5, 10, and 15, respectively [122]. Other values in this range have been reported by Simmons et al., who estimated surface pKa0 s of 6.31, 6.80, 6.67, and 7.16 for n ¼ 1, 2, 10, and 15, respectively.Other factorshavealsoappeared tohaveane¡ectontheobserved surface pKa0 s, including alkanethiol surface coverage [123] and the mole fraction of carboxylic acid-derivatized alkanethiols in mixed SAM systems [117,118,123]. 2.5

Mixed Monolayers

The study of mixed alkanethiol SAMs has been driven in large part by the possibility of advanced technological applications that may require systematic control over interfacial ¢lm properties. Early investigations of the simplest mixed SAM systems involved the co-adsorption of binary mixtures of alkanethiols (or dialkyldisul¢des) on Au, and centered on mixtures containing either di¡ering terminal groups (CH3 versus OH) [42,44], di¡ering chain-length components [43,128], or both [129,130]. In many of these early studies, changes in ¢lm properties were monitored by examining variations in ellipsometric ¢lm thickness, contact angle, or vibrational peak positions=intensities as a function of alkanethiol solution mole ratio. For SAMs composed of di¡ering-chain-length alkanethiols only, Bain et al. concluded that the ¢lms consisted of a densely packed molecular region near the Au surface and a disordered outer region in contact with the deposition solution [43,128]. Thus, by varying the solution ratio of the two components, the degree of disorder at the ¢lm=solution interface could be reproducibly controlled. Preferential adsorption of the longer-chain component was

78

Niki and Gregory

observed from ethanolic solutions (due to a somewhat low solution solubility), and substantial variations in ¢lm composition with solvent were noted, both arguing in favor of thermodynamic control of the composition of the SAM [43]. Although large-scale phase segregation into single-component domains was not observed in these studies, the authors speculated as to the improbability of a completely random distribution of both components in the ¢lm. Similar conclusions regarding component distributions within these ¢lms have been made in studies of mixed-composition SAMs fabricated either from mixtures of alkanethiols di¡ering in their terminal functionalities [42,131], or from asymmetric disul¢des having di¡erent end groups [131]. Additionally, a systematic investigation of equimolar, binary mixtures of HOCH2^(CH2)15SH and CH3^(CH2)(15 þ m)SH (where 6 < m < 6) self-assembled onto Au from ethanolic solutions have led to a general structural model in which the ¢lm organization evolves from one exhibiting random mixing (at low m) to one evidencing phase segregation (at high m) [130]. Electrochemical methods have been recently employed to address issues of phase segregation in mixed SAMs, and have taken advantage of the fact that reduction potentials (which result in desorption) of the alkanethiols in strongly basic solution are dependent on both alkyl chain length [99^105] and terminal functionality [104,105]. The obvious utility underlying an electrochemical approach to such a problem lies both in its ability to distinguish between and to quantify such components at the surface. Nishizawa et al. have examined the electrochemical desorption of a SAM formed from a binary mixture of 3-mercaptopropionic acid (3MPA) and C16SH, which differ signi¢cantly both in alkyl chain length and in the nature of the end group [132]. In these studies, two separate cathodic stripping peaks were observed, and were attributed to reductive desorption of each component separately, implying phase segregation of the components within the ¢lm [132]. Longer exposure times to the deposition solution were shown to yield mixed ¢lms exhibiting larger fractions of C16SH. Furthermore, porous SAMs could be generated through selective desorption of the more positive desorption feature (belonging to the shorter-chain component, 3MPA)[132]. In contrast, based on measured oxidation peak currents for cyt c and FeðCNÞ4 6 as a function of C18SH mole fraction in binary SAMs of 3MPA and C18SH on Au, Sato et al. concluded that rather uniform mixing of the co-adsorbed thiols was occurring [133]. This was corroborated with the observation that although two separate reductive desorption peaks were seen in the mixed monolayer, the one attributed to 3MPA shifted to more negative potentials as the fraction of C18SH increased [133]. Arnold et al. have demonstrated (also via electrochemical desorption) the presence of homogeneous mixing in binary alkanethiol SAMs whose components exhibit similar chain lengths

Electrochemistry of Redox-Active Protein Films

79

but highly dissimilar end groups (CH3 versus COOH) [134]. Subsequent approaches to electrochemically mediated, selective removal of constituents from phase-segregated SAMs have led to methods for their selective replacement [135^137]. Questions of miscibility and lateral distribution within various mixed SAMs have also been addressed recently by high-spatial-resolution techniques such as STM, which yield real-space structural information at the molecular level. This is particularly important since many of the aforementioned methods for analyzing ¢lm composition and distribution yield information about the macroscopic properties of the ¢lm. Stranick et al. observed phase separation by STM in binary mixed monolayers of C16SH and CH3O2C ^ C15SH on Au, resulting in the formation of nanometer-scale domains [138]. These components, di¡ering only in the nature of their end groups, were chosen based not only on their inability to hydrogen-bond but also on their identical hydrocarbon chain lengths. The authors concluded that the phase-separation process must be driven entirely by terminalgroup interactions, despite the fact that di¡erences in their relative interaction strengths must be small. Further studies by Stranick et al. with mixed monolayers having di¡ering end groups (including OH and CN) have also demonstrated some phase separation [139]. Internal functionalization has also been shown to drive the phase-separation process in binary mixed SAMs of internally functionalized alkanethiols (e.g., amide-containing alkanethiols) and nonfunctionalized alkanethiols (e.g., C10SH) [140]. 3 3.1

SAM/PROTEIN FILMS Voltammetric Studies

3.1.1 Cytochrome c at Bare Electrodes Cytochrome c in solution often exhibits an irreversible voltammetric response at gold electrodes, and an adsorbed layer of cyt c on an electrode surface blocks the ET reaction of cyt c in solution. The only exception to this involves cytochromes c3 extracted from the sulfate-reducing bacteria Desulfo vibrio [3]. This unique class of heme proteins contains four hemes in the molecule and exhibits a reversible voltammetric response at various electrode materials without surface modi¢ers [3]. Hinnen and Niki examined the formal potential of horse heart cyt c adsorbed on polycrystalline gold, ruthenium, and glassy carbon electrodes by electrore£ectance (ER) voltammograms. The formal potential was measured to be around 0.18 V (versus NHE) in 30 mM phosphate bu¡er at pH 7.0, which is 0.44 V more negative than that of the native one; however, the adsorbed cyt c undergoes a

80

Niki and Gregory

rapid (reversible) ET reaction at this potential [141,142]. On the other hand, the formal potential of cyt c3 adsorbed on metal electrodes is nearly the same as that of the native protein in solution [143]. 3.1.2 Proteins Immobilized Electrostatically on Self-Assembled Monolayers Cytochrome c at 4,40 -Bypyridine and 4-Pyridinethiol SAMs. Hinnen and Niki [142] and Sagara et al. [144] studied the interfacial properties of cyt c at a gold electrode in the presence of various surface modi¢ers by the ER technique and found that there are two types of surface modi¢ers. Cytochrome c immobilized on bis(4-pyridyl) disul¢de (4-PySSPy) or 4-mercaptopyridine (PySH)-modi¢ed gold electrodes undergoes electrode reactions through the layer of the surface modi¢er (SAM ¢lm).The formal potential of cyt c immobilized on the 4-pyridinethiolate SAM is slightly more negative than that of the native one.On the other hand, the formal potential of cyt c coadsorbed with 4-40 -bipyridine (in 10 mM 4,40 -bipy solution) is 0.03 V (versus NHE), which is about 0.23 V more negative than that of the native one and 0.21 V more positive than that adsorbed on the bare gold electrode. It is also interesting to note that the electrode reaction of cyt c co-adsorbed with 4,40 bipy undergoes a rapid (reversible) electrode reaction at the formal potential in its adsorbed state.The formal potential of cyt c co-adsorbed with 4,40 -bipy on the gold electrode is strongly dependent on the concentration of 4,40 -bipy in solution, and varies from 0.18 V (versus NHE, without 4,40 -bipy) to 0.03 V (saturated 4,40 -bipy solution) [145]. Cytochrome c at Carboxylic Acid-Terminated Alkanethiol SAMs. Tarlov and Bowden ¢rst demonstrated the feasibility of using carboxylic acid-terminated alkanethiol SAMs to study the ET reaction of cyt c [146]. Cytochrome c electrostatically immobilized on carboxylic acid-terminated alkanethiol SAMs exhibits a reversible voltammetric response at its formal potential, but the ET rate constant varies with the chain length of alkanethiols. The formal potential of cyt c depends on the degree of electrostatic interaction between it and the carboxylic acid terminus and on the amount of adsorbed cyt c [7,147^152]; the measured formal potential is shifted negatively by approximately 10^50 mV. The full-width-at-half-maximum peak width of the cyclic voltammogram is somewhat larger than that of the theoretical value, and the deviation from the ideal value has been explained in terms of the distribution of the ET rate constants of cyt c at the SAM surface [148,149,153]. Noting that peak potential separations are always greater than 57 mV and the di¡usion coe⁄cients are smaller than those obtained from studies at simple electrodes, Honeychurch and Rechnitz have analyzed the

Electrochemistry of Redox-Active Protein Films

81

cyclic voltammograms of cyt c in solution at 4-pyridinethiol-modi¢ed gold electrodes by taking into account the potential distribution at the electrode=solution interface [154]. Cytochrome c at Mixed SAMs. Electrode reactions of horse heart cyt c in solution at mixed SAMs of 4-pyridinethiol and methyl-terminated alkanethiols on gold electrodes have been investigated. It was found that the voltammogram becomes sigmoidal with an increase in the fraction of methyl-terminated alkanethiols in the SAMs, and the peak current decreases, accordingly. Alkanethiols with shorter chains were observed to be much more e¡ective at suppressing the peak current than those with longer chains. This is likely due to island formation in the mixed SAMs, which results from weak interaction between 4-pyridinethiol and the long-chain alkanethiols (or due to the strong lateral interactions between long-chain alkanethiols) [155]. On the other hand, the e¡ect was pronounced as n increased from 3 to 18 in the mixed SAMs of HOOC(CH2)2SH and HOOC(CH2)nSH [156]. These results can be explained in terms of the stability of n-alkanethiol SAMs: shorter-chain alkanethiols produce a less stable SAM layer because the hydrophobic e¡ect is smaller [157]. Cytochrome b5 and Enzymes at Poly-L-Lysine=Carboxylic Acid-Terminated Alkanethiol SAMS. A monolayer of cyt b5 exhibits a well-de¢ned voltammetric response at multilayers of cationic poly-L -lysine=carboxylic acid-terminated alkanethiol-modi¢ed gold electrodes in 8.8 mM phosphate bu¡er solution at pH 7.0 [158]. Its formal potential was 0.040 V (versus NHE), which is about 30 mV more positive than the native one in the same solution. The measured ET rate of cyt b5 through the multilayer was 1.2 0.3 s1. Both glucose oxidase (GOD) [159] and lactate oxidase [160] immobilized on poly-L -lysine-covered carboxylic acid-terminated alkanethiol SAMs show voltammetric responses and exhibit linear relationships with their substrates up to 2 and 0.2 mM respectively. Glucose Oxidase Immobilized on DTSSP and DSP SAMs. Glucose oxidase immobilized at SAMs of either 3,30 -dithiobis-sulfocinnimidylpropionate (DTSSP) or dithiobis(cinnimidylpropionate) (DSP) undergoes a reversible charge transfer reaction at E1=2  0:282 V (versus Ag=AgCl), and its ET rate constant has been estimated to be 0.026 s1 [161]. Enhancement of the oxidation current of the GOD=DTSSP=Au assembly (on addition of glucose) was observed under anaerobic conditions. Immobilization of

82

Niki and Gregory

FAD on the DTSSP-modi¢ed gold surface gives rise to well-de¢ned, reversible electrochemistry at E1=2 ¼ 0.395 V (versus Ag=AgCl) [161]. 3.1.3 Immobilized Proteins Covalently Attached to Self-Assembled Monolayers Cytochrome c. Collinson and Bowden studied cyt c covalently attached to HOOC(CH2)15SH=Au electrodes via an electrostatically guided carbodiimide coupling. The cyclic volammetric results revealed that the coverage of electrochemically active cyt c was about one-third monolayer and the formal potential was slightly more negative than that in the solution. The ET rate constants of cyt c immobilized electrostatically and bonded covalently to HOOC(CH2)15S=Au electrode were almost the same [147]. Cytochrome c Oxidase. Dong and co-workers have reported that cyt c oxidase covalently bonded to a HOOC(CH2)2S=Au electrode by a carbodiimide coupling exhibits a reversible voltammetric response due to the redox reaction of cyt a3 in the molecule and can mediate the redox reaction of cyt c in the solution [162]. The ET rate constant depends strongly on the ionic strength and composition of the supporting electrolyte. Spatially Ordered Catalytic Enzyme Assemblies on Electrodes. Redox enzymes organized onto electrodes as monolayer assemblies with covalently attached redox relay groups yield electrochemically active enzyme electrodes exhibiting bioelectrocatalytic features [163]. Enzymes or proteins have been covalently bonded to SAMs on electrodes via (1) coupling of the carboxylic acid termini of SAMs on electrodes (or their active esters) to lysine residues of the proteins, (2) direct coupling of the amine termini of SAMs to glutamic=aspartic residues of proteins, (3) or stepwise coupling of the lysine residues of an enzyme to the SAMs by bifunctional reagents. For example, enzyme electrodes consisting of glutathione reductase coupled to a carboxylic acid-functionalized monolayer via lysine residues (surface coverage is 2 1011 mol cm2) have been electrically contacted with the electrode via covalently attached N-methyl-N0 -carboxyalkyl-4,40 -bipyridium electron relay units. Several methods have been developed to increase biocatalytic activities of enzyme electrodes.One approach includes the assembly of an enzyme multilayer network on the electrode surface. Other methods to assemble multilayer enzyme electrodes include the interlayer linkage of the enzyme layers by biotin ^ avidin, antibody ^ antigen, or lectin ^ sugar interactions [164].

Electrochemistry of Redox-Active Protein Films

3.2

83

Structural Studies

3.2.1 Orientation of the Prosthetic Group Cytochrome c. Yeast cyt c has been covalently bound to the surface of multilayer ¢lms consisting of even numbers of arachidic acid monolayers (COOH groups are exposed at the surface) plus a surface monolayer of thioethyl stearate. Optical linear dichromism was used to determine the average orientation of the heme group within cyt c relative to the multilayer surface plane. Orientations of the heme were observed to be similar for both electrostatically immobilized (on the acid terminus of the multilayer) and covalently bound yeast cyt c; in both cases, the heme was observed to be nearly co-planar with the surface plane [165,166]. Glass slides or Ge=Si multilayer substrates coated with a covalently bound SAM (11-siloxyundecanethiol, which exposes a free thiol end group) covalently binds yeast cyt c molecules through the cysteine-102 residue, forming a disul¢de linkage. The cyt c monolayer binds to the surface at approximately closed-packed inplane densities, and also binds photosynthetic reaction centers electrostatically [167]. The orientation distribution of the heme group in adsorbed cyt c was determined by a combination of absorption linear dichromism and emission anisotropy. A Langmuir-Blodgett ¢lm consisting of three layers of cadmium arachidate, followed by a single arachidic acid layer, were ¢rst deposited on silicon oxynitride or sol-gel glass planar waveguides [168,169]. Electrostatic adsorption of the positively charged protein (cyt c) to the negatively charged head group of the exterior arachidic acid monolayer resulted in a narrow orientation distribution of the heme (tilt angle ¼ 46 6 ). Cytochrome b5. Sligar and co-workers have successfully utilized denovo gene synthesis to selectively modify the properties of cyt b5 building blocks, to allow fundamental measurements of ET and construction of mesoscale biological assemblies with de¢ned two-dimensional orientation and structure [170^174]. A unique surface cysteine residue was introduced into the cyt b5 surface by genetic engineering techniques, and this cysteine was used to covalently attach the protein to a silica surface derivatized with a thiol-reactive alkylsilane, or to atomically £at gold on mica. Linkers have included alkyl chains with a terminal halide to provide the electrophilic center for protein derivatization. Surface coverages are typically greater than 90%. TheT65C mutant of cyt b5 (threonine-65 is replaced by cysteine) has been covalently attached through Cys-65 to bromine-terminated alkanethiol SAMs on gold electrodes; however, no voltammetric response was obtained [173]. The heme orientation of these mutants was determined by linear dichromism. TheT8C (threonine-8 is replaced by cysteine) and T65C

84

Niki and Gregory

cyt b5 mutants exhibit very distinct linear dichromism e¡ects, with opposite relative signs for the optical signal representing parallel minus perpendicular light component absorption. The average tilt angles forT8C and T65C lie on either side of 54.7 [172]. Myoglobin. Myoglobin (Mb), modi¢ed by site-directed mutagenesis to include a unique cysteine residue, covalently binds selectively to selfassembled haloalkylsilylated silica surfaces. The Mb A126C (alanine-126 in myoglobin is replaced by cysteine) orients the heme plane nearly parallel to the point of attachment, thereby producing alignment of the heme prosthetic group essentially consistent with the surface normal. The results indicate that on average, the heme normal of Mb A126C is oriented 44 2 relative to the surface normal [174]. 3.2.2 The Spin State of the Heme Chromophore: SERRS (SERS) Studies Cytochrome c at Silver Electrodes. Hildebrandt and Stockburger have extensively studied the conformation and redox properties of horse heart cyt c adsorbed on silver electrodes by surface-enhanced resonance Raman scattering (SERRS), and have found that cyt c exhibits various conformational states upon adsorption on the bare silver electrode [175^180]. An adsorption-induced transformation from the six-coordinate low-spin state (6cLS) to the ¢ve-coordinate high-spin state (5cHS) was observed.The formal potential of the 5cHS state was estimated to be 0.17 V, in agreement with results obtained by Hinnen and Niki [141,142]. The SERRS spectrum of cyt c immobilized on a carboxylic acid-terminated alkanethiol modi¢ed gold or silver electrode reveals that both oxidized and reduced forms of cyt c are in the 6cLS state [181,182], and that its formal potential is 0.26 V (versus NHE) in 10 mM phosphate bu¡er solution at pH 7.0 [147,149]. These results strongly support the retention of the native state of cyt c on the carboxylic acid-terminated alkanethiol modi¢ed electrodes. Song et al. reported the formal potential of cyt c of the cyt c=HOOC(CH2)nSH=Au (n ¼ 5, 10, and 15) system to be 0.215 V in 4.4 mM phosphate bu¡er solution at pH 7.0 [148]. Cytochrome c in the Presence of 4,40 -Bipyridine. Fan et al. studied the conformation of pyridine derivatives by SERS at a silver electrode in the presence of cyt c [181].When 4,40 -bipy-modi¢ed silver electrodes are transferred to a cyt c solution, 4,40 -bipy is partially displaced by cyt c from the electrode and a mixed adsorbed layer of 4,40 -bipy and cyt c is formed. The spin state of the co-adsorbed cyt c with 4,40 -bipy is a mixture of the 5cHS and 6cLS states, and the formal potential of the co-adsorbed cyt c is 0 V (versus NHE).

Electrochemistry of Redox-Active Protein Films

85

Cytochrome c at Carboxylic Acid-Terminated Alkanethiol SAMs. The SERRS spectrum of cyt c immobilized on carboxylic acid-terminated alkanethiol-modi¢ed gold or silver electrodes reveals that both oxidized and reduced forms of cyt c are in the 6cLS state [182,183], and that its formal potential is 0.26 V (verus NHE) in 10 mM phosphate bu¡er solution at pH 7.0 [150,184]. These results strongly support the retention of the native state of cyt c on the carboxylic acid-terminated alkanethiol-modi¢ed electrodes. The electrostatic binding of cyt c to the SAM-modi¢ed electrode yields a highly oriented con¢gurational state with the heme edge directed toward the electrode surface [152]. On the other hand, covalent binding of cyt c to the carboxylic acid-terminated SAM-modi¢ed electrode yields a randomly oriented con¢gurational state with no preferred direction between the heme edge and the electrode surface [152]. Cytochrome c3 at 4,40 -Bipyridyl and Carboxylic Acid-Terminated Alkanethiol SAMs. The SERS bands attributable to both cyt c3 (a tetraheme protein,with the hemes well exposed to the solvent) from Desulfovibrio desulfuricans (Norway strain) and 4,40 -bipy suggest that 4,40 -bipy is coadsorbed with cyt c3 on the silver electrode surface [185]. Ferri-cyt c3 exists in the mixed 6cLS and 5cHS spin state, which is similar to that observed in the absence of 4,40 -bipy. The transformation from the oxidized form (the mixed-spin state) to the reduced form (6cLS state) is completed at 0.0 V (versus NHE), which is 0.2 V more positive than the formal potential of the most positive redox site of the native cyt c3. The intensity of the SERRS spectrum of the cyt c3 co-adsorbed with 4,40 -bipy in the low-frequency region (370^430 nm) is considerably weak compared to that for RRS in solution. These results suggest that 4,40 -bipy has a signi¢cant e¡ect on the heme environment of the adsorbed cyt c3. The formal potential of cyt c3 coadsorbed with 4,40 -bipy is monitored by the shift of the oxidation-state marker band, which becomes more positive with increasing concentrations of 4,40 -bipy. The shift of the formal potential may be explained in terms of a decrease in the solvent exposure of the heme due to increased shielding of the exposed heme edge by 4,40 -bipy molecules. Both SERRS and ER voltammetric techniques have been used to investigate the e¡ect of surface modi¢ers [11-mercaptoundecanoic acid (11MUDA) and 4,40 -bipy] on the structure and redox properties of cyt c3 at a silver electrode [185]. The ER voltammograms of cyt c3 immobilized on 11MUDA exhibit a small hysteresis between the forward and backward potential scans, suggesting that the redox process is reversible and cyt c3 retains its native structure. The SERRS spectra of cyt c3 (both oxidized and reduced forms) immobilized on 11-MUDA clearly indicate that the hemes are in the 6cLS state. The formal potential of cyt c3 immobilized on the

86

Niki and Gregory

11-MUDA-modi¢ed silver electrode was monitored by the shift of the oxidation-state marker band and is 0.20 V (versus NHE) , which is about 0.10 V more negative than that observed on a bare silver electrode [144]. 3.3

Electron Transfer Kinetics of Proteins at Modified Electrodes

Intermolecular ET processes between the positively charged and negatively charged sites of electron-transfer protein molecules such as cyt c=cyt b2 [186] cyt c=cyt c peroxidase [187], cyt c plastcyanine [188^190], and cyt c=cyt b5 [191^194] couples have been investigated in solutions having di¡erent ionic strength, pH, and viscosity. E¡ects of solution viscosity on the con¢gurational rearrangement reaction rate between ETprotein complexes have been studied theoretically [195,196] and experimentally [188^192,194,197^200]. The nonuniform charge distributions on the surface of protein molecules often lead to multiple con¢gurational states in the protein coupling. The initial step in the formation of the protein complex is a nonspeci¢c electrostatic association between the two proteins, followed by rotational di¡usion on the molecular surface to reach the proper con¢guration for the ET event. When the protein couple forms multiple con¢gurations, the intermolecular ET rate is limited by the rotational di¡usion of the molecules. This process can be viewed as directional ET regulated by a ‘‘gating mechanism’’ [186^ 189,191^194,201^213]. Intermolecular ET rates depend strongly on the ionic strength, pH, and temperature of the solution and are reported to be in the range of 103^105 s1. Details of the molecular structure of such complexes and intermolecular ET reactions have not been fully understood by using physiological redox complexes because of the complexity of biological systems. In electrochemical systems, on the other hand, the electrodes act as both electron acceptor and electron donor, and are considered to be a simple model system to mimic a charged interface of the physiological binding domain. The electrode reaction of cyt c immobilized electrostatically to carboxylic acid-terminated alkanethiol SAMs is considered to be the simplest model for biological ET processes. The advantages of an electrochemical approach to investigate the ETmechanism are that one can regulate the ETpath length by using alkanethiol SAMs of varying chain lengths and deconvolute the intermolecular ET event at the interface from the intramolecular ET event, provided that one can measure ET rate constants up to 105 s1. The charging current of the electrical double layer at the electrode interface limits the measurable ET rate when traditional electrochemical techniques (cyclic voltammetry and AC impedance) are used. At well-de¢ned surfaces having geometric areas of approximately1 cm2, the fastest reliable ETrates that have

Electrochemistry of Redox-Active Protein Films

87

been reported are around 100 s1, which are much slower than intermolecular ET rates for protein complexes [188,189,194,201,202]. Thus, it is di⁄cult to measure intermolecular ETevents by traditional electrochemical techniques. The UV-visible potential-modulated re£ectance (ER) spectroscopic technique, on the other hand, involves the measurement of a faradaic current as an optical response generated by an AC modulation of the electrode potential, and the e¡ect of double-layer charging on the optical response can be minimized [214].This technique enables one to measure ET rates up to approximately 10 4 s1. 3.3.1 Electron Transfer Reaction Rates Through Alkanethiol SAMs Horse Heart Cytochrome c at Carboxylic Acid-Terminated Alkanethiol SAMs. It has been shown that the long-range ET rate between metal electrodes and electroactive species through alkanethiol SAMs decreases with the chain length of the SAM [13,150,215^219]. For these systems, the ET reaction rate constant at the binding site between cyt c and the carboxylate terminus depends on the number of methylene groups n in the alkyl chain: ks ¼ kðn¼0Þ expðbnÞ

ð4Þ

where k(n ¼ 0) is the apparent ET rate constant extrapolated to n ¼ 0 [220]. The exponential decay factor b has been found to be 1.09 0.02 per methylene group regardless of the type of redox species at the terminus of alkanethiol SAMs when n  10 [13,150,215^219]. The ET reaction rates between cyt c (horse heart) and a Au(111) electrode through carboxylic acid-terminated alkanethiol [HOOC(CH2)nSH] self-assembled monolayers have been measured by the ER technique as a function of the chain length of alkanethiols (n ¼ 2 ^ 11) [214,221]. The ET reaction rate constant decreases exponentially with the chain length as shown in Fig. 2, and the exponential decay factor given by the Marcus theory is b ¼ 1.09 0.02 per methylene group (0.71 0.01 —1). The ET rates through the long-chain alkanethiol SAMs are controlled by the ET rates through the alkyl chain. The ET reaction rates through the short-chain alkanethiol monolayers, on the other hand, are nearly independent of the chain length. It has been assumed that there is a con¢gurational rearrangement of cyt c on the SAM prior to the ET reaction. Given by reaction (5), a thermodynamically stable adsorbed structure of cyt c(ox)(I), which is formed upon the adsorption of cyt c from the solution to the carboxylate termini, undergoes a con¢gurational rearrangement to cyt c(ox)(II), at which the most e⁄cient ET reaction takes place [given by reaction (6)]. The ET reaction is followed by a second con¢gurational rearrangement given by reaction (7) to form a thermodynamically stable binding state, cyt c(red)(I). The rate-controlling step of the ET reaction through

88

Niki and Gregory

FIG. 2 Logarithmic plot of the ET rate constant (ket) of horse heart cyt c through carboxylate-terminated alkanethiol SAMs on a Au electrode with respect to alkanethiol chain length n: ( ) data from Ref. 3 in 10 mM phosphate buffer solution at pH 7.0; (~) data from Ref. 19 in 4.4 mM phosphate buffer solution at pH 5.

short alkanethiol chains is very likely to be the conformational rearrangement of cyt c on the SAM surface given by reaction (5), and the transformation rate constant k1 of the forward reaction is estimated to be 2.6  103 s1. k1

* cyt cðoxÞðIIÞ cyt cðoxÞðIÞ ) k2

kf

* cyt cðredÞðIIÞ cyt cðoxÞðIIÞ þ e ) k3

kb

* cyt cðredÞðIÞ cyt cðredÞðIIÞ )

ð5Þ ð6Þ ð7Þ

k4

The transformation rate constant k1 is markedly suppressed with increasing solution viscosity, and accords with the intermolecular ETrate of the protein couple limited by the rotational di¡usion of the molecules [221]. The ET rate through longer-chain alkanethiols decreases with increasing ionic strength

Electrochemistry of Redox-Active Protein Films

89

or decreasing pH of the solution. The rate-limiting ET step through short alkyl chains results from a con¢gurational rearrangement process of cyt c from a stable binding form on the carboxylic acid terminus to a con¢guration which facilitates the most e⁄cient ET pathway, and is represented by reaction (5) [221]. It is possible to elucidate the binding site of cyt c(ox)(II) that facilitates the most e⁄cient ET pathway from the extrapolated value of ks to n ¼ 0 at low ionic strength, the value of which is about 5  10 6 s1.The ETrates from the heme edge to various lysine residues on the surface of cyt c can be estimated from the results of Gray and Winkler [222], and the intramolecular ET rate from the heme edge to lysine-13 is estimated to be 2  105 s1. It is very probable that Lys-13 hydrogen bonds to the carboxylate terminus of the alkanethiol SAMs. The ET rate constant of RC9-K13A (lysine-13 of rat cyt c, RC9, is replaced by alanine) of rat cyt c expressed in yeast [223] immobilized on 3mercaptopropionic acid SAMs on gold electrodes was measured to be 0.2 0.05 s1 [224]. In contrast, the ET rate of native horse heart cyt c at 3-mercaptopropionic acid SAMs on gold electrodes has been estimated to be 8.4  105 s1 from Fig. 2. This result suggests that Lys-86 is a potential candidate to facilitate the ETpathway of RC9-K13A cyt c. Iso-1 Cytochrome c at Carboxylic Acid-Terminated Alkanethiol SAMs. Cytochrome c extracted from yeast (iso-1 cyt c) exhibits a di¡erent kinetic nature from that of horse heart cyt c at carboxylic acid-terminated alkanethiol-modi¢ed gold electrodes. While the three-dimensional structures of iso-1 cyt c and horse heart cyt c are very similar, the ET rate of iso-1 cyt c is approximately three times slower than that of horse heart cyt c [225]. The exponential decay factor of the ET rates through longer alkyl chains (n ¼ 7 and 10) is similar to that of horse heart cyt c. The slow ET rate may be explained by assuming the ET pathway that lies from the heme edge to the carboxylic acid terminus of the SAM occurs via Lys-11of iso-1cyt c [224]. This is likely due to the inability of Arg-13 to form stable hydrogen bonds to the carboxylic acid terminus of the alkanethiol because of its resonance structure. Cytochrome c at Mixed Carboxylic Acid-Terminated Alkanethiol SAMs: The ET rate constants of cyt c (horse heart) immobilized on mixed SAMs of carboxylic acid- and methyl-terminated alkanethiols [HOOC(CH2)10SH= CH3(CH2)9SH] with varying mole fractions were determined by the ER spectroscopic technique [134].The mixed monolayers were co-deposited from ethanolic solutions containing both components at speci¢c mole fractions. The mixed monolayers were found to be homogeneous, with compositions that were very similar to those in the solution

90

Niki and Gregory

phase. The ET rate constant showed a maximum when the mole fraction of the carboxylic acid-terminated alkanethiol was 0.8, and was approximately six times faster than one comprised only of HS(CH2)10 COOH. Bowden and co-workers have reported an interesting result obtained for the voltammetry of horse heart and yeast cyt c immobilized at mixed SAMs [225].With horse heart cyt c, the ET rate constant through the mixed SAM deposited from a solution containing an equimolar mixture of HS(CH2)10COOH and HS(CH2)8OH (the composition of the SAM on the electrode is considered to be an equimolar mixture) is approximately ¢ve times faster than one comprised only of HS(CH2)10COOH. In contrast, with yeast cyt c, there is a 2500-fold rate enhancement when switching to the mixed HS(CH2)10COOH= HS(CH2)7OH monolayer (again considered to be a 1:1 mixture). The two proteins have similar ET reorganization energies and surface charge distributions, but there are large di¡erences in amino acid composition that may a¡ect the ETcoupling to the electrode. Cytochrome c Covalently Attached to Carboxylic Acid-Terminated Alkanethiol SAMs. Collinson et al. measured the ETrate of cyt c covalently attached to 16-mercaptohexadecanoic acid SAMs on gold by cyclic voltammetry and found that the ET rate is similar to that for cyt c immobilized electrostatically on the same electrode [147]. Cytochrome c at N-Acetylcysteine-Modified Electrodes. Ruzgas et al. proposed a modi¢ed approach to evaluate the ET rate constant of cyt c at modi¢ed electrodes from the ER response [151]. The ET rate constant evaluated by using the modi¢ed equivalent circuit, however, was unreasonably small. In contrast, the ET rate constant evaluated by using the equation of Feng et al. [214] was reasonable. Cytochrome c at Hydroxy-Terminated Alkanethiol SAMs. Miller and co-workers studied the electrode reactions of horse heart, tuna, and yeast cyt c in solution at hydroxy-terminated alkanethiol-modi¢ed gold electrodes to determine thermodynamic and kinetic parameters [226]. Well-de¢ned voltammograms were obtained at HO(CH2)3SH SAMs, whereas those at HO(CH2)11SH SAMs were drawn out and ill-de¢ned. After corrections for di¡usion limitation and double-layer charging, the kinetic parameters were calculated and it was concluded that the reorganization energies and the maximum ET rates among these cyts c were indistinguishable. However, both the calculated reorganization energies and di¡usion coe⁄cients of the cyts c in this study di¡er signi¢cantly from the results of others [220,227].

Electrochemistry of Redox-Active Protein Films

91

3.3.2 Azurin Immobilized (by Hydrophobic Interaction) on Hexanethiol SAMs The ER spectroscopic technique has been applied to the study of the redox reaction kinetics of the blue copper protein azurin. Azurin was irreversibly adsorbed on the hexanethiol SAM by hydrophobic interactions; the SAM itself was deposited on a polycrystalline gold electrode from 1^5 mM ethanolic solutions of hexanethiol. Azurin adsorbed on the SAM was shown to be intact and electrochemically reversible [228]. The formal potential of the adsorbed azurin was 95 8 mV (versus Ag=AgCl), which is nearly identical to that of the native azurin in solution. The ETrate was determined by cyclic voltammetry, AC impedance techniques, and ER spectroscopy. The ET rate constants determined byAC impedance and ER techniques were 300 s1 and 150^200 s1, respectively, which were considerably higher than the values determined voltammetrically (4^12 s1). The disparity between the ET rates measured voltammetrically and by ER suggests that the redox reaction may not be a single-step process.The reorganization energy determined from the temperature dependence was found to be l ¼ 1 0.2 eV with a frequency factor 2.2  10 6 s1. REFERENCES 1. 2. 3. 4.

P Yeh,T Kuwana. Chem Lett 1145, 1977. MA Eddows, HAO Hill. J Chem Soc Chem Commun 722:772^773, 1977. K Niki,T Yagi, H Inokuchi, K Kimura. J Electrochem Soc 124:1889, 1977. I Taniguchi, K Toyosawa, H Yamaguchi, K Yasukouchi. J Chem Soc Chem Commun. 1032, 1982. 5. I Taniguchi, K Toyosawa, H Yamaguchi, K Yasukouchi. J Electroanal Chem 140:187, 1982. 6. FA Armstrong. Struct Bond 72:137, 1990, and references therein. 7. FA Armstrong, GS Wilson. Electrochim Acta 45:2623, 2000, and references therein. 8. MJ Tarlov, EF Bowden. J Am Chem Soc 113:1847, 1991. 9. FM Hawkridge, I Taniguchi. Comments Inorg Chem 17:163, 1995. 10. A Ulman. An Introduction to Ultrathin Organic Films. San Diego, CA: Academic, 1991, and references therein. 11. A Ulman. Chem Rev 96:1533, 1996. 12. LH Dubois, RG Nuzzo. Annu Rev Phys Chem 43:437, 1992. 13. HO Finklea. In AJ Bard, I Rubinstein, eds. Electroanalytical Chemistry. Vol. 19. New York: Marcel Dekker, 1996, pp 109^335, and references therein. 14. GE Poirier. Chem Rev 97:1117, 1997. 15. A Ulman, ed. Self-Assembled Monolayers of Thiols. Thin Films. Vol. 24. San Diego, CA: Academic, 1998. 16. GM Whitesides, JP Mathias, CT Seto. Science 254:1312, 1991.

92 17. 18. 19. 20. 21.

22. 23. 24. 25. 26. 27. 28. 29. 30. 31.

32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50.

Niki and Gregory GM Whitesides, EE Simanek, JP Mathias, CT Seto, DN Chin, M Mammen, DM Gordon. Accts Chem Res 28:37, 1995. J Cheng, G Sa'ghi-Szabo¤, JA Tossell, CJ Miller. J Am Chem Soc 118:680, 1996. JN Richardson, SR Peck, LS Curtin, LM Tender, RH Terrill, MT Carter, RW Murray, GK Rowe, SE Creager. J Phys Chem 99:766, 1995. PE Laibinis, GM Whitesides. J Am Chem Soc 114:9022, 1992. PE Laibinis, BJ Palmer, S-W Lee, GK Jennings. In: A Ulman, ed. Self-Assembled Monolayers of Thiols. Thin Films. Vol. 24. San Diego, CA: Academic, 1998, and references therein. SE Creager, LA Hockett, GK Rowe. Langmuir 8:854, 1992. L-H Guo, JS Facci, G McLendon, R Mosher. Langmuir 10:4588, 1994. CD Bain, EB Troughton, Y-T Tao, J Evall, GM Whitesides, RG Nuzzo. J Am Chem Soc 111:321, 1989. M Buck, F Eisert, J Fischer, M Grunze, F Tra«ger. Appl Phys A 53:552, 1991. M Buck. Appl Phys A 55:395, 1992. M Buck, F Eisert, J Fischer, M Grunze, F Tra«ger. J Vacuum Sci Technol A 10:926, 1992. G Ha«hner, Ch Wo«ll, M Buck, M Grunze. Langmuir 9:1955, 1993. RC Thomas, L Sun, RM Crooks, AJ Ricco. Langmuir 7:620, 1991. DS Karpovich, GJ Blanchard. Langmuir 10:3315, 1994. DS Karpovich, HM Schessler, GJ Blanchard. In: A Ulman, ed. Self-Assembled Monolayers of Thiols. Thin Films.Vol. 24. San Diego, CA: Academic, 1998, p 43. KD Truong, PA Rowntree. J Phys Chem 100:19917, 1996. GE Poirier, ED Plyant. Science 272:1145, 1996. GE Poirier. Langmuir 15:1167, 1999. R Yamada, K Uosaki. Langmuir 14:855, 1998. J Noh,T Murase, K Nakajima, H Lee, M Hara. J Phys Chem B 104:7411, 2000. N Camillone III,TYB Leung, P Schwartz, P Eisenberger, G Scoles. Langmuir 12:2737, 1996. P Schwartz, F Schreiber, P Eisenberger, G Scoles. Surface Sci 423:208, 1999. DM Collard, MA Fox. Langmuir 7:1193, 1991. JB Schleno¡, M Li, H Ly. J Am Chem Soc 117:12528, 1995. Y-T Tao, K Pandian,W-C Lee. J Am Chem Soc 122:7072, 2000. CD Bain, J Evall, GM Whitesides. J Am Chem Soc 111:7155, 1989. CD Bain, GM Whitesides. J Am Chem Soc 111:7164, 1989. CD Bain, HA Beibuyck, GM Whitesides. Langmuir 5:723, 1989. RG Nuzzo, BR Zegarski, LH Dubois. J Am Chem Soc 109:733, 1987. RG Nuzzo, FA Fusco, DL Allara. J Am Chem Soc 109:2358, 1987. PE Laibinis, GM Whitesides, DL Allara, Y-T Tao, AM Parikh, RG Nuzzo. J Am Chem Soc 113:7152, 1991. MM Walczak, CA Alves, BD Lamp, MD Porter. J Electroanal Chem 396:103, 1995. DE Weisshaar, BD Lamp, MD Porter. J Am Chem Soc 114:5860, 1992. RG Nuzzo, LH Dubois, DL Allara. J Am Chem Soc 112:558, 1990.

Electrochemistry of Redox-Active Protein Films 51. 52. 53. 54. 55.

56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79.

93

PE Laibinis, GM Whitesides, DL Allara, Y-T Tao, AN Parikh, RG Nuzzo. J Am Chem Soc 113:7152, 1991. MA Bryant, JE Pemberton. J Am Chem Soc 113:3629, 1991. MA Bryant, JE Pemberton. J Am Chem Soc 113:8284, 1991. P Fenter. In: A Ulman, ed. Self-Assembled Monolayers of Thiols. Thin Films. Vol. 24. San Diego, CA: Academic, 1998, and references therein. TM Schultz. 2 and 3 dimensional systems studied using X-ray crystallographic techniques. PhD thesis, Riso National Laboratory and Aarhus University, 1998. AL Harris, CED Chidsey, NJ Levinos, DN Loiacono. Chem Phys Lett 141:350, 1987. MS Yeganeh, SM Dougal, RS Polizzotti, P Rabinowitz. Phys Rev Lett 74:1811, 1995. G Ha«hner, M Kinzler, C Thu«mmler, ChWo«ll, M Grunze. J Vacuum Sci Technol A 10:2758, 1992. A-S Duwez, S Di Paolo, J Ghijsen, J Riga, M Deleuze, J Delhalle. J Phys Chem B 101:884, 1997. MD Porter, TB Bright, DL Allara, CED Chidsey. J Am Chem Soc 109:3559, 1987. SD Evans, A Ulman. Chem Phys Lett 170:462, 1990. S Abrahamsson, B Dahle¤n, H Lo«fgren, I Pascher. Prog Chem Fats Other Lipids 16:125, 1978. GE Poirier, MJ Tarlov. Langmuir 10:2853, 1994. GE Poirier, MJ Tarlov. Langmuir 10:3383, 1994. GE Poirier, MJ Tarlov, HE Rushmeier. Langmuir 10:3383, 1994. N Camillone III, P Eisenberger,TYB Leung, P Schwartz, G Scoles, GE Poirier, MJ Tarlov. J Chem Phys 101:11031, 1994. E Delamarche, B Michel, Ch Gerber, D Anselmetti, H-J Gu«ntherodt, H Wolf, H Ringsdorf. Langmuir 10:2869, 1994. D Anselmetti, A Barato¡, H-J Gu«ntherodt, E Delamarche, B Michel, Ch Gerber, H Kang, H Wolf, H Ringsdorf. Europhys Lett 27:365, 1994. F Tera¤n Arce, ME Vela, RC Salvarezza, AJ Arvia. J Chem Phys 109:5703, 1998. J Kang, PA Rowntree. Langmuir 12:2813, 1996. P Fenter, P Eisenberger, KS Liang. Phys Rev Lett 70:2447, 1993. N Camillone III, CED Chidsey, G-Y Liu, G Scoles. J Chem Phys 98:3503, 1993. LH Dubois, BR Zegarski, RG Nuzzo. J Electron Spectrosc Rel Phenom 54=55:1143, 1990. P Fenter, A Eberhardt, P Eisenberger. Science 266:1216, 1994. P Fenter, F Schreiber, L Berman, G Scoles, P Eisenberger, MJ Bedzyk. Surface Sci 412=413:213, 1998. T Ishida, S Yamamoto, W Mizutani, M Motomatsu, H Tokumoto, H Hokari, H Azehara, M Fujihira. Langmuir 13:3261, 1997. JJ Gerdy,WA Goddard III. J Am Chem Soc 118:3233, 1996. L-J Wan,Y Hara, H Noda, M Osawa. J Phys Chem B 102:5943, 1998. LH Dubois, BR Zegarski, RG Nuzzo. J Chem Phys 98:678, 1993.

94

Niki and Gregory

80. L Strong, GM Whitesides. Langmuir 4:546, 1988. 81. N Camillone III, CED Chidsey, G-Y Liu, G Scoles. J Chem Phys 98:4234, 1993. 82. J Li, KS Liang, N Camillone III,TYB Leung, G Scoles. J Chem Phys 102:5012, 1995. 83. GE Poirier. J Vac Sci Technol B 14:1453, 1996. 84. PE Laibinis, GM Whitesides, DL Allara, Y-T Tao, AN Parikh, RG Nuzzo. J Am Chem Soc 113:7152, 1991. 85. PE Laibinis, GM Whitesides. J Am Chem Soc 114:1990, 1992. 86. GK Jennings, PE Laibinis. J Am Chem Soc 119:5208, 1997. 87. MM Walczak, C Chung, SM Stole, CA Widrig, MD Porter. J Am Chem Soc 113:2370, 1991. 88. R Sobocinski, MA Bryant, JE Pemberton. J Am Chem Soc 112:6177, 1990. 89. M Schoen¢sch, JE Pemberton. Langmuir 15:509, 1999. 90. CJ Sandro¡, S Garo¡, KP Leung. Chem Phys Lett 96:547, 1983. 91. TO Deschaines, KT Carron. Appl Spectrosc 51:1355, 1997. 92. JYGui, DA Stern, DG Frank, F Lu, DC Zapien, AT Hubbard. Langmuir 7:955, 1991. 93. R Heinz, JP Rabe. Langmuir 11:506, 1995. 94. A Dhirani, MA Hines, AJ Fisher, O Ismail, P Guyot-Sionnest. Langmuir 11:2609, 1995. 95. P Fenter, P Eisenberger, J Li, N Camillone III, S Bernasek, G Scoles, TA Ramanarayanan, KS Liang. Langmuir 7:2013, 1991. 96. PE Laibinis, GM Whitesides. J Am Chem Soc 114:9022, 1992. 97. Y Yamamoto, H Nishihara, K Aramaki. J Electrochem Soc 140:436, 1993. 98. CW Sheen, J-X Shi, J Martensson, AN Parikh, DL Allara. J Am Chem Soc 114:1514, 1992. 99. CA Widrig, C Chung, MD Porter. J Electroanal Chem 310:335, 1991. 100. MM Walczak, DD Popenoe, RS Deinhammer, BD Lamp, C Chung, MD Porter. Langmuir 7:2687, 1991. 101. DE Weisshaar, BD Lamp, MD Porter. J Am Chem Soc 114:5860, 1992. 102. C-J Zhong, J Zak, MD Porter. J Electroanal Chem 421:9, 1997. 103. C-J Zhong, MD Porter. J Electroanal Chem 425:147, 1997. 104. S Imabayashi, M Iida, D Hobara, ZQ Feng, K Niki,T Kakiuchi. J Electroanal Chem 428:33, 1997. 105. D Hobara, K Miyake, S Imabayashi, K Niki, T Kakiuchi. Langmuir 14:3590, 1998. 106. TW Schneider, DA Buttry. J Am Chem Soc 115:12391, 1993. 107. P Krysinski, RV Chamberlain II, M Majda. Langmuir 10:4286, 1994. 108. BD Lamp, D Hobara, MD Porter, K Niki,TM Cotton. Langmuir 13:736, 1997. 109. I Taniguchi, S Yoshimoto, M Yoshida, S Kobayashi, T Miyawaki, W Aono, Y Sunatsuki, H Taira. Electrochim Acta 45:2843, 2000. 110. T Sawaguchi, F Mizutani, S Yoshimoto, I Taniguchi. Electochim Acta 45:2861, 2000. 111. S Yoshimoto,T Sawaguchi, F Mizutani, I Taniguchi. Electochem Comm 2:39, 2000.

Electrochemistry of Redox-Active Protein Films 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127.

128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144.

95

I Taniguchi. Denki Kagaku 56:158, 1988. C Zhou, S Ye,TM Cotton, X Yu,T Lu, S Dong. J Electroanal Chem 319:71,1991. I Taniguchi, S Yoshimoto, K Nishiyama. Chem Lett 353, 1997. CD Bain, GM Whitesides. Langmuir 5:1370, 1989. EB Troughton, CD Bain, GM Whitesides, RG Nuzzo, DL Allara, MD Porter. Langmuir 4:365, 1988. TR Lee, RI Carey, HA Biebuyck, GM Whitesides. Langmuir 10:741, 1994. SE Creager, J Clarke. Langmuir 10:3675, 1994. HS White, JD Peterson, Q Cui, KJ Stevenson. J Phys Chem B 102:2930, 1998. T Kakiuchi, M Iida, S Imabayashi, K Niki. Langmuir 16:5397, 2000. J Wang, LM Frostman, MD Ward. J Phys Chem 96:5224, 1992. K Shimazu,T Teranishi, K Sugihara, K Uosaki. Chem Lett 669, 1998. K Sugihara and K Shimazu, Annual Meeting of the Electrochemical Society of Japan, Nagaoka, Japan, private communication October 1998. K Hu, AJ Bard. Langmuir 13:5114, 1997. DV Vezenov, A Noy, L Rozsnyai, CM Lieber. J Am Chem Soc 119:2006, 1997. JF Smalley, K Chalfant, SW Feldberg,TM Nahir, EF Bowden. J Phys Chem B 103:1676, 1999. N Simmons, J Zak, CC Wu, MM Walczak, MD Porter. Symposium on Langmuir-Blodgett and Self-assembled Films, abstr 12, 205th ACS National Meeting, Denver, CO, March 1993. PE Laibinis, RG Nuzzo, GM Whitesides. J Phys Chem 96:5097, 1992. JP Folkers, PE Laibinis, GM Whitesides. Langmuir 8:1330, 1992. SVAtre, B Liedberg, DL Allara. Langmuir 11:3882, 1995. T Takami, E Delamarche, B Michel, Ch Gerber, H Wolf, H Ringsdorf. Langmuir 11:3876, 1995. M Nishizawa,T Sunagawa, H Yoneyama. J Electroanal Chem 436:213, 1997. Y Sato, F Mizutani. J Electroanal Chem 438:99, 1997. S Arnold, ZQ Feng,T Kakiuchi,W Knoll, K Niki. J Electroanal Chem 438:91, 1997. S Imabayashi, N Gon, T Sasaki, D Hobara, T Kakiuchi. Langmuir 14:2348, 1998. D Hobara,T Sasaki, S Imabayashi,T Kakiuchi. Langmuir 15:5073, 1999. T Kakiuchi, K Sato, M Iida, D Hobara, S Imabayashi, K Niki. Langmuir 16:7238, 2000. SJ Stranick, AN Parikh,Y-T Tao, DL Allara, PS Weiss. J Phys Chem 98:7636, 1994. SJ Stranick, SV Atre, AN Parikh, MC Wood, DL Allara, N Winograd, PS Weiss. Nanotechnology 7:438, 1996. RK Smith, S Reed, J Monnell, PA Lewis, RS Clegg, KF Kelly, LA Bumm, JE Hutchison, PS Weiss. J Phys Chem B, in press. C Hinnen, R Parsons, K Niki. J Electroanal Chem 147:329, 1983. C Hinnen, K Niki. J Electroanal Chem 264:157, 1989. K Niki,Y Kawasaki,Y Kimura,Y Higuchi, N Yasuoka. Langmuir 3:982, 1987. T Sagara, K Niwa, A Sone, C Hinnen, K Niki. Langmuir 6:254, 1990.

96 145. 146. 147. 148. 149. 150.

Niki and Gregory

T Sagara, H Murakami, S Igarashi, H Sato, K Niki. Langmuir 7:3190, 1991. MJ Tarlov, EF Bowden. J Am Chem Soc 113:1847, 1991. M Collinson, EF Bowden, MJ Tarlov. Langmuir 8:1247, 1992. S Song, RA Clark, EF Bowden, MJ Tarlov. J Phys Chem 97:6564, 1993. RA Clark, EF Bowden. Langmuir 13:559, 1997. Z-Q Feng, S Imabayashi, T Kakiuchi, K Niki. J Chem Soc Faraday Trans 93:1367, 1997. 151. T Ruzgas, LWong, AK Gaigalas,VLVilker. Langmuir 14:7298, 1998. 152. LA Dick, AJ Haes, RP Van Duyne. J Phys Chem B 104:11752, 2000. 153. TM Nahir, EF Bowden. J Electroanal Chem 410:9, 1996. 154. MJ Honeychurch, GA Rechnitz. J Phys Chem B 101:7472, 1997. 155. Y Sato, F Mizutani. Denki Kagaku 63:1173, 1995. 156. Y Sato, F Mizutani. J Electroanal Chem 438:99, 1997. 157. KJ Stevenson, M Mitchell, HS White. J Phys Chem B 102:1235, 1998. 158. JDH Glenn, EF Bowden. Chem Lett 399, 1996. 159. F Mizutani,Y Sato, S Yabuki,Y Hirata. Chem Lett 251, 1996. 160. F Mizutani,Y Sato,Y Hirata, S Yabuki. Denki Kagaku 64:1266, 1996. 161. L Jaing, CJ McNeil, JM Cooper. J Chem Soc Chem Commun 1293, 1995. 162. J Li, G Cheng, S Dong. J Electroanal Chem 416:97, 1996. 163. I Willner, E Katz, B Willner. Electroanalysis 9:965, 1997, and references therein. 164. C Bourdillon, C Demaille, J Moiroux, J-M Save¤ant. Accts Chem Res 29:529, 1996, and references therein. 165. JM Pachence, RF Fischetti, JK Blasie. Biophys J 56:327, 1989. 166. JM Pachence, SM Amador, G Maniara, J Vanderkooi, JM Dutton, JK Blasie. Biophys J 58:379, 1990. 167. SM Amador, JM Pachence, RF Fischetti, JP McGauley Jr, AB Smith III, JK Blasie. Langmuir 9:812, 1993. 168. PL Edmiston, JE Lee, S-S Cheng, SS Saavedra. J Am Chem Soc 119:560, 1997. 169. JE Lee, SS Saavedra. In: TA Horbett, JL Brash, eds. Proteins at Interfaces II. ACS Symposium Series 602. Washington, DC: American Chemical Society, 1995, p 269. 170. K Rogers, SG Sligar. J Am Chem Soc 113:9419, 1991. 171. PS Stayton, JM Olinger, M Jiang, PW Bohn, SG Sligar. J Am Chem Soc 114:9298, 1992. 172. H-G Hong, SG Sligar, PW Bohn. Anal Chem 65:1635, 1993. 173. H-G Hong, M Jiang, SG Sligar, PW Bohn. Langmuir 10:153, 1994. 174. MA Firestone, M Shank, SG Sligar, PW Bohn. J Am Chem Soc 118:9033, 1996. 175. P Hildebrandt, M Stockburger. Biochemistry 28:6722, 1989. 176. P Hildebrandt, M Stockburger. J Phys Chem 90:6017, 1986. 177. P Hildebrandt, M Stockburger. In: AJ Alix, PL Bernard, M Manfait, eds. Spectroscopy of Biological Molecules: Proceedings of the First European Conference on Spectroscopy and Biological Molecules. Chichester, UK: Wiley, 1985, p 25.

Electrochemistry of Redox-Active Protein Films 178.

179. 180. 181.

182. 183. 184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194. 195. 196. 197. 198. 199. 200. 201. 202. 203. 204. 205. 206. 207. 208. 209.

97

P Hildebrandt, M Stockburger. In: HD Bist, JR Durig, JF Sullivan, eds. Raman Spectroscopy: Sixty Years on Vibrational Spectra and Structure. Vol 17A. Amsterdam: Elsevier, 1989, p 443. P Hildebrandt, M Stockburger. Biochemistry 28:6710, 1989. P Hildebrandt. J Mol Struct 242:379, 1991. K-J Fan, I Satake, K Ueda, H Akutsu, K Niki. In: G Dryhurst and K Niki, eds. Redox Chemistry and Interfacial Behavior of Biological Molecules. NewYork: Plenum, 1988, p 125. D Hobara, K Niki,TM Cotton. Denki Kagaku 61:776, 1993. D Hobara, K Niki, C-L Zhou, G Chumanov,TM Cotton. Colloids Surfaces A: Physicochem. Eng. Aspects 93:241, 1994. Z-Q Feng, S Imabayashi, T Kakiuchi, K Niki. J Electroanal Chem 394:149, 1995. D Hobara, K Niki,TM Cotton. Biospectroscopy 4:161, 1998. G McLendon, K Parude, P Bak. J Am Chem Soc 109:7540, 1987. JT Hazzard, G McLendon, M Cusanovich, G Tollin. Biochim Biophys Res Commun 151:429, 1988. JS Zhou, NM Kostic¤. J Am Chem Soc 114:3562, 1992. JS Zhou, NM Kostic¤. J Am Chem Soc 115:10796, 1993. JS Zhou, NM Kostic¤. Biochemistry 32:4539, 1993. A Willie, P Stayton, SG Sligar, B Durham, F. Millett. Biochemistry 31:7237, 1992. A Willie, M McLean, R-Q Liu, S Hilgen-Willis, AJ Saunders, GJ Pielak, SG Sligar, B Durham, F Millett. Biochemistry 32;7519, 1993. SH Northrup, KA Thomasson, CM Miller. Biochemistry 32:6613, 1993. Q Ling, NM Kostic¤. Biochemistry 33:12592, 1994. HA Kramers. Physica (Utrecht) 7:284, 1940. A Ansari, CM Jones, ER Henry, J Hofrichter, WA Eaton. Science 256:1796, 1992. MM Cornogorac, C-Y Shen, S Young, O Hansson, NM Kostic¤. Biochemistry 35:16465, 1996. JS Zhou, NM Kostic¤. Biochemistry 31:7543, 1992. L Qui, NM Kostic¤. Biochemistry 33:7543, 1994. MM Ivkovic¤ -Jensen, NM Kostic¤. Biochemistry 36:8135, 1997. BM Ho¡man, MA Ratner. J Am Chem Soc 109:6237, 1987. BM Ho¡man, MA Ratner. J Am Chem Soc 110:8267, 1988. JT Hazzard,TL Poulos, G Tollin. Biochemistry 26:2836, 1987. CC Moser, PL Dutton. Biochemistry 27:2450, 1988. G McLendon. Accts Chem Res 21:160, 1988. BS Brunschwig, N Sutin. J Am Chem Soc 111:7454, 1989. LD Eltis, RG Herberty, PG Barker, AG Mauk, SH Northrup. Biochemistry 30:3663, 1991. J Fietelson, G McLendon. Biochemistry 30:5051, 1991. JM Nocek, ED Stemp, MG Finnegan,TI Koshy, MK Johnson, E Margoliash, AG Mauk, H Smith, BM Ho¡man. J Am Chem Soc 113:6822, 1991.

98 210. 211. 212. 213. 214. 215. 216. 217. 218. 219. 220. 221. 222. 223. 224. 225. 226. 227. 228.

Niki and Gregory MC Waker, G Tollin. Biochemistry 30:5546, 1991. EP Sullivan Jr, JT Hazzard, G Tollin, JH Enemark. J Am Chem Soc 114:9662, 1992. MC Waker, G Tollin. Biochemistry 31:2798, 1992. SM Andrew, KA Thomasson, SH Northrup. J Am Chem Soc 115:5516, 1993. Z-Q Feng,T Sagara, K Niki. Anal Chem 67:3564, 1995. AM Becka, CJ Miller. J Phys Chem 96:2657, 1992. HO Finklea, DD Hanshew. J Am Chem Soc 114:3173, 1992. J Xu, H-L Li,Y Zhang. J Phys Chem 97:11497, 1993. JF Smallet, SW Feldberg, CE Chidsey, MR Linford, MD Newton, YP Liu. J Phys Chem 99:13141, 1995. HO Finklea, L Liu, MS Ravenscroft, S Puntri. J Phys Chem 100:18852, 1996. RA Marcus, N Sutin. Biochim Biophys Acta 811:265, 1985, and references therein. A Avila, BW Gregory, K Niki,TM Cotton. J Phys Chem B 104:2759, 2000. HB Gray, JRWinkler. Annu Rev Biochem 65:537,1996, and references therein. Y Wang, E Margoliash. Biochemistry 34:1948, 1995. K Niki, KP Pressler, JR Sprinkle, H-Y Li, E Margoliash. Elektrokhimiya, in press. AE Kasmi, JM Wallace, EF Bowden, SM Binet, RJ Linderman. J Am Chem Soc 120:225, 1998. S Terrettaz, J Cheng, CJ Miller, RD Guiles. J Am Chem Soc 118:7857, 1996. A Ehrenberg, S Paleus. Acta Chem Scand 9:1955, 1955. AK Gaigalas, G Naiura. J Colloid Interface Sci 193:60, 1997.

3 Biomimetic Membranes on Metal Supports John T. Elliott, Curtis W. Meuse, Vitalii Silin, Susan Krueger, John T. Woodward, Teresa Petralli-Mallow, and Anne L. Plant National Institute of Standards and Technology, Gaithersburg, Maryland, U.S.A.

1 1.1

INTRODUCTION The Biological Membrane

Biological membranes are complex and dynamic structures. The biological functions associated with membranes involve a number of di¡erent molecular species, and theories of how the molecular species are organized are still evolving.The ‘‘£uid mosaic’’ model of freely di¡using proteins and lipids in a two-dimensional £uid [1] has been re¢ned to account for the presence of functional domains [2^4]. The composition, dynamics, source, and even existence of discrete segregated features, such as membrane ‘‘rafts,’’are currently sources of signi¢cant controversy [5,6]. As a result of ambiguities in the complex nature of biomembranes, better physical and theoretical models are constantly under development. Both lipid and protein components of membranes are responsible for membrane function as well as structure. In addition, membrane protein function can be in£uenced by the lipid matrix that surrounds it. Attempts to 99

100

Elliott et al.

understand integral membrane protein structure at the atomic level by crystallization and di¡raction remains a critical challenge.Only a few membrane protein structures have been determined by X-ray di¡raction [7^9]. It is assumed that the protein’s crystal structure is likely to be the same as that of the active protein in its native membrane environment; in fact, the function of bacteriorhodopsin in crystals has been demonstrated directly [10]. Nevertheless, part of the challenge associated with membrane protein structures is that their native structure and function can be highly dependent on an appropriate lipid environment. Two-dimensional protein crystallization provides some promise of a general way of crystallizing membrane proteins in the presence of lipid or detergent [11]. In general, the lipid environment is highly dynamic, and moving molecules are not good subjects for X-ray di¡raction. Understanding how membrane protein structural changes are responsible for functions such as transport and signal transduction is a critical issue. Active biological membranes are dynamic on a time scale of minutes and hours at the cellular level (e.g., membrane tra⁄cking), and on a scale of milliseconds and nanoseconds on the intermolecular and intramolecular levels. The function of membrane proteins during transport of molecules across the lipid barrier, or during intracellular kinase activation resulting from an extracellular ligand binding to a receptor, likely occurs via protein conformational changes. In addition to such intramolecular dynamics, protein and lipid components di¡use laterally in the plane of the membrane, and di¡erent approaches to study di¡usion can provide information on appropriately di¡erent time scales [12,13]. Furthermore, lipid components can di¡use between the membrane and the surrounding aqueous milieu, by a process that is well explained by thermodynamic considerations [14]. 1.2

Traditional Model Membranes for Studying Biology and Biophysics

Model membrane systems have historically been essential to the development of our understanding of biological membranes. Their use allows us to study isolated features of these otherwise highly complicated structures. Knowledge of physical constants and principles provide the backdrop for careful measurements in limited-component systems, which in turn provide the basis of hypotheses that can be tested in more complex systems. Lipid vesicles or liposomes [15], black (or more recently, bilayer) lipid membranes [16], multilayers of lipid for di¡raction studies [17], planar lipid bilayers on solid supports [18^19], and more recently, bicelles [20] are all forms of model systems that have distinct advantages and disadvantages as models of biological membranes. The patch clamp technique [21] is probably the approach

Biomimetic Membranes on Metal Supports

101

that allows the study of membranes in their most native form, but its use is limited mainly to electrical measurements and it is technically challenging. The liposome is probably the most popular model membrane, and has contributed greatly to our understanding of biological membranes. Liposomes are usually made from puri¢ed lipids or their mixtures, and can be fabricated to incorporate membrane proteins (proteoliposomes). The structure of the spherical bilayer surrounded by and encapsulating an aqueous solution is thermodynamically stable and assembles spontaneously when phospholipids are mixed with water. As a result of its stability and ease of formation, it is a model membrane system that has found application in arenas outside basic research, namely, in drug delivery [22] and, to a lesser extent, in diagnostics. Supported lipid bilayers on glass are a model membrane system on a planar support. The planar geometry provides the opportunity to examine membranes with techniques that are not amenable to vesicles. They have been employed in many important biophysical studies (for examples, see Ref. 23 and the many other contributions that followed this seminal work [24^26]), and they are most often probed with £uorescence microscopy [23] and infrared spectroscopy [27]. These membranes are composed of two lea£ets of natural lipid, and are associated with the surface through electrostatic interactions. Although this bilayer is probably £oating above the surface on a water layer of several tens of angstroms, it does not provide an optimal biomimetic matrix for all transmembrane proteins. Data indicate that some transmembrane proteins associated with planar-supported bilayers on glass do not di¡use readily [28,29], while others appear to di¡use freely [30]. In addition, the association of these layers with the surface is sensitive to the solution ion strength, and so forming the bilayer to take background data before adding protein is often not successful. Neutron re£ectivity measurements indicate that incomplete surface coverage can be a problem with these bilayers [31]. However, supported lipid bilayers on glass and on electrode surfaces supply many opportunities in experimental biophysics, biology and materials science. 1.3

The Hybrid Bilayer Membrane

If one could design the ideal model membrane, it would form by selfassembly, be robust enough to be probed for long periods of time, be appropriately £exible and dynamic on a molecular scale to allow lateral di¡usion and insertion of molecules, and be addressable analytically by a number of techniques. This latter point is extremely important, because the use of multiple techniques can reduce ambiguity and prevent erroneous conclusions.

102

Elliott et al.

FIG. 1 The simplest form of an alkanethiol/phospholipid hybrid bilayer membrane on a gold-coated support. Thiols with many different length alkane chains have been used to form the proximal leaflet closest to the gold layer. Many different compositions of lipid and lipid plus protein have been used to form the distal leaflet. (From Ref. 33.)

The focus of this chapter will be bilayers that are formed at metal surfaces. Metal-tethered membranes are similar to glass-supported bilayers, but they are composed of two di¡erent lipid lea£ets.What we have called the hybrid bilayer membrane (HBM) [32] is composed of a monolayer of alkanethiol, which is tightly bound to the surface through a strong association between sulfur and gold, and a second monolayer of phospholipid. This simple construct is depicted in Fig. 1. There are many variations on this theme, which di¡er from this most simple formulation. These HBMs use other sulfur-containing molecules for the tethering layer, and contain di¡erent lipids, proteins, and other molecules in the bilayer. In this chapter, we will refer to all forms of bilayers that involved thiol-gold tethers as hybrid bilayers or HBMs. A short review of this topic has been presented previously [33]. 1.3.1 Advantages and Potential Disadvantages of Hybrid Bilayers on Metal Supports Hybrid bilayers on metal supports are easy to prepare and are very stable. These attributes make this model membrane adaptable to application in sensors, diagnostics, and assays. Optimization of thiol-tethered membranes to make them more biomimetic is an area of active research. Focus has been placed on introducing an aqueous compartment between the metal support and the alkane chains of the bilayer, and increasing the £uidity of the lipidic moieties in the tethering layer. The use of the exquisite properties associated

Biomimetic Membranes on Metal Supports

103

with biological membranes, such as molecular recognition, dynamic reorganization, and formation of channels and transport systems within an otherwise insulating layer, can now be employed in sensor applications.This has been accomplished quite elegantly by the Cornell group, which has produced robust, sensitive, and commercially viable devices based on metalsupported hybrid bilayers [34]. Hybrid bilayers on metal are more robust than glass-supported bilayers. The high degree of stability is provided by covalently tethering a hydrophobic layer to the solid support through a sulfur and gold interaction. (Whether or not a covalent bond is formed between sulfur and gold is debatable; however, the free energy of this interaction is some 40^50 kcal/mol [35]). Covalent tethering can also be achieved through silane bond formation to glass and other materials, but tethering molecules using thiol groups is often easier and more reproducible than silane chemistry. Hybrid bilayers on metal supports are analytically addressable by a wide range of surface analytical techniques. Because the alkanethiol/lipid bilayers are on metal, highly informative techniques such as surface plasmon resonance (SPR) impedance analysis, and electrochemistry can be used to probe kinetics and to provide structural information. These techniques are particularly powerful when used simultaneously. Simultaneous use of SPR and impedance analysis to measure the kinetics of bilayer formation showed di¡erences in the density of lipid layers that added to di¡erent thiol-tethered monolayers [36], demonstrated the hydration of a cholera toxin layer bound to an HBM containing GM1 ganglioside [37], and permitted discrimination between speci¢c and nonspeci¢c binding of ligands to the supported membrane surface [38]. These analytical techniques are highly amenable for use in routine screening and sensing applications. HBMs can be characterized by infrared spectroscopy, atomic force microscopy, ellipsometry, and with techniques that have not been typically used in the study of biology. For example, the high-vacuum technique, ESCA (electron spectroscopy for chemical analysis), can be used to de¢ne the surface properties of an HBM at the atomic level [39]. The use of nontraditional methods for querying model membranes provides the opportunity to ask (and answer) questions about membranes that have been di⁄cult or impossible to address previously. A distinct disadvantage associated with the use of the metal support is the limitations to £uorescence microscopy measurements, which are imposed by the electromagnetic characteristics of the metal. However, recent advances in surface plasmon enhanced £uorescence [40,41] may lead to the wider use of £uorophores with metal-supported bilayers. To the extent that hybrid bilayers are biomimetic, their use can enhance the understanding of membranes by simplifying experiments,

104

Elliott et al.

reducing ambiguity in data interpretation, and allowing for greater experimental control. By combining experimental techniques, this approach may eventually allow simultaneous monitoring of membrane lipid and protein structural changes, and membrane function, so that the relationship between structure and function can be better understood. Because of the stability and ease of formation, hybrid bilayers are also excellent model systems for studying and developing new measurement techniques. The robustness of this well-behaved model system simpli¢es experimental complexities. Two examples we will discuss are neutron re£ectivity of single bilayer membranes, and nonlinear optical spectroscopy. In this chapter,we will focus primarily on fabrication, characterization tools, and applications for supported model membranes on metal surfaces. This relatively narrow topic has been strongly contributed to during this past decade. However, it should be appreciated that the literature describing planar nontethered bilayers and bilayers tethered on glass supports is vast, and that these are important alternative approaches to the study of biological membranes. 2

FABRICATION, STABILITY, AND CHARACTERIZATION OF HYBRID BILAYERS

The structure of the alkanethiol/phospholipid hybrid bilayer as shown in Fig. 1 has been examined using the techniques described below. We have prepared HBMs with a variety of alkanethiols and phospholipids, such as 1,2-dimystroyl-sn-glycero-3-phosphocholine (DMPC), 1,2-dipalmitoylsn-glycero-3-phosphocholine (DPPC), 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), and 1,2-dioleoyl-sn-glycero-3-phosphoserine (DOPS). Construction of a simple hybrid bilayer membrane is a two-step process. First, an alkanethiol self-assembled monolayer [42] is formed on a bare gold surface. Next, a phospholipid monolayer is transferred onto the hydrophobic alkanethiol monolayer. Addition of the phospholipid monolayer can be achieved by Langmuir-Blodgett trough transfer, solvent painting, detergent dilution, or by exposing the alkanethiol monolayer to phospholipid vesicles [33]. The ability to use phospholipid vesicles to form an HBM is convenient, since vesicles are easy to prepare and use in many experimental procedures. When phospholipid vesicles are added to an aqueous solution above an alkanethiol-covered surface, a phospholipid monolayer spontaneously assembles onto the alkanethiol lea£et.The hydrophobicity of the alkanethiol layer provides the thermodynamic driving force for bilayer formation. The exact molecular details of the process have not been elucidated, although the kinetics of the process are consistent with a vesicle-dependent rather than a

Biomimetic Membranes on Metal Supports

105

monomer-dependent process [43]. A popular schematic depicting the monolayer formation process as an unfolding of vesicles at a hydrophobic surface has been presented [44]. Several di¡erent methods have been used to follow bilayer formation in real time. Studies using surface plasmon resonance, impedance spectroscopy, and the quartz crystal microbalance o¡er mechanistic detail about the rate-determining step and/or information about the resulting structure.

2.1

Surface Plasmon Resonance

2.1.1 Introduction to Surface Plasmon Resonance (SPR) Surface plasmons are electron density oscillations at an interface between a metal and a dielectric. For surface plasmons to exist, the two materials must have dielectric permeabilities, e, of opposite sign [45]. Plasmons propagate in the plane of the interface, and the electromagnetic ¢eld decays exponentially with distance, z, from the interface. The real part of the dielectric constant for most metals is negative and opposite in sign to air or water in the spectral range from infrared to visible. For biotechnology applications, gold is probably the most suitable metal because of its optical and chemical properties. Light impinging on a metal surface can couple into surface plasmons only if its electromagnetic wave momentum is equivalent to that of the surface plasmon. The surface plasmon momentum, ksp, is dependent on em and ed (for the metal and dielectric media, respectively), and is proportional to the angular frequency of the light, o. The surface plasmon momentum is larger than the momentum of the incident light, kb, as follows: ksp ¼ o=c ½em ed =ðem þ ed Þ1=2 1=2 kb ¼ o=c ed

ð1Þ ð2Þ 1=2

1=2

Using a prism coupler with refractive index, np ¼ ep > ed , as shown in Fig. 2, is one way of achieving the resonance condition. To achieve resonance, the angle, j, of the incident light must be adjusted such that ksp ¼ kb np sin j

ð3Þ

A sharp minimum in re£ection coe⁄cient occurs at angles at which the incident light is coupled into surface plasmon excitation, as shown in Fig. 2. The angle at which this minimum occurs is strongly dependent on the complex refractive indices of all the media, and the thickness of the layers that are deposited at the gold surface. As layer thickness grows, the SPR minimum shifts correspondingly to larger angles, as shown in Fig. 2. This shift can be

106

Elliott et al.

FIG. 2 (a) Diagram of surface plasmon excitation. The bulk propagation vector, kb, increases np times inside a prism with refractive index np. By changing the angle of incidence, j, the component of the propagation vector parallel to surface ðnp kb sin jÞ can be matched to the surface plasmon wave vector, kSP, on a metal surface with a refractive index, nm, and thickness, d. I0 and I are intensities of the incident and reflected laser beam, and R ¼ I=I0 is the reflection coefficient. (b) Dependence of the reflection coefficient on the angle of incidence. The solid line is before, and the dotted line is after addition of material to the surface. The sharp minimum appears due to surface plasmon excitation. The SPR line width is denoted w. (From Ref. 47.)

followed over time as biological molecules interact with the surface. The technique can be sensitive to less than a 0.5 — change in layer thickness. The sharpness of the angular dependence of the re£ectivity is dependent on how well the resonance condition is achieved, and on the smoothness of the surface. Roughness of the metal, or inhomogeneity in the lateral distribution of material binding to the surface, can result in broadening of the angular dependence of the re£ectivity. This can reduce sensitivity, particularly if the roughness is associated with the metal layer, because it can make the measurement of the angle of minimum re£ectivity ambiguous. On the other hand, the broadening can in some cases provide information about

Biomimetic Membranes on Metal Supports

107

how homogeneously species from solution bind to the reactive surface [46,47]. The power of the SPR technique is that it provides high sensitivity and label-free detection of analyte interaction with the surface. As a result, it allows real-time measurements of the kinetics of the interaction. Furthermore, it can be combined with other methods such as electrochemistry for simultaneous measurements [36]. SPR is suitable for studying hybrid bilayer formation and protein ^ membrane interactions. 2.1.2 Kinetics of Formation of HBMs The kinetics of HBM formation can be easily followed in real time using SPR. Phospholipid vesicles are allowed to £ow across a hydrophobic alkanethiol monolayer, and as lipid adds to the surface, the angle of minimum re£ectivity from the surface shifts correspondingly to higher angles.The ¢rst example of using SPR to observe the kinetics of formation of a surfacesupported bilayer was published by Spinke et al. [48]. We have used SPR in an attempt to elucidate the physicochemical mechanisms by which these bilayers form. The process of HBM formation is an example of the e¡ect of interfacial free energy on molecular reorganization at a surface.We can envision at least two mechanisms of bilayer formation, as shown in Fig. 3. The bilayer could form by reorganization of intact phospholipid vesicles, as was originally proposed by Kalb et al. [44]. However, since phospholipid molecules have a small but ¢nite solubility in water, it is possible that the formation of the bilayer occurs by addition of individual phospholipid molecules to the surface. If this is the case, then the composition of the resulting bilayer may be signi¢cantly di¡erent than one that would be obtained if a cooperative deposition of molecules comprising the vesicle lamellae is responsible for bilayer formation. We distinguished these two mechanisms by examining the kinetics of bilayer formation at di¡erent concentrations of vesicles [43]. The experimental results for addition of DMPC to a hydrophobic alkanethiol surface are presented in Fig. 4. The data show that the time dependence for the increase in surface-layer thickness is strongly dependent on the concentration of phospholipid vesicles in the solution. For this study, experiments were conducted under static, non£owing, conditions. The kinetics of phospholipid addition to the alkanethiol layer are generally more rapid under conditions of continuous £ow, where the rate of mass transport of vesicles to the surface is enhanced, but these data are more di⁄cult to model than data obtained under the conditions used in this study. The model presented is intended to rationalize the kinetics of the bilayer-forming process. We consider the following probable sequence of events: di¡usion of vesicles and/or lipid monomer molecules to the surface

108

Elliott et al.

FIG. 3 Two mechanisms for the formation of HBMs from vesicles were considered: (1) that intact vesicles add to the hydrophobic surface, or (2) monomer phospholipid transfers from vesicles, through the aqueous phase to the hydrophobic surface. (From Ref. 33.)

FIG. 4 Concentration-dependent kinetics of HBM formation. Concentrations listed to the right of each curve are in mg lipid/mL. Addition of lipid to the surface was allowed to occur under unstirred conditions. Dotted lines are the results of fitting Eq. (4). (From Ref. 43.)

Biomimetic Membranes on Metal Supports

109

where they are either repelled or retained, followed by adsorption and/or reorganization at the surface, and with time, saturation of the surface with a monolayer of lipid. The mathematical form of the model that was used to describe the data expresses the time-dependent surface concentration in terms of the maximum surface coverage of phospholipid, G, the bulk concentration of an adsorbing species, C0, the bulk di¡usion of an adsorbing species, D, and a surface reorganization rate constant, K: " ( " )  2 1=2 # KC0 D K t K 2t GðtÞ ¼ Gm 1  exp 1 exp erfc K2 D D Gm   !# 2 Dt 1=2 þ ð4Þ K p where erfc indicates the complementary error function [49]. The rate constant K characterizes the in£uence of surface reorganization kinetics on the surface concentration of lipid. We emphasize that K refers not only to the surface adsorption of intact vesicles, but also incorporates the vesicle disruption process and the surface migration dynamics of lipid monomer. K therefore designates an entire set of quite complex surface kinetic processes. Large values of KðK 2 =D  1Þ correspond to instantaneous surface kinetics (compared to di¡usion) and therefore perfect adsorption at available surface sites. Small values of K imply that bulk material is partially re£ected or repelled by the active surface, and that surface kinetic processes are slow relative to di¡usion. The results of ¢tting the kinetic data to Eq. (4) are shown in Fig. 4. As can be seen, the model ¢ts all the data sets well. Further examination of the data by comparing them to expressions for the limiting cases of the expression described in Eq. (4) indicated that there are two apparent concentration regimes in which the rate of bilayer formation is limited by di¡erent processes. At high vesicle concentrations, the rate of bilayer formation is dependent on the di¡usion constant of the vesicles. This indicates the development of a time-dependent concentration gradient pro¢le at the surface, which develops because the reorganization process at the surface is occurring quickly relative to the rate of di¡usion. However, at lower vesicle concentrations, the rate of vesicle di¡usion to the surface is not kinetically limiting. Surface reorganization is apparently slow relative to the rate of vesicle di¡usion, suggesting that perhaps under these conditions, we observe nonproductive association of lipid vesicles with the surface, i.e., approach of vesicles that fail to reorganize at the surface. Interestingly, ¢tting the data to this model that explicitly took into account contribution from monomer phospholipid revealed that lipid

110

Elliott et al.

monomer does not appear to play a signi¢cant role in bilayer formation. The ¢tted parameters resembled those found for the model that considered the contribution from vesicles only.The apparent unimportance of phospholipid monomer in the addition of lipid to the surface may re£ect that vesicles contribute much more to covering surface area per ‘‘event’’ than do individual lipid molecules. The fact that formation of the bilayer does appear to be a vesicle-dependent process makes it reasonable to estimate the HBM composition based on the composition of the vesicle precursors. This is important for applications that use ligand- or receptor-containing vesicles for the construction of an HBM surface. 2.1.3 Using SPR to Measure Specific Interactions at the Membrane Surface Given this information about the mechanism of HBM formation from lipid vesicles, it is easy to prepare membrane mimics that contain biological recognition moieties. There are many examples of biomembrane-surface speci¢c processes being investigated with the use of HBMs and SPR. Several of these studies are described in the applications section of this chapter. Figure 5 shows SPR monitoring of cytochrome c adsorption to HBMs formed from vesicles containing phosphatidylcholine, a net neutral

FIG. 5 The formation of an HBM by the addition of phospholipid vesicles to an alkanethiol monolayer and the adsorption of protein to the HBM can be monitored by SPR. Cytochrome c adsorbs to the surface of HBMs prepared from vesicles containing negatively charged phospholipid. The w indicates the flow of buffer to wash the surface. (From Ref. 50.)

Biomimetic Membranes on Metal Supports

111

phospholipid, and phosphatidic acid, a negatively charged phospholipid [50].Cytochrome c did not bind to HBMs that were formed from vesicles that contained only the neutral phospholipid. Although many of the published SPR studies have been performed on custom-built instrumentation, it was demonstrated that HBMs could be formed and studied in a commercial SPR device [51]. Since then, ‘‘sensor chips’’ designed for HBM-based studies have been made commercially available, facilitating studies such as antibody ^ antigen recognition on immobilized receptors [52] and protein ^ membrane interactions [53] in commercial SPR devices. 2.2

Electrochemistry and Impedance Analysis

Electrochemistry and impedance analysis techniques are highly applicable to the study of membranes tethered to a metal surface. HBMs are completely supported on an electrode that is parallel to the membrane surface. The unique electrical and structural properties of phospholipid membranes and the thiol tethering layer provide the opportunity to use electrical techniques to characterize fabrication approaches, self-assembly processes, and structural details of the model membranes. In some cases, these techniques can also be used to monitor the activity of peptides, proteins, and other biomolecules introduced into the membrane. 2.2.1 Impedance Analysis Impedance spectroscopy is an AC method involving the application of small AC potentials to charge and discharge the surface with solution electrolyte ions.The amplitude and delay in surface charging with respect to the applied potential are examined by applying small positive and negative potentials at various frequencies. The physical structure of the material on the electrode surface can be approximated as a combination of electrical circuit components in an equivalent circuit model.The frequency dependence of the impedance response can be simulated for a chosen equivalent circuit model, and compared with the impedance response observed for the actual sample. In an iterative process, values for equivalent circuit components can be adjusted until good correspondence between the data and the model are achieved. Impedance studies have been used to characterize alkanethiol monolayers [54^57], and both simple and complex HBMs [34,58]. It is also important to note that there is a large body of data that describes the electrical properties of actual biological membranes [59,60], and these data aid in predicting the electrical response of the metal-tethered model membranes. When a metal surface coated with a thin organic ¢lm, such as an alkanethiol monolayer or an HBM, is introduced into an electrochemical cell

112

Elliott et al.

containing an electrolyte solution, its cross section is similar to that of a plane plate capacitor. The organic coating acts as an insulating dielectric ¢lm parallel the metal electrode. The complex impedance response of this electrode is de¢ned by i ð5Þ oC where Rs is the solution resistance and i=oC is the frequency-dependent impedance due to capacitance. Capacitance is dependent on the physical properties of the insulating layer in the following way: Z ¼ Rs 

A ð6Þ d where e0 is the permittivity of free space (8.85 1014F/cm), e is the dielectric constant of the organic ¢lm, d is the dielectric layer thickness, and A is the area of the electrode. Thus the value of e=d for the dielectric ¢lm can be determined. If the dielectric constant of the ¢lm is known, the impedance analysis provides information about the structure (i.e., the thickness) of the dielectric layer. Therefore, if the impedance response can be ¢tted to simple equivalent circuit modes, these data can be used to obtain gross structural information about the molecular ¢lms that are used to fabricate HBMs. The capacitance values for the HBM and alkanethiol layer can be used to calculated the capacitance of the lipid layer by C ¼ e0 e

1 1 1 ¼ þ CT Cal Cpl

ð7Þ

where CT is the total capacitance, Cal corresponds to the capacitance of the alkanethiol monolayer,which can be determined before adding phospholipid to the layer, and Cpl is the capacitance of the phospholipid layer. Infrared and Raman spectroscopy have shown that structural changes in the alkanethiol layer upon addition of the phospholipid layer are negligible [61], thus supporting the validity of using this additive model. Determination of Cpl allows assessment of the phospholipid monolayer thickness and dielectric constant. Examples of impedance data and a simple equivalent circuit model for alkanethiol monolayers and HBMs are shown in Fig. 6. The Bode plot (Fig. 6c) shows absolute impedance (jZj) versus frequency, where rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 ð8Þ jZj ¼ Rs2 þ 2 2 o C When the frequency of the applied potential is large, the capacitance element is negligible, and the jZj is equal to the solution resistance. At lower

Biomimetic Membranes on Metal Supports

113

FIG. 6 Impedance response from a simple alkanethiol-based HBM. The impedance response of the HBM can be modeled with the circuit shown (A). The capacitance elements correspond to the alkane chain regions of the thiol and phospholipid, respectively. The impedance responses before and after fusion of DMPC vesicles to an octadecanethiol monolayer represented in complex capacitance (B), Bode (C), and phase (D) plots. The capacitance of the octadecanethiol monolayer is 0.89 mF/cm2. The total capacitance drops to a value of 0.58 mF/cm2 after the addition of DMPC phospholipid vesicles.

frequencies, the capacitive impedance becomes much greater the solution resistance and 1 or logðjZjÞ   logðoÞ  logðCÞ ð9Þ Z oC A plot of logðjZjÞ versus log(o) at frequencies where the capacitive element dominates the impedance response is then a line with slope 1 and intercept at  logðCÞ. A plot of phase versus frequency is presented in Fig. 6d. The phase difference between the applied potential and potential across the capacitor is described by     ImðZÞ 1 y ¼ tan1 or y ¼ tan1 ð10Þ ReðZÞ oðRs CÞ

114

Elliott et al.

where Im(Z) and Re(Z) are the imaginary and real parts of the complex impedance, respectively. In the limiting case where o ! 1, the phase angle is 0 ; and in the case where o ! 0, the phase angle is 90. In a plot of phase versus log(o), the phase angle should approach 90 as the applied potential frequency approaches 0. The complex capacitance plot shown in Fig. 6b is another useful way to represent impedance data, since the capacitance value can be read directly from the plot. The axes of this plot are Im(Y)/o and Re(Y)/o, where Y is the admittance (Z1) and, Im(Y) and Re(Y) are the imaginary and real parts of the admittance: ReðY Þ oRs C 2 ¼ 2 2 2 o o Rs C þ 1

and

ImðY Þ C ¼ 2 2 2 o o Rs C þ 1

ð11Þ

In the limiting case where o ! 0, Im(Y)/o approaches C. The impedance response of the circuit on the complex capacitance plane is a semicircle with a radius of C/2 o¡set from the origin on the imaginary axis by C/2 (Reference 62 shows how these variables ¢t into an equation of a circle). The impedance data shown in Fig. 6 were collected before and after DMPC vesicles were fused to a C18SH alkanethiol layer. The Bode plot exhibits a slope of 1, the complex capacitance plot is a semicircle, and the phase is at 90 at low frequencies. By examining the data using di¡erent plotting methods, it is possible to conclude that the spectra can be ¢tted with a simple capacitor in series with a solution resistor model. In practice, simple equivalent circuit models such as that shown in Fig. 6 do not always describe the data adequately. Even alkanethiol monolayers are known to deviate from ideal capacitor behavior at low frequencies, and it has been shown that they are better represented by complex phase elements (CPEs) instead of capacitors [63]. Unfortunately, values obtained with the use of CPE elements in the equivalent circuit cannot be easily interpreted as well-de¢ned structure-based electrical elements. Clearly, equivalent circuit models must be chosen with discretion. Simply adding more electrical elements to a model can often improve the ¢t to the data but without necessarily improving physical insight. Another caveat to impedance analysis is that only e=d can be obtained directly from capacitance measurements, and the calculation of either value will depend on information about one of the values. The exact dielectric constant of the phospholipid monolayer is unclear, as it appears to depend on the phospholipid acyl chain and the method of lipid deposition. A dielectric constant of 2.1 is obtained with unsupported black lipid membranes (BLMs), and this value is often used to describe a phospholipid monolayer [64,65]. Plant et al. have shown that the dielectric constant for a series of monounsaturated lipid monolayers prepared by fusing vesicles to an alkanethiol

Biomimetic Membranes on Metal Supports

115

surface is 2.7 [58]. Lingler et al. used simultaneous SPR and impedance measurements to show that a DMPC monolayer formed by vesicle fusion onto a thiol phospholipid monolayer also has a dielectric constant of 2.7 [36]. Discrepancies may arise from the use of organic solvent (e.g., decane) in the BLM membrane preparation [58]. The solvent remains dissolved in the bilayer and reduces the dielectric constant towards that of a pure alphatic oil (e  2:0). Plant et al. showed that HBMs formed with vesicles prepared with 5:1 phospholipid:decane exhibited lower capacitance than those prepared with lipid vesicles [32]. Furthermore, when this HBM was kept under the electrolyte solution for 16 h, its capacitance increased to near expected values as decane partitioned out of the membrane.Using Eqs. (6) and (7), and the data shown in Fig. 6, the capacitance of the DMPC layer is 1.6 mF/cm2 and the thickness of the hydrocarbon region of the DMPC layer is 14 —, using a dielectric constant of 2.7. Upon addition of phospholipid vesicles to a solution above an alkanethiol monolayer, one can monitor changes in the impedance of the surface ¢lm as the HBM is formed. The kinetics of this monotonic response are highly analogous to the shape of the respective SPR response, suggesting that both techniques are reporting on the average change in surface area covered by the additional dielectric phospholipid layer [36,51]. Impedance analysis has been used extensively to evaluate the formation and physical properties of simple HBMs formed on well-packed and well-de¢ned alkanethiol surfaces. For example, Steinem et al. evaluated HBMs prepared by vesicle fusion, Langmuir-Blodgett trough transfer, and detergent dialysis techniques and found that there is little variation in the capacitance of the resulting phospholipid monolayer [66]. Ding et al. reported the use of a paint-freeze technique to remove organic solvent from painted monolayers [67]. An interesting study using a combination of SPR and impedance measurements on a single sample demonstrated that the organization of the phospholipid monolayer is dependent on the chain length of the underlying alkanethiol monolayer [36]. Phospholipid layers on C2^ C7 alkanethiol monolayers exhibited high capacitance and low thickness values, suggesting some level of disorder. The disorder appears to decrease as the thickness of the alkanethiol monolayer increases. Other studies have shown that HBM capacitance has a temperature dependence that nearly re£ects the transition temperature of the phospholipid monolayer [36,68]. The resistivities of an HBM have been measured and are as high as 5 107 O-cm2 [58]. This is within an order of magnitude of the resistivity observed across unsupported model membranes [69]. Nevertheless, although the upper phospholipid lea£et has characteristics similar to those in biological membranes, the lower alkanethiol lea£et of the simple HBM is signi¢cantly di¡erent from that expected in a biological membrane. More

116

Elliott et al.

advanced HBMs that contain hydrated reservoirs between the phospholipid membrane and the gold electrode have been fabricated to more closely mimic true biological membranes [34]. The alkylthiolates used to prepare these HBMs include a hydrophilic ethylene oxide [70^75] or peptide [76] linker between the thiol and the hydrophobic alkyl chains. Once membranes are formed on the alkylthiolate layer, the hydrophilic region de¢nes the reservoir characteristics. The predicted electrical model for tethered membranes with a proximal aqueous reservoir is shown in Fig.7. A solution resistor is in series with a parallel resistor/capacitor to represent the insulating alkane regions of the

FIG. 7 Impedance response from an alkylthiolate-based HBM with a thin hydrated reservoir proximal to the electrode. HBMs with hydrated reservoirs between the membrane and the solid support are intended to accommodate the extramembraneous regions of transmembrane proteins (A). The circuit model used to fit these data is also shown (A). Complex capacitance (B), Bode (C), and phase (D) plots of a mixed b-mercaptoethanol/1-thia-hexa(ethylene oxide)-octadecane tethering monolayer before and after phospholipid addition by ethanol painting are shown. The capacitance transition in the Bode plot, the dip in the phase plot, and the dual semicircle-like features in the complex capacitance plots are typical features in the impedance response of this type of HBM. The values for the equivalent circuit determined from fitting routines are CM ¼ 1.0 mF/cm2, RM ¼ 5000 O-cm2, CHYD ¼ 9 mF/cm2.

Biomimetic Membranes on Metal Supports

117

membrane. A second capacitor in series with the parallel RC components represents the Helmholtz capacitance in the hydrated reservoirs.There are a number of descriptions of the construction and characterization of HBMs containing aqueous reservoirs [34,77,78]. Figure 7 shows the impedance analysis of an HBM with an aqueous reservoir that was prepared in our laboratory. The structure is created by applying POPC in ethanol over a mixed monolayer of b-mercaptoethanol and thia-hexa(ethylene oxide)-octadecane [70] and e¡ecting membrane formation by the addition of bu¡er [34]. The b-mercaptoethanol is intended to prevent tight alkane packing in the thiol-containing monolayer, and to allow the thia-hexa(ethylene oxide)-octadecane to assume a more disordered conformation than it does as a single-component monolayer [70]. The impedance response from these structures shows distinct spectral features in both the Bode, phase, and complex capacitance plots that are clearly di¡erent than those for the simple HBM shown in Fig. 6. In Fig. 7, the impedance response shows a transition at approximately 100 Hz that is a characteristic of the tethered membranes containing an aqueous reservoir. The Bode plot shows that between 60,000 and 100 Hz, the HBM impedance is dominated by the membrane capacitance. Below 10 Hz, the impedance is dominated by the Helmholtz capacitance of the reservoir region. This is represented as two adjacent semicircular-like features in the complex capacitance plot. Fitting the data to the equivalent circuit model results in a membrane capacitance that corresponds to 1.0 mF/cm2, a membrane resistance of  5500 O-cm2 and a Helmholtz capacitance of  9 mF/cm2. The capacitance and speci¢c conductance of the membrane is signi¢cantly higher than that observed with both black lipid membranes and alkanethiol-based HBMs. This is likely a result of defects in the membrane. The Helmholtz capacitance is lower than the  20^35 mF/cm2 typically observed at a gold electrode. This di¡erence is likely due to the reduction of gold surface area by the alkylthiolates and possibly due to partial ordering of the water molecules so that the e¡ective dielectric constant is signi¢cantly less than that in bulk solution [72]. Steinem et al. formed similar reservoir-containing tethered membranes by POPC vesicle fusion to thia-tri(ethylene oxide)-phospholipid monolayers and obtained membranes with resistance values of 11,000 O-cm2 [73]. This low value suggests defects are present in the phospholipid surface. Stora et al. formed tethered bilayers via detergent dialysis techniques and found a speci¢c membrane resistance of 5000 O-cm2 [78]. This is similar to the values we obtain with ethanol-painted POPC membranes as shown in Fig. 7. Raguse et al. used impedance analysis to examine thiol-tethered membranes with di¡erent reservoir volumes [72]. The phytanoyl-based

118

Elliott et al.

phosphatidylcholine/diacylglycerol ether lipid mixture and the ethanol painting method used in their studies result in highly resistive phospholipid membranes (>5 10 6 O-cm2) that are comparable to alkanethiol-supported phospholipid layers. Their results clearly show that the hydrated region proximal to the gold electrode responds to both the length of the hydrophilic tether and the spacer concentration. They note that the use of other lipids, such as 20:1 diacyl phosphatidylcholine, results in a membranes with resistance values of  200,000 O-cm2, suggesting that the unsaturated lipids form less insulating layers. Overall, this illustrates that the type of phospholipid, the underlying tethering architecture, and the method by which phospholipids are adsorbed to alkylthiolate layers in£uences the electrical properties of these HBM. It also suggests that more structurally complex tethered bilayers may be prone to defects. A variation of the reservoir-containing tethered membrane that contains micropatterned regions of b-mercaptoethanol surrounded by thiapoly(ethylene oxide)-cholesterol has been used to form micropatterned suspended bilayers membranes [79]. The capacitance drops by a factor 10 to a value of  0.9 mF/cm2 after vesicle addition, suggesting that phospholipid monolayers are adsorbed on the cholesterol-based anchors and suspended bilayers are formed over the b-mercaptoethanol regions. A multielement electrical model was applied to the micropatterned HBM, and results indicated that the membrane resistance was high enough to detect conductance changes induced by valinomycin and gramicidin. 2.2.2 Electrochemical Analysis The integral gold electrode also allows examination of HBMs with surfacesensitive electron transfer techniques. While optical techniques such as SPR, ellipsometry, impedance analysis, and infrared provide information about the surface average, electrochemistry provides information about microscopic coverage details present in the monolayer components of the HBM [57]. Electron transfer between the gold electrode and a solution redox species is dependent on the di¡usion of a redox species near the gold electrode (i.e., mass transfer), and the electron-transfer rate constant. The faradaic currents that arise from electron transfer between a redox species and the electrode is maximal at a bare clean electrode. When a gold electrode is covered with a highly insulating organic layer, such as a well-packed alkanethiol or alkanethiol/phospholipid bilayer, then the faradaic currents are typically suppressed [58]. Residual currents result from electron transfer through pinholes or defects in the insulating surface [57]. Pinholes are regions in the ¢lm at which the bare gold electrode is exposed, and electron transfer in these regions is controlled by mass transfer. Defects are considered regions where ¢lm coverage is such that redox species can transfer

Biomimetic Membranes on Metal Supports

119

electrons to the gold electrode, but at a diminished rate [57]. In alkanethiols SAMs, defect sites are thinner (e.g., collapsed alkane chains) than the wellpacked regions [62]. Electrochemical analysis methods such as cyclic vol tammetry and AC impedance spectroscopy can provide structural information about the molecular layers that compose HBMs. An excellent review of electrochemistry on organized monolayers on gold electrodes can be found in Ref. 57. Cyclic voltammetry monitors the faradic currents as the potential of the gold electrode is swept from one potential to another and back to the original potential at a linear rate [80]. Figure 8 shows a cyclic voltammogram of a bare gold electrode in the presence of 1 mM K3Fe(CN)6. The large negative and positive peaks in the voltammogram re£ect the large oxida tion and reduction currents at potentials near the formal potential of Fe(CN)63/4. The peak-to-peak separation is  59 mV, indicating a reversible reaction at the gold electrode [80]. A cyclic voltammogram of a gold electrode with a alkanethiol monolayer in the presence of 1 mM K3Fe(CN)6 [58] is also shown in Fig. 8. The absence of a peak near the formal potential suggests that the monolayer is free of pinholes. Larger potentials are required to produce currents, indicating that the redox species is unable to make close contact with the electrode surface. Fe(CN)63/4 may not be the

FIG. 8 Cyclic voltammetry of an HBM, an alkanethiol monolayer, and bare gold. Voltammograms were collected at 50 mV/s in the presence of K3Fe(CN)6 in 1 M KCl. (From Ref. 58.)

120

Elliott et al.

best probe of pinholes due to its relatively slow electron transfer kinetics at the gold electrode [81,82]. In a study evaluating pinholes in monolayers and HBMs with Fe(CN)63/4, Ru(NH3)63 þ /2 þ and hydroxymethylferrocene (FcCH2OH) redox couples, Brevnov and Finklea showed that pinholes are more readily observed with the kinetically fast Ru(NH3)63 þ /2 þ and FeCH2OH probes [81]. Figure 8 also shows a voltammogram obtained after the addition of a phospholipid layer to the alkanethiol monolayer [58]. When phospholipids are added to an alkanethiol monolayer, the observed faradaic currents are further suppressed. Studies with kinetically fast redox probes show that the phospholipid layer suppresses currents that arise from electron transfer at both defects and pinholes in longer-chain alkanethiol monolayers [81]. This indicates that the phospholipid layer is able to span the defects in the alkanethiol monolayer and act as an e¡ective barrier to both electron transfer and ion penetration [36,83]. Interestingly, in alkanethiol monolayers with alkane chain lengths of seven carbons or less, the phospholipid monolayers do not completely suppress faradaic currents in the presence of ferrocenecarboxylic acid [36].These data support the results of impedance and optical measurements that suggest that phospholipid monolayers on short-chain thiols are not fully ordered and have a greater number of defects [36]. AC impedance spectroscopy in the presence of a redox species is an additional technique that allows one to address the nature of the structural defects in the HBM ¢lm. AC impedance spectroscopy can be performed at low overpotentials, allowing one to observe electron transfer at pinholes [57].The application of low overpotentials may also minimize damage to the alkylthiolate layers [84]. In the simplest case, a Nyquist plot [Re(Z) versus Im(Z)] of the impedance response will contain a semicircle with a diameter that is the apparent charge transfer resistance, RCT, at the electrode [85]. If the pinholes are modeled as a microelectrode array, the fraction of pinhole coverage can be calculated by y¼1

0 RCT RCT

ð12Þ

where y is the fraction of pinhole coverage, RCT0 is the measured charge transfer resistance on bare gold and RCT is the charge transfer resistance obtained in the presence of the insulating layer [84]. Further analysis of the impedance response allows characterization of the size and distribution of pinholes [85]. For a pinhole-free covered electrode, AC impedance spectroscopy can be used to characterize the collapsed sites that are present in the ¢lm [62,86]. Several studies have used AC impedance to characterize pinholes and defects in alkylthiolate monolayers [84^88]. Excellent examples of the

Biomimetic Membranes on Metal Supports

121

experimental methods and observations can be found in these references. Only a few studies have used AC impedance to characterize the defects in HBM ¢lms. Plant et al. evaluated an HBM with AC impedance in the presence of the Fe(CN)63/4 redox species [58]. The resulting spectrum was semicircular, with no evidence of mass transfer-limited faradaic processes. This is characteristic of a pinhole-free membrane in which electron tunneling (presumably at collapsed sites) is the predominant faradaic process. Diao et al. utilized AC impedance to determine the mechanism by which electron tunneling occurs through an HBM on a pinhole-free alkanethiol layer [62]. Based on measured rate constants and the AC impedance spectra, they concluded that Fe(CN)63/4 di¡uses through pinholes in the phospholipid layer over the collapsed sites and transfers electrons through the collapsed alkanethiol layer. 2.3

Infrared Spectroscopy

Infrared (IR) spectroscopy has long been used to characterize the molecular structure, conformation, and orientation of planar lipid monolayers, bilayers and their composite proteins and carbohydrates [27,89,90]. IR spectral signatures describe the normal modes of vibrations caused by displacements in the permanent bond dipole moments of the chemical bonds in the sample molecules. These displacements are sensitive to di¡erences in the con¢guration and conformations of the chemical bonds, and as a result, infrared spectroscopy can be used to characterize the number, composition, structure, and conformation of the molecules that comprise an HBM. For example, in a supported bilayer system, the inner lea£et, the outer lea£et, and the resident protein can all be observed simultaneously, allowing complex interactions between these components to be characterized. IR spectroscopy can be used as a noninvasive method to study the formation and structure of an HBM. The use of this technique can facilitate the application of new and advanced analytical tools to HBMs by removing ambiguities and supplying corroborative evidence to assist data interpretation. IR spectroscopy operates on a short time scale ( 1012 s, to prevent the time averaging of molecular motions) and does not require invasive and possibly perturbing molecular tags. 2.3.1 The Molecular Structure of HBMs in Air The simplest sample con¢guration for studying HBMs is external re£ection in air [61,91], referred to as re£ection absorption infrared spectroscopy (RAIRS). In our studies, we have employed a high angle of incidence and re£ection from an HBM on a 150-nm-thick gold ¢lm to maximize the surface ¢eld strength and the area sampled. In this con¢guration, the re£ected

122

Elliott et al.

light that is polarized parallel to the substrate undergoes a nearly 180 phase change. With this phase change, the parallel polarized components of the incident and re£ected light destructively interfere with each other and the net result is that the observed spectrum of the HBM is almost entirely due to vibrations that occur perpendicular to the surface. In our laboratory, a variety of HBMs with di¡erent alkanethiols (hexanethiol, decanethiol, and octadecanethiol) and phospholipid with di¡erent chain lengths (DMPC, POPC, DOPC, DPPC, and DSPC) have been characterized. RAIRS data clearly show bands from the alkane chains and the choline head groups of the phospholipids in HBMs and con¢rm the presence of the phospholipid layer [61,91]. We have also characterized the e¡ect the addition of phospholipid has on the structure of the underlying alkanethiol monolayer using surface-enhanced Raman spectroscopy and RAIRS spectra of alkanethiols in the presence and absence of an added layer of deu terated phospholipid [61]. The results indicated that the addition of a phospholipid monolayer caused only very small structural changes that are consistent with increased order in the underlying alkanethiol monolayer. Importantly, these data also suggest that the structure-dependent electrical properties of the alkanethiol layer are not signi¢cantly a¡ected during the formation of the bilayer.The information obtained in this study validates the use of the equivalent circuit models for impedance analysis of simple HBMs [see Eqs. (6) and (7)]. RAIRS structural studies of HBMs constructed with hydrogenated phospholipids [91] reveal a series of CH2 wagging progression bands in the 1000^1500 cm1 region. Since wagging progressions are too weak to be observed in alkanethiols [92], these progression bands are probably from the lipid layer. If the progression bands are from the lipid layer, they suggest that the phospholipid portion of the HBMs has a high degree of intramolecular order [93]. To evaluate more clearly the separation of the wagging progressions, a spectrum of d62DPPC in an HBM was subtracted from a spectrum of DPPC in an HBM. After subtraction, only the features due to the hydrogenated alkane chains remained, supporting the idea that the wagging progressions are from the lipid layer. The separation between these wagging progression bands, Dn, can be related to the number of trans bonds [92] contributing to the wag-twist modes, m, by Dn ¼ 326=m þ 1

ð13Þ

The values derived indicated that for HBMs in air, phospholipids of several di¡erent chain lengths average one gauche bond per chain. Thus, HBMs in air appear to contain a well-ordered lipid layer similar to that in a gel-phase lipid bilayer.This would be expected for DPPC and DSPC,with gel-to-liquid

Biomimetic Membranes on Metal Supports

123

crystalline transition temperatures of 41 and 55 C, respectively. However, the gel-to-liquid crystalline transition temperature of DMPC is 23 C, and at room temperature a disordered DMPC layer might be expected. One explanation for the highly ordered structure of DMPC in an HBM is that the stability of the octadecane thiol lea£et stabilizes the DMPC lea£et, leading to a higher transition temperature. Another possibility is that a lower degree of lipid hydration that results from the samples being in air could raise the e¡ective transition temperature. The highly ordered nature of the phospholipids in the HBMs allows their acyl chains to be described as if they were rigid. This approximation allows the optical constants of the phospholipids to be obtained from IR measurements of bulk polycrystalline lipid samples, which would be expected to have similar numbers of gauche defects.These optical constants allow electromagnetic wave theory to be used to simulate the CH stretching region of the RAIRS spectra of DPPC layers. The angles for the tilt, twist, and azimuth of the phospholipid acyl chains were found to be 34, 38, and 50, respectively [91]. These models also reveal that the lipid layers can be described as uniaxial, with the azimuthal angle having little e¡ect on the simulated spectra. 2.3.2 HBMs Characterized by IR Spectroscopy in situ An advanced type of external-re£ection infrared spectroscopy, polarizationmodulation infrared re£ection adsorption spectroscopy, has been utilized to characterize HBMs under aqueous solutions [94]. This study examined HBMs consisting of DMPC and hexadecanethiol with and without gramicidin D. The resulting spectra show clearly the di¡erence between the hexadecanethiol monolayer, the hexadecanethiol/DMPC HBM, and the HBM containing gramacidin.Changes in the absorption intensities corresponding to the CH stretching region (2800^3000 cm1) in the alkanethiol monolayer and the HBM spectra are consistent with the formation of a complete bilayer. For the HBM, the frequency of the CH2 asymmetric stretching band, 2917 cm1, indicates that the lipid layer acyl chains are quite ordered, as disordered chains are expected at a frequency of around 2924 cm1. In the HBMs containing gramicidin D, characteristic amide A, amide I, and amide II protein bands are shifted to frequencies lower than those observed for gramicidin in water. These frequency shifts likely indicate that the environment or conformation of gramicidin in water is altered upon incorporation into the HBM. HBMs can also be studied under aqueous solution using total internal re£ection approaches. The major advantage to using internal re£ection is that the HBM can be monitored in situ under £ow conditions. IR spectra of

124

Elliott et al.

the HBM can be collected during the fabrication stage (e.g.,vesicle addition) and during the addition of membrane-active biomolecules. A second advantage is that both the parallel and the perpendicular components of the absorption spectrum are available. This information is valuable when it is used to describe structural details of the molecular constituents (i.e., lipids and proteins) that comprise an HBM. In total internal re£ection studies, the gold ¢lm that supports the HBM is deposited on an internal re£ection crystal. Two signi¢cant problems arise from using gold ¢lms on the re£ection surface. First, the metal ¢lm greatly attenuates the re£ected light. This results in the need to use ultrathin gold ¢lms (<20 nm) on the internal re£ection element.The second problem is that can be di⁄cult to coat a surface reproducibly with a continuous ultrathin metal ¢lm. The presence of a discontinuous gold layer can be problematic when determining the structure of an HBM. Since the thiols adsorb only to the gold-coated portion of the surface, it is unclear whether the HBM consists of a proximal alkanethiol and a distal phospholipid or if the proximal layer is a mixture of alkanethiol and phospholipid. Furthermore, optical properties of the re£ection surface are not uniform. The nonuniformity of the electromagnetic ¢eld at the surface makes quantitative and even qualitative interpretations of the absorption band intensities di⁄cult, compromising the information that can be obtained about the molecular properties of the HBM components. Sevin-Landais et al. used internal re£ection studies to evaluate the structure of the nicotinic acetylcholine receptor immobilized in a HBM [96]. The methodology used to prepare the ultrathin gold layer on the internal re£ection crystal [98] may result in a discontinuous gold ¢lm that covers only 80^85% of the crystal surface. As described above, the presence of an incomplete gold ¢lm may in£uence the structure information in the observed IR spectrum. Cheng et al. [97] used a 14-re£ection crystal coated with 20 nm of gold to examine thia-poly(ethylene oxide)-cholesterol-based HBMs.The multiple re£ections result in a total path length of at least 140 nm through the gold layer. The decay length for the intensity of the light through the gold is given by 4*p*k/l, where k is the extinction coe⁄cient for gold at wavelength l. In the C-H stretching spectral region (2800^3000 cm1), the decay length of gold is  13 nm. If the light had to travel through a path length of 140 nm, essentially no light would be left to detect. However, if the gold layer is discontinuous, the path length through the gold could be substantially less and the signal proportionally greater. This is similar to the phenomenon exploited in surface-enhanced infrared spectroscopy, an internal re£ection technique that takes advantage of the higher plasmon density at the surface of discontinuous gold layers (see Ref. 95 as an example).

Biomimetic Membranes on Metal Supports

125

Using a multiple internal re£ection element, Cheng et al. [97] found that the orientation of the phospholipid lipid acyl chains on the cholesterolbased HBM was 12 relative to normal. This value is di¡erent than the 26 orientation angle determined in nuclear magnetic resonance (NMR) studies of cholesterol/phospholipid mixtures that are representative of the lipids on the HBM surface [97]. Although there are several possible reasons for the di¡erences in orientation angles, this ambiguity illustrates the di⁄culty of quantifying the IR spectra in the presence of an inhomogeously coated surface. Recently, our group has minimized the problem of signal attenuation by the thin gold layer in the multiple internal re£ection geometry by using a single internal re£ection. A silicon crystal was used as the support for the gold ¢lm, and this improved the integrity of the layer. Figure 9 is an AFM image of a 10-nm gold layer on a silicon wafer with a  0.5-nm chromium adhesion layer. This layer has a root-mean-square roughness of <0.3 nm for a 1-mm2 area. The electric ¢eld strengths at the gold surface for di¡erent gold ¢lm thicknesses can be calculated, as shown in Fig. 10, by adapting the three-layer method of Hansen [99] as implemented by Axelsen [100] to accommodate a gold layer. Using the data in Figure 10 and this new

FIG. 9 (a) Atomic force microscopy shows the continuity of a 10-nm gold film on a 0.5-nm chromium support layer on a silicon wafer. (b) Histogram shows the distribution of the surface roughness of the film equivalent to  0.3-nm RMS roughness for a 1-mm2 area.

126

Elliott et al.

FIG. 10 The field strengths Ex2 , Ey2 , and Ez2 at the air/gold interface for a single reflection at a three-layered silicon/gold/air sample as a function of the gold thickness.

three-layer model, the orientation of lipid acyl chains during addition of the phospholipid layer to an alkanethiol layer can be determined from the parallel and perpendicularly polarized spectra shown in Fig. 11. Using the single internal re£ection technique, the average orientation of DMPC acyl chains in an HBM was calculated to be  25 relative to

FIG. 11 Single internal reflection spectra of the CD stretching region polarized parallel (s) and perpendicular (p) to the crystal surface for a deuterated d54DMPC layer under aqueous solution.

Biomimetic Membranes on Metal Supports

127

normal. This value compares nicely with the NMR value reported by Cheng et al. [97] and with the 26 tilt for DMPC in the gel phase on an ATR crystal reported by Hubner and Mantsch [101]. The current challenge for the infrared characterization of HBMs is in developing the methods and models required to derive the quantitative structural information that is available in the infrared spectra. While some methods have been developed for the characterization of lipid/protein layers at the surfaces of internal re£ection crystals [89,100,102], the presence of the gold ¢lm adds an additional level of complexity. However, we have shown that the use of single-bounce techniques and homogenous thin gold ¢lms can allow for successful quantitative measurements. 2.4

In Situ Ellipsometry and Quartz Crystal Microbalance

Other techniques that allow the observation of kinetic changes in HBMs include in-situ ellipsometry and the quartz crystal microbalance. 2.4.1 In Situ Ellipsometry When linearly polarized light is re£ected from a surface, its polarization is altered by virtue of its interaction with the surface. In ellipsometry, two parameters, phase (D) and relative amplitude (c), are measured to de¢ne the polarization change in the re£ected polarized light. For a three-phase sample
i.e., a supporting base layer, a ¢lm layer of unknown properties, and an ambient layer (of air or water, for example)
the ellipsometric values for the ¢lm layer can be calculated by measuring the ellipsometric values of the surface before and after ¢lm deposition [103]. The ratio of the complex re£ection coe⁄cients for the parallel and perpendicular polarized components of the re£ected light can be calculated from these ellipsometric values and used with the Fresnel equations to solve for the optical properties of the ¢lm (index of refraction n, extinction coe⁄cient k, and thickness d), provided the optical constants of the support (n and k), the index of refraction of the ambient layer, and the incident angle of the probe beam are known [103]. The solution is found numerically by adjusting the optical properties of the ¢lm with computer software until a best-¢t answer is obtained. In the case of ultrathin ¢lms, the index of refraction (n) and the extinction coe⁄cient (k) of the organic ¢lm are required to determine its thickness (d). For the organic monolayers used for HBM construction, n  1:45 and k  0 are generally acceptable optical constants [55,91]. Alternatively, if the thickness of the organic material is known, then the optical properties of the organic ¢lm can be determined. Ellipsometric values can be collected at a single wavelength or at multiple wavelengths. The latter technique, called spectroscopic ellipsometry, increases the number of known quantities about the ¢lm. If n and k

128

Elliott et al.

are assumed to be constant for a particular wavelength range, then d for the ¢lm can be obtained. If the ¢lm thickness is considered constant for each measurement, then n and k of the ¢lm in a di¡erent wavelength range can be determined [104]. Spectroscopic ellipsometry provides the opportunity to characterize absorbing species (i.e., proteins or photophores) that can be used in HBM construction. Ex situ ellipsometry is used routinely to measure the thickness of ultrathin dielectric ¢lms such as alkanethiol monolayers on gold, and these results compare favorably with those obtained with impedance studies [55,57]. In-situ ellipsometry provides the opportunity to monitor the growth of a dielectric ¢lm under aqueous conditions [105]. Unlike SPR, it can be performed on thick gold ¢lms supported on opaque substrates. In situ ellipsometry has been used to examine the formation of DOPG layers on decanethiol monolayers [106]. In these studies, ellipsometric measurement from a single-wavelength ellipsometer (632.8 nm) suggested that between 2.1and 2.5 nm of lipid were deposited on the alkanethiol ¢lm.This is in agreement with the values measured with other techniques. Our laboratory has been examining HBM formation with in situ spectroscopic ellipsometry. Our results are in agreement with other measurements that indicate the equivalent of  2.0 nm of phospholipid absorbs onto alkanethiol monolayer. 2.4.2 Quartz Crystal Microbalance A quartz crystal microbalance (QCM) is a piezoelectric acoustic wave device that can be used to monitor mass changes at a surface. If the QCM crystal is coated with a gold ¢lm and modi¢ed with thiols, the technique provides an alternative for monitoring the formation of HBMs. When the QCM is operating in shear mode, the resonance frequency and dissipation of the shear wave are dependent on the total mass and the viscoelastic properties of the mass coupled to the surface [107]. For HBM assembly,the ability to measure each of these properties provides the ability to distinguish between lipid adsorption in the form of a monolayer or in the form of vesicles. In general, the resonance frequency is proportional to the mass that becomes mechanically coupled to the surface. Depending on the morphology of the lipid adsorbed to the surface (i.e., monolayer or vesicles), the coupled mass will include lipid, and may include water trapped inside and between vesicles [108]. The dissipation energy, re£ected in the resonance amplitude decay, is dependent on the viscoelastic properties of the coupled mass. Keller and Kasemo used QCM to monitor the fusion of small unilameliar vesicles to SiO2, oxidized Au, and alkanethiol surfaces [109]. Their results show that they can easily distinguish between absorbed intact vesicles on the oxidized Au surface and the phospholipid monolayer on the

Biomimetic Membranes on Metal Supports

129

alkanethiol SAM by monitoring the resonance frequency and dissipation of the shear wave. Interestingly, the results show a two-step process for vesicle fusion on the SiO2 surface: vesicles adsorb to the surface, and when they reach a critical concentration, they fuse to form a bilayer [108]. Two-step behavior was not observed during vesicle fusion onto alkanethiol SAMs, indicating that preadsorption of intact vesicles does not occur at a hydrophobic surface. In addition to monitoring phospholipid adsorption, QCM can be used to study the preparation of the supporting alkanethiol monolayers and in biosensor applications [110]. Burgess et al. utilized QCM measurements to prepare an octadecanethiol submonolayer for HBMs that could accommodate transmembrane proteins [111,112]. Steinem et al. have used QCM to monitor the binding of protein toxins, lectins, and antibodies to ligands integrated into an HBM surface [113,114]. 2.5

Atomic Force Microscopy

Unlike the techniques discussed above, which provide information over a scale of micrometers to millimeters, atomic force microscopy (AFM) allows a surface to be imaged at the nanometer scale. AFM is a local probe that provides detailed information about a small area of the sample. By combining the information from the AFM with that from the nonlocal probes discussed in this chapter, it is possible to gain a far more complete understanding of the structure of the supported membrane systems. AFM is a critical technique for characterizing HBMs and for helping to verify the interpretation of data from other measurements. The atomic force microscope belongs to the class of scanning probe microscopes that originated with the Nobel Prize-winning invention of the scanning tunneling microscope in 1982 [115]. The introduction of the AFM in 1986 [116] eliminated the need for an electrically conductive sample and opened the ¢eld to biologically relevant samples. Continued improvements since that time, including intermittent contact imaging and scanning probe microscopes designed speci¢cally for imaging in liquid environments, have allowed for the routine use of AFM with biological samples [117]. The AFM functions by scanning a microfabricated probe over the sample surface (see Fig. 12). The scanning is performed with a piezoelectric tube scanner that moves in the x, y, and z directions. The probe consists of a sharp tip at the end of a thin cantilever that is typically several hundred micrometers in length and has a spring constant in the range 0.01^1 nN/m. The de£ection of the cantilever is measured using an optical lever in which a laser beam is focused on the end of the cantilever and re£ected into a split photodiode.When the tip contacts the sample, the cantilever de£ects and the

130

Elliott et al.

FIG. 12 Schematic of an atomic force microscope. Refer to text for a full description.

beam moves across the split photodiode.The voltage di¡erence from the two halves of the photodiode is then used as a feedback signal to move the tip (or sample) up and down to maintain a constant cantilever de£ection. As the surface is scanned, the distance the sample must be raised or lowered is recorded and used to make an image. The resulting image is of a surface that applies a constant force to the tip and is usually interpreted as a topographic image of the sample, although some caveats to this interpretation will be discussed. The lateral resolution of the AFM is determined largely by the shape of the tip.Tips with a nominal radius of curvature of less than 10 nm are commercially available; however, radius of curvature is not always the relevant measure. For atomically smooth samples, it is only a few atoms at the end of the tip that interact appreciably with the sample, while for a steeply sloped sample, the aspect ratio of the tip may be more important. The majority of AFM studies on supported lipid layers have been performed on glass, silicon, or mica substrates.While a general discussion of this work is outside the purview of this chapter, many of the caveats of applying AFM to supported membrane systems are the same regardless of the underlying support. From the earliest attempts to image a supported membrane it was recognized that even forces less than a nanonewton exerted by the tip on the sample were likely to deform and disrupt the membrane [118]. While catastrophic damage is easy to detect by repeated scanning of the same area, noncatastrophic compression of the sample can give misleading height measurements while leaving the sample intact. Additionally, if one or more components of the supported membrane is charged, there can be electrostatic forces between the AFM tip and the sample. Again, this leads to a

Biomimetic Membranes on Metal Supports

131

surface of constant force that is di¡erent from the topography of the sample as the tip is repelled by long-range electrostatic forces instead of short-range van der Waals forces. Both the experimental and theoretical aspects of these issues are well treated in a pair of recent articles [119,120]. The advantage of using mica, silicon, or glass as a substrate is that they are extremely £at over large areas. By using a £at substrate, any features observed by the AFM can be attributed to the sample.While metal ¢lms can be annealed to give atomically £at planes, these are relatively small laterally and typically have large valleys between £at areas. One imaginative way around this problem has been to use the AFM in nontopographic modes. Both phase-contrast and frictional force modes are sensitive to sample compliance, and to tip and sample surface chemistry, while they are relatively insensitive to topography. Burgess et al. have used phase contrast to locate cytochrome c oxidase in a supported membrane [121]. By comparing the phase images from samples with and without the oxidase, the phase signal characteristic of the oxidase was established. The corresponding locations in the topographic image could then be identi¢ed as the oxidase despite its similarity to the roughness of the substrate. Jenkins et al. used frictional force measurements on a substrate patterned with b-mercaptoethanol and a thiol-linked cholesterol [79]. Di¡erences in the friction between the two regions, and changes in this friction when lipid was added,were used to infer the formation of a suspended bilayer over the b-mercaptoethanol. We have taken another approach originally inspired by our desire to perform neutron scattering experiments on supported bilayers. As discussed below, the root-mean-square (RMS) roughness of each interface is a critical parameter in extracting information from a neutron scattering curve. To make smooth substrates,we deposit gold ¢lms on the order of 10 nm thick on silicon wafers with a 1.5-nm chromium adhesion layer. We have prepared such ¢lms using two di¡erent techniques [122,123], both with an RMS roughness <0.3 nm over a 1-mm2 area (see Fig. 9). While somewhat rougher than the nonmetallic substrates, they are smoother than thermally evaporated or annealed gold samples. Using these substrates, we can easily distinguish holes in monolayers or additional overlayers from the substrate roughness when samples are imaged in topographic mode. The smooth gold substrates allow the AFM to assess the quality of each layer of the HBM at the nanoscale.This has proved to be extremely valuable when combined with other techniques. We o¡er below several examples from our own work in which the combination of the AFM and a nonlocal probe can give information that neither can provide alone. In our early experiments, we frequently needed to reuse gold substrates, due to the expense of the smooth gold ¢lms. The samples were cleaned in a UV/ozone cleaner, rinsed, and placed in an alkanethiol solution

132

Elliott et al.

FIG. 13 AFM images of hexadecanethiol self-assembled monolayers on UV/ ozone cleaned gold. (a) The gold was soaked in ethanol after cleaning and before incubation in the hexadecanethiol solution, giving a uniform monolayer. (b) The gold was placed in hexadecanethiol solution immediately after UV/ozone cleaning, leading to the formation of islands on top of the monolayer. The z scale is 8 nm black to white in both images.

overnight to form a new monolayer. A contact angle with water of 110 was taken to indicate that a good monolayer had formed.When a new researcher changed the rinsing protocol, there was no observable di¡erence in the contact angle. However, the AFM showed the presence of small islands with uniform height on the surface, as seen in Fig. 13b [124]. Note that the AFM image of the properly prepared monolayer in Fig. 13a, as well as the area between the islands in Fig.13b, is indistinguishable from that of the bare gold substrate. Thus, while the AFM was required to see the islands, the contactangle measurement was required to know that the alkanethiol monolayer had formed on the surface. In a similar vein, we can use the AFM to image the completed HBM in situ. However, we are again faced with the problem that a well-formed HBM is expected to have a featureless surface that di¡ers little from that of the supporting layer. How can we be certain that the lipid monolayer has formed on top of the alkanethiol monolayer? One answer is to use the tip to scrape a hole in the lipid.We have done this successfully. However, we frequently ¢nd that the subsequent images are of poor quality, due to either damage or contamination of the tip, preventing further measurements. Since the sample is already in a £uid cell and has a gold substrate, we can add a platinum counterelectrode to the cell and use impedance analysis. As discussed

Biomimetic Membranes on Metal Supports

133

FIG. 14 (a) AFM image of an octadecanethiol monolayer exposed to a DMPC vesicle solution for several hours. The uniform monolayer of DMPC leaves the topography of the surface essentially unchanged from that of the monolayer. (b) A complex capacitance plot shows that after exposure to the DMPC vesicles the capacitance of the surface has decreased, indicating that a layer of DMPC has adsorbed on the alkanethiol monolayer.

above, as a layer of phospholipid adsorbs to the alkanethiol surface, it increases the thickness of the insulating layer between the gold and the buffer solution. This causes a reduction in the capacitance of the layer that can be monitored on-line. Figure 14 shows an in situ image of an octadecanethiol monolayer after incubation under DMPC vesicles. The image is similar to that of a bare monolayer. However, by applying the two techniques simultaneously, we can con¢rm that this topographically £at surface is indeed an intact HBM: the impedance plots show that the capacitance of the layer has changed from 0.97 to 0.70 mF, indicating the presence of an adsorbed layer of phospholipid. 3

NEW TECHNIQUES APPLIED TO HBMS

The development of measurement tools for probing the structure and function of engineered membrane mimics on metal supports and the cell membrane components incorporated into them is essential for optimizing their biomimetic character and for using them to learn more about biological membranes. The stability of HBMs and their ease of formation make them ideal model systems for study with emerging measurement techniques.

134

3.1

Elliott et al.

Neutron Reflectivity

Neutron re£ectivity has been thoroughly described in the context of the characterization of condensed-matter thin-¢lm systems at interfaces [125^ 128]. The specular re£ection of neutrons, R(Q), in which the angle of re£ection is equal to the angle of incidence, is de¢ned as the intensity of the specularly re£ected beam divided by the intensity of the incident beam. Here, Q ¼ ð4p=lÞ sin y, where l is the incident neutron wavelength and y is the angle of incidence relative to the plane of the ¢lm. The measured re£ectivity, R(Q), is given by jrðQÞj2, where r(Q) is the complex re£ection amplitude. Thus, the phase information is lost in a typical measurement of R(Q). The neutron scattering length density (SLD) pro¢le of a ¢lm, often represented by r(z), is a function of ¢lm depth, z. For a given layer de¢ned over a range of z values, the SLD is de¢ned as the sum of the coherent neutron scattering lengths of all of the atoms in that layer, divided by the volume of the layer. Thus, the SLD pro¢le is a continuous function of z that relates directly to the chemical composition of the ¢lm. For this reason, it is sometimes termed a compositional depth pro¢le. While r(Q), and thus R(Q), can be calculated exactly from the SLD pro¢le of a ¢lm, the SLD pro¢le cannot be obtained directly from the measured R(Q). Thus, data-¢tting methods are usually used to obtain SLD pro¢les that are consistent with the re£ectivity data. Both standard model-dependent ¢tting methods using stepped functions based on the theoretical lipid composition, and model-independent ¢tting methods using randomly generated smooth functions represented, for example, by parametric B-splines [129], are used for this purpose. An advantage of using neutron re£ectivity for the study of biological systems is that neutrons are sensitive to the light elements, such as carbon, hydrogen, nitrogen, and oxygen, which are so important in biological materials. Furthermore, neutrons are sensitive to isotopic di¡erences in these elements. Thus, a simple substitution of deuterium for hydrogen can substantially alter the SLD pro¢les of biological ¢lms while having a minimal e¡ect on their biochemistry. By making several re£ectivity measurements on the same system, but with di¡erent components deuterated, a more accurate model structure can be determined, since it must be consistent with all of the re£ectivity data sets simultaneously. This ‘‘contrast variation’’ method is used widely in the characterization of biomimetic ¢lms, and its importance would be di⁄cult to overstate. Finally, because neutrons interact weakly with materials, a solid support can be used as the incident medium in neutron re£ectivity experiments.This allows a supported bilayer to be measured in situ, in full contact with water. Over the past decade, the neutron re£ectivity technique has been used increasingly to assist in the structural characterization of biomimetic

Biomimetic Membranes on Metal Supports

135

membranes that are in contact with water. Achieving the sensitivity that allowed single bilayer membranes to be studied has been on ongoing challenge. It was demonstrated that decreasing the volume of aqueous solution in contact a biological membrane greatly decreased the background from incoherent scattering from hydrogen and deuterium, and allowed measurements to be made out to Q ¼ 0:25 —1 [130]. This was more than twice the Q value that was achieved in earlier experiments on supported single bilayers [131]. This new experimental setup was subsequently used to study DPPC phospholipid bilayers adsorbed onto planar silicon substrates from vesicles in solution [31]. Neutron re£ectivity was used to study the structural details of supported hybrid bilayer membranes (HBMs) in D2O solution [91]. The structural parameters of the DPPC monolayer in contact with water were found to agree with those from partially hydrated multibilayer systems. However, the thickness and hydration of the lipid head-group region was found to be in better agreement with the results of the previous supported single-bilayer experiment [31]. The interpretation of the resultant SLD pro¢les was simpli¢ed, since the HBM is formed on the gold-coated silicon substrate with complete surface coverage and since complementary re£ection-absorption infrared spectroscopy measurements allowed an independent determination of the tilt angle of the alkane chains. In more recent HBM experiments [132], a monolayer of chain-deuterated DMPC (d54DMPC) was deposited onto monolayers of two di¡erent types of alkanethiol. The ¢rst HBM studied was formed from octadecanethiol and d54DMPC monolayers and was studied in the presence of melittin, a peptide toxin. The second HBM used a novel thia-hexa(ethylene oxide)-alkane (THEO-C18) molecule [70] in which a hydrophilic hexa (ethylene oxide) moiety is incorporated between the sulfur moiety at the gold surface and the alkane chain. Since neutron re£ectivity is extremely sensitive to the presence of D2O in the THEO-alkane monolayer, one of the purposes of the experiments was to ascertain the degree of hydration of the ethylene oxide region. In order to obtain angstrom-level information about the structure of these HBMs, improvements in instrumentation and the sample environment were made to allow re£ectivity data to be obtained down to 108 in re£ected intensity and out to Q values as high as 0.7 —1. Under these conditions, the resolution of the neutron re£ectivity experiment is now limited only by the roughness of the supporting substrate.Thus, like in the studies of HBMs with AFM, smooth gold-coated substrates were essential for successful highresolution neutron re£ectivity experiments. The structures of both HBMs were quanti¢ed by comparing the resultant SLD pro¢les to pro¢les generated by molecular dynamics simulations of octadecanethiol/DPPC HBMs.

136

Elliott et al.

It was found that the octadecanethiol and THEO-C18 lea£ets of the HBMs exhibit roughness similar to that of the gold surface. However,this roughness does not propagate throughout the d54DMPC phospholipid lea£et in either HBM. Since the phospholipid is in the £uid state, the d54DMPC lea£et is much more disordered than either the octadecanethiol or THEO-C18 leaflets. Therefore, the d54DMPC essentially anneals, smoothing out the roughness present in the distal lea£et of the bilayers. The measured neutron re£ectivities from the THEO-C18/d54DMPC HBMs in the absence and presence of melittin are shown in Fig.15. Extensive model ¢tting [129] allowed the SLD pro¢les for the HBMs to be obtained with unprecedented accuracy and detail. Since care was taken to obtain a ‘‘family’’of ¢tted SLD pro¢les in each case, the variations in the sensitivity of the neutron re£ectivity technique from region to region could easily be tracked. The SLD pro¢les obtained for the HBM in both the absence and presence of melittin are shown in Fig. 16. The results show unequivocally that the ethylene oxide region of the THEO-C18 layer is not hydrated with bulk D2O from the solution. Furthermore, it is evident that melittin does not alter the HBM in a way that allows bulk water to penetrate into this region. The location of melittin appears to be primarily in the DMPC head groups.

FIG. 15 Reflectivity from thia-hexa(ethylene oxide)-C18/DMPC HBMs. Measured neutron reflectivity versus Q from thia-hexa(ethylene oxide)-alkane/C18 HBMs in the presence ( ) and absence ( ) of the membrane-active peptide, melittin, in contact with D2O solution. Chain-deuterated DMPC molecules (d54DMPC) were used to increase the contrast between the alkanethiol and phospholipid layers. (From Ref. 132.)

Biomimetic Membranes on Metal Supports

137

FIG. 16 Boundaries of the families of PBS-fitted scattering-length density profiles obtained from the neutron reflectivity data for thia-hexa(ethylene oxide)-C18/ d54DMPC HBMs in D2O solution with (light shaded area) and without (darker shaded area) melittin. The chromium and gold regions are not shown in the scattering length density profiles for clarity. (From Ref. 132.)

The large change in SLD suggests that there is signi¢cantly less water in this region when melittin is present. However, melittin does have a signi¢cant e¡ect on the alkanethiol layer, since it is found to be approximately 3 — thicker in the presence of melittin. This can occur if the thickness of the alkane chains has increased, presumably through a decrease in tilt angle. In spite of the fact that SLD pro¢les have been determined with unprecedented con¢dence and detail, the simultaneous presence of water and melittin in the DMPC acyl chain region could not be ruled out in this work. Thus, contrastvariation experiments are still needed to determine the amount of melittin and water in all of the layers. Recently, a new phase-sensitive neutron re£ectivity method that allows the determination of the compositional depth pro¢les of biomimetic membranes from ¢rst principles, without the need for ¢tting, has been reported [133]. In this technique, two measurements are made on identical samples deposited on two di¡erent gold-coated substrates, such as silicon and sapphire. The neutrons are incident through the highly transparent sin gle-crystal silicon or sapphire substrate, which becomes the ‘‘fronting’’ medium in these experiments. By so varying the composition, and thus the SLD, of the fronting medium, it is possible to obtain the complex re£ection amplitude, rðQÞ, which includes the phase information, for the HBM.

138

Elliott et al.

Determination of the complex re£ection amplitude using this surround variation method [133] allows the related SLD pro¢les to be obtained by ¢rstprinciples inversion without the need for ¢tting or adjustable parameters. The SLD pro¢le obtained in this way is ensured to be unique, and it compared favorably with that calculated from the chemical compositional pro¢le predicted by a molecular dynamics simulation of an HBM of similar structure [134], as shown in Fig. 17. Again, it should be emphasized that this new experimental method, which yields exact SLD pro¢les, must still be combined with the contrastvariation technique to obtain information on the location of protein and water in all of the regions of the HBM. It should also be pointed out that a direct inversion method such as this, which allows for a unique determination of a unique SLD pro¢le, is absolutely essential in order to model the structure of more complicated biomimetic systems, especially those containing integral membrane proteins.

FIG. 17 Scattering-length density profile (gray line) resulting from a direct inversion of real r(Q) for a thia-hexa(ethylene oxide)-C18/d54DMPC HBM in 38% D2O solution, compared with that predicted by a molecular dynamics simulation (black line) of an octadecanethiol/DPPC HBM. The ethylene oxide region was modeled as part of the substrate. Data truncation effects are responsible for the oscillation in the inversion profile in the gold regions. (From Ref. 133.)

Biomimetic Membranes on Metal Supports

3.2

139

Nonlinear Optical Spectroscopy

Vibrational sum frequency generation (SFG), a nonlinear optical vibrational spectroscopic technique, is among several methods we are developing as new in situ probes of HBMs. SFG o¡ers several advantages due to its high surface selectivity, submonolayer sensitivity, and excellent spatial, spectral, and temporal resolution [135,136]. SFG is capable of in situ measurements at biological interfaces and is potentially able to give direct information about the structure, orientation, aggregation, and organization of surfaceassociated biomolecules. Historically used by physicists and chemists for studies of small molecules at interfaces [137], vibrational SFG and second harmonic generation (SHG), an electronic spectroscopy version of SFG, have only recently been applied to surface studies of molecules of biological interest. Some aspects of interest to the biological spectroscopists have been reviewed [138].We have applied SHG-circular dichroism to probe the interaction of cytochrome c with HBMs [50]. In SFG, two laser beams are simultaneously re£ected from an interface, as depicted in Fig.18. Due to the nonlinear properties of the surface, the two photons, at frequencies o1 and o2, can interact to create a third photon with a frequency that is at the sum of the two photon frequencies, o3 ¼ o1 þ o2.This occurs through a nonlinear polarization P(2)of the surface caused by the electric ¢elds (E) of the incident light and the second-order nonlinear susceptibility x(2)of the surface: Pð2Þ ðo3 Þ ¼ xð2Þ : Eðo2 Þ : Eðo1 Þ

ð14Þ

For example, broad-band infrared photons (2700^3100 cm1) can be summed with 800-nm photons to produce SFG photons at 650 nm [139]. Due to the vibrational resonance of the infrared photons, the 650-nm photons

FIG. 18 Schematic showing the process of sum frequency generation using incident light of frequencies o1 (IR) and o2 (visible) and producing SFG light at a frequency o3 (sum).

140

Elliott et al.

convey the vibrational response, and thereby, information regarding molecular structure within the surface. The symmetry properties of x(2) are the origin of the exceptional surface speci¢city of SFG [137].Within the electric dipole approximation, x(2) is zero for centrosymmetric media,which include achiral solutions such as air, water, etc. An interface between two media, such as gas/solid or liquid/solid, naturally breaks centrosymmetry, which allows x(2) to be nonzero, and SFG may occur. Therefore, SFG signals are only generated by molecules at the interface, and there is zero background from molecules in solution. Taking the symmetry requirements to the level of individual molecules, SFG activity also requires noncentrosymmetric vibrations, or noncentrosymmetric arrangements of vibrations,which leads to the requirement of an simultaneous nonzero IR dipole and a nonzero Raman transition dipole for SFG activity [137].The resultant SFG spectra can therefore be less complex than IR or Raman spectra and, in principle, yield more information about molecular orientation. Furthermore, since the relationship between SFG activity and symmetry is closely tied to molecular structure, both the presence and absence of vibrational bands in SFG spectra can be helpful in interpreting complex surface IR and Raman spectra. Due to the symmetry properties of x(2), SFG is sensitive to the orientation (up/down, tilt angle) of interfacial molecules [137]. Due to the asymmetric nature of an interface, amphiphilic molecules often adsorb with a net orientation in the direction normal to the interface. As x(2) is a third-rank tensor, the sign of x(2) inverts with a 180 rotation of molecular orientation, and thus the sign of a SFG band changes.Consequently, the net orientation of an adsorbate can often be inferred from the observed sign of the SFG bands. Additionally, a layer of adsorbates with random orientations (in the direction normal to the surface) will mutually cancel and yield no net SFG bands. For surface coverages up to a monolayer, x(2) can be modeled [137] as xð2Þ ¼ Ns hbi

ð15Þ

where Ns is the number density of molecules on the surface and b is the second-order optical polarizability of an individual molecule. The brackets indicate an average over all orientations of the molecules.The intensity of the SFG bands is proportional to the square of x(2). Therefore, SFG band intensities can be used to follow the adsorption of molecules at a surface, e.g., the adsorption of phospholipids during the formation of an HBM. 3.2.1 The HBM as a Model System for SFG Technique Development As an experimental system for the development of new techniques, HBMs have the obvious advantages of robustness and ease of preparation. In

Biomimetic Membranes on Metal Supports

141

addition, the underlying gold support of the HBM gives large SFG signals that are not vibrationally resonant and are convenient for heterodyning with a vibrationally resonant SFG signals [135]. The HBM’s alkanethiol monolayer is particularly amenable to SFG studies, and it has already been extensively characterized in linear vibrational [140] and SFG studies. Furthermore, the alkanethiol monolayer can be deuterated,which allows it to be probed independently of other molecules as needed. 3.2.2 In Situ SFG During HBM Formation We have used the formation of HBMs by vesicle fusion to demonstrate the ability of SFG to gain in situ spectroscopic information [141].The kinetics of vesicle fusion to an alkanethiol monolayer during HBM formation have been studied by surface plasmon resonance [43], but understanding the molecular level details of the process has been an ongoing challenge. Using a broadband SFG approach [139], it is possible to follow, spectroscopically, the evolution of the phospholipid monolayer in real time. Figure 19 shows SFG spectra before and throughout the interaction of deuterated phospholipid vesicles at the D2O bu¡er/deuterated alkanethiol monolayer interface. This isotope-labeled version of phospholipid (d13DMPC) has a fully deuterated head group and has hydrogenated acyl chains. The strong features in the spectra in Fig. 19 are assigned to vibrational modes of the acyl-chain methyl groups. The spectra show only weak methylene resonances (at 2850 cm1 and broad features in the region of 2890^2910 cm1). The presence of strong methyl modes and weak methylene modes indicates that the phospholipid alkyl chains are highly symmetric, and thus highly ordered. A symmetric bilayer would be expected show no vibrational bands in SFG spectra due to destructive interference between the two layers. An ordered monolayer of phospholipid on an alkanethiol monolayer would be expected to have its methyl groups pointing toward the gold surface, and would show an SFG spectrum with bands pointing up.We observed positive bands in our SFG spectra of adsorbed d13DMPC, indicating that the terminal methyl groups have a net orientation toward the alkanethiol monolayer. This is consistent with the expected orientation of the phospholipid acyl chains due to van der Waals’ interactions with the thiol alkane chains. The observation of only weak methylene resonances indicates that the acyl chains of DMPC are highly ordered, and are predominantly in a trans con¢guration, as CH 2 modes of trans acyl chains are symmetry-forbidden in SFG. The time frame of HBM formation observed by our SFG studies, roughly 45 min, is within the time frame of HBM formation observed by surface plasmon resonance. SPR has demonstrated the phospholipid monolayers in HBMs are stable under continuous £ow, but that unfused

142

Elliott et al.

FIG. 19 SFG spectra before and throughout d13DMPC vesicle fusion at the D2O buffer/d37octadecanethiol monolayer interface. The spectra shown are at 10-min intervals over 90 min after the addition of vesicles to the sample reservoir. Each spectrum was ratioed to the SFG spectrum of the buffer/d37octadecanethiol interface to remove the nonresonant background of the underlying gold film. Each spectrum was collected over 60 s at a resolution of 4^5 cm1. The input beams were plane-polarized.

vesicles and/or multilayers are washed away after a bu¡er rinse.We observe an increase in the SFG band intensities after rinsing with bu¡er. This likely indicates that a signi¢cant amount of adsorbed phospholipid was not perturbed by the bu¡er washes, but perhaps unfused vesicles or multilayers (which could destructively interfere with the SFG signal) were removed.The observation that our SFG spectral features nearly disappear after an air and ethanol rinse indicates signi¢cant disruption and removal of the adsorbed phospholipid, in agreement with SPR results. 4

APPLICATIONS OF HBMS

The variety of analytical methods that can be used to address the physical properties of HBMs makes them ideal as biomimetic platforms for studying membrane biology and biosensor development. The ability to control the fabrication details of the individual phospholipid and alkylthiolate lea£ets allows construction of an HBM tailored to optimize a particular analytical

Biomimetic Membranes on Metal Supports

143

signal that corresponds to the molecular interaction being studied. This section focuses on studies that have utilized HBMs as active biomimetic membranes.We have attempted to include unique issues addressed in fabrication and experimental design to provide a general view of the role that HBMs can play in biological studies. The section is divided into two parts. The ¢rst part deals with sensing and screening applications, in which HBMs are used as sensor platforms to report the function of a biomolecule (proteins, peptides, or small molecules). These studies illustrate the possibilities of using HBMs to construct biosensors for clinical, industrial, and pharmaceutical applications. The second part of this section describes studies in which new information about biomolecules are obtained with the use of HBMs as biomimetic surfaces. In these applications, HBMs o¡er the ability to study structure and/or function of biomolecules with methods that are not readily available with other model membrane systems. 4.1

HBMs in Screening and Sensing Applications

One of the most common uses for HBMs is to monitor the interaction between protein receptors and biological ligands at a surface. The HBM o¡ers several features that are useful in these studies. 1. The HBM is a highly reproducible, rugged, well-de¢ned planar surface. The supported planar con¢guration provides analytical accessibility not available for some other model systems (i.e., vesicles). Assumptions about the chemistry and physical structure of the experimental test surface can be con¢rmed through analytical studies. 2. The lipid layer can be deposited onto the alkanethiol layer by vesicle fusion. Fusion of lipid vesicles containing various components (lipids, ligands, or proteins) provides a convenient method to transfer the components to the test surface. 3. The phospholipid layer resists nonspeci¢c protein adsorption, and the alkanethiol/phospholipid bilayer covers the entire analytical surface [37,51]. This combination signi¢cantly reduces the level of nonspeci¢c protein absorption during binding assays. 4. Alkanethiol monolayer chips are commercially available for surface plasmon resonance instrumentation that is routinely used in many laboratories. Among the ¢rst uses of HBMs was to examined the binding of cholera toxin to GM1 ganglioside ligands immobilized in the phospholipid matrix [37,142]. In these experiments, the HBM was formed by fusing vesicles containing the ganglioside ligands to an alkanethiol monolayer, and SPR was

144

Elliott et al.

used to measure mass changes at the HBM surface. The surface concentration of ligand could be controlled by varying the ligand concentration in the vesicles [37]. In addition to SPR, QCM has been used to monitor the binding of cholera, tetanus, and pertussis toxins, and peanut lectins to HBMs containing various gangliosides [113,143]. HBMs should be useful for screening binding interactions with the variety of lipid ligands that exist in native cellular membranes, such as phosphatidylinositols, gangliosides, phosphatidylserine, phosphatidylcholine, etc. Plant et al. used SPR to monitor the binding of streptavidin to biotinDPPE ligands immobilized in an HBM surface [51]. The ability to bind streptavidin to the HBM expands its use for a number of applications. The streptavidin tetramer is thought to bind with two sides directed toward the surface and two sides toward the bathing solution [51]. Biotinylated ligands or receptors can then be immobilized on the sensor surface and used in other mass-sensitive screening assays. Experiments with lipopeptides [144] suggest that a variety of peptides and other biomolecules can be immobilized in HBMs through covalent attachment to lipid moieties. For example, lipid-linked antibodies and proteins have been immobilized in HBMs [145,146]. Double-stranded DNA fragments have been immobilized into HBM surfaces via cholesterol-based membrane anchors and the resulting membranes were used to study DNA/ protein-binding interactions [147]. Glycosylphosphatidylinositol (GPI)linked proteins, aminopeptidase N [144], and promastigote surface protease [148] have also been functionally anchored into HBMs. It is important that GPI-linked proteins can be immobilized in HBMs, since almost any soluble protein can be made to exhibit a GPI link with the use of molecular biology methods [149]. A number of potential screening applications with lipidderivitized molecules can be envisioned. A study that examined the binding kinetics of the therapeutic coagulant factor VIII from di¡erent sources to HBMs containing negatively charged lipids [53] found that the membrane-binding a⁄nity was dependent on the puri¢cation level and whether the protein was derived from natural or recombinant sources. The study suggests that the HBM may be useful for quality control screening purposes. The gold surface associated with HBMs provides the opportunity to electrically monitor changes that occur at the membrane surface. Electrical readout of protein function or membrane structure is convenient for biosensor applications and for studying fundamental structure/function relationships. Variations in HBM design allow for di¡erent insulating and conductivity characteristics as appropriate for the components and responses being studied. It is important to note that an HBM can be integrated into instruments that simultaneously monitor optical and electrical

Biomimetic Membranes on Metal Supports

145

properties. This feature provides the opportunity to make independent measurements of the di¡erent kinds of changes that occur in the membrane [38]. The excellent insulating properties and high speci¢c capacity of an alkanethiol-supported membrane allows the HBM to be used as a sensitive capacitive electrode. The function of ion-translocating membrane proteins including bacteriorhodopsin, Naþ ,Kþ -ATPase, Hþ ,Kþ -ATPase, and Ca2 þ -ATP can be detected when membrane fragments containing these proteins are adsorbed onto an HBM [150].When a short light pulse is used to stimulated pump activity (i.e., release caged-ATP), ion transport into the interstitial space between the added membrane fragment and the HBM results in a measurable change in charge at the gold electrode. Fast £ow methods were used to study Naþ ,Kþ -ATPase in response to rapid concentration changes of ATP and ions in the bathing solution [151].These types of experiments are only possible with rugged solid-supported membranes. The use of HBMs to monitor the function of ion pumps could have a number of pharmaceutical or clinical screening applications. Alkanethiol-supported HBMs are excellent barriers that prevent redox species from encountering the gold electrode. In many cases, this feature limits the use of these HBM to electrical measurements that monitor impedance changes that occur in the membrane. Although impedance measurement can be used to monitor biological activity at a membrane surface (e.g., PLA2 activity [38] and ion-channel formation [34]), it is often desirable to reduce the barrier properties of the supporting thiol layer so that redox species can e⁄ciently transfer electrons to the gold electrode. Electroactive HBMs are prepared by introducing molecular defects in the alkanethiol layer that allow the redox species to participate in electron transfer processes with the gold electrode [57,152]. Ideally, the defects are large enough to provide electrode access, but small enough to allow a continuous monolayer of phospholipid to cover the holes. Phospholipids adsorbed onto disordered short-chain alkanethiol monolayers or longer-chain alkylthiol monolayers containing defects allow the use of lipophilic redox species to mediate electron transfer between redox-active peripheral membrane proteins and the gold electrode. The activities of urate oxidase [153], fructose dehydrogenase [154], and Escherichia coli pyruvate oxidase [155] have been monitored with electroactive HBMs. In addition to functioning as a membrane surface for protein immobilization, the bilayer structure acts as an e¡ective barrier preventing extraneous charged hydrophilic redox species, such as ascorbic acid, from interacting with the gold electrode. This is important for the development of biosensors that will assay complex solutions (i.e., blood, urine, or fruit extracts) containing additional redox-active compounds.

146

Elliott et al.

Our laboratory has developed an electroactive HBM that allows us to monitor the activity of biomolecules that induce structure defects in the supported phospholipid membrane. The alkylthiolate monolayer layer is formed from 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-[3-[2pyridyldithio]propionate] as shown in Fig. 20a.The resulting monolayer has extensive pinholes as determined from the redox currents observed in the presence of Fe(CN)63/4 (Fig. 20b). When POPC vesicles are added, the redox currents drop >40-fold, indicating the adsorbed phospholipid layer prevents permeation of the charged redox species (Fig. 20b). Impedance analysis and in situ ellipsometry agree with the CV results and indicate the formation of a phospholipid monolayer with the correct structure. This HBM is sensitive to structure defects in the phospholipid layer since the supporting alkylthiolate layer has a large number of pinholes. We used this surface to detect the function of peptides and proteins that may induce structural perturbations in the phospholipid layer. Figure 20c shows a plot of the peak faradaic current near the formal potential of Fe(CN)63/4 after two di¡erent concentrations of bacterial phospholipase C have been added. As observed, phospholipase C appears to induce structure defects in the HBM (Fig. 20d) that allow the charged redox species access to the gold electrode.We were also able to use this membrane system to detect low concentrations (30 nM) of the peptide toxin, melittin. Our results support the prevailing idea that melittin increases membrane permeability. We are presently attempting to determine the types of structure defects that are formed in the phospholipid surface using a combination of electrochemistry, infrared spectroscopy, and neutron re£ectivity. This study illustrates how the fabrication versatility of the HBMs can be used to tailor the surface for coupling enzymatic function to electrical readouts. A number of di¡erent chemistries have been employed in HBMs to provide a hydrated region between the alkane chains of the membrane and the solid surface. These include ethylene oxide units [34,70^75], hydrophilic peptides and similar linker chemistries [76,156], and polymeric materials [39,45,157,158]. The thin aqueous reservoir may better accommodate the extramembraneous regions of transmembrane proteins, and these regions may act as an ion reservoir proximal to the metal electrode. Cornell et al. developed an impedance biosensor based on analytedirected association or disassociation of gramicidin subunits in each lea£et of a tethered HBM [34]. This sensor was the ¢rst example of a hybrid bilayer membrane constructed with ethylene oxide-containing alkylthiolates to form a thin aqueous reservoir between the phospholipid membrane and the solid surface. The detection limit for various drugs, proteins, hormones, antibodies, and DNA [159] has been reported to be in the 107^1013 M range [160]. A schematic of this construct, shown in Fig. 21a, illustrates the

Biomimetic Membranes on Metal Supports

147

FIG. 20 Detection of phospholipase C activity with an electroactive HBM. A monolayer of 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-[3-[2-pyridyldithio]propionate] (A) containing a large number of defects as observed by cyclic voltammetry in the presence of Fe(CN)63/4. When POPC phospholipid vesicles are added to the monolayer, the redox current is reduced greater than 40fold after 7 h (B). SPR and impedance analysis confirmed the formation of a phospholipid monolayer on the alkylthiolate surface. When phospholipase C is added to the HBM, faradaic current near the formal potential is observed (C), suggesting that PLC introduces defects in the membrane surface (D).

high level of engineering that is possible with self-assembly techniques. The long-term stability, ease of construction, and high sensitivity suggest that the device may be suitable for commercial pharmaceutical and clinical applications. The bacterial ion channel OmpF was reconstituted into HBMs formed with thia-poly(ethylene oxide)-phospholipids [78]. Simultaneous impedance spectroscopy and SPR measurements were used to monitor channel gating when a fragment of colicin N, a ligand for the OmpF porin, bound to the receptor. Glazier et al. examined the insertion of a-hemolysin into HBMs prepared from thia-hexa(ethylene oxide)-octadecane [161].The data suggest that the channel-forming protein at least partially reconstitutes in the

148

Elliott et al.

FIG. 21 Complex HBMs. (A) Multiple of poly(ethylene oxide)-based alkylthiols and spacer thiols were used to construct the lower leaflet of an analyte sensor HBM. A gramicidin hemichannel in the lower leaflet was immobilized through a thia(ethylene oxide) linker. The upper monolayer contains lipid and a second gramicidin hemi channel. Dimerization of gramicidin was prevented by an immobilized ligand^receptor interaction. The addition of excess ligand competitively displaces the upper gramicidin channel from the anchor, allowing the formation of a conducting gramicidin dimer. (Adapted from Ref. 34.) (B) UV lithography was used to prepare patterned thiol monolayers containing thia-poly(ethylene oxide)phospholipid and 11-mercaptoundecanoic acid. The spanning bilayers containing rhodopsin and its associated G proteins were then formed by detergent dilution techniques. (Adapted from Ref. 170.)

Biomimetic Membranes on Metal Supports

149

supported membrane. HBMs have also been used in simple ion sensor applications. For example, potassium sensors that utilize the ionophore valinomycin have been demonstrated [72,73,162,163]. Several studies have demonstrated functional reconstitution of complex transmembrane proteins in HBMs. Mammalian cytochrome c oxidase (COX) is a large 28-transmembrane helix protein with 13 subunits. It is involved in the oxidative phosphorylation pathway and catalyzes the reduction of oxygen by cytochrome c. Burgess et al. used detergent dilution methods to reconstituted COX into a submonolayer of octadecanethiol on silver/ gold-coated quartz crystals for QCM and electrochemistry studies [112].The protein exhibited electron transfer to the gold electrode, indicating that it was reconstituted in a functional form. The COX-modi¢ed electrode could also be used to monitor the oxidation of reduced cytochrome c. AFM examination of the HBM surface suggested the presence of monomers and small aggregates of COX [121]. Importantly, the electrodes were stable and could withstand temperatures in excess of 80 C, suggesting that surface immobilization stabilizes the COX protein structure [164]. COX was also reconstituted into peptide-supported HBMs [165]. Immobilization, structure, and function of the COX protein were examined simultaneously with SPR, surface plasmon £uorescence, and electrochemistry in the same sample cell. Evidence for COX insertion was provided by SPR and surface plasmon £uorescence using £uorescently labeled antibodies. Impedance analysis indicated that only  70% of the surface was covered with a lipid bilayer, but electron transfer and proton pumping in the presence of reduced cytochrome c, and its dependence on cyanide concentration, were demonstrated. Electron transfer during oxidation of cytochrome c was monitored at þ 300 mV and transmembrane proton accumulation was measured at 700 mV, where Hþ is reduced to H2. Several other transmembrane proteins have been reconstituted into peptide and ethylene oxide-linked membranes. Naumann et al. immobilized Hþ -ATPases into peptide-supported bilayers and were able to monitor ATPdependent proton pumping with square-wave voltammetry and doublepulse chronoamperometry techniques [166,167]. Functional reconstitution of bacteriorhodopsin on thiophospholipid HBMs has been attained [168]. In this study, pump currents were measured by impedance analysis. The nicotinic acetylcholine receptor (NAR) has been reconstituted into HBMs and was shown to retain the binding properties of structuresensitive antibodies, a-bungarotoxin, and a small molecule agonist, carbamoylcholine [96,169]. Internal re£ection IR studies suggest the structural properties of the protein in the supported membranes are similar to those in native membranes. Yu et al. immobilized NAR into alkanethiol HBMs to monitor real-time kinetics of an immunogen complex reaction [52].

150

Elliott et al.

Although NAR ion-channel activity was not demonstrated in either of these studies, the results suggest that the structural features of the channel are conserved on the supported membrane systems. G-protein-coupled receptors (GPCR) are an important class of drug targets for the pharmaceutical industry. Rhodopsin, a light-activated GPCR, and its associated G proteins have been reconstituted into a patterned HBM [170]. The use of detergent dialysis techniques resulted in the formation of suspended bilayers containing rhodopsin over micropatterned regions of 11-mercaptoundecanoic acid as depicted in Fig. 21b. Using imaging SPR, light- and GTP-dependent G-protein release from the receptor was demonstrated.The patterned HBM used in this study was prepared with UV lithography techniques. This demonstrates the fabrication versatility that is available with HBM systems. HBMs can also be prepared with cell membrane proteins that are not easy to purify. An e⁄cient way of producing a supported membrane containing transmembrane proteins is to fuse whole cell membranes, isolated by osmotic cell lysis, to alkanethiol monolayers. Cell membrane hybrids from erythrocyte [171] and mammalian COS cell membranes have been prepared in our laboratory. Approximately 35 — is added to the thickness of the surface layer, suggesting the addition of the equivalent of a single lea£et of the cell membrane. The kinetics of addition of the membrane layer are determined by the di¡usion constant of the cell membrane vesicles, suggesting that the mechanism for cell membrane hybrid formation is the same as that for HBM formation [43]. Cell membrane hybrids may be useful for screening biological ligands. Figure 22 shows that cell membrane hybrids prepared from the membranes of COS cells transfected with CCR5, a chemokine receptor, bind antibodies to both the N terminus and the C terminus of CCR5. This result indicates the receptor has a bidirectional orientation at the surface.These antibodies do not bind to cell membrane hybrids prepared from untransfected COS cell membranes, as shown in the inset of Fig. 22. More recent data indicate that these cell membrane hybrids speci¢cally recognize their chemokine ligand, RANTES. 4.2

Supported Membranes for Understanding Biological Molecule Function

Hybrid bilayer membranes provide the opportunity to examine membranebound proteins with methods that are not readily accessible to other model membrane systems. For example, the electrode allows the ability to apply a potential across the membrane surface while surface features are being examined with other physical techniques. Real-time binding kinetics can be obtained with SPR and QCM techniques. Changes in membrane structure in

Biomimetic Membranes on Metal Supports

151

FIG. 22 Cell membrane hybrids for chemokine screening. Mammalian COS cells transfected with CCR5 chemokine receptor were the source of membrane for forming the cell membrane hybrid. The kinetics of the formation of the hybrid are analogous to that for HBMs from vesicles. Antibodies to the N terminus of the genetically engineered chemokine, and to the C terminus, bind to the cell membrane hybrid, but not to hybrids prepared from membranes of cells that were mock-transfected (inset).

response to a protein can be measured by AFM, IR spectroscopy, impedance, and electrochemical techniques. These data can provide new information about the structure and function of biomolecules at membrane interfaces. A variety of interesting observations were made in the studies described in the previous section. For example, Cooper et al. [144,172] reported that the binding a⁄nity for glycopeptide antibiotics to the HBM-anchored C22 -Lys(Ac)-D -Ala-D -Ala mucopeptide correlated with their relative antimicrobial activity. This is in contrast to the binding a⁄nities obtained from the antibiotics and a soluble form of the mucopeptide. This result suggests that, in at least some cases, membrane-immobilized ligands are a more representative model of the native proteins present in the bacterial cell wall. Similar results were observed for the binding a⁄nity and kinetics between the Cry1Ac toxin and HBM-immobilized GPI-linked aminopeptidase M [144]. The binding a⁄nity was measured to be 3.0 nM and is in agreement with vesicle studies, but this value is 30-fold lower than that reported with the respective soluble versions. The dissociation constant between solubilized bacterial porin OmpF and the R domain of its ligand, colicin N, is 100 mM,

152

Elliott et al.

but when OmpF was immobilized in an HBM, the dissociation constant was measured to be  0.6 mM [78], similar to the value measured in vivo. In addition, it was found that the Rdomain of colicin N caused closure of the OmpF porin channel. These data are examples of the importance of immobilizing proteins and ligands into membrane mimic environments. Details about the speci¢c binding and electrogenic transport properties of Kþ (and Naþ in Naþ ,Kþ -ATPase were identi¢ed with a rapid solution-exchange technique that is possible for protein only on a solidsupported membrane [173]. The intrinsic disassociation rate constant for the transducin/rhodopsin complex (  2.5 mM1-s1) was measured directly by SPR on the patterned GPCR-containing HBMs [170].The availability of a well-de¢ned membrane surface allowed a direct measure of the approximate number of phospholipids (  208lipids/protein) associated with a membrane-bound phospholipase A2 protein [174]. Burgess et al. were able to use cytochrome c oxidase immobilized into HBMs to study the kinetics and conditions that are required to convert the enzyme between a resting and pulse state [175]. They found that resting-state activities are observed when the electron transfer rate is 0.4^0.7 s1, values  10-fold lower than those previously reported. A signi¢cant number of studies on the structure and function of the peptide toxin melittin have been performed with HBMs. Melittin is a poreforming peptide that is responsible for lysing cellular membranes. Although it is known to form pores large enough to cause the release of hemoglobin in erythrocytes, the exact action of the molecule is unknown. Some studies suggest that melittin resides perpendicular to the phospholipid membrane, and others suggest that it lies parallel to the membrane. Plant et al. used capacitance and electrochemical measurements to examine the e¡ect of melittin on HBMs [32,58]. Melittin did not signi¢cantly change the capacitance of the membrane, suggesting that it did not cause signi¢cant membrane disruption, but increases in electron transfer with Fe(CN)63/4 were observed, suggesting that melittin does form lesions in the phospholipid surface. Recent electrochemical work demonstrated that some of the e¡ect of melittin on increasing electron transfer for negatively charged redox species could be explained by electrostatic e¡ects [81]. In these studies, it was found that melittin addition to an HBM completely suppressed pinhole redox currents from Ru(NH3)63 þ /2 þ , did not change the residual redox currents from hydroxymethylferrocine (FcCH2OH), and increased the redox currents associated with Fe(CN)63/4. These data suggest that the basic residues in melittin at least in part alter the redox processes by electrostatic screening interactions. In this study, the capacitance of the HBM increased upon the addition of melittin, suggesting that the toxin perturbs the membrane structure. Steinem et al. found that the addition of up to 6 mM

Biomimetic Membranes on Metal Supports

153

melittin does not signi¢cantly alter the capacitance nor the resistance across an ethylene oxide-supported HBM [77]. Interestingly, they showed that phospholipase A2 and melittin act synergistically to greatly reduce membrane capacitance and increase conductance. A QCM study by the same group suggests that melittin forms multilayers on the HBM surface [176]. From the results in these studies and those from many di¡erent bilayer systems, it is still unclear how melittin functions at the membrane surface. We are attempting to put to rest many controversies surround melittin action with the use of neutron re£ectivity [91,132,133], SPR, and other electrochemical measurements. We anticipate that the use of a rugged supported membrane will continue to allow such detailed questions to be addressed and eventually lead to an unraveling of the parameters associated with melittin behavior, including concentration dependence, and the e¡ects of charge and applied potential.

5

CHALLENGES

For many applications, including the molecular recognition of membrane surface ligands, and studies of peripheral membrane enzyme activity, the structure of the simple HBM is probably su⁄cient. However, for applications involving transmembrane proteins, and particularly for studies involving membrane protein structure and function, it may be necessary to prepare a membrane that has a more biomimetic character. The fabrication chemistry and methodology may need to be tailored not only to optimize bioactivity, but also to achieve a speci¢c assay readout. For example, to read out protein function electrically, the insulating properties of the membrane may be very important. If optical readouts are used to monitor protein function, then the insulating properties of the membrane may be irrelevant. For transmembrane proteins, a hydrated compartment proximal to the gold surface is expected to be required for correct reconstitution. Examples of sophisticated chemistries and fabrication approaches have been discussed in the previous section. However, new approaches to solving this issue are still under development, and include chelator-based linkers for supporting planar bilayers [156,177], covalent linker chemistries [178,179], streptavidinsupported membranes [180], and polymer-based supports [19]. Another approach that is being examined is the fabrication of surfaces that are patterned with areas of thiol-linked alkanes and areas of short polar moieties [79,170,181]. This approach is an e¡ort to produce areas of tethered alkanethiol-phospholipid bilayers and areas of suspended phospholipidphospholipid bilayers that are not tethered to the surface. Methods to form

154

Elliott et al.

membrane micropatterned surfaces are signi¢cant because they have applications in microarray technologies for multiplexing, and high- and medium-throughput analyte analyses.Unambiguous characterization of the results of such fabrication approaches is not trivial. One study examined vesicle association and fusion to a patterned octadecanethiol and b-mer captoethanol surface by SPR microscopy [182]. The phospholipid adsorption kinetics on the two di¡erent areas are very di¡erent, but the total amount of lipid added to the surface appears to be consistent with the addition of phospholipid monolayer on the octadecanethiol regions and bilayer over the b-mercaptoethanol region. Impedance analysis of patterned surface requires the use of complicated equivalent circuit models to describe various types of defects that are possibly present at the surface [163]. In our laboratory we have observed that measurements from a single technique are not always su⁄cient to provide a correct interpretation of the data. For example, experiments were performed to examine the e¡ect of surface free energy on the kinetics of formation of HBMs from vesicles. Mixtures of ^ CH3 - and ^ OH-terminated thiols were used to vary the surface energy, and vesicle adsorption to the surface was monitored by SPR as shown on the left in Fig. 23. When vesicles were added to surfaces containing a fraction of 30% or less ^ OH alkylthiol, SPR indicated a change in thickness that was consistent with the addition of a single layer of phospholipid.The graph in Fig. 23 shows that the apparent change in thickness that occurred when vesicles were added to a surface that was 30% ^ OH alkylthiol was approximately 23 —. From these data alone, one would conclude that a monolayer of phospholipid added to the surface. However, impedance analysis indicated a capacitance for the putative ‘‘monolayer’’ of phospholipid of greater than 16 mF/cm2, which is approximately 10-fold higher than expected for an insulating lipid layer. Panel a on the right of Fig. 23 shows the result of addition of vesicles to the 100% ^ CH3 -terminated surface. The AFM image shows the surface is topographically smooth, consistent with the formation of a phospholipid monolayer. The addition of vesicles to the surface composed of 50% ^ OH alkylthiol (panel b) produced a rough surface, suggesting the presence of unfused or partially fused vesicles. Interestingly, at the 100% ^ OH-terminated surface, SPR data suggest addition of a complete bilayer of phospholipid. However, impedance analysis shows that there was e¡ectively no change in the capacitance of the surface after addition of phospholipid vesicles. The initial AFM scans of this surface were poorly resolved, although subsequent scanning produced a smooth topology as shown in panel c.Thus, in some circumstances, it will be critical to examine the results from several techniques in order to characterize the structural features of the surfaces correctly. This will be particularly important as fabrication approaches

Biomimetic Membranes on Metal Supports

155

FIG. 23 Fusion of POPC vesicles to mixed monolayers containing HSC11OH and HSC11CH3. SPR data are shown on left. The fraction of methyl terminated thiol in the monolayer is indicated by x. Vesicles are added at time 0. The break in the curves indicates the flow of buffer without vesicles into the cell. On the right are shown AFM images of surfaces after exposure to vesicles. (a) x ¼ 1:0. (b) x ¼ 0:5. (c) The initial AFM scans on the 100% hydroxyl terminated surface ðx ¼ 0Þ are not well resolved (not shown), but after repeated scans, the lipid appears to organize into layer with a smooth topology. Scale at bottom is represents the z dimension.

become increasingly complex in our attempts to reconstitute active transmembrane proteins.

6

CONCLUSIONS

The hybrid bilayer membrane is composed of both synthetic and natural components that assemble at a solid support. While the composition and arrangement of the molecules of HBMs precludes a truly ‘‘native’’ structure, this approach does provide advantages such as ease of fabrication and stability of the structure. For these reasons, the HBM provides the opportunity to take advantage of the functions of biological membranes for applications such as sensing, screening, and diagnostics. The ruggedness of the HBM lends itself to a variety of analytical techniques, the combination of which makes it possible to know unequivocally

156

Elliott et al.

many details of the physical nature of the structure. Together with the fact that the preparation of the hybrid membrane is spontaneous and reproducible, HBMs can be easier to use as research tools and as commercial analytical devices than other model membranes. Many of the tools and approaches that have been used in the development, characterization, and application of HBMs involve materials science and engineering. For scientists interested in the materials properties of biologically derived structures, the hybrid membrane presents a healthy challenge as an ultrathin material, with features of interest that are on the nanometer scale. Importantly, the use of hybrid bilayer membranes is providing opportunities to examine biological membranes in new ways, and is beginning to provide new insight into biological molecules and biologically relevant phenomenan. ACKNOWLEDGMENTS We would liketoacknowledge the manycontributions to the HBMproject that have been made by colleagues at NISTand other who have visited NISTduring experimental studies. We would like to thank Dr. Joseph Hubbard, Dr. N. Madhusudhana Rao and Dr. Herbert Wieder for their contributions to the work on the mechanism of hybrid bilayer formation, cell membrane hybrids, and vesicle fusion to mixed monolayer surfaces, respectively. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9.

10. 11. 12.

SJ Singer, GL Nicolson. Science 175:720, 1972. GL Nicolson. In: (CR Richmond et al., eds. Mammalian Cells: Probes and Problems. Oak Ridge,TN: Technical Information Center, pp 246^253, 1975. Y Schi¡mann. Mol Cell Biochem 42:117, 1982. K Jacobson, ED Sheets, R Simson. Science 268:1441, 1995. K Jacobson, C Dietrich. Trends Cell Biol 9:87, 1999. AK Kenworthy, N Petranova, M Edidin. Mol Biol Cell 11:1645, 2000. C Ostermeier, H Michel. Curr Opin Struct Biol 7:697, 1997. A Pautsch, GE Schulz. J Mol Biol 298:273, 2000. K Palczewski,T Kumasaka,T Hori, CA Behnke, H Motoshima, BA Fox,T Le, I, DC Teller, T Okada, RE Stenkamp, M Yamamoto, M Miyano. Science 289:739, 2000. J Heberle, G Buldt, E Koglin, JP Rosenbusch, EM Landau, J Mol Biol 281:587, 1998. J Rigaud, M Chami, O Lambert, D Levy, J Ranck. Biochim Biophys Acta 1508:112, 2000. M Edidin. J Cell Sci Suppl 17:165, 1993.

Biomimetic Membranes on Metal Supports 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49.

157

M Bloom, JLThewalt. Mol Membr Biol 12:9, 1995. JB Massey, DH Bick, HJ Pownall. Biophys J 72:1732, 1997. AD Bangham. Chem Phys Lipids 64:275, 1993. M Montal. J Membr Biol 98:101, 1987. YK Levine. Progr Surface Sci 3:279, 1973. E Sackmann. Science 271:43, 1996. E Sackmann, M Tanaka. Trends Biotechnol 18:58, 2000. CR Sanders, RS Prosser. Struct Fold Des 6:1227, 1998. BE Ehrlich. Meth Enzymol 207:463, 1992. A Chonn, PR Cullis. Curr Opin Biotechnol 6:698, 1995. LK Tamm, HM McConnell. Biophys J 47:105, 1985. JT Groves, C Wul¢ng, SG Boxer. Biophys J 71:2716, 1996. HM McConnell, TH Watts, RM Weis, AA Brian. Biochim Biophys Acta 864:95, 1986. M Seul, S Subramaniam, HM McConnell. J Phys Chem 89:3592, 1985. E Goormaghtigh, V Raussens, JM Ruysschaert. BBA- Rev Biomembr 1422:105, 1999. MLWagner, LK Tamm. Biophys J 79:1400, 2000. J Salafsky, JT Groves, SG Boxer. Biochemistry 35:14773, 1996. EM Erb, K Tangemann, B Bohrmann, B Muller, J Engel. Biochemistry 36:7395, 1997. BW Koenig, S Kruger, WJ Orts, CF Majkrzak, NF Berk, JV Silverton, K Gawrisch. Langmuir 12:1343, 1996. AL Plant. Langmuir 9:2764, 1993. AL Plant. Langmuir 15:5128, 1999. BA Cornell, VL Braach-Maksvytis, LG King, PD Osman, B Raguse, LWieczorek, RJ Pace. Nature 387:580, 1997. RG Nuzzo, BR Zegarski, LH Dubois. J Am Chem Soc 109:733, 1987. S Lingler, I Rubinstein,W Knoll, A O¡enhausser. Langmuir 13:7085, 1997. S Terrettaz,T Stora, C Duschl, H Vogel. Langmuir 9:1361, 1993. M Stelzle, G Weissmuller, E Sackmann. J Phys Chem 97:2974, 1993. KG Marra, DDA Kidani, EL Chaikof. Langmuir 13:5697, 1997. T Liebermann,W Knoll. Colloid Surface A 171:115, 2000. J Enderlein. Biophys J 78:2151, 2000. RG Nuzzo, DL Allara. J Am Chem Soc 105:4481, 1983. JB Hubbard,V Silin, AL Plant. Biophys Chem 75:163, 1998. E Kalb, S Frey, LK Tamm. Biochim Biophys Acta 1103:307, 1992. I Pockrand. Surface Sci 72:577, 1978. V Silin, H Weetall, DJ Vanderah. J Colloid Interface Sci 185:94, 1997. V Silin, A Plant. Trends Biotechnol 15:353, 1997. J Spinke, J Yang, H Wolf, M Liley, H Ringsdorf, W Knoll. Biophys J 63:1667, 1992. M Abramowitz, IA Stegun. Handbook of Mathematical Functions, National Bureau of Standards Applied Mathematics. Series 55, Chapter 7. Washington, DC: U.S. Government Printing O⁄ce, 1966.

158

Elliott et al.

50. TP Petralli-Mallow, AL Plant, ML Lewis, JM Hicks. Langmuir 16:5960, 2000. 51. AL Plant, M Brigham-Burke, EC Petrella, DJ O’Shannessy. Anal Biochem 226:342, 1995. 52. YY Yu, BJ VanWie, AR Koch, DF Mo¡ett,WC Davis. Anal Biochem 263:158, 1998. 53. S Raut, LWeller,TW Barrowcli¡e, Br J Haematol 107:323, 1999. 54. C Miller, P Cuendet, M Gratzel. J Phys Chem 95:877, 1991. 55. MD Porter, TB Bright, DL Allara, CED Chidsey. J Am Chem Soc 109:3559, 1987. 56. JR Macdonald. Impedance Spectroscopy: Emphasizing Solid Materials and Systems. New York: Wiley, 1987. 57. HO Finklea. In: (AJ Bard, I Rubinstein, eds. Electroanalytical Chemistry. New York: Marcel Dekker, pp 109^335, 1996. 58. AL Plant, M Gueguetchkeri,W Yap. Biophys J 67:1126, 1994. 59. B Hille. Ionic Channels of Excitable Membranes. Sunderland: Sinauer Associates, 1997. 60. RB Gennis. Biomembranes. New York: Springer-Verlag, 1992. 61. CW Meuse, G Niaura, ML Lewis, AL Plant. Langmuir 14:1604, 1998. 62. P Diao, DL Jiang, XL Cui, DP Gu, RT Tong, B Zhong. Bioelectrochem Bioenerget 48:469, 1999. 63. E Boubour, RB Lennox. Langmuir 16:4222, 2000. 64. JP Dilger, SG McLaughlin,TJ McIntosh, SA Simon. Science 206:1196, 1979. 65. R Benz, O Frohlich, P Lauger, M Montal. Biochim Biophys Acta 394:323, 1975. 66. C Steinem, A Jansho¡,WP Ulrich, M Sieber, HJ Galla. Biochim Biophys Acta 1279:169, 1996. 67. L Ding, JH Li, SJ Dong, EK Wang. J Electroanal Chem 416:105, 1996. 68. EL Florin, HE Gaub. Biophys J 64:375, 1993. 69. HT Tien, J Kutnik, P Krysinski, ZK Lojewska. In: M Blank, ed. Electrical Double Layers in Biology. New York: Plenum, pp 149^166, 1985. 70. DJ Vanderah, CW Meuse,V Silin, AL Plant. Langmuir 14:6916, 1998. 71. DJ Vanderah, CP Pham, SK Springer, V Silin, CW Meuse. Langmuir 16:6527, 2000. 72. B Raguse, V Braach-Maksvytis, BA Cornell, LG King, PDJ Osman, RJ Pace, LWieczorek. Langmuir 14:648, 1998. 73. C Steinem, A Jansho¡, K dem Bruch, K Reihs, J Goossens, HJ Galla. Bioelectrochem Bioenerget 45:17, 1998. 74. LM Williams, SD Evans, TM Flynn, A Marsh, PF Knowles, RJ Bushby, N Boden. Langmuir 13:751, 1997. 75. H Lang, C Duschl, H Vogel. Langmuir 10:197, 1994. 76. N Bunjes, EK Schmidt, A Jonczyk, F Rippmann, D Beyer, H Ringsdorf, P Graber,W Knoll, R Naumann. Langmuir 13:6188, 1997. 77. C Steinem, HJ Galla, A Jansho¡. Phys Chem Chem Phys 2:4580, 2000. 78. T Stora, JH Lakey, H Vogel. Angew Chem Int Ed 38:389, 1999. 79. ATA Jenkins, N Boden, SD Bushby, PF Evans, PF Knowles, RE Miles, SD Ogier, H Schonherr, GJ Vansco. J Am Chem Soc 121:5274, 1999.

Biomimetic Membranes on Metal Supports 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114.

159

J Wang. Analytical Electrochemistry. New York: Wiley, 2000. DA Brevnov, HO Finklea. Langmuir 16:5973, 2000. O Chailapakul, RM Crooks. Langmuir 9:884, 1993. P Krysinski, MR Moncelli, F Tadini-Buoninsegni. Electrochim Acta 45:1885, 2000. E Sabatani, J Cohenboulakia, M Bruening, I Rubinstein. Langmuir 9:2974, 1993. HO Finklea, DA Snider, J Fedyk, E Sabatani,Y Gafni, I Rubinstein. Langmuir 9:3660, 1993. TM Nahir, EF Bowden. Electrochim Acta 39:2347, 1994. E Sabatani, I Rubinstein, R Maoz, J Sagiv. Electroanal Chem 219:365, 1987. E Sabatani, I Rubinstein. J Phys Chem 91:6663, 1987. LK Tamm, SA Tatulian. Q Rev Biophys 30:365, 1997. RA Dluhy. Appl Spectrosc Rev 35:315, 2000. CW Meuse, S Krueger, CF Majkrzak, JA Dura, J Fu, JT Connor, AL Plant. Biophys J 74:1388, 1998. AN Parikh, B Liedberg, SVAtre, M Ho, DL Allara. J Phys Chem 99:9996,1995. RG Snyder, JH Schachtschneider. Spectrochim Acta A 19:85, 1963. J Ha, CS Henry, I Fritsch. Langmuir 14:5850, 1998. R Kellner, B Mizaiko¡, M Jakusch, HD Wanzenbock, N Weissenbacher. Appl Spectrosc 51:495, 1997. A Sevin-Landais, P Rigler, S Tzartos, F Hucho, R Hovius, H Vogel. Biophys Chem 85:141, 2000. YL Cheng, N Boden, RJ Bushby, S Clarkson, SD Evans, PF Knowles, A Marsh, RE Miles. Langmuir 14:839, 1998. M Liley,TA Keller, C Duschl, H Vogel. Langmuir 13:4190, 1997. WN Hansen. J Opt Soc Am 58:380, 1968. MJ Citra, PH Axelsen. Biophys J 71:1796, 1996. W Hubner, HH Mantsch. Biophys J 59:1261, 1991. P Wenzl, M Fringeli, J Goette, UP Fringeli. Langmuir 10:4253, 1994. S Gottesfeld. Electroanal Chem 15:143, 1989. V Reipa, AK Gaigalas,VLVilker. Langmuir 13:3508, 1997. C Striebel, A Brecht, G Gauglitz. Biosens Bioelectron 9:139, 1994. PN Bartlett, K Brace, EJ Calvo, R Etchenique. J Mater Chem 10:149, 2000. M Rodahl, B Kasemo. Rev Sci Instrum 67:3238, 1996. CA Keller, K Glasmastar, VP Zhdanov, B Kasemo. Phys Rev Lett 84:5443, 2000. CA Keller, B Kasemo. Biophys J 75:1397, 1998. A Jansho¡, HJ Galla, C Steinem. Angew Chem Int Ed 39:4004, 2000. JD Burgess, FH Hawkridge. Langmuir 13:3781, 1997. JD Burgess, MC Rhoten, FM Hawkridge. Langmuir 14:2467, 1998. A Jansho¡, C Steinem, M Sieber, HJ Galla. Eur Biophys J Biophys 25:105, 1996. C Steinem, A Jansho¡, J Wegener, WP Ulrich, W Willenbrink, M Sieber, HJ Galla. Biosens Bioelectron 12:787, 1997.

160

Elliott et al.

115. 116. 117.

G Binnig, H Rohrer, Ch Gerber, E Weibel. Phys Rev Lett 49:57, 1982. G Binnig, CF Quate, Ch Gerber. Phys Rev Lett 56:930, 1986. VJ Morris, AR Kirby, AP Gunning. Atomic Force Microscopy for Biologists. London, UK: Imperial College Press, 1999. JA Zasadzinski, CA Helm, ML Longo, AL Weisenhorn, SA Gould, PK Hansma. Biophys J 59:755, 1991. YF Dufrene, T Boland, JW Schneider, WR Barger, GU Lee. Faraday Discuss 111:79, 1998. J Schneider,YF Dufrene,WR Barger, GU Lee. Biophys J 79:1107, 2000. JD Burgess, VW Jones, MD Porter, MC Rhoten, FM Hawkridge. Langmuir 14:6628, 1998. J Pedulla, RD Deslattes. In: RB Hoover, AB Walker, eds. Multilayer and Grazing Incidence X-Ray/EUV Optics II. Proc SPIE 2011:299, 1994. JT Woodward,V Silin, AL Plant, S Krueger. In preparation, 2001. JT Woodward, ML Walker, CW Meuse, DJ Vanderah, GE Poirier, AL Plant. Langmuir 16:5347, 2000. CF Majkrzak, GP Felcher. Mater Res Bull 15:65, 1990. J Penfold, RK Thomas. J Phys
Condens Matter 2:1369, 1990. CF Majkrzak. Mater Res Soc Symp Proc 376:143, 1995. J Penfold, RM Richardson, A Zarbakhsh, JRP Webster, DG Bucknall, AR Rennie, RAL Jones, T Cosgrove, RK Thomas, JS Higgins, PDI Fletcher, E Dickinson, SJ Roser, IA McLure, AR Hillman, RW Richards, EJ Staples, AN Burgess, EA Simister, JW White. J Chem Soc Faraday Trans 93:3899, 1997. NF Berk, CF Majkrzak. Phys Rev B 51:11296, 1995. S Krueger, JF Ankner, SK Satija, CF Majkrzak, D Gurley, M Colombini. Langmuir 11:3218, 1995. SJ Johnson,TM Bayerl, DC McDermott, GWAdam, AR Rennie, RK Thomas, E Sackmann. Biophys J 59:289, 1991. S Krueger, CW Meuse, CF Majkrzak, JA Dura, NF Berk, M Tarek, AL Plant. Langmuir 17:511, 2001. CF Majkrzak, NF Berk, S Krueger, JA Dura, M Tarek, D Tobias, V Silin, CW Meuse, J Woodward, AL Plant. Biophys J 79:3330, 2000. M Tarek, K Tu, ML Klein, DJ Tobias. Biophys J 77:964, 1999. CD Bain. J Chem Soc FaradayTrans 91:1281, 1995. GL Richmond. Anal Chem 69:A536^ A543, 1997. YR Shen. Nature 337:519, 1989. V Vogel. Curr Opin Colloid Interface 1:257, 1996. LT Richter,TP Petralli-Mallow, JC Stephenson. Opt Lett 23:1594, 1998. CD Bain, EB Troughton, YT Tao, J Evall, GM Whitesides, RG Nuzzo. J Am Chem Soc 111:321, 1989. T Petralli-Mallow, KA Briggman, LJ Richter, JC Stephenson, AL Plant. In: M Fallahi, B Swanson, eds. Advanced Materials and Optical Systems for Chemical and Biological Detection. Proc SPIE 3858:25, 1999. GM Kuziemko, M Stroh, RC Stevens. Biochemistry 35:6375, 1996.

118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128.

129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141.

142.

Biomimetic Membranes on Metal Supports 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157. 158. 159. 160. 161. 162. 163. 164. 165. 166. 167. 168. 169.

170. 171. 172. 173.

161

A Jansho¡, C Steinem, M Sieber, A elBaya, MA Schmidt, HJ Galla. Eur Biophys J Biophys 26:261, 1997. MA Cooper, AC Try, J Carroll, DJ Ellar, DH Williams. BBA-Biomembranes 1373:101, 1998. MA Cooper, DH Williams. Anal Biochem 276:36, 1999. I Vikholm, K Gyo«rvary, J Peltonen. Langmuir 12:3276, 1996. CI Webster, MA Cooper, LC Packman. DH Williams, JC Gray. Nucleic Acids Res 28:1618, 2000. M Liley, J Bouvier, H Vogel. J Colloid Interface Sci 194:53, 1997. ME Medof, S Nagarajan, MLTykocinski. FASEB J 10:574, 1996. K Seifert, K Fendler, E Bamberg. Biophys J 64:384, 1993. J Pintschovius, K Fendler. Biophys J 76:814, 1999. P Krysinki, HT Tien, A Ottova. Biotechnol Progr 15:974, 1999. T Nakaminami, S Ito, S Kuwabata, H Yoneyama. Anal Chem 71:4278, 1999. KT Kinnear, HG Monbouquette. Anal Chem 69:1771, 1997. O Pierrat, N Lechat, C Bourdillon, JM Laval. Langmuir 13:4112, 1997. U Radler, J Mack, N Persike, G Jung, R Tampe. Biophys J 79:3144, 2000. W Knoll, CW Frank, C Heibel, R Naumann, A O¡enhausser, J Ruhe, EK Schmidt,WW Shen, A Sinner. J Biotechnol 74:137, 2000. J Radler, E Sackmann. Curr Opin Solid State Matter 2:330, 1997. SW Lucas, MM Harding. Anal Biochem 282:70, 2000. BA Cornell, VL Braach-Maksvytis, LG King, PD Osman, B Raguse, LWieczorek, RJ Pace. Novartis Found Symp 225:231, 1999. SA Glazier, DJ Vanderah, AL Plant, H Bayley, G Valincius, JJ Kasianowicz. Langmuir 16:10428, 2000. L Ding, JH Li, EK Wang, SJ Dong. Thin Solid Films 293:153, 1997. ATA Jenkins, RJ Bushby, N Boden, SD Evans, PF Knowles, QY Liu, RE Miles, SD Ogier. Langmuir 14:4675, 1998. MC Rhoten, JD Burgess, FM Hawkridge. Electrochim Acta 45:2855, 2000. R Naumann, EK Schmidt, A Jonczyk, K Fendler, B Kadenbach, T Liebermann, A O¡enhausser,W Knoll. Biosens Bioelectron 14:651, 1999. R Naumann, A Jonczyk, C Hampel, H Ringsdorf,W Knoll, N Bunjes, P Graber. Bioelectrochem Bioenerget 42:241, 1997. R Naumann, A Jonczyk, R Kopp, J Esch, H Ringsdorf, W Knoll, P Gra«ber. Angew Chem Int Ed 34:2056, 2001. C Steinem, A Jansho¡, F Hohn, M Sieber, HJ Galla. Chem Phys Lipids 89:141, 1997. EK Schmidt, T Liebermann, M Kreiter, A Jonczyk, R Naumann, A O¡enhausser, E Neumann, A Kukol, A Maelicke, W Knoll. Biosens Bioelectron 13:585, 1998. S Heyse, OP Ernst, Z Dienes, KP Hofmann, H Vogel. Biochemistry 37:507, 1998. NM Rao, AL Plant,V Silin, S Wight, SW Hui. Biophys J 73:3066, 1997. MA Cooper, DH Williams. Chem Biol 6:891, 1999. J Pintschovius, K Fendler, E Bamberg. Biophys J 76:827, 1999.

162

Elliott et al.

174.

LS Jung, JS Shumaker-Parry, CT Campbell, SS Yee, MH Gelb. J Am Chem Soc 122:4177, 2000. JD Burgess, MC Rhoten, FH Hawkridge. J Am Chem Soc 120:4488, 1998. C Steinem, A Jansho¡, HJ Galla. Chem Phys Lipids 95:95, 1998. T Stora, Z Dienes, H Vogel, C Duschl. Langmuir 16:5471, 2000. G Brink, L Schmitt, R Tampe, E Sackmann. BBA-Biomembranes 1196:227, 1994. ATA Jenkins, J Hu, YZ Wang, S Schiller, R Foerch, W Knoll. Langmuir 16:6381, 2000. C Bieri, OP Ernst, S Heyse, KP Hofmann, H Vogel. Nat Biotechnol 17:1105, 1999. J Lahiri, P Kalal, AG Frutos, ST Jonas, R Schae¥er. Langmuir 16:7805, 2000. ATA Jenkins,T Neumann, A O¡enha«usser. Langmuir 17:265, 2001.

175. 176. 177. 178. 179. 180. 181. 182.

4 Peptide- and Protein-Based Biomolecular Assemblies: Physical and Chemical Characterization for Optimal Function Paul W. Bohn University of Illinois at Urbana-Champaign, Urbana, Illinois, U.S.A.

1 1.1

INTRODUCTION Why Biomolecular Films?

Recently, a great deal of attention has been focused on the development of biomaterials derived from peptides, oligopeptides, and proteinacious materials for a variety of technological applications, including chromatographic supports
especially in a⁄nity chromatography
bioreactors, biosensors, and as functionalized building blocks for intelligent materials. There are numerous driving motivations supporting the development of peptide and protein-bearing interfacial structures. Proteins are Mother Nature’s worker bees. A very large number of naturally occurring proteins and protein subunits have been separated, identi¢ed, and characterized structurally and functionally. Thus, they constitute an advantageous starting point for fabrication of interfacial structures which can e¡ect molecular recognition and couple that molecular recognition to a physical or chemical signal. Also, since the study of proteins has a long history and is well developed, many of the tools which have been developed to understand structure and function of proteins in homogeneous solution and in vivo are directly transferable to arti¢cial assemblies, where planar or more complex 163

164

Bohn

interfacial structures are decorated with protein assemblies. Finally, the set of experimental tools known collectively as genetic engineering has been exploited to extend the distribution of naturally occurring proteins, to elaborate their structural and functional capabilities and to provide a rich array of engineered molecular architectures for construction of biomolecular assemblies. One signi¢cant challenge to realizing the full promise of protein-based biomaterials is the development of strategies to assemble complex molecular species into structures with order on the supermolecular length scale, i.e., the mesoscale. On the mesoscale, not only is the inherent structure of each individual molecule important, but a host of other structural details must be considered. How are neighboring molecules oriented, one relative to the next? Do they present the same three-dimensional motifs to the solvent accessible portion of the structure? Is there two-dimensional or even threedimensional crystalline order in the assembly? Are the molecules su⁄ciently close and oriented properly such that intermolecular communication can be accomplished electrically, optically, or mechanically? Finally, there is the overriding concern that the native functionality of the immobilized protein or peptide is retained after being incorporated in the assembly. 1.2

Importance of Proper Characterization

Given the complexity of the structures which are contemplated, and the degree of detail about structure and function which is required, it is critical that a suite of powerful characterization methodologies be developed in concert with the synthetic and assembly strategies used to fabricate biomolecular ¢lms.These characterization strategies are required to yield di¡erent information in some circumstances than in others, but in all cases must draw a tight connection between the structure of the biomolecular assemblies and their designed function. There are a number of hurdles to overcome to properly characterize biomolecular assemblies. The ¢rst of these clearly is measurement sensitivity. Interfacial structures inherently are composed of a very small number of molecules. In the case of a close-packed monolayer of a protein with a characteristic 1000 —2 molecular footprint and areal density of 1013cm2, a characterization experiment which accesses a 100 mm 100 mm area of the interface, to the exclusion of all the background material surrounding it, is still only interrogating ca. 1 femtomole ( ¼1015 mol) of material. Of course, as the area probed becomes smaller, the number of molecules addressed decreases proportionately. Fortunately, experimental tools have been developed to address interfaces which generate signals large enough to distinguish small numbers of analyte species, even individual molecules in

Peptide and Protein-Based Biomolecular Assemblies

165

favorable cases, so the small number of molecules available to be probed is not an insurmountable hurdle per se. It does, however, place signi¢cant constraints on the background generated by the surrounding materials, both solid and liquid, which may be used while still permitting useful characterization. Fluorescence spectroscopy, for example, is easily capable of generating signals which allow an individual molecule to be detected and distinguished from its neighbors. However, the background must either be completely devoid of emissive or scattering centers, or the experiment must reject signals generated from volumes of space other than the region being probed. Another key consideration is the length scale of the experiments which are used to characterize the biomolecular assemblies. It is important to distinguish between experiments that are characteristically molecular but provide information averaged over a macroscopic area and those which intrinsically sample the system on the nanometer length scale. In the depth dimension this issue is straightforward, because there are a number of localization phenomena, primarily optical, which can be used to provide a small and readily locatable sampling volume. Many of these optical localization phenomena scale with excitation wavelength, so it is possible to tune the volume probed simply by tuning the excitation wavelength. Achieving high spatial resolution with molecular speci¢city in the transverse, or in-plane direction, is much more di⁄cult, however. Optical experiments are limited by the di¡raction associated with the far-¢eld projection of sub wavelength image details. Tremendous advances have occurred in the past 15 years, primarily in the introduction of scanned probe microscopies which can elucidate molecular detail on the nanometer length scale and below. Scanned probe microscopies have rapidly come to the fore to play a major role in the characterization of interfacial structures generally, and biomolecular assemblies in particular. Finally, the very nature of peptide- and protein-containing biomolecules leads to another measurement challenge. In nature, most proteins derive their characteristic function from only a small portion of the molecule. In order to characterize biomolecular ¢lms composed of peptides and proteins properly, it is important to understand how the molecular sca¡olding interacts with the functional portion of the molecule to moderate its functionality in a biomolecular assembly. As but one example, surface-immobilized metalloproteins can undergo partial denaturation which disrupts the molecular sca¡olding directing the precise three-dimensional placement of the heme moiety, consequently disturbing, or in severe cases completely eliminating, the characteristic binding and spectroscopic properties associated with the heme prosthetic group.

166

1.3

Bohn

Scope of the Review

This chapter will focus on the physical and chemical characterization tools which have been successfully applied to characterization of bio¢lms composed of peptide and protein-based molecules. There are a variety of interesting biomolecular structures composed of sugar-based and oligonucleotide-based materials that have achieved both scienti¢c and technological interest. Although inclusion of these materials is beyond the scope of this chapter, in many cases the techniques and approaches applied to protein-containing ¢lms are also germane to characterizing biomolecular assemblies of these materials. The chapter will focus primarily on characterization, although some attention will be given to the methods of assembly which have been used to prepare complex monolayer and multilayer biomolecular assemblies. Although fabrication is not the major focus of this chapter, in some cases the method of assembly places constraints on the characterization tools which have to be considered in order to construct a detailed picture of how biomolecular structure a¡ects function. The characterization tools themselves are divided into those tools which are primarily physical and those which yield chemical information, although clearly some experiments have both characteristics. For example, the use of chemically derivatized tips in scanned force microscopies (SFMs) imparts molecular selectivity to the signal, yet the signal-generation process is primarily a physical event. Finally, the examples chosen to highlight the points made in the review are meant to be illustrative
this is not an exhaustive review of the current literature. If it were, the subject chosen for the current chapter could easily justify its own monograph. 2 2.1

METHODS OF ASSEMBLY Nonspecific Adsorption

From the contemporary perspective, nonspeci¢c adsorption is seen as an unwanted side product of a molecular assembly process which was designed to yield a speci¢c architecture in three dimensions. Nonspeci¢c adsorption thus gives rise to a dispersion in the structural presentation of protein and peptide motifs which is typically unwanted. The earlier literature, however, treats the native adsorption of proteins at surfaces as an interesting phenomenon to be studied in and of itself. Much of the pre-1985 interest in nonspeci¢c adsorption of proteins was driven by the need to understand plasma protein interactions with biomaterials which came into contact with blood. It was recognized early that protein transport to and from, and adsorption on, a foreign surface was much more rapid than the subsequent colonization by cells, e.g., platelets, leukocytes, erythrocytes, etc. These early studies had

Peptide and Protein-Based Biomolecular Assemblies

167

a clear biomedical motivation to develop nonthrombogenic materials to be used in implanted devices, and they motivated a large number of studies of the fundamental mechanisms driving adsorption of plasma-borne proteins onto progressively more carefully prepared surfaces. Fluorescence spectroscopy and microscopy were key characterization tools used to study both kinetic and equilibrium properties of native protein adsorption in these systems [1^3]. A particularly important aspect of these early studies was the parallel development of total internal re£ection £uorescence (TIRF) spectroscopy as a tool for the direct interrogation of protein surface interactions [4,5].The development of TIRF methods was especially important because it allowed the interface between two disparate materials, e.g., the solid ^ liquid interface, to be investigated while rejecting the background arising from either of the two surrounding bulk phases. Furthermore, it is useful in the study of adsorbed proteins at biointerfaces, because a signal can be obtained which is quantitatively related to the number density of probed protein species at the interface. This quantitative capability can be used to monitor the amount and kinetics of adsorbed proteins, their conformational and orientation states, and how the interaction of an adsorbed protein with a solution-borne ligand a¡ects the £uorescence intensity. The advances in characterization capability represented by the TIRF technique was one development which spurred additional advances in the sophistication of biomolecular assembly methods, for example, the development of wettability gradient surfaces as a mechanism to display di¡erential protein surface interactions spatially [6,7]. 2.2

Langmuir-Blodgett Deposition

Given the interest in preparing well-characterized interfaces for studies of protein and peptide adsorption, it is natural that the Langmuir-Blodgett (LB) techniques were the ¢rst of the directed assembly techniques to be exploited.The LB process inherently a¡ords two-dimensional organization, i.e., polar order, at the air ^ water interface,which can be readily exploited for the preparation of organized biomolecular assemblies. LB techniques can be powerfully ampli¢ed by using them to position and orient molecular recognition moieties, and a great deal of early work focused on the avidin ^ biotin and streptavidin ^ biotin systems. Both avidin and streptavidin are strong biotin-speci¢c binding proteins. Streptavidin is composed of four subunits arranged such that two binding pockets each are arrayed on opposite sides of the protein, each subunit capable of binding one biotin moiety with remarkable speci¢city [8]. Furthermore, it was discovered that interaction with biotin-containing monolayers can facilitate two-dimensional

168

Bohn

crystallization of streptavidin at the air ^ water interface. The attainment of two-dimensional crystalline order at an interface was a key step, because for the ¢rst time it allowed translational symmetry to be attained in biomolecular assemblies,thereby opening these materials to the powerful techniques capable of characterizing crystalline order (Scheme 1) [9^11]. LB techniques are also advantageous, because the surface pressure at which biomolecular recognition events are studied at the air ^ water interface, or at which biomolecular ¢lms are transferred to a solid substrate, go a long way toward determining the density and molecular organization with which the active biomolecules are presented. Thus, the physical state of the lipid matrix elements are coupled to the binding characteristic of the receptors which are incorporated in the bio¢lms. LB techniques also allow for two-dimensional ordering and presentation of intermediate reactive species. An early and elegant demonstration of this principal was obtained by Samuelson and coworkers in the their development of a two-dimensionally ordered photodynamic protein system based on speci¢c molecular recognition utilizing either biotin ^ avidin or biotin ^ streptavidin [10].

SCHEME 1 Schematic representation of a biotin^streptavidin two-dimensional crystal at the air^water interface. The tetrameric streptavidin protein layer is indicated as partially filled with biotin molecules, which are terminated either with reactive moieties, X, or lipid tails. (Reprinted and adapted with permission from Ref. 9. Copyright 1993 American Association for the Advancement of Science.)

Peptide and Protein-Based Biomolecular Assemblies

169

The immunoglobulins constitute another attractive system for exploiting LB technology. Dubrovsky and coworkers exploited the immunological reaction mediated by immunoglobulin G (IgG) organized at the air ^ water interface by binding it to the protein A [12]. They found that protein A organized the molecular assembly by binding only to the Fc arm of IgG molecules, increasing the number of active antibody elements which had their active components in sterically unhindered positions for solution binding, and thereby enhancing the density of ligand binding after transfer of the LB layer to a solid surface. Increases in binding between 10- and 100-fold of rabbit-derived anti-mouse IgG toward mouse IgG were found relative to the nonspeci¢c case. 2.3

Self-Assembly

The development of self-assembly was a truly revolutionary development in the capabilities of researchers to prepare interfacial assemblies with desired properties. Initial e¡orts focused largely on preparation of derivatized silicon oxide surfaces using silane-based nucleophilic substitution chemistry [13^17]. While a number of innovative self-assembly schemes were implemented based on this chemistry, the introduction of self-assembly based on Au ^ thiol chemistry by Whitesides, Nuzzo, and coworkers [18^21] had a much broader impact on the assembly of biomolecular architectures generally and protein-based architectures in particular. There are a number of reasons for the wide implementation of the Au ^ thiol chemistry. Au ¢lms can be prepared readily with a predominant (111) surface orientation by simple thermal evaporation, and the crystallography can be further re¢ned by postdeposition annealing, meaning that large terraces with well-de¢ned crystallographic properties are accessible with modest equipment investment. Au, as a coinage metal, is particularly resistant to chemical degradation of the surface, meaning that oxidation, which is a problem for transition metals generally, and more of a problem for Ag and Cu, is less an issue in Au chemistry.The ability to elaborate the o end of mercaptans synthetically gives rise to a very large number of potential reactive moieties which can be investigated and greatly enhances the power of the Au-thiol self-assembly process for producing biomolecular assemblies. Furthermore, Au can also be electrolessly deposited, so that it is possible to prepare Au coatings on nonplanar surfaces and even surfaces to which a line-of-sight path is not available. One ¢nal advantage of the Au ^ thiol system derives from the ready application of a number of widely used characterization tools, e.g., electrochemistry, surface plasmon resonances (SPRs), and external re£ection infrared absorption measurements, to characterize the assemblies on Au surfaces and thin Au ¢lms [22^26].

170

2.4

Bohn

Multilayer Assemblies

The real power of both LB and self-assembly techniques lies in the ability to place reactive moieties in solvent-accessible regions of molecules where they are available for subsequent reaction with species in the adjacent phase. By placing the reactive moiety at a particular location in the molecule and presenting the molecule to the interface in a well-de¢ned geometry, it is possible to direct reactions spatially to occur at a well-de¢ned location. The reactions themselves may be either thermal or, in a particularly powerful approach, photochemical. An exemplar of the photochemical approach is contained in work by Michel and coworkers [27]. They adsorbed an ofunctionalized dialkyl disul¢de on Au and in two steps converted this to a photoactive benzophenone derivative with good overall yield. Subsequently, antibodies were photoimmobilized on the surface in regions which were irradiated, resulting in a covalently linked antibody as determined by a number of di¡erent assays (cf. Scheme 2) The obvious advantage of introducing photoreactive moieties is the ability to exploit optical lithography to produce laterally patterned structures with micrometer and submicrometer spatial extent. Two other methods of multilayer assembly, which are related to but su⁄ciently distinct from self-assembly to merit mention on their own, are molecular grafting and electrostatic self-assembly. Molecular grafting starts with an appropriate surface-bound monomer, and by controlling chemistry repeat units may be added one at a time, thereby successively growing a polymer from the surface up. Directionally aligned helical peptides on Au have been obtained in this manner by directly grafting and growing an oligopeptide up from the surface [28]. Starting with a trithiolamine, N-carboxyanhydride coupling was used to add sequential layers of alanine to build up a poly(alanine) helical peptide. The great power of this technique is that molecular control is exerted on a peptide-by-peptide basis on the structure of the polymer, thereby acquiring control over primary structure and in many cases over secondary structure as well. Electrostatic selfassembly is a relatively new development in which polyelectrolytes of a de¢ned charge are added, layer by layer, in an alternating charge fashion, in order to create multilayer structures with de¢ned geometric characteristics. A particularly elegant use of this strategy has been demonstrated by Corn and coworkers using positively charged poly(lysine) in order to direct the assembly of subsequent negatively charged DNA oligomers [29,30]. The advantage accruing to the electrostatic self-assembly process is that it can be applied to almost any surface. It does not require a given starting surface, such as Au or SiO2, in order to build up sophisticated multilayer assemblies.

SCHEME 2 Schematic representation of the preparation of a photoreactive monolayer for IgG immobilization. (I) Assembly of the activated ester. (II) Reaction with the diamine. (III) Reaction with the isothiocyanate derivative of benzophenone. (IV) UV immobilization of the IgG. (Adapted from and reprinted with permission from Ref. 27. Copyright 1996 American Chemical Society.)

Peptide and Protein-Based Biomolecular Assemblies 171

172

3 3.1

Bohn

PHYSICAL CHARACTERIZATION Microscopy

Optical microscopy is the oldest of the techniques used to bring structural detail out of biomolecular assemblies. Yet it has proven to be an evergreen, especiallyasitis applied tostructuresinwhich lateralchangesinchemistryare used to e¡ect di¡erential organization of biomolecular assemblies [31^33],or situations in which the behavior of an external agent such as biological cell is of interest [34^38]. Especially important in this regard is the coupling of the basic microscopy experiment with a large number of image-enhancement methodologies and data treatments, for example, £uorescence microscopy, confocal microscopy, di¡erential interference contrast work, and the powerful array of image processing tools which have become available with ready access to imaging detectors. The power of simple optical microscopy to provide information on surface chemistry, as it can be used to direct biomolecular assembly, is illustrated by work performed by Friedman and collaborators [33]. These investigators were interested in identifying solution conditions which would lead to the growth of large optically accessible crystals of proteins such as lysozyme, a-lactalbumin, ribonuclease, hemoglobin, thaumatin, and catalase. Optical microscopy easily allowed areal density of crystal growth and crystal morphology to be characterized, so the development of certain crystalline motifs could be correlated with the solution conditions used in their preparation. In another illustrative example, Keana and coworkers investigated the photochemical derivatization of polymers to produce micrometer-scale structures decorated with biomolecules using deep UV-lithography of substituted per£uorophenyl azides.They used £uorescence microscopy of labeled horseradish peroxidase, bound selectively in areas of a micrometer-scale test pattern, to characterize the contrast obtained by their derivatization procedures and were able to correlate the £uorescence obtained from their images with spectrophotometric measures of enzyme activity for the bound species [31]. While optical microscopy has proven to be a valuable adjunct to other chemical and physical tools for characterizing simple surface chemistries, it is absolutely indispensable for studies of large supermolecular structures, such as biological cells, on surfaces. There has been a great deal of activity aimed at understanding the adhesion and spreading properties of various types of cells on model surfaces as a means to advance the design of biomaterials for implantable prostheses. Whitesides and co-workers studied the wetting and protein adsorptive behavior of alkanethiolate self-assembled monolayers (SAMs) on Au ¢lms thin enough that optical microscopy could be done in transmission [32]. They successfully obtained phase-contrast micrographs of MG63 osteosarcoma cells on a patterned SAM and were able

Peptide and Protein-Based Biomolecular Assemblies

173

to demonstrate dramatic di¡erences in attachment between hydroxylterminated SAMs and adjacent methyl-terminated SAMs. Ratner, Allara, and their co-workers studied the relationship between protein adsorption and endothelial cell growth on o-functionalized alkanethiolate SAMs on Au [35].Using optical microscopy as the primary tool for evaluating cell density and cell growth as a function of time and conditions, they were able to show that cells seated on a substrate in serum-containing medium interact with an adsorbed protein layer rather than with the substrate on which the protein rests, and they were able to make general conclusions about the rates of cell growth and their dependence on the terminal functionality of the alkanethiolate SAM. These simple experiments demonstrated conclusively that cell growth on tissue culture polystyrene outstripped growth rates on all of the monofunctional alkanethiolate SAMs studied, thereby implicating multiple chemical functionalities in the promotion of cell growth and spreading. Beyer and co-workers studied the derivatization of poly[(1-methylvinylisocyanate)-alt-(maleic anhydride)] with a variety of bioactive side chains, including a 22-amino acid peptide derived from the a chain of mouse laminin and an aminopropyl poly(ethylene glycol) monomethyl ether, as supports for hippocampal cell cultures [37]. Using both phase-contrast and £uorescence microscopy, they characterized the development of neural processes and showed that neural cell adhesion requires the presence of the polypeptide derived from laminin. Cell ^ semiconductor hybrids constitute an exciting new area of investigation in which one seeks to establish electrical communication between a biological cell, typically a neuron, and a prefabricated semiconductor device on which the cell is placed and grows.Gaub and coworkers examined hybrids of normal rat kidney ¢broblasts on semiconductors as a model system to replace heart muscle cells or nerve cells [36]. Fibroblasts have the advantage that they may be easily maintained in cell culture, eliminating the need to work from primary culture. These investigators used re£ection interference contrast microscopy and comparisons to atomic force microscopy (AFM) to characterize the behavior of heart muscle cells and normal rat kidney ¢broblasts side by side. Re£ection interference contrast microscopy is particularly powerful in these applications because the micrographs show regions of close contact, allowing investigators to identify focal contact regions and compare them with the global cell morphology. Optical microscopy of biomolecular assemblies and biological cells is particularly powerful when coupled with £uorescence derived from labeled antibodies to structural proteins characteristic of speci¢c elements of cytoskeletal organization. Tresco and coworkers studied the relationship among cell attachment, spreading, cytoskeletal organization, and migration rate for MC3T3-EL oesteoblasts on a variety of glass surfaces di¡ering in the

174

Bohn

nature of the terminal functional moieties of the organosilane SAMs [38]. It was especially important to be able to observe di¡erences in actin and vinculin organization of these 3T3 oesteoblasts in delineating di¡erences between cell attachment and spreading behavior on thiol and oxidized-thiol surfaces versus those surfaces terminated with quaternary amines and methyl groups. The results clearly illustrated the importance of the speci¢c chemistry of the surface rather than generic parameters, such as wettability, in determining cellular behavior. 3.2

Optical Properties

3.2.1 Ellipsometry Re£ection of electromagnetic radiation from smooth interfaces invariably alters the polarization properties of the radiation and provides a powerful mechanism to characterize the optical frequency dielectric response of the interface and the surrounding media. The amplitude re£ectivity, r, can be written as jrp j iðyp ys Þ e ¼ tan ceiD ð1Þ r¼ jrs j where p and s refer to the polarization state of the radiation, y is the phase shift upon re£ection, and r is the amplitude re£ectivity. The right-hand side of the above equation is the formal de¢nition of amplitude re£ectivity in terms of the ellipsometric parameters D and c. Single-wavelength null ellipsometry is the most common variant of the ellipsometric experiment and is performed by varying the general state of elliptically polarized light in the incident beam, until the re£ection from the surface produces a linearly polarized output which is then characterized. Once the ellipsometric parameters of the system have been determined, a generalized version of the ellipsometry equation is combined with the Fresnel relations for multilayer interface systems, and the entire system of equations is inverted to obtain the dielectric response and thicknesses of the relevant layers of the biomolecular assembly. Clearly, the complexity of this inversion process depends directly on the number of layers in the sample and the number of layers in the system for which the optical properties are known prior to the experiment. For reasons that will be apparent below, the applications of ellipsometry have been far outstripped by measurements of surface plasmon resonances. Nevertheless, ellipsometry has been used in a number of systems to characterize biomolecular assemblies. Lu and coworkers examined the oriented immobilization of IgG Fab0 fragments on SiO2 surfaces using ellipsometry to determine the surface concentration of the immobilized fragments [39]. Both oriented and randomly coupled IgG fragments were

Peptide and Protein-Based Biomolecular Assemblies

175

covalently bound to derivatized silica surfaces containing pyridyl disul¢de groups. By comparing the total amount of immobilized immunoglobulin fragment to the active component determined from immunoassays, the authors were able to demonstrate large di¡erences in antigen-binding activity for fragments which are oriented with their active surfaces available to solution, thereby emphasizing the power of oriented molecular assemblies. Suci and coworkers demonstrated the utility of ellipsometry for characterizing sophisticated multilayer structures in their examination of biotinylated lipid ^ streptavidin Fab multilayer construction [40]. Ellipsometry was used to characterize the uptake of a biotinylated Fab fragment onto the preorganized streptavidin template by monitoring changes in the ellipsometric parameters ddD and ddc as a function of the active component in the solution phase. Comparison of the optical parameters derived from ellipsometric measurements indicated a dense attachment of the biotinylated Fab fragment to the surface. One important advantage retained by ellipsometry over SPR measurements is that it may be applied to a large number of substrate materials, whereas SPR measurements are largely con¢ned to systems fabricated on the coinage metals Au, Ag, and Cu. However, high accuracy is achieved only when working near the Brewster angle. Given the fact that optimal performance is obtained near a speci¢c angle, ellipsometric imaging can only be achieved through the aegis of imagining detectors. Recently, Sackmann and coworkers developed a high-resolution imaging microellipsometer which promises to advance the state of characterization of biofunctionalized surfaces, cf. Fig. 1 [41]. In their apparatus, the re£ected light from a collimated source is collected with a long-distance working objective and passed through a rotating analyzer before being projected onto the active area of a CCD camera by a tube lens. They were able to achieve lateral resolution of 3 mm and a height resolution of 5 — working at angles far from the Brewster angle and imaging areas of the order of a few tens to hundreds of square micrometers. The performance achieved in these studies promises to be a helpful adjunct to studies in which the biomolecular assembly of interest cannot be formed on a substrate suitable for SPR imaging. 3.2.2 Surface Plasmon Resonances (SPRs) The collective electron resonances which are pinned at the interface between metals and dielectrics can be e⁄ciently excited for a select group of metals, namely, Ag, Cu, and Au. These resonances, termed surface plasmon resonances, derive from the resonance conditions in the bulk dielectric functions of the two materials, and are contained implicitly in the Fresnel relations describing the optical response of a multilayer stack containing the

176

Bohn

FIG. 1 (a) Schematic diagram of the microimaging ellipsometer. (b) Ellipsometric images of n-C16H34 dewetting transition on thermally oxidized Si at 5 min (left) and 8 min (right) after deposition. (Adapted from Ref. 41. Copyright 1998 American Institute of Physics.)

Peptide and Protein-Based Biomolecular Assemblies

177

metal and dielectric of interest. Surface plasmons are useful for the characterization of interfaces, because (1) they provide an enhanced electromagnetic ¢eld amplitude at the interface, (2) surface plasmons decay over a relatively short distance, with electric ¢elds having 1=e distances of a few hundred angstroms in the adjacent dielectric, and (3) they are completely p-polarized, leading to the acquisition of spatial orientation information about the dipoles with which they interact. In the characterization of biomolecular assemblies, surface plasmons (SPs) are typically excited in the Kretschmann geometry shown in Fig. 2.The molecular assembly of interest is usually applied as an adlayer on the SP supporting metal. Since, at a given frequency, the wave vector of a photon instant from vacuum is smaller than that of the surface plasmon with the same frequency, it is necessary to increase the momentum so that both energy and momentum conservation can be satis¢ed simultaneously. Kretschmann developed a simple and elegant solution to this conservation condition based on placing the SP supporting metal ¢lm on the base of a

FIG. 2 Schematic diagram of the sample configuration in a typical SPR experiment used to monitor protein binding events at a solid^liquid interface. The specific system depicted is meant to convey streptavidin (tetrameric solution phase component) binding to a biotinylated lipid present in the outer portion of a phospholipid double layer assembled on SiOx. (Adapted from Ref. 47. Copyright 1992 Biophysical Society.)

178

Bohn

high-index prism or hemicylinder,the refractive index of the prism serving to increase the wave vector of the incident bulk electromagnetic mode so that its p-component may be matched to the real part of the SP resonance of the same frequency. When the resonance is satis¢ed, excitation is coupled into the collective oscillation of electrons at the interface, thereby frustrating the near-total re£ection which would otherwise apply for the metal ^ dielectric interface. The coupling of energy into the surface plasmon is accompanied by a decrease in the bulk re£ectivity and an increase in Rayleigh scatter. The position of the SP resonance may be calculated from "   # osp em es 1=2 ¼ k0 np sin yp ð2Þ Refksp g ¼ Re c em þ es where subscripts m, s, and p stand for metals, superstrate, and prism, respectively, and k0 is the vacuum wave vector of the instant radiation. An excellent summary of the underlying physics supporting the use of surface plasmons for characterizing biomolecular assemblies is contained in a recent review [22]. The most prevalent use of surface plasmons has involved characterizing interfacial binding events between immobilized receptors and solutionborne ligands [30,42^49]. Ligand binding and=or formation of a macromolecular complex with a solution-borne component naturally changes the optical dielectric function of the interfacial layer, which is most commonly characterized by a shift in the position of the SP resonance in momentum space, cf. Fig. 3. Qualitatively, the shift in surface plasmon resonance may be used to monitor the extent of modi¢cation of the interfacial layer. Quantitative measurements, however, are less straightforward. While the position and shape of the SPR resonance depend directly on the optical properties of the interface, inverting the equations in order to obtain the optical dielectric response function, including both refractive index and thickness information, requires rather more e¡ort than is typically devoted to the experiment in studies of biomolecular binding. In experiments where optical property data are obtained, it is common to ¢x either the refractive index or the thickness and to calculate changes in the other parameter to ¢t the shifts in the SPR resonance code. To determine both parameters simultaneously for a single ¢lm requires either measurements at two wavelengths or measurements under di¡erent superstrate refractive index conditions [50,51]. Whitesides and coworkers demonstrated that SPR measurements could be used to monitor nonspeci¢c adsorption of proteins in situ and in real time, by monitoring the deposition of RNase A, lysozyme, ¢brinogen, and pyruvate kinase on a set of mixed SAMs comprised of ethylene

Peptide and Protein-Based Biomolecular Assemblies

179

FIG. 3 SPR reflectivity curves for a bare Au surface (*) and after deposition of sequential adlayers of HS(CH2)10CO2H (~), poly(lysine) ( ), and avidin ( ffl ). (Adapted and reprinted with permission from Ref. 30. Copyright 1995 American Chemical Society.)

oxide- and methyl-terminated alkanethiolates [45]. Their experimental methodology, which is now common, used an array detector to monitor a range of angles around the central portion of the plasmon resonance curve containing the minimum. Then the position of the minimum was ¢tted onthe-£y to determine the shift in optical properties, and this was then plotted in real time to follow the biomolecular events of interest. They and others have used this approach, or closely related variants, to follow a large number of biochemically interesting surface interaction systems, including the binding of carbonic anhydrase to self-assembled monolayers of alkanethiolates terminated with benzenesulfonamide groups on Au, and a number of studies in which surfaces designed to resist protein adsorption are characterized by SPR [42,44]. Corn and coworkers demonstrated how powerful it is to use SPR in conjunction with in situ Fourier transform-infrared spectroscopy for the characterization of multilayer assemblies. They exploited the electrostatic multilayer assembly technique to develop structures which present DNA oligomers for biosensing and DNA computing applications [30,52]. In these studies, an ionized carboxylic acid-terminated alkanethiolate was used as a template on which poly(lysine), a polycationic electrolyte, was adsorbed simultaneously with a positively charged avidin layer. The

180

Bohn

avidin then provided a template for the adsorption of biotinylated oligonucleotides. One of the most powerful factors in explaining the popularity of SPR measurements is the fact that thin Au ¢lms can also serve as substrates for in-situ electrochemistry and for external re£ection IR measurements, as recently demonstrated by work from Mrksich’s laboratory [49]. They assembled mixed alkanethiolate SAMs in which a fraction of the terminal functional groups were composed of hydroquinone moieties which could undergo reversible electron transfer to shuttle them between dihydroxy and quinoid-like forms, the quinoid-like form subsequently being reactive in a Diels-Alder addition with a biotinylated cyclopentadiene. The ability to follow the appearance of the quinoid-like form electrochemically, and to monitor the subsequent immobilization of streptavidin, through interaction with the biotinylated reaction product was a powerful adjunct in being able to work out the biomolecular assembly in this instance. It must also be said that much of the popularity of SPR measurement to monitor biomolecular binding is due to the advent of commercial instrumentation. The most prevalent commercial instrument is based on work done at Pharmacia in which hydrogel matrices are formed from carboxymethylated dextran to provide a matrix on the Au surface which can mediate reversible biochemical binding. Although the measurement principles are just as described above, the presence of the hydrogel matrix transforms the measurement. It impedes the direct adsorption of proteins at the interface, thereby avoiding problems with protein denaturation, and it greatly increases the mass sensitivity by increasing the volume sensed from a little more than a molecular monolayer to a volume extending an appreciable portion of the 1=e decay length of the surface plasmon ¢eld [53]. Relatively early in the use of surface plasmons as a tool for investigating biomolecular interfaces, it became clear that a simple change in the optics could transform the SPR measurement from a spatially integrating measurement, to one in which microscopic image information could be obtained, cf. Fig. 4. Gaub and coworkers exploited this capability and used SPR imaging to monitor the assembly of Fab0 fragments of a monoclonal antibody raised against a dinitrophenol hapten covalently bound to a phospholipid [54]. Dinitrophenol-labeled bovine serum albumin (BSA) was used as an antigen and monitored as it bound to the Fab0 fragment of the derivatized phospholipid. Step heights between protein-rich and surrounding phospholipid matrix domains were measured, and the data were found to agree well with the dimensions of crystallized Fab0 fragments. The spatial resolution in SP microscopy is limited not by the optical di¡raction limit, but by the spatial decay length of the SP resonances themselves. The

Peptide and Protein-Based Biomolecular Assemblies

181

FIG. 4 Schematic diagram of the surface plasmon resonance microscope. Note the incident radiation is well collimated, and the image is processed by the collection lens (a microscope objective) onto an imaging detector of high spatial fidelity. (Adapted and reprinted with permission from Ref. 54. Copyright 1993 American Chemical Society.)

SP resonance is a propagating resonance with a characteristic in-plane decay length typically from a few micrometers up to many tens of micrometers depending on the optical parameters of the interface. Because the spatial resolution in an experiment in which the SP resonance is monitored cannot be any better than its propagation length, the spatial resolution in SP microscopy is typically degraded relative to that of phase-contrast optical microscopy. Nevertheless, the unique nature of the information which is obtained from the SP experiment, as well as its ability to monitor interfacial biochemical events, maintains high interest in this and continued e¡orts to develop microscopy exploiting the surface plasmon phenomenon [55,56]. A somewhat more challenging application of SPR involves using the shifts produced in the SP resonance to follow conformational or orientational changes in biomolecular components, because the resulting changes in optical properties are typically much smaller than for corresponding addition of a separate ligand or binding partner. Nevertheless, a number of groups have successfully exploited SPR to do just this. The simplest approach to using surface plasmons to monitor molecular orientation uses a subsequent recognition event to detect that orientation. In work from Gerisch’s laboratory a biosensor was modi¢ed to monitor the contact site A

182

Bohn

glycoprotein from the aggregating cells of Dictyostelium discoideum [57]. The protein was ¢xed in a known orientation by cross-linking with per£uorophenylazide, and the surfaces thus obtained were challenged with antibodies raised against either the native or denatured glycoprotein. Here the orientation was detected by the presence or absence of the immunocomplex. Direct determination of protein conformation was obtained by Sota and co-workers, who monitored changes in immobilized dihydrofolate reductase derived from Escherichia coli and attached to a carboxymethyl dextran matrix [58]. Exposure of the immobilized protein to an acidic environment resulted in changes in the SPR resonance position which corresponded to similar pH-induced changes in the molar ellipticity of the protein in homogeneous solution. Dufour and coworkers studied the use of SPR measurements and epitope mapping to monitor protein conformation as displayed through binding of ¢ve di¡erent anti-b-lactoglobulin monoclonal antibodies. Binding of the monoclonal antibodies was used as evidence of the presence of the native epitopes accruing from the original b-lactoglobulin [59]. Given the great success of SPR measurements in addressing a wide range of biomolecular assemblies, it is not surprising that SPR resonance measurements have made their way into a number of speci¢cally targeted biosensing platforms [43,60^62]. Bergener and coworkers developed an SPR-based biosensor to study the hybridization characteristics of peptide nucleic acid ligands, immobilized on sensor surfaces [60]. They found that covalently immobilized sensing molecules exhibited superior performance to those simply physically adsorbed, and they demonstrated that a perfectly matched peptide nucleic acid allowed the detection of singlestrand DNA at a sensitivity better than 1% of the background in a singlestranded DNA having a single C-to-T point mutation in the complement to the peptide nucleic acid. Kooyman and coworkers demonstrated another advantageous feature of SP resonances in their application to biosensing by showing how a parallel array of sensing modalities could be constructed in a single monolithic device [61]. Finally, a very interesting development in the use of SPR for biosensing by Natan and coworkers involved enhancement of SPR sensitivity by coupling the ligand to colloidal Au particles to create an antigen ^ Au colloid conjugate which, when arrayed near the original Au ^ dielectric interface, caused a larger shift in SP resonance than would have been observed by simple molecular binding alone. This enhancement in the shift and the change in the shape of the SP resonance is general and potentially quite powerful, since a great deal of chemistry exists, primarily derived from the electron microscopy literature, for speci¢cally coupling a wide variety of biomolecules to Au colloids [62].

Peptide and Protein-Based Biomolecular Assemblies

3.3

183

Mechanical Properties

3.3.1 Quartz Crystal Microbalance (QCM) Measurements Another exceptionally sensitive and simple approach to monitoring changes in the composition of interfacial biomolecular layers is obtained through the piezoelectric e¡ect, in which an electric ¢eld applied across a piezoelectric material induces a stress in that material. When driven at the vibrational resonance frequency, the piezoelectric material oscillates, and any change in the mass of the surface of the oscillator results in a change in the vibrational frequency, as has been known since the pioneering work of Sauerbrey in the 1950s [63]. A popular piezoelectric material is ATcut quartz, and the ability to deposit thin gold layers on the surface of the quartz without signi¢cantly disrupting its piezoelectric properties means that all of the chemical and biochemical advantages which accrue to the use of AU ¢lms in SPR measurements also apply to QCM measurements [64]. The key ¢gures of merit in judging the QCM measurement are sensitivity to mass adsorption and ability to distinguish authentic mass-induced frequency shifts from non-signal-related sources of frequency shift. Mass sensitivity is easily capable of achieving submonolayer detection limits for typical biomolecular binding events. In cases where direct binding is not adequate to achieve frequency shift of the magnitude needed, sandwich assays can be implemented in which a secondary species is generated and used to accentuate the frequency shift, cf. Fig. 5.The inability to discriminate unambiguously between mass-induced changes in resonance frequency and frequency shifts due to environmentally induced changes in viscoelastic properties of materials adsorbed to the surface is a distinct disadvantage to the QCM measurement. However, frequently in studies of biomolecular assemblies, the changes in the mechanical response of the adsorbed layers is minor compared to changes in mass, so this disadvantage is not as troublesome as it might be. Many of the same biomolecular motifs used in studies performed with SPR are used in QCM measurements, i.e., immunochemically linked molecules and molecules linked to surfaces via biotin ^ avidin or biotin ^streptavidin chemistry. For example, Hu and coworkers demonstrated a highly sensitive immunosensor for detection of human complement C4, based on immobilization of an anti-C4 antibody on the surface of a QCM [65]. Protein A was bound to the surface to immobilize the antibody in an oriented fashion, giving results superior to the nonspeci¢c binding of IgG obtained via gluteraldehyde coupling or by physical adsorption. Nicolini et al. also noted the importance of orientation in studying the immunological activity of immunoglobulin Langmuir ¢lms oriented by a protein A sublayer [12]. Kunitake and coworkers used QCM measurements to follow the

184

Bohn

FIG. 5 Ability of the sandwich-type assay to enhance the frequency shift associated with the biomolecular binding event of interest. In this assay monitoring the binding of adenosine phosphosulfate reductase (APS reductase) to its rabbitderived antibody, the sandwich is formed with an enzyme^antibody conjugate which catalyzes the dimerization of 5-bromo-4-chloroindolylphosphate (BICP). (Adapted and reprinted from Ref. 64. Copyright 1990 American Association for the Advancement of Science.)

Peptide and Protein-Based Biomolecular Assemblies

185

layer-by-layer assembly of alternating polycationic and polyanionic electrolytes in the electrostatic assembly process employing poly(ethyleneimine) and poly(styrene sulfonate) [66]. They used electrostatically assembled SAMs to support the adsorption of charged proteins, including cytochrome c, myoglobin, lysozyme, histone F3, glucoamylase, and glucose oxidase. More sophisticated biomolecular motifs have been implemented and monitored by QCM measurements byAizawa and coworkers. They have, for example, made extensive use of genetic engineering to produce fusion proteins in which one component is designed for binding to a surface and the other component for biomolecular binding to a target substrate. In one set of experiments, the B domain of staphylococcal protein A was repeated ¢ve times and terminated with a cysteine residue to facilitate self-assembly on Au. The B domain is a speci¢c IgG-binding domain and was used to immobilize antibody molecules in a highly oriented fashion [67,68].These workers also enhanced the sensitivity of QCM mass changes by binding the receptor, not to a single antibody molecule, but to a lipid-derived antibody which had been inserted into the outer lea£et of a large vesicle, cf. Fig. 6, thereby greatly amplifying the mass change. In structures in which the bacterially produced, lipid-tagged single-chain antibody against 2-phenyloxazolone was incorporated in phosphotidylcholine liposomes, the presence of soluble hapten could inhibit immunoliposome binding in a concentration-dependent manner and resulted in a competitive assay in which antigen was measured in the concentration range down to 108M [69]. Similar strategies were employed in studies of a polyfunctional arti¢cial proteins derived from calmodulin and glutathione-s-transferase. Glutathione was ¢rst self-assembled on Au and used to template the binding of calmodulin through ligand interaction with its partner glutathione-s-transferase. The fusion protein was reversibly adsorbed and desorbed by the binding of solution-borne glutathione in a competitive binding assay [70]. In a separate study, a related fusion protein of calmodulin was used directly by introducing a cysteine residue at the terminus, which could then mediate self-assembly at a Au surface. The authors demonstrated that the engineered calmodulin bound to the surface retained its Ca2þ -responsive modulation of enzyme function [71]. The fact that SPR measurements and QCM measurements both provide physical insight into the addition of molecules to Au ^ dielectric interfaces begs the question of how they compare with respect to critical measurement ¢gures of merit. The question has been addressed from a practical perspective by Sackmann and co-workers, who compared QCM and SPR £ow injection analysis systems for observing nonspeci¢c binding of BSA at Au surfaces and speci¢c binding of BSA to monoclonal antibodies immobilized on Au [72]. They concluded that neither measurement o¡ers a distinct advantage over the other in terms of sensitivity or background

186

Bohn

FIG. 6 Lipid-tagged antibody strategy for enhancing the response of a QCM to a selected binding event. (a, top) Genetic construction of the lipid-tagged antibody. (a, middle) Schematic diagram of the lipid-tagged antibody. (a, bottom) Schematic representation of the competitive binding of free hapten (Ox-CA) and lipid-tagged antibody bearing liposome. (b) Typical QCM response upon being challenged with various solutions. Note the disparity between signal magnitudes of the liposome exposure and exposure to molecular constituents. (Adapted and reprinted with permission from Ref. 69. Copyright 1998 American Chemical Society.)

Peptide and Protein-Based Biomolecular Assemblies

187

FIG. 6 Continued.

rejection. Johannsmann and co-workers addressed the question from a more theoretical perspective and made several interesting observations [73]. First, they noted that the contrast in acoustics is usually far larger than it is in optics, owing to the large di¡erences in shear moduli among di¡erent materials, while refractive indices generally change by only a few percent. This naturally mitigates in the favor of QCM measurements. However, the acoustic thickness approaches the geometric thickness at rather low surface coverages and tends not to increase, as the ¢lm is densi¢ed by prolonged exposure to solution. These authors compared the two measurement principles in the adsorption of streptavidin on self-assembled monolayers of a biotinylated alkanethiol. Although the sensitivities to adsorption are similar, the authors noted an interesting di¡erence in the kinetics of assembly, presumably owing to the geometric thickness saturation e¡ect in QCM measurements, cf. Fig. 7. 3.3.2 Scanned Probe Microscopies Unquestionably, the invention of scanned probe microscopies (SPMs)
primarily scanning tunneling microscopy (STM) and atomic force

188

Bohn

FIG. 7 Comparison of SPR (a) and QCM (b) measurements of the assembly of streptavidin onto a biotinylated SAM. Panel compares the assembly kinetics as measured through the optical and acoustic thicknesses. (Adapted from Ref. 73. Copyright 2000 American Institute of Physics.)

microscopy (AFM) and then later a whole set of derivative techniques based on these two basic principles
has revolutionized the manner in which interfacial chemistry, physics, and biology is explored. Proteins and peptides immobilized at interfaces tend to develop structure on mesoscopic length scales which cannot be probed by optical microscopy, because the structural details are below the di¡raction limit and are therefore not amenable to direct optical imaging whether through phase-contrast optical microscopy, surface plasmon microscopy, or ellipsometric microscopy. To characterize structure on the sub-micrometer length scale, a wholly new approach is

Peptide and Protein-Based Biomolecular Assemblies

189

required, an approach which is well ¢lled by the scanned probe microscopies. The general principles of scanned probe microscopies are well known, and an excellent recent review exists [74]. The vast majority of SPM studies to characterize biomolecular assemblies make use of AFM or some variant of it. This is understandable, inasmuch as the tunneling requirement of STM limits the ¢lm thicknesses of structures to be studied to 3 nm at most, far smaller than the thickness of typical biological assemblies. Nevertheless, in some limited applications STM can be useful, as demonstrated by the study of the assembly of l-cysteine on Au (111) by Kolb and co-workers [75]. Among the force microscopy e¡orts, two general themes emerge. In one, scanned force microscopies, AFM, tapping mode AFM, etc., are used to make topographic, or force map, images of surfaces and how those surfaces evolve upon exposure to biological agents. In the other set of measurements, the forces which exist between derivatized tips and biologically decorated surfaces are measured as a function of the tip-to-surface separation. A number of instructive AFM-based studies of protein assemblies are available in the literature. Patel and coworkers studied the binding of catalase to Au surfaces modi¢ed by SAMs of acid-terminated alkanethiols of varying chain lengths, in pure ¢lms and in ¢lms with mixtures of chain lengths [76]. They determined that the surfaces with mixed composition facilitated adsorption of catalase relative to pure carboxylate-terminated layers of either chain length. Work in Knobler’s laboratory examined how phase-separated two-component self-assembled organosilane monolayers could direct the selective adsorption of BSA [77]. They prepared twocomponent Langmuir ¢lms composed of octadecyltrichlorosilane and 1H,1H,2H,2H-per£uorodecyltrichlorosilane and showed that the BSA was preferentially adsorbed as aggregates on the CH3 -terminated regions of the patterned monolayers, cf. Fig. 8. AFM also has the advantage that it is applicable to a wider range of underlying substrates, as demonstrated by the study performed in Cullen’s laboratory examining time-resolved adsorption of IgG and glucose oxidase to highly oriented pyrolytic graphite [78]. Aizawa and coworkers also used HOPG in a study of an antibody-binding protein, which they termed (E12B2)n , fabricated to contain a hydrophobic peptide, E12, at one terminus and an antibody-binding peptide, B2, at the other [79]. AFM imaging was used to demonstrate e⁄cient self-assembly on a hydrophobic solid surface, i.e., HOPG. An issue of some concern in the study of soft materials, such as the biologically active assemblies of interest here, is the destruction of the assemblies by the force-sensing cantilevers when they are applied at forces beyond the yield limit of the monolayers being studied. When this is a concern, derivatives of the standard AFM techniques, such as tapping-mode AFM, can be invaluable. They were used

FIG. 8 Topographic AFM images of BSA adsorbed on mixed LB layers of OTS and FTS. Films in (a) and (b) were transferred at 8 mN=m but were compressed at 2 mm=min and 5 mm=min, respectively. (c) Higher-resolution image shows the height of the adsorbed BSA is relatively monodisperse at 25 3 nm. (Adapted and reprinted with permission from Ref. 77. Copyright 1996 American Chemical Society.)

190 Bohn

Peptide and Protein-Based Biomolecular Assemblies

191

byViitala and coworkers in the study of the attachment of Fab0 fragments of a polyclonal anti-human IgG molecule to a polymerizable lipid with the linker group in the terminal position [80]. The linker, attached to a lipid, was imbedded in a monolayer of dilinoleoylphosphatidylethanolamine. AFM conclusively demonstrated that when the monomeric monolayer was used, large apertures were opened in the as-deposited monolayer, thereby completely obviating its usefulness, cf. Fig. 9. In contrast, cross-linking the monomer to produce a two-dimensional sheetlike polymer resulted in planar ¢lms with good structural integrity. Even more information can be obtained by characterizing the distance dependence of the forces generated when a chemically derivatized tip comes near a surface presenting the biological molecule of interest. Mueller and coworkers used force measurements between myelin basic protein (MBP) adsorbed on mica and lipid bilayer surfaces to demonstrate that the presence of myelin basic protein always led to an attractive force between tip and sample [81]. When retracting the tip, MBP molecules were found to exhibit elastic stretching behavior consistent with a wormlike chain model, yielding speci¢c values for persistence lengths and average contour lengths. While attached to a lipid bilayer, MBP did not show elastic stretching behavior. These observations were taken to indicate that MBP adopts a di¡erent conformation when in contact with lipids and that the lipid bilayer structure is strong modi¢ed by MBP attachment. Grunze and coworkers used functionalized scanning force microscope tips to investigate the resistance to protein adsorption of oligo(ethylene oxide)-terminated SAMs [82]. They

FIG. 9 A comparison of tapping-mode AFM images of monomeric (left) and cross-linked (right) linker phospholipid assemblies. (Adapted in part and reprinted with permission from Ref. 80. Copyright 2000 American Chemical Society.)

192

Bohn

observed a long-range hydrophobic attractive interaction for a SAM with three repeat units of ethylene oxide, i.e., (CH2CH2O)3, on Ag, whereas a repulsive force was measured for the same molecule on Au. Furthermore, in the Au system the repulsive force displayed a strong dependence on the intervening solution ionic strength, indicating a signi¢cant electrostatic contribution. The observed di¡erences in behavior between Ag and Au were attributed to distinct molecular conformations of the ethylene oxide units on Au (helical) and Au (all trans). Finally, Stewart and Hlady integrated a standalone scanning force microscope with a re£ection interference contrast microscope to measure the separation between the SFM probe and the sample surface directly. These measurements were then used to interpret a number of seemingly anomalous distance separation results which had been obtained for binding of immunologically active antigen ^ antibody complexes [83]. 4

CHEMICAL CHARACTERIZATION

As powerful as the above tools are, for monitoring changes in the building up of biomolecular assemblies at interfaces, they all su¡er from one critical £aw
chemical or molecular information must be inferred; it is not developed directly from the signals obtained. Although a very great deal can be learned about biomolecular structures through these techniques, it is evident that experiments providing information about functional groups, prosthetic groups, and speci¢c chemical moieties within a biomolecular assembly will be invaluable as an adjunct to the physical characterization tools discussed above. 4.1

Electronic Structure

4.1.1 Absorption Measurements There are a number of requirements for chemical characterization methods to be used for biomolecular assemblies. First, given the fact that planar interfaces contain a very small number of molecules, high sensitivity is a prerequisite.The experiment must be su⁄ciently unobtrusive that it does not damage or induce change in the biomolecular structure during the course of the experiment. Spectral probes, sensitive to small elements of the biomolecule, are important, because information concerning the disposition of the active site or prosthetic group in peptide and protein assemblies is critical to understanding the mode of interaction with solution-borne species. Finally, since all biological assemblies considered in this chapter are intended for applications in speci¢c biochemical environments, in situ spectral probes are highly desirable.

Peptide and Protein-Based Biomolecular Assemblies

193

For a long time, traditional UV-visible absorption spectroscopy saw only limited application in the study of monolayer assemblies in general, and biological assemblies in particular, due to the small optical absorption cross section associated with typical transitions in the accessible region and the small number of molecules probed in the transmission geometry. Recently, a number of localized excitation strategies have been developed which have greatly enhanced the sensitivity of biomolecular assemblies to optical absorption, thereby placing the direct measurement of optical absorption into the realm of routine characterization probes of these systems. Optical absorption in biomolecular assemblies is particularly powerful when coupled with polarized-light transmission measurements, because the orientation of the electronic transition moment of the probed species can be measured, and thereby provide order parameters for oriented peptide and protein sequences at interfaces. Much of the early work examining monomolecular layers of biologically active molecules relied on prosthetic groups with extremely high inherent molar absorptivity, such as the heme moiety present in the cytochrome metalloproteins [84,85]. Kunitake and coworkers took advantage of the strong absorption of the heme moiety in myoglobin to examine the functional conversion of myoglobin from an oxygen storage protein to a redox enzyme upon exposure to NADH and £avin mononucleotide [86].This work presaged later oriented metalloprotein assembly work in that these investigators cast multibilayer ¢lms which were used to organize the proteins of interest relative to a macroscopic surface which could be used as a geometric ¢ducial plane. Another way in which molecular absorption can be used is demonstrated by the elegant set of experiments conducted by Charych and coworkers, in which they used optical spectroscopic changes in poly (diacetylene) ¢lms as a sensing mechanism for binding of glycolytic-linked moieties such as proteoglycans and glycoproteins [87^89]. They took advantage of the fact that the multivalent character of viral binding at a synthetic membrane, in which the organized supermolecular sensor structure is contained, can give rise to large conformational changes in the polymer side chains, which then disrupt the conjugation in the poly(diacetylene) backbone, shifting the absorption spectrum and resulting in a color change. In recent work, the conformationally induced changes in poly(diacetylene) spectroscopic properties have been used to detect the binding of lectins such as wheat germ agglutinin, concanavalin A, in£uenza virus, and entire E. coli organisms [87]. Planar interfaces can also act as a reference plane for organized and oriented assemblies of metalloproteins. Bohn, Sligar, and their coworkers were the ¢rst to exploit genetic engineering principles to create unique

194

Bohn

reactive sites on the solvent-accessible surface of metalloproteins, in order to link the protein to a reactive site of a self-assembled organosilane monolayer on SiO2 [90^94]. Several di¡erent organosilane-based reactive terminal functional groups were exploited, for reactive coupling to unique cysteine residues, introduced by genetic engineering, into metalloproteins, such as cytochrome b5, cytochrome c, and myoglobin [91,92]. A complete evaluation of all of the factors giving rise to uncertainty in the orientation of myoglobin mutants was performed by characterizing the order of the alkyl linker moieties by Fourier transform-infrared spectroscopy for short, C2 and C3, intermediate, C8, and long, C15, linker chains. The orientation of the terminal carbon ^ carbon bond was monitored by linking the terminal moiety to a dimethylaminopyridine (DMAP) chromophore. This moiety exhibits a linear transition moment pointing along the long molecular axis, so that performing polarized absorption measurements yields the orientation of the transition moment.With these two pieces of information in hand, linear dichroism measurements on the xy-degenerate Soret transition of the heme prosthetic group in myoglobin yielded results which could be compared to model calculations (Fig. 10). These investigators found surface roughness to be a minor contributor in determining overall macromolecular ordering, but the nature of the underlying silane self-assembled coupling layer strongly in£uenced the spatial and functional properties of the chemisorbed protein. Silane layers with short aliphatic chains and long aliphatic chains were found to produce ordered structures, whereas those produced from intermediate-chain-length organosilanes were conformationally disordered. Furthermore, the orientation of myoglobin predicted from the measured orientations of the linkers with the DMAP linear transition moment chromophore agreed exceptionally well with the orientations of the resulting myoglobin SAM, obtained from the orientation of the heme prosthetic group [90]. One of the impediments to studying protein and peptide molecular monolayers and multilayers arises from the sensitivity constraints described above. One obvious solution to the low spectroscopic sensitivity of transmission-based measurements is simply to increase the e¡ective pathlength. Bohn and coworkers were the ¢rst to show how integrated optical waveguide sample geometries could be used to measure absorption properties of adsorbed layers and orientations of molecular transition moments via waveguide linear dichroism measurements [95^97]. This work was later extended in the laboratories of Reichert [98,99] and Saavadra [100,101]. The work of Saavadra and coworkers is particularly interesting in this regard, because they used emission anisotropy obtained from total internal re£ection £uorescence measurements to obtain the distribution of orientations of cytochrome c immobilized on self-assembled monolayers. Their results

FIG. 10 (a, top) Geometric model of a noninteracting chromophore with a transition moment along the long molecular axis. b is the spherical polar angle between the molecular transition moment axis z0 and the surface normal z. (a, bottom) Geometric model of an x^y degenerate oscillator, e.g., the heme prosthetic group. y is the spherical polar angle between the heme normal and the surface normal. (b) Dichroic ratio as a function of the angle of incidence for polarized absorption of the A126C mutant of myoglobin. Solid squares are experimental data, and the solid lines are model calculations for various orientations, y. (Adapted and reprinted with permission from Ref. 90. Copyright
1996 American Chemical Society.)

Peptide and Protein-Based Biomolecular Assemblies 195

196

Bohn

indicated that a macroscopically ordered ¢lm of cyt c could be produced when a single high-a⁄nity but noncovalent binding motif, for example, the electrostatic attraction between the positively charged surface of cyt c and the negatively charged head groups of a Langmuir-Blodgett ¢lm of arachidic acid, was present between the protein surface and the substrate. These techniques are especially powerful, because the same long path-length advantages which are enjoyed by linear dichroism measurements in the integrated optical waveguide geometry can also be exploited in emission anisotropy measurements. Circular dichroism (CD) is a powerful tool for the study of secondary structure of proteins in solution, but relatively little has been done to date to exploit CD measurements of immobilized peptides and proteins at planar interfaces. De Jong and coworkers studied the protein mellitin on phospholipid layers [102]. Using literature reference circular dichroism spectra for the various possible secondary structures they expected to see, they ¢tted the experimental interfacial spectra to determine the orientation of the helical part of mellitin in oriented ¢lms. They found good agreement with the orientations produced from their circular dichroism measurements and those obtained by Fourier transform-IR measurements. Dutton and co-workers used interfacial circular dichroism measurements to study arti¢cial 4-a-helix bundle proteins assembled to accommodate heme prosthetic groups [103]. They reasoned that the orientation of these molecular maquettes at the air ^ water interface could be controlled by choosing the distribution of the charged amino acids used to make up the four helixes. Surprisingly, the heme moiety was found to dictate the orientation of a-helical axes relative to the surface plane
with the parallel orientation obtained in the absence of the heme converting to a perpendicular orientation when heme was present. CD measurements were used to demonstrate that the molecular orientations obtained at the air ^ water interface were faithfully reproduced on solid substrate surfaces. Since CD is so sensitive to the presence and nature of secondary structure in proteins and polypeptides, and since the extent of denaturation in adsorbed proteins is a key question, one can expect greater e¡orts in the future to sidestep the sensitivity issues associated with transmission-based absorption measurements of circular dichroism. 4.1.2 Fluorescence Fluorescence spectroscopy can circumvent some of the sensitivity issues associated with absorption, because the low background associated with the £uorescence measurement means that quite low detection limits can be reached. In characterizing biomolecular ¢lms, £uorescence is used in three principal ways: (1) as a mechanism to follow the adsorption of solution

Peptide and Protein-Based Biomolecular Assemblies

197

components to an interface; (2) with imaging techniques as a mechanism to follow patterning of biomolecular components in two dimensions; and (3) using the £uorescence recovery after photobleaching (FRAP) technique as a mechanism to follow the dynamics of motion of biomolecular constituents in the plane. One of the most powerful modalities for studying the £uorescence properties of interfacially bound biomolecules is the Total Internal Re£ection Fluorescence (TIRF) experiment. In TIRF the evanescent wave generated upon total re£ection at a dielectric ^ dielectric interface is available for excitation of molecular electronic transitions within a decay length of the interface, typically of the order of the wavelength of the excitation radiation, i.e., 200^500 nm. Because the number density of target molecules at the interface is usually much larger than the bulk concentration in solution, interfacial molecules give rise to the majority of the signal, meaning that the interface is interrogated preferentially to solution. In addition, because experiments may be performed as a function of the excitation and emission polarization, emission anisotropy measurements may readily be accomplished in the TIRF con¢guration. Thus, it is not surprising that TIRF or some variant of it is a commonly encountered experimental paradigm in £uorescence spectroscopy of biomolecular assemblies. For example, Hlady and coworkers exploited the native £uorescence from the single tryptophan of human growth hormone (HGH), in order to monitor its adsorption to a variety of self-assembled surfaces [7,103]. They noted that the adsorbed amount of HGH was large when the alkyl chains were in an ordered structure and considerably lower where adsorption to disordered alkyl chains of fatty acid or phospholipid layers occurred. They were also able to follow adsorption kinetics on the hydrophilic head groups of self-assembled monolayers and compare them to kinetics for adsorption to hydrophilic solid surfaces. An elegant multilayer molecular construction and scheme for utilizing ligand ^ receptor binding for biosensing was developed by Aizawa and coworkers [104,105]. As shown in Scheme 3, the F(ab)2 fragment of antihuman IgG was bound to the surface followed by human IgG, then a fusion protein of protein A and luciferase was bound to the Fc fragment of the human IgG, thereby presenting the luciferase portion of the fusion protein to the ambient solution. Binding of the ligand luciferin then resulted in excitation-free bioluminescence which was used to sense the presence of luciferin or of the cofactors required in the bioluminescence from luciferase. Karyakin and coworkers demonstrated another interesting use of chemiluminescence by e¡ecting the biocatalytic reduction of H2O2 in the presence of catechol [106]. Horseradish peroxidase (HRP) was bound to the sensing arm of an anti-HRP antibody, which was then bound either to an alkanethiolate monolayer, or after disruption of the disul¢de bonds holding

198

Bohn

SCHEME 3 Molecular construction of the multilayer fusion protein based biosensor for luciferin. (Adapted with permission from Ref. 105. Copyright 1998 Elsevier Science.)

the Fc fragment together, directly to the Au surface through Au ^ thiol chemistry. The resulting bioluminescence was adequate for observation of the binding, despite the fact that Au is potentially an e⁄cient quencher of luminescence via coupling of the molecular excited state to bulk plasmons in the Au substrate.The observation of e⁄cient bioluminescence indicated that the active site of the HRP was su⁄ciently distant from the Au surface to allow e⁄cient radiative decay to occur, thus con¢rming one of the potential binding motifs for the HRP=anti-HRP conjugate.The above applications all rely on observing an increase in £uorescence upon binding of the target molecule. It is also possible to design assays in which the native £uorescence is diminished by the binding of a ligand. This approach was taken by Lo¤pez et al. in their study of the binding of streptavidin to a biotinylated surface containing a native population of synthetically introduced bodipy £uorophores [107]. Addition of the protein was e¡ective in changing both the lifetime of £uorescence and the overall £uorescence intensity, thereby signaling the binding event of interest. Fluorescence microscopy is a powerful technique for examining the patterning of biologically active surfaces with spatially dispersed £uorophores. It was used by Biebuyck and coworkers in their experiments to use micro£uidic networks to pattern biomolecules on Au, glass, and polystyrene [27]. Fluorescently labeled immunoglobulins were used as marker molecules and delivered to the substrates by micro£uidic networks in ways that resulted in spatially variegated patterns. Prestwich and coworkers

Peptide and Protein-Based Biomolecular Assemblies

199

developed an elegant multilayer construction for photopatterning of antibodies on solid surfaces [108]. Their photoimmobilization started with an ethylene glycol-linked heterobifunctional molecule with a benzophenone moiety at one end, and a maleimide moiety at the other. The benzophenone was photoprocessed to cross-link it to poly(styrene), leaving the maleimide residue available for binding the free thiol of a Fab0 fragment of an antiBSA immunoglobulin. Then, a £uorescently tagged BSA was used to bind to the Fab0 fragment, giving rise to £uorescence at the polystyrene surface. Boxer’s group has been active in nonphotochemical patterning of surfaces utilizing supported lipid bilayers with immobilized £uorophores as a mechanism of visualizing the patterns which are produced. In one example from their laboratory, corrals, formed in one of two ways, were used to direct molecular self-assembly spatially. In the ¢rst method, poly(dimethylsiloxane) stamps were used to deposit arbitrarily shaped patterns of immobilized protein onto glass, followed by vesicle fusion into the regions which are not protein-covered. In the second method, supported bilayer membranes were blotted to remove the patterned region of the membrane, and the blotted regions were ¢lled in with protein from solution. In both methods, epi-£uorescence microscopy, using a dilute £uorophore probe, provided a potent mechanism to visualize the progress of the fabrication strategy [109]. Fluorescence recovery after photobleaching (FRAP) is a powerful tool for studying the motion of £uorescently tagged molecules at interfaces [1,2,110,111]. An example of the information which can be gleaned in FRAP experiments on biomolecular assemblies was demonstrated by Yu and coworkers in their study of the interactions of £uorescently tagged BSA with pegylated trimethylsiloxanes [111]. These materials were of interest because of their biocompatibility and their ability to mimic the behavior of ethylene oxide-containing SAMs. The basics of the FRAP technique are shown in Fig.11.Overlapped lasers are brought into focus at an interface, and the areas of constructive interference provide a high local electric ¢eld amplitude, which can then be used to photobleach the £uorophores in that region. After a period of photobleaching, the sample consists of lines of bleached and unbleached molecules. Because photochemically reacted and unreacted molecules are chemically di¡erent, interdi¡usion of the pattern occurs, with a characteristic time behavior which depends on the twodimensional di¡usion properties of the di¡using molecules. These workers discovered that both high- and low-molecular-weight poly(ethylene-glycol)derivatized silanes signi¢cantly reduced the interaction between protein and substrate. The mobility and aggregation behavior of the tagged BSAwas found to be very di¡erent in the two kinds of pegylated layers, however, leading the authors to conclude that the adsorption behavior of proteins

FIG. 11 (a) Schematic diagram showing the elements of the FRAP measurement. (b) Representative FRAP signal for the diffusion of FITC-labeled BSA on pegylated SAMs. (Adapted in part and reprinted with permission from Ref. 111. Copyright 1999 American Chemical Society.)

200 Bohn

Peptide and Protein-Based Biomolecular Assemblies

201

depends not only on interactions with the polymer-graft modi¢ed surface, but also on di¡usion and self-aggregation behavior. 4.2

X-Ray Photoelectron Spectroscopy

One of the workhorse techniques in the characterization of biomolecular assemblies is X-ray photoelectron spectroscopy (XPS). Three types of information are readily available from straightforward XPS experiments. Information about the atomic composition and oxidation state of atoms in the near-surface region are derived from energy-resolved XPS spectra. In addition, by comparing the magnitudes of signals of various species at different electron takeo¡ angles, it is possible to gain an understanding of the relative placement of atomic constituents in the direction perpendicular to the surface plane. Thus, in cases where some components of the biomolecular assembly have unique identifying atoms, its possible to learn a great deal about the structure of the assembly by careful analysis of XPS data. XPS studies have proven to be especially powerful in the analysis of multistep syntheses of multilayer biomolecular assemblies, especially those containing speci¢c polypeptide residues [112^117]. Healy and coworkers made extensive use of XPS in their multistep scheme to build up polypeptide surfaces on titanium oxide [112]. Starting with an amine-terminated organosilane, they used a heterobifunctional linker to provide a thiol-reactive moiety at the distal end which was then coupled to a cysteine-terminated polypeptide. By following spectra of the N1s, C1s, Si2p, and Ti2p photoelectrons, they were able to assign de¢nitively the components of the multilayer structure to the desired molecular precursors. XPS data were combined with spectroscopic ellipsometry to assess the identity, thickness, and surface density of the grafted layers thus produced.The surfaces were then subjected to adhesion studies with rat calvaria osteoblast-like cells. Menzel and coworkers used XPS in a similar study in which polypeptides were grafted to the surface of amino-terminated SAMs by N-carboxyanhydride polymerization [113]. Holland and co-workers also made extensive use of XPS in characterizing a synthetic peptide, ^ GRGDSPK ^, which was covalently bound to a dialdehyde starch coating on a polymer surface [114]. The RGDcontaining peptides were then used to modulate the attachment of human umbilical vein endothelial cells (EC) and to show that EC adhesion, spreading, and growth was similar for cells prepared on these surfaces as for cells prepared on a ¢bronectin-coated surface. Matsuta and coworkers made extensive use of angle-resolved and highresolution X-ray photoelectron spectroscopy in characterizing their derivatized poly(vinyl alcohol) (PVA) ¢lms [118,119]. In their scheme, an isocyanate reaction was used to develop linkers of di¡erent lengths, for

202

Bohn

derivatization of a variety of RGD-containing oligopeptides. The presence of the isocyanate moiety in their reaction scheme was critical and its availability was followed by careful analysis of the line shape in the C1s spectral region, cf. Fig. 12, and by following the time dependence of the N1s=C1s signal ratio at grazing and near-normal electron takeo¡ angles. Using XPS data, these workers were able to show that the surface-modi¢ed PVA had a density of RGD functionalities that was controlled by the degree of isocyanation of the PVA, and that bovine endothelial cells adhered well and grew on these derivatized polymers independent of the presence of serum. 4.3

Vibrational Spectroscopy

4.3.1 Infrared Spectroscopy The workhorse of molecular characterization in homogeneous solution is magnetic resonance spectroscopy. A wealth of detailed structural information can be obtained from relatively simple and straightforward experiments. Unfortunately, the sensitivity of NMR measurements is too low to make it generally applicable to low-surface-area materials, such as planar interfaces.

FIG. 12 The C1s region of the XPS spectrum of an isocyanate-derivatized poly(vinyl alcohol) surface as a function of reaction time. Note the high binding energy peak assigned to C¼O and due to the isocyanate moiety increases at long reaction times. (Adapted from Ref. 118 and reprinted with permission. Copyright 1993 John Wiley & Sons.)

Peptide and Protein-Based Biomolecular Assemblies

203

Thus, vibrational spectroscopy, which is not nearly as information-rich as magnetic resonance, is the method of choice for extracting detailed structural information at the level of individual functional groups in molecules. Standard, transmission-based infrared spectra su¡er from the same limitations as transmission-based absorption measurements in the UV-visible region of the spectrum. However, since the 1960s it has been known that external re£ection at a grazing angle can be used to build up a strong surfacecon¢ned electric ¢eld with p-polarized excitation. The fact that the ¢eld is signi¢cant only with p-polarized excitation means that there is a surface dipole selection rule which can be used to great advantage in characterizing the orientation of dynamic transition dipoles in the infrared [120]. One of the reasons that Au is a particularly fortuitous choice for biomolecular selfassembly is that it is an excellent substrate for surface IR measurements in the external re£ection geometry in an experiment termed infrared re£ection-absorption spectroscopy (IRRAS). The other highly popular geometry for surface infrared spectroscopy makes use of in attenuated total re£ection (ATR). In ATR measurements the substrate has to be transparent over large regions of the infrared spectrum to be useful. This requirement limits the number of materials which can be used as substrates, but within these constraints, both p- and s-polarized absorption can be measured, making it possible to derive detailed structural and orientational information about the biomolecules of interest in the ATR geometry. There are a large number of papers which could serve as exemplars of the type of detailed information which can be extracted from external re£ection IR spectroscopy in the development of biomolecular multilayer assemblies [121^129]. Tengvall and collaborators made extensive use of infrared re£ection-absorption spectroscopy (IRRAS) to characterize in-vitro plasma protein adsorption on o-functionalized alkanethiolate SAMs [124]. They compared SAMs functionalized with terminal methyl, tri£oromethyl ester, sulfate, carboxyl, and hydroxyl functional groups. The low-energy surfaces obtained from methyl and tri£oromethyl ester terminations showed a⁄nity for ¢brinogen, while the more hydrophilic sulfate ^ and carboxylate-terminated surfaces facilitated the deposition of the coagulation proteins, high-molecular-weight kininogen, factor XII, and prekallikrein. IRRAS measurements have been particularly powerful in characterizing directionally aligned helical peptides on Au surfaces. Whitesell and Chang developed a layer-by-layer growth technique which allowed them to assemble the helical peptide poly(alanine) by N-carboxyanhydride chemistry, following the step-by-step synthesis by IRRAS (vide supra) [28]. Fujita and coworkers studied several synthetic octapeptides based on a dimeric repeat unit composed of alanine and a-amino isobutyric acid [127]. In these studies, the authors made use of the fact that helical peptides have a

204

Bohn

speci¢c relationship to the transition moments of the amide I and amide II bands,which are located in a plane formed by the C, O, N, and H atoms of the amide group. Thus, in the approximation that a particular peptide adopts an entirely a-helical structure, it is possible to use polarized IR measurements to extract information about the orientation of the helix relative to the surface, from the ratio amide I to amide II absorbances. Fujita and coworkers exploited this capability to make a detailed structural model including thickness of the peptide adlayer, tilt of the helix axis with respect to the surface normal, and surface packing relative to the Au top layer lattice. Kimura and colleagues took a similar tack in examining a variety of alanine=a-amino isobutyric acid helix-forming peptides with di¡erent terminal functionalities for derivatization of the solid substrate [122,125,126]. Despite using very di¡erent surface binding chemistry, they arrived at an orientation for the helix axis very similar to that in the system studied by Fujita and co-workers. Vogel and co-workers took advantage of ATR infrared measurements in order to characterize the self-assembly of a monolayer of a cheating alkanethiol. Even though the ATR elements in their experiments were coated with  2^3 nm of Au, they were still able to acquire high-quality infrared spectra on SAMs in this con¢guration. They found that the approach compared favorably with IRRAS spectroscopy, and they were able to use it to study metal-ion binding and the e¡ects of metal ions on the secondary conformation of histidine-tagged proteins bound to the SAM [123]. Yarwood and co-workers used a more traditional ATR approach in which self-assembled multilayers of the phospholipid dipalmitoyl-L-a-phosphatidic acid were shown to be well ordered and packed principally in the all-trans con¢guration in a hexagonal lattice. Incorporating the polypeptide gramicidin D into the phospholipid layer produced a largely unperturbed lipid lattice, implying strong peptide aggregation, as was found for the same peptide in Langmuir-Blodgett multilayers [130,131]. 4.3.2 Raman Scattering Compared with the use of IR measurements, Raman scattering is relatively infrequently used to characterize biomolecular assemblies,with almost all of the attention being devoted to assemblies of highly electron-delocalized heme-containing metalloproteins. The reason is straightforward to understand. Given the exceptional weakness of the Raman scattering cross sections of typical molecules, without some type of extraneous enhancement mechanism the signals expected from molecular monolayers would in most cases be well below the limit of detection. Just such a mechanism is employed by taking advantage of the electromagnetic and chemical enhancements

Peptide and Protein-Based Biomolecular Assemblies

205

associated with the formation of roughened surfaces of the coinage metals, establishing surface-enhanced Raman scattering (SERS) and surfaceenhanced resonance Raman scattering (SERRS) as useful tools for the characterization of metalloprotein assemblies. Kitano and coworkers examined cytochrome c bound to self-assembled monolayers of mercaptoalkanoic acids on colloidal Ag comparing the spectra obtained at 488 nm to that at 514.5 nm [132]. They observed that vibrational bands originating from both protein and underlying SAM were observed in the 514.5-nm spectra,whereas at the shorter wavelength only the mercaptoalkanoic acid bands were detectable. Given the strong distance dependence of the SERS excitation phenomenon, it is highly likely that the cytochrome c was entirely contained within a second layer on top of the selfassembled monolayer and distal to the underlying silver substrate. Rospendowski, Smith, and their coworkers examined liver microsomal cytochrome P450 enzymes adsorbed onto citrate-reduced Ag colloids [133]. They observed di¡erences in the spin states of the heme groups in each of the four proteins studied. Cytochrome P450s are well known to denature at the surfaces of sols, so extensive e¡orts were made by these workers to ensure the preparation of biocompatible sols. They concluded that by careful preparation of the sol, the cytochrome P450 could be adsorbed without denaturation or a change in spin state, although in the same system, denaturation to cyctochrome P420 could be forced by exposure to acid or alkali. Consideration of the intensities of characteristic bands suggested that the heme orientation was more perpendicular to the surface of the metal than is the corresponding orientation in adsorbed cytochrome c. Cotton and her coworkers were responsible for a substantial literature addressing the surface-enhanced Raman scattering of a variety of cytochrome and related proteins [134^136]. For example, in their study of the tetra-heme protein cytochrome c3 isolated from Desulfovibrio desulfuricans, a comparison of the protein in solution and at the citrate-reduced Ag sol surface shows that the native structure of cyt c3 is retained at the SERRS active substrate. No evidence for high-spin states was obtained in the spectra of the adsorbed cyt c3. Splitting of redox-state marker bands was observed, however, and was correlated with heme ^ heme interactions in the multisubunit protein. The results were interpreted to mean that the cytochrome is adsorbed to the Ag in an ordered fashion and that the heme exhibiting the most positive redox potential was closest to the surface [135]. Finally, Bohn and coworkers made extensive use of the Raman spin and oxidation state markers in various genetically engineered mutants of the heme protein cytochrome b5. By preparing an ultrathin silicon oxide overlayer on a Ag island structure, they were able to demonstrate surface-enhanced Raman scattering at distances from the silver islands as far as 5^60 —, thereby allowing the

206

Bohn

study of surface chemistry characteristic of the oxide while still retaining the ability to enhance Raman scattering from surface-adsorbed proteins. The spin-state marker bands were found to be consistent with a heme which was high-spin, but six-coordinate. Proposed interpretations for such an intermediate result in the Raman spectrum were (1) a slight alteration in the bonding of the heme pocket for all molecules, leading to an average band position characteristic of the high-spin six-coordinate species, or (2) the existence of two populations on the surface, high-spin ¢ve-coordinate, and low-spin six-coordinate,with the resulting band position being the weighted average of the two components [137]. 5

CONCLUSIONS

The study of biomolecular assemblies is clearly a challenging and fascinating area from an intellectual point of view, and the potential impact of such assemblies on technology is vast. Yet we ¢nd ourselves in much the same situation as the great synthetic chemists of the 1940s, in that much of what can be fabricated or synthesized is still laborious to characterize.There have been, to be sure, great advances in characterization technology, especially in the last 15 years. In 1985, scanned probed microscopies had just been invented and were beginning to be applied in the most rudimentary of model systems. Single-molecule spectroscopy and all of the technology which it presages for high-sensitivity, high-background-rejection spectroscopy of interfaces was still on the horizon, and much of the sophistication which is evident in the studies reviewed here was yet to be approached. So, what are the great lessons we can take from this review of modern characterization tools as applied to biomolecular assemblies? First and foremost, despite the fact that individual tools were discussed one at a time in this chapter, one should realize that routine biomolecular assembly characterization is not accomplished with a single tool. In the vast majority of studies, researchers make use of not one, not two, but in many cases four, ¢ve, or six di¡erent tools to bring di¡erent pieces of structural and functional information about the biomolecular assembly to light. It is instructive to go over one paper in some detail, not because it is extraordinary, but rather because it is quite typical of contemporary research into biomolecular assemblies. In their development of a strategy for immobilizing immunoglobulins on a photoactivateable benzophenone-terminated SAM, Michel and coworkers made use of a large number of disparate characterization strategies [94]. Ellipsometry was used to characterize the thickness of the evolving layers at each stage of the fabrication process, and to contrast the e¡ectiveness of competing procedures at each step in the multilayer assembly. XPS of SAMs terminated in succinimidyl esters,

Peptide and Protein-Based Biomolecular Assemblies

207

amines, and benzophenone provided chemical detail on the type and degree of transformation of the surface and con¢rmed the conclusions reached about the reaction-induced thickness changes from ellipsometry. Clear evidence for the success of several of the reactions was obtained. For example, condensation of the succinimidyl ester with ethylenediamine was responsible for loss of a carbonyl group from the N-hydroxy succinimidyl ester as subsequently was evidenced by a diminution in the high-binding-energy C1s signal from the carbonyl carbon. Scanning electron microscopy was also used as a semiquantitative probe of thickness and molecular composition. SEM demonstrated that washing of pegylated methyl-terminated or benzophenone-terminated SAMs did not result in any signi¢cant change in sample morphology, con¢rming the robustness of the sample preparation procedure. Tapping-mode AFM measurements were used to compare the topography of benzophenone-terminated monolayers on Au(111), both before exposure to immunoglobulin and UV irradiation, and after. Clear evidence for the decoration of the surface with immunoglobulins was obtained as seen by a ¢lling out and rounding of the high spots on the structure observed in SEM. Audioradiography was used to follow a set of radiolabels in a classic radiolabeling experiment. Both pegylated and benzophenone-terminated SAMs were exposed to a 14C-radiolabeled version of an immunoglobulin in the presence and absence of UV irradiation. Clear di¡erences in the radioactivity of samples prepared by the di¡erent procedures pointed to the success of the immobilization when the immunoglobulin was UV-irradiated in the presence of benzophenone. Finally, after the entire structure was assembled, the immobilized immunoglobulins, which were obtained from a mouse monoclonal preparation, were challenged with anti-mouse IgG, labeled with an alkaline phosphatase. The enzyme label signaled the presence of a successful coupling reaction by converting a colorless substrate into a highly absorbing one through enzyme- catalyzed phospholysis. This latter assay cannot be emphasized strongly enough. All of the structural characterization in the world is for naught, if at the end of the preparation the molecular components do not retain their native functionality or the functionality for which they have been engineered. Retention of functionality is most commonly and straightforwardly interrogated by reproducing the catalysis, ligand binding, or other native biochemical behavior observed in homogeneous solution or in the native environment. If those working to prepare and characterize biomolecular assemblies could have one wish for enhanced capabilities on the characterization front, it would undoubtedly involve some measurement by which the kind of structural information available from NMR could be obtained from the relatively small number of molecules present at planar interfaces. While some very early e¡orts are underway to utilize polarization transfer to

208

Bohn

enhance the sensitivity of NMR measurements and bring planar interfaces into the realm in which NMR can be useful, the successful realization of this dream appears to be some time in the future. In the meantime, we expect continued advances on the characterization front and are comforted to know we are far ahead of where the ¢eld stood in 1985. Nevertheless, each successive advance in characterization capabilities is just keeping step with enhanced and ever more sophisticated strategies for chemical manipulation and fabrication of spatially patterned multilayer biomolecular assemblies. As seen above, it is not uncommon for assemblies to be composed of four or more individual, oriented, and carefully chemically prepared layers
each of which is composed of an intermediate which itself involved signi¢cant e¡ort to obtain in pure form.Thus,while characterization e¡orts continue to advance, they seem destined to constantly pursue the sophisticated chemistry that they exist to characterize. ACKNOWLEDGMENTS The portions of this work that were carried out in the author’s laboratories were supported by the National Science Foundation, the Biotechnology Research and Development Corporation, the Department of Energy, and the National Institutes of Health. The author wishes to thank Amy Lynch for aid in preparing the manuscript. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.

R Tilton, C Robertson, A Gast. J Colloid Interface Sci 137:192, 1990. R Tilton, C Robertson, A Gast. Langmuir 7:2710, 1991. B Lok,Y Cheng, C Robertson. J Colloid Interface Sci 91:104, 1983. T Burghardt, D Axelrod. Biochemistry 22:979, 1983. M Garrison, D Iuliano, S Saavedra, G Truskey,W Reichert. J Colloid Interface Sci 148:415, 1992. V Hlady. Appl Spectosc 45:246, 1991. V Hlady, JN Lin, JD Andrade. Biosensors Bioelec 5:291^301, 1990. NM Green. Adv Protein Chem 29:85, 1975. W Mu«ller, H Ringsdorf, E Rump, G Wildburg, X Zhang, L Angermaier, W Knoll, M Liley. J Spinke. Science 262:1706, 1993. L Samuelson, P Miller, D Galotti, K Marx, J Kumar, S Tripathy, D Kaplan. Langmuir 8:604, 1992. R Reiter, H Motschmann,W Knoll. Langmuir 9:2430, 1993. T Dubrovsky, A Tronin, S Dubrovskaya, S Vakula, C Nicolini. Sensors and Actuators B 23:1^7, 1995. R Maoz, J Sagiv. J Colloid Interface Sci 100:465, 1984.

Peptide and Protein-Based Biomolecular Assemblies 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36.

37. 38. 39. 40. 41. 42. 43. 44. 45.

209

R Maoz, J Sagiv. Langmuir 3:1045, 1987. L Netzer, J Sagiv. J Am Chem Soc 105:674, 1983. E Polymeropoulis, J Sagiv. J Am Phys 69:1836, 1978. R Popovitz-Biro, K Hill, E Landau, M Lahav, L Leiserowitz, J Sagiv. J Am Chem Soc 110:2672, 1988. R Nuzzo, D Allara. J Am Chem Soc 105:4481, 1983. E Troughton, C Bain, G Whitesides, R Nuzzo, D Allara, M Porter. Langmuir 4:365, 1988. C Bain, E Troughton,YT Tao, J Evall, G Whitesides, R Nuzzo. J Am Chem Soc 111:321, 1989. P Laibinis, G Whitesides, D Allara, YT Tao, A Parikh, R Nuzzo, J Am Chem Soc 113:7152, 1991. W Knoll. Annu Rev Phys Chem 49:569, 1998. P Harder, M Grunze, JH Waite. J Adhesion 73:161, 2000. L-H Guo, G McLendon, H Raza¢trimo,Y Gao. J Mater Chem 6:369, 1996. RA Clark, EF Bowden. Langmuir 13:559, 1997. JDH Glenn, EF Bowden. Chem Lett 399, 1996. E Delamarche, G Sundarababu, H Biebuyck, B Michel, C Gerber, H Sigrist, H Wolf, H Ringsdorf, N Xanthopoulos, H Mathieu. Langmuir 12:1997, 1996. JK Whitesell, HK Chang. Science 261:73, 1993. CEJRM Corn. Anal Chem 69:1449, 1997. B Frey, C Jordan, S Kornguth, R Corn. Anal Chem 67:4452, 1995. M Yan, S Cai, M Wybourne, J Keana. J Am Chem Soc 115:814, 1993. PA DiMilla, JP Folkers, HA Biebuyck, R Harter, GP Lopez, GM Whitesides. J Am Chem Soc 116:2225, 1994. D Ji, CM Arnold, M Graupe, E Beadle, RV Dunn, MN Phan, RJ Villazana, R Benson, R.Colorado Jr, TR Lee, JM Friedman, J Cryst Growth 218:390, 2000. S Robertson, S Bike. Langmuir 14:928, 1998. CD Tidwell, SI Ertel, BD Ratner, BJ Tarasevich, S Atre, DL Allara. Langmuir 13:3404, 1997. WJ Parak, J Domke, M George, A Kardinal, M Radmacher, HE Gaub, ADGD Roos, APR Theuvenet, G Wiegand, E Sackmann, JC Behrends. Biophys J 76:1659, 1999. D Beyer, M Matsuzawa, A Nakao,W Knoll. Lanmuir 14:3030, 1998. K Webb,V Hlady, PAJ Tresco. Biomed Mater Res 49:362, 1999. B Lu, J Xie, C Lu, C Wu,Y Wei. Anal Chem 67:83, 1995. J Herron, W Muller, M Paudler, H Riegler, H Ringsdorf, P Suci. Langmuir 8:1413, 1992. A Albersdo«rfer, G Elender, G Mathe, KR Neumaier, P Paduschek, E Sackmann. Appl Phys Lett 72:2930, 1998. RG Chapman, E Ostuni, S Takayama, RE Holmlin, L Yan, GM Whiteides. J Am Chem Soc 122:8303, 2000. E Kaganer, R Pogreb, D Davidov, I Willner. Langmuir 15:3920, 1999. M Mrksich, J Grunwell, GJ Whitesides. J Am Chem Soc 117:12009, 1995. M Mrksich, G Sigal, G Whitesides. Langmuir 11:4383, 1995.

210

Bohn

46. H Morgan, D Taylor, C D’Silva, H Fukushima. Thin Solid Films 210^211:773, 1992. 47. A Schmidt, J Spinke,T Bayerl, E Sackmann,W Knoll. Biophys J 63:1385, 1992. 48. J Spinke, M Liley, HJ Guder, L Angermaier,W Knoll. Langmuir 9:1821, 1993. 49. MN Yousaf, M Mrksich. J Am Chem Soc 121:4286, 1999. 50. K Peterlinz, R Georgiadis. Langmuir 12:4731, 1996. 51. K Peterlinz, R Georgiadis, Opt Commun 130:260, 1996. 52. CE Jordan, AG Frutos, AJ Thiel, RM Corn. Analyt Chem 69:4948, 1997. 53. S Lo«fs, B Johnsson. J Chem Soc Chem Commun 1526, 1990. 54. B Fischer, S Heyn, M Egger, H Gaub. Langmuir 9:136, 1993. 55. H Kano,W Knoll. Opt Commun 182:11, 2000. 56. H Kano,W Knoll. Opt Commun 153:253, 1998. 57. T Stein, G Gerisch. Anal Biochem 237:252, 1996. 58. H Sota,Y Hasegawa, M Iwakura. Anal Chem 70:2019, 1998. 59. AVenien, D Levieux, E Dufour. J. Colloid Interface Sci. 223:215, 2000. 60. M Bergener, M Sa«nger, U Candrian. Bioconj Chem 11:749, 2000. 61. CEH Berger,TAM Beumer, RPH Kooyman, J Greve. Anal Chem 70:703, 1998. 62. L Lyon, M Musick, M Natan. Anal Chem 70:5177, 1998. 63. G Sauerbrey. Z Phys 155:206, 1959. 64. M Ward, D Buttry. Science 249:1000, 1990. 65. R Pei, J Hu,Y Hu,Y Zeng. J Chem Technol Biotechnol 73:59, 1998. 66. Y Lvov, K Ariga, I Ichinose,T Kunitake. J Am Chem Soc 117:6117, 1995. 67. S Kanno, Y Yanagida, T Haruyama, E Kobatake, M Aizawa. J Biotechnol 76:207, 2000. 68. S Kanno, Y Yanagida, T Haruyama, E Kobatake, M Aizawa. J Biotechnol 76:207, 2000. 69. K Yun, E Kobatake, T Haruyama, M-L Laukkanen, K Keinanen, M Aizawa. Anal Chem 70:260, 1998. 70. N Damrongchai, K Yun, E Kobatake, M Aizawa. J Biotechnol 55:125, 1997. 71. E Kobatake, S Suzuki, Y Yanagida, T Haruyama, M Aizawa. J Intelligent Mater Syst Struct 10:446, 1999. 72. C Ko«blinger, E Uttenthaler, SDF Aberl, H Wolf, G Brink, A Stanglmaier, E Sackmann. Sensors Actuators B 24^25:107, 1995. 73. A Laschitsch, B Menges, D Johannsmann. Appl Phys Lett 77:2252, 2000. 74. RJ Hamers. J Phys Chem 100:13103, 1996. 75. A Dakkouri, D Kolb, R Edelstein-Shima, D Mandler. Langmuir 12:2849,1996. 76. N Patel, MC Davies, RJ Heaton, CJ Roberts, SJB Tendler, PM Williams. Appl Phys A 66:S569, 1998. 77. J Fang, CM Knobler. Langmuir 12:1368, 1996. 78. DC Cullen, CR Lowe. J Colloid Interface Sci 166:102, 1994. 79. T Sugihara, GH Seong, E Kobatake, M Aizawa. Bioconj Chem 11:789, 2000. 80. T Viitala, I Vikholm, J Peltonen. Langmuir 16:4953, 2000. 81. H Mueller, H Butt, E Bamberg. Biophys J 76:1072, 1999. 82. K Feldman, G Hahner, ND Spencer, P Harder, M Grunze. J Am Chem Soc 121:10134, 1999.

Peptide and Protein-Based Biomolecular Assemblies 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117.

211

JK Stuart,V Hlady. Biophys J 76:500, 1999. M Collinson, E Bowden. Anal Chem 64:1470, 1992. M Collinson, E Bowden. Langmuir 8:2552, 1992. I Hamachi, S Noda,T Kunitake. J Am Chem Soc 113:9625, 1991. M Baek, RC Stevens, DH Charych. BIoconjugate Chem 11:777, 2000. A Lio, A Reichert, DJ Ahn, JO Nagy, M Slameron, DH Charych. Langmuir 13:6524, 1997. JJ Pan, D Charych. Langmuir 13:1365, 1997. MA Firestone, M Shank, SG Sligar, PW Bohn. J Am Chem Soc 118:9033, 1996. HG Hong, PW Bohn, SG Sligar. Anal Chem 65:1635, 1993. HG Hong, M Jiang, SG Sligar, PW Bohn. Langmuir 10:153, 1994. PS Stayton, JM Olinger, M Jiang, PW Bohn, SG Sligar. J Am Chem Soc 114:9298, 1992. PS Stayton, JM Olinger, ST Wollman, PW Bohn, SG Sligar. ACS Symp Ser 240:475, 1994. DA Stephens, PW Bohn. Anal Chem 61:386, 1989. DA Stephens, PW Bohn. Anal Chem 59:2563, 1987. DM Cropek, PW Bohn. J Phys Chem 94:6452, 1990. S Saavedra,W Reichert. Langmuir 7:995, 1991. D Walker, H Hellinga, S Saavedra,W Reichert. J Phys Chem 97:10217, 1993. PL Edmiston, JE Lee, S-S Cheng, SS Saavedra. J Am Chem Soc 119:560, 1997. LLWood, S-S Cheng, PL Edmiston, SS Saavedra. JAm Chem Soc119:571,1997. HHJd Jongh, E Goormaghtigh, JA Killian. Biochemistry 33:14521, 1994. X Chen, CC Moser, DL Pilloud, BR Gibney, PL Dutton. J Phys Chem B 103:9029, 1999. K Owaku, M Goto,Y Ikarlyama, M Aizawa. Anal Chem 67:1613, 1995. M Aizawa, Y Yanagida, T Haruyama, E Kobatake. Sensors Actuators B 52:204, 1998. AA Karyakin, GV Presnova, MY Rubtsova, AM Egorov. Anal Chem 72:3805, 2000. MMA Sekar, PD Hampton, T Buranda, GP Lopez. J Am Chem Soc 121:5135, 1999. X Liu, H-K Wang, JN Herron, GD Prestwich. Bioconj Chem 11:755, 2000. LA Kung, L Kam, JS Hovis, SG Boxer. Langmuir 16:6773, 2000. Z Yang, H Yu. Advanced Mater 9:426, 1997. Z Yang, JA Galloway, H Yu. Langmuir 15:8405, 1999. A Rezania, R Johnson, AR Lefkow, KE Healy. Langmuir 15:6931, 1999. A Heise, H Menzel, H Yim, M Foster, R Wieringa, A Schouten, V Erb, M Stamm. Langmuir 13:723, 1997. J Holland, L Hersh, M Bryhan, E Onyiriuka, L Ziegler. Biomaterials 17:2147, 1996. BD Lamp, D Hobara, MD Porter, K Niki,TM Cotton, Langmuir 13:736, 1997. M Matsuzawa, S Tokumitsu,W Knoll, H Sasabe. Langmuir 14:5133, 1998. VA Tegoulia, SL Cooper. J Biomed Mater Res 50:291, 2000.

212 118. 119. 120. 121.

Bohn

A Kondoh, K Makino,T Matsuda. J Appl Polymer Sci 47:1983, 1993. T Sugawara,T Matsuda. J Biomed Mat Res 29:1047, 1995. R Greenler. J Chem Phys 44:310, 1966. D Beyer, TM Bohanon, W Knoll, H Ringsdorf, G Elender, E Sackmann. Langmuir 12:2514, 1996. 122. Y Miura, S Kimura,Y Imanshi, J Umemura. Langmuir 14:6935, 1998. 123. M Liley,TA Keller, C Duschl, H Vogel. Langmuir 13:4190, 1997. 124. M Lestelius, B Liedberg, P Tengvall. Langmuir 13:5900, 1997. 125. Y Miura, S Kimura,Y Imanshi, J Umemura. Langmuir 15:1155, 1999. 126. Y Miura, S Kimura,Y Imanishi, J Umemura. Langmuir 14:2761, 1998. 127. K Fujita, N Bunjes, K Nakajima, M Hara, H Sasabe, W Knoll. Langmuir 14:6167, 1998. 128. AG Frutos, JM Brockman, RM Corn. Langmuir 16:2192, 2000. 129. Y Ito,YS Park,Y Imanishi. Langmuir 16:5376, 2000. 130. S Naydenova, AG Petrov, J Yarwood. Langmuir 11:3435, 1995. 131. P Lukes, M Petty, J Yarwood. Langmuir 8:3043, 1992. 132. Y Maeda, H Yamamoto, H Kitano. J Phys Chem 99:4837, 1995. 133. B Rospendowski, K Kelly, C Wolf,W Smith. J Am Chem Soc 113:1217, 1991. 134. A Avila, BW Gregory, K Niki,TM Cotton. J Phys Chem B 104:2759, 2000. 135. LH Eng, V Schlegel, D Wang, HY Neujahr, MT Stankovich, T Cotton. Langmuir 12:3055, 1996. 136. R Picorel, G Chumanov, TM Cotton, G Montoya, S Toon, M Siebert. J Phys Chem 98:6017, 1994. 137. PS Stayton, JM Olinger, ST Wollman, EM Thurman, PW Bohn, SG Sligar. In: Synthetic Microstructures in Biological Research, JM Schnur, M Peckerar, eds. New York: Plenum, 1992.

5 Surface Plasmon Resonance Spectroscopy: Applications in Protein Adsorption and Electrochemistry Shaopeng Wang, Salah Boussaad, and Nongjian J. Tao Arizona State University, Tempe, Arizona, U.S.A.

1

INTRODUCTION

Understanding the interactions of proteins with surfaces is important in many ¢elds of biomedical science, from biosensors to biocompatible materials. In the case of biosensor applications, a critical step is to immobilize proteins with intact functions onto the surface of a transducer. Although many proteins can be immobilized by simply letting them adsorb onto the surface, the strong protein ^ surface interactions often cause denaturation of the proteins [1^4]. To remedy this problem, many methods have been developed [5], one of which is to coat an electrode with an appropriate layer of organic molecules [6^10]. A novel extension of this method uses alkylthiol monolayers self-assembled on Au electrodes [11^15]. Recently, membranes of surfactants [16^22] and natural lipids [23^ 26] have been used to coat electrodes by self-assembly or LangmuirBlodgett methods. Because these systems mimic the physiological environment of many membrane-associated proteins, they provide not only an e¡ective way to immobilize proteins onto electrodes for biosensor applications, but also nice model systems to study the biological functions of these proteins. The latter point is important because studying membrane proteins directly in live cells is rather di⁄cult and the model systems can be studied in great detail with surface analytic techniques. These 213

214

Wang et al.

techniques include scanning probe microscopy (SPM), quartz crystal microbalance (QCM), surface plasmon resonance (SPR), and a variety of electrochemical methods. SPR is a rather simple but useful technique that has received increasing attention in recent years [27,28]. A good example is the a⁄nity biosensor, which allows real-time analysis of biospeci¢c interactions without the use of labeled molecules [27,29,30]. SPR has also found applications in protein adsorption [31,32], protein folding=unfolding [33], protein ^ membrane interactions [25,26], and antibody characterizations [34,35]. Similar to QCM, SPR can easily detect adsorbed molecules with submonolayer coverage, and it can be combined with electrochemical techniques to study the electrochemical behavior of proteins immobilized on electrodes. When operated in the multiwavelength mode, SPR provides electronic states of the adsorbed species [36,37]. Other important features are simplicity and a¡ordability. A good SPR setup can be assembled and operated by someone with no extensive optics background. SPR applications have been growing rapidly in recent years. Several thousands of publications are already in existence and many more are coming out quickly. Instead of a comprehensive review, we hope to provide here an introduction to SPR using applications in electrochemistry and protein adsorption as primary examples. The chapter is organized in the following way. A brief description of the fundamentals of SPR is given in Sec. 2. The electrochemical aspect of SPR is discussed in Sec. 3, followed by a description of di¡erent experimental setups of SPR in Sec. 4. Some examples relevant to the protein ^ surface interactions are provided to illustrate SPR applications in Sec. 5. The chapter ends with a brief discussion of future trends in SPR in Sec. 6. Two appendixes are included at the end of the chapter. Appendix A is a supplement to Sec. 2 which provides a more detailed treatment of the SPR fundamentals, and Appendix B describes a simple SPR setup in detail. 2

FUNDAMENTALS OF SPR

Surface plasmons are collective oscillations of free electrons in a metallic ¢lm. Under an appropriate condition, the plasmons can be set to resonate with light, which results in the absorption of light [38,39]. Because the resonance condition is extremely sensitive to the refractive index of the medium adjacent to the metallic ¢lm, presence of molecules on the surface of the metallic ¢lm can be accurately detected. We provide a brief description of the basic principle of SPR below and a more detailed treatment in Appendix A. Thorough treatment of SPR can be found in the literature [40,41].

Surface Plasmon Resonance Spectroscopy

215

FIG. 1 Surface plasmon propagating along (x direction) the interface of a metal and transparent dielectric medium. The associated electromagnetic wave is a transverse magnetic wave (TM) in which the magnetic field is perpendicular to the direction of propagation. The electric field starts from the positively charged regions and ends in the negatively charged regions, which has components in the x and z directions. The amplitude of the electromagnetic wave decays exponentially away from the interface into both media.

Let us consider the interface between a metal (e.g., silver) and a transparent dielectric medium (e.g., water) with dielectric constants, em and el , respectively (Fig. 1). The surface charge density wave associated with the surface plasmons propagating along the interface is given by sðx; t Þ ¼ s0 exp½iðkx x  otÞ, where k x and o are wavevector and angular frequency of the plasmon, respectively. According to the Maxwell equations [42], the surface charge density generates transverse magnetic ¢eld (TM) electromagnetic waves in both the metal and the dielectric medium, whose electric ¢eld components are given by E1 ¼ ½ðE10x x^ þ E10z ^z Þ expða1 zÞ exp½iðkx x  otÞ E ¼ m

^ ½ðEm 0x

þ

z Þ expðam zÞ exp½iðkx x Em 0z ^

 otÞ

for z > 0

ð1Þ

for z < 0

ð2Þ

where a1 and am are two positive constants. According to Eqs. (1) and (2), the amplitudes of the TM waves decay exponentially away from the interface in both media, so theTM waves are surface waves.This is the reason that SPR is sensitive only to molecules at or near the metal ^ dielectric medium interface. In order to excite surface plasmons with a bulk electromagnetic wave (light), one has to use a clever scheme to overcome the wavevector matching problem (see Appendix A). The most popular scheme is the so-called Kretschmann con¢guration [43], in which a bulk electromagnetic wave is

216

Wang et al.

coupled to the surface plasmons via a prism on which a thin metal ¢lm is coated (Fig. 2). At resonance, the angle (yR ) of light incident onto the metal ¢lm from the prism is related to the optical constants of the prism, metal ¢lm, and dielectric medium on the other side of the metal ¢lm, by (see Appendix A) rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi e1 em sinðyR Þ ¼ ð3Þ ðe1 þ em Þe2 For a BK7 prism, silver ¢lm, and water as the dielectric medium, e2 ¼ n22 ¼ 2:30; e1 ¼ n21 ¼ 1:515; and em ¼ 11:7. For an incident light of wavelength 630 nm, yR ¼ 72:2 . Hence, the surface plasmons at the silver ^ water interface can be easily excited with the help of a BK7 prism. Using Eqs. (A4) and (A8) in Appendix A,we found a1  0:005 nm1 , and the decay length is about 200 nm which means that the presence of molecules within  200 nm from the surface can be detected with SPR. Strictly speaking, the prism would modify the dispersion relation of the surface wave, but the modi¢cation is usually very small. A more rigorous derivation that includes the prism can be readily carried out in the same spirit as we have given above.Such a derivation includes three phases,the prism,the metal ¢lm, and the dielectric medium (air or water).The input parameters are the wavelength of the incident light, the dielectric constant (or refractive index) of each phase, and the thickness of the metal ¢lm. When a layer of molecules adsorb on the metal surface, an additional phase needs to be

FIG. 2 Kretchman configuration of SPR. A beam of light incident upon a metal film from a prism, and the reflection detected with a photodetector. When the incident angle reaches the resonance angle given by Eq. (3), the reflection reaches a minimum. Molecules adsorbed onto the metal surface change the resonance angle, which can be accurately detected.

Surface Plasmon Resonance Spectroscopy

217

included as a dielectric slab with two parameters, the refractive index, n, and the average thickness, d. We note that when the coverage is less than one monolayer, d should be understood as an e¡ective thickness that is proportional to the coverage of the adsorbed molecules. Light re£ection from such a four-phase system can be calculated in the frame of Fresnel optics using a matrix method [44]. In fact, the matrix method can be readily applied to a system with an arbitrary number of phases. Using the method, the re£ectivity as a function of the incident angle or wavelength can be calculated numerically. Experimentally one can measure the re£ectivity, RðyÞ, as a function of the incident angle, y, at a ¢xed wavelength. By ¢tting the plot of RðyÞ versus y near the minimum where the surface plasmon resonance occurs, n and d of the adsorbed molecular layer can be extracted. A more common approach is simply to determine the resonance angle, yR, at which the re£ectivity is minimum. When molecules adsorb on the metal ¢lm, yR shifts to a larger angle and the shift, DyR, is related to d according to DyR ¼ cðnÞd

ð4Þ

where cðnÞ depends on the thickness of the metal ¢lm, the refractive indices of the prism, metal ¢lm, bu¡er solution, and the adsorbed molecules. For many applications, such as kinetics of a binding process, only the relative change in DyR is needed. However, if one is interested in an accurate measurement of d, the optical constants of the prism, metal ¢lm, bu¡er solution, and the adsorbed molecules will be needed.The thickness of the metal ¢lm is usually known during the ¢lm evaporation process. The refractive indices of metals and bu¡er solutions can be determined experimentally [45,46] or found in the literature [47]. If the refractive index of the sample molecules is also known, then d can be accurately determined from the measured DyR . In many cases, the refractive index of the sample is not available and one often assumes a ¢xed refractive index based on the known chemical properties of the molecule, or one estimates the refractive index using the Lorentz-Lorenz relation [48]. A more accurate way to resolve this problem is to measure DyR under two di¡erent conditions [45^51]. For example, Peterlinz and Georgiadis [50,51] have measured DyR at two di¡erent wavelengths (twocolor SPR) to determined n and d independently. Chinowsky and Yee [52] have shown that d determined from the two-color SPR can be one order of magnitude more accurate than a single-color SPR. 3

ELECTROCHEMICAL SPR

In an electrochemical environment, the electrochemical potential of the metal ^ electrolyte interface can be controlled, which can induce various

218

Wang et al.

interesting phenomena, from simple double-layer charging to complex electrochemical reactions. A small modulation in the electrode potential (DV ) can shift the resonance angle (DyR ), via changes in the refractive index (Dn), average thickness (Dd) of the adsorbed molecular layer, and surface charge density (Ds) of the electrode. The potential-induced shift is given by DyR ðlÞ DnðlÞ Dd Ds ¼ c1 þ c2 þ c3 DV DV DV DV

ð5Þ

where c1, c2, and c3 are constants and l is the wavelength of the incident light. In the absence of electrochemical induced adsorption or desorption, the ¢rst term in Eq. (5) describes changes in the electronic states of the adsorbed molecules upon modulation of the electrode potential, which can arise from the Stark e¡ect or chemical reactions. The second term describes the change of thickness of the molecular layer, which may arise from adsorption=desorption, structural changes, and the piezoelectric e¡ect. The latter is usually a small constant. Because the refractive index is related to the structure of the molecules, a structural change in molecular layer as described by the second term also induces a change in the ¢rst term. We will discuss how these two e¡ects may be separated in Secs. 5.5 and 5.6. The third term has been studied previously by Kotz et al. [53]. They found no di¡erence in the potential-dependent SPR between two electrolytes, NaClO4 and NaF, and thus ruled out the possibility of anion adsorption being responsible for the SPR shift. They further proposed that the potential changed the dielectric constant of the metal ¢lm, em, via changing the free electron density, ne, according to the simple free-electron model of metals [48], em ðoÞ ¼ 1 

ne e 2 e0 me o2

ð6Þ

where e and me are the charge and mass of electron, respectively. The change of em due to ne is Dem Dne ¼ em  1 ne

ð7Þ

where Dne is related to the potential induced excess surface charge, Ds ¼ C DV , according to Dne ¼ Ds=et ¼ C DV =edm (dm is the thickness of the metal ¢lm). For a 50-nm silver ¢lm, dm ¼ 50 nm, ne ¼ 5.85  1022=cm2, em ¼ 11:8, and C ¼ 20 mF=cm2, Eq. (7) leads to Dem  0:002 per volt. Using Eq. (4), we ¢nd that a 1-V potential can induce an angular shift of  0.02 ,which is in good agreement with the experimental data [54].

Surface Plasmon Resonance Spectroscopy

219

Since the third term of Eq. (5) is proportional to the surface capacitance,which varies as 1=d,where d is the thickness of the adsorbed molecular layer, we have DyðlÞ 1 / const þ ð8Þ DV d if the ¢rst two terms of Eq. (5) are small. In order to examine this relation, Wang et al. [37] have varied d using thiol molecules [HS ^(CH2)n ^ CH3, n ¼ 8, 10, 12, 14, 16, 18] self-assembled on silver electrodes. For a given wavelength (l ¼ 670 nm), the measured Dy=DV is plotted as a function of 1=d in Fig. 3, where d was calculated from the chain length of each thiol molecule using CS Chem3D Pro software. The plot shows that the SPR angular shift in response to the potential modulation is linearly proportional to 1=d, as predicted by Eq. (8). 4

EXPERIMENTAL SETUPS FOR SPR

As mentioned in the previous section, in order to excite surface plasmons with light, one has to overcome the wavevector mismatch problem. In addition

FIG. 3 Potential-induced SPR angular shift of thiol SAMs (CnH2nþ1SH, n ¼ 8, 10, 12, 14, 16, 18) on silver film as a function of thickness. Potential 0.2 V, modulation 10 mV, 200 Hz. (From Ref. 37.)

220

Wang et al.

to the Kretschmann con¢guration [43], in which the coupling of the bulk electromagnetic wave is introduced from the metal side using a prism, the coupling can be introduced from the dielectric-medium side as in the Otto con¢guration [38].The Otto con¢guration requires a small and constant gap between the metal and the prism surfaces, which is not as convenient as the Kretschmann con¢guration. Both the Kretschmann and Otto con¢gurations are based on the attenuated total internal re£ection (ATR) from a highrefractive-index prism. A metal grating with a speci¢c period can also be used to solve the wavevector mismatch problem [55].The incident beam of light is di¡racted by the grating to form a series of beams directed away from the surface at various angles. The wavevector of the di¡racted beam along the interface di¡er from that of the incident one by multiples of the grating wavevector, 2p=a.The surface plasmon resonance occurs only if the component of the wavevector of the di¡racted beam matches that of the surface plasmons. Grating-based SPR has been demonstrated by measuring light intensity or wavelength modulation [56]. A recent report [57] found that the grating method is as sensitive as the ATR methods when using the angle-scan setup, but much less sensitive when using the wavelength-scan setup (both setups are described below). Since the Kretschmann con¢guration is so far the most popular setup,we will discuss it in detail below. 4.1

Rotating Prism Method

One possible setup based on the Kretschmann con¢guration is to measure the re£ection of a p-polarized (TM wave) light incident upon the metal ¢lm as a function of incident angle (Fig. 2).When the incident angle reaches the resonance angle, the re£ection decreases sharply to a minimum, corresponding to the resonance of surface plasmons in the metal ¢lm. The sharpness of the resonance depends on the thickness of the metal ¢lm and the wavelength of the incident light. At 670 nm a 50-nm-thick silver ¢lm can produce a sharp resonance dip with near-zero re£ection at the dip angle. Because the resonance angle (dip position) is sensitive to the adsorption of molecules on the metal ¢lm, accurately measuring the shift of the resonance angle is the basic task. One way is to measure the re£ection as a function of the incident angle with a photodetector by rotating both the prism and the photodetector [58]. This rotating prism method has a typical angular resolution of 102^103 degrees. To compare di¡erent SPR detection methods, the resolution is often described in terms of the smallest detectable change in the refractive index of an analyte (refractive index units, or RIU) [59]. The above angular resolution corresponds to 105^106 RIU. For higher angular resolution, a large distance between the prism and the photodetector is required, making the setup not only bulky but also susceptible to mechanical

Surface Plasmon Resonance Spectroscopy

221

noise and thermal drift.The response time is slow because of the mechanical movements in the setup. 4.2

Intensity at Fixed Angle Method

The mechanical movements in the above setup can be avoided by ¢xing the photodetector at an angle near resonance and measuring the intensity change of the re£ected beam due to a shift in the resonance angle [60^62]. A major advantage of this approach is that the response time is limited only by the speeds of the photodetector and the associated electronics, which can be as fast as a few nanoseconds [62]. A drawback, however, is that the relationship between the intensity and the resonance angle is sensitively dependent on the angle at which the photodetector is ¢xed. Major limitations in the resolution come from the intensity £uctuation of the laser and from thermal and mechanical drifts in the setup. 4.3

Linear Diode Array=Charge-Coupled Device Method

The collimated incident light in the above setups can be replaced with a convergent beam that covers a range of incident angles. The re£ections from di¡erent incident angles are then collected simultaneously with a linear diode array (LDA) or charge-coupled device (CCD) [43,63,64].This method involves no mechanical movements, but simultaneously detection of many channels (e.g., 1024 in a typical LDA) slows down the response time. The typical angular resolution is 102^103 degrees or 105^106 RIU. As in the rotating-prism method, high angular resolution requires large distance between the prism and the photodetector. However, it has been shown that numerically ¢tting the experimental data using, for example, locally weighted parametric regression, can signi¢cantly improve the resolution [65^67]. Garabedian et al. [68] have developed a microfabricated SPR setup in which a position-sensitive photodetector was used to measure the resonance angle. 4.4

Bi-cell Detection Method

Tao et al. used a bi-cell detection method that has achieved an angular resolution of 105 degrees (or 108 RIU). The method uses a convergent beam focused onto a thin metal ¢lm, but the total internal re£ection is collected by a bi-cell photodetector, instead of the CCD or LDA used in the existing setups (Fig. 4, left). The re£ected light falling onto the two cells of the photodetector is ¢rst balanced such that the SPR dip is located near the center of the photodetector. A small shift in the dip position causes a large imbalance in the two cells, which is detected as a change in the di¡erential signal of the bi-cell photodetector (Fig. 4, right). Because the di¡erential signal is linearly

222

Wang et al.

FIG. 4 (Left) Schematic diagram of the proposed SPR setup. The reflected light falling onto the two cells of the bi-cell photodetector is first balanced by adjusting the position of the photodetector. A small shift in the SPR dip causes a large change in the differential signal, AB. (Right) Intensity profiles on the two cells before (solid line) and after (dashed line) a shift in the SPR angle.

proportional to the SPR angular shift and can be easily ampli¢ed without saturation problem, it provides an accurate detection of SPR. The bi-cell photodetector has been used in atomic force microscopy (AFM) to detect small de£ections of a laser beam caused by the bending of the AFM cantilever [69,70]. In the SPR application, the intensity distribution due to the shift in the resonance angle is measured, rather than the movement of the laser beam. The high sensitivity of the SPR is based on the di¡erential detection with the bi-cell photodetector.The shift in the resonance angle is determined from the ratio of the di¡erential signal, AB, to the sum signal, A þ B,where A and B are the output signals of the two cells. Linearity, resolution, and response time, three important parameters of the bi-cell SPR setup, are discussed in details in Appendix B. Because of the simplicity of the method, we also provide a detailed description of the setup in Appendix B for those who are interested in building one. 4.5

Wavelength Modulation Method

The above setups involve re£ection intensity versus incident angle at ¢xed wavelength of incident light,which is referred to as angle-scan systems. SPR has also been detected by measuring the re£ection versus wavelength at ¢xed angle of incidence or the so-called wavelength-scan system. One accurate method involves modulating the wavelength of the incident light [71,72].The wavelength modulation provides an accurate measurement of SPR by modulating the re£ected intensity, which can be monitored with a lock-in ampli¢er. Using an acoustooptic tunable ¢lter (AOTF), it has been demonstrated that wavelength changes of 0.0005 nm around 630 nm [72] correspond to 5 107 RIU [59]. When applied to DNA-SH adsorption on gold, the

Surface Plasmon Resonance Spectroscopy

223

signal-to-noise ratio of the AOTF SPR is six times better than that achieved by an angle-scan system [71]. The wavelength-dependent SPR has also been used to build an optical ¢ber setup [73]. 5

APPLICATIONS OF SURFACE PLASMON RESONANCE SPECTROSCOPY

SPR has become a powerful optical technique in recent years for a variety of applications, from the measurement of physical quantities to chemical and biological sensing [59,74^76]. By coating the metal ¢lm with various optical transducer materials that are sensitive to di¡erent physical parameters, SPR can be used to detect these parameters. For example, utilizing humidityinduced refractive index changes in porous thin layers and polymers, SPR can be transformed into a humidity sensor [77]. Based on the thermooptic e¡ect in hydrogenated amorphous silicon, an SPR temperature sensor was reported [78]. One simple SPR chemical sensing application is to monitor the distillation processes of £avor mixtures [79], in which the concentration of an analyte is determined from the refractive index. Most chemical SPR sensors for complex systems are based on the measurement of the change in the resonance angle due to the adsorption of an analyte onto or a chemical reaction of an analyte with a transducer layer. For the detection of various organic vapors, a layer of appropriate materials is coated onto the metal ¢lm, and the adsorption of each vapor onto the ¢lm results in a change of the resonance angle. For example, polyethylene glycol ¢lms were used to detect aldehydes and alcohols [80], and a gold active layer was used to detect NO2 [81]. Other examples include the detection of toluene by copper or nickel phthalocyanine ¢lms [82] and the detection of NO2 and H2S by polyaniline ¢lms [83]. Combined with cyclic voltammetry, SPR has been used to detect Cu and Pb ions [84,85]. The ¢rst biosensor application of SPR was reported in1983 [86], and an early survey on real-time biospeci¢c interactions analysis using SPR was published in 1994 [87]. Since then, numerous studies of the kinetic and the thermodynamic constants of biological interactions have been reported with SPR [59,88]. Several di¡erent SPR biosensor approaches have been used. One of the widely used approaches is to detect the binding reaction of an analyte directly. This approach works well for large molecules, but for small molecules, indirect detection methods such as sandwich SPR [89] or competitive assay [90] methods seems to work more e¡ectively. Early SPR applications for protein studies were focused on the methodology and simple sample systems such as antigen ^ antibody binding. The example of biotin ^ avidin reaction was studied extensively. With the commercialization and wide use of SPR, current research covers a broad

224

Wang et al.

range of biological interactions [51,91,92]. Among all the systems, the binding kinetics is the mostly studied problems, followed by qualitative detection and equilibrium studies, as reported in the recent survey of Myszka [76].One of the keys in the SPR biosensor applications is the appropriate coating layer that provides speci¢c binding or interaction of an analyte with the layer, which can be detected. A number of di¡erent surfaces have been developed to accomplish di¡erent types of application, and many of them are commercially available from Biacore or A⁄nity Sensors. These include di¡erent types of hydrophilic or hydrophobic surfaces, positive or negative charged surfaces, biotinylated surface for Streptavidin conjugation, nickel chelation (NTA) surface for His-tagged protein conjugation, and lipophilic dextran surface to capture liposomes [75]. 5.1

Binding Kinetics

SPR has been widely used to determine the binding kinetics of biological interactions, such as antibody ^ antigen [93], ligand ^ receptor [94], and DNA hybridization [51]. Here we give a few examples of such applications to illustrate the method. Altschuh and coworkers [93] studied the di¡erence in the a⁄nity of a monoclonal antibody raised against the protein of tobacco mosaic virus for 15 related peptides carrying single-residue modi¢cations. An analysis of the peptide ^ antibody interaction in real time with SPR allowed fast and reproducible measurements of both the association and the dissociation rate constants. Out of the 15 mutant peptides analyzed, seven were recognized as well as the wild-type peptide. The pattern of residue recognition suggests that a helical conformation formed by three residues mimics the structure of the protein. Even a minor modi¢cation to these residues totally prevents the recognition by the antibody. Modi¢cation of adjacent residues results in signi¢cant di¡erences in the binding constants. Figure 5a shows a typical kinetic measurement. The association rate constant ka and the dissociation rate constant kd were calculated from the slope and the intercept of the dRA =dt versus RA plot (Fig. 5b), where RA is the SPR response to the antibody bound to the immobilized ligand obtained from the kinetic curves after subtraction of the background signal. Most SPR kinetic measurements involve either stepwise titration or kinetic analysis (such as the example described above), in which the SPR response is monitored as a function of time after the injection of the analyte. However, these methods have their limitations in the types of interactions that can be studied. The stepwise titration is time-consuming due to the requirement for multiple-step measurements. Denaturation of biological samples and nonspeci¢c binding also increase with time, making the result

Surface Plasmon Resonance Spectroscopy

225

FIG. 5 Example of a typical kinetic run analysis. (a) Kinetic run at a 37.5 nM concentration of Mab 57P. (A) 21 mL of antibody followed by (B) 6 mL of 50 mM HCl is injected at a flow rate of 3 mL=min. (b) dRA=dt versus RA plot. (From Ref. 93.)

226

Wang et al.

di⁄cult to interpret. The kinetic analysis can be performed very quickly, but it must be repeated at multiple £ow rates and analyte concentrations to ensure accurate results and to isolate the e¡ects of mass transport. The sensor surface must be regenerated or replaced by a new one, which is either time-consuming or costly. Recently, Shank-Retzla¡ and Sligar have reported an improved one-step method [88], to determine the kinetic rates and the equilibrium binding a⁄nities, called analyte gradient-surface plasmon resonance (AGSPR). They used a high-performance liquid chromatography (HPLC) pump system to create a continuous gradient £ow that made the concentration of the analyte passing over the SPR sensing surface increase linearly with time. The binding rate between the analyte and the immobilized receptors was measured by monitoring the change in the resonance angle. Kinetic rates were determined by ¢tting the data with a modi¢ed version of a two-compartment £uid cell [95]. In the ¢rst compartment, the applied analyte concentration was assumed homogeneous and equal to the applied concentration. The second compartment was the region next to the surface of the sensor, where there was no £ow, making the analyte enter or leave the compartment only by di¡usion. The transport of material into this region depends on the £ow rate, the di¡usion constant, and the geometry of the £uid cell. The binding reaction can be described as a twostep process in which the ¢rst step is the transport of the analyte from the ¢rst compartment to the second, and the second step is the binding of the analyte to the surface receptors. The £ux of the inbound analyte can be described with an ordinary di¡erential equation and the rate constant can be resolved from there. Because only the £ow rate and the slope of the gradient determine the experiment time, problems with the receptor dissociation and sensor fouling are minimized. One AG-SPR experiment provides su⁄cient information to determine both the kinetic rates and the binding a⁄nity. Simulations indicate that the method can be used not only for accurate determination of both the kinetic rates and the equilibrium a⁄nities, but also for the characterization of the interaction, which does not obey pseudo-¢rst-order kinetics due to the presence of heterogeneous receptor population. This method was used to characterize the interaction between cytochrome c and cytochrome b 5 ; both the speci¢c and nonspeci¢c interactions were quantitatively analyzed. A limitation of the method is the aggregation of analytes at high concentrations, especially when the reactivity of the analyte is signi¢cantly altered by the aggregation. In this case, the concentration of the analyte that is available for binding will not increase linearly, and the model is no longer suitable.

Surface Plasmon Resonance Spectroscopy

5.2

227

Protein–Lipid Interactions

As mentioned in the introduction, the interactions of proteins with solidsupported lipid bilayers is important not only for biosensor applications but for a better understanding of the functional role of such interactions in biological systems. SPR is a highly sensitive technique for studying protein ^ lipid interactions [96^101]. One interesting example is the interactions of cytochrome c with the bilayer of cardiolipin, an anionic diphosphoglyceride lipid located in the inner membrane of mitochondria. The redox potential measurements, nuclear magnetic resonance (NMR) [102,103], and resonant Raman spectroscopy [104] have shown strong interactions between cytochrome c and cardiolipin. Salamon and Tollin [99] studied the complex formation between cytochrome c and a solid-supported phosphatidylcholine lipid bilayer containing varying amounts of cardiolipin.They used a self-assembly method to form the lipid bilayers on silver electrodes. The ¢rst step of the preparation involves spreading a small amount of cardiolipin solution across an ori¢ce in a Te£on sheet between the silver electrode surface and the solution. The hydrophilic silver surface attracts the charged polar groups of cardiolipin and results in a cardiolipin monolayer with the hydrocarbon tails pointing into the bulk solution. The SPR cell is then ¢lled with the appropriate aqueous solution that allows formation of both the second monolayer and a Plateau-Gibbs border that anchors the bilayer ¢lm to the Te£on sheet. Using SPR, they studied the protein-binding kinetics over a wide range of protein concentration and ionic strength conditions. They found two distinct binding processes. The initial binding involves a purely electrostatic interaction between cytochrome c and the charged cardiolipin head groups.The second process is a hydrophobic interaction that accompanies penetration of the protein into the membrane interior. The conclusions were supported by direct AFM images of the cytochrome c binding to the cardiolipin bilayer supported on a graphite electrode [105,106]. 5.3

Protein Conformational Changes

Membrane proteins respond to a variety of extracellular stimuli, such as hormones, ions, light, and electron transfer reactions. These proteins are believed to switch from one conformation to another in order to selectively bind ligands and activate important functional systems. Immobilized membrane proteins on modi¢ed solid substrates are nice model systems that mimic the in-vivo system, and SPR is a nice tool that allows the conformational changes to be studied. Salamon et al. [98] applied SPR to bovine rhodopsin immobilized in phosphatidylcholine bilayers. They monitored the process of rhodopsin

228

Wang et al.

incorporation into the bilayer and the light-induced conformational changes. The magnitude of the SPR change indicated an increase of 4 — in the thickness of the proteolipid ¢lm, which agrees with the £ash photolysis experiments. Heyse et al. [107] monitored the coupling reactions between rhodopsin incorporated into patterned phospholipid membranes and transducin with SPR.The coupling reactions determined by SPR resemble closely the native system, which indicates that rhodopsin had preserved its native functionality. Ozawa et al. [108] examined the complex formation of calmodulin ^ metal ^ peptide. Large changes in SPR were observed for several cations including Ca2 þ , indicating that these cations favor the complex calmodulin ^ metal ^ peptide. On the other hand, some cations such as Mg2 þ , Cu2 þ , and Zn2 þ displayed no SPR signal. The selectivity of recoverin is similar to that of calmodulin. Upon Ca2 þ binding, large SPR change occurs due to the recoverin conformational changes, which extrudes the myristoyl group. Mg2 þ induced no SPR signal even though recoverin accommodates two metal ions. This con¢rms that recoverin does not undergo conformational changes in the presence of Mg2 þ [109]. Sota et al. [33] used SPR to investigate conformational changes in Escherichia coli dihydrofolate reductase induced by acid denaturation. The protein, immobilized on the surface through a disul¢de linkage, exhibited a larger response to acid treatment compare to the bare surface used as a reference. Furthermore, the SPR response to the changes of pH is similar to that of CD,which makes SPR a potential sensing device switched by changes of conformation. Redox-induced conformational changes are more di⁄cult to measure because they are very small. Boussaad et al. [36] used high-resolution SPR, which can achieve a resolution of 108 RIU, to monitor the redox-induced conformational changes of cytochrome c. The measured SPR shows a sigmoid decrease as the protein is switched from the oxidized to the reduced states. The related results are discussed in more detail below. 5.4

Electrochemical Surface Plasmon Resonance Spectroscopy

Several groups have recently started to use SPR in combination with electrochemical techniques. The metal ¢lms in the SPR are naturally used as electrodes onto which various electrochemical phenomena take place. Hanken and Corn [110] performed electrochemically modulated surface plasmon resonance (EM-SPR) on organic thin ¢lms. They used SPR to measure the changes in the index of refraction of a noncentrosymmetric thin organic ¢lm upon the application of an external electrostatic ¢eld that was

Surface Plasmon Resonance Spectroscopy

229

controlled by the electrochemical potential of the electrode. The pro¢le of the electric ¢eld was measured inside the ¢lm. A conventional rotation plate setup was used to scan the SPR spectrum at a ¢xed electrode potential. The di¡erential re£ectivity due to the electrooptical e¡ect was measured by modulating the electrode potential around the ¢xed potential. Fig. 6 shows the di¡erential re£ectivity curves observed for the ¢lm at di¡erent potentials.The x axis of Fig. 6 represents the di¡erence between the incident angle and the surface plasmon angle.This study shows that EM-SPR is an e¡ective method for monitoring the electrostatic ¢eld strength inside thin organic ¢lms on electrodes. Iwasaki and coworkers [111] analyzed the electrochemical reaction of Fe(CN)63=4 on gold electrodes using SPR combined with cyclic voltammetry (CV). The time derivative of the SPR signal was correlated to the electrochemical current. They found from the SPR response that phosphate

FIG. 6 Differential reflectivity (D%R) obtained by EM-SPR measurements in 0.2 M tetrabutylammonium bromide for the 1 HAPA þ 4 DBP ZP film. The curves are the overlay of increasing electrode modulation potentials (Dfm) at 1 kHz. The differential reflectivities measured in situ are converted to a change in electric field strength (DE) with the self-assembled multilayer. (From Ref. 110.)

230

Wang et al.

ions replacement by Fe(CN)63=4 tend to increase with the concentration of Fe(CN)63=4. At high concentration the SPR signal re£ects the di¡erence between the refractive indices of the reduced and oxidized states of Fe(CN)63=4. The potential dependence of the resonance angle increased with time while the CV remained almost unchanged (Fig. 7). This observation was attributed to the formation of a surface ¢lm that changes the SPR signal but does not participate in the electron transfer reaction probed by CV. This study demonstrated that the correlation between the electrochemical current and SPR could be a molecular-speci¢c detection method. Badia, Knoll, and coworkers [112] used SPR and AFM in combination with CV to probe the electrochemically driven deposition of organic molecules onto electrode surfaces. They monitored the self-assembly and desorption of alkanethiols on gold under potential control. This revealed the potential dependency of the self-assembly of alkanethiols on gold surface. For example, C16H33SH does not adsorb on a gold surface at surface potentials below 0.8 V. The experiment demonstrated the possibility of preparing organic monolayers with low defect densities by controlling the self-assembly process with the electrochemical potential. Schlereth [113] used SPR in combination with CV to characterize monolayers of cytochrome c and cytochrome c oxidase on a gold surface modi¢ed with self-assembled alkanethiol monolayer. The SPR angle and

FIG. 7 Cyclic voltammograms (10 mV=s) of Fe(CN)3=4 (9.5 mM in 2 M KCl) in 6 initial scan (solid line) and after 2500 s (dashed lines). (From Ref. 111.)

Surface Plasmon Resonance Spectroscopy

231

the electrochemical current were recorded while cycling the potential for each species. Protein coverage calculated from SPR and CV gave information about the amount and orientation of cytochrome c adsorbed on modi¢ed gold. Fig. 8 shows the SPR and time-di¡erential SPR-CV plots of Au=MPA surfaces covered with a monolayer of cytochrome c oxidase. The results were interpreted as a potential-dependent conformational change

FIG. 8 (A) Potential-dependent SPR angle shift (Dypl) for Au=MPA surfaces modified with a monolayers of cytochrome c oxidase recorded at 1 mV=s (a) in 5 mM Na-phosphate, pH 7.0, and (b) after addition of 0.1 mM ferrocytochrome c in the solution. Negative (full line) and positive (dotted line) scans. (B) Time differential SPR-CV plots for Au=MPA surfaces modified with a monolayers of cytochrome c oxidase recorded at 1 mV=s in 5 mM Na-phosphate, pH 7.0 (gray) and after addition of 0.1 mM ferrocytochrome c in the solution. Negative (full line) and positive (dotted line) scans. (From Ref. 113.)

232

Wang et al.

between the ‘‘resting’’ and the ‘‘pulsed’’ states of the adsorbed cytochrome c oxidase, which gives rise to two species with di¡erent electrochemical behavior. Schlereth has also characterized self-assembled monolayers with biospeci¢c a⁄nity for NAD(H)-dependent dehydrogenases using SPR combined with electrochemistry [114]. Boussaad et al. studied the redox-induced conformational changes in the redox protein, cytochrome c. A basic question about this protein is whether there is a large conformational di¡erence between the oxidized and the reduced states [115^119]. Experiments, such as hydrodynamic [117,118] and small-angle X-ray scattering measurements [116], detected large di¡erences between the two states, which were attributed to large conformational changes. However, X-ray crystallography [120] and NMR spectroscopy [121] observed only small structural di¡erences between the two states. The electron transfer current (Fig. 9a) and the resonance angle (Fig. 9b) as functions of potential for cytochrome c immobilized on MPA-coated gold were simultaneously recorded in 50 mM phosphate bu¡er (pH ¼ 7.0). The peaks in the CV plots are due to the well-known reduction and oxidation of cytochrome c, involving Fe3þ $ Fe2þ þ 1e (Fig. 9a). The simultaneously measured resonance angle shows a sigmoid decrease as the protein is switched from the oxidized to the reduced state. The change was reversible when switching the protein back to the oxidized state. The noise in the resonance angle versus potential plot came largely from the noise in the electrochemical potential. The SPR shift was attributed to a conformational change in the protein induced by the electron transfer. This change can a¡ect both the average thickness and the refractive index of the protein layer. For the purpose of a rough estimate, however, one can use the Lorentz-Lorenz relation [48], which relates the change in the refractive index to the change in the conformation of the protein. Using this relation, they estimated that 0:008 shift (after background correction) corresponds to 0.2^0.4 — change in the dimension of the protein. This is in good agreement with X-ray crystallography [120] and NMR studies [121]. 5.5

Multiwavelength Electrochemical SPR

The most frequent optical measurement performed on biological molecules is perhaps the absorption of visible or ultraviolet light.This technique can be used for many purposes, ranging from simple concentration determination to resolving complex structural questions. Pockrand et al. [122] showed that the absorption properties of dye monolayers on silver ¢lms could be determined by SPR. They measured the SPR angular shift, dip width, and intensity as a function of wavelength, from which they extracted the real and

Surface Plasmon Resonance Spectroscopy

233

FIG. 9 (a) The cyclic voltammograms with of cytochrome c immobilized on the surface in 50 mM phosphate solution, where the arrows point to the oxidation and reduction of the protein. (b) Corresponding shift in the resonant angle due to the electron transfer.

imaginary parts of the dielectric constant of the dye molecules. The wavelength-dependent dielectric constant provided absorption spectroscopic properties of the dye molecules. Using high-resolution SPR, Boussaad et al. recently obtained absorption spectroscopy of cytochrome c [36,37]. The absorption bands provide detailed information about the electronic states of molecules, which allow one to identify them directly. This is important because, as discussed in previous sections, most SPR identi¢es molecules

234

Wang et al.

based on speci¢c bindings with their respective ligands immobilized on the surface. When applied to the study of conformational changes, SPR measures the overall changes in the refractive index and the thickness of the adsorbed molecular layer. This capability helps to pin down the parts of the molecule involved in the conformational changes. One possible way to obtain absorption spectroscopy of adsorbed molecules is to measure the resonance angle as a function of the incidentlight wavelength.When the wavelength is scanned across an optical absorption band of the molecule, a kink centered at the peak position occurs in the refractive index of the molecule,which is accurately measured with highresolution SPR. Because the kink is related directly to the absorption spectrum of the molecule, information about the electronic states of the molecule is readily obtained.This is di¡erent from the previously reported wavelength modulation method, in which the wavelength is varied at a ¢xed incident angle. The basic principle is brie£y discussed below. According to Eq. (5), the resonance angle shift, DyR, depends on the changes of the surface charge density, the thickness of the adsorbed protein layer, and the refractive index of the protein.The former two are independent of the wavelength, thus the wavelength-dependent part of DyR is proportional to the change in the refractive index of the protein. The dielectric constants of the metal and glass prism are weakly dependent on the wavelength within the studied range. By measuring the resonance angle at various wavelengths, the refractive index as a function of wavelength can be determined. Because the change in the refractive index, Dn, is related to the absorption coe⁄cient, De, by the Kramers-Kronig relation [123], 2l DeðlÞ ¼  P p

Z1 0

Dnðl0 Þ 0 dl l2  l0 2

ð9Þ

where P in front of the integral sign implies how the in¢nity at l0 ¼ l should be treated, the absorption coe⁄cient can be determined from the resonance angle at all wavelengths. In practice, even if data are available only for a limited wavelength range, the relation can still provide a rather accurate result [123]. Therefore, the optical absorption properties of proteins adsorbed on the surface can be obtained from the multiwavelength SPR measurement.This is important because directly measuring the absorption spectra of large biological molecules adsorbed on the surface is known to be a di⁄cult task. Even for cytochrome c, which has a large molar extinction coe⁄cient (10 4 M1 cm1 at 550 nm) (Fig. 10a), the re£ectivity measurement must have a minimum accuracy of 105 in order to detect the absorption spectrum of one monolayer. Such high accuracy is not easy to achieve. Using multiwavelength SPR, the absorption band at 550 nm leads to a kink of

Surface Plasmon Resonance Spectroscopy

235

FIG. 10 (a) Absorption peak near 550 nm for reduced cytochrome c. (b). The calculated refractive index variation due to the absorption using the Kramers-Kronig relation exhibits a kink centered at the absorption peak. Because the shift in the resonant angle is proportional to the change in the refractive index, the kink also appears in the resonant angle versus wavelength plot. The proportionality is about 6 per RIU for a 5-nm-thick protein layer as found by a numerical calculation based on Fresnel optics. (From Ref. 36.)

amplitude 0:04 at the resonance angle as predicted by the Kramers-Kronig relation (Fig. 10b). Such a large angular shift is much easier to detect with SPR. For an angular resolution of 105 degrees,which was achieved with the preliminary setup,we can measure the absorption spectrum of cytochrome c up to a coverage of 1%. One example of such an application of multiwavelength SPR is the study of the redox reaction of cytochrome c [36]. Reduced cytochrome c

236

Wang et al.

(solid line) has two pronounced absorption peaks at 520 and 550 nm respectively,while oxidized cytochrome c (dashed line) is relatively £at in the same wavelength window (Fig. 11a). The shift of the resonance angle as a function of the wavelength is plotted in Fig. 11b. Far from the absorption band, the shift is independent of the wavelength and re£ects conformational changes of the protein, as discussed before. However,when the wavelength is close to the absorption peaks, two interesting kinks centered at 520 and 550 nm appear, as expected from the Kramers-Kronig relation.Using the absorption spectra as an input, the resonance angle shift (Fig. 11c) was calculated using the Kramers-Kronig relation, and quantitative agreement was found between the theory and the experimental data (Fig. 11b). 5.6

SPR Stark Spectroscopy

The e¡ect of an applied electric ¢eld on the absorption or emission spectra of molecules is known as Stark e¡ect. Stark spectroscopy measures the change in the absorption spectra of the molecule as a function of the applied electric ¢eld. This technique has been widely used to study di¡erent molecular systems and materials [124]. In order to study the Stark spectra of molecules adsorbed onto a surface, the ability to measure small amounts of light absorption is required. One way to achieve high sensitivity is to use a socalled potential-modulated electrore£ectance absorption spectroscopy in which a small change in the re£ectance is measured with a lock-in technique [125^127]. Wang et al. [37] have shown that the multiwavelength SPR described above can be used to obtain Stark spectroscopy of adsorbed molecules. The basic principle is described by Eq. (5), in which the ¢rst term is due to changes in the electronic states of the adsorbed molecules upon modulation of the electrode potential, which can arise from the Stark e¡ect. Knowing Dy=DV as a function of l, the change in the molar extinction coef¢cient (De) as measured by conventional Stark spectroscopy can be obtained according to the Kramers-Kronig relation. Stark spectroscopy of Ni(II)-phthalocynine-tetrasulfonic acid tetrasodium salts (NiPh) and other organic molecules have been obtained using this method. The SPR response was ¢rst measured (Dy=DV ) with blank NaClO4. Then a small amount of 104 M aqueous solution of each of the molecules was introduced into the SPR cell. The SPR shift was monitored during the adsorption of molecules on the surface of the electrode. The shift stopped after 30 min, corresponding to the maximum coverage of adsorbed molecules. Replacing the sample solution with a blank electrolyte caused no further changes of the resonance angle, which indicates that the adsorbed molecules were rather stable on the surface of the electrode. A potential modulation was then applied to the electrode, and the induced SPR shift

Surface Plasmon Resonance Spectroscopy

237

FIG. 11 (a) Absorption spectra of reduced (solid line) and oxidized (dashed line) cytochrome c. (b) experimental SPR shift of cytochrome c (open and filled circles) as it is switched from oxidized to reduced states. The kinks occur at absorption peaks, 550 and 520 nm. The shift in pure phosphate buffer (open squares). (c) Theoretical SPR shift based on the absorption peaks and the Kramers-Kronig relation. (From Ref. 36.)

238

Wang et al.

ðDy=DV Þ was determined as a function of l. The result, plotted in Fig. 12 for NiPh, shows pronounced dips near 640 nm. In sharp contrast, the plot of the thiol molecules, HOOC ^(CH2)10^ SH, shows only a weak and smooth dependence on l within the same range of wavelength. The dip for NiPh is due to the Stark e¡ect because no detectable electrochemical reactions or desorption take place upon a 20-mV potential modulation at 0.2 V. By integrating DyðlÞ=DV according to the Kramers-Kronig relation [Eq. (9)] DeðlÞ=DV was extracted. Previous Stark spectroscopy studies found that the applied electric ¢eld often induces a dipole moment only along one direction, regardless of the molecular orientation, and the absorption shifts toward lower energy. Consequently, the ¢eld shifts the entire absorption spectrum to a higher or lower wavelength without changing its shape, and the Stark spectrum is given by the ¢rst-order derivative of the absorption spectrum with respect to the wavelength. Figure 13 compares the Stark spectrum (bottom) obtained with the SPR to the solution-phase absorption spectrum (top) and the ¢rst derivative (middle) of the absorption spectrum. The solution-phase absorption spectrum was measured at 104 M NiPh þ 0.1 M NaClO4 and shows a typical peak of phthalocynine at  610 nm. The ¢rst-order derivative of this spectrum shows a kink around 610 nm. The

FIG. 12 Potential-induced SPR angular shift of HOOC(CH2)10SH and NiPh as a function of wavelength. (From Ref. 37.)

Surface Plasmon Resonance Spectroscopy

239

FIG. 13 (Top) Solution-phase absorption spectrum of NiPh in 0.1 M NaClO4. (Middle) derivative of the absorption spectrum. (Bottom) Stark spectrum obtained from SPR angular shift and Kramers-Kronig relation. (From Ref. 37.)

240

Wang et al.

Stark spectrum is similar to the ¢rst derivative of the absorption spectrum, indicating that the electrode potential shifts the absorption energy of the molecule. 5.7

SPR Absorption Spectroscopy

The above methods of SPR-based absorption spectroscopy rely on measuring the resonance angle that is related to the change of the refractive index as a function of wavelength. The absorption coe⁄cient of the molecules is extracted from the wavelength-dependent refractive index using the Kramers-Kronig relation. Another way to obtain absorption spectroscopy is to measure the re£ectivity changes at the resonance angle due to the absorption of light by molecules on or near the metal ¢lm. Kano et al. [128] showed that the re£ectivity changes could be much more than simple optical absorption due to an enhancement e¡ect by the surface plasmons. Kolomenskii et al. [129] and Wang et al. [130] measured the re£ectivity as a function of wavelength, from which they obtained absorption spectroscopy with a signal enhancement of at least one order of magnitude over conventional absorption spectroscopy. The changes in the re£ectivity, RðyR Þ, at the resonance angle when tuning the wavelength to the absorption band of the molecules can be attributed to two e¡ects. The ¢rst one is direct absorption of light by the adsorbed molecules, which always decreases the re£ectivity. Since SPR intensi¢es the optical ¢eld near the surface, the re£ectivity decrease is much greater than that in the conventional re£ectance measurement. The second e¡ect is that the absorption of light by the molecules perturbs the resonance condition and causes an increase in the re£ectivity. Consequently, the net re£ectivity change is either positive or negative, depending on the e¡ect that dominates. In addition, the net re£ectivity is controlled by the thickness of the metal ¢lm. The absorption spectrum is obtained by simply measuring RðyR Þ as a function of the wavelength.Using the right thickness of metal ¢lm, an enhancement factor of 40 times over conventional absorption spectroscopy can be achieved. In addition to this enhancement, SPR absorption spectroscopy has an advantage in the dynamic range of the photodetector. The conventional approach measures the light re£ected from (or transmitted through) the adsorbed molecules. The intensity change in the re£ected (or transmitted) light due to the optical absorption of the molecules is usually very small comparing to the total intensity. In order to detect this small change from the intense re£ected (or transmitted) beam, the photodetector and associated electronics must have a large dynamic range. In contrast, SPR absorption spectroscopy measures the intensity change at the SPR dip, which is much weaker than the re£ected or transmitted beam. Therefore, for

Surface Plasmon Resonance Spectroscopy

241

a given dynamic range, SPR absorption spectroscopy is more sensitive than the conventional approach even without enhancement. 6

SOME FUTURE TRENDS OF EC-SPR

SPR has found applications in many ¢elds, from determining optical properties of thin ¢lms to biological sensors. It has been combined with other techniques, such as AFM [131], mass spectroscopy, and electrochemical techniques, and this trend is expected to continue. For example, SPR coupled to mass spectroscopy (MALDI-TOF, matrix-assisted laser desorption= ionization time-of-£ight) enables the characterization of small amounts of bioactive substances immediately following an SPR experiment [132]. The e¡ort of combining SPR with electrochemical techniques has also just started. Many fundamental issues on the SPR response to various electrochemical phenomena that take place on or near the electrode surfaces need to be resolved. Many well-de¢ned crystal surfaces are reconstructed and the reconstructed phase may transform into the ideal 11 phase at certain potentials. It is not clear how SPR depends on the surface reconstruction. Another basic issue is the relationship between the surface charge and SPR angular shift. Earlier work has shown that the surface charge changes SPR angle via changing the plasmon frequency. A systematic study of the e¡ect for various well-de¢ned single-crystal surfaces has not yet been carried out. In this regard, the Otto con¢guration is probably more appropriate than the more widely used Kretschmann con¢guration because it allows singlecrystal surfaces to be studied directly. Other basic issues, including the e¡ects of double-layer and solvent structure near the electrode surfaces on the SPR, also need to be understood. Technical improvement in the present SPR technique is expected to continue, which will widen the applications. There is still room to improve both the resolution and the response time of current SPR setups. Better resolution will allow SPR to detect smaller amounts of adsorbed molecules, or smaller amounts of changes in the adsorbed molecules. Faster response time will allow very fast dynamic processes taking place on electrode surfaces to be measured. Another direction to improve the current SPR technique is to introduce new capabilities. Using multiwavelength SPR to study the electronic states of adsorbed molecules is one possibility. Other useful improvements for SPR technique are the utilization of waveguide technology on SPR technique, which could help the development of compact and rugged sensing elements with the possibility of fabricating multiple sensors on one chip. Finally, it is important to note that SPR can be used as surface plasmon microscopy [133^135],which allows one to map local SPR change. Although

242

Wang et al.

the lateral resolution is limited by di¡raction, the vertical resolution is enough to resolve monolayer coverage of molecules. This approach is particularly useful for quick screening of complex systems. Another variant of SPR technique is coupling of plasmon resonance in a metallic thin ¢lm and waveguide modes in a dielectric layer coated on the metallic ¢lm [101,136]. The technique has been referred to as coupled plasmon-waveguide resonance (CPWR) spectroscopy. In addition to enhanced sensitivity and resolution, it has the capability to measure anisotropies in the refractive index and the optical absorption constants of adsorbed molecules. APPENDIX A: FUNDAMENTALS OF SPR Figure 1 depicts an interface between a metal and a transparent dielectric medium,with dielectric constants, em and e1 , respectively.The surface charge density wave associated with the surface plasmon propagating along the interface (de¢ned as the x direction) is given by sðx; t Þ ¼ s0 exp½iðkx x  otÞ, where kx and o are wavevector and angular frequency of the plasmon, respectively. The surface charge density generates electromagnetic waves in both the metal and the dielectric medium. Because the charge is symmetric along the y direction (Fig. 1), any electric ¢eld in the y direction would break the symmetry, which means that Ey ¼ 0 in both media. The remaining components, Ex and Ez , induce a magnetic ¢eld component in the y direction or perpendicular to the direction of propagation [42]. Such an electromagnetic wave is called transverse magnetic ¢eld (TM) mode. The electric ¢eld components of theTM mode can be written as 1 1 ^z Þ exp½iðkx x þ kz1 z  otÞ E1 ¼ ðE0x x^ þ E0z for the dielectric medium ðz > 0Þ

ðA1Þ

m m ^z Þ exp½iðkx x þ kzm z  otÞ Em ¼ ðE0x x^ þ E0z

for the dielectric medium ðz < 0Þ

ðA2Þ

Substituting Eqs. (A1) and (A2) into the wave equation [42], H2 E 

e @2E ¼ 0; c 2 @t 2

we have  2 o2 e1 kx2 þ kz1 ¼ c  2 o2 kx2 þ kzm ¼ em c

in the metal

ðA3Þ

in the dielectric medium

ðA4Þ

Surface Plasmon Resonance Spectroscopy

243

The ¢elds given by Eqs. (A1) and (A2) must also match the boundary condi? tions, E==1 jz¼0 ¼ E==m jz¼0 and e1 E? 1 jz¼0 ¼ em Em jz¼0 at the interface, so we have 1 m ¼ Ex0 Ex0

and

1 m e1 Ez0 ¼ em Ez0

ðA5Þ

Since there is no external charge in both media, H E ¼ 0 (Gauss’s law) [42], into which we substitute Eqs. (A1) and (A2) and obtain 1 1 ikx Ex0 þ ikz1 Ez0 ¼0

and

m m ikx Ex0 þ ikzm Ez0 ¼0

ðA6Þ

Combining Eqs. (A5) and (A6),we have kz1 em ¼ kzm e1,which we combine with Eqs. (A3) and (A4) to obtain the dispersion relation of the surface plasmon:

o2 e e 1 m ðA7Þ kx2 ¼ c e1 þ em 1 For a typical metal (em ¼ 11:7 for silver [47]) and transparent medium 2 o 2 for BK7 glass), e < 0 and e , so k > e and (e1 ¼ 1:515 > e j j m m 1 1 x c 2  2 kx2 > oc em . This leads to two important conclusions. First, both kz1 2 and kzm are negative, and kz1 and kzm are imaginary according to Eqs. (A3) and (A4). Letting kz1 ¼ ia1 and kzm ¼ iam, Eqs. (A1) and (A2) becomes E1 ¼ ½ðE10x x^ þ E10z ^z Þ expða1 zÞ exp½iðkx x  otÞ ^ þ Em z Þ expðam zÞ exp½iðkx x  otÞ Em ¼ ½ðEm 0x x 0z ^ where a1 and am are given by rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

o2 a1 ¼ kx2  e1 c rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

o2 ffi 2 am ¼ kx  em c

for z > 0 ðA10 Þ for z < 0 ðA20 Þ

ðA30 Þ ðA40 Þ

according to Eqs. (A3) and (A4). It is clear that the amplitude of the TM mode decays exponentially away from the interface in both media, so theTM waves associated with the surface plasmons are surface waves. Sommerfeld predicted the existence of surface electromagnetic waves as early as 1909 [137]. The decay lengths of the TM waves into the metal and the transparent dielectric medium are 1=a1 and 1=am, respectively. The second important conclusion is that the wavevector kx of the surpffiffiffiffiffi face TM mode is always greater than the wavevectors, k1bulk ¼ oc e1 and ffiffiffiffiffi p bulk ¼ oc em of the bulk electromagnetic waves in the two media. Because km of the wavevector mismatch, it is impossible to excite the surface plasmons with a bulk electromagnetic wave. However, one can use a number of ingenious schemes to excite surface plasmons. The popular one is the so-called Kretschmann con¢guration [43], in which a bulk electromagnetic wave is coupled to the surface plasmons via a prism (Fig. 1). The x component of the wavevector of the bulk electromagnetic wave is

244

Wang et al.

opffiffiffiffiffi ðA8Þ e2 sinðyÞ c where y is the angle of the incident light and e2 is the dielectric constant of the pffiffiffiffiffi prism (refractive index n2 ¼ e2).When the refractive index is large enough, the x component of the wavevector (kx2 bulk ) of the bulk electromagnetic wave can match the wavevector (kx ) of the surface plasmon, and

opffiffiffiffiffi

orffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi e1 em e2 sinðyR Þ ¼ c c e1 þ em bulk k2x ¼

or sinðyR Þ ¼

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi e1 em ðe1 þ em Þe2

ð3Þ

yR in Eq. (3) is the incident angle at which the surface plasmon resonates with the incident light or the so-called resonant angle. For a BK7 prism, silver ¢lm, and water as the dielectric medium, e2 ¼ n22 ¼ 2:30, e1 ¼ n21 ¼ 1:515, and em ¼ 11:7 for an incident light of wavelength 630 nm, yR ¼ 72:2 . Thus, the surface plasmons at silver ^ water interface can be easily excited with the help of a BK7 prism. Using Eqs. (4) and (8), we found a1  0:005 nm1, and the decay length is about 200 nm,which means that the presence of molecules within  200 nm from the surface can be detected with SPR. APPENDIX B: A SIMPLE SPR SETUP BASED ON BI-CELL DETECTION As discussed in Sec. 4, there are many di¡erent SPR setups. Here we provide a detailed description of the experimental setup of the bi-cell SPR that we have used in our laboratory. We start with a summary of some basic characteristics of the bi-cell SPR, and then discuss experimental details. B.1

Linearity

The ratio of the di¡erential to sum signals, (AB)=(A þ B), is the measured quantity in the bi-cell SPR, which is linearly proportional to the shift of the resonance angle, Dy, for small Dy. This can be understood by the following simple analysis. Near the resonance angle, the re£ectivity versus incident angle is given by RðDyÞ ¼ Rð0Þ þ

dRð0Þ 1 d 2 Rð0Þ Dy þ ðDyÞ2 þ dDy 2 dDy2

ðB1Þ

Surface Plasmon Resonance Spectroscopy

245

The ¢rst term is the re£ection at the resonance angle, which is a small constant, and the second term is zero because the re£ectivity at the resonance angle is minimum. By keeping the quadratic term, simple integrations lead to DyR 1 A  B ðB2Þ ¼ 6A þ B y0 where y0 is the convergent angle of the incident light. The relation of (AB)=(A þ B) versus DyR has been examined by numerical simulations based on Fresnel optics using a matrix method [54]. For a relative small angular shift, DyR=y0 < 0.2, the ratio (AB)=(A þ B) is linearly proportional to DyR. For y0 ¼ 5, the linear angular range is 1, which is large enough for most biosensor applications. For applications that require larger angular range, one can increase the convergent angle of the incident light, y0. The linear relation was also con¢rmed experimentally by comparing the data measured by the bi-cell detection setup with those obtained with a home-made conventional LDA SPR [54]. B.2

Response Time

The response time of the setup reported by Tao et al. was 1 ms, which was limited by the bandwidth of the preampli¢er. The bi-cell photodetector they used actually has a response time of 10 ns, so further improvement can be made. The LDA- or CCD-based method has a typical response time of milliseconds. The response time of the AOTF (acoustooptic tunable ¢lter) method is currently limited by the response time of the photodetector as well as by the modulation frequency of AOTF and the bandwidth of the lock-in ampli¢er [71]. B.3

Angular Resolution

The angular resolution of the bi-cell detection method is about 5  106 degrees, or 3  108 RIU for a bandwidth of 1 Hz. To the best of our knowledge, this angular resolution is much higher than the reported resolution achieved with other angle-scan systems [86,138]. By eliminating various errors discussed below, further improvement of the resolution is possible. Errors in the SPR methods come from three main sources: intensity £uctuation in the incident light; noise of the photodetector and related electronics circuit; and mechanical vibration and thermal drift of the setup. The intensity £uctuation in a typical diode or He-Ne laser is between 0.01% and 1%,which is a serious source of errors.This problem is greatly reduced in the bi-cell detection because the noise is largely subtracted out in the di¡erential signal. The detection of di¡erential signal also makes the bi-cell setup

246

Wang et al.

immune to noise due to ambient background light. These advantages are not shared by other methods using either a single or an array of photodetectors. Noise in the photodetector and its electronics circuit is another source of errors. The noise in AB due to the photodetector and electronics reported by Tao et al. has a root-mean-square (RMS) value of 1 mV, corresponding to an angular resolution of 107 degrees (109 RIU) for a 5-mW diode laser. The intrinsic statistical £uctuation in the number of photons pffiffiffiffi limits the ultimate resolution of SPR, which is given by DN =N ¼ 1= N , where N is the average number of photons entering the detector and DN is the standard deviation from the average. For a 5-mW laser diode emitting at a wavelength offfi 670 nm, this intrinsic £uctuation gives (AB)=(A þ B)  108 pffiffiffiffiffiffi degrees= Hz. This noise is clearly not yet the limiting factor of the angular resolution. The last source of errors is from mechanical vibrations and thermal drift of the system, which depend on the design of the setup. The mechanical vibrations come from the table on which the SPR setup is supported and from acoustic noise in the surrounding environment. Both can cause £uctuations in the positions of the optical components. The thermal drift includes thermal expansion of the prism and the holders as well as temperaturedependent refractive indexes of the prism, metal ¢lm, and solution. Temperature change due to laser heating is expected initially, but it should eventually reach equilibrium. The mechanical vibrations and thermal drift are the major resolution-limiting factors of the setup of Tao et al. [54]. B.3

Experimental Details

The bi-cell SPR setup consists of three major parts, a light source, a prism, and a photodetector. A laser diode with a few milliwatts output power can be used as light source. The wavelength of the laser can be either red (e.g., 635 nm) or infrared. Much shorter than 635 nm is not desirable because of increasing absorption of the incident light by Au and Ag ¢lms at short wavelengths. Furthermore, short wavelengths result in shifts of the re£ected light at resonance angle to the edge of the prism, thus causing di⁄culty in the measurement. The output beam from the diode is usually elliptical and polarized either along or perpendicular to the long axis of the ellipse.We have used a diode with a polarization along the short axis. The output beam is typically divergent and needs to be focused with a lens. Many vendors (e.g., Thorlabs, Inc., www.thorlabs.com) sell laser diode modules that include the laser diode, the related lens assemblies, as well as a built-in power supply and stabilization circuits. A BK-7 hemicylindrical lens ( 1 in.) from Melles Griot (part number 01LCP004) can be used as a prism. Onto the £at surface of the prism a

Surface Plasmon Resonance Spectroscopy

247

microscope cover slide (18 18  0.2 mm, e.g., Fisher Scienti¢c, #12-540A) coated with either Au or Ag ¢lm is placed with the help of a drop of indexmatching £uid used for oil-immersion objectives in optical microscopes. The optimal thickness of the metal ¢lm that gives the best SPR dip depends on the material and wavelength of the incident light. At 635-nm incident light, the optimal thickness is  48.5 nm for Au and  49 nm for Ag. Although Au and Ag ¢lms are commercially available, an inexpensive sputtering coater (from Kurt J. Lesker Company, model 108; with a thickness monitor from Cressington, model MTM10) used for scanning electron microscopy produces good metal ¢lms for SPR experiments. In the case of Au ¢lms, it is desirable to heat each ¢lm with H2 £ame for a few seconds before the experiment.On the metal ¢lm, a solution cell that is appropriate to each experiment is attached. Light re£ected from the metal ¢lm into the prism is detected with a bicell photodetector that consists of two rectangular photocells integrated on a single chip with a  50 mm separation (Hamamatsu, model S2721-02). The two photocells are nearly identical (<1%) so that common noise is largely subtracted out. The outputs, A and B, from the photocells are connected to an electronics circuit that calculates AB with at variable gain (10^100) and A þ B. Both AB and A þ B are recorded, by a digital oscilloscope or a computer equipped with a data acquisition board, and the ratio AB=A þ B is calculated digitally. Before the experiment, the re£ected light falling on the photocells A and B is ¢rst balanced such that AB is close to zero with a translation stage (Melles Griot, 07TMC501 or o7TMC502) on which the bicell detector is mounted. Because of the high sensitivity of the setup, drift in the signal usually follows the initial alignment, but it eventually settles down to a tolerable level. REFERENCES 1.

EF Bowden, FM Hawkridge, HN Blount. Comprehensive Treatise of Electrochemistry.Vol. 10. New York: Plenum, 1985. 2. A Heller. Accts Chem Res 23:128^132, 1990. 3. DE Reed, FM Hawkridge. Anal Chem 2334^2339, 1987. 4. AEF Nassar,WS Willis, JF Rusling. Anal Chem 67:2386^2392, 1995. 5. WH Scouten, JHT Luong, RS Brown. Tibtech 13:178^185, 1995. 6. MJ Eddowes, HAO Hill. J Chem Soc Chem Commun 771^772, 1977. 7. MJ Eddowes, HAO Hill. J Am Chem Soc 101:4462^4464, 1979. 8. I Taniguchi, K Toyosawa, H Yamaguchi, K Yasukouch. J Chem Soc Chem Commun 1032^1033, 1982. 9. FA Armstrong, HAO Hill, NJ Walton. Accts Chem Res 21:407^413, 1988. 10. AM Bond, HAO Hill, S Komorsky-Lovric, M Lovric, ME McCarthy, SM Psalti, NJ Walton. J Phys Chem 96:8100^8107, 1992.

248

Wang et al.

11.

SM Amador, JM Pachence, R Fischetti, JP McCauley, AB Smith, JK Blasie. Langmuir 9:812^817, 1993. ZQ Feng, S Imabayashi,T Kakiuchi, K Niki. J Electroanal Chem 394:149^154, 1995. TM Nahir, EF Bowden. J Electroanal Chem 410:9^13, 1996. PL Edmiston, JE Lee, S-S Cheng, SS Saavedra. J Am Chem Soc 119:560^570, 1997. ZQ Feng, S Imabayashi, T Kakiuchi, K Niki. J Chem Soc, Faraday Trans 93:1367^1370, 1997. JK Cullison, FM Hawkridge, N Nakashima, S Yoshikawa. Langmuir 10:877^ 882, 1994. K Tanaka, R Tamamushi. J Electroanal Chem 236:305^307, 1987. MT Rojas, M Han, AE Kaifer. Langmuir 8:1627^1633, 1992. J-M Kaufmann, O Chastel, G Quarin. Bioelectrochem Bioenerget 23:167^172, 1990. J Wang, Z Lu. Anal Chem 62:826^829, 1990. AEF Nassar, Z Zhang, N Hu, JF Rusling, TF Kumosinski. J Phys Chem B 101:2224^2231, 1997. J Rusling, A-EF Nassar. J Am Chem Soc 115:11891^11897, 1993. Z Salamon, LB Vitello, JE Erman. Bioelectrochem Bioenerget 21:213^224, 1989. Z Salamon, JT Hazzard, G Tollin. Proc Natl Acad Sci USA 90:6420^6423, 1993. Z Salamon,Y Wang, G Tollin. Biophys J 71:283^289, 1996. Z Salamon, G Tollin. Biophys J 71:858^867, 1996. RJ Fisher, M Fivash. Curr Biotechnol 45:65^71, 1994. K Mo¡at, G Wagner. Curr Opin Struct Biol 5:637^639, 1995. A Szabo, L Stolz, R Granzow. Curr Opin Struct Biol 5:699^705, 1995. TD Barley, RW Hunt, AA Welcher,WJ Boyle,VP Parker. Nature 368:558^560, 1994. M Nrksich, GB Sigal, GM Whitesides. Langmuir 11:4383^4385, 1995. CE Jordan, LF Brian, S Kornguth, RM Corn. Langmuir 10:3642^3648, 1994. H Sota,Y Hasegawa, M Iwakura. Anal Chem 70:2019^2024, 1998. SM Barbas, HJ Ditzel, EM Salone, W Yang, GJ Silverman, DR Burton. Proc Natl Acad Sci USA 92:2529^2533, 1995. L Haussling, H Ringsdorf, F-J Schmitt,W Knoll. Langmuir 7:1837^1840, 1991. S Boussaad, J Pean, NJ Tao. Anal Chem 72:222^226, 2000. S Wang, S Boussaad, S Wong, NJ Tao. Anal Chem 72:4003^4008, 2000. A Otto. Z Phys 216:398^407, 1968. E Kretschmann. Z Phys 241:313^324, 1971. H Raether. PhysThin Films 9:145^261, 1977. YR Shen. The Principles of Nonlinear Optics. New York: Wiley, 1984. DJ Gri⁄ths. Introduction to Electrodynamics. Englewood Cli¡s, NJ: Prentice Hall, 1999. E Kretschmann. Opt Commun 26:1^11, 1978.

12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43.

Surface Plasmon Resonance Spectroscopy

249

44. WN Hansen. J Opt Soc Am 58:380^390, 1969. 45. HE Bruijn, BSF Altenburg, RPH Kooyman, J Greve. J Opt Commun 82:425^ 432, 1991. 46. JH Grassi, RM Georgiadis. Anal Chem 71:4392^4396, 1999. 47. ED Palik, ed. Handbook of Optical Constants of Solids II. Boston: Academic, 1991. 48. C Kittle. Solid State Physics. New York: Wiley, 1996. 49. HE Bruijn, M Minor, RPH Kooyman, J Greve. Opt Commun 95:183^188, 1993. 50. KA Peterlinz, R Georgiadis. Opt Commun 130:260^266, 1996. 51. KA Peterlinz, RM Georgiadis. J Am Chem Soc 119:3401^3402, 1997. 52. TM Chinowsky, SS Yee. Sensors Actuators B 51:321^330, 1998. 53. R Kotz, DM Kolb, JK Sass. Surface Sci 69:359^364, 1977. 54. NJ Tao, S Boussaad, WL Huang, RA Arechabaleta, J D’Agnese. Rev Sci Instrum 70:4656^4660, 1999. 55. RW Wood. Phil Mag 4:396^402, 1902. 56. MJ Jory, PS Vukusic, JR Sambles. Sensors Actuators B 17:1203^1209, 1994. 57. J Homola, I Koudela, SS Yee. Sensors Actuators B 17:16^24, 1999. 58. JGIaS Gordon, JD Opt. Commun 22:374^377, 1977. 59. J Homola, SS Yee, G Gauglitz. Sensors Actuators B 54:3^15, 1999. 60. C Nylander, B Liedberg,T Lind. Sensors Actuators 3:79^88, 1982. 61. MMB Vidal, R Lopez, S Aleggret, J Alonso-Chamarro, I Garces, J Mateo. Sensors Actuators B 11:455^459, 1993. 62. S Herminghaus, P Leiderer. Appl Phys Lett 58:352^354, 1991. 63. K Masubara, S Kawataa, S Minami. Appl Opt 27:1160^1163, 1988. 64. W Hickel,W Knoll. Thin Solid Films 199:367^374, 1991. 65. KS Johnston, KS Booksh,TM Chinowsky, SS Yee. Sensors Actuators B 54:80^ 88, 1999. 66. TM Chinowsky, LS Jung, SS Yee. Sensors Actuators B 54:89^97, 1999. 67. K Johansen, R Stalberg, I Lundstrom, B Lieberg. Measure Sci Technol 11:1630^1638, 2000. 68. R Garabedian, C Gonzalez, J Richards, A Knoesen, R Spencer, SD Collins, RL Smith. Sensors Acutators A 43:202^207, 1994. 69. G Meyer, NM Amer. Appl Phys Lett 53:1045^1047, 1988. 70. S Alexander, L Hellemans, O Marti, J Schneir, V Elings, PK Hansma, M Longmire, J Gurley. J Appl Phys 65:164^171, 1989. 71. F Caruso, MJ Jory, GW Bradberry, JR Sambles, DN Furlong. J Appl Phys 83:1023^1028, 1998. 72. MJ Jory, GW Bradberry, PS Cann, JR Sambles. Measure Sci Technol 6:1193^ 1200, 1995. 73. RC Jorgenson, SS Yee. Sensors Actuators B 12:213^220, 1993. 74. T Weimar. Angew Chem Int Ed 39:1219^1221, 2000. 75. RL Rich, DG Myszka. Curr Opin Biotechnol 11:54^61, 2000. 76. DG Myszka. J Mol Recognit 12:390^408, 1999. 77. MN Weiss, R Srivastava, H Groger. Electron Lett 32:842^843, 1996.

250

Wang et al.

78. 79.

B Chadwick, M Gal. Jpn J Appl Phys 32:2716^2717, 1993. EG Ruiz, I Garcez, C Aldea, MA Lopez, J Mateo, J Alonso-Chamarro, S Alegret. Sensors Actuators A 37^38:221^225, 1993. S Miwa,T Arakawa. Thin Solid Films 281^282:466^468, 1996. GJ Ashwell, MPS Roberts. Electron Lett 32:2089^2091, 1996. C Granito, JN Wilde, MC Petty, S Houston, PJ Iradale. Thin Solid Films 284^285:98^101, 1996. NE Agbor, JP Cresswell, MC Petty, MAP. Sensors Actuators B 41:137^141, 1997. CC Jung, SB Saban, SS Yee, RB Darling. Sensors Actuators B 32:143^147,1996. TM Chinowsky, SB Saban, SS Yee. Sensors Actuators B 35:37^43, 1996. B Liedberg, C Nylander, I Lundstrom. Sensors Actuators 4:299^304, 1983. I Lundstrom. Biosensors Bioelectron 9:725^736, 1994. ML Shank-Retzla¡, SG Sligar. Anal Chem 72:4212^4220, 2000. AH Severs, RBM Schasfoort, MHL Salden. Biosensors Bioelectron 8:185^189, 1993. A Brecht, G Gauglitz. Anal Chim Acta 347:219^233, 1997. DR Mernagh, P Janscak, K Firman, GG Kneale. Biol Chem 379:497^503, 1998. RJ Pei, XQ Cui, XR Yang, EK Wang. Talanta 53:481^488, 2000. D Altschuh, M-C Dubs, E Weiss, G Zeder-Lutz, MHV Van Regenmortel. Biochemistry 31:6298^6304, 1992. W Knoll et al. Colloids Surfaces A
Physicochem Eng Aspects 161:115^137, 2000. P Schuck, AP Minton. Anal Biochem 240:262^272, 1996. Z Salamon, Y Wang, G Tollin, HA Macleod. Biochim Biophys Acta
Biomembranes 1195:267^275, 1994. Z Salamon,TE Meyer, G Tollin. Biophys J 68:648^654, 1995. Z Salamon, Y Wang, MF Brown, HA Macleod, G Tollin. Biochemistry 33:13706^13711, 1994. Z Salamon, G Tollin. Biophys J 71:848^857, 1996. JL Soulages, Z Salamon, MA Wells, G Tollin. Proc Natl Acad Sci USA 92:5650^5654, 1995. Z Salamon, D Huang,WA Cramer, G Tollin. Biophys J 75:1874^1885, 1998. A Muga, HH Mantsch,WK Surewicz. Biochemistry 30:7219^7224, 1991. LR Brown, K Wuthrich. Biochim Biophys Acta 468:389^410, 1977. T Heimburg, P Hildebrandt, D Marsh. Biochemistry 30:9084^9089, 1991. S Boussaad, L Dziri, R Arechabalea, NJ Tao, RM Leblanc. Langmuir 14:6215^ 6219, 1998. S Boussaad, NJ Tao, R Arechabaleta. Chem Phys Lett 280:397^401, 1997. S Heyse, OP Ernst, Z Dienes, KP Hofmann, H Vogel. Biochemistry 37:507^ 522, 1998. T Ozawa, K Sasaki,Y Umezawa. Biochim Biophys Acta
Protein Struct Mol Enzymol 1434:211^220, 1999.

80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108.

Surface Plasmon Resonance Spectroscopy 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138.

251

T Ozawa, M Fukuda, M Nara, A Nakamura, Y Komine, K Kohama, Y Umezawa. Biochemistry 39:14495^14503, 2000. DG Hanken, CRM. Anal Chem 69:3665^3673, 1997. Y Iwasaki,T Horiuchi, M Morita, O Niwa. Surface Sci 427^428:195^198, 1999. A Badia, S Arnold, V Scheumann, M Zizlsperger, J Mack, G Jung, W Knoll. Sensors Actuators B 54:145^165, 1999. DD Schlereth. J Electroanal Chem 464:198^207, 1999. DD Schlereth, RPH Kooyman. J Electroanal Chem 444:231^240, 1998. P Chin, SS Brody. Biochemistry 14:1190^1193, 1975. E Margoliash, A Schejter. Adv Protein Chem 21:113^283, 1966. WR Fisher, H Taniuchi, CB An¢nsen. J Biol Chem 248:3188^3195, 1973. D Eden, JB Matthew, JJ Rosa, FM Richards. Proc Natl Acad Sci USA 79:815^ 819, 1982. J Trewhella,VAP Carlson, EH Curtis, DB Heidon. Biochemistry 27:1121^1125, 1988. T Takano, RE Dickerson J Mol Biol 153:79^115, 1981. Y Feng, H Roder, SW Englander. Biochemistry 29:3494^3504, 1990. I Pockrand, JD Swalen, R Santo, A Brillante, MR Philpott. J Chem Phys 69:4001^4011, 1978. CRaS Cantor, PR Biophysical Chemistry, Part II: Techniques for the Study of Biological Structure and Function. San Francisco: Freeman, 1980. GU Bublitz, SG Boxer. Annu Rev Phys Chem 48:213^242, 1997. T Satoshi, S Igarashi, H Sato, K Niki. Langmuir 7:1005^1012, 1991. E Ngameni, LAM L’Her. J Electroanal Chem 301:207^226. T Sagara, HX Wang, K Niki. J Electroanal Chem 364:285^288, 1994. H Kano, S Kawata. Appl Opt 33:5166^5170, 1994. AA Kolomenskii, PD Gershon, HA Schuessler. Appl Opt 39:3314^3320, 2000. S Wang, S Boussaad, S Wong, NJ Tao. Unpublished. X Chen, MC Davies, CJ Robert, KM Shakeshe¡, SJB Tendler, PM Williams. Anal Chem 68:1451^1455, 1996. CP Sonksen, E Nordho¡, O Jansson, M Malmqvist, P Roepstor¡. Anal Chem 70:2731^2736, 1998. B Rothenha«usler,W Knoll. Nature 332:615^617, 1988. H Morgan, DM Taylor. Appl Phys Lett 64:1330, 1994. W Hickel,W Knoll. Appl Phys Lett 57:1286^1288, 1990. Z Salamon, HA Macleod, G Tollin. Biophys J 73:2791^2797, 1997. A Sommerfeld. Ann Phys 28:665, 1909. B Liedberg, C Nylander, I Lundstrom. Biosensors Bioelectron 10:i ^ ix, 1995.

6 Characterization of Biomolecular Interfaces with Scanning Electrochemical Microscopy: From Model Monolayers to Tissues and Cells Anna L. Barker, Catherine E. Gardner, Julie V. Macpherson, Patrick R. Unwin, and Jie Zhang University of Warwick, Coventry, United Kingdom

1

INTRODUCTION

Scanning electrochemical microscopy (SECM) has developed into a powerful technique for quantitative investigations of interfacial physicochemi cal processes, in a wide variety of areas, as considered in several recent reviews [1^6]. This chapter will provide a background to SECM and its application in characterizing biomolecular ¢lms and related interfaces. Since many SECM techniques are generic, we consider their application to biomolecular systems in general. In the simplest terms, SECM involves the use of a mobile ultramicroelectrode (UME) probe, either amperometric or potentiometric, to investigate the activity and/or topography of an interface on a localized scale. The origins of the technique can be traced, at least in part, to initial work by Engstrom’s group in the mid-1980s [7,8], who showed that an amperometric UME could be used to probe the concentration boundary layer of a larger active electrode. Simultaneously, Bard’s group was developing in-situ electrochemical scanning tunneling microscopy (STM), to permit local electrochemical measurements at the STM tip [9]. Outside the 253

254

Barker et al.

electrochemistry ¢eld, tiny mobile amperometric electrodes were being used even earlier, for example, to measure gas transfer close to water surfaces [10]. The attractive features of SECM, for the study of biomolecular ¢lms on a local scale, were recognized soon after the technique was formally established [11^13]. Early applications included quantitative studies of immobilized enzyme activity [14^18], photosynthetic processes on leaves [19,20], and investigations of polymer ¢lms in general [21^27]. These studies provided the foundations for a myriad of further related applications, which have served to diversify the range of processes and interfaces that can be studied. It is the aim of this chapter to provide an overview of the application of SECM to biologically relevant ¢lms and interfaces, as well as highlighting promising areas for the future. Many of the biomolecular interfaces amenable to study by SECM can be understood by reference to Fig. 1, which shows examples of the various levels of complexity of biologically relevant interfaces that can be investigated by SECM and UME methods.These range from various tissues to cells, bilayers, and monolayer systems. At the same time, biomolecular interfaces relevant to analysis, sensor design, and micropatterning may also be investigated.Consequently, this chapter provides a holistic view of the application of SECM to biomolecular interfaces, covering systems that are not

FIG. 1 Pictorial representation of the various levels of complexity of a biologically relevant interface, in this case articular cartilage, that can be investigated by SECM.

Characterization of Biomolecular Interfaces with SECM

255

considered in other chapters in this volume, but are necessary to treat SECM comprehensively. We begin with a basic overview of the principles of SECM, including the instrumentation and methods. Applications of SECM are then considered, starting with studies of immobilized biomolecules and simple monolayers, progressing to bilayers, and ¢nally covering synthetic membranes, tissues, and cells. In this way we begin with the simplest molecular interfaces that are amenable to study with SECM and ¢nish with living systems,which represent a huge area for potential applications of the technique.

2

PRINCIPLES

Several modes of SECM have been developed to allow the local chemical properties of interfaces to be investigated. A comprehensive review of all of the techniques can be found in [28]. Here we consider those methods that are most important in the study of biomolecular interfaces. 2.1

Concentration Mapping

As mentioned brie£y above, both potentiometric and amperometric probe electrodes may be used in SECM. At the simplest level, both types of probes can be used to map local variations in concentrations above a target inter face. In this type of application, the detector probe is generally assumed to be passive (i.e., nonperturbing to the interfacial process).This usually applies to potentiometric tips, provided they are su⁄ciently small not to impede mass transport to and from the target interface. Amperometric tips may also be considered to be relatively noninvasive when the di¡usion ¢eld associated with the tip is small compared to the tip ^ substrate separation and the size of the active site being investigated. Methods for making small tips are considered in Sec. 3.3. Alternatively, enhanced spatial resolution can be obtained by operating the probe electrode in chronoamperometric mode and deriving information from the form of the short-time response [29]. As discussed in Sec. 8, tip detection (or collection) measurements have proved particularly powerful for identifying localized transport pathways in synthetic membranes and biological tissues. The approach is also useful for investigating the activity of immobilized enzymes, when the kinetics are slow and traditional techniques, such as the feedback mode, are dominated by a di¡usional background signal (see Sec. 5.1). Tip collection measurements have also been used to probe the ingress and egress of counterions at conducting polymers [25,26,30,31]. Nanoscopic amperometric tips have allowed the redox properties of a polymer ¢lm to be probed directly, by controllably positioning the tip in the ¢lm, in penetration experiments [23,32].

256

2.2

Barker et al.

Negative Feedback Mode

Many applications of SECM involve using the tip to locally perturb an interfacial process, by electrolysis or ion transfer, and determining the kinetic e¡ect from the resulting tip current. In this situation, the tip is usually held at a potential to drive the detection of a target analyte at a di¡usionlimited rate. The baseline response for these measurements, when the interface is inert with respect to the tip-detected species, is termed ‘‘negative feedback’’ [11,13] and it is useful to consider this, by way of introduction. When the tip is positioned a long way from the target interface, d > 10a, where d is the tip ^ interface distance and a is the electrode radius, it behaves as a conventional UME. In this situation, a steady-state current, ið1Þ, is rapidly established due to hemispherical di¡usion of the target species (Red in Fig. 2a). As the tip is brought close to an interface which is inert with respect to the species involved in the electrode process, di¡usion to the UME becomes hindered (Fig. 2b) and the steady-state current, i, decreases compared to ið1Þ. In general, measurements of i=ið1Þ as a function of d are termed ‘‘approach curves.’’ Since the dependence of the i=ið1Þ ratio on d and the tip geometry can be calculated theoretically [11,12],simple current measurements with mediators whichdo not interact at the interface canbe used to provide information on the topography of the sample of interest. In this application, an amperometric UME is typically scanned at a constant height above the target interface (x ^ y plane) and the di¡usion-limited current for electrolysis of the target species is measured.This, in turn, can be related to the distance between the tip and the interface, from which topographical information is obtained. When either the solution species of interest, or tip product(s), interact with the target interface, the hindered mass transport picture of Fig. 2b is modi¢ed. The e¡ect is manifested in a change in the tip current, which is the basis of using SECM to investigate interfacial reactivity. Under these conditions, independent methods for determining topography of the sample are often useful. Recently developed nonelectrochemical methods of imaging sample topography, appear to be particularly promising (see Sec. 10). There are primarily three ways in which an amperometric electrode can be used to simultaneously induce and monitor interfacial processes. These are illustrated schematically in Fig. 3 for the most general case, where di¡usion may occur in both of the phases,which comprise the interface of interest. These basic mass transport pictures are applicable to the situation where the liquid phase containing the UME is in contact with a second phase,which has £uidlike transport properties (e.g., a second immiscible liquid, biomaterial, or gas). Although a redox reaction is considered at the tip, similar experiments may be carried out with ion-transfer voltammetric probes. Transport

Characterization of Biomolecular Interfaces with SECM

257

FIG. 2 (a) Schematic of the hemispherical diffusion-field established for the steady-state diffusion-limited oxidation of a solution species, Red, at a disk-shaped UME, giving rise to a current ið1Þ. (b) When the UME is positioned close to an inert target interface, diffusion of Red is hindered and the current, i decreases. (Reproduced with permission from Ref. 5. Copyright 1999 Elsevier.)

processes in phase 2 can usually be neglected when phase 2 is a solid or a gas (due to the rapidity of gas transport compared to di¡usion in liquids). 2.3

Feedback Mode

Feedback mode, depicted in Fig. 3a, is one of the most widely used SECM techniques, applicable mainly to the study of interfacial redox processes [11],

258

Barker et al.

FIG. 3 Principal methods for inducing and monitoring interfacial processes with SECM: (a) feedback mode, (b) induced transfer, and (c) double-potential step chronoamperometry.

although feedback based on ion transfer has also been reported [33,34]. For redox processes, the basic idea is to generate a species at the tip in its oxidized or reduced state (generation of Ox1 in Fig. 3a), typically at a di¡usioncontrolled rate, by electrolysis of the other half of a redox couple (Red1). The tip-generated species di¡uses from the UME to the target interface. If it undergoes a redox reaction, which converts it to the original form, the

Characterization of Biomolecular Interfaces with SECM

259

mediator di¡uses back to the tip, thereby establishing a feedback cycle and enhancing the current at the UME. The redox reaction could occur at a ¢xed site on the interface, as in the case of immobilized oxidoreductase enzymes [14,15,35]. Alternatively, the reaction could require the di¡usion of a partner species in phase 2 to the interface (Red2 in Fig. 3a), as in the case of electron transfer at immiscible liquid/liquid interfaces. Both cases are considered in Secs. 5 and 6, respectively. 2.4

SECM Induced Transfer (SECMIT)

This technique, depicted schematically in Fig. 3b, can be used to characterize reversible phase-transfer processes at a wide variety of interfaces [36]. The basic idea is to perturb the process, initially at equilibrium, through local amperometry at the UME. Hitherto, this technique has mainly been used in conjunction with metal tips [36], but ion-transfer voltammetric probes can also be used [37]. The application of a potential to the tip, su⁄cient to deplete the species of interest in phase 1 (oxidation of Red1 to Ox1 in Fig. 3b), drives the transfer of species Red from phase 2 to phase 1.Collection of this species at the tip enhances the current £ow, compared to the situation where there is no net transfer across the target interface and species Red reaches the tip by hindered di¡usion through phase 1 only. For a given tip ^ interface separation, the overall current response is governed by di¡usion in the two phases and the interfacial kinetics [36].The technique was originally employed in a time-dependent potential step chronoamperometric mode to probe desorption processes at solid/liquid interfaces [38], and was subsequently shown to be a powerful probe of dissolution kinetics under both steady-state and time-dependent operation [39^45]. SECMIT was subsequently used to investigate the kinetics of solute transfer across interfaces formed between (1) biological tissues and a bathing solution [46^48], (2) two immiscible liquids [36,49], and (3) a liquid phase (sometimes with an adsorbed layer) in contact with a gas [50]. Many of these applications are considered in later sections of this chapter. When there are no kinetic limitations to the interfacial transfer process, SECMIT is also an e¡ective analytical technique for determining the permeability, concentration, and di¡usive properties of a solute in a target phase, without the UME having to enter or contact that phase, as discussed in Sec. 8. This is obviously advantageous for situations where direct voltammetric measurements would otherwise be impractical, for example, due to high resistivity or a limited solvent window of the sample. This aspect of SECMIT has also been used to great e¡ect when the intimate presence of the UME would damage the structural integrity of the sample under investigation, as in the case of biological tissues [47,48].

260

2.5

Barker et al.

Multistep Transient Methods

Both double potential step chronoamperometry (DPSC) [51^53] and triple potential step methods [54] have been used to investigate various interfacial processes in the SECM con¢guration. In homogeneous phases, DPSC at UMEs has proven powerful for characterizing the lifetimes of transient species down to the microsecond time scale [55] and the di¡usion coe⁄cients of electrogenerated species, independent of knowledge of the concentration and number of electrons transferred [56]. In the SECM geometry, the followup chemical reaction involving the tip-generated species is e¡ectively con¢ned to the interface under study.The basic concept, Fig. 3c, is to employ the UME to generate a reactive species in an initial (forward) step for a ¢xed period. The potential is stepped from a value where there are no redox reactions to one where Red1 is oxidized to Ox1 at a di¡usion-controlled rate. During this step,the tip-generated species (Ox1) di¡uses away from the UME and intercepts the interface. If Ox1 interacts with the interface (e.g., by adsorption, absorption, or a chemical reaction), its concentration pro¢le is modi¢ed compared to the situation where there is no interaction and Ox1 simply leaks out of the tip ^ interface gap by hindered di¡usion. Consequently, when the potential is reversed, in a ¢nal step, to collect Ox1 by electrolysis, the £ux at the UME, and the corresponding current ^ time characteristics, depend strongly on the nature of the interaction of Ox1 with the target interface. 3

INSTRUMENTATION

Although commercial instruments for SECM have recently become available from several companies, including CH Instruments (USA), Quesant Instruments (USA), and Uniscan (UK), the majority of instrumentation has been constructed by individual research groups, to a variety of speci¢cations. The advantage of building instruments from scratch is that the researcher can tailor the apparatus to particular applications. This is an important point, since the types of problems to which SECM can be applied continues to grow at a rapid rate. At the heart of SECM is the amperometric or potentiometric tip, whose location is controlled remotely, with appropriate positioners, relative to the sample interface. The type of cell or vessel in which measurements can be made depends on the type of interface to be addressed, ranging from a simple beaker [52] to a Langmuir trough in a controlled atmosphere [50,57]. Electrochemical control and measurement in SECM is relatively simple, as discussed in the next section. 3.1

Electrochemical Instrumentation

For amperometric control of the tip, with externally unbiased interfaces, a simple two-electrode system su⁄ces (Fig. 4a). A potential is applied to the

Characterization of Biomolecular Interfaces with SECM

261

FIG. 4 (a) Block schematic of the typical instrumentation for SECM with an amperometric UME tip. The tip position may be controlled with various micropositioners, as outlined in the text. The tip potential is applied, with respect to a reference electrode, using a potential programmer, and the current can often be measured with a simple amplifier (current follower) device. The tip position may be viewed using a video microscope. (b) Schematic of the ‘‘submarine’’ UME configuration, which facilitates interfacial electrochemical measurements when the phase containing the UME is more dense than the second phase. In this case, the glass capillary is attached to suitable micropositioners and electrical contact is made via an insulated copper wire. (Reproduced with permission from Ref. 5. Copyright 1999 Elsevier.)

tip, with respect to a suitable reference electrode, to drive the process of interest at the tip and the corresponding current that £ows is typically ampli¢ed by a current-to-voltage converter. If the sample is also to be biased externally, a bipotentiostat is required. For some studies of membrane transport, ion £ow is driven from a donor to a receptor compartment galvanostatically, and a potentiostatically controlled tip serves as a detector [58]. Potentiometric detection with UMEs of various types is readily

262

Barker et al.

FIG. 4

Continued.

accomplished [59,60], typically using a voltage follower with input impedance appropriate to the type of indicator electrode used. 3.2

Positioning

The tip is attached to positioners,which allow it to be moved and positioned relative to the interface under investigation. A variety of positioners have been employed in SECM instruments,with the choice depending on the type of measurement and spatial resolution required. For the highest (nanometer) resolution, piezoelectric positioners similar to those used in STM are mandatory [9]. Piezoelectric ‘‘inchworm’’ motors developed by Burleigh (Fishers, NJ) have proved the most popular choice for SECM instruments, o¡ering high-resolution positioning capabilities with long-range travel [11]. A photograph of a typical SECM device, built in our laboratory, based on this type of positioner (with integrated optical encoders), is shown in Fig. 5. Some of the key components involved in an SECM setup are identi¢ed and labeled in this ¢gure. There has also been some use of stepper motors to control the position of the tip in the x ^ y plane [61^63], parallel to the interface of interest.This approach has proved to be successful when the position of the translational stages can be monitored independently with encoder devices.

Characterization of Biomolecular Interfaces with SECM

263

FIG. 5 Photograph of the stages and cell for a typical SECM experiment: (a) vibrationally isolated breadboard, (b) stainless steel cell mounting poles, (c) aluminum plate, (d) closed-loop translational stages, (e) aluminium UME holder, (f) UME, (g) Ag quasi-reference electrode, and (h) fully detachable cell. (Reproduced with permission from Ref. 5. Copyright 1999 Elsevier.)

In the application of SECM at solid/liquid interfaces, high-resolution x, y, z positioning and scanning is usually required. However, many SECM measurements, e.g., at air/liquid interfaces, simply involve the translation of a tip toward and/or away from a speci¢c spot on an interface, in the perpendicular (z) direction. In this situation, it is only necessary to have high resolution z control of the tip, typically using a piezoelectric positioner, while manual stages su⁄ce for the other two axes [36,49]. It has further been shown that SECM measurements can be made with manual stages on all axes, with the z axes driven by a di¡erential micrometer and the x ^ y stages controlled by ¢ne adjustment screws.This simple cost-e¡ective setup allows tip approach measurements to be made with a spatial resolution of 0.25 mm [64,65]. The use of a video microscope, aligned such that the electrode may be observed from the side, has proved useful in facilitating the positioning of the UME probe relative to the interface of interest [36,49].

264

3.3

Barker et al.

Probes

The type of probe electrode used in SECM depends on the particular process under investigation. A diversity of probes is available for amperometry and potentiometry; since these often have to be prepared in house, we highlight some of the most important tip designs in this section. 3.3.1 Micrometer-Sized Disk-Shaped Electrodes Sealed in Glass Typically, amperometry involves electrolysis at a solid UME, usually a diskshaped electrode, with a diameter of 0.6^25 mm. This type of electrode is readily fabricated by sealing a wire of the material of interest in a glass capillary, making an electrical connection, and polishing the end £at [11,66,67]. Pt, Au, and C electrodes have been successfully fabricated in this way. For most SECM studies, the ratio of the diameter of the tip (electrode plus surrounding insulator, 2rs) to that of the electrode itself, 2a, RG ¼ rs/a, is typically around 10.This minimizes e¡ects from back di¡usion (from behind the probe), making the electrode response most sensitive to the surface process of interest and also simplifying the treatment of mass transport. It should be noted, however, that the larger the RG value, the more di⁄cult it is to place the active portion of UME close to the target interface, since the end of the probe and the interface can never be truly parallel. The smallest distance the tip can be placed with respect to the interface, in practice, is termed the ‘‘distance of closest approach.’’ SECM images may be convoluted with the both activity and topographical contributions.To resolve such e¡ects, it may be possible to scan the sample twice, with the mediator of interest and then with a moiety that is inert with respect to the sample, so mapping the topography [47,48]. Dual amperometric probes with one channel serving as a topographic sensor and the other to determine activity have also been crafted [68]. Such probes have found application for dual potentiometric amperometric/conductivity sensing [60], as described in Sec. 3.3.5. 3.3.2 Submarine Probes For some liquid/liquid interfaces and for studies at the water/air interface, a ‘‘submarine’’ electrode can be deployed [49^51], depicted schematically in Fig. 4b. In this case, the electrode is inverted in the cell, such that the tip points upwards, and an insulated connection is made through the solution. 3.3.3 Submicrometer and Nanometer-Scale Electrodes In order to improve the spatial resolution of SECM, there is much interest in reliable methods for shrinking the size of the probe electrode. Probably the simplest approach is to electrochemically etch a length of metal microwire to

Characterization of Biomolecular Interfaces with SECM

265

a sharp point, and insulate o¡ all but the end of the probe, leading to a conically shaped tip. This methodology is often employed in the fabrication of electrochemical scanning tunneling microscopy tips. A number of di¡erent coating procedures have been investigated, such as a simple dipping technique with a varnish [69] or molten para⁄n [70].Translation of the tip through a molten bead of glass [71,72], poly(a-methylstyrene) [72], or apiezon wax [73] held on a heated support, has also been adopted as a method for applying an insulating coating to etched metal wires. At present, the electrophoretic deposition of an insulting polymer ¢lm appears to be a popular choice for coating tips [74^77]. In this case, shrinkage of the polymer coating from the tip end, during curing at high temperature, results in the formation of a submicrometer-sized electrode. Full details on this fabrication strategy are given elsewhere [74,75]. Several groups have worked on the production of tiny disk-shaped UMEs, sealed in glass [78^80]. Electrodes of this type are fabricated by heating and pulling metal wires inserted into quartz or borosilicate glass capillaries. As the glass is drawn out, the metal thins, resulting in the formation of a needle-shaped electrode, as shown in Fig. 6. A detailed procedure for producing this type of tip with a high yield has been described recently by Schuhmann’s group [80]. As the electrode dimensions shrink, characterization of the probe geometry becomes more challenging. High-resolution imaging techniques, such as scanning electron microscopy, are often needed, in conjunction with voltammetry and SECM approach curve measurements, where the tip feedback current is recorded as a function of distance, d, from either an inert interface or conducting surface. The shape of the curve is characteristic of the probe geometry [12,81^83] 3.3.4 Micro-ITIES Probes As mentioned earlier, amperometry is not limited to electron transfer reactions between a metal electrode and a redox moiety. A considerable amount of research has been carried out on electron transfer and ion transfer at the polarized interface between two immiscible electrolyte solutions (ITIES) [84]. Girault’s group ¢rst demonstrated that amperometric ion transfer measurements could be made at a liquid/liquid interface formed at the opening of tapered glass capillary [85]. The successful deployment of this type of probe in SECM has expanded the range of species that can be detected [33,86]. Both electron transfer and ion transfer processes can be driven at a micro-ITIES probe by polarizing the interface formed between the liquid in the capillary and the immiscible solution into which the probe is placed. For example, electron transfer between the reduced form of an aqueous redox

266

Barker et al.

FIG. 6 (a) SEM image of a pulled Pt-disk nanoelectrode. (b) SEM image of the front end of a pulled nanoelectrode. (Reproduced with permission from Ref. 80. Copyright 2001 Wiley.)

Characterization of Biomolecular Interfaces with SECM

267

couple, at a high concentration in a capillary, and the oxidized form of a second redox couple in an organic solution has been demonstrated [86]. The voltammetry at this type of polarized ITIES is similar to that at a metal UME, with the current ultimately governed by the di¡usion of the species in the organic phase, provided that the aqueous couple is at a su⁄ciently high concentration relative to that in the organic phase. The use of a polarized ITIES to induce ion transfer provides a route for injecting or depleting speci¢c ions, such as Kþ [33,37,87], on a local scale close to a target interface. 3.3.5 Potentiometric Probes The simplest potentiometric probes are made from metal disks sealed in glass capillaries or an insulating polymer sheath, prepared in a similar way to the amperometric tips described above. For example, silver and silver chloride-coated disk-shaped UMEs have found application in the potentiometric monitoring of Agþ and Cl [25,88]. Nanometer-scale versions of these UMEs can be fabricated from etched wires [89,90] or by pulling a metal wire inside a glass capillary, as described in Sec. 3.3.3 [79]. Antimony UMEs have also proved to be a powerful pH probe over the pH range 5^9 [91]. Advantages of these electrodes are that they are easy to make and have a fast response time. Moreover, such probes can be used in amperometric as well as potentiometric mode, opening up the possibility of determining tip ^ interface distances (from the hindered di¡usion, negative feedback mode, described in Sec. 2.2). To expand the range of species detectable by potentiometry, it is necessary to move toward liquid membrane glass micropipette-based ionselective UMEs, which have found considerable application in the life sciences [59].There are, however, speci¢c considerations when employing such probes in SECM. In particular, it is di⁄cult to fabricate electrodes that allow high (micrometer or submicrometer) spatial resolution and have a fast response time. Ion-selective UMEs for K þ , NH4 þ, and Zn2þ have been fabricated (diameters in the range 1^20 mm [60]),with response times that allow tip scanning at 10 mm/s. These UMEs comprise a selective liquid membrane in the end of a pulled capillary, which separates an internal reference solution from an external test solution. Additionally, pH sensitive electrodes of similar design have been reported [92]. It is important that the absolute distance between the UME and the interface is known, both to avoid tip crashes and to ensure that quantitative information about near-interface concentrations can be obtained. In this case, dual-tip sensors have been developed, with a potentiometric indicator electrode and a second sensor which monitors the distance. This type of double-barrel electrode has employed either conductivity or amperometry to maintain the tip-to-sample separation [60].

268

Barker et al.

It is worth noting that promising nonelectrochemical methods for maintaining a constant tip-to-interface distance are being developed (see Sec.10). Such strategies may prove particularly useful for distance regulation with this type of probe. 3.4

Cells

A wide variety of SECM cells has been described, given the diversity of applications of the technique. Detachable cells are particularly useful, as they allow ready exchange between di¡erent types of experiments. Cylindrical cells comprising a base, a body, and a lid, such as that shown in Fig. 5, are readily constructed.The bases are typically manufactured fromTe£on or Plexiglas. The diameter used depends on the substrate or target interface to be examined, but we have generally found 40 mm to be acceptable for most purposes. The base includes a Viton O-ring, which is resistant to attack by most common solvent of interest. The cell is assembled by pushing the body, constructed from a precision-bore glass cylinder, over the base. An optical window can be incorporated into the cell body, if a microscope is to be used to view the tip and/or target interface. A Te£on or Plexiglas lid, containing holes for the tip, reference (and counter) electrodes, and lines for purging with gases (when required) completes the cell. For studies of liquid/liquid interfaces or liquid/air interfaces, one-piece glass cells, or even small beakers, can also be used, as already mentioned. Solid substrates are secured to the base of the cell, so that the top face of the substrate lies parallel to the cylindrical axis of the UME probe. Depending on the solvent used, the substrate can be (1) secured with tape or a glue, (2) pushed into a tight-¢tting recess machined into the base of the cell, or (3) gently secured under anchors incorporated in the cell base. When the substrate is an electrode, a hole is typically drilled through the cell base to accommodate it. A convenient way to study membranes, which e¡ectively separate a donor solution from a receptor solution, is to mount the membrane of interest on the end of a glass capillary and push this vertically through a hole drilled in the cell base, so that the membrane lies perpendicular to the tip.The other end of the tube is then connected to a reservoir, containing the donor solution. This simple setup allows the study of transmembrane transport by convection (with a hydrostatic or osmotic pressure across the membrane),di¡usion (with a concentration di¡erence between the donor and receptor phases), and ion migration (with a potential applied across the membrane). The integration of SECM with a Langmuir trough to permit the study of monolayers at water/air (W/A) interfaces is an area of considerable promise [50,54,57]. In this case a submarine UME (Fig. 4b), controlled remotely

Characterization of Biomolecular Interfaces with SECM

269

by appropriate micropositioners, is deployed in the dipping well of a conventional Langmuir trough. 4

THEORY

The success of SECM in providing quantitative information on interfaces and interfacial processes rests on the availability of accurate theoretical models for mass transport and coupled kinetics.Whereas mass transport for many macroelectrode geometries and simple mechanisms may be solved analytically in one space dimension, the two-dimensional geometry of SECM is not conducive to exact analytical solution and hence a number of semianalytical [93,94] and numerical [13,38,95^99] methods have been introduced. Quantitative treatments are available for a variety of SECM operating modes for di¡usion-controlled processes, as well as more complicated mechanisms involving, for example, adsorption/desorption at interfaces [38,57] and heterogeneous or homogeneous kinetics [96,97]. Most of the recent numerical approaches for solving the di¡erential equations describing mass transport in microelectrode problems mirror those used in engineering to treat £uid £ow [100] and heat transfer [101], namely, ¢nite-element methods (FEM) [102,103] or ¢nite-di¡erence methods (FDM) [104]. In the application of these approaches to electrochemical problems, the continuous di¡usion ¢eld (concentration as a function of space and time) is described in terms of discrete values at prescribed locations, i.e., at the nodes of a grid dissecting the di¡usion ¢eld. In general, the structure of the grid is important in determining the accuracy and e⁄ciency of these types of simulation [105^107]. The alternating direction implicit ¢nite-di¡erence method (ADIF DM) [108] is an e⁄cient digital technique for solving two-dimensional timedependent problems. It has been applied successfully to model di¡usion at UMEs [109^113] and has proved to be a versatile method for treating a wide variety of steady-state and transient problems in the SECM geometry, particularly by our group. The advantages of ADIFDM are typical of most successful digital methods: relatively simple algorithms are employed with a high e⁄ciency of computation and £exibility that allows easy adaptation to di¡erent kinetic situations and SECM operating modes. ADIFDM also facilitates the treatment of implicit boundary conditions and permits the use of optimized space grids and time steps, so that near-steady-state conditions can be simulated with good e⁄ciency. The ¢rst treatment of mass transfer in the SECM geometry used the FEM to calculate the steady-state tip current response and concentration pro¢le, for an UME operating in the SECM feedback mode positioned close to an in¢nite, planar, conducting or inert substrate [12]. The

270

Barker et al.

di¡usion-controlled chronoamperometric behavior of the SECM feedback experiment was subsequently computed [95] using an integrator based on a Krylov algorithm [114].This approach was, however, limited to simulation of the tip-current response in the absence of homogeneous and heterogeneous kinetic complications. A semianalytical method was used to simulate the steady-state and chronoamperometric response of a SECM tip for irreversible, reversible, and quasi-reversible kinetics at a substrate of ¢nite size, through the derivation and solution of multidimensional integral equations [93,94]. ADIFDM was ¢rst used to simulate the SECM feedback response for the case where the tip-generated species undergoes homogeneous chemical reaction in solution [96]. The method was subsequently employed to model the SECM feedback mode with heterogeneous kinetics for in¢nite and arbitrary-sized substrates [97]. Since these initial applications, ADIFDM has been used to model a variety of kinetic situations for several di¡erent SECM modes, incorporating heterogeneous [14,38,40^44,51] or homogeneous [115^117] kinetics. The e¡ect on the SECM chronoamperometric response of allowing the components of the mediator redox couple to have arbitrary di¡usion coe⁄cients has been assessed through model calculations using the ADIFDM, for the positive feedback [65] and generation/collection [64] modes and for the reverse transient behavior of SECM DPSC measurements in bulk solution [56]. Most SECM theoretical treatments to date have been developed for an inlaid disk microelectrode tip; however, in some situations it may be necessary to use probes with a more complex geometry. Several theoretical treatments for these cases are available, including hemispherical electrodes [83], cone [81,82], recessed electrodes [79], and thinner insulating sheaths [118]. Recently, the boundary-element method (BEM) has been shown to be an e⁄cient approach for simulating the steady-state SECM response for a range of complex tip and substrate geometries [98,99]. FDMs have been used to simulate the degradation of SECM images in the feedback mode due to the roughness of the substrate, which was incorporated into the model as circularly symmetric steps [119]. ADIFDM has also been used to assess the e¡ect of substrate geometry on the SECM tip current response for permeable substrates containing cylindrical pitlike depressions [47]. More recently, two-phase SECM problems have been tackled using ADIFDM. Unless conditions are selected so that the phase that does not contain the tip (denoted as phase 2 throughout this chapter) is maintained at a constant composition, the treatment of SECM problems requires consideration of mass transfer in both phases.These simulations are appropriate to liquid/liquid interfaces, discussed later in this chapter, and to the case

Characterization of Biomolecular Interfaces with SECM

271

where phase 2 is a permeable material such as a polymeric ¢lm or tissue sample. Some of the main conclusions arising from this theory for two-phase systems are highlighted in the following sections. 4.1

Feedback Mode

Empirical expressions describing the steady-state di¡usion-limited current as a function of tip ^ substrate distance have been determined for in¢nite planar inert (no mediator regeneration) and conducting substrates (di¡usion-controlled mediator regeneration) from the respective numerical simulations [81]. For an inert substrate, ITin ðLÞ ¼

1 ð0:15 þ 1:5385=L þ 0:58 expð1:14=LÞ þ 0:0908 exp½ðL  6:3Þ=ð1:017LÞÞ ð1Þ

where ITin ðLÞ is the steady-state tip current, i, normalized by the tip current at an e¡ectively in¢nite tip ^ substrate separation, i(1), and L ¼ d/a is the normalized distance between the substrate and the tip of radius, a.This equation was accurate to 0.5% over the interval 0.05 L 20 [81]. Similarly, for a conductive substrate the dimensionless tip current, ITc ðLÞ, was given by   0:78377 1:0672 þ 0:3315 exp þ 0:68 ð2Þ ITc ðLÞ ¼ L L which ¢ts within 0.5% over the same L interval [81]. Figure 7 shows the approach curves calculated using Eqs. (1) and (2). An empirical approximation for the steady-state tip current has also been determined for the case of ¢nite irreversible kinetics for the heterogeneous redox reaction at the substrate [84] by ¢tting the family of working curves derived numerically for di¡erent values of the rate constant [97]. For this case,   ITin k k ð3Þ IT ¼ IS 1  c þ ITin IT ISk ¼

0:78377 ½0:68 þ 0:3315 expð1:0672=LÞ þ Lð1 þ 1=LÞ ½1 þ F ðL; LÞ

ð4Þ

where ITin and ITc are given by Eqs. (1) and (2), respectively, and ISk is the kinetically controlled normalized substrate current. L ¼ kf d=D, where kf is the apparent heterogeneous rate constant (cm/s) for the ¢rst-order process of interest that occurs at the substrate and D is the di¡usion coe⁄cient of the redox mediator. F ðL; LÞ is given by F ðL; LÞ ¼ ð11 þ 7:3LÞ=Lð110  40LÞ

ð5Þ

272

Barker et al.

FIG. 7 Approach curves calculated from Eq. (1) (lower curve) and Eq. (2) (upper curve).

Equation (3) ¢ts the numerical results within 1^2% over 0.1 L 1.0, 2 log k 3, where k ¼ kf a=D [84]. A set of working curves for di¡erent values of L and 2 log k 3 are shown in Fig. 8. The rate of an irreversible heterogeneous reaction occurring at a substrate can be extracted by ¢tting experimental approach curves to those calculated using Eqs. (3)^(5). Equations (3)^(5) have been used to determine electron transfer (ET) rates at ITIES from SECM approach curves under conditions where mediator di¡usion in phase 2(the phase that does not contain the tip) and ion transport across the ITIES are nonlimiting [120,121]. Although the use of this model considerably simpli¢es the quantitative analysis of data, it implicitly assumes that phase 2 is maintained at a constant composition during a measurement. Recently, a numerical model has been formulated that fully treats di¡usional mass transfer in the two phases [122]. In addition to extending the range of conditions under which SECM feedback measurements can be made at the interface between two discrete phases, lifting the restriction on the composition of phase 2 is particularly bene¢cial for enhancing both the range and precision with which fast kinetics can be investigated. By decreasing the ratio of the redox-active species in two phases, Kr ¼ c1 =c2, where c1 is the concentration of the precursor reactant in phase 1 and c2 is the bulk concentration of the reactant in phase 2,

Characterization of Biomolecular Interfaces with SECM

273

FIG. 8 SECM feedback working curves of IkT versus log (k) for several tip^ substrate distances, logðLÞ ¼ 1:0 (upper curve), 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2, 0.1, 0.0 (lower curve), calculated from Eqs. (3)^(5).

steady-state approach curves in the fast kinetic limit are more readily distinguished [122]. This is clearly seen in Fig. 9, which shows SECM approach curves for identical ET reaction rates, with a nonlimiting reactant supply in phase 2(Kr ! 1) and in the second case Kr ¼ 3. In this ¢gure K represents the normalized rate constant de¢ned by K ¼ k12 ac2 =D1, where k12 is the biomolecular interfacial electron transfer rate constant, and D1 is the di¡usion coe⁄cient of the mediator in phase 1. These attributes have been exploited in the practical measurement of rapid redox kinetics at liquid/ liquid interfaces [122], which were inaccessible to earlier SECM studies which required a high concentration of phase 2 reactant [120,121]. 4.2

SECMIT and SECM-DPSC

The SECMIT mode, discussed in Sec. 2.4, has been treated by numerical simulation [36]. With no limitations on the current response from kinetic e¡ects at the target interface, the normalized steady-state current is governed primarily by the value of the relative permeability of the solute in two phases. This is conveniently de¢ned as the product of the partition

274

Barker et al.

FIG. 9 Simulated approach curves i=ið1Þ versus normalized tip^interface separation, d=a, for a feedback experiment: (i) under constant composition conditions with normalized rate constant K ¼ (a) 100, (b) 50, (c) 20, (d) 10, (e) 5, (f) 2, and (g) 1; (ii) full (two-phase) model conditions with Kr ¼ 3 and K ¼ (a)1000, (b) 100, (c) 50, (d) 20, (e) 10, (f) 5, (g) 2, and (h) 1. (Reproduced with permission from Ref. 122. Copyright 1999 American Chemical Society.)

Characterization of Biomolecular Interfaces with SECM

275

FIG. 10 SECMIT chronoamperometric characteristics for logðL ¼ d=aÞ ¼ 0:8 and Ke g ¼ 2:0. The curves correspond to (a) Ke ¼ 40:0; g ¼ 0:05, (b) Ke ¼ 20:0, g ¼ 0:01, (c) Ke ¼ 4:0; g ¼ 0:5, (d) Ke ¼ 2:0; g ¼ 1:0, (e) Ke ¼ 1:0; g ¼ 2:0, (f) Ke ¼ 0:5, g ¼ 4:0, (g) Ke ¼ 0:1; g ¼ 20:0, and (h) Ke ¼ 0:05; g ¼ 40:0.

coe⁄cient, Ke (the ratio of bulk analyte concentration in phase 2 to that in phase 1) and the ratio of the di¡usion coe⁄cients in the two phases (phase 2 relative to phase 1), g. In contrast, the current ^ time characteristics are highly dependent on the individual values of Ke and g. Figure 10 illustrates the chronoamperometric behavior for non-limiting interfacial kinetics and Ke g ¼ 2 at a ¢xed tip ^ substrate separation, log(d/a) = 0.8. These results consider the current ^ time behavior when the tip potential is suddenly jumped from a value where there is no reaction to one where the analyte of interest is removed at a di¡usion-limited rate. The tip current, i, normalized by i(1), has been plotted as a function of inverse square root of normalized time, t ¼ tD1 =a2, where t is time following the potential step, so as to emphasize the short-time characteristics. It can be seen clearly from this plot that whereas the current ^ time behavior is sensitive to the value of Ke and g, the curves tend to the same steady-state current in the long-time limit. This di¡erence between the steady-state and chronoamperometric characteristics could, in principle, be exploited in determining the concentration and di¡usion coe⁄cient of a solute in a phase that is not in direct contact with the UME probe. It should be possible to determine

276

Barker et al.

independently both Ke and g by correlating measurements of the steadystate current, as a function of distance of the tip from the interface, with chronoamperometric measurements (if there is no interfacial kinetic barrier). Alternatively, steady-state measurements alone provide a powerful approach to determining the product Ke g, since the shape of an approach curve is sensitive to relative permeability [36]. These observations are of considerable practical importance, opening up a new route for measuring concentrations and di¡usion coe⁄cients in phases that have hitherto been di⁄cult to study with dynamic electrochemistry [36,47,48]. More recently, the SECMIT model has been extended to reversible transfer processes across a molecular monolayer adsorbed at the interface between two immiscible liquids [123]. The SECM-DPSC response has been simulated for the cases of both irreversible [51] and reversible [53] phase transfer across an interface. Applications of these techniques are considered in Sec. 6.2. As well as treating an in¢nite second phase, the e¡ect of a second phase of ¢nite thickness has received attention [124]; it has been shown that the DPSC response for reversible phase transfer can be used to determine the dimensions of a thin layer. A triple potential step method developed for investigations of lateral di¡usion has also been treated [54]. 5

BIOMOLECULES IMMOBILIZED ON SOLID SUPPORTS

SECM has been applied extensively to immobilized enzymes deposited on both dielectric supports and electrode surfaces. The focus has been on characterizing enzyme activity and developing assays. Microfabrication and patterning of surfaces, to create array structures, has also been considered. The focus of this work has again been on enzymes, but the approaches developed have been extended to other chemistry. 5.1

Immobilized Enzyme Activity

The investigation of catalytic processes occurring at the surface of solidsupported enzymes and enzyme labels has been a rich area for the application of SECM. Systems studied include glucose oxidase (GOx) [14,15,35,125^ 127], urease, horseradish peroxidase (HRP) [16,128], diaphorase [16,129], nitrate reductase [130], and alkaline phosphatase [131]. A popular method for investigating surface kinetics is the feedback mode, but when the kinetics are too slow, generation-collection strategies are optimal because of the negligible baseline signal for the case of an inactive surface. The two approaches are compared in Fig. 11, for the case where GOx catalyzes the oxidation of b-D-glucose to D-glucono-d-lactone in the presence of a suitable electron acceptor.

Characterization of Biomolecular Interfaces with SECM

277

FIG. 11 Detection of GOx activity: (a) feedback mode with GOx on an insulating support, (b) feedback mode with GOx on a conductive support, and (c) generatorcollector mode. (Reproduced with permission from Ref. 126. Copyright 1997 American Chemical Society.)

In the feedback mode, the electron acceptor, such as an oxidized ferrocene derivative, Fcþ , is generated at the tip and di¡uses to the target surface. Here, it is reduced by GOx in the presence of glucose, forming Fc. Consequently, the measurement of the catalytic process depends on detecting the £ux of Fc from the surface on top of a background signal associated with hindered di¡usion of Fc through the tip ^ interface gap to the electrode. This requires that the enzymatic reaction results in a su⁄cient regeneration rate [left-hand side of Eq. (6)] compared to mass transport from solution [right-hand side of Eq. (6)]: kcat Genz 

103 Dc  a

ð6Þ

where kcat ðs1 Þ de¢nes the catalytic reaction rate under substrate-saturated conditions (e.g., high glucose concentration in the example considered), so that the reaction of the mediator with reduced GOx is rate-limiting. Genz is the e¡ective enzyme surface density (mol/cm2), D is the concentration of the mediator, which has a bulk concentration c  . Notice that there are implications from Eq. (6) for the detection of enzyme activity at high spatial resolution, since a small electrode promotes a high di¡usive £ux,which may swamp the kinetic signal. In a generation-collection experiment, the bulk solution contains the mediator in its oxidized form and the role of the tip is to detect the reduced moiety. This includes Fc in the example, or other reaction products, e.g.,

278

Barker et al.

H2O2 could be detected if O2 was employed as the electron acceptor. In this case, the enzymatic reaction occurs over the whole sample, and the UME is a probe of the di¡usion layer established at the reactive interface. It follows that for macroscopic samples, there may be signi¢cant overlap and crosstalk between neighboring sites, leading to di¡usional blurring of the SECM signal. Moreover, a de¢ned steady-state pro¢le will not prevail at such surfaces and there will be much greater problems of stirring from the movement of the scanning tip in imaging applications, compared to the feedback mode. Tip collection imaging experiments are thus only advisable when the sample comprises remotely located spots of active sites. On the other hand, the detection limit for the generation-collection mode is greatly enhanced compared to the feedback mode. In this case, de¢ning c 0 as the detection limit for the probe electrode, surface enzyme activity should be observable provided that [132] kcat Genz 

Dc 0 rS

ð7Þ

Here, rS is the specimen radius (assuming a disk-shaped sample source). Although Eq. (7) has a similar form to Eq. (6), it should be possible to detect much smaller activity because the tip signal arises only from the product detected. Based on values of D ¼ 5106 cm2 =s, rS ¼ 5103 cm, Genz ¼ 1012 mol=cm2 ; c 0 ¼ 1 mM, it has been suggested that enzyme reactions may be monitored with SECM, provided that kcat > 1 s1 [132]. It follows from this discussion that, to study reactions of immobilized enzymes, consideration must be given to the importance of detection sensitivity, compared to spatial resolution, in determining the most appropriate SECM method. The enhanced sensitivity of the generation-collection mode has proved most useful in enzyme-linked immunosorbent assays (ELISAs), where SECM has been used to measure the activity of small spots of peroxidase-labeled antibody [17]. This type of application is considered further below. Early feedback mode studies [14] established that SECM could be used to measure the rate of oxidation of b-D-glucose to D-glucono-d-lactone, catalyzed by GOx. The enzyme was either covalently attached to a nylon support, trapped within a hydrogel, or deposited as a Langmuir-Blodgett (LB) ¢lm on a glass slide. The former approach proved most sensitive, since it served to e¡ectively enhance the surface concentration of enzyme catalyst Genz . Experiments utilized ferrocene monocarboxylic acid as a mediator and solutions were deaerated, so that the natural electron acceptor, O2, was excluded from the solution. The kinetics of the reaction were deduced using feedback-mode approach curves, in which the tip generated the oxidant, Fcþ ,which di¡used to the target surface,where it was reduced by GOx in the

Characterization of Biomolecular Interfaces with SECM

279

FIG. 12 Normalized feedback current^distance curves recorded with 0.5 mM ferrocene monocarboxylate, in 0.1 M pH 7.0 phosphate/perchlorate buffer with 50 mM D-glucose. With the tip near 50 wt% GOx hydrogel membranes, curves a^d correspond to 1.0-mm-thick membranes composed of 50, 5, 1, and 0 wt% GOx, respectively. Circles denote the theoretical current^distance behavior for an insulating (nonreactive) substrate. (Reproduced with permission from Ref. 14. Copyright 1992 American Chemical Society.)

presence of glucose.Typical approach curves, shown in Fig.12, illustrate that the catalytic current measured by the SECM tip depends on the amount of enzyme deposited on the surface and the presence of glucose. These data, and results for other mediators, were found to ¢t a model for zero-order heterogeneous kinetics, which can be derived from the Michaelis-Menten scheme, when the glucose concentration is much larger than the Michaelis constant, KM [14]. The feedback mode proved useful in monitoring the activity of diaphorase immobilized on a glass slide [16,133]. In the presence of excess bnicotinamide adenine dinucleotide (NADH) in solution, the reaction was probed by oxidizing ferrocenemethanol at the tip UME,which was converted back to its reduced form at the target interface.The reaction was again found to be in a zero-order regime and the tip ^ current response was used to infer the surface coverage, given the known kinetic parameters for this system [134]. The use of enzyme-labeling techniques, to generate redox-active moieties, has opened up the possibility of investigating antibody ^ antigen complexation using SECM. For example, an SECM immunoassay of the toxic protein, leukocidin, produced by methicillin-resistant Staphylococcus

280

Barker et al.

aureus has been developed [135]. A dual immunoassay for the human hormones, placental lactogen and chorionic gonadotropin, has also been reported [136]. Although most investigations of immobilized enzymes have focused on oxidoreductases, it is possible to study the activity of other classes of enzymes using potentiometric detection. For example, an ammoniumselective electrode was used to follow the production of NH4þ in the hydrolysis of urea by urease immobilized on a gold electrode [137].The e¡ect of the potential of the gold electrode on activation ^ deactivation of the enzyme was investigated, and it was hypothesized that changes in the electric ¢eld and the ionic environment of the interface had a huge in£uence on enzyme activity. 5.2

Micropatterning and Microfabrication

Local modi¢cation of a substrate surface with scanned probe microscopy methods has received considerable attention [138^145]. In terms of spatial resolution, SECM does not compete with STM and atomic force microscopy (AFM). However, SECM o¡ers the possibility of carrying out speci¢c chemical and biochemical transformations on a surface, so allowing the construction of functional patterns. The use of SECM for microfabrication has been reviewed [146]; here we focus on processes related to the creation of enzyme-patterned surfaces. Schuhmann’s group introduced a method for creating polypyrrole structures on an electrode surface, using the SECM tip as a mobile counterelectrode [61,62,147].With this approach, the surface electropolymerization reaction was largely con¢ned to the portion of the substrate electrode adjacent to the counter electrode. By using N-(o-aminoalkyl) pyrrole as a monomer, a derivatized polypyrrole containing amino functionality was produced, to which periodate-oxidized GOx could be attached covalently. In this way, an electrode surface was patterned with discrete domains of immobilized enzyme. The creation of polymer towers [147] has been investigated in order to enhance the concentration of immobilized enzyme, so increasing the sensitivity of amperometric assays. By using the SECM tip as a counter electrode it is possible to promote the local desorption of alkanethiol self-assembled monolayers (SAMs) on gold surfaces [148]. The resulting zones of bare gold were used for the selfassembly of an o-functionalized thiol, cystaminium dihydrochloride. GOx patterns were then created in these regions, by the covalent attachment of periodate-oxidized GOx. The basic steps involved in this microfabrication strategy are summarized in Fig. 13. The method opens up the possibility of patterning a surface with many types of enzymes, by repeating desorption and enzyme attachment steps at di¡erent regions of a sample [126]. It has

Characterization of Biomolecular Interfaces with SECM

281

FIG. 13 Reaction sequence for creating a microscopic spot of immobilized GOx on a patterned gold alkanethiolate layer. (Reproduced with permission from Ref. 126. Copyright 1997 American Chemical Society.)

further been shown that pattern size and de¢nition can be enhanced by using an alternating current (kilohertz range) to e¡ect the desorption process [149]. As well as using the SECM tip as a minute counter electrode, it may be used to generate a chemical reagent, which opens up the possibility of carrying out local modi¢cation reactions on both electrode surfaces and dielectric materials. Matsue’s group [150] has controllably destroyed alkylsilanes in selected regions of a monolayer on a glass surface, using an SECM tip to generate OH radicals via the Fenton reaction. It was shown that diaphorase could subsequently be attached to the surface by either physical adsorption to methylene-terminated thiols (Fig.14a) or covalent attachment to amino-terminated thiols, in regions of the monolayer that remained intact (Fig. 14b). Further, by treating the damaged regions of a sul¢de-terminated monolayer with aminopropyltriethoxysilane, patches of amino-terminated monolayer were created to which diaphorase could subsequently be attached (Fig. 14c). Wipf ’s group has employed the reagent-generation mode and biotin/ avidin chemistry to selectively derivatize carbon surfaces [151]. The process was achieved locally by activating biotin hydrazine using an SECM tip, which then bound to the carbon surface and formed a microspot of similar

282

Barker et al.

FIG. 14 Preparation of micropatterns with diaphorase at SAM-immobilized glass surfaces by electrogenerated hydroxyl: (a) diaphorase, E, is physically adsorbed onto the hydrophobic area; (b) diaphorase is chemically linked to the hydroxyl radical-nonattacked area to give a negative pattern; and (c) diaphorase is chemically linked to the hydroxyl radical-attacked area to give a positive pattern. (Reproduced with permission from Ref. 150. Copyright 1997 American Chemical Society.)

Characterization of Biomolecular Interfaces with SECM

283

size to the UME probe. It was also demonstrated that bound biotin could be selectively desorbed by the generation of hydroxide ions at the tip, resulting in a surface pattern comprising an underivatized patch surrounded by the biotin-derivatized surface. The latter idea, of patterning a surface by selective deactivation of a zone of a uniformly active surface, has been developed by considering alternative denaturing reagents, such as Br2/HOBr [16]. This approach has been used to denature immobilized diaphorase; the resulting active ^ deactivated surface pattern was imaged using the SECM feedback mode [16]. Local activation also appears to be possible, as evidenced by studies in which alcohol dehydrogenase, immobilized on agarose beads, was maintained in a deactivated state, at bulk pH 6, well below the optimum value (pH 9). Reduction of oxygen at the tip resulted in the injection of hydroxide ions, raising the pH, and switching on the enzymatic activity in a local spot,which was monitored by simultaneous £uorescence microscopy of the reduced cofactor, NADH [152]. Surface-modi¢ed paramagnetic beads represent a potentially attractive support for biochemical analyses, since they combine features of both homogeneous and heterogeneous assays. A suspension of beads can be treated as simply as a liquid, in terms of mixing with reagents,etc., but can then be deposited on a support on the microscopic scale to create agglomerates for surface analysis by SECM. By immobilizing antibodies or other functionality on magnetic beads, the mounds produced can be used for local assays. Beads coated with anti-mouse antibody, saturated with alkaline phosphatase, have been imaged in the tip collection mode using the oxidation of enzyme-generated 4-aminophenol as the signature [125]. Feedback-mode imaging has also been considered with glucose oxidase as the enzyme label [125]. 6

MONOLAYERS AT LIQUID/LIQUID AND WATER/GAS INTERFACES

As illustrated in Fig. 1, a monolayer represents one-half of a bilayer membrane, and so constitutes a useful model system for investigating physicochemical processes pertinent to cellular membranes [153,154]. SECM has found successful application in the study of physicochemical processes at liquid interfaces modi¢ed with monolayers [1,155]. Much of this work has considered the kinetics of molecular, ion, and electron transfer processes, with the monolayer formed at a liquid/liquid interface or water/gas interface. There is also general interest in creating nanostructured biomolecular interfaces using metal nanoparticles as building blocks [156^161]. Pertinent preliminary work using SECM to investigate bare nanoparticle assemblies will be mentioned brie£y in Sec. 6.2.4.

284

6.1

Barker et al.

Monolayers at Liquid/Liquid Interfaces

Monolayers at oil/water interfaces constitute an attractive and simple model for a biomolecular interface, since the potential drop across the interface can readily be controlled and varied, thereby allowing the e¡ect on charge transfer kinetics to be identi¢ed [162,163]. Distance e¡ects on electron transfer across an oil/water interface have been investigated using SECM, with variable-chain-length phospholipid monolayers adsorbed at the liquid/liquid interface serving to separate the reactants in the two immiscible phases [164] (Fig. 15).These investigations involved the reaction between tip-generated 5,10,15,20-tetraphenyl21H,23H-porphine zinc cation (ZnPorþ .) in a benzene phase and various aqueous-phase reductants (Rw in Fig. 15). The electron transfer rate constants measured in the presence of the phospholipid monolayer were lower than for the lipid-free interface and generally decreased as the number of methylene groups in the hydrocarbon chain of the phospholipid increased. Some deviations from this trend were observed,which were attributed to the partial penetration of the ZnPorþ . species into the lipid monolayer. An important outcome of these studies was that at high overpotentials the electron transfer rate appeared to decrease with increasing driving force, con sistent with the predictions from Marcus theory of an inverted-reaction freeenergy pro¢le [165]. Results obtained in this study are summarized in Fig.16, in which the heterogeneous electron transfer rate constant, k f, is plotted as a function of driving force. The latter quantity is written in terms of the di¡erence in redox potentials between the two reactants, measured against the same reference electrode, DEo, and the potential across the oil/water interface, Dow j. In a separate study [166], the rate of electron transfer across an oil/ water interface in the presence of adsorbed conjugated phospholipids was found to be at least twice as rapid as that measured when saturated phospholipids were used. This e¡ect was interpreted in terms of the delocalized conjugated chain acting as a conductive wire, so increasing the rate of electron transfer. The di¡erence between electron transfer rates with the di¡erent types of phospholipids was su⁄ciently high to enable the use of the SECM feedback mode to image the reactivity of mixed monolayers [166]. When a 25-mm-diameter disk UME generating ZnPorþ . was scanned laterally across a mixed monolayer comprised of a saturated and a conjugated phospholipid, regions of relatively high and low ET rate, with reductants in the aqueous phase, were detected from changes in the SECM feedback current. These zones had dimensions of tens of micrometers and were considered to be associated with domains that were rich in one of the types of lipid.

Characterization of Biomolecular Interfaces with SECM

285

FIG. 15 (A) Probing the kinetics of ET beween ZnPorþ and various aqueous redox species at the ITIES with the SECM feedback mode. OW/RW is an aqueous redox couple, such as Ru(CN)63/4, Mo(CN)83/4, Fe(CN)63/4, Fe3/2þ , V3/2þ or Co(III)/(II) sepalchrate. Of the ionic species contained in the system, only ClO4 can readily cross the interface to maintain electroneutrality. (B) ET across an ITIES modified by a monolayer of phospholipid. The inset shows the structure of a synthetic phosphatidylcholine lipid studied by this method. (Reproduced with permission from Ref. 164. Copyright 1997 American Chemical Society.)

286

Barker et al.

FIG. 16 Driving force dependence of the interfacial ET rate constant, kf , between ZnPor þ in benzene and various aqueous redox species across a monolayer of C-10 lipid. The organic phase contained 0.25 M THACIO4, 0.5 mM ZnPor, and 100 mM C-10. The aqueous solution contained 7 mM of (1) Fe(CN)64, (2) Co(II) sepalchrate, and (3) V2 þ . (Reproduced with permission from Ref. 164. Copyright 1997 American Chemical Society.)

The e¡ect of temperature on interfacial electron transfer rates for saturated phospholipids has also been investigated [166]. A sharp decrease in the rate constant at a critical temperature was attributed to a phase transition changing the tunneling distance between the redox species contained in the two contacting solutions. In related work [167], SECM was used to study the adsorption of the nonionic surfactant, Triton X-100, at the ITIES, and its e¡ect on the oxidation of decamethylferrocene in 1,2-dichloroethane by aqueous Ru(CN)63. An observed blocking e¡ect of adsorbed Triton X-100 on the interfacial redox reaction was successfully analyzed in terms of surfactant adsorption following a Langmuir isotherm. 6.2

Monolayers at Water/Air (W/A) Interfaces

A wide range of di¡usion processes can be investigated in molecular monolayers at the W/A interface, by combining SECM with a Langmuir trough.

Characterization of Biomolecular Interfaces with SECM

287

The use of the Langmuir trough enables the e¡ect of monolayer compression on the process of interest to be readily investigated. 6.2.1 Transfer of Neutral Molecules Across Langmuir Monolayers at a W/A Interface Transfer of oxygen across a W/A interface in the absence and presence of a monolayer of 1-octadecanol was employed as a model system to investigate the e¡ect of a monolayer on molecular transfer rates [50].This study provided information on the e¡ect of the monolayer on reaeration rates, which is of importance in natural environments [168,169].The con¢guration also represented a simple model for the study of oxygen transfer across biomembranes. For these investigations, the UME probe of the SECM was deployed in the aqueous subphase, which contained 0.1 M KNO3, and held at a potential to reduce oxygen at a di¡usion-controlled rate. With the probe positioned close to the W/A interface, the electrochemical process promoted the transfer of O2 from air (phase 2) to the aqueous solution (phase1),with subsequent collection at the tip UME (Fig. 17). Given the high di¡usion coe⁄cient and concentration of oxygen in the air phase, depletion e¡ects in phase 2 were unimportant. The results of the study demonstrated that the rate of oxygen transfer across a clean W/A interface was di¡usion-controlled on the time scale of SECM measurements.The rate of this transfer process was, however,

FIG. 17 Schematic illustration (not to scale) of the SECM induced transfer of oxygen across a 1-octadecanol monolayer at the air/water interface.

288

Barker et al.

FIG. 18 Normalized steady-state diffusion-limited current versus UME^interface separation for the reduction of oxygen at an UME approaching a W/A interface with a 1-octadecanol monolayer (O). From top to bottom, the curves correspond to an uncompressed monolayer and surface pressures of 5, 10, 20, 30, 40, and 50 mN/ m. The solid lines represent the theoretical behavior for reversible transfer in an aerated atmosphere, with zero-order rate constants for oxygen transfer from air to water, k0 /108 mol/cm2-s of 6.7, 3.7, 3.3, 2.5, 1.8, 1.7, and 1.3. (Reproduced with permission from Ref. 50. Copyright 1998 American Chemical Society.)

signi¢cantly reduced with increasing compression of a 1-octadecanol monolayer assembled at the W/A interface. Figure 18 illustrates this point, showing experimental approach curves for oxygen reduction, recorded with the monolayer at di¡erent surface pressures. The best ¢ts through the data are obtained from an SECM mass transport model that involves only the £ux, k0 , of oxygen from air to water as the only adjustable parameter. The transfer rates measured in this way were successfully interpreted using an accessible area model [170^172] described by the following equation:   Ay¼1 0 0 k 0 ¼ ky¼0 ð1  yÞ ¼ ky¼0 1 ð8Þ A

Characterization of Biomolecular Interfaces with SECM

289

where y represents the fraction of the interface that is covered by surfactant, A is the surface area, usually written as the area per molecule of surfactant, and Ay¼1 is the area for which the transfer rate is zero, typically when the surface is completely covered with a close-packed arrangement of surfactant molecules. The £ux, k0 , can be expressed in terms of ¢rst-order interfacial rate constants for oxygen transfer from water to air, k wa , and air to water, kaw, and the concentration of oxygen in the air and water phases, ca and cw, [50]: k 0 ¼ kaw ca  kwa cw

ð9Þ

Typical rate constants are plotted as function of A1 in Fig. 19. In these studies, it was found that the molecular area at which the monolayer collapsed corresponded well with Ay¼1 (deduced from Fig. 19), thus indicating that at the area of collapse, the monolayer was close to a state where oxygen transfer was completely inhibited. In order to investigate the applicability of more sophisticated models in describing passive di¡usion across a monolayer, the transfer of Br2 across aW/A interface modi¢ed with di¡erent-chain-length fatty alcohols, has been

FIG. 19 Oxygen transfer rate constants as a function of the reciprocal of the interfacial area per molecule. (Reproduced with permission from Ref. 50. Copyright 1998 American Chemical Society.)

290

Barker et al.

considered more recently [173]. A homologous series of aliphatic alcohols was investigated: C14H29OH, C16H33OH, C18H37OH, and C20H41OH. Kinetic data were interpreted in terms of three di¡erent theories, in addition to the accessible area model [170]: the energy barrier model [174^176], the density £uctuation model [177,178], and the solubility-di¡usion model [154]. The principles of these experimental measurements, which employed the SECM-DPSC mode, are illustrated in Fig. 20. Br2 was electrogenerated in an initial (forward) potential step by the di¡usion-controlled oxidation of Br in an aqueous sulfuric acid subphase. Tip-generated Br2 di¡used to and

FIG. 20 Schematic (not to scale) of SECM-DPSC measurements. Molecular Br2 is generated by oxidizing Br in the forward potential step (a) and collected when the direction of the potential step is reversed to reduce Br2 to Br (b).

Characterization of Biomolecular Interfaces with SECM

291

transferred across the fatty alcohol monolayer irreversibly. Br2 was subsequently collected back by di¡usion-controlled reduction to Br in a second (reverse) potential step. The resulting current ^ time behavior provided information on both the tip ^ interface separation (forward step) and the kinetics of Br2 transfer (reverse step) [51]. Kinetic data obtained from these measurements were used to examine each of the models. None was found to give a complete description of all the data, although features of the experiments were amenable to interpretation. A clear outcome of these studies was the need for more sophisticated models
for example, derived from molecular dynamics. 6.2.2 Lateral Amphiphile Diffusion A new SECM approach for studying the lateral di¡usion of redox-active amphiphiles in Langmuir monolayers at a W/A interface has recently been developed [54]. Analogous to £uorescence recovery after photobleaching (FRAP) [179^183], this approach involves an ‘‘electrochemical bleaching’’ step, a ‘‘recovery’’ step, and a ¢nal ‘‘analysis’’ step, as illustrated schematically in Fig. 21. Practically, a triple potential step is applied at a submarine UME placed in the aqueous subphase of the Langmuir trough, close (1^2 mm) to the monolayer. In the ¢rst potential step, an electroactive species is generated at the UME by di¡usion-controlled electrolysis of a precursor. This species di¡uses to, and reacts with, the redox-active amphiphile at the W/A interface, resulting in the formation of the initial solution precursor,

FIG. 21 Schematic (not to scale) of the arrangement for SECM triple potential step measurements of lateral diffusion processes at the W/A interface. (Reproduced with permission from Ref. 54. Copyright 2001 American Chemical Society.)

292

Barker et al.

which undergoes di¡usional feedback to the UME. In this ¢rst step, the rate constant, k, for electron transfer between the solution mediator and the surface-con¢ned species can be measured from the UME current ^ time transient. In the second period, the potential step is reversed to convert the electrogenerated species to its initial form. Lateral di¡usion of electroactive amphiphile into the interfacial zone probed by the UME occurs simultaneously in this recovery period. In the third step, the potential is stepped in the same direction as for the ¢rst step. The corresponding UME current ^ time transient can be used to determine either the distance between the UME tip and the monolayer at the water surface (if an extensive ¢rst step and short second step is utilized), or the lateral di¡usion coe⁄cient of the amphiphile (if a longer recovery period is set). This method was demonstrated experimentally with measurements on the lateral di¡usion of N-octadecylferrocenecarboxamide(C18Fc) in a 1:1 Langmuir monolayer with 1-octadecanol. The di¡erent transient responses for the generation and collection of Ru(bipy) 3 3þ [from a bulk solution of Ru(bipy)3 2þ ] are considered separately in Fig. 22.These data are for the case where the mean area per molecule in this bimolecular system was 50 —2. The distance, d, of the probe electrode from the interface was established from the current ^ time transient recorded during the third potential step, with time scales for the potential steps de¢ned by the switching times, t1 ¼ 0.4 s and t2 ¼ 0.44 s. The long ‘‘bleaching’’ time and short ‘‘recovery’’ time permitted a d value of 1.25 mm to be obtained for the case depicted in Fig. 22a, by ¢tting the ¢nal step data to theory for an inert interface [95]. The outlying theoretical curves in this ¢gure correspond to ca. 0.1 mm, indicating the high precision of distance measurements with this approach. With knowledge of d, the rate constant, k, for the interfacial ETprocess was obtained by ¢tting the transient for the ¢rst potential step to theoretical curves using k as the only unknown. It can be seen in Fig. 22b that there is a vast enhancement in the current £owing during the ¢rst step, compared to that found for an inert interface, due to the redox reaction between electrogenerated Ru(bipy)3 3þ and C18Fc, resulting in positive feedback. Analysis of the data shown resulted in k ¼ 0.035 cm/s as the best ¢t, with G ¼ 1.66  1010 mol/cm2 (de¢ned by the A value). The outlying theoretical curves are for k 0.005 cm/s, which suggests that the precision of the k measurement is within 10%. The ET rate constant and lateral di¡usion coe⁄cient measured for A ¼ 40 —2 were 0.045 0.006 cm/s and (5 1)  107 cm2/s, respectively. The ET rate constant measured for A ¼ 30 —2 was 0.06 0.01 cm/s, while lateral di¡usion was undetectable, suggesting a lateral di¡usion coe⁄cient less than 1 107 cm2/s. These data are consistent with earlier work using a microline electrode technique [184]. There are prospects for applying this

Characterization of Biomolecular Interfaces with SECM

293

FIG. 22 Transients for the C18Fc-Ru(bipy)32 þ system, with A ¼ 50—2. (a) Measurement of the tip^interface separation, d. The solid curve is the experimental characteristic during the third potential step, with t1 ¼ 0:4 s and t2 ¼ 0:44 s, which is effectively coincident with theory for an inert interface with log L ¼ 1:0 (dashed curve). The outlying upper and lower dashed theoretical curves are for log L ¼ 0:97 and 1.04. (b) Measurement of the ET rate constant, k. The solid curve is a typical experimental transient for the first potential step, which fits well to k ¼ 0.035 cm/s (d ¼ 1.25 mm), shown as a coincident dashed line. The outlying upper and lower dashed theoretical curves are for k ¼ 0.040 and 0.030 cm/s, respectively. (c) Measurement of the relative diffusion coefficients, Dr (lateral diffusion coefficient compared to that of the redox mediator). The solid curve is the experimental result, which fits well to Dr ¼ 0:2. The outlying upper and lower dashed theoretical curves are for Dr ¼ 0:24 and 0.16, respectively. (Reproduced with permission from Ref. 54. Copyright 2001 American Chemical Society.)

approach to redox reactions in biologically relevant assemblies and also to study lateral charge transfer in ultrathin ¢lms. 6.2.3 Lateral Proton Hopping An approach similar to SECMIT has been used to investigate lateral proton di¡usion processes in acidic monolayers [57]. Lateral di¡usion processes of this type are crucial in de¢ning the activity of membrane-bound reactive centers in cells [185]. A controversial aspect of prior work in this ¢eld has been the magnitude of the lateral proton di¡usion coe⁄cient, since different techniques have provided contradictory results [185^192]. Many of

294

FIG. 22

Barker et al.

Continued.

the earlier measurements were made over centimeter length scales [185^ 190]; a key advantage of SECM is the ability to make measurements with high spatial and temporal resolution, pertinent to cellular membranes. SECM was initially used to investigate lateral proton di¡usion at stearic acid assembled at theW/A interface on an aqueous subphase containing a low concentration of protons (20^50 mM). The UME was biased at a potential suitable to reduce protons to hydrogen at a di¡usion-controlled rate. The resulting local depletion drove the interfacial acid dissociation reaction,

Characterization of Biomolecular Interfaces with SECM

295

FIG. 23 Schematic (not to scale) of the arrangement for SECM measurements of proton transport at a stearic acid monolayer deposited at the W/A interface. The UME diameter, 2a, is typically in the range 10^25 mm, and the tip^interface distance, d, is typically 2a.

which in turn created a proton di¡usion gradient, both in solution and at the interface (Fig. 23). The transport-limited current £owing at the electrode provided a measure of the rates of the solution and surface processes, which were investigated as a function of the surface coverage of stearic acid, by controlling the monolayer compression. These measurements indicated that inplane lateral proton di¡usion was facilitated at W/A interfaces on which stearic acid monolayers were formed.The lateral proton di¡usion coe⁄cient was found to depend critically on the physical state of the monolayer and was at most ca. 15% of the magnitude in bulk solution. Most recently, a new SECM proton feedback method has been developed for investigating lateral proton di¡usion at phospholipid assemblies, speci¢cally, Langmuir monolayers at the W/A interface [193]. In this approach, a base is electrogenerated by the reduction of a weak acid at a ‘‘submarine’’ UME placed in the aqueous subphase of a Langmuir trough close to a monolayer. The base di¡uses to the interface and titrates monolayer-bound protons and is thus converted back to the acid form, so enhancing the current response at the UME. Lateral proton di¡usion has

296

Barker et al.

been investigated in monolayers comprising either acidic DL-a-phosphatidyl-L-serine, dipalmitoyl (DPPS) or zwitterionic L-a-phosphatidylcholine, dipalmitoyl (DPPC) monolayers at a range of surface pressures. Lateral proton £uxes at DPPS were found to be signi¢cant, but the lateral proton di¡usion coe⁄cient was lower than in bulk solution. In contrast, lateral proton di¡usion could not be detected at DPPC, suggesting that the acid/ base character of the phospholipid is important in determining the magnitude of interfacial proton £uxes [193]. 6.2.4 Nanoparticle Assemblies As highlighted earlier, metal nanoparticles are ¢nding application as a building block in bottom-up approaches for creating functionalized biomolecular interfaces [156^161]. SECM could have a key role to play in characterizing charge transport at this type of interface, as it is not necessary for the assembly of interest to be directly attached to an electrode. Initial studies of assemblies of bare metal nanoparticles at nonconductive interfaces have already appeared, which highlights the type of information that might be obtained. Bard and co-workers extended the combined SECM-Langmuir trough approach to investigate the lateral conductivity of nanoparticles [194]. Zhang et al. [195] used SECM to study the kinetics of proton reduction by methyl viologen, catalyzed by bare gold nanoparticles (ca. 6 nm diameter) absorbed at the surface of a glass slide silanized with 3-mercaptopropyl trimethoxysilane (Fig. 24). A hydrogen discharge rate of approximately (2.6 0.5)  10 4 molecules of H2 per second per particle was measured at pH 2.

FIG. 24 Schematic of SECM measurements on nanoparticle catalyzed H2 evolution; MV2 þ denotes methyl viologen dication. (Reproduced with permission from Ref. 195. Copyright 2001 Royal Society of Chemistry.)

Characterization of Biomolecular Interfaces with SECM

7

297

BILAYER STUDIES

The properties of bilayer lipid membranes (BLMs) are of great interest, as BLMs are considered to be suitable arti¢cial analogues for cellular membranes [196]. Transport processes at cell membranes are of vital biological importance, and arti¢cially constructed BLMs can be modi¢ed to mimic selected properties of living membranes [196]. BLMs can readily be prepared in a form suitable for study by a variety of techniques, including SECM. For example, phospholipid solutions in decane can simply be ‘‘painted’’ across an aperture in a supporting sheet (usuallyTe£on) positioned in an aqueous solution and allowed to thin out to form a bilayer [197]. Alternatively, BLMs can be formed via a ‘‘monolayer folding’’ technique from thin ¢lms spread at the W/A interface [198]. The latter method has the advantage that it can be used to produce bilayers composed of two di¡erent monolayers. Moreover, this technique minimizes solvent inclusion e¡ects that may result from painting techniques. BLMs can be modi¢ed to mimic a speci¢c property of a cell membrane; they are often more robust than living cells and may be made on a larger scale for physicochemical studies. BLMs have been studied extensively since the 1960s, with electrical (impedance and capacitance) methods proving popular [199] for characterizing the thickness and successful formation of a bilayer [199]. Other early studies included an examination of the change in the electrical properties of bilayer membranes with the addition of salts [200].The advent of SECM and related microelectrode techniques has further advanced the study of BLMs and their properties at the local level. 7.1

Experimental Design for the Study of BLMs

The relative ease of formation and robustness of arti¢cial BLMs means that they are amenable to study by a range of experimental techniques. Consequently, di¡erent apparatus have been designed to be ‘‘¢t for purpose,’’ depending on the requirements of the experiment. Usually, these design features relate to the cell in which the supporting membrane containing the bilayer aperture is held. For example, does the bilayer need to be monitored visually during the experiment? Where must the microelectrodes be placed in relation to the bilayer? Some researchers have utilized electrochemical cells in which the BLM is vertically oriented [201^204]. A suitable cell design for this arrangement (believed to be the most stable con¢guration) is shown in Fig. 25 [202]. In this case, SECM approach curves are recorded with the tip translated along the x (or y) axis, contrasting to the usual direction of approach curve operation, where the sample is oriented in the x ^ y plane and the tip is translated along

298

Barker et al.

FIG. 25 An experimental apparatus for the SECM investigation of vertically oriented bilayers. (From Ref. 202.)

z axis. The addition of glass windows to the cell allows the position of the electrode with respect to the bilayer to be monitored visually and recorded on video via a CCD camera. Bard and co-workers have developed a di¡erent system in which the bilayer is oriented horizontally within the experimental apparatus (Fig. 26) [205]. This enables the use of a conventional SECM for imaging and approach curve measurements. 7.2

Topographical Imaging of BLMs Using SECM

Bard and co-workers have employed the horizontally oriented bilayer apparatus discussed in Sec. 7.1 to image the topography of BLMs [205]. Membranes were prepared using the paintbrush technique and the cell design ensured equal pressure either side of the membrane, resulting in a stable BLM. To image the bilayer topographically, the SECM probe was held at a potential suitable for the di¡usion-limited oxidation of ferrocyanide, which was employed as the mediator. Unmodi¢ed BLMs are inert to the transport of ferrocyanide, thus, as the electrode approached the bilayer, the current decreased as di¡usion to the electrode became more hindered (negative feedback response [13]). The electrode was moved toward the bilayer until a current that was approximately 80% of that in bulk solution was registered and then scanned, at a ¢xed height, from one side of the bilayer to the other. The current changes recorded could be related directly to a change in the

Characterization of Biomolecular Interfaces with SECM

299

FIG. 26 Schematics detailing SECM experimental apparatus used for the study of horizontal bilayers. A shows the entire experimental setup and B shows the SECM cells used in detail: (1) screw, (2) O-ring, (3) PTFE sheet with two holes, (4) PTFE sheet with one hole, and (5) holes in the inner wall of the PTFE barrier. (Reproduced with permission from Ref. 205. Copyright 1999 American Chemical Society.)

300

Barker et al.

FIG. 27 The upper image shows a diffusion-limited current linescan across a BLM. The lower image shows this data interpreted as a representation of the BLM topography. The image was obtained by scanning a 25-mm Pt UME parallel to the bilayer surface. (Reproduced with permission from Ref. 205. Copyright 1999 American Chemical Society.)

tip ^ membrane separation, using Eq. (1), thus revealing the topography of the membrane. From this, the £at, true bilayer portion of the membrane and the annulus could readily be distinguished, as shown in Fig. 27. Initial approach curve studies conducted on BLMs identi¢ed a key factor in this type of analysis. If the £at apex of the SECM tip was not coplanar with the BLM, then the current ^ distance data obtained could be erroneous. For example, a situation could be imagined whereby the edge of the glass sheath of the electrode touched and broke the bilayer before the electrode itself made contact. Although the distances involved were only on a scale of a few micrometers, it is nevertheless a factor that must be borne in mind during experiments of this type. 7.3

SECM and Microelectrode Studies of Charge Transfer Across BLMs

Matsue and coworkers were the ¢rst to use linear-sweep voltammetry at UMEs to examine the permeation of ions through a BLM [201]. Speci¢cally,

Characterization of Biomolecular Interfaces with SECM

301

Ru(NH3)63þ and ferrocenecarboxylic acid (FCA) were used as target ions. In this system, a two-electrode system was used, with a 65-mm-diameter Pt working electrode and an Ag/AgCl reference electrode. Both electrodes were immersed in solution on the same side of the BLM, with the working electrode held in close proximity to the membrane and attached to a motordriven positioning stage. As shown in Fig. 28, a series of voltammograms were recorded at di¡erent electrode-bilayer distances, and the change in the magnitudes of the limiting currents for the reduction of Ru(NH3)63þ and the oxidation of ferrocenecarboxylic acid (FCA) were monitored. As the electrode ^ BLM distance decreased, there was a marked diminution in current for Ru(NH3)63þ, leading to the conclusion that the

FIG. 28 Voltammograms recorded at a Pt disk UME for a solution containing 1.0 mM FCA and 1.0 mM Ru(NH3)63 þ at two different pH values, (a) pH 5.89, and (b) pH 7.30. For each voltammogram, the solid line represents experimental data and the open symbols represent simulated data. Each voltammogram was recorded at a different electrode^membrane distance. (a) 1, 1 ; 2, 51 mm; 3, 27 mm (b) 1, 1 ; 2, 45 mm; 3, 27 mm. (Reproduced with permission from Ref. 201. Copyright 1991 Elsevier.)

302

Barker et al.

BLM was practically impermeable to Ru(NH3)63þ transport. This allowed data obtained from the Ru(NH3)63þ experiments to be used as a measure of the tip ^ substrate distance, as might be done in the analysis of an SECM hindered di¡usion curve [12,13]. In contrast, there was only a slight decrease in FCA signal with distance, suggesting that the membrane was permeable to FCA. A permeation coe⁄cient of 4.0  103 cm/s was determined by correlation between experimental data and theoretical simulation. As FCA is a weak acid, further experiments were carried out examining BLM permeability as a function of pH. As the pH was increased, the permeation coe⁄cient, Pm, decreased, indicating that the Pm value for the anionic form of FCAwas small compared to the neutral form. This study was aimed at testing Overton’s rule [206],which asserts that permeation through the hydrocarbon chains of the BLM is the rate-determining step in BLM permeation with equilibrium partitioning of the solute between the membrane and the solution. Experimental data for FCA were found to be consistent with this rule. Matsue and co-workers made further SECM studies of the transport properties of a BLM containing ion channels formed from alamethicin [204], which is selective to cations.The permeation of Ru(NH3)63þ and Fe(CN)63 through the ion channels in the BLM were investigated and reported.Using a four-electrode system, a schematic of which is shown in Fig. 29, potentials were applied across the membrane, to facilitate opening of the channels which occurred at a membrane potential of 50 mV.The number of alamethicin molecules forming an average channel was determined by monitoring the dependence of the total ionic current on alamethicin concentration. In order to monitor permeation, a voltage pulse was applied to the membrane and the reduction current recorded for the detection of Ru(NH3)63þ by amperometry at a Pt UME held at selected distances from the trans side of the membrane.These data were used to calculate the relative permselectivity, PR , of the redox ion compared to that of Kþ . For Ru(NH3)63þ, PR was determined to be 0.27. This lower value was attributed to the di¡erences in di¡usion coe⁄cients between the two ions and thus variations in ion mobility. By reversing the potential across the membrane, it was possible to obtain, by the same method, relative permeabilities for Fe(CN)63 and I. The relative permeability of Fe(CN)63 was found to be much lower then for Ru(NH3)63þ,which was attributed to electrostatic repulsion e¡ects between Fe(CN)63 and the carbonyl groups on the walls of the ion channel.This was consistent with previous ¢ndings that showed that the weakly cation-selective almethicin channels [207] resisted the permeation of multicharged negative ions.

Characterization of Biomolecular Interfaces with SECM

303

FIG. 29 Schematic showing the apparatus used to make simultaneous ionic and redox current measurements of transport across a BLM. BP ¼ bipotentiostat; W1 ¼ Pt microdisk electrode, W2 ¼ Ag/AgCl electrode; RE ¼ Ag/AgCl connected to virtual ground; CE ¼ Pt wire. (Reproduced with permission from Ref. 204. Copyright 1994 American Chemical Society.)

This study also demonstrated that in the case of I, there was charge transfer across the membrane by virtue of the oxidation of I to I3, which deposits in the BLM as I2. The enhanced rate of charge transfer is because this species acts as an I carrier across the BLM. Bard and coworkers also investigated the transport of I in BLM systems using SECM approach curves [205]. It was shown that the £ux of I to the electrode tip was considerably enhanced by the addition of I2 to the membrane-forming solution. Furthermore, when the amount of I2 dopant was varied, the net £ux of I to the tip changed accordingly. Further work by Bard and coworkers saw the use of voltammetric ionselective micropipette electrodes for probing BLMs. These were based on valinomycin and used to investigate the transport of Kþ ions through gramicidin channels included in BLMs [87]. Electrodes were fabricated from pulled glass capillaries in a similar way to those described in Sec. 3.3.5. After pulling to the dimensions of short patch-clamp pipettes, the inner walls of the capillaries were silanized, by ¢lling with a toluene solution of

304

Barker et al.

trimethylchlorosilane, in order to render them hydrophobic. To make the Kþ -selective microelectrodes, the capillaries were ¢lled with a DCE solution of 10 mM valinomycin and 10 mM ETH 500 as ionophore and supporting electrolyte. Electrical connection was made to the electrodes via a silver wire reduction-coated in silver tetrakis(4-chlorophenyl) borate. A specially designed SECM cell allowed an external potential to be applied. Moreover, a pumping system was utilized to allow the solution in the upper compartment of the cell to be changed, thus enabling the cell to hold solutions of di¡erent composition on either side of the membrane, as shown in Fig. 30. In some studies, a concentration gradient in Kþ was established across the BLM, from the lower to the upper solution compartment. The dynamics

FIG. 30 Schematic of the SECM cell built for the study of horizontally oriented bilayers with ion-selective micropipette electrode probes. Integration of an HPLC pump allows the cell to hold solutions of differing composition on either side of the bilayer membrane. (Reproduced with permission from Ref. 87. Copyright 2000 American Chemical Society.)

Characterization of Biomolecular Interfaces with SECM

305

of Kþ transfer were obtained by analyzing the steady-state tip current. In the absence of a concentration gradient, as the electrode approached the bilayer, a signi¢cant decrease in current was observed and the normalized approach curve ¢tted hindered di¡usion theory [12]. With a concentration gradient, however, the decrease in current was less appreciable. This experiment probed the conductivity of BLMs, with and without gramicidin channels, and showed that, without an externally imposed Kþ concentration gradient, transport of ions across the membrane was very slow. The driving force for Kþ transport was also varied, by application of a potential across the membrane. These experiments were akin to SECM generation-collection experiments,with the ions generated at the bilayer and collected at the tip. By applying di¡erent membrane potentials, and recording approach curves, it was possible to establish how the membrane potential controlled transport. 7.4

Permeation of Uncharged Solutes in BLMs

SECM has been employed to examine the permeation of oxygen through BLMs formed from L-a-phosphatidylcholine [202]. For these studies the bilayer was oriented vertically using the apparatus shown in Fig. 25 and measurements made via approach curves. Initially, the bilayer was approached with the electrode held at a suitable potential for the di¡usion-limited oxidation of IrCl63 (deliberately added to the solution). As the BLM was inert toward this ion, the resulting approach curves provided information on the tip ^ bilayer distances. The tip was then retracted and the potential changed to one suitable for the reduction of oxygen (naturally present in the aerated solution). In this case, the BLM was permeable to the transport of oxygen, as shown in Fig. 31. The current remained similar to that obtained in bulk solution throughout the approach curve, due to the transport of O2 across the BLM,with no apparent interfacial resistance. An advantage of SECM in this application is that the tip induces the transport process and so there are no signi¢cant stagnant layers of the type involved in conventional transport measurements, so allowing much faster interfacial processes to be characterized. These stagnant or ‘‘unstirred’’ layers (USLs) can, in fact, be probed using microelectrodes, as discussed in the following section. 7.5

Factors Influencing Membrane Transport Processes

Pohl and coworkers undertook a series of electrochemical studies, employing UMEs, to investigate BLM transport processes. For these studies the physical environment of the bilayer and/or the mechanism of the

306

Barker et al.

FIG. 31 Approach curves recorded with the UME translated towards bilayer membranes:hindered diffusion theory; , ~, experimental data for the one-electron oxidation of IrCl63;experimental data for the reduction of O2. (From Ref. 202.)



transport process were of interest, as well as the properties of the unstirred ‘‘di¡usion layer.’’ This latter work extended existing biological and physiological models to provide a more ‘‘realistic’’ interpretation of processes occurring in USLs. For example, in one study, concentration pro¢les were recorded in the immediate vicinity of a BLM, i.e., across the USL, and the e¡ect of a transmembrane osmotic £ux was measured [208]. The established physiological model theorized that osmotic advection was countered only by back di¡usion of the solute. The work showed, however, that the model did not accurately describe true concentration pro¢les. A more accurate picture was provided by invoking a stagnant-point £ow model, based on gradual changes in stirring velocity in the proximity of the membrane, rather than there being a discrete boundary between the stirred and unstirred layers [209]. The importance of the presence of USLs has been cited in biological and physiological systems [210^212]. For example, the existence of a USL around the erythrocyte cell membrane is believed to slow the O2 uptake in humans by a factor of at least 1.8^2.0 [213], and the di¡usion of cholesterol

Characterization of Biomolecular Interfaces with SECM

307

molecules through an extracellular USL in£uences cholesterol e¥ux from cell membranes [214]. In a study by Pohl and coworkers, the relationship between the size of the USL and the di¡usion coe⁄cient of the solute was examined [215].These factors had commonly been assumed to be independent of one another. Although equations describing simultaneous convection and di¡usion processes have been solved for BLM systems, existing equations developed to describe solute concentration in the USL region break down when a discrete boundary between the stirred and unstirred region in postulated, as mentioned above. Pohl thus proposed the following equation to describe the unstirred layer thickness, d, based on the stagnant-point £ow model [215]: d ¼ 1:6

 1=3  D n 1=2 n a

ð10Þ

where D is the di¡usion coe⁄cient, n is the kinematic viscosity, and a is a stirring parameter. This equation relates the thickness of the USL to the diffusion coe⁄cient of the solute. To prove the validity of Eq. (10) experimentally, concentration pro¢les for two di¡erent ions, Ca2 þ and either Naþ or Kþ , were recorded simultaneously (via potentiometry) using double-barreled ion-selective microelectrodes, at the trans side of the membrane. These microelectrodes were fabricated using double-barreled capillaries which were pulled to a point, to yield a tip diameter typically 1^2 mm, and then back-¢lled with ionophore cocktails according to the procedure described by Amman [59]. The advantage of making simultaneous measurements in this way is that error is greatly reduced compared to the situation where the concentration pro¢les are recorded consecutively. For example, the e¡ects of temperature £uctuations or a change of stirring conditions need not be considered in relation to these measurements. As the electrode probe was moved perpendicular to the face of the BLM, a steep potential change identi¢ed the point where the electrode touched the BLM. As the velocity of the electrode movement was known, it was possible to establish the position of the tip in relation to the BLM at any point in the experiment. Examples of the concentration pro¢les determined by these experiments can be found in Fig. 32.These pro¢les demonstrate that di¡erent ionic species have di¡erent USL thicknesses.Over a series of experiments,the value of d obtained for a particular ion varied by less than 10%. Further concentration pro¢les, recorded in conjunction with pH pro¢les (see Fig. 33), in the presence of the net ion £ux, rather than an osmotic gradient, veri¢ed the cube-root dependence of the USL on di¡usion

308

Barker et al.

FIG. 32 (a) Concentration profiles of Ca2 þ and K þ recorded at the trans side of a bilayer membrane. The addition of 0.8 M urea at the cis side of the BLM induced a transmembrane flux. (b) Concentration profiles of Ca2 þ and K þ recorded at the cis side of a bilayer membrane. Here transmembrane flux was induced by addition of 0.3 M choline. Use of double-barreled electrodes facilitated simultaneous collection of K þ and Ca2 þ concentration data. (Reproduced with permission from Ref. 208. Copyright 1997 Biophysical Society.)

Characterization of Biomolecular Interfaces with SECM

309

FIG. 33 Ca2 þ concentration and pH profiles induced by the addition of A23187 at the cis side of a BLM recorded using a double-barreled microelectrode positioned on the trans side of the membrane. (Reproduced with permission from Ref. 208. Copyright 1997 Biophysical Society.)

coe⁄cient [215].These experiments were carried out after the incorporation of A23187 or nigericin into the BLM.These agents facilitated a net ion £ux in the presence of a 10-fold ion gradient when there was no net volume £ow across the BLM. The proton £ux measured in the pH pro¢les in fact represented a £ux of bu¡er-bound molecules. This study showed that the physiological model for USLs, which excludes the e¡ect of the di¡usion coe⁄cient, was inadequate. In fact, smaller molecules with a higher D value experienced a thicker USL than larger molecules, which di¡used more slowly.

310

Barker et al.

When water molecules move across BLMs, they ‘‘drag’’ molecules of solvent with them, and this methodology has been extended to explore solvent drag [216]. This phenomenon, where solute £ux is enhanced in the direction of water movement and retarded in the opposite direction, is known as ‘‘true’’ solvent drag. There is also, however, a competing process which can make the e¡ects of true solvent drag di⁄cult to resolve experimentally [212]. This pseudo-solvent drag results from additional solute £ux, due to a concentration gradient across the BLM, arising from permeating water concentrating the solution it leaves and diluting the solution it enters. The aim of the study [216] was to resolve the e¡ect of true solvent drag in the transport of ions across gramicidin channels in a BLM, via the simultaneous measurement of concentration pro¢les of channel-permeable andimpermeable cations. This experiment again utilized double-barreled, potentiometric ion-selective microelectrodes to record concentration pro¢les for Kþ and Ca2 þ in the vicinity of a BLM. The microelectrode probe was placed, with a reference electrode, on the trans side of the BLM. Transmembrane current was monitored using two pairs of Ag/AgCl electrodes, across the BLM, before and after each volume £ow measurement. The ¢rst pair of electrodes were used to generate a current step and the second measured the resulting potential di¡erence. Measurements were made which showed the e¡ects of both true and pseudo-solvent drag, and it was possible to resolve the two processes through investigations in a system where the concentrations of solute on either side of the membrane were identical. This experiment showed that microelectrodes could be used to monitor and resolve the e¡ects of the competing processes of ion and water £uxes across BLMs. From the concentration pro¢le and membrane potential difference data, it was possible to determine the hydraulic membrane permeability, the single-pore water permeability coe⁄cient, and the number of water molecules transferred per ion in single-¢le transport [216]. This methodology has been used by Pohl and coworkers [217,218] to examine related aspects of membrane transport. 8

INVESTIGATIONS OF THICK MEMBRANES: SYNTHETIC MATERIALS AND TISSUES

As already highlighted, the transport of species between two solutions separated by a membrane or porous material is a general phenomenon underpinning a number of major processes in biological systems. Solute transport can be driven in several ways: an applied electric ¢eld will lead to migration of ions, a pressure di¡erence will result in transport by convection, and a concentration gradient will promote di¡usion. There are several key questions to address when investigating permeability and transport

Characterization of Biomolecular Interfaces with SECM

311

properties in synthetic membranes and biological tissues. Is the rate of transport uniform or spatially localized? If transport is localized, where are the active sites and can the local transport rate be measured quantitatively? SECM is particularly suited to addressing these questions, since
under conditions where the UME is held at a potential to detect a target solution species by di¡usion-controlled electrolysis
the magnitude of the current £owing at the tip depends only on the local rate of mass transport. Thus, SECM can be used to obtain quantitative data on transmembrane £uxes, with micrometer to submicrometer resolution. Figure 34 illustrates schematically one approach for locating permeable regions in a target membrane which separates a donor and receptor compartment, across which mass transport occurs, by one of the processes identi¢ed above. Measurement of the tip current, as a function of position in the x ^ y plane, allows active regions to be identi¢ed from an enhancement in the current, compared

FIG. 34 Examining local membrane transport with SECM. Transport of species by convection, diffusion, or migration (promoted by pressure, concentration, or electric field gradientsthe latter for ions only) can be detected as an increase in the transport-limited current at a UME. A map of diffusion-limited current versus tip position can be converted to a corresponding interfacial transport map. (Reproduced with permission from Ref. 5. Copyright 1999 Elsevier.)

312

Barker et al.

to the response when the tip sits above an area of the interface where transport is purely by hindered di¡usion in the receptor compartment. The ¢rst studies of this type were made by White and coworkers [219^ 223], who used SECM to determine the major transport pathways of solutes across: track-etched polycarbonate (0.1-mm pore size) [219], punctured mica samples (pore sizes in the range 1^7 mm) [219,220], and excised hairless mouse skin [219,221,222]. These studies utilized an applied electric ¢eld to drive the transport of ions and were aimed at providing a better understanding of the processes involved in iontophoretic transdermal drug delivery. Studies of skin, in particular, were able to identify through which skin appendages transport occurred. White’s group demonstrated that, upon application of an electric ¢eld, the follicles (typical pore diameter 15^20 mm; density of several hundred/ cm2) in hairless mouse skin provided the dominant transport pathway, ultimately carrying almost 75% of the current (after an initial period of low activity). Moreover, these measurements were quantitative and down to the level of a single pore. Recently, the role of applied electric ¢eld on the transport of neutral molecules, through porous membranes, has been investigated [223]. In particular, protocols were devised for measuring electroosmotic drag coe⁄cients and convective velocities of neutral electroactive molecules emerging from an individual pore. Recent developments have seen this methodology extended to explore the solute transport properties of ion-selective membranes [224]. A further advance, to make the technique more applicable to real-life biological systems, has seen the introduction of the reverse imaging mode [225]. In this method, the electrode is positioned in the donor compartment of the iontophoresis cell, just above a pore in the membrane, and used to monitor the £ux of electroactive species at the membrane entrance. Work from our group has used the imaging capabilities of SECM to quantify the rates of convective £ow through tubules in dentine slices [226^ 228], subjected to £uid pressures similar to in vivo pulpal pressures. Complementary studies [229] examined di¡usive transport through dentine slices. Fluid £ow through exposed dentinal tubules in the tooth is important in the condition of hypersensitivity, and it is necessary to gain a fundamental understanding of £uid movement, at a local level, to develop e¡ective treatments. SECM studies demonstrated, for the ¢rst time, that convective rates across dentine varied dramatically at the local level [226,227]. By using UMEs with small radii (down to 1 mm), it was possible to determine the rate of convection of an electrolyte solution containing ferrocyanide ions to the level of single tubules (2-mm diameter) [227]. Figure 35 shows typical images of normalized transport-limited current for the one-electron oxidation of ferrocyanide as a function of tip position in the xy plane parallel to a

Characterization of Biomolecular Interfaces with SECM

313

FIG. 35 Images of the variation of normalized transport-limited current for the oneelectron oxidation of ferrocyanide with a tip (1.0-mm radius) scanned in a plane parallel to a dentine surface. The data were obtained with: (a) a pressure of 20-cm aqueous solution across a 50-mm-thick dentine slice and (b) no solution pressure across the slice. The current difference image (c) highlights the areas of the sample through which localized mass transport occurs. (Reproduced with permission from Ref. 5. Copyright 1999 Elsevier.)

dentine surface.The data were obtained with (a) a hydrostatic pressure of 20cm aqueous solution across a dentine slice and (b) in the absence of a pressure di¡erence. In the latter case, the current was due solely to hindered di¡usion of ferrocyanide, in the receptor phase containing the tip. By subtracting the image in Fig. 35b from that in Fig. 35a, a di¡erence plot of the normalized current, Di=ið1Þ, Fig. 35c, highlights clearly the regions of £ow. For this particular case, £ow was predominantly through one tubule, with

314

FIG. 35

Barker et al.

Continued.

several others showing limited activity. Since the 50-mm2 area contained ca. 80 tubules, it was concluded that the majority of tubules in this region showed no detectable £ow, probably due to occlusions subsurface. These results demonstrated that the localized £ow rates in a single tubule may be signi¢cantly di¡erent from the mean £ow rate obtained from measurements over the bulk sample. Subsequent work investigated the e¡ectiveness of blocking agents such as calcium oxalate [226] in occluding tubules and retarding £uid £ow. SECM imaging of transport in biological materials has been further extended to investigations of osmotically induced convective transport of solutes through soft tissues, using Ru(NH3)63þ in pig laryngeal cartilage as a model system [46]. SECM models allowed lateral variations of mass transport rates across the surface of cartilage to be visualized and compared directly to the corresponding topography, for the ¢rst time. Permeable areas of the tissue were identi¢ed and local £uid velocities on the mm/s scale determined. It was established that the interterritorial regions, i.e., the areas between cells in the surface of the cartilage presented to the UME, provided the most facile transport pathways. The SECMIT technique is particularly attractive for mass transport measurements in biological tissues. The technique is able to locally induce and probe solute permeability without the UME having to contact or enter the sample, so that the structural integrity of the tissue is maintained. SECMIT has been used successfully to measure the di¡usion coe⁄cient of oxygen in pig laryngeal cartilage. Figure 36 shows a typical steady-state approach curve for the di¡usion-limited reduction of oxygen at a 25-mmdiameter Pt disk electrode approaching a thin slice of cartilage in aerated aqueous electrolyte. Close to the interface, the measured currents are higher

Characterization of Biomolecular Interfaces with SECM

315

FIG. 36 Approach curve of normalized steady-state current versus probe/interface separation for the diffusion-controlled reduction of oxygen at a UME scanned toward a sample of laryngeal cartilage (O). The dashed line shows the theoretical response for an inert interface (hindered diffusion only of oxygen in the aqueous phase containing the UME), while the solid line shows the behavior for induced transfer with the oxygen diffusion coefficient having a value of 50% of that in aqueous solution. The partition coefficient for oxygen between the aqueous and cartilage phases is considered to be unity. (Reproduced with permission from Ref. 5. Copyright 1999 Elsevier.)

than predicted for an inert surface, since the electrolysis process promotes the transfer of oxygen from the cartilage matrix to the aqueous solution, enhancing the £ux at the UME. Through these measurements, the spaceaveraged di¡usion coe⁄cient of oxygen in cartilage was estimated to be ca. 50% of that in aqueous solution [46]. A key issue in the study of surfaces using SECM is resolving sample topography and reactivity. This can be achieved with dual-mediator imaging, in which one (impermeable) chemical species is used to give a measure of tip ^ substrate distance, by way of a hindered di¡usion response. The response for the second mediator then provides information on the

316

Barker et al.

activity of the sample.This approach has been used to good e¡ect to measure the distribution and permeability of molecular oxygen in cartilage samples [47]. A ¢xed-height x--y scan detecting the impermeable mediator, Ru(CN)6 4, in the bathing solution provided information on the sample topography. Rescanning the same area, but detecting oxygen by di¡usionlimited reduction, provided a permeability response, convoluted by topographical e¡ects. With knowledge of the topography, maps of oxygen permeability were obtained, as shown in Fig. 37. In this ¢gure, permeability

FIG. 37 Contour plots showing (a) a normalized current map over a cartilage surface, obtained via the oxidation of Ru(CN)64; (b) the corresponding topography plot; (c) a normalized current map for the reduction of oxygen; and (d) the corresponding oxygen permeability map of the tissue.

Characterization of Biomolecular Interfaces with SECM

FIG. 37

317

Continued.

is expressed as Ke g,where Ke is the partition coe⁄cient of the solute between the two phases and g is the ratio of the di¡usion coe⁄cients of the solute in the two phases. This technique can be further enhanced by using local ‘‘electrochemical staining,’’ e.g., the electrodeposition of palladium [48], to mark the area of the sample imaged for subsequent histochemical and structural analysis. 9

CELLS

SECM is being used increasingly to image the topography and biochemical activity of biological cells and other living entities. Advantages of SECM in

318

Barker et al.

this ¢eld are typical of many of the applications already mentioned in this chapter. The noninvasive nature of the probe, and small tip sizes available, enable biochemical processes to be studied down to the level of individual cells,while the cell is maintained in a functioning state.The SECM tip can be used as a local in situ probe of the response of the cell to external stimuli, such as irradiation or adding chemical species to a medium. An important factor to consider when making SECM measurements on living systems is that many of the conventional electrochemical redox mediators are often toxic, and thus the choice of mediator may be restricted. In early work, the surface topography of detached leaves was examined using the SECM feedback mode, with ferrocyanide as the electroactive mediator [19]. SECM current images revealed individual stoma (corresponding to peaks in the tip current), £anked by guard cells that bulged above the surrounding epidermal cells (corresponding to regions of low tip current). SECM maps of the variation in oxygen concentration over the upper surface of an Elodea leaf, in the dark and under illumination, demonstrated locations of enhanced oxygen £ux due to photosynthesis, corresponding to dilated stomata. The surface topography and photosynthetic oxygen evolution of a leaf of Tradescantia £uminensis has been studied in vivo for an intact plant [20]. The leaf was immersed in bu¡er solution to maintain the surface pH constant.To improve the sensitivity to the low concentrations of oxygen generated by photosynthesis, the whole experimental apparatus and plant were brie£y placed in a glove bag to control the background oxygen concentration levels. The variegated leaf studied contained green regions where chloroplasts where present in mesophyll and in guard cells and white regions where functional chloroplasts were present exclusively in guard cells. Probing oxygen generation over the white regions of the leaf enabled direct observation of the photosynthetic electron transport of single guard cells. The photosynthetic activity of speci¢c white and green regions was followed over several minutes in response to switching the light irradiation on and o¡. The respiratory and photosynthetic activities of single living protoplasts derived from marine alga,Bryopsis plumosa, have also been investigated [230].The cellular activity was imaged in the dark and under illumination, by monitoring the concentration of oxygen near to the protoplast via the oxygen reduction current at an UME tip. In the dark, the SECM image corresponded to oxygen consumption near to the protoplast, due to respiration, whereas under illumination oxygen was generated by photosynthesis. SECM images showed a decline in the photosynthetic activity of the protoplast, following injection of 3-(3,4-dichlorophenyl)-1,1-dimethylurea,which is a known inhibitor of electron transfer in the photosynthetic chain. Figure 38 shows approach curves for the localized concentration of oxygen obtained for dif-

Characterization of Biomolecular Interfaces with SECM

319

FIG. 38 Variation of oxygen concentration as a function of the UME^protoplast (center) distance under light irradiation with different intensity. Probe: Pt UME (radius 1.1 mm). Radius of the protoplast: 50 mm; light intensity: j 25,  15, r 10, m 5, d 0 kLx. (Reproduced with permission from Ref. 2. Copyright 2000 Wiley.)

ferent light intensities [2]. Modeling mass transport gave an oxygen consumption rate of 2 1014 mol/s in the dark due to respiration. Oxygen generation in the light became saturated for light intensities >15 kLx, corresponding to an evolution rate of  7.5 1014 mol/s at a single cell. The in£uence of p-benzoquinone,which can permeate through the cell membrane and act as an e⁄cient electron acceptor in the photosynthetic electron-transport chain, has been investigated by detecting the localized concentration of p-benzoquinone/p-hydroquinone near to a protoplast upon irradiation [231]. With 1 mM p-benzoquinone in the medium, oxygen production in the light (25 kLx) increased to a steady state, indicating that the process was not limited by the rate of regeneration of photosynthesisrelated chemicals by the dark reaction, which is otherwise important in the absence of p-benzoquinone. Dual microdisk electrodes have been employed to image topography and cellular activity simultaneously in an SECM scan. The topographic pro¢le of a single algal protoplast was obtained by measuring the variation of the oxidation current of ferrocyanide at one microdisk electrode, while photosynthetic oxygen generation was followed at the second microdisk [68] (Fig. 39). In these constant-height images, the ferrocyanide oxidation current decreases as the tip UME passes over the cell, due to hindered di¡usion. In contrast, the current for O2 reduction increases over the cell due to the photosynthetic process.

320

Barker et al.

FIG. 39 Dual-probe SECM images of a single, living protoplast (radius, 25 mm) based on the oxidation current for Fe(CN)64 (a), and the reduction current for oxygen (b). Probe, dual-microdisk electrode with a leveled (a) and a beveled (b) disk. Scan rate: 9.8 mm/s. Concentration of Fe(CN)64, 1.0 mM. The radii of leveled and beveled disks were 2.8 and 1.2 mm, respectively. (Reproduced with permission from Ref. 68. Copyright 1999 American Chemical Society.)

Characterization of Biomolecular Interfaces with SECM

321

A di¡erent strategy involved developing C-¢ber-based UMEs which would enable shear-force distance regulation or topographic measurement (see Sec.10.2) to be used in combination with electrochemical imaging [232]. The shear-force response was used to image the topography of a single cell of a pheochromocytoma tumor cell line and to position the electrode close to a region on the cell of interest. The electrochemical response of the tip was then used to follow the secretion of neurotransmitter molecules into extracellular space, following an induced exocytosis event. SECM has been used to image activity of NADH-linked reductases of individual immobilized rat liver mitochondria [15]. The principles of this approach are similar to the feedback imaging experiments involving GOx described in Sec. 5.1. Reductase enzymes residing on, or within, the outer membrane catalyze reduction of a moiety in the optimal pH range of 5.5^8.2 by accepting two electrons from NADH and transferring the electrons to the oxidant in two one-electron steps. In these studies, oxidized N,N,N0,N0 ^ tetramethyl ^ p ^ phenylenediamine (TMPDþ ) generated at a UME tip functioned as the electron acceptor, competing with the natural acceptors involved in the catalytic cycle. TMPD was regenerated by the catalytic reaction at the mitochondrion membrane, and thus detected as a feedback current at the tip. Enhancements in the tip current, indicating an active enzymatic site, were observed when the tip was positioned directly above mitochondria. The enhancements were small, however, due to the concentration of TMPD used being slightly above that de¢ned by the detection limit criteria for this system [see Eq. (6)]. Respiratory activity of single cells has been investigated using cultivated cancer cell lines [233].The current for oxygen reduction directly above a cell was used as a measure of cellular respiratory activity. Cyanide was added to block the mitochondrial electron transport chain, ultimately resulting in cessation of respiration and consequent cell death. For an initial period of 500 s following exposure to KCN solution, the respiratory rate remained stable, then increased until 800 s had elapsed, at which point respiration decreased terminally. This trend was attributed to the concentration of CN within the mitochondrial membrane lagging behind the extracellular concentration, as CN has to cross both cell and mitochondrial membranes and a critical intracellular concentration of CN is required to prove lethal to the cell. The permeability of the cell and mitochondrial membranes to CN was estimated to be of the order of 107cm/s, using a simple model for CN transport into the cell. Redox activity of individual metastatic and nonmetastatic human breast cells has been investigated by SECM feedback experiments [234]. When highly charged hydrophilic mediators such as Fe(CN)63/4 and Ru(NH)63/2þ were used, negative feedback behavior was observed,

322

Barker et al.

indicating that these mediators were not able to cross the cell membrane and undergo intracellular redox processes. Use of the neutral lipid-soluble mediators, napthaquinone and menadione, gave enhanced currents when the SECM tip was positioned directly over a cell. Approach curves to single cells allowed extraction of apparent reoxidation rates of these species by utilizing the analytical expressions for ¢nite heterogeneous kinetics, as described by Eqs. (3)^(5). Di¡erent responses were observed for (1) nonmotile, nontransformed human breast epithelial cells, (2) motile breast cells which overexpress protein kinase Ca, and (3) highly metastatic breast cancer cells. Lower mediator regeneration rates were observed for the motile cells,which may re£ect the lower metabolic activity of these cells compared to normal cells. Shiku et al. have recently used SECM to probe the gradient of oxygen near to single living bovine embryos, during di¡erent developmental stages, as a re£ection of metabolic activity [235]. Using oxygen reduction approach curves, the di¡erence in localized oxygen concentration at the surface of the embryo to that in bulk solution was compared for 6-day morulae embryos (50^100 cells) and the inner cell mass and trophoblast regions of 8-day blastocyst embryos (110^200 cells). Oxygen consumption of individual morula varied between embryos and showed a strong correlation with cell morphology. Morulae with higher oxygen consumption tended to show stronger potential for further development to the blastocyst stage and generally gave rise to larger embryos. In a quite di¡erent application, functioning osteoclast cells have been studied by SECM [236]. These cells regulate dissolution of calcium from bone (resorption) and are implicated in bone-wasting diseases such as osteoporosis. Berger et al. utilized a calcium ion-selective liquid-membrane potentiometric microelectrode to monitor calcium dissolution from bone slices in the presence of adherent osteoclasts [236]. Presoaking bone slices with ammonium £uoride gave resistance to osteoclast-mediated resorption. The presence of a calcium £ux showed that osteoclasts were still attached to the bone slices and mediating resorption. Comparative studies using scanning electron microscopy showed that the number of bone-resorbing pits was not signi¢cantly a¡ected, but the excavated area was reduced by £uoride pretreatment [236]. Thus, the microelectrochemical probe allowed real-time analysis of calcium £ux during the resorption process. 10

LOOKING TO THE FUTURE

The increasing application of SECM to living systems is likely to be a major area of application in the future; the preceding section has highlighted some

Characterization of Biomolecular Interfaces with SECM

323

of the important problems that SECM can address. A key methodological advance that will facilitate many future applications, across a range of areas, will be the development of probes that allow the ready separation of topographical and activity e¡ects in SECM experiments. Some of the recent advances in SECM instrumentation that should realize this goal are considered in this section. 10.1

SECM-Atomic Force Microscopy (AFM)

As illustrated amply throughout this chapter, SECM imaging most often utilizes the ¢xed-height mode, in which the tip is held at a certain distance above the substrate while imaging in a ¢xed plane. Consequently, the tip is unable to follow the contours of the surface and topographical information is most often obtained by SECM in the negative feedback mode [13]. However, it is not always possible to ¢nd a mediator suitable for this purpose; moreover, the resolution of this approach is limited by the size of the UME probe. Recently, nonelectrochemical methodologies have been developed which lift this constraint. AFM maps the topography of a substrate with nanometer vertical resolution, by monitoring the interaction force between the sample and a sharp tip, which is attached to the end of a force-sensing cantilever [237^239]. By integrating an electrode into the AFM probe design it is possible to have a probe which provides both electrochemical and high-resolution height information, thus enabling structure ^ activity-related problems to be mapped at the submicrometer level. To data, two approaches to fabricating SECM-AFM probes have been employed. The ¢rst employs homemade probes produced by coating an etched and £attened microwire with an electrophoretic paint [240,241]. The £attened section provides a £exible cantilever (force sensor), and the coated etched tip acts as an electrode. SECM-AFM probes with conically shaped electrodes of size 10^1000 nm have been fabricated in this way. The probes can be used in conjunction with any commercially available AFM instruments. Via the attached cantilever, it is possible to determine precisely the point of contact of the electrode with the substrate, and control the tip ^substrate separation with nanometer resolution.The tip can either be scanned in contact with the surface (to provide electrical information [242]), at a ¢xed height, or with a constant separation, i.e., following the surface contours [241]. Figure 40 shows a typical SEM image of an SECM-AFM probe, along with tip de£ection and SECM approach curves, obtained with such a probe. In this particular case, the probe was translated to and from an inert glass surface while recording simultaneously the tip current for the di¡usionlimited oxidation of IrCl63 and the tip de£ection.

324

Barker et al.

FIG. 40 (a) Scanning electron micrograph of a typical SECM-AFM probe. Simultaneously recorded cantilever deflection, (b), and the diffusion-controlled limiting current, (c), for the oxidation of 10 mM IrCl63 in 0.5 M KCl, as a function of distance, z, of an SECM-AFM tip from an inert glass substrate. The surface was scanned toward and away from the tip, biased at þ 1.0 V versus AgQRE, at a rate of 0.2 mm/s. The theory lines()were obtained assuming a conical tip geometry characterized by the (radius of cone)/(height of cone) ¼ 1.0 (lower curve), 1.5, 2.0, 2.5, and 3.0 (upper curve). [(a) reproduced with permission from Ref. 241. Copyright 2001 American Chemical Society. (b) and (c) reproduced with permission from Ref. 240. Copyright 2001 American Chemical Society.]

Characterization of Biomolecular Interfaces with SECM

FIG. 40

325

Continued.

SECM-AFM tips of this type can be used to correlate surface activity with surface structure. For example, the probes have been used to investigate the di¡usional transport of an electroactive mediator through a track-etched polycarbonate membrane (average pore diameter 600 nm) [240]. The SECM-AFM images obtained demonstrated a correlation between surface topography and di¡usional activity, at a submicrometer level [240]. The second method for producing SECM-AFM tips employs a conventional AFM probe as the starting platform. Electrochemical capability is built into the device in one of two ways. The simplest involves sputtering or evaporating a thin metal ¢lm (typically Pt) onto the AFM probe [45,243]. By coating the main body of the probe with a suitable insulator such as nail polish or polystyrene, the electroactive area of the device will be governed by the dimensions of the exposed metal cantilever. A single or double beam of length 100^200 mm, width 20^40 mm, represents the geometry of the cantilever. Although this area is much larger than that for a typical SECM electrode, probes of this kind enable electrochemical control to be exerted on an interfacial processes, while recording corresponding changes in topography at ultrahigh spatial resolution [45,244]. This technique has been used successfully to elucidate the dissolution mechanisms of various ionic crystals [45,244]. This type of probe can be used for high-resolution electrochemical imaging in air [243]. A more sophisticated approach involves depositing a second layer of an insulating material over the entire model-coated probe and then exposing the electrode and remodeling and sharpening the original AFM tip using a focused ion beam [245]. In this way, a square-shaped electrode, positioned at the base of the tip is produced, as shown in Fig. 41. SECM imaging with this

326

Barker et al.

FIG. 41 (a) Schematic top view of the geometric factors of a frame microelectrode. (b) Scanning electron micrograph of an integrated frame microelectrode having an edge length of 1.5 mm. (Reproduced with permission from Ref. 245. Copyright 2001 American Chemical Society.)

Characterization of Biomolecular Interfaces with SECM

327

probe at a constant distance from the substrate surface thus occurs with the separation de¢ned by the distance between the electrode and the apex of the tip. 10.2

Shear Force Topographical Imaging

An alternative method for obtaining information on the tip ^ substrate separation can be achieved by ‘‘dithering’’ the electrode, via a small oscillation in the x ^ y plane. As the electrode is brought close to a surface, the oscillation is damped to a degree depending on the tip ^ substrate separation [246]. Images are usually acquired at a constant damping amplitude, analogous to constant-distance imaging. The oscillation amplitude is mainly monitored in one of two ways, either: (1) using a laser which is focused at the end of the tip electrode, with the signal detected by a split photodiode [246,247]; or (2) by monitoring the vibration amplitude of a tuning fork attached to the electrode [248,249]. With this experimental arrangement, it is possible to replace conventional SECM probes with an open glass capillary, opening up the possibility of ¢lling the capillary with a myriad of ‘‘chemical cocktails.’’ For example, in one study [247], the glass capillary was ¢lled with a biocatalyst, which could be released and detected at an underlying Pt electrode. Since the topographical resolution of this type of approach is determined by the diameter of the overall probe, the use of small UMEs, such as those discussed in Sec. 3.3.3, is essential for high-resolution electrochemical and topographical imaging [80]. 10.3

SECM-Scanning Near-Field Optical Microscopy (SNOM)

SNOM provides a means of probing the optical properties of substrates at the nanometer level [250], and holds tremendous possibilities for biological systems in particular. SNOM utilizes a near-¢eld light source, scanned close to the surface of the sample.The aperture of the SNOM probe is smaller than the optical wavelength, so that at very close tip ^ substrate separations, the interaction with the sample is not limited by di¡raction. In this way, the spatial resolution of the technique is limited only by the size of the aperture and the tip ^ substrate separation and not the wavelength of the light source. Optical ¢bers etched to a sharp point, or pulled glass capillaries, are often employed as the light guides [250].Control of the tip ^ substrate separation in SNOM often incorporates shear force or AFM feedback methods. To minimize the loss of light in transmission from the laser source to the tip end, the probe is often coated with a thin metallic layer (such as A1).

328

Barker et al.

The addition of an insulating coating to the metallic ¢lm results in the formation of a ring UME [251], and the prospect of simultaneous electrochemical and optical measurements under tip ^ substrate distance regulation. The formation of ultrasmall ring electrodes of this type is challenging, but there has been some recent encouraging progress in this area [251]. Using a varnish insulating coat, SECM and photoelectrochemical measurements were reported [252,253]. Only far-¢eld optical imaging has been discussed so far, with a spatial resolution in the micrometer range, but further developments should see this technology extended to the near ¢eld. ACKNOWLEDGMENTS We are grateful to the EPSRC and BBSRC for support of various aspects of our work on SECM. Anna L. Barker thanks the Wellcome Trust for a postdoctoral research fellowship, Catherine E. Gardner is the recipient of an EPSRC/RSC Analytical Sciences studentship, Julie V. Macpherson thanks the Royal Society for a University Research Fellowship, and Jie Zhang thanks the ORS Committee, the University of Warwick, and Avecia for scholarships. Helpful discussions with past members of the Warwick Electrochemistry and Interfaces Group, especially Phil Dobson, Lou Gonsalves, Nicky Gray, and Chris Slevin, are much appreciated. REFERENCES 1.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

AL Barker, CJ Slevin, PR Unwin, J Zhang. In: AG Volkov, ed. Liquid Interfaces in Chemical, Biological and Pharmaceutical Applications. New York: Marcel Dekker, 2001, p 283. T Yasukawa,T Kaya,T Matsue. Electroanalysis 12:653, 2000. S Amemiya, ZF Ding, JF Zhou, AJ Bard. J Electroanal Chem 483:7, 2000. MV Mirkin, BR Horrocks. Anal Chim Acta 406:119, 2000. AL Barker, M Gonsalves, JV Macpherson, CJ Slevin, PR Unwin. Anal Chim Acta 385:223, 1999. MV Mirkin. Mikrochimica Acta 130:127, 1999. RC Engstrom, M Weber, DJ Wunder, R Burgess, S Winquist. Anal chem 58:844, 1986. RC Engstrom,T Meaney, R Tople, RM Wightman, Anal Chem 59:2005, 1987. HY Liu, F-RF Fan, CW Lin, AJ Bard. J Am Chem Soc 108:3838, 1986. YH Lee, GT Tsao, PC Wankat. Ind Eng Chem Fundam 17:59, 1978. AJ Bard, F-RF Fan, J Kwak, O Lev. Anal Chem 61:132, 1989. J Kwak, AJ Bard. Anal Chem 61:1221, 1989. J Kwak, AJ Bard. Anal Chem 61:1794, 1989. DT Pierce, PR Unwin, AJ Bard. Anal Chem 64:1795, 1992. DT Pierce, AJ Bard. Anal Chem 65:3598, 1993. H Shiku,T Takeda, H Yamada,T Matsue, I Uchida. Anal Chem 67:312, 1995.

Characterization of Biomolecular Interfaces with SECM 17. 18.

19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49.

329

H Shiku,T Matsue, I Uchida. Anal Chem 68:1276, 1996. H. Shiku,Y Hara,T Takeda,T Matsue, I Uchida. In: G Jerkiewicz, MP Soriaga, K Uosaka, A Wieckowski, eds. Solid-Liquid Electrochemical Interfaces. Washington, DC: American Chemical Society, 1997, p 202. C Lee, J Kwak, AJ Bard. Proc Natl Acad Sci, USA 87:1740, 1990. M Tsionsky, ZG Cardon, AJ Bard, RB Jackson. Plant Physiol 113:895, 1997. J Kwak, C Lee, AJ Bard. J Electrochem Soc 137:1481, 1990. C Lee, AJ Bard. Anal Chem 62:1906, 1990. MV Mirkin, F-R Fan, AJ Bard. Science 257:364, 1992. IC Jeon, FC Anson. Anal Chem 64:2021, 1992. G Denuault, MH Troise-Frank, LM Peter. Faraday Discuss Chem Soc 94:23, 1992. MH Triose-Frank, G. Denuault. J Electroanal Chem 354:331, 1993. M Pyo, AJ Bard. Electrochim Acta 42:3077, 1997. AJ Bard, MV Mirkin, eds. Scanning Electrochemical Microscopy. New York: Marcel Dekker, 2001. C Amatore, S Szunerits, LThouin, J-S Warkocz. Electrochem Commun 2:353, 2000. MH Troise-Frank, G Denuault. J Electroanal Chem 379:399, 1994. MV Mirkin, M Arca, AJ Bard. J Phys Chem 97:10790, 1993. F-RF Fan, MV Mirkin, AJ Bard. J Phys Chem 98:1475, 1994. Y Shao, MV Mirkin. J Electroanal Chem 439:137, 1997. Y Shao, MV Mirkin. J Phys Chem 102:9915, 1998. C Kranz, G Wittstock, H Wohlschla«ger, W Schuhmann. Electrochim Acta 42:3105, 1997. AL Barker, JV Macpherson, CJ Slevin, PR Unwin. J Phys Chem B 102:1586, 1998. NJ Evans, M Gonsalves, NJ Gray, AL Barker, JV Macpherson, PR Unwin. Electrochem Commun 2:201, 2000. PR Unwin, AJ Bard. J Phys Chem 96:5035, 1992. JV Macpherson, PR Unwin. J Chem Soc FaradayTrans 89:1883, 1993. JV Macpherson, PR Unwin. J Phys Chem 98:1704, 1994. JV Macpherson, PR Unwin. J Phys Chem 98:11764, 1994. JV Macpherson, PR Unwin. J Phys Chem 99:3338, 1995. JV Macpherson, PR Unwin. J Phys Chem 99:14824, 1995. JV Macpherson, PR Unwin. J Phys Chem 100:19465, 1996. JV Macpherson, PR Unwin, AC Hillier, AJ Bard. J Am Chem Soc 118:6445, 1996. JV Macpherson, D O’Hare, PR Unwin, CP Winlove. Biophys J 73:2771, 1997. M Gonsalves, AL Barker, JV Macpherson, PR Unwin, D O’Hare, CP Winlove. Biophys J 78:1578, 2000. M Gonsalves, JV Macpherson, D O’Hare, CP Winlove, PR Unwin. Biochim Biophys Acta 1524:66, 2000. CJ Slevin, JA Umbers, JH Atherton, PR Unwin. J Chem Soc Faraday Trans 92:5177, 1996.

330

Barker et al.

50. 51. 52. 53. 54.

CJ Slevin, S Ryley, DJ Walton, PR Unwin. Langmuir 14:5331, 1998. CJ Slevin, JV Macpherson, PR Unwin. J Phys Chem B 101:10851, 1997. J Zhang, AL Barker, PR Unwin. J Electroanal Chem 483:95, 2000. AL Barker, PR Unwin. J Phys Chem B 105:12019, 2001. J Zhang, CJ Slevin, C Morton, P Scott, DJ Walton, PR Unwin, J Phys Chem B 105:11120, 2001. CP Andrieux, P Hapiot, JM Save¤ant. J Phys Chem 92:5992, 1988. JV Macpherson, PR Unwin. Anal Chem 69:2063, 1997. CJ Slevin, PR Unwin. J Am Chem Soc 122:2597, 2000. BD Bath, HS White, ER Scott. In: AJ Bard, MV Mirkin, eds. Scanning Electrochemical Microscopy. New York: Marcel Dekker, 2001, p 343. D. Amman. Ion-Selective Microelectrodes: Principles, Design and Application. Berlin: Springer-Verlag, 1986. C Wei, AJ Bard, G Nagy, K Toth. Anal Chem 67:1346, 1995. C Kranz, M Ludwig, HE Gaub,W Schuhmann. Adv Mater 7:38, 1995. C Kranz, M Ludwig, HE Gaub,W Schuhmann. Adv Mater 7:568, 1995. P Hliva, I Kapui, L Czako, G Nagy. Magyar Kemiai Folyoirat 104:223, 1998. RD Martin, PR Unwin. Anal Chem 70:276, 1998. RD Martin, PR Unwin. J Electroanal Chem 439:123, 1997. AJ Bard, F-RF Fan, MV Mirkin. In: AJ Bard, ed. Electroanalytical Chemistry. Vol. 18. New York: Marcel Dekker, 1993, p 243. RM Wightman, DO Wipf. In: AJ Bard, ed. Electroanalytical Chemistry. Vol. 15. New York: Marcel Dekker, 1989, p 267. T Yasukawa,T Kaya,T Matsue. Anal Chem 71:4637, 1999. AA Gewirth, DH Craston, AJ Bard. J Electroanal Chem 261:477, 1989. B Zhang, E Wang. Electrochim Acta 39:103, 1994. RM Penner, MJ Heben,TL Longin, NS Lewis. Science 250:1118, 1990. RM Penner, MJ Heben, NJ Lewis. Anal Chem 61:1630, 1989. LA Nagahara,T Thundat, SM Lindsay. Rev Sci Instrum 60:3128, 1989. CJ Slevin, NJ Gray, JV Macpherson, MA Webb, PR Unwin. Electrochem Commun 1:282, 1999. JL Conyers, HS White. Anal Chem 72:4441, 2001. CE Bach, RJ Nichols, H Meyer, JO Besenhard. Surf Coatings Technol 67:139, 1994. K Potje-Kamloth, J Janata, M Josowicz. Ber Bunsenges Phys Chem 93:1480, 1993. BD Pendley, HD Abrun‹a. Anal Chem 62:782, 1990. Y Shao, MV Mirkin, G Fish, S Kokotov, D Palanker, A Lewis. Anal Chem 69:1627, 1997. B B Katemann,W Schuhmann. Electroanalysis 14:22, 2002. JM Davis, F-RF Fan, AJ Bard. J Electroanal Chem 238:9, 1987. MV Mirkin, F-RF Fan, AJ Bard. J Electroanal Chem 328:47, 1992. Y Selzer, D Mandler. Anal Chem 72:2383, 2000. AG Volkov, ed. Liquid Interfaces in Chemical, Biological and Pharmaceutical Applications. New York: Marcel Dekker, 2001.

55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84.

Characterization of Biomolecular Interfaces with SECM 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119.

331

MC Osborne, Y Shao, CM Pereira, HH Girault. J Electroanal Chem 364:155, 1994. T Solomon, AJ Bard. Anal Chem 67:2787, 1995. S Amemiya, AJ Bard. Anal Chem 72:4940, 2000. TJ Kemp, PR Unwin, LVincze. J Chem Soc FaradayTrans 91:3893, 1995. NJ Gray, PR Unwin. Analyst 125:889, 2000. A Eftekhari. Anal Lett 34:1087, 2001. BR Horrocks, MV Mirkin, DT Pierce, AJ Bard, G Nagy, K Toth. Anal Chem 65:1213, 1993. E Klusmann, JW Schultze. Electrochim Acta 42:3123, 1997. MV Mirkin, AJ Bard. J Electroanal Chem 323:1, 1992. MV Mirkin, AJ Bard. J Electroanal Chem 323:29, 1992. AJ Bard, G Denuault, RA Friesner, BC Dornblaser, LS Tuckerman. Anal Chem 63:1282, 1991. PR Unwin, AJ Bard. J Phys Chem 95:7814, 1991. AJ Bard, MV Mirkin, PR Unwin, DO Wipf. J Phys Chem 96:1861, 1992. Q Fulian, AC Fisher, G Denuault. J Phys Chem B 103:4387, 1999. Q Fulian, AC Fisher, G Denuault. J Phys Chem B 103:4393, 1999. RW Johnson, ed. The Handbook of Fluid Dynamics. Boca Raton, FL: CRC Press, 1998. GF Hewitt, GL Shires,YV Polezhaev, eds. International Encyclopedia of Heat and MassTransfer. Boca Raton, FL: CRC Press, 1997. SS Rao. The Finite Element Method in Engineering. New York: Pergamagon, 1982. KJ Huebner, EA Thornton,The Finite Element Method for Engineers. 2nd ed. New York: Wiley-Interscience, 1982. S. Feldberg. In: Computers in Chemistry and Instrumentation. New York: Marcel Dekker, 1972, p 185. T Joslin, D Pletcher. J Electroanal Chem 172:49, 1974. SW Feldberg. J Electroanal Chem 127:1, 1981. AJ Bard, JA Crayston, GP Kittlesen,T Varco Shea, MS Wrighton. Anal Chem 58:2321, 1986. DW Peaceman, HH Rachford. J Soc Ind Appl Math 3:28, 1955. J Heinze. J Electroanal Chem 124:73, 1981. J Heinze. Ber Bunsen-Ges. Phys Chem 85:1096, 1981. J Heinze, M Sto«rzbach, Bunsen-Ges. Phys Chem 90:1042, 1986. G Taylor, HH Girault, J McAleer. J Electroanal Chem 293:19, 1990. DJ Gavaghan, JS Rollett. J Electroanal Chem 295:1, 1990. RA Friesner, LS Tuckerman, BC Dornblaser, TY Russo. J Sci Comput 4:327, 1989. F Zhou, PR Unwin, AJ Bard. J Phys Chem 96:4917, 1992. C. Demaille, PR Unwin, AJ Bard. J Phys Chem 100:14137, 1996. RD Martin, PR Unwin. J Chem FaradayTrans 94:753, 1998. JL Amphlett, G Denuault. J Phys Chem B 102:9946, 1998. KA Ellis, MD Pritzker,TZ Fahidy. Anal Chem 67:4500, 1995.

332

Barker et al.

120. C Wei, AJ Bard, MV Mirkin. J Phys Chem 99:16033, 1995. 121. M Tsionsky, AJ Bard, MV Mirkin. J Phys Chem 100:17881, 1996. 122. AL Barker, PR Unwin, S Amemiya, J Zhou, AJ Bard. J Phys Chem B 103:7260, 1999. 123. JStrutwolf,JZhang,ALBarker,PRUnwin.Phys ChemChemPhys3:5553, 2001. 124. J Strutwolf, AL Barker, M Gonsalves, DJ Caruana, PR Unwin, DE Williams, JRP Webster. J Electroanal Chem 483:163, 2000. 125. CA Wijayawardhana, G Wittstock, HB Halsall, WR Heinenman. Anal Chem 72:333, 2000. 126. G Wittstock,W Schuhmann. Anal Chem 69:5059, 1997. 127. I Turyan,T Matsue, D Mandler. Anal Chem 72:3431, 2000. 128. H Zhou, S Kasai,T Yasukawa. Electrochemistry 67:1135, 1999. 129. T Yasukawa, N Kanaya, D Mandler,T Matsue. Chem Lett 458, 2000. 130. J Zaumseil, G Wittstock, S Bahrs, P Steinru«cke. Anal Chem 367:352, 2000. 131. G Wittstock, K Yu, HB Halsall, TH Ridgway, WR Heineman. Anal Chem 67:3578, 1995. 132. BR Horrocks, G Wittstock. In: AJ Bard, MV Mirkin, eds. Scanning Electrochemical Microscopy. New York: Marcel Dekker, 2001, p 445. 133. H Yamada, H Shiku, T Matsue, I. Uchida. Bioelectrochem Bioenerget 33:91, 1994. 134. T Sawaguchi,T Matsue, I Uchida. Bioelectrochem Bioenerget 29:127, 1992. 135. S Kasai, AYokota, H Zhou, M Nishizawa, K Niwa,T Onouchi,T Matsue. Anal Chem 72:5761, 2000. 136. H Shiku, Y Hara, T Matsue, I Uchida, T Yamauchi. J Electroanal Chem 438:187, 1997. 137. BR Horrocks, MV Mirkin. J Chem Soc FaradayTrans 94:1115, 1998. 138. W Schindler, D Hofmann, J Kirschner. J Appl Phys 87:7007, 2000. 139. DM Kolb, R Ullmann,T Will. Science 275:1097, 1997. 140. F Marchi, D Tonneau, H Dallaporta, V Safarov, V Bouchiat, P Doppelt, R Even, L Beitone. J Vacuum Sci Technol B 18:1171, 2000. 141. J Hartwich, L Dreeskornfeld,V Heisig, S Rahn, O Wehmeyer, U Kleineberg, U Heinzmann. Appl Phys A 66:S685, 1998. 142. R Po«tzschke, G Staikov,WJ Lorenz,W Wiesbeck. J Electrochem Soc 146:141, 1999. 143. R Schuster, V Kirchner, XH Xia, AM Bittner, G Ertl. Phys Rev Lett 80:5599, 1998. 144. W Li, GS Hsiao, D Harris, RM Ny¡enegger, JAVirtanen, RM Penner. J Phys Chem 100:20103, 1996. 145. S Hong, CA Mirkin. Science 288:1808, 2000. 146. D Mandler. In: AJ Bard, MV Mirkin, eds. Scanning Electrochemical Microscopy. New York: Marcel Dekker, 2001, p 593. 147. C Kranz, HE Gaub,W Schuhmann. Adv Mater 8:634, 1996. 148. G Wittstock, R Hesse,W Schuhmann. Electroanlysis 9:746, 1997. 149. T Wilhelm, G Wittstock. Mikrochim Acta 133:1, 2000. 150. H Shiku, I Uchida,T Matsue. Langmuir 13:7239, 1997.

Characterization of Biomolecular Interfaces with SECM 151. 152. 153.

333

WB Nowall, DO Wipf,WG Kuhr. Anal Chem 70:2601, 1998. JC O’Brian, J Shumakerparry, RC Engstrom. Anal Chem 70:1307, 1998. H Mo«hwald. In: R Lipowsky, E Sackmann, eds. Structure and Dynamics of Membranes.Vol. 1A. Amsterdam: Elsevier, p 161. 154. RB Gennis. Biomembranes. New York: Springer-Verlag, 1989. 155. AJ Bard, F-RF Fan, MV Mirkin. In: I Rubinstein ed. Physical Electrochemistry: Principles, Methods and Applications. New York: Marcel Dekker, 1995, p 209. 156. AN Shipway, I Willner. Chem Commum 2035, 2001. 157. S Santra, P Zhang, KM Wang, R Tapec,WH Tan. Anal Chem 73:4988, 2001. 158. JJ Storho¡, AA Lazarides, RC Mucic, CA Mirkin, RL Letsinger, GC Schatz. J Am Chem Soc 122:4640, 2000. 159. AA Lazarides, CC Schatz. J Phys Chem B 104:460, 2000. 160. R Mahtab, HH Harden, CJ Murphy. J Am Chem Soc 122:14, 2000. 161. T Torimoto, M Yamashita, S Kuwabata, H Mori, H Yoneyama. J Phys Chem B 103:8799, 1999. 162. AG Volkov, DW Deamer, DL Tanelian, VS Markin. Liquid Interfaces in Chemistry and Biology. New York: Wiley, 1998. 163. HH Girault, DJ Schi¡rin. In: AJ Bard ed. Electroanalytical Chemistry.Vol. 15. New York: Marcel Dekker, 1989, p 1. 164. M Tsionsky, AJ Bard, MV Mirkin. J Am Chem Soc 119:10785, 1997. 165. RA Marcus, N Sutin. Biochim Biophys Acta 811:265, 1985. 166. MH Delville, M Tsionsky, AJ Bard. Langmuir 14:2774, 1998. 167. J Zhang, PR Unwin. J Electroanal Chem 494:47, 2000. 168. LJ Thibodeaux. Environmental Chemodynamics: Movement of Chemicals in Air,Water and Soil. 2nd ed. New York: Wiley, 1996. 169. RP Schwarzenbach, PM Gschwend, DM Imboden. Environmental Organic Chemistry. New York: Wiley, 1993. 170. GT Barnes,TI Quickenden, JE Saylor. J Colloid Interface Sci 33:236, 1970. 171. GT Barnes. J Colloid Interface Sci 65:566, 1978. 172. GT Barnes, DS Hunter. J Colloid Interface Sci 136:198, 1990. 173. J Zhang, PR Unwin. Langmuir. 18:1218, 2002. 174. I Langmuir, DB Langmuir. J Phys Chem 31:1719, 1927. 175. I Langmuir,VJ Schaefer. J Franklin Inst 235:119, 1943. 176. GT Barnes, VK La Mer. In: VK LaMer, ed. Retardation of Evaporation by Monolayers. New York: Academic, 1962, p 9. 177. M Blank, JS Britten. In: F Snell, J Wolken, G Iverson, J Lam, eds. Physical Principles of Biological Membranes. New York: Gordon & Breach, 1968. 178. M Blank. J Phys Chem 68:2793, 1964. 179. R Peters. Cell Biol Int Rep 5:733, 1981. 180. M Edidin. In: JB Finean, RH Michell, eds. Membrane Structure. New York: Elsevier, 1981, p 37. 181. WLC Vaz, ZI Derzko, KA Jacobson. Cell Surface Rev 8:83, 1982. 182. EL Elson. Annu Rev Phys Chem 36:379, 1985.

334

Barker et al.

183.

EL Elson. In: PD Weer, BM Salzberg. eds. Optical Methods in Cell Physiology. New York: Wiley, 1986, p 367. DH Charych, EM Landau, M Majda. J Am Chem Soc 113:3340, 1991. J Teissie¤, B Gabriel, M Prats. Trends Biochem Sci 18:243, 1993. J Teissie¤, M Prats, P Soucaille, F Tocanne. Proc Natl Acad Sci USA 82:3217, 1985. M Prats, J Teissie¤, JF Tocanne. Nature 322:756, 1986. B Gabriel, J. Teissie¤. J Am Chem Soc 113:8818, 1991. B Gabriel, J Teissie¤. Proc Natl Acad Sci USA 93:14521, 1996. VBP Leite, A Cavalli, ON Oliveria Jr. Phys Rev E 57:6835, 1998. M Gutman, E Nachliel. Biochim Biophys Acta 1231:123, 1995. E Nachliel, M Gutman. J Am Chem Soc 110:2629, 1988. J Zhang, PR Unwin. J Am Chem Soc 124:2379, 2002. BM Quinn, I Prieto, SK Haram, AJ Bard. J Phys Chem B 105:7474, 2001. J Zhang, RM Lahtinen, K Kontturi, PR Unwin, DJ Schi¡rin. Chem Commun 1818, 2001. HT Tien. Bilayer Lipid Membranes (BLM), Theory and Practise. New York: Marcel Dekker, 1974. P Mueller, DO Rudin, HT Tien,WC Westcott. J Phys Chem 67:534, 1963. M Montal, P Mueller. Proc Natl Acad Sci USA 69:3561, 1972. T Hanai, DA Haydon, J Taylor. Proc R Soc Lond 281:377, 1964. HT Tien, AL Diana. J Colloid Interface Sci 24:287, 1967. H Yamada,T Matsue, I Uchida. Biochem Biophys Res Commun180:1330,1991. CE Gardner, PR Unwin. MChem report, University of Warwick, 2000. H Yamada, H Shiku,T Matsue, I Uchida. J Phys Chem 97:9547, 1993. T Matsue, H Shiku, H Yamada, I Uchida. J Phys Chem 98:11001, 1994. M Tsionsky, J Zhou, S Amemiya, F-RF Fan, AJ Bard. Anal Chem 71:4300, 1999. E Overton.Vierteljahrsschr Naturforsch Ges Zuerich 44:88, 1899. G Menestrina, KP Voges, G Jung, G Boheim. J Membr Biol 93:111, 1986. P Pohl, SM Saparov,YN Antonenko. Biophys J 72:1711, 1997. TJ Pedley. J Fluid Mech 101:843, 1980 A Finkelstein. J Gen Physiol 68:127, 1976. P La«uger. Biochim Biophys Acta 455:493, 1976. A Finkelstein. Water Movement Through Lipid Bilayers, Pores and Plasma Membranes. New York: Wiley, 1987. R-AB Holland, H Shibata, P Scheid, J Piiper. Respir Physiol 59:71, 1985. GH Rothblat, FH Halberg, WJ Johnson, MC Phillips. J Lipid Res 33:1091, 1992. P Pohl, SM Saparov,YN Antonenko. Biophys J 75:1403, 1998. P Pohl, SM Saparov. Biophys J 78:2426, 2000. P Pohl, T Rokitskaya, EE Pohl, SM Saparov. Biochim Biophys Acta 1323:163, 1997. P Pohl, SM Saparov, EE Pohl, VY Evtodienko, II Agapov, AG Tonevitsky. Biophys J 75:2868, 1998.

184. 185. 186. 187. 188. 189. 190. 191. 192. 193. 194. 195. 196. 197. 198. 199. 200. 201. 202. 203. 204. 205. 206. 207. 208. 209. 210. 211. 212. 213. 214. 215. 216. 217. 218.

Characterization of Biomolecular Interfaces with SECM 219. 220. 221. 222. 223. 224. 225. 226. 227. 228. 229. 230. 231. 232. 233. 234. 235. 236. 237. 238. 239. 240. 241. 242. 243. 244. 245. 246. 247. 248. 249. 250. 251. 252. 253.

335

ER Scott, HS White, JB Phipps. Anal Chem 65:1537, 1993. ER Scott, HS White, JB Phipps. J Membr Sci 58:71, 1991. ER Scott, AI Lapalaza, HS White, JB Phipps. Pharm Res 10:1699, 1993. ER Scott, JB Phipps, HS White. J Invest Dermatol 104:142, 1995. BD Bath, RD Lee, HS White, ER Scott. Anal Chem 70:1047, 1998. BD Bath, HS White. Anal Chem 72:433, 2000. OD Uitto, HS White. Anal Chem 73:533, 2001. JV Macpherson, MA Beeston, PR Unwin, NP Hughes, D Littlewood. J Chem Soc FaradayTrans 91:1407, 1995. JV Macpherson, MA Beeston, PR Unwin, NP Hughes, D Littlewood. Langmuir 11:3959, 1995. PR Unwin, JV Macpherson, MA Beeston, NJ Evans, NP Hughes, D Littlewood. Adv Dental Res 11:548, 1997. S Nugues, G Denuault. J Electroanal Chem 408:125, 1996. T Yasukawa,T Kaya,T Matsue. Chem Lett 975, 1999. T Yasukawa, I Uchida,T Matsue. Biophys J 76:1129, 1999. A Hengstenberg, A Blo«chl, ID Dietzel, W Schuhmann. Angew Chem Int Ed 40:905, 2001. T Yasukawa,Y Kondo, I Uchida,T Matsue. Chem Lett 767, 1998. B Liu, A Rotenberg, MV Mirkin. Proc Natl Acad Sci USA 97:9855, 2000. H Shiku,T Shiraishi, H Ohya,T Matsue, H Abe, H Hoshi, M Kobayashi. Anal Chem 73:3751, 2001. CEM Berger, BR Horrocks, HK Datta. Electrochim Acta 44:2677, 1999. G Binnig, CF Quate, C Gerber. Phys Rev Lett 56:930, 1986. D Rugar, PK Hansma. PhysToday 43:23, 1990. R Wiesendanger. Scanning Probe Microscopy and Spectroscopy. Cambridge, UK: Cambridge University Press, 1994. JV Macpherson, PR Unwin. Anal Chem 72:276, 2000. JV Macpherson, PR Unwin. Anal Chem 73:550, 2001. JV Macpherson, J-P Gueneau de Mussy, J-L Delplancke. Electrochem Solid State Lett 4:E33, 2001. CE Jones, JV Macpherson, ZH Barber, RE Somekh, PR Unwin. Electrochem Commun 1:55, 1999. CE Jones, JV Macpherson, PR Unwin. J Phys Chem B 104:2351, 2000. C Kranz, G Friedbacher, B Mizaiko¡, A Lugstein, J Smoliner, E Bertagnolli. Anal Chem 73:2491, 2001. M Ludwig, C Kranz,W Schuhmann, HE Gaub. Rev Sci Instrum 66:2857, 1995. A Hengstenberg, C Kranz,W Schuhmann. Chem Eur J 6:1547, 2000. PI James, LF Gar¢as-Mesias, PJ Moyer, WH Smyrl. J Electrochem Soc 145:L64, 1998. M Bu«chler, SC Kelley,WH Smyrl. Electrochem Solid State Lett 3:35, 2000. H Shiku, RC Dunn. Anal Chem 71:23A, 1999. Y Lee, S Amemiya, AJ Bard. Anal Chem 73:2261, 2001. G Shi, LF Gar¢as-Mesias,WH Smyrl. J Electrochem Soc 145:3011, 1998. P James, N Casillas,WH Smyrl. J Electrochem Soc 143:3853, 1996.

7 Layered Protein Films: Quartz Crystal Resonator Frequency and Admittance Analysis Ernesto J. Calvo, Claudia Danilowicz*, Erica Forzani, Alejandro Wolosiuk, and Marcelo Otero University of Buenos Aires, Buenos Aires, Argentina

1

INTRODUCTION

In this chapter we describe recent advances in the use of quartz crystal microbalance (QCM) frequency shift and quartz crystal admittance= impedance analysis of self-assembled protein ¢lms complemented by ellipsometric thickness, and comparison of acoustic and optical mass=thickness. These protein ¢lms can be regarded as model systems among other applications of electrochemical biosensors based on molecular recognition and electrical signal generation through electrocatalysis of ‘‘wired’’ (electrically connected) enzyme electrodes. Organized multilayers formed by stepwise alternated electrostatic adsorption of anionic and cationic polyelectrolytes provide a simple way to fabricate ultrathin functional ¢lms on solid surfaces with nanometer resolution [1]. Self-assembled protein ¢lms on electrode surfaces have the advantage over hydrogels of the same components, obtained either by random electropolymerization or sol-gel transformation, of better control of the molecular orientation and organization in the nanoscale.

*Present a⁄liation: Harvard University, Cambridge, Massachusetts, U.S.A.

337

338

Calvo et al.

The method of layer-by-layer (LBL) electrostatic adsorption between a charged surface and oppositely charged molecules in solution with surface charge reversal brings the possibility of: 1. 2.

Adsorption regulation and restriction to monolayer by repulsion of soluble molecules of equal charge Adsorption of oppositely charged soluble molecules on top of the surface layer

In order to characterize the resulting organized multilayer ¢lms, it is very convenient to have at hand in situ methods to determine the ¢lm structure and dynamics. Among these methods, acoustic and optical techniques have been extensively used in recent years. The quartz crystal microbalance based on the thickness shear-mode wave resonator provides gravimetric information and reveals volume and viscoelastic changes. Optical techniques such as ellipsometry, surface plasmon resonance (SPR), and scanning angle re£ectometry (SAR) complement acoustic methods, and ¢lm thickness and refractive index values can be determined. Both acoustic and optical techniques result in di¡erent values of protein mass or thickness since they sense in di¡erent ways the interactions among biomolecules, polymers, and water. 2

SELF-ASSEMBLED POLYELECTROLYTE AND PROTEIN FILMS

The buildup of organized protein ¢lms on surfaces represents one of the major growing ¢elds in the boundaries between materials science, biology, biomedicine, and microelectronics. Among other applications, protein ¢lms have been used in biosensors, medical diagnostics, bioseparations, and catalysis [2^6], biocompatibility to produce surfaces that prevent adsorption of proteins [7], to induce time-delayed speci¢c responses at the cellular level, to modify the surface of human repair devices, or to be in contact with biological mediums (catheters, protheses, storage vessels, etc.). Proteins immobilized on solid surfaces can be accomplished by different techniques such as physical adsorption, entrapment, chemical binding, and electropolymerization [2]. Among the problems to overcome in these surface protein ¢lms are inhomogeneity, immobilized protein denaturing, and restricted ¢lm permeability to substrates. Ordered protein multilayer ¢lms with a high protein density have been constructed by using: (1) Langmuir-Blodgett deposition [2,8]; (2) biospeci¢c interactions [9^12] such as avidin-biotinylated protein and

Layered Protein Films

339

antigen-antibody; and (3) electrostatic LBL deposition of proteins and polyelectrolytes of di¡erent electrical charge [13]. The alternating adsorption of charged macromolecules or LBL adsorption has recently emerged as a promising simple method to fabricate controlled and highly ordered molecular assemblies in a predesigned architecture [14,15]. This strategy is based on the stepwise adsorption of charged species onto a charged substrate, primarily by electrostatic interactions with charge overcompensation in each deposition cycle, thereby inducing the adsorption of the next layer of oppositely charged macromolecules. The LBL strategy was ¢rst introduced almost a decade ago for the creation of pure polyelectrolyte multilayer ¢lms on macroscopically £at surfaces [14,15]. A variety of multicomponent ¢lms of inorganic particles [16^19], dye molecules [20^24], and water-soluble proteins [11,24^29] alternating with oppositely charged polyelectrolytes have been assembled by the LBL electrostatic self-assembly (ESA) technique. The most extensively studied system has been the poly(styrenesulfonate) anion (PSS) and poly(allylamine) cation (PAH) ESA multilayer, which can be taken as a model ¢lm. Decher and coworkers [30] have recently studied the in situ buildup of a PSS=PAH multilayer by means of streaming potential measurements (SPM) and by scanning angle re£ectometry (SAR). The results were described in the framework of a schematic representation of the multilayer in three zones: a precursor zone (I), a core zone (II), and an outer zone (III). This view seems to be supported by their experimental ¢ndings: the zeta potential of the multilayer determined by SPM shows a symmetrical and constant charge inversion during the multilayer buildup, which seems to indicate an exact charge compensation in zone II and an excess charge located entirely in the outer zone III. It was also shown by SAR that a regular buildup regime, in which the thickness increment per layer is constant, is reached after the deposition of the ¢rst six polyelectrolyte layers,which gives an indication of the extent of zone I. Using neutron re£ectometry, Decher and coworkers [31] were able to resolve the internal structure of self-assembled polyelectrolyte multilayer ¢lms to high resolution. According to the molecular picture that results from the interpretation of such studies by analyzing the data with a compositionspace re¢nement technique, the ESA ¢lms consist of strati¢ed structures with interdigitated layers of polyanions and polycations. For alternating layers of PSS and PAH adsorbed onto atomically £at surfaces, a roughening of successively deposited layers leads to a progressively larger number of adsorption sites for consecutive generations of adsorbed polymer. This leads to an increase in layer thickness with an increasing number of deposited layers, but because of the interpenetration of adjacent polyelectrolyte species this increase settles quickly into an

340

Calvo et al.

equilibrium thickness (zone II). In fully hydrated ¢lms (100% relative humidity), water occupies 40% of the volume within the ¢lms and about twice as much water (by volume) is associated with PSS rather than with PAH. The equilibrium thickness of the PSS-PAH multilayer structure depends strongly on the external electrolyte solution ionic strength. For weak polyelectrolytes such as poly-(allylamine) the solution pH regulates the polymer surface charge and therefore the ¢lm thickness [32]. There are examples of protein self-assembled ¢lms [11,26,27,33^35] such as immunoglobulin G, anti-immunoglobulin G, cytochrome C, lysozyme, myoglobin, hemoglobin, peroxidase, and glucose oxidase. It is generally accepted that when biomolecules such as enzymes or antibodies are inserted between di¡erent polyelectrolyte layers, the biological activity results are additive. In some examples two oppositely charged proteins separated by at least one polyelectrolyte layer pair were inserted in the ¢lm [27,33], while in the case of antibodies inserted in the polyelectrolyte ¢lm, it was shown that the embedded proteins kept their reactivity with respect to their antigenic reaction when no more than four polyelectrolyte layers were deposited over them [26]. Decher and coworkers [36,37] studied the adsorption of negatively charged human serum albumin (HSA) onto a positively charged poly(ethylenimine) (PEI)-(PSS-PAH)3 ¢lm or a negatively charged PEI-(PSSPAH)3 -PSS ¢lm. They observed that HSA adsorbs on both types of surfaces and that on PSS-terminating ¢lms which are similarly charged as albumin, only monolayer adsorption of HSA was found whereas on the oppositely charged PAH-terminated ¢lms the adsorbed protein layers extended over thicknesses larger than four times the largest dimension of the HSA molecule. They also demonstrated that the intermolecular interactions involved in these processes are mostly of electrostatic origin. Di¡erent strategies have been used to design modi¢ed surfaces for immunosensing. Particularly, immobilization of antibody or antigen on transducer has proved to be e¡ective in several applications [38^43]. The design of immunosensor devices is therefore possible through the combination of the immunochemical reaction and a transducer which transforms the recognition event into an electronic signal. Such devices have been of great interest in the last decades due to the potential applications in environmental, food, and drug monitoring, as well as in clinical diagnostics, mainly point-of-care testing. A simple way of achieving biospeci¢c activity is through adsorption on clean surfaces [44,45] and polymers [36,46,47]. In either case the modi¢ed surface is stable and allows e⁄cient antigen ^ antibody binding. One

Layered Protein Films

341

attractive approach has involved self-assembled layers for protein deposition on solid surfaces. A simple procedure requires the initial adsorption of thiotic acid and further covalent immobilization of antibody molecules onto a microporous gold membrane [48]. Similar self-assembled layers of sulfur containing molecules have also been used for immunosensor design [49,50]. The direct assembling of an antigen monolayer is also possible and has been applied to antibody detection in a reusable sensor [51]. More precise control of these surfaces can possibly be attained through LBL deposition of charged molecules from aqueous solutions. It has been proposed that the electrostatically driven assembly of multilayered structures may lead to the fabrication of multicomponent nano¢lms [15]. This strategy has proved to be e¡ective for enzyme electrode design [52,53]. Furthermore, precise control over the distance of the active layers can be achieved with directional electron transfer through the layers [54]. The use of alternating polyelectrolyte ¢lms has been reported for designing immunosensing surfaces. An increased response was obtained through deposition of several antibody layers [46] and therefore the sensitivity was controlled by the number of protein layers. A detailed characterization of polyelectrolyte-protein (IgG) multilayer was achieved employing several techniques [36]. It was observed that either layered or disordered ¢lms were obtained depending on the number of polymer layers between the protein layers. For the ¢lms assembled with one polymer layer separating each protein layer, a disordered structure was found.On the other hand, ¢lms including ¢ve polyelectrolyte layers between each IgG layer showed that a dense layered structure was formed. The di¡erent biosensing applications of these structures such as the recognition event involved in antigen ^ antibody binding were also evaluated. One major concern relates to molecule orientation. The immobilization of biomolecules onto surfaces by simple adsorption from solution results in random orientation on the surface, leading to ine⁄cient binding of target molecules. Several publications have focused on the possibility of improving orientation during the deposition of antibodies onto solid supports [55^59]. The orientational aspects of immobilization and immunological activity were studied for several strategies such as simple adsorption, antibody thiolation, and protein A binding [56]. It was observed that protein A coupling improves antibody binding. However, it was not possible to separate the contributions from nonspeci¢c binding for an unambiguous conclusion. The use of covalent coupling of Fab0 fragments to liposomes via linker lipids demonstrates that it is possible to control the orientation of antibodies and therefore increase the overall binding e⁄ciency. In this work it is shown that nonspeci¢c binding depends critically on the monolayer matrix [57].

342

Calvo et al.

Studies of site-oriented deposition of antibodies showed that the oxidation of carbohydrate residues also allowed a controlled deposition upon covalent coupling [58]. A di¡erent strategy for oriented immobilization involves using native sul¢de groups of immunoglobulins in order to attach them to gold surfaces. The immobilized antibodies exhibit high antigen-binding constants and no distribution of the a⁄nity constant, proving site-oriented adsorption on the surfaces [59]. The adsorption of proteins onto polyelectrolyte assembled ¢lms was investigated at di¡erent pH values. It was found that proteins interact with these polymers independently of the charge of the multilayer and the protein. However, when the charges of protein and polyelectrolyte are opposite, larger amounts of absorbed protein were observed. Piezoelectric transducers coupled to biosensors are sensitive to mass changes since their response is in£uenced by interfacial phenomena, volume, and viscoelastic changes of surface bio¢lms [60]. However, they can monitor in real time protein adsorption on di¡erent solid substrates, providing information on the amount of adsorbed material as well as on the adsorption kinetics [61^63]. It should be stressed that the applicability of this technique has been argued, considering the e¡ect of trapped liquid within the protein molecules and additional viscoelastic e¡ects [64]. A combination of the conventional QCM and dissipation measurements has proved to be successful in obtaining information about elastic and inelastic components of the shear-wave propagation through an adsorbed viscoelastic ¢lm [65]. 3

SELF-ASSEMBLED ENZYME FILMS

Electrically ‘‘wired’’ enzymes have attracted attention recently because of their potential applications in biosensors and molecular devices. Molecular recognition with enzymes and with enzyme-labeled immuno and genomic electrodes based respectively on the antigen ^ antibody interaction and single-stranded DNA (ss-DNA) hybridization with self-immobilized redox relays to generate an electrical signal can be integrated in circuits [66]. In large proteins such as glucose oxidase (GOx) (186,000 g=mol), direct electron transfer from the electrode surface to the prosthetic group, FADH2, buried inside the protein structure, is hindered. Heller and coworkers demonstrated that electrical communication between the FADH2 in glucose oxidase and electrodes can be facilitated by electrostatically complexing the negatively charged enzyme in a solution of pH above the isoelectric point (4.2) with a cationic quaternized poly(vinylpyridine) and poly(vinylpyridine) Os(bpy)2Cl redox mediator polyelectrolyte copolymer

Layered Protein Films

343

[67]. With this strategy in mind, Heller proceeded to introduce a twocomponent epoxy technique combining GOx and other oxidases with the polycationic redox mediator cross-linked with a bifunctional reagent [68^71]. In the hydrogel there is a random distribution of components with no control of the molecular orientation; furthermore, at high-proteinconcentration segregation of hydrophobic phases may occur. Spatially ordered enzyme assemblies o¡er several advantages over random polymers with the same active components [52,72,73] for understanding the mechanisms involved in molecular recognition by the enzyme, redox mediation by the redox polymer, and signal generation. Among these advantages, Calvo and coworkers have found that the enzyme concentration can be quanti¢ed during the deposition process, and the redox mediator charge can be obtained by integration of voltammetric surface waves in the absence of substrate [52,54]. In addition, well-de¢ned spatial distribution of enzyme with respect to the mediator can be achieved and substrate mass transport limitations can be minimized in nanometer-scale ultrathin ¢lms. In organized multilayer systems with more than one enzyme, precise control of distances of the active layers can be achieved with directional electron transfer to the sequential enzyme reactions. Chemically modi¢ed electrodes [74] have evolved in recent years into integrated chemical systems [75^77] with high degree of organization of the di¡erent components in supramolecular architectures. The LBL thin-¢lm fabrication technique introduced by Decher and co-workers [14a,15,78^80] to produce complex multilayer thin ¢lms with molecular-level thickness control has been extended to enzyme-organized multilayers. Enzymes deposited in ordered monolayers and multilayer systems have been described using di¡erent assembling techniques for enzyme immobilization, such as Langmuir-Blodgett [81], self-assembled monolayers [82^84], step-by-step electrostatic adsorption of alternate multilayers [2,52,27,29], antigen ^ antibody interaction [85^89], avidin ^ biotin interaction [73,90,91], surfactants ¢lms [92,93], electrostatic adsorption of hyperbranched polyelectrolytes [94], etc. Step-by-step deposition of glucose oxidase monolayers on carbon surfaces was achieved by the antigen ^ antibody binding strategy [86]. Three kinetic barriers in redox mediation to glucose oxidase are proposed: di¡usion to the enzyme surface, positioning with respect to the redox site FADH2, and electron transfer FADH2 to the redox molecule (k). Savea¤nt and co-workers have concluded that since the di¡usion of the mediator towards the enzyme is kD  5.108 M1=s, the positioning of the mediator with respect to the prosthetic site (FADH2) is the rate-determining step [95]. A remarkably open structure was obtained after multilayer deposition of several antigen ^ antibody layers for enzyme conjugate immobilization.

344

Calvo et al.

Furthermore, the resulting ¢lm was fully active upon substrate addition [87]. The characterization of these assembled structures containing active and inactivated enzyme layers also showed the low compactness of the enzyme ¢lms. In this case the ¢lm was enlarged, and therefore it was possible to evaluate the competition between the enzymatic reaction and mass transport within the multilayer structure [88].The glucose oxidase conjugate was also applied to the study of biorecognition dynamics in antibody ^ antigen binding [96]. The design of spatially ordered assemblies was also possible using the avidin ^ biotin binding reaction [73]. A monolayer assembled ¢lm using this interaction allowed the simultaneous immobilization of glucose oxidase and an electron mediator. In this work, glucose oxidase-conjugated avidin was initially attached to covalently bound biotin and further interaction of a long-chain biotinylated ferrocene to the remaining vacant sites determined mediator co-immobilization. The catalytic currents were at least 20 times larger when soluble mediator was used compared to a self-contained monolayer of ferrocene and GOx. For the long-chain biotinylated ferrocene in solution, a decrease in k value by a factor of 17 was reported and attributed to its larger viscosity with respect to ferrocene methanol. Furthermore, a decrease by a factor of 60 in k value was found in the GOx monolayer with coimmobilized ferrocene. In this work the authors assumed that the same concentration of GOx that could be oxidized by the ferrocene methanol was oxidized by the immobilized long-chain ferrocene. The dynamics of the biotin ^ avidin multilayer ¢lm was evaluated in terms of catalytic responses in order to prove the spatial order and to estimate the average distance between layers [90]. Avidin ^ biotin technology has also proved to be useful for a step-by-step construction of bienzyme electrodes [97]. Calvo and coworkers [52] have described the redox mediation of glucose oxidase in self-assembled structures of cationic poly(allylamine) modi¢ed by ferrocene and anionic GOx deposited stepwise in alternate polymer=enzyme multilayers. Figure 1 depicts schematically an organized multilayer composed of a redox polymer and GOx on a £at gold surface.The redox charge and the amount of enzyme increase in step with the number of multilayers deposited. However, it has been found that only a small fraction of the active assembled GOx molecules are ‘‘electrically wired’’ by the ferrocene polymer in the integrated chemical structure. The behavior of redox enzymes (GOx, LOx, and SBP) in multilayers self-assembled with PAH-Os by electrostatic adsorption through sequential immersion in the respective redox polycation and enzyme solutions of pH above the isoelectric point, where the enzymes carry a net negative charge, were investigated [54]. The method allows the construction of well-ordered

Layered Protein Films

345

FIG. 1 Schematic representation of LBL self-assembled enzyme in polyelectrolyte multilayer on a surface.

systems and quantitative determination of the enzyme electrical ‘‘wiring’’ e⁄ciency, kinetic coe⁄cients of the reactions involved, and the e¡ect of the number of adsorbed layers. Calvo and coworkers reported combined electrochemistry, electrochemicalquartz crystal microbalance (EQCM),and atomic force microscopy (AFM) studies of layer-by-layer deposited PAH-Os redox-active polymer and GOx in glucose-free solutions and the enzyme electrocatalysis mediated by an osmium redox polymer in b-D -glucose-containing solutions [98]. 4

THICKNESS SHEAR-MODE RESONATORS

The quartz crystal microbalance (QCM), the most extensively studied shearmode AT-cut quartz resonator, is formed by a thin slice of quartz single crystal with two metal electrodes deposited on both faces of the crystal. The device is therefore an electromechanical transducer in which the excitation electrodes generate a transverse standing shear wave across the thickness of the crystal, which propagates into the deposited ¢lm immobilized onto the crystal surface [99^101]. For rigidly coupled ¢lms moving synchronously with the quartz crystal, the oscillating resonant frequency of the piezoelectric device decreases linearly with the mass loading [102].When the ¢lm is nonrigidly coupled to the quartz crystal, on the other hand, the response

346

Calvo et al.

also depends on the thickness and viscoelastic properties of the surface layer [103], due to a mismatch in the sound velocity in the quartz and surface ¢lm and attenuation of the shear wave. Sauerbrey [102] in 1959 related the change in resonance frequency of a piezoelectric quartz crystal with the mass deposited onto or removed from the crystal surface. This approach has been used extensively in microgravimetric measurements in the gas phase, such as metal evaporation. For (Df << fs ) the Sauerbrey equation states that Dfs ¼

2f 2s Dm pffiffiffiffiffiffiffiffiffiffiffi  mQ rQ A

ð1Þ

with Df the measured frequency shift, fS the fundamental resonant frequency of the quartz crystal, Dm the mass loading, A the piezoelectrically active area, rQ the quartz density (2.648 g=cm3), and mQ the shear modulus of AT-cut quartz (2.947 1011 dyn=cm2). The sensitivity factor increases at high resonant frequency of the resonator, which corresponds to a thinner quartz crystal; for example, a 10-MHz AT-cut quartz crystal of 0.017cm thickness has a nominal sensitivity of 0.226 Hz cm2=ng. The Sauerbrey equation assumes a rigid ¢lm with density and transverse velocity of the acoustic wave identical to those of the quartz crystal (equal acoustic impedance of the quartz and the ¢lm overlayer). It also assumes that the deposit is uniform while the sensitivity of the QCM is nonuniform across the radial direction of the resonant quartz crystal, with maximum sensitivity at the crystal center [104]. Piezoelectric acoustic wave devices also respond to small changes in mass at surfaces immersed in viscous liquids [105],where the shear coupling of the standing wave and the propagating acoustic wave in the liquid result in a velocity shift attenuation of the thickness shear-mode device. The resonance frequency of AT-cut quartz resonators immersed in liquids depends on the bulk properties of the liquid (density, viscosity, and conductivity) and the interfacial properties of the sensor ^ liquid interface (wettability, surface roughness, shear modulus, etc.). In addition to mass changes at the quartz crystal surface and density and viscosity of the liquid, the resonant frequency can also be a¡ected by several factors such as liquid conductivity [106], the hydrostatic pressure di¡erence across both crystal surfaces and lateral stress at the overlayer deposited onto the quartz crystal [107], surface roughness [108], longitudinal waves in the liquid [109], and viscoelastic properties and volume change of ¢lms immobilized at the quartz crystal [110]. QCM has extraordinary sensitivity and hence permits the kinetic examination of processes that involve events at the monolayer level and thin ¢lms, which would be otherwise silent to electrochemical methods, i.e., £uxes of solvent, neutral molecules, or ion pairs. Bruckenstein weighed a

Layered Protein Films

347

single adsorbed oxygen monolayer (20 ng) on a gold surface during simultaneous recording of the current ^ voltage curve [111]. Partial electron transfer, i.e., electrosorption valency, during electrosorption phenomena could be detected by simultaneous measurement of the electrical charge in the electrochemical experiments and frequency change in the electrochemical quartz crystal microbalance during the adsorption of ionic species [112]. The electrical properties of the unloaded quartz crystal can be described by an equivalent circuit as depicted in Fig. 2, known as a ButterworthVan Dyke (BVD) circuit [99^101]. This BVD equivalent circuit is formed by a motional series RLC circuit in parallel with a static capacitance C0. The electrical equivalence to the mechanical model (mass, elastic response, and friction losses of the quartz crystal) are the series inductance L, capacitance C, and a resistance R. The static capacitance in parallel with the series motional RLC arm, C0 represents the electrical capacitance of the parallelplate capacitor formed by both metal electrodes that sandwich the thin

FIG. 2 Equivalent circuit models of the quartz crystal resonator: (A) general equivalent circuit representation of unloaded quartz resonator, BVD circuit; (B) lumped-element model, LEM valid near resonance with surface impedance load, Zs.

348

Calvo et al.

quartz crystal (C00 ), the stray capacitance due to the connectors,Cp. In addition, when metal-covered quartz crystals are in contact with an aqueous electrolytic solution, the double layer, Cdl, a¡ects the value of C0. This static 1 Þ is not related to the piezoelectric capacitance C0 ðC01 ¼ C001 þ Cp1 þ Cdl e¡ect re£ected in the motional arm series RLC, but in£uences the resonant frequency f p, as will be shown below. In the series branch of the equivalent electrical circuit, L is proportional to the quartz inertial mass displaced by the shear oscillation, C is the energy stored in the quartz crystal during oscillation, and R is the energy frictional loses of the quartz crystal. AT-cut quartz crystals immersed in liquids are also a¡ected by the liquid density, viscosity, and conductivity. The ¢rst adds to the mass of the resonator and the viscosity contributes to the shear wave damping in the liquid. For a Newtonian liquid, additional inductance L1 and resistance R1 are needed in the BVD circuit in series with the motional arm in Fig. 2. Some typical values for the electrical equivalent elements of a 10 MHz AT-cut quartz crystal are C ¼ 10 fF, R ¼ 100 O, L ¼ 8.4 mH, C0 ¼ 1^10 pF, Ll ¼ 3.5 mH, and R l ¼ 214 O for water. Figure 3 depicts a typical admittance parametric plot (suceptance Y 00 versus conductance Y 0 ) for a quartz crystal resonator. Note that the static capacitance C0 in the parallel branch produces a vertical shift of the circle by oC0 and gives rise to a new resonance frequency op which depends on C0 , in addition to the series resonance frequency os ¼ 2pf0. The circle diameter is

FIG. 3

Admittance polar plot.

Layered Protein Films

349

inversely proportional to the motional resistance which measures the shearwave damping in the liquid. Mason [113] ¢rst realized that the viscoelastic properties of a £uid in contact with quartz crystals can a¡ect the resonant properties. However, Mason’s work was ignored for a long time and there were no studies of piezoelectric acoustic wave devices in contact with liquids until Nomura and Okuhara [114] found an empirical expression that described the changes in the quartz resonant frequency as a function of the liquid density, viscosity, and conductivity in which the crystal was immersed. Shortly after that, the empirical observations of Nomura were described in terms of physical models by Kanazawa [110] and Bruckenstein [115]. These authors derived the equation that describes the changes in resonant frequency of a lossless quartz crystal in contact with an in¢nite, nonconductive, and perfect Newtonian £uid. Both treatments predict that the frequency shift is proportional to the square root of the density ^ viscosity product, in agreement with the empirical approach of Nomura and Okuhara [114]: fs 3=2 pffiffiffiffiffiffiffiffiffi Dfl ¼  pffiffiffiffiffiffiffiffiffiffiffiffiffiffi rl Zl prQ mQ

ð2Þ

In many cases the frequency shift produced by the liquid adds to the frequency shift due to a change in areal mass at the crystal surface. In those cases it has been possible to follow adsorption, electrodeposition, and stripping of material to or from the quartz crystal surface, allowing microgravimetry in contact with liquids to be done with the quartz oscillator as the working electrode in electrochemical experiments and biomolecule adsorption. More recently, the treatment was extended to piezoelectric devices in contact with viscoelastic media (i.e., liquids and polymers). It was then realized that if the deposited mass was not rigidly coupled to the oscillating quartz crystal, separation of inertial mass and energy loss measurements was not possible with the resonant frequency alone. Quartz crystal impedance in the acoustic frequency range was introduced in order to study mass and viscoelastic changes, and a full electrical characterization of the crystal behavior near resonance was employed. This has been accomplished by measurements with impedance analyzers or network analyzers operating in the megahertz frequency range [60]. Changes in the resonance frequency are related to changes in the equivalent inductance L and broadening of the admittance curve near resonance (decrease in the circle diameter 1=R in Fig. 3) are related to equivalent resistance R.This can be seen in Fig. 4,which depicts real (Y 0 ) and imaginary (Y 00 ) components of the electrical admittance versus oscillation

350

Calvo et al.

FIG. 4 Real and imaginary components of the admittance spectra around the series resonant frequency versus frequency for a 10-MHz TSM resonator crystal coated with (PAH-Os)7(GOx)7 self-assembled film on a quartz crystal electrode Au-coated and thiolated with MPS: (A) in air; (B) under water. (Reprinted with permission from Ref. 178. Copyright 2002 American Chemical Society.)

frequency for a 10-MHz-thickness shear-mode resonator crystal coated with Au, and modi¢ed with mercaptopropane sulfonate (MPS) and a selfassembled (PAH-Os)7(GOx)7 ¢lm multilayer in air (A) and in contact with water (B). The resonant admittance for the ¢lm in contact with the liquid decreases and the admittance curve broadens, while the resonant frequency shifts to lower values. The complex surface admittance of the composite resonator (ZS ) is determined by the surface thin ¢lm and the viscous liquid, since the shear wave penetrates the ¢lm (¢lm thickness, df ¼ 217 nm) and is further attenuated in the viscous liquid. A further shift of the resonant frequency, fs , toward lower values is apparent in successive adsorbed layers due to mass increase and both peak height and peak width change during the stepwise enzyme adsorption,

Layered Protein Films

351

re£ecting energy storage and dissipation during the shear-wave perturbation by the thin ¢lm. Buttry [116] measured the admittance around resonance of a quartz crystal coated with polynitrostyrene and related those measurements to the reological changes due to ¢lm swelling. Muramatsu et al. [117] used the resonant resistance in addition to the resonant frequency of EQCM as a criterion to evaluate the ¢lm nonrigidity for several electroactive polymer systems, including poly-pyrrole, Na¢on, Langmuir-Blodgett ¢lms, and other deposits. Muramatsu [117] et al. monitored the reological changes of polypyrrole during electrodeposition and cyclic voltammetry in liquid. Since the transverse shear wave penetrates the damping surface layer and the viscous liquid, additivity of the equivalent electrical elements in the BVD circuit is valid only under certain particular conditions. Martin and Frye studied the impedance near resonance of polymer ¢lm-coated resonators in air with a lumped-element BVD model, modi¢ed to account for the viscoelastic properties of the ¢lm [118,119]. Extensive literature covers the use of quartz acoustic impedance techniques to describe the volume and viscoelastic changes in polymer ¢lms, both in air or in contact with a liquid, electrolytes, and undergoing simultaneously electrochemical transformations [119^139]. 4.1

QCM Complex Admittance=Impedance Analysis

The general electrical representation of a piezoelectric quartz resonator is given by the transmission line model (TLM) of Mason [140], in which complex input impedance is given by " # 1 K 2 2 tanðf=2Þ  jðZS =Zq Þ ð3Þ Z¼ 1 joC0 fq 1  jðZS =Zq Þ cotðfq Þ where K2 is the quarts electromechanical coupling coe⁄cient, f is the complex acoustic-wave phase shift across the quartz,C0 is the static capacitance of the resonator, o ¼ 2pf with f the excitation qffiffiffiffiffiffiffiffiffiffiffiffiffi frequency, where Zq is the quartz characteristic impedance Zq ¼ ðrq mq Þ, rq is the quartz density and mq the quartz sti¡ness, ZS is the surface mechanical impedance due to the surface ¢lm and viscous liquid in contact with it. The motional impedance arising from the mechanical motion of the quartz in parallel with C0 [123] is " #   fq fq ðZS =Zq Þ jðZS =Zq Þ 1 1 1 þ 1 Zm ¼ joC0 2K 2 tanðfq =2Þ 4K 2 oC0 2 tanðf=2Þ ð4Þ

352

Calvo et al.

The ¢rst term in Eq. (4) represents the unloaded quartz crystal impedance and the second term the lumped element surface (LEM) impedance ZLEM, given by " #1 fq ðZS =Zq Þ jðZS =Zq Þ ZLEM ¼ 1 ð5Þ 2 tanðfq =2Þ 4K 2 oC0 Furthermore, near resonance, o ’ os , where os ¼ 2pfs , with fs the series resonance frequency and N the harmonic resonance [141] and if jZS j= Zq << 2 tanðfq =2Þ, then   Np ZS ZLEM ’ ð6Þ ¼ RS þ jXLS 4K 2 oS C0 Zq The electroacoustic impedance of a lumped-element BVD resonator comprised of a viscoelastic ¢lm on a quartz crystal immersed in liquid depends on the liquid density (rl ) and viscosity (Zl ) and the ¢lm thickness (df ), density (rf ), and shear (storage G0 and loss G00 ) moduli. The validity of the LEM equivalent circuit to within 1% of the transmission line model [142] is ful¢lled, since the ratio of the surface ¢lm and=or liquid impedance ðZS Þ to the quartz crystal impedance (ZQ ) is ZS =ZQ < 0:005 for protein ¢lms in contact with water. If two nonpiezoelectric layers (¢lm f, and liquid, l ) are successively attached to the crystal, the following expression describes the surface impedance [142]: " # Zf tanhðkf df Þ þ Zl tanhðkl dl Þ ZS ¼ ð7Þ 1 þ ðZl =Zf Þ tanhðkf df Þ tanhðkl dl Þ where the subscript f denotes the viscoelastic ¢lm underlayer and l the liquid overlayer. The characteristic ¢lm and liquid impedances are, respectively, Zf and Zl ; and kf is the ¢lm complex wave propagation constant, kf ¼ joðrf =Gf Þ1=2 . The complex shear modulus is G ¼ G0 þ G00 and the characteristic impedance of the Newtonian liquid is pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð8Þ Zl ¼ r1 Z1 oj ¼ ð1 þ jÞ r1 Z1 o=2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  and Rl ¼ XLl ; the characteristic ¢lm impedance is Zf ¼ ðGf rf Þ. If the liquid layer is much thicker than the acoustic wave penetration, then tanhðkl dl Þ ! 1, and Eq. (6) becomes ZLEM ¼ RS þ jXLS

where LQ ¼

N p2 8K 2 o2S C0

  Zf tanhðkf df Þ þ Zl 2oLQ ¼ pffiffiffiffiffiffiffiffiffiffiffi p mQ rQ 1 þ ðZl =Zf Þ tanhðkf df Þ

ð9Þ

Layered Protein Films

353

Inspection of Eq. (9) shows that one can access experimentally two quantities of the surface impedance i.e., RS and XLS , but three ¢lm material properties, namely, rf , Gf0 , and Gf00 , and df need to be evaluated [139,141,143]. Several approximations have been made on at least two parameters, normally thickness and density or loss tangent, tan a ¼ G00 =G0 [137] or constant Faradaic e⁄ciency and constant density during electrodeposition, and the electrical charge has been taken as a measure of the ¢lm thickness [144]. Calvo et al. [139] introduced ¢lm impedance polar parametric plots of Rf versus XLf in order to characterize the viscoelastic behavior of the nonpiezoelectric ¢lm layers on quartz crystal resonators. Figure 5 depicts one such polar parametric plot calculated with Eq. (9) for a thin Newtonian £uid of variable viscosity for three di¡erent thicknesses assuming constant density and zero G 0 and in the absence of liquid. At low ¢lm viscosity (G00 ! 0) the penetration depth of the shear wave, l [where l ¼ Zf =ð2prf oÞ1=2 ] is much less than the ¢lm thickness, df , and semi-in¢nite Newtonian £uid behavior is apparent as described by the Kanazawa-Bruckenstein equation (9) with Rl ¼ XLl and a slope of unity in Fig. 5. Under these conditions no mass information can be derived from the acoustic impedance. As the ¢lm viscosity increases (Zf increases in the clockwise direction in Fig. 5), at constant density the semicircles in Fig. 5 describe the combined e¡ects of viscosity and mass per unit area (rf df ). Note that the slope of the parametric plot changes sign at the plot apex and ¢nally reaches the rigid mass condition (R ¼ 0) at very high viscosity [validity of Sauerbrey, Eq. (1)].

FIG. 5 Parametric film impedance plane plots calculated with Eq. (9) for the variable G00 using 103 jG=N=m2 j 109 with G0 = 0, df = 0.8, 1.0, and 1:2 mm, and rf = 1 g=cm3 in the absence of liquid phase. (From Ref. 139).

354

Calvo et al.

Other formalisms, equivalent to the one described above, have been reported.While the impedance depends on the quartz crystal resonator area, Johannsmann [135,136] introduced an equivalent formalism based on the series resonant frequency shift Df and the impedance line width G, both in hertz and independent of the electrode area. The complex resonant frequency f  ¼ f þ DG, with 2G the bandwidth. The complex frequency shift is then ð10Þ Df  ¼ Df þ i DG and pffiffiffiffiffiffiffiffiffiffi Df ifs ZL fs ¼ ¼  pffiffiffiffiffiffiffiffiffiffi ðrGÞ tan f p ZQ p m q rq

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi " s # r jGod

ð11Þ

Kasemo et al. [145] developed an experimental setup, QCM-D, for simultaneous measurement of the frequency, the absolute quality factor Q, and the amplitude of oscillation of a quartz crystal microbalance covered by a surface ¢lm ( f ) and immersed in a Newtonian liquid (l).The authors de¢ned an energy dissipation factor D ¼ 1=Q and applied this to the study of several biomolecular ¢lms on quartz crystal [146]. The measured quantities, resonance frequency shift Df and dissipation factor shift DD, were measured simultaneously with the adsorption of biomolecular ¢lms. ImðbÞ ReðbÞ Df ¼ and DD ¼  ð12Þ 2pdq rq pdq rq where jGf 1  a expð2kf df Þ b ¼ kf o 1 þ a expð2kf df Þ

and

 jkf Gf =kl Gl  1  a¼ jkf Gf =k1 Gl þ 1 ð13Þ

with the respective complex wave propagation constants for the ¢lm and liquid, respectively (Gl ¼ oZl and Gf ¼ Gf0 þ jGf00 ), sffiffiffiffiffiffi rffiffiffiffiffi rf rl kf ¼ jo and kl ¼ jo ð14Þ Gf Gl Another method to obtain the quartz crystal impedance is the transfer function method (TFM) introduced by Muramatsu [117]. The method consists of an oscillator that drives a crystal through a known measuring impedance and a radiofrequency voltmeter which measures the transference modulus of the system. In this paper, however, Muramatsu neglected the e¡ect of the parasitic capacitance, and his expression for the quartz impedance resulted in a nonlinear relationship between the measured resistance

Layered Protein Films

355

R with an AC voltage divider and the value of R measured by an impedance analyzer. Calvo and Etchenique [137] improved the method and introduced an analytical expression to ¢t the entire transfer function around resonance in order to obtain the same values or R, L, and C0 as measured by a frequencyresponse analyzer. The complex voltage divider used to measure the resonant frequency and both components of the quartz crystal Butterworth-Van Dyke equivalent circuit are shown in Fig. 6. The ratio of input (Vi ) to output ðV0 Þ AC voltage at 10 MHz is Vi Zm ¼1þ ð15Þ Vo ZQ where Zm is the measuring impedance and ZQ is the quartz acoustic impedance of the loaded quartz crystal. Note that the quantities in bold are vectors represented in the complex plane. In terms of the BVD equivalent electrical circuit of the quartz crystal, near resonance is ZQ ¼

ðL=C0 Þ  ð1=o2 CC0 Þ  ð jR=oC0 Þ R þ joL  j ½ð1=oCÞ þ ð1=oC0 Þ

ð16Þ

The absolute value of the transfer function spectrum, jVo =Vi jðoÞ measured in a narrow frequency interval around the resonant frequency is then

FIG. 6 Voltage divider measuring circuit and BVD equivalent electrical circuit (box), ZQ, and measuring impedance, Zm.

356

Calvo et al.

nonlinear ¢tted to the transfer function analytical expression obtained after substitution of ZQ and Zm ¼ j=oCm in Eq. (15): qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðoL  1=oC Þ2 þR2 Vo ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi V i ½oL  ð1=oCÞ þ ðoLC0 =Cm Þ þ ðC0 =oCCm Þ  ð1=oCm Þ2 þ½R þ ðRC0 =Cm Þ2

ð17Þ

Although this voltage divider method does not measure the phase of the transfer function, it has the advantage of speed, and the equivalent circuit parameters of the quartz crystal impedance can be determined in real time in 10 ms. This method has been employed to monitor in real time the quartz crystal equivalent circuit parameters RS and XLS of enzymes entrapped in redox hydrogels [137,138,147,148] and self-assembled enzyme multilayers [52^54], and their time and electrode potential dependence in electrochemical experiments. Experimental data for selected systems
viscoelastic ¢lms and liquids, self-assembled layers, and electroactive polymers
were reported, and these results have been analyzed on the basis of the Gransta¡ and Martin viscoelastic model, separating the inertial (rigid) mass and the viscoelastic components of quartz impedance during dynamic processes. Gravimetric measurements during electrochemical redox switching of an Os-containing enzyme hydrogel ¢lm in contact with aqueous electrolyte have been obtained beyond the Sauerbrey rigid mass limit for an acoustically nonrigid ¢lm [147]. In that case the frequency-to-mass Sauerbrey relationship is not valid, and the mass can been obtained from the electroacoustic impedance data measured simultaneously in electrochemical experiments by using Martin’s viscoelastic model. Film thickness and shear modulus have been obtained from imaginary and real components of the complex acoustic impedance in transient and steady-state oxidation of Os-containing hydrogels of poly(allylamine) and glucose oxidase [148]. 5

QCM STUDIES OF PROTEIN FILMS

Avery important ¢eld of application of the quartz crystal microbalance is the investigation of protein adsorption at functionalized surfaces, including piezoimmunosensors in the detection of the binding of antibodies to surfacecon¢ned antigens. A recent review article on the quartz crystal microbalance in life science summarizes the state of the art in this ¢eld [60]. In gravimetric terms, the QCM in principle permits quantifying the amount of adsorbed protein, the protein uptake kinetics in real time, and the binding constants from adsorption isotherms. The widespread use of this technique is based on its simplicity, the high mass sensitivity in the

Layered Protein Films

357

nanogram range, and the possibility of implementing gravimetric measurements in real time with the adsorbate in contact with liquids. Many studies of protein adsorption with QCM have been carried out by simply measuring the resonant frequency shift and assuming a direct quanti¢cation of the adsorbed protein using the Sauerbrey equation [149^165]. However, a number of reports have established that protein adsorption in contact with liquid leads to larger frequency shifts than the same protein ¢lm produces in air [150,166]. Two di¡erent sources have been identi¢ed for the failure of the Sauerbrey equation to account for the mass-to-frequency relationship in adsorbed protein ¢lms: (1) the viscoelastic properties of the hydrated protein ¢lm, which a¡ect the propagation of the shear wave; and (2) water and ions entrapped in the protein ¢lm,which are weighed by the acoustic sensor.This later aspect is highlighted by the comparison of acoustic and optical methods as discussed below. Muratsugu et al. [164] compared the mass load quanti¢ed by using radioisotope-labeled human serum albumin (HSA) and the frequency shift during the adsorption of HSA in a quartz crystal resonator. While the frequency-to-mass conversion factor predicted by the Sauerbrey equation is 0.183 Hz-cm2=ng for a 9-MHz quarz crystal, the radiolabeling results yielded much larger values of 0.375 Hz-cm2=ng. Go«pel et al. have reported that the QCM frequency shifts of rigid protein ¢lms immersed in aqueous electrolyte are 4^5 times larger than the values found for the same ¢lms under dry conditions [150]. It has been suggested that the excess mass is due to water molecules and ions associated with the protein. The sensitivity of multilayers composed of alternating biotinylated BSA and streptavidin layers is four times larger than the mass sensitivity as predicted by the Sauerbrey equation. By determining the thickness of the layer, Go«pel et al. demonstrated that approximately 75% of the overall protein ¢lm mass is water. By using complex network analysis, Thompson and coworkers [154] interpreted the frequency responses of thickness-shear mode (TSM) devices immersed in water upon interfacial immunochemical reactions in terms of the viscous properties of the protein layer, rather than changes in mass or bulk solution properties. Protein adsorption and denaturation on sensor surfaces [158] and interfacial nucleic acid chemistry have been studied by complex admittance of the composite resonator in contact with liquid [160^163], and the rigid mass uptake concept appeared inadequate in explaining the results. When a protein that carries charge (the surface charge depends on the protein isoelectric point and the solution pH) is adsorbed on an electri¢ed interface, a perturbation of the electrical double layers of the surface and the

358

Calvo et al.

protein takes place. Thompson and coworkers [158], using network analysis, also investigated the perturbation of the electri¢ed interface and the response of the TSM acoustic wave sensor during adsorption of a-chymotrysinogen A on a gold surface. The authors showed that the static parallel capacitance C0 and the motional resistance R of theTSM device vary during the protein adsorption. The variation of C0 is due to the capacity of the electrical double layer at the charged interface,which is modi¢ed by the charged protein adsorbate with probable conformational changes and redistribution of charge. Kasemo and coworkers, using QCM-D that allows frequency shifts to be monitored simultaneously with the energy dissipation factor D with a time resolution of 1 s [63], studied the adsorption of hemoglobin and methemoglobin. Plotting the frequency shift versus change in D for the binding of methemoglobin (met-Hb) and hemoglobin-CO (HbCO) on a hydrophobic methyl-terminated monolayer, two distinct slopes indicative of a two-step adsorption process were obtained. Note that Df D plots are equivalent to the parametric polar plots Rf  XLf in Fig. 5, but Df and D carry di¡erent units while XLf and Rf are both expressed in ohms. These authors concluded that energy loss occurs not only at the interfaces but also in the protein layer itself, probably due to conformational changes within the adsorbed protein layer and tightly bound water. As reported in a recent paper [65], Kasemo and coworkers measured the time-resolved adsorption kinetics of the mussel adhesive protein (Mefp1) on a nonpolar, methyl-terminated (thiolated) gold surface. The authors combined a quartz crystal microbalance with dissipation monitoring (QCM-D) with two independent optical techniques, surface plasmon resonance and ellipsometry. By measuring the frequency and energy dissipation at multiple harmonics (frequency dispersion of the shear modulus was neglected) and employing theoretical simulations based on a viscoelastic model, the authors were able to show quantitative data on the thickness, shear elastic modulus, and shear viscosity of the protein ¢lm even beyond the Sauerbrey regime. Substitution of H2O by D2O did not in£uence the adsorption behavior, but resulted in the expected di¡erences in the estimated e¡ective density and shear viscosity. Figure 7 depicts an example of this work with the time course of Df and DD during protein adsorption on a methyl-terminated surface, for di¡erent harmonics (n ¼ 1, 3, and 5). 6

QCM STUDIES OF SELF-ASSEMBLED PROTEIN LAYERS

The frequency shift in the quartz crystal microbalance has been extensively employed to monitor the buildup of electrostatic self-assembled polyelectrolytes of di¡erent charges [27,46,97,167^170].

Layered Protein Films

359

FIG. 7 Time course of 4fn and dissipation factor 4Dn at different harmonics (n = 1, 3, 5) for the adsorption of Mefp-1 on methyl-terminated surface from a buffer solution containing 25 mg=mL in 0.1 M acetate buffer (0.75 M NaCl, pH 5.5) followed by exchange of the protein solution for a pure buffer. (Reprinted with permission from Ref. 65. Copyright 2001 American Chemical Society.)

Caruso et al. [26,34,56,61,171^173] pioneered the use of the quartz crystal microbalance to study immunological activity of antibody layers constructed by alternate adsorption of oppositely charged polyelectrolytes and antibodies. The frequency changes of modi¢ed surfaces were studied with the QCM for successive adsorption of PAH and PSS layers on gold pretreated with 3-mercaptopropionic acid (MPA) for the immobilization of IgG and anti-IgG in multilayers.The assembly process of the multilayer ¢lms was monitored, recording the resonant frequency shift with a quartz crystal microbalance and complemented with plasmon surface resonance (SPR). Danilowicz et al. [47] studied the formation of a multilayer of electrically wired horseradish peroxidase biotin conjugate (HRP-b) by an antibiotin IgG structure built by sequential alternate adsorption of Os(bpy)2ClPyCH2NH-poly(allylamine) (PAH-Os) and goat antibiotin immunoglobulin, IgG-ab. The binding of antibiotin IgG-ab on PAH-Os redox polymer surface self-assembled on thiolated gold was proved by QCM, AFM, and by

360

Calvo et al.

the generation of an electrocatalytic signal upon addition of the enzyme substrate, hydrogen peroxide, to the solution in contact with the sensing layer. Figure 8 shows the time course of XLS and RS during the binding of HRP-b on IgG-ab self-assembled on the PAH-Os modi¢ed surface. The thiolated gold layer on the quartz crystal surface was modi¢ed with eight

FIG. 8 Time course of the thickness shear-mode impedance components XLS and RS during adsorption of 0.23 mM HRP-b conjugate onto (PAH-Os)8(IgG-ab)8 multilayer self-assembled structure LBL on thiolated gold. Inset: Polar parametric plot of RS versus XLS .

Layered Protein Films

361

PAH-Os=IgG-ab bilayers from 50 nM antibody solution and exposed, after rinsing with bu¡er, to 0.23 mM HRP-b conjugate solution. Both DXLS and DRS transients could be ¢tted to a double-exponential law with k1 ¼ 8:4  103 s1 and k2 ¼ 3:6  104 s1 . The mass uptake from HRP-b solution reached a ¢nal value of 0.69 mg=cm2 for the eight IgG layers, which corresponds to a molar surface coverage of 1.5 1011 mol=cm2 and the molar concentration ratio HRP-b to IgG-ab resulted close to one showing that the multilayer is permeable to the small di¡using HRP-b. It can be seen that RS represents almost one-half of the XLS , and therefore an important volume=viscoelastic e¡ect is present in the layer.This can be clearly seen in the polar parametric plot in the inset of Fig. 8. Anzai and coworkers [174,175] reported the LBL construction of multilayer ¢lms composed of avidin and biotin-labeled poly(amines) (a) or polyanions (b). In their study they followed the layer buildup by the frequency shift of the quartz crystal resonator. In the interaction of avidin ^ biotin conjugates, the frequency shift expected for the adsorption of a monolayer of avidin would be 22 Hz at 97 MHz, while the frequency drop observed by Thompson and Ghafouri [166] was 100^500 Hz for various layer structures. Not only the water of protein hydration but also interfacial properties such as viscoelasticity and acoustic coupling are likely to be responsible for the discrepancy. Figure 9, for example, depicts the TSM response in frequency change and the motional resistance associated with the introduction of sequences of avidin followed by biotinylated bovine serum albumin (a) or biotinylated insulin (b) with intermittent washing with bu¡er. A typical result obtained in our laboratory is shown in Fig. 10 for PAHOs and avidin self-assembled with biotinylated glucose oxidase, GOx-b. These self-assembled layers were employed as biosensors for the detection of b-D -glucose by glucose by oxidase generating an electrical signal by ‘‘wiring’’ the enzyme with the redox mediator polymer, PAH-Os. A thickness shear-mode quartz crystal resonator was employed, following in real time the variation of the surface complex acoustic impedance real and imaginary components, DXLS and DRS , after the addition of avidin to the osmium polyelectrolyte-covered surface, rinsing with bu¡er and adding biotinylated GOx to the liquid electrolyte. The uptake of biomolecules follows single exponential transients with time constants of 216 s for avidin of PAH-Os and 62 s for biotinylated GOx-b on the avidin-modi¢ed surface, respectively, as observed in other studies of these molecules. The DXLS saturation values in gravimetric mass are much less for avidin, 0.145 mg=cm2 (2.2 pmol=cm2, MW 68,000 Da), than for the heavier enzyme GOx-b, 1.12 mg=cm2 (6.0 pmol=cm2, MW 186,000 Da). The ratio of the surface molar concentrations is 2.73, as compared to the theoretical four avidin sites to bind GOx-b.

362

Calvo et al.

FIG. 9 Response of TSM device to multiple layers of avidin^biotin conjugates. The plots show the frequency change and the motional resistance, Rs , associated with the introduction of sequences of avidin followed by (a) biotinylated bovine albumin (BBA) and (b) biotinylated insulin (BI) with intermittent washing with buffer on electrodes of 9 MHZ quartz crystal sensor. (Reprinted with permission from Ref. 166. Copyright 1999 American Chemical Society.)

Whilethe valuesof the surface motional resistance RS are much lessthan the corresponding XLS , it is interesting to analyze the variation of RS during adsorption of these biomolecules. An increase of RS is obtained while adsorbing avidin on the polyelectrolyte-modi¢ed surface, but a small decrease appears during GOx-b adsorption while ellipsometric measurements indicate an increase in ¢lm thickness after adsorption of GOx-b on the avidin ^ polymer surface. It should be emphasized, however, that the optical and acoustic techniques sense di¡erent thickness, as will be discussed below. These results can be rationalized with the model proposed by Fawcett et al. for the adsorption of biomolecules on hairy polymer surfaces [17]. The oscillation of a thin dry polymer ^ biomolecule layer on the quartz crystal

Layered Protein Films

363

FIG. 10 Time course of the thickness shear-mode impedance components XLS and RS during adsorption of 0.33 mM avidin from HEPES buffer 50 mM pH 7.50 on Au=MPS=(PAH-Os) modified surface and further adsorption of 23 mM GOx-b conjugate.

(Fig.11A) is decoupled at the layer ^ air interface,which moves synchronously with the crystal. The shear wave damps rapidly to zero in air unless it is not perfectly elastic or it is thick ascompared tothe shear-wave penetrationdepth. On the other hand,when the polymer ^ avidin layer is in contact with an aqueous liquid (Fig. 11B), hydration and swelling of the ¢lm will occur with increase in ¢lm thickness. Acoustic wave attenuation occurs in the hydrated viscoelastic ¢lm and viscous liquid which is characterized by df (hyd), Gf0 (hyd), Gf00 (hyd), and rf (hyd). These quantities and the liquid density and viscosity de¢ne the surface acoustic impedance, ZS , according to Eq. (8). It is most likely that the decrease in motional resistance occurs and the hairy ¢lm (Fig. 11B) becomes more compact upon GOx-b binding and less energy is dissipated in the structure depicted schematically in Fig. 11C. Similar decrease in RS has been recorded under certain conditions during the adsorption of GOx multilayers on the PAH-Os polyelectrolyte without avidin.

364

Calvo et al.

FIG. 11 Schematic representation of a dry thin polyelectrolyte layer deposited on a quartz crystal and exposed to air (A), and the same ’’hairy‘‘ film immersed in liquid before (B) and after (C) adsorption of biomolecules.

Hodak et al. [52] deposited successive layers of poly(allylamine) derivatized with ferrocene, PAH-Fc, and GOx by alternate immersion of thiolmodi¢ed Au in the respective polyelectrolyte and enzyme solutions. The uptake of thiol, redox polymer, and enzyme on the surface was monitored by QCM with acoustic admittance analysis of the viscoelastic e¡ects [52]. During the adsorption of 3-mercaptopropane sulfonate on gold, a resonant frequency shift of 40 Hz was recorded,which corresponds to a mass uptake of 174 ng=cm2 (0.98 nmol=cm2). The use of the QCM technique to measure poly(allylamine) uptake on the thiolated gold by electrostatic adsorption was complicated by a strong viscoelastic e¡ect due to the rheological properties of the PAH solution. Using a Pierce-Miller oscillator, which operates at the parallel resonant frequency, op (see Fig. 2), a positive frequency shift was observed which would correspond to a decrease in mass. This result could not be explained by either a simple mass change or an increase in the Newtonian viscosity of the liquid.

Layered Protein Films

365

To overcome this problem, Hodak et al. employed acoustic impedance analysis, which showed an increase of DXLS and DRS on addition of PAH-Fc to the bu¡er solution in contact with the thiol-modi¢ed gold-coated quartz crystal.When the quartz crystal was then removed from the PAH-Fc solution and thoroughly rinsed with water, electrochemical measurements indicated that the redox polymer was present on the surface. Measurements of the quartz impedance parameters for the modi¢ed quartz crystal in water gave changes in DXLS and DRS of 1.6 and 0.2 O, respectively,when compared to the corresponding values for the crystal in water before adsorption of the PAHFc. From that it could be concluded that the abnormal increase in DXLS and DRS with DRS > DXLS was due entirely to the rheological properties of the polymer solution (Maxwell £uid) in contact with the quartz crystal.This was further con¢rmed by repeating the experiment with addition of PAH-Fc to water, but now using a crystal which was already coated with the redox polymer.Then, as expected, the same rheological changes in the solution but with no polymer adsorption onto the crystal were observed, DRS increased in approximately the same manner and by the same amount as in the ¢rst experiment, but DXLS increased by a lesser amount than in the ¢rst case. By comparison of the steady-state values of DXLS , the mass of adsorbed PAH-Fc could be estimated to be 46 ng=cm2 (0.49 nmol=cm2 of monomer units). The electrostatic adsorption of polyanionic GOx on the poly (allylamine) top layer was shown by a large negative frequency shift observed by the QCM. The saturation values for di¡erent numbers of polymer ^ enzyme bilayers resulted similar to those reported by Kunitake et al. for multilayers of GOx self-assembled with PEI [27]. Having established that the quartz crystal microbalance could be used with success to follow the stepwise deposition of enzyme molecules in multilayer structures, Pietrasanta et al. [98] further investigated the adsorption kinetics of GOx on a modi¢ed surface with a similar polycationic osmium redox polymer, PAH-Os. The mass transient of adsorbed glucose oxidase from 50 nM GOx onto the Au=thiol=PAA-Os surface obtained from the frequency shift using the Sauerbrey equation can be ¢tted to a doubleexponential law: Dm ¼ Að1  ek1 t Þ þ Bð1  ek2 t Þ

ð18Þ

with most of the enzyme uptake corresponding to a fast simple exponential dependence with a time constant k1 ¼ 1:2  102 s1 and a much smaller second rate constant k2 ¼ 1:9  106 s1 for the second exponential term. Subsequent alternate layers Au=thiol=(PAH-Os)n=(GOx)m showed a similar adsorption kinetic pattern. The maximum surface concentration of GOx in the ¢rst layer resulted in 6.2 1012 mol=cm2 for both 50 nM and 1 mM solutions of the enzyme, indicating surface saturation.

366

Calvo et al.

In order to establish the role played on the enzyme adsorption by the adsorbed polyelectrolyte with multiple positive charges, a clean Au surface was modi¢ed with single positively charged 2,20 -diaminoethyldisul¢de (cystamine) with subsequent adsorption of GOx. Unlike the adsorption on the PAH-Os-modi¢ed surface, on cystamine the enzyme uptake showed monoexponential kinetics with k1 ¼3.5 103 s1 and a maximum of 4.6  1012 mol=cm2, which compares very well to the value expected for a densely packed monolayer, 4.7 1012 mol=cm2, based on the X-ray crystallographic data of the GOx enzyme. Two conclusions have been drawn from that experiment: the polycationic PAH-Os can accommodate more GOx molecules than cystamine adsorbed on gold; and for the polymer-modi¢ed surface, a more complex adsorption kinetic pattern is apparent. AFM studies of the di¡erent layers showed the existence of large twodimension enzyme aggregates on the osmium polymer for1 mM GOx and less aggregation for 50 nM GOx solutions.When the short alkanethiol, 2,20 -diaminoethyldisul¢de, was preadsorbed on gold, a monoexponential adsorption law was observed and single GOx enzyme molecules could be seen on the surface where the enzyme was adsorbed from 50 nM GOx in water. Furthermore, it has been found that there is an important di¡erence in the adsorption of GOx on a polymer-modi¢ed surface from water solutions or from solutions containing electrolyte, as previously reported by Hodak et al. [52].While a simple adsorption pattern was found for the adsorption of glucose oxidase from water solution, for the adsorption from GOx solutions containing electrolyte an overshot was observed, after which a constant surface concentration was reached. A consistent explanation for this overshot is as follows: in the presence of electrolyte the protein charges are screened, and therefore a larger amount of GOx is required to saturate the excess positive charge on the electrode surface. However, the ¢nal amount of GOx on the surface will be determined by the positive charge of the poly(allyamine) on the surface, and any excess of GOx will be released into the solution. The study of the motional resistance transient RS in both cases also shows a distinctive pattern for the adsorption of GOx from the electrolyte or from pure water. In the ¢rst case a decrease in DRS was observed during the enzyme mass overshot and a regular increase in DRS was found in water solutions of GOx. The steady-state mass uptake from 1 mM GOx solutions on the PAHOs-terminated surface did not show a constant mass increase in the ¢rst few layers, but subsequent layers showed a constant mass increase per enzyme layer. This phenomenon, which is typical of LBL adsorption of polyeletrolytes, has been reported by several authors and can be explained by the three-zone model proposed by Decher [13]. From the QCM sensitivity factor

Layered Protein Films

367

at 10 MHz for the ¢rst enzyme layer, 6.2 1012 mol=cm2 was found, while for the fourth and subsequent layers a constant increment per layer of 2.0  1011 mol=cm2 was observed, much larger than the expected value for a GOx monolayer, 4.7 1012 mol=cm2. Go«pel [150] observed that the adsorption of proteins in contact with water and ions, monitored by QCM, results in a mass increase between 2 and 4 times the respective mass for the dry protein, and suggested that the excess mass is due to solvent molecules and ions associated with the protein. In a recent work [176], the Calvo group studied in detail the quartz crystal acoustic admittance in thickness shear-mode resonators loaded with self-assembled multilayers comprised of alternate layers of GOx and PAHOs, deposited on MPS-modi¢ed gold on the quartz crystal. The complex acoustic impedance parameters, RS and XLS ,were determined for organized thin ¢lms of di¡erent thicknesses obtained by varying the number of enzyme layers, n, in (PAH-Os)n(GOx)n structures. The ellipsometric ¢lm thickness and mass for dry enzyme multilayer ¢lms and ¢lms in contact with aqueous electrolyte were evaluated and the average ¢lm density estimated. Both ellipsometric thickness and gravimetric QCM measurements revealed that the self-assembled PAH-Os=GOx multilayers are highly hydrated in contact with aqueous electrolyte. For the ¢lm in contact with aqueous electrolyte, the steady-state mass was 35% larger than for the same dry ¢lm. By combination of the estimated ¢lm thickness and density, an expression for the surface mechanical impedance of the lumped-element modi¢ed resonator [Eq. (9)]
the Gransta¡ and Martin model
and the electrolyte density and viscosity, the values of storage (G0 ) and loss shear modulus (G 00 ) at 10 MHz for di¡erent ¢lms were estimated as a function of ¢lm thickness and electrode potential on the basis of the measured electroacoustic impedance. The experimental data for XLS and RS for a multilayer enzyme ¢lm (PAH-Os)14 (GOx)14 are depicted in Fig. 12; both components of the complex acoustic impedance increase with the number of self-assembled layers, with the motional resistance very low
indicating that these multilayers behave as acoustically thin ¢lms (rigid ¢lms). Assuming constant shear moduli for all self-assembled layers,to a good approximation, then, the results depicted in Fig. 12 can be explained by an increase in ¢lmthicknessevery time a newenzymelayer is added.Equation (9) describes XLS and RS for a viscoelastic model based on four parameters: ¢lm thickness, density, and shear moduli. The calculated values of XLS and RS with viscoelastic values for the ¢lm G 0 ¼ 9:9 MPa, G00 ¼ 0.7 MPa, rf ¼ 1:5 g=cm3, and ¢lm ellipsometric thickness, liquid density, and viscosity are shown by the solid line in Fig. 12.

368

Calvo et al.

FIG. 12 Experimental data (open circles) and simulations of XLS and RS (solid line) layers of (PAHOS)14(GOX)14 under water with Gf0 ¼ 9.9 MPa, G00 ¼ 0.7 MPa, rf ¼ 1.5 g=cm3, rH2O ¼ 1 g=cm3, and G00H2 O ¼ Z ¼ 4:6  104 Pa. Inset: Parametric plot of Rf versus XLf and simulation (solid line).

Except for the ¢rst deposited layers,very good agreement with the calculated data con¢rms that each enzyme and polyelectrolyte bilayer contributes approximately the same shear modulus and density. The values of XLS and RS increase with the number of deposited PAHOs=GOx pairs of layers, i.e., with the ¢lm thickness at constant density. The inset is a parametric plot of RS versus XLS for the experimental impedance data and the solid line calculated with Eq. (9) and the above ¢lm parameters. Good agreement with the viscoelastic model is apparent given the approximations, since a single set of G0, G00, and rf was used to simulate all the experimental data for 14 enzyme layers. The discrepancy between experimental and simulated data for low and high thickness may result in

Layered Protein Films

369

slight changes of viscoelastic parameters with thickness not taken into account in the simulation. The magnitude found for G0 and G00 is also in the range previously reported for much thicker hydrogels and two orders of magnitude larger than for pure water, ca. G00 ¼ 5.4  10 4 Pa. Therefore, the self-assembled layerby-layer enzyme thin ¢lms behave as lossy viscoelastic ¢lms at 10 MHz, where the ionic atmosphere relaxation of ionized groups is responsible for the storage modulus, and the friction of ions and solvent with the moving polymer chains is responsible for the loss modulus, but the ¢lm thickness is much less than the shear-wave penetration depth in the ¢lm. Under a sinusoidal perturbation the ions in the polyelectrolyte at the surface exhibit an oscillatory motion around their equilibrium positions. At low frequency the ionic atmosphere associated to a polyion has an asymmetry due to the external force balanced by internal friction with the solvent. At higher frequencies, when the ionic atmosphere relation time, t, is such that ot  1, the asymmetric ionic atmosphere has less chance to be formed. This is expected to occur in the megahertz frequency range where the QCM works. Finite reorientation time, t, of molecules under shear stress characterize Maxwellian £uids [177]: Z¼

Z0 1 þ jot

ð19Þ

with Z0 the low-frequency viscosity. For ot << 1ZðoÞ ! Z0 ; for ot  1, on the other hand, the elastic energy is not totally dissipated in viscous £ow and some is stored elastically by the polymer. In the reduced-state Os(II) polymer, the (PAH-Os)n(GOx)n multilayers with n 14 in contact with electrolyte behave as acoustically thin ¢lms and the increase of surface acoustic impedance corresponds simply to an increase of the hydrated ¢lm thickness. When the ¢lms are switched between reduced [PAH-Os(II)] and oxidized [PAH-Os(III)] states, however, a pronounced change in the complex surface acoustic impedance is observed, as shown in Fig. 13 for ¢lms of di¡erent thickness, with n ¼ 4, 7, and 14 [178]. In all three cases the electrochemical oxidation of the osmium polymer results in an increase in the imaginary part of the complex surface acoustic impedance (negative shift of the motional resonant frequency).The motional resistance RS also increases during ¢lm oxidation for thickness larger than 200 nm (seven bilayers),while thinner ¢lms appear to be acoustically thin at 10 MHz. The values of XLS and RS and the ellipsometric thickness and ¢lm density at each electrode potential were used to solve Eq. (9) for G0 and G00 using a two-dimensional root-¢nding algorithm (secant method) with Mathematica

370

Calvo et al.

FIG. 13 Real, RS , and imaginary, XLS , components of the surface acoustic impedance at 10 MHZ for (PAH-OS)n(GOX)n in contact with 0.12 M KNO3 and 0.10 M Tris buffer electrolyte of pH 7.5 for different film thickness, with n = 4, 7, and 14 during oxidation of the OS(II) in the film under cyclic voltammetry conditions at 5 mV=s.

4.0 with error less than 0.01 O for RS and XLS. For the liquid electrolyte viscosity and density, Z1 ¼ 0:01 cm2 =s and r1 ¼ 1 g=cm3 were used. During electrochemical oxidation of the ¢lms the increase in thickness measured by ellipsometry is not su⁄cient to explain the change in the observed complex impedance components. An increase of G00 upon oxidation and a relatively constant G0 with a slight decrease in the fully oxidized ¢lm is observed. The mass increases due to the exchange of anions and

Layered Protein Films

371

solvent with the electrolyte and the viscoelastic changes during oxidation ^ reduction cycles of self-assembled ¢lm result from swelling and conformational changes of the enzyme structure. 7

COMBINATION OF QCM AND ELLIPSOMETRY

Film thickness, refractive index, and extinction coe⁄cient can be assessed with ellipsometry using equations derived from Fresnel equations that relate the changes of elliptically polarized light parameters measured after incidence on the sample [179^182]. Although other related optical methods such as surface plasmon resonance and scanning angle re£ectometry have been used for the study of layered protein ¢lms, only ellipsometric studies of self-assembled protein ¢lms will be discussed in this chapter. Ellipsometry has been extensively employed to study the growth of multilayer self-assemblies composed of lineal [32,183^186] and hyperbranched [187] polycations and polyanions. Except in a few cases where spectroscopic ellipsometry was employed [184,185] or ellipsometric measurements were supported by XRR [186], the value of a complex refractive index has been assumed on the basis of literature data to extract the ¢lm thickness values. Ellipsometric studies of self-assembled multilayers with proteins have also been reported in the literature [81,182,188,189]. They comprised spectroscopic ellipsometry studies of avidin ^ biotin self-assembled multilayers [188,189] as well as thickness estimations for GOx self-assembled layer by layer with a redox polymer [182] and in LB ¢lms [81]. Forzani et al. [179] studied self-assembled multilayers composed of alternated layers of GOx and PAH-Os, deposited on a MPS-modi¢ed gold surface under ex-situ and in-situ conditions by ellipsometry and QCM. The ellipsometric parameters of the thiol ¢lm on gold in the ¢rst layer were analized in terms of an anisotropic single-layer model. For the subsequent (PAH-Os)n(GOx)n multilayers on Au, a two-layer model with the anisotropic thiol ¢lm and the isotropic enzyme=polyelectrolyte ¢lm yielded identical results as an isotropic one-layer model with the substrate parameters measured after thiol adsorption to o¡set any e¡ect due to the Au ^ S bond. Film thickness and complex refractive index for each adsorbed layer in Au=MPS=(PAH-Os)n (GOx)n multilayers were determined for dry ¢lms and for ¢lms in contact with water, revealing the importance of water content control in these selfassembled structures. The drying steps change the system hydration state and consequently its optical properties. Complex refractive indices of 1.432^0.000i, 1.439^0.005i, and 1.435^0.008i, and 1.55^0.021i, 1.58^0.030i, and 1.54^0.046i were obtained for in situ and ex situ experiments, respectively, at 628.3, 546.1, and 405.0 nm. The values of complex refractive index

372

Calvo et al.

determined from in-situ experiments suggested that the enzyme kept its native state after the adsorption step. Gottesfeld et al. [181] studied GOx entrapped in polypyrrole ¢lms. They showed that the combined information of ellipsometric ¢lm thickness and QCM ¢lm mass makes it possible to assess the apparent mass density of the GOx ^ polymer ¢lms during their growth. As an example, Fig. 14 shows acoustic mass ^ ellipsometric thickness plots obtained for di¡erent enzyme=polycation self-assembled multilayers built up under similar conditions. It can be observed that QCM and ellipsometry yield concordant results and therefore it is possible to estimate the average apparent ¢lm density. Moreover, in this case of L-B-L self-assembled multilayers, an apparent density value for each enzyme=polycation bilayer can be calculated, so the models for the enzyme= polycation distribution in the multilayer structure can be assessed. In the case of PAH=GOx self-assembled systems, Forzani et al. [179] found that the amount of adsorbed GOx during each step ranges from one to two times the value of a monolayer in the proximity of the substrate to a value corresponding to 10 times a monolayer once the fourth

FIG. 14 Adsorbed enzyme acoustic mass versus ellipsometric thickness for PAHOS=GOX,PAH=polyphenoloxidase(PPO)andpoly(diallyldimethylamine)(PDDA)= PPO self-assembled multilayers. Data taken after each enzyme adsorption and washing step under water or 10 mM Tris buffer pH ¼ 6.4.

Layered Protein Films

373

PAH-Os=GOx bilayer is built up. Therefore, the enzyme spatial distribution in this self-assembly remains heterogeneous. It should be noticed that the ¢lm thicknesses derived from acoustic and optical measurements do not coincide since water molecules, which are not polarized in the visible spectral region, are not detected in re£ectance methods, unlike the shear wave, which senses water in the ¢lm as a coupled mass. The acoustic contrast between the ¢lm and liquid depends on the di¡erence in shear modulus (i.e., 106 Pa for the ¢lm and 10 4 Pa for water), and this is more pronounced than the optical contrast given by the di¡erence in refractive indices of the hydrated ¢lm and water (ca. 1.43 for the protein ¢lm and 1.33 for water). Therefore, the most highly hydrated outer regions of the ¢lm (see Fig. 11B) cannot be distinguished from the aqueous electrolyte by the optical methods but can be sensed by the acoustic wave. An ellipsometric areal mass density, Dme , can be derived from the optical ¢lm thickness (df ) and ¢lm refractive index(nf ) through an equation proposed by Feijter et al. [190]: nf  n0 ð20Þ Dme ¼ df qn=qc where n0 is the electrolyte solution refractive index, nf is the ¢lm refractive index, and qn=qc is the refractive index increment of the adsorbed substance measured. Ho«o«k and Kansemo [65] have represented the protein ¢lm with an e¡ective hydrodynamic thickness, de¡, and an e¡ective density, reff . The e¡ective thickness can be expressed in terms of the Sauerbrey mass, DmQCM, the ellipsometric mass, Dme , and the respective densities of the dry protein ¢lm, rdry ¼ 2:0 g=cm3, and water, rwater, as follows: deff ¼

DmQCM DmQCM ¼ reff rdry ðDme =DmQCM Þ þ rwater ½1  ðDme =DmQCM Þ ð21Þ

The e¡ective thickness is always larger than the ellipsometric thickness and smaller than the QCM thickness as depicted in Fig. 15. In the few ¢rst layers the e¡ect of the ‘‘hairy’’ external ¢lm=water structure depicted in Fig. 11B, which cannot be seen by ellipsometry, becomes dominant and, as more layers are built up, the e¡ect of the interface with the external electrolyte becomes less important. 8

CONCLUDING REMARKS

The use of shear thickness-mode resonators with either frequency shift and bandwidth, real and imaginary components of the acoustic surface

374

Calvo et al.

FIG. 15 Plot of the effective thickness to ellipsometric thickness ratio, deff=de versus the number of self-assembled (PAH-OS)n(GOX)n multilayers in contact with water. qn=qc (0.177 cm3=g) was measured in solutions of GOX concentrations range between 64 and 321 mg=mL.

impedance=admittance, or frequency shift and energy dissipation factor, represents an invaluable tool to obtain gravimetric, thickness, and viscoelastic information of layered protein ¢lms. These techniques can be applied to ¢lms either under steady state or to study protein ¢lm dynamics in real time and in situ, with the bio¢lms in contact with liquids. Combining acoustic and optical techniques, which sense di¡erent properties of these layered protein ¢lms, better insight into the ¢lm structure and thickness can be gained in addition to protein ¢lm mass. Furthermore, measurement of the optical ¢lm thickness is very useful for the protein ¢lm viscoelastic model to be applied and hence to obtain the mass beyond the Sauerbrey limit. The evaluation of the gravimetric error with the quartz crystal microbalance is important when an inventory of active components such as antibodies or enzymes in the multilayers is needed. One such application is the comparison of the number of enzyme molecules ‘‘wired’’ by molecular polymer redox mediators with the total number of enzyme molecules in the layered polymer ¢lm, or the number of antibodies or avidin molecules in the multilayer that are e¡ectively participating in antigen or biotinylated molecule binding.

Layered Protein Films

375

ACKNOWLEDGMENTS Financial support from the University of Buenos Aires, CONICET, ANPCyT, and Motorola Semiconductor Sector, Arizona, are gratefully acknowledged. E. Forzani, M. Otero, and A.Wolosiuk also thank CONICET for their fellowships. E. J. Calvo acknowledges a Guggenheim Fellowship (2001).

REFERENCES 1. A Ullman. Chem Rev 96:1533, 1996. 2. Y Lvov. In: Y Lvov, H. Mo«hwald, eds. Protein Architecture: Interfacing Molecular Assemblies and Immobilization Biotechnology. New York: Marcel Dekker, 2000. Ch. 6, pp 125^167. 3. TA Horbett, JL Brash. In: TA Horbett, JL Brash, eds. Proteins at Interfaces: Fundamental and Applications. ACS Symposium Series 602.Washington, DC: American Chemical Society, 1995, p 1. 4. D Catty, C Raykundalla. In: D Catty, ed. Antibodies: A Practical Approach. Vol. II. Oxford, UK: IRL, 1989, p 97. 5. I Karube. In: APF Turner, I Karube, G Wilson, eds. Biosensors: Fundamentals and Applications. Oxford, UK: Oxford University Press, 1987, p 471. 6. JL Guesdon, S Avrameas. In: LB Wingard, E Katchalski-Katzir, L Goldstein, eds. Applied Biochemistry and Bioengineering. New York: Academic, 1981, 207. 7. R Ros Seigel, P Harder, R Dahint, M Grunze, F Josse, M Mrksich, GM Whitesides. Anal Chem 69:3321, 1997. 8. IV Turko, IS Yurkevich,VL Chashchin. Thin Solid Films 210=211:710, 1992. 9. M Ahlers,W Mu«ller, A Reichert, H Ringsdorf, J Venzmer. Angew Chem Int Ed Eng 129:1269, 1990. 10. W Mu«ller, H Ringsdorf, E Rump, G Wildburg, X Zhang, L Angermaier, W Knoll, M Liley, J Spinke. Science 262:1706, 1993. 11. T Cassier, K Lowack, G Decher. Supramol Sci 5:309, 1998. 12. PL Edmiston, SS Saavedra. J. Am Chem Soc 120: 1665, 1998. 13. G Ladam, P Schaaf, FJG Cuisinier, G Decher, JC Voegel. Langmuir 17:878, 2001. 14. (a) G Decher, J-D Hong, J. Schmitt. J Thin Solid Films 210=211:831, 1992. (b) G Decher, J Schmitt. J Prog Colloid Polymer Sci 89:160, 1992. 15. G Decher. Science 277:1232, 1997. 16. J Schmitt, G Decher, WJ Dressick, SL Brandow, RE Geer, R Shashidhar, JM Calvert. Adv Mater 9:61, 1997. 17. DL Feldheim, KC Grabar, MJ Natan, TE Mallouk. J Am Chem Soc 118:7640, 1996. 18. NA Kotov, I Dekany , JH Fendler. J Phys Chem 99:13065, 1995. 19. MD Musick, CD Keating, MH Keefe, MJ Natan. Chem Mater 9:1499, 1997. 20. TM Cooper, AL Campbell, RL Crane. Langmuir 11:2713, 1995.

376

Calvo et al.

21. 22. 23. 24. 25. 26. 27. 28. 29. 30.

K Araki, MJ Wagner, MS Wrighton, Langmuir 12:5393, 1996. K Ariga,Y Lvov,T Kunitake. J Am Chem Soc 119:2224, 1997. D. Yoo, AP Wu, J Lee, MF Rubner. Synth Met 85:1425, 1997. C Tedeschi, F Caruso, H Mo«hwald, S Kirstein. J Am Chem Soc 122:5841, 2000. M Onda, K Ariga,T Kunitake. J Biosci Bioeng 87:69,1999. F Caruso, K Niikura, DN Furlong,Y Okahata. Langmuir 13:3427, 1997. Y Lvov, K Ariga, I Ichinose,T Kunitake. J Am Chem Soc 117:6117, 1995. M Onda,Y Lvov, K Ariga,T Kunitake. Biotechnol Bioeng 51:163, 1996. M Onda,Y Lvov, K Ariga,T Kunitake. J Ferment Bioeng 82:502, 1996. G Ladam, P Schad, JC Voegel, P Schaaf, G Decher, E Cuisinier. Langmuir 16:1249, 2000. G Decher. Macromolecules 31:8893, 1998. SS Shirartori, MF Rubner. Macromolecules 33:4213, 2000. Y Lvov, K Ariga,T Kunitake. Chem Lett 6117, 1994. F Caruso, DN Furlong, K Ariga, I Ichinoze, T Kunitake. Langmuir 14:4559, 1998. SW Keller, H-N Kim,TE Mallouk. J Am Chem Soc 116:8817, 1994. G Ladam, C Gergely, B Senger, G Decher, JC Voegel, P Schaaf, FJG Cuisinier. Biomacromolecules 1:674, 2000. G Ladam, P Schaaf, FJG Cuisinier, G Decher, JC Voegel. Langmuir 17:878, 2001. L Alfonta, AK Singh, I Willner. Anal Chem 723:91, 2001. J Wang, PAV Pamidi, KR Rogers. Anal Chem 70:1171, 1998. AJ Killard, S Zhang, H Zhao, R Hohn, EI Iwuoha, MR Smyth. Anal Chim Acta 400:109, 1999. P Treloar, J Kane, P Vadgama. In: CP Price, DJ Newman, eds. Principles and Practice of Immuno-assay. London: Macmillan Reference, 1997, Ch 19. M Liu, QX Li, GA Rechnitz. Anal Chim Acta 387:29, 1999. B Ko«ning, M Gra«tzel. Anal Chem 66:341, 1994. E Uttenthaler, C Ko«llinger, S Drost. Anal Chim Acta 362:91, 1998. M Minunni, M Mascini, RM Carter, MB Jacobs, GJ Lubrano, GG Guilbault. Anal Chim Acta 325:169, 1996. YM Lvov, Z Lu, JB Schenkman, X Zu, JF Rusling. J Am Chem Soc 120:4073, 1998. C Danilowicz, JM Manrique. Electrochem Commun 1:22, 1999. C Duan, ME Meyerho¡. Anal Chem 66:1369, 1994. C Berggren, B Bjarnason, G Johansson. Biossensor Bioelectron 13:1061, 1998. M Dijksma, B Kamp, JC Hooguliet,WP Bennekom. Anal Chem 73:901, 2001. R Blonder, S Levi, G Tao, I Ben-Dov, I Willner. J Am Chem Soc 119:10467, 1997. J Hodak, R Etchenique, EJ Calvo, K Singhal, PN Bartlett. Langmuir 13:2708, 1997. ES Forzani,VM Solis, EJ Calvo. Anal Chem 72:5300, 2000. EJ Calvo, F Battaglini, C Danilowicz, A Wolosiuk, M Otero. Faraday Discuss 116:47, 2000. K Nakanishi, H Mugurama, I Karube. Anal Chem 68:1695, 1996.

31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55.

Layered Protein Films 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88.

377

F Caruso, E Rodda, D Neil Furlong. J Colloid Interface Sci 178:104, 1996. I Vikholm,WM Albers, Langmuir 14:3865, 1998. W Qian, D Yao, B Yu, F Yu, Z Lu,W Knoll. Chem Mater 11:1399, 1999. AA Karyakin, GV Presnova, MY Robtsova, AM Egorov. Anal Chem 72:3805, 2000. A Jansho¡, HJ Galla, C Steinem. Angew Chem Int Ed 39:4004, 2000. F Caruso, D Neil Furlong, P Kingshott. J Colloid Interface Sci 186: 129, 1997. F Ho«o«k, M Rodahl, B Kasemo, P Brzezinski. Proc Natl Acad Sci USA 95: 12271, 1998. F Ho«o«k, M Rodahl, P Brzezezinski, B Kasemo. Langumuir 14:729, 1998. J Rickert, A Brecht,W Go«pel. Anal Chem 69: 1441, 1997. F Ho«o«k, B Kasemo, T Nylander, C Fant, K Scott, H Elwing. Anal Chem 73:5796, 2001. R Rajagopalan, A Heller. In: J Jortner, M Ratner, eds. Molecular Electronics. Oxford, UK: Blackwell, 1997, p 241. Y Degani, A Heller. J Am Chem Soc 111:2357, 1989. BA Gregg, A Heller. J Phys Chem 95:5970, 1991. BA Gregg, A Heller. J Phys Chem 95:5976, 1991. A Heller. Accts Chem Res 23:128, 1992. MV Pishko, I Katakis, SE Lindquist, L Ye, BA Gregg, A Heller. J Phys Chem 96:3579, 1992. J Willner, A Riklin, BShoham, DRivenzon, E Katz. Adv Mater 5(12):912,1993. N Anicet, A Anne, J Moiroux, JM Save¤ant. J Am Chem Soc 120:7115, 1998. RW Murry. In: RW Murray, ed. Molecular Design of Electrode Surfaces. New York: Wiley, 1992, P 1. A Ulamnn. In: An Introduction to Ultrathin Organic Films: From LangmuirBlodgett to self-Assembly. Boston: Academic, 1991. I Willner, A Riklin, Anal Chem 66:1535, 1994. I Willner, N Lapidot, A Riklin, R Kasher, E Zahavy, E Katz. J Am Chem Soc 116:1428, 1994. G Decher, JD Hong. Ber Bunsenges Phys Chem 95:1430, 1991. Y Lvov, G Decher, G Sukhorukov. Macromolecules 26:5396, 1993. Y Lvov, G Decher, H Mohwald. Langmuir 9:481, 1993. S Sun, PH Ho-Si, DJ Harrison. Langmuir 7:727, 1991. F Mizutani,Y Sato, S Yabuki,Y Hirata. Chem Lett 251, 1996. KT Kinnear, HG Monbouquette. Langmuir 9:2255, 1993. AJ Guiomar, JT Guthrie, SD Evans. Langmuir 15:1198, 1999. R Blonder, E Katz,Y Cohen, N Itzhak, A Riklin, I Wilner. Anal Chem 68:3151, 1996. C Bourdillon, C Demaille, J Gueris, J Moiroux, JM Save¤ant. J Am Chem Soc 115:1264, 1993. C Bourdillon, C Demaille, J Moiroux, JM Save¤ant. J Am Chem Soc 116:10328, 1994. C Bourdillon, C Demaille, J Moiroux, JM Save¤ant. J Am Chem Soc 117:11499, 1995.

378

Calvo et al.

89. C Bourdillon, C Demaille, J Moiroux, JM Save¤ant. Accts Chem Res 29:529, 1996. 90. N Anicet,C Bourdillon, J Moiroux, JM Save¤ant. J Phys Chem B102:9844,1998. 91. C Bourdillon, C Demaille, J Moiroux, JM Save¤ant. J Phys Chem B 103:8532, 1999. 92. JF Rusling. Accts Chem Res 31:363, 1998. 93. JF Rusling. In: Y Lvov, H Mohwald, eds. Protein Architecture, New York: Marcel Dekker, 2000, p 337. 94. JG Franchina, WM Lackowski, DL Dermody, RM Crooks, DE Bergbreiter, K Sirkar, RJ Russell, MV Pishko. Anal Chem 71:3133, 1999. 95. A’P Alzari, N Anicet, C Bourdillon, J Moiroux, JM Saveant. J Am Chem Soc 118:6788, 1996. 96. C Bourdillon, C Demaille, J Moiroux, JM Saveant. J Am Chem Soc 121:2401, 1999. 97. N Anicet, C Bourdillon, J Moiroux, JM Saveant. Langmuir 15:6527, 1999. 98. EJ Calvo, R Etchenique, L Pietrasanta, AWolosiuk, C Danilowicz. Anal Chem 73:1161, 2001. 99. DA Buttry, MD Ward. Chem Rev 92(6):1355, 1992. 100. DA Buttry.In:H Abrun‹a,ed.Electrochemical Interfaces. Weinheim,Germany: VCH, 1991, p 531. 101. MD Ward. In: J Rubinstein, ed. Physical Electrochemistry. New York: Marcel Dekker, 1995, p 293. 102. G Sauerbery. Z Phys 155:206, 1959. 103. M Yang, M Thomson,WC Duncan-Hewitt. Langmuir 9:802, 1993. 104. HD Hillier, MD Ward. Anal Chem 64:2539, 1992. 105. KK Kanazawa, JG Gordon. Anal Chim Acta 175:99, 1985. 106. F Josse, ZA Shana, DE Radke, DT Haworth. IEEE Trans Ultrasound, Ferroelectrics, Freq Control 37:359, 1990. 107. KE Heusler, A Grzegorzewski, L Jackel, J Pietrucha. Ber Bunsenges Ges Phys Chem 92:1218, 1988. 108. SJ Martin, HE Hager. J Appl Phys 65:2627, 1989. 109. Z Lin, MD Ward. Anal Chem 67:685, 1995. 110. CE Reed, KK Kanazawa, JG Gordon. Anal Chem 57:1770, 1985. 111. S Bruckenstein, S Swathirajan. Electrochim Acta 30:851, 1985. 112. M Deakin,T Li, O Melroy. J Electroanal Chem 243:343, 1988. 113. WP Mason,WO Baker, HJ McSkimin, JH Heiss. Phys Rev 75(6):936, 1949. 114. R Nomura, M Okuhara. Anal Chim Acta 147:365, 1982. 115. S Bruckenstein, M Shay. Electrochim Acta 30:1295, 1985. 116. R Borjas, DA Buttry. J Electroanal Chem 280:73, 1990. 117. H Muramatsu, X Ye, M Suda, T Sakuhara, T Ataka. J Electroanal Chem 322:311, 1992. 118. SJ Martin,GC Frye.Proc.IEEE Ultrasonics Symp.NewYork:IEEE1991, p 393. 119. SJ Martin,VE Gransta¡, GC Frye. Anal Chem 63:2272, 1991. 120. H Muramatsu, K Kimura. Anal Chem 64:2502, 1992. 121. H Muramatsu, A Egawa,T Ataka. J Electroanal Chem 388:89, 1995.

Layered Protein Films 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155. 156. 157.

379

V Grasta¡, S Martin. J Appl Phys 75:1319, 1994. R Lucklum, C Behling, RW Cernosek, SJ Martin. J Appl Phys 30:346, 1997. S Martin,V Grasta¡, G Frye. Anal Chem 63:2272, 1991. MA Noel, PA Topart. Anal Chem 66:484, 1994. PA Topart, AM Noel. Anal Chem 66:2926, 1994. DM Kelly, JG Vos, RA Hillman. J Chem Soc FradayTrans 92:4121, 1996. N Oyama,T Tatsuma, K Takahashi. J Phys Chem 97:10504, 1993. S Ikeda, N Oyama. Anal Chem 65:1910, 1993. N Oyama, K Takada, T Tatsuma, K Naoi, T Okajmia. T Oksaka, Sensors Actuators B 13^14:372, 1993. S Yamaguchi,T Shimomura,T Tatsuma, N Oyama. Anal Chem 65:1925, 1993. H Inaba, M Iwaku,T Tatsuma, N Oyama. J Electroanal Chem 71:387, 1995. C Frubose, K Doblhofer, DM Soares. Ber Bunseges Phys Chem 97:475, 1993. DM Soares, W Kautek, C Frubose, K Doblhofer. Ber Bursenges Phys Chem 98:219, 1994. D Johannsmann, K Mathauer, G Werner,W Knoll. Phys Rev B 46:7808, 1992. D Johannsmann, J Gru«ner, J Wesser, K Matlhauer, G Wagner, W Knoll. Thin Solid Films 210=211:662, 1992. EJ Calvo, C Danilowicz, R Etchenique. J Chem Soc Faraday Trans 91:4083, 1995. RA Etchenique, EJ Calvo. Anal Chem 69:4833, 1997. EJ Calvo, R Etchenique, PN Bartlett, K Singhal, C Santamaria. Faraday Discuss 107:141, 1997. WP Mason, RN Thurston, AD Pierce. In: WP Mason, ed. Physical Acoustics. Vol. 2. Part A. New York: Academic Press, 1965. RW Cernosek, RW Martin, AR Hillman, HL Bandey. IEEE Trans Ultrasonics Ferro-Electronics Ferq Control 45:1399, 1998. VE Gransta¡, SJ Martin. J Appl Phys 44:209, 1994. R Lucklum, P Hauptmann. Electrochim Acta 45:3907, 2000. AR Hillman, A Jackson, SJ Martin. Anal Chem 73:540, 2001. M Rodahl, F Ho«o«k, A Krozer, P Brzezinski, B Kasemo. Rev Sci Instrum 66:3924, 1995. M Rodahl, F Ho«o«k, C Fredriksson, CA Keller, A Krozer, P Brezezinski, M Voinova, B Kasemo. Faraday Discuss 107:229, 1997. RA Etchenique, EJ Calvo. Electrochem Commun 1:167, 1999. E.J Calvo, R Etchenique. J Phys Chem B 103:8944, 1999. J Rickert,T Weiss,W Gopel. Sensors Actuators B 31:45, 1996. J Rickert, A Brecht,W Go«pel. Anal Chem 69:1441, 1997. AF Collings, F Caruso. Rep Prog Phys 60:1397, 1997. B Kasemo. Curr. Opin. Solid State Mater Sci 3:451, 1998. K Niikura, H Matsuno,YJ Okahata. J Am Chem Soc 120:8537, 1998. M Thompson, AC Arthur, GK Dhaliwal. Anal Chem 58:1206, 1986. AL Kipling, M Thompson. Anal Chem 62:1514, 1990. M Yang, M Thompson. Anal Chem 65:1158, 1993. M Yang, FL Chung, M Thompson. Anal Chem 65:3713, 1993.

380

Calvo et al.

158. 159. 160. 161. 162. 163. 164.

H Su, M Thompson. Biosensors Bioelectron 2:161, 1997. H Su, M Yang, MR Kallury, M Thompson. Analyst 118:309, 1993. H Su, MR Kallury, M Thompson. Anal Chem 66:769, 1994. H Su,Williams, M Thompson. Anal Chem 67:1010, 1995. H Su, S Chong, M Thompson. Langmuir 12:2247, 1996. H Su, M Thompson. Can J Chem 74:344, 1996. M Muratsugu, F Ohta, Y Miya, T Hosokawa, S Kurosawa, N Kamo, H Ikeda. Anal Chem 65:2933, 1993. F Ho«o«k, M Rodahl, P Brzezinski, B Kasemo. J Colliod Interface Sci 208:63, 1998. S Ghafouri, M Thompson. Langmuir 15:564, 1999. Y Lvov, K Ariga,T Kunitake. Chem Lett 2323, 1994. Y Lvov, K Ariga, I Ichinose, T Kunitake. J Chem Soc Chem Commun 2312, 1995. Y Lvov, K Ariga, I Ichinose,T Kunitake. Thin Solid Films 284:797, 1996. Y Lvov, K Ariga, I Ichinose, T Kunitake. J Chem Soc Chem Commun 2313, 1995. F Caruso, HA Rinia, DN Furlong. Langmuir 12:2145, 1996. F Caruso, K Niikura, DN Furlong,Y Okahata. Langmuir 13:3422, 1997. AF Collings, F Caruso. Rep Prog Phys 60:1397, 1997. J Anzai,Y Kobayashi, N Nakamura, M Nishimura,T Hoshi. Langmuir 15:221, 1999. J Anzai,T Hoshi, N Nakamura. Langmuir 16:6306, 2000. EJ Calvo, E Forzani, M Otero. Anal. Chem. 74:3281, 2002. JD Ferry. In: Viscoelastic Properties of Polymers. 3rd ed., New York: Wiley, 1980. EJ Calvo, E Forzani, M Otero. J. Electroanal. Chem, Langmuir, 18:4020, 2002. E Forzani, M Otero, MA Perez, M Lopex Teijelo, EJ Calvo. Submitted 2001. RMA Azzam, NH Bashara. In: Ellipsometry and Polarized Light. 1st ed. Amsterdam: North-Holland, 1977. S Gottesfeld,Y-T Kim, A Redondo. In: Rubinstein, ed. Recent Applications of Ellipsometry and Spectroellipsometry in Electrochemical Systems; Physcial Electrochemistry: Principles, Methods and Applications. New York: Marcel Dekker, 1995. Chap 9. H Arwin. Thin Solid Films 377^378:48, 2000. SW. Keller, H-N Kim,TE Mallouk. J Am Chem Soc 116:8817, 1994. V pardo-Yissar, E Katz, O Lioubashevsky, I Willner. Langmuir 17:1110, 2001. JJ Harris, ML Bruening. Langmuir 16:2006, 2000. D Li, M Lu«tt, MR Fitzsimmons, R Synowicki, ME Hawley, GW Brown. J Am Chem Soc 120:8797, 1998. JG Franchina, WM Lackowski, DL Dermody, RM Crooks, DE Bergbreiter, K Sirkar, RJ Russll, MV Pishko. Anal Chem 71:3133, 1999. K Spaeth, A Brecht, G Gauglitz. J Colloid Interface Sci 196:128, 1997. K Sirkar, A Revzin, MV Pishko. Anal Chem 72:2930, 2000. F Tiberg, M Landgren. Langmuir 9:927, 1993.

165. 166. 167. 168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180. 181.

182. 183. 184. 185. 186. 187. 188. 189. 190.

8 Nano-Sized Thin Films for Enzyme Reactors Katsuhiko Ariga Japan Science and Technology Corporation (JST), Tokyo, Japan

Yoshihiro Sasaki and Jun-ichi Kikuchi Nara Institute of Science and Technology (NAIST), Nara, Japan

1

INTRODUCTION

Nobody would deny the huge contribution of computers to current civilization. The computer is only a small box that can do complicated calculations and information processing within a short time, but it is not a ‘‘magic box’’or a‘‘black box.’’ Sophisticated design of electrical circuits allows the computer to process information logically. In order to produce wiser and lighter computers, a large number of circuits are intended to be integrated in an incredibly small area by microfabrication techniques. The ¢nal goal of this approach is to develop an information-processing device having structural precision at atomic or molecular levels. In this regard, nanotechnology, which aims to establish nanometer-size fabrication, has rapidly attracted attention [1,2]. Methodology for the nanometer-size fabrication can be classi¢ed into two categories, i.e., a top-down approach and a bottom-up approach. The top-down approach is based on techniques to convert a bulk material into precisely fabricated pieces. Sophisticated techniques such as photolithography, laser manipulation, and cantilever technique, have been developed to provide ultra¢ne structures. However, these techniques do not 381

382

Ariga et al.

usually supply molecular resolution, because the fabricated size is limited by the wavelength of light and the radius of curvature of the tip on the cantilever. In contrast, the bottom-up approach based on self-assembling processes has not been well developed but has the potential to achieve molecularly precise fabrication. A properly controlled self-assembly of functional molecules would be a key technique for fabrication with molecular precision. Supramolecular chemistry plays an important role in the bottom-up approach. Speci¢c recognition and the resulting interaction sometimes provide useful information conversion. Therefore, the accumulation of supramolecular approaches would ¢nally provide ultrasmall computer-like devices with molecular-level structural precision. One of the most successful results in this approach can be seen in biological systems (Fig. 1) [3,4]. Ultra¢ne energy converters, information processors, and machines are working in our body.They all consist of functional molecules through speci¢c molecular recognition. Therefore, we have to learn biological systems and mimic them to develop molecular devices by the bottom-up approach. One of the most pronounced di¡erences between man-made machines and biological systems exists in how to transmit signals from one site to another site. In an arti¢cial system, electrons play a major role in signal transmission and photons will make an important

FIG. 1 Cooperative functions by proteins immobilized on a lipid bilayer membrane. This is a model for nano-sized enzyme reactors.

Nano-Sized Thin Films for Enzyme Reactors

383

contribution in future devices. In contrast, signals are transmitted through chemical reaction and interaction between chemicals in biological processes. De¢ned chemicals can transmit signals in an incredibly speci¢c way, and the signals are frequently modulated through conversion of the chemicals. This mechanism creates huge numbers of diversities in signal transmission modes. Precision in the chemical conversion is indispensable for error-free signal transmission and information conversion in biological systems. Nature relegates this important role to enzymes which catalyze chemical conversions with high speci¢city and e⁄ciency. In biological systems, large numbers and kinds of enzymes often work sequentially and/or cooperatively to achieve highly sophisticated functions such as energy conversion and electron transport. Speci¢ed spatial organization of enzymes is essential to perform these functions with excellent e⁄ciency. Creation of arti¢cial enzyme reactors has been paid much attention in biorelated chemistry and in biotechnology. Development of methodologies for arti¢cial enzyme organization is one of the most important targets in current protein engineering [5^10]. It is indispensable for nano-sized enzyme reactors which are sometimes developed as molecular devices (Fig. 1). Immobilization of enzymes in ultrathin ¢lms is a promising approach to produce ultrahigh-density enzyme organizations (Fig. 2). Immobilization of enzymes has been conducted by a variety of techniques such as solvent casting, a sol-gel methods, and cross-linking [11]. However, controlled spatial arrangements of the enzyme molecules cannot be achieved by these methods. A self-assembled monolayer (SAM) technique can provide a covalently immobilized monolayer of functional molecules including enzymes, but the organization of the multiple kinds of enzymes is usually di⁄cult by the SAM method. The Langmuir-Blodgett (LB) method provides ordered molecular layers and has been applied to protein immobilization. Direct spreading of protein molecules on water frequently causes loss of protein activity through dissolution and surface denaturation [12]. Protein adsorption on lipid monolayers and preparation of a monolayer from lipid ^ enzyme complexes may avoid surface denaturation [13^15]. Okahata et al. reported that a glucose oxidase/lipid monolayer deposited on a Pt electrode acted as a glucose-sensing ultrathin membrane with quick response [16,17]. Another disadvantage of the LB ¢lm is slow di¡usion of substrates through the ¢lm due to dense packing of the lipid molecules. Recently, alternate layer-by-layer adsorption based on stepwise charge reversal has attracted much attention as a novel means of molecular ¢lm preparation [18^20]. This method was pioneered by Iler [21] and extended recently by Decher et al. [18]. According to the latter studies,

384

FIG. 2

Ariga et al.

Typical methods for protein immobilization as a thin film.

alternating adsorption of anionic and cationic polyions (polyelectrolytes) leads to formation of regular multilayer assemblies. Linear relations were found between ¢lm thickness and the number of adsorbed layers, and the ¢lms were remarkably uniform [22]. This technique has been applied to conventional and functional polymers [22^44], biopolymers [45^73], inorganic materials [74^88], and small molecules [89^98]. We conducted an extensive investigation of polyion ^ protein alternate assemblies and found that molecular ¢lms of most water-soluble proteins could be assembled if the proteins were properly charged under the adsorption conditions [45^ 49]. Because molecular ¢lms are assembled from protein solution under mild conditions, denaturation during immobilization would be minimal. Recent experiments demonstrated that the polyion layers are more permeable to small molecules than lipid ¢lms are [54]. The abovementioned characteristics encouraged us to prepare ultrathin-¢lm-type enzyme reactors through the alternate layer-by-layer adsorption between polyions and proteins.

Nano-Sized Thin Films for Enzyme Reactors

385

In the ¢rst part of this chapter, preparation of an enzyme reactor by the alternate layer-by-layer adsorption is mentioned [50^54]. Initially, we explain the outline and practical technique of this method. We next show glucose oxidation catalyzed by glucose oxidase (GOD) in alternately assembled ¢lms. The practical usefulness of the nano-sized enzyme reactors depends on the stability of the enzymes in the ¢lm. Increased storage stability and thermostability of GOD in the ¢lms are also described [53]. Improvement of the enzymatic activity was also achieved by a premixing alternate assembly [47,53]. As an example of a multienzyme reactor, layerby-layer ¢lms containing glucoamylase (GA) and GOD on an ultra¢lter were prepared, and sequential reaction of hydrolysis of starch and oxidation of the resulting glucose was demonstrated. The studies described in the ¢rst part were performed in the Fundamental Design Group of the Supermolecules Project, Japan Science and Technology Corp. (JST). In the latter part of this chapter, we describe another type of nanosized enzyme reactor. A lipid bilayer vesicle is one of the potential media for immobilizing enzymes (Fig. 2). Although spatial organization of enzymes is di⁄cult on the bilayer membrane, incorporation of other components is relatively easy in this medium. In the latter section, we present preliminary results on switchable enzyme reactors [99^103]. Lactate dehydrogenase (LDH) and arti¢cial receptors were co-immobilized on the lipid bilayer membrane. The arti¢cial receptor can trap an enzyme inhibitor upon addition of a chemical signal and a photostimulus. As a result, the enzymatic activity can be controlled by the addition of these signals. The presented results are still preliminary, but it will open a way to developing a novel type of nano-sized enzyme reactors and enzyme-based molecular devices. 2 2.1

MULTIENZYME REACTOR Preparation of Thin-Film Reactors by Alternate Layer-by-Layer Adsorption

2.1.1 Characteristics and Application Areas of Alternate Layer-by-Layer Adsorption The basic procedure of this method is illustrated in Fig. 3. The fundamental driving forces of this technique are charge neutralization and resaturation upon adsorption of charged materials on oppositely charged surfaces. These processes result in an alternative change in the surface charge and lead to a continuous assembly between positively and negatively charged materials. The adsorption amount is spontaneously regulated by the charge repulsion in an excessively adsorbed layer.

386

Ariga et al.

FIG. 3 Basic procedure and assembling mechanism for alternate layer-by-layer adsorption. Assembly between a cationic polyion and an anionic protein on a negatively charged support is shown as an example.

The concept of the layer-by-layer alternate adsorption was ¢rst proposed for charged colloidal particles in 1966 by Iler [21]. Multilayer assembly of charged proteins by the alternate adsorption was proposed by Fromherz who, however, did not experimentally demonstrate this idea [13]. Decher and coworkers introduced the concept of the alternate adsorption to a ¢lm assembly a linear polyions or bola-amphiphiles [94^96]. The alternate adsorption of Zr4þ ions and diphosphonic acid was developed by Mallouk et al. [104]. Kunitake et al. proved the charge neutralization and resaturation process by a surface force measurement [105].

Nano-Sized Thin Films for Enzyme Reactors

387

The major advantages of the layer-by-layer alternate adsorption are its simplicity and versatility. One of its most prominent advantages is the simplicity of the assembling procedure. In the ¢rst step, a solid support with a charged surface is immersed in a solution containing oppositely charged polyions. Relatively high concentrations of the polyions are used in this technique; therefore, excess adsorption of the polyion occurs, leading to e¡ective reversal of the surface charge through a number of ionic groups remaining at the ¢lm surface. After the solid support is rinsed in pure water, the support is immersed in a solution of an oppositely charged polyion. Adsorption of the latter polyion results in reversal of the surface charge again. Multilayer assemblies are obtained by alternately repeating both steps. The only instruments required in this procedure are beakers and tweezers. Although the procedure is quite simple, the ¢lm thickness is easily controlled by the number of dipping cycles, and the ¢lms were remarkably uniform [22]. Shiratori et al. reported an automatic dipping machine for this procedure [106]. The machine performs the polycation/polyanion assembly based on a ¢lm-mass control mechanism. Another advantage of this technique is versatility of the applicable materials. Because the driving force of the assembly is a simple electrostatic interaction, various charged materials can be used in this method. Assemblies of conventional functional polymers [22^44], biopolymers [45^73], and inorganic materials [74^88] have been demonstrated.We demonstrated that a large number of water-soluble proteins were assembled in combination with oppositely charged polyions [45^49]. Sequential enzymatic reaction by preparation of anisotropic enzyme layers and precise control of distances between active layers is explained in this chapter [50^54]. Shen and coworkers showed that enzymes and bolaamphiphile were alternately assembled on a solid support without losing the enzyme activity [70^73]. This technique is applicable to small molecules which spontaneously form aggregate structures. Decher et al. reported the assembly of a bolaamphiphile monolayer with a polyion [94^96]. Kunitake et al. reported the alternate assembly of bilayer-forming lipids with polyions [97]. This technique can be combined with the Langmuir-Blodgett (LB) technique to provide an LB ¢lm ^ polyion alternate assembly [98]. Self-assembly of the lipid molecules probably leads to stabilization of the adsorbed layer. Multicharged dye molecules were also assembled with polyions due to stacking of the dye molecules [89^93]. Alternate assemblies based on an interaction other than an electrostatic interaction were demonstrated. For example, a biological speci¢c interaction can be applied to this technique. Concanavaline A was alternately assembled with polysaccharides [56]. Streptavidine was successfully

388

Ariga et al.

assembled with biotinated poly ( L -lysine) [69]. Poly(vinylpyrrolidone), poly(vinyl alcohol), poly(acrylamide), and poly(ethylene oxide) were alternately assembled through hydrogen bonding [34]. A polymer carrying a dinitrophenyl group was assembled with a polymer with a carbazole side group through formation of a charge transfer complex [107]. Reactivity of fullerenes toward amines allows alternate assembly between them through a covalent linkage [108]. Recently, Serizawa, Akashi, and coworkers demonstrated that isotactice poly(methyl methacrylate) and syndiotactic poly(methyl methacrylate) were alternatively assembled through a stepwise stereocomplex formation [109,110]. Versatility of the alternate layer-by-layer adsorption technique also appears in the selection of solid supports [52,72]. Flat plates are not always necessary in this method unlike other thin ¢lm preparations. Caruso, Mo«hwald, Lvov, and coworkers prepared an alternately assembled ¢lm on colloidal particles and crystals as a template [111^120]. A hollow capsule having the alternate assembled ¢lms as a skin was prepared by destroying the colloidal particles after the assembling process. 2.1.2 Practical Procedure of Alternate Layer-by-Layer Adsorption Materials used in this study are summarized in Fig. 4. Glucose oxidase Aspergillus niger (GOD) (EC:1.1.3.4), peroxidase from horseradish (POD) (EC:1.11.1.7), glucoamylase (GA) (EC:3.2.1.23), poly(ethylenimine) (PEI, MW 70,000), and poly(styrenesulfonate) (PSS, MW 70,000) were used for assembly.Water used in this study has a speci¢c resistance of ca. 18 MO cm. A GOD ¢lm preparation on a quartz plate is described below as a typical example of enzyme assembly. Step 1.

Step 2.

Step 3. Step 4.

Aqueous solutions of polyions (PEI and PSS) and GOD were prepared. The pH of aqueous GOD solution (2 mg/ mL) was adjusted to pH 6.5 with hydrochloric acid. The solution pH should be set for the enzyme to have a su⁄cient amount of surface charges without causing denaturation. GOD is negatively charged at this pH because its isoelectric point is 4.2. A negatively charged quartz slide was immersed into aqueous PEI (1.5 mg/mL) for 20 min and then rinsed with water for 2 min. The slide was next immersed into aqueous PSS (3 mg/mL) for 20 min with a 2-min intermediate water washing. Steps 2 and 3 were repeated eight or nine times in order to prepare precursor layers. The precursor layer provides a su⁄cient amount of surface charges which allow further

Nano-Sized Thin Films for Enzyme Reactors

389

FIG. 4 Materials used for enzymes reactors in this study. The advantages of the alternate layer-by-layer adsorption for protein immobilization are also listed.

Step 5.

protein assembly.The outermost layer becomes ‘‘negative’’or ‘‘positive’’ by choosing either PSS or PEI as the lastly adsorbed layer. For assembly of negatively charged GOD, the outermost layer must be cationic PEI. The slide having the positively charged ¢lms was immersed in a negatively charged GOD solution for 20 min with a 2-min intermediate water washing.

390

Ariga et al.

Step 6. The ¢lm was then immersed in PEI solution in a similar manner. Step 7. Steps 5 and 6 were repeated until the desired numbers of the alternate layers were formed. The charge at the outermost layer can be converted to negative by adsorption of anionic PSS. In principle, construction of ¢lms with any kind of positively charged and negatively charged enzymes is possible. This methodology has the following advantages in enzyme immobilization in nano-sized thin ¢lms: (1) it is applicable to most water-soluble proteins, (2) the layer organization such as the number of layers and arrangement of di¡erent layers is £exibly designed, and (3) the assembly conditions are mild and denaturation can be avoided. Therefore, multienzyme ¢lms with a predetermined sequence of enzyme layers can be prepared. 2.1.3 QCM Evaluation of Alternate Layer-by-Layer Adsorption The alternate layer-by-layer adsorption was quantitatively evaluated using a quartz crystal microbalance (QCM) technique. Because the resonant frequency of the QCM changed sensitively due to the mass adsorption on its electrodes, the QCM is widely known as a useful device for the detection of small-quantity adsorptions in both the air phase and in the liquid phase [121^ 133]. The frequency shift (DF) is given as below for the 9-MHz AT-cut QCM [134]: DF ðHzÞ ¼ 18:3 M =A ðng=m2 Þ

ð1Þ

where M and A represent the mass increase and the area of the electrode on the QCM plate, respectively. A frequency change of 1 Hz corresponds to ca. 0.9 ng of adsorption over the entire electrode. This remarkable sensitivity is useful for evaluation of the growth of molecular ¢lms. As shown in our previous report [45], comparing scanning electron microscopic (SEM) images with QCM frequency changes provides the following relation between ¢lm thickness (d ) and frequency change (DF). 

dðAÞ ¼ 0:16 DF ðHzÞ

ð2Þ

As the ¢rst example, alternate assemblies of PEI-GOD and POD-PSS ¢lms were investigated by the QCM method. PEI-GOD ¢lms were assembled on the precursor ¢lm (four layers of PEI-PSS ¢lm). Film growth was constant except for the ¢rst few steps and the average frequency change is ca. 2000 Hz per cycle for adsorption of a single GOD-PEI layer. This assembly process was repeated and stopped at the PEI adsorption step of the 10th cycle. The surface charge was converted to negative by adsorption of PSS.

Nano-Sized Thin Films for Enzyme Reactors

391

A POD-PSS assembly was carried out on the former ¢lm. Constant ¢lm growth was repeatedly observed with an average frequency change of ca. 180 Hz per cycle.The thickness of the POD and GOD layers were estimated from the frequency shifts to be 35 and 344 —, respectively, upon subtraction of the thickness of the PSS or PEI layer. The diameters of POD (MW 40,000) and GOD (MW 186,000) were estimated to be 35 and 85 —, respectively, by assuming a spherical shape.The frequency change for POD corresponds to a single monolayer adsorption, but that for GOD is equal to the simultaneous adsorption of 3^4 layers. GOD molecules probably form aggregates under this assembly condition. 2.1.4 Evaluation of Catalytic Ability of GOD-Based Reactor The PEI-GOD ¢lms were assembled on a precursor ¢lm on a quartz plate and an enzymatic reaction with the GOD ¢lms was examined. The coupled enzymatic reaction of GOD and POD illustrated in Fig. 5A was used for evaluation of the enzymatic activity of the assembled ¢lms. GOD converts D-glucose and O2 to D-glucose-d-lactone and hydrogen peroxide (H2O2). In the next step, POD oxidizes DA67 (indicator) using H2O2 as oxidant. Because oxidized DA67 has a large molar extinction coe⁄cient of 90000 [135], the reaction can be easily followed by monitoring the absorbance at 665 nm. The activity of GOD immobilized in the ¢lms was assayed using Dglucose as substrate with the aid of dissolved POD and DA67. The ¢lm was immersed in the upper space of a cuvette containing 3.0 mL of substrate solution (25 C, pH 7.0 with 0.1 M PIPES, immersion area 5  5 mm2  2 faces). The separation between the ¢lm and the light beam was approximately 2.0 cm; thus the light path was not interrupted. The solution was stirred continuously during each experiment. In order to examine the features of the enzymatic reaction, we added each component step-by-step to the solution. Figure 5B shows the time course of the absorbance change at 665 nm. An aqueous mixture of POD (4 mg/L) and DA67 (100 mM) was incubated at 25 C (t ¼ 0 to A). The absorbance was not increased during this period, indicating that spontaneous air oxidation of DA67 can be ignored. A one-layer GOD ¢lm was immersed into the solution (A). Changes in absorbance were not observed, because the substrate (glucose) was absent at this stage. After removal of the ¢lm from the solution (B), glucose (56 mM) was added (C). Any absorbance change was detected, indicating that GOD did not leak from the ¢lm. The GOD ¢lm was again immersed into an aqueous solution that contained glucose, POD, and DA67 (D). The absorbance at 665 nm increased linearly after the ¢lm immersion, because all components required for the whole reaction were supplied. A dark blue color was observed originally in the vicinity of the ¢lm and dispersed into the entire solution. This indicates that

392

Ariga et al.

FIG. 5 (A) Sequential enzymatic reaction for investigation of GOD activity in the assembled film. (B) Absorbance changes at 665 nm upon DA67 oxidation are plotted to evaluate the GOD activity. Reagents for the reaction were added stepwise.

Nano-Sized Thin Films for Enzyme Reactors

393

GOD immobilized in the ¢lm is active. When the GOD ¢lm was removed from the solution (E), the increase in the absorbance ceased. An additional immersion/removal cycle (F and G) reproducibly caused a change in the absorbance (experimental error < 5%). When a two-layer GOD ¢lm was immersed in the reaction mixture (H), a greater absorbance increase was observed. The apparent reaction rate increased as the number of layers increased, i.e., the initial rate of the glucose oxidation was calculated to be 2 and 7 108 M/s for one- and two-layer GOD ¢lms, respectively. It is clear that GOD molecules immobilized in the inner layer also contributed to glucose oxidation. 2.2

Improved Stability of Reactor

2.2.1 Storage Stability The long-term stability of an immobilized enzyme is one of the most important requirements in practical applications. Precipitation of oxidized DA67 on the ¢lm sometimes caused loss of the apparent activity, when an identical plate of immobilized GOD was repeatedly used more than four or ¢ve times. Therefore, storage stability was measured using separate samples that had been stored for di¡erent periods of time. The one-layer GOD ¢lms were prepared on quartz plates (5  5 mm2  2 faces), and they were stored in water at 25 C, stored in 0.1 M PIPES bu¡er (pH 7) at 4 C, and allowed to stand in air at 4 C. Each sample was washed with water, and its activity was measured by the standard procedure after given periods of time. The ¢lm samples stored in water at 25 C showed drastic decreases in activity, and approximately 70% of the activity was lost after 4 weeks. Although the reason for this deterioration is unclear at present,we speculate that bacterial growth is the cause of the deterioration. In contrast, the ¢lms kept in the bu¡er at 4 C, did not show a signi¢cant decrease in enzymatic activity over 14 weeks. A 10% decrease in the activity was observed in the ¢rst week for the ¢lms kept in air at 4 C, but the activity was maintained during the following 13 weeks. The initial activity loss was probably due to drying of the ¢lm. In the alternate assembly, GOD is immobilized through electrostatic interaction, and the activity loss from structural strain and deformation appears absent. 2.2.2 Thermostability Most enzymes are denaturated and lose their activity at high temperature. Some studies reported that immobilization suppressed the structural deformations of enzymes with enhancement of their stabilities [136]. Improvement of thermostability is expected for GOD immobilized in the alternately assembled ¢lm. The GOD-PEI ¢lm was immersed in water at 50^60 C for

394

Ariga et al.

various time periods, and the QCM frequency was measured after N2 drying. Weight loss was not detected even after 2-h immersion, indicating that GOD is not released into water by such treatment. Next, we examined the thermostability of enzymatic activity. A onelayer GOD ¢lm on a quartz plate (5  5 mm2  2 faces) was incubated in 3 mL of water at a given temperature for 10 min. The ¢rst measurement of the enzymatic activity was performed at 25 C immediately after the incubation period. After keeping the ¢lm in air room temperature for 30^40 min, the second measurement was carried out. Aqueous GOD (1 mg/mL in 0.5 mL of 0.1 M PIPES bu¡er at pH 7) was kept at a given temperature for 10 min, and the activity measurement was similarly performed. Aqueous GOD lost activity on incubation even at 30^40 C (Fig. 6a), and it became inactive at 50 C. The activity loss was partially recovered by returning the solution to room temperature (Fig. 6b). This activity recovery decreased with the incubation temperature and totally disappeared at 70 C. The enzymes in solution would lose their activity by both

FIG. 6 Thermostability of aqueous GOD and GOD assembled in the film: (a) relative activity of aqueous GOD immediately after incubation; (b) relative activity of aqueous GOD 30 min after incubation; (c) relative activity of GOD in the film immediately after incubation; (d) relative activity of GOD in the film 30 min after incubation. The activity relative to that at 22 C is plotted as a function of incubation temperature.

Nano-Sized Thin Films for Enzyme Reactors

395

reversible unfolding and irreversible changes [137,138]. The recovered activity in our case may be related to partial unfolding. The thermostability of GOD assembled with PEI was remarkably improved (Fig. 6c). A signi¢cant decrease in the activity was not detected even after incubation at 50 C. Interestingly, recovery of the reduced activity was not observed in the ¢lm sample at any incubation temperature (Fig. 6d). The improved enzymatic activity and the absence of the activity recovery could be attributed to suppression of conformational mobility of GOD upon complex formation with surrounding polymer chains. 2.2.3 Improvement of Enzyme Activity by Premixing Method As demonstrated by the QCM experiment, GOD was adsorbed as multilayers in one step and it would exist as aggregates in the ¢lm. This aggregation would a¡ect the GOD activity.When we used high GOD concentrations (0.5^2 mg/mL) for assembly with PEI, the frequency change upon GOD adsorption corresponded to 3^4 GOD layers. The frequency change was comparable to monolayer adsorption (670 Hz) when a low GOD concentration (0.05 mg/mL) was utilized. However, the relative activity of these ¢lms was almost identical. Therefore, a simple decrease in the GOD concentration does not improve the enzymatic activity. The aggregation state of GOD would not be changed by simple dilution. In order to molecularly disperse GOD in the ¢lm, we prepared GODcontaining ¢lms by a premixing method. Anionic GOD was mixed with a large excess of cationic PEI to form a positively charged protein ^ polyion complex, and the formed complex was then assembled alternately with anionic PSS (Fig. 7). The assembling behavior and GOD reactivity of the ¢lms were investigated. The QCM frequency changes due to adsorption of the premixed complexes were signi¢cantly larger than those of the PEI-PSS standard ¢lm. Increased frequency changes for the premixed complexes originate from the incorporation of GOD. The relative activity of the premixed ¢lms is calculated based on the assumption that ratios of GOD and polyions are unchanged in solution and in the assembled ¢lms.The ¢lms with GOD-PEI premixed layers showed signi¢cant enhancement in activity ( > 67 times). Better molecular dispersion of GOD in the ¢lms prepared by the premixing method probably improves the enzymatic activity. 2.3

Sequential Catalysis by Multienzyme Reactor

2.3.1 Preparation of GA-GOD Reactor on an Ultrafilter Properties of a nano-sized reactor of a single-component enzyme are described in the previous sections. In this section, we describe multienzyme

396

Ariga et al.

FIG. 7 Premixing method of alternate layer-by-layer adsorption. An anionic enzyme is first mixed with an excess amount of cationic polyion, and the resulting complex is then alternately assembled with a polyanion.

reactors prepared on an ultra¢lter through which a substrate solution was passed.The reaction e⁄ciency would be determined by the relation between the direction of substrate £ow and the arrangement of the enzyme layers. We chose coupled catalyses by GA and GOD as a model reactor. The substrate of the coupled catalyses is starch, which has di⁄culty di¡using through the ¢lm because of its high molecular weight. The assembling behavior of GA and GOD molecules in the same ¢lm was ¢rst examined by the QCM method. As described above, constant ¢lm growth was observed for the assembly of the PEI-GOD layers with a frequency change of ca. 2000 Hz per cycle. GA was then alternatively assembled with PEI on top of the PEI-GOD multilayer ¢lm. Constant ¢lm growth was again observed. The frequency decrease upon PEI-GA adsorption was ca. 145 Hz per cycle, while the frequency shift for PEI was only 40 Hz. The reversed protein assembly was also carried out, i.e., PEI-GA layers were ¢rst assembled and then PEI-GOD layers were prepared on the PEI-GA layer. Both of these two processes showed constant ¢lm growth, and their frequency changes were the same as those in the oppositely sequenced ¢lm. Therefore, the assembly processes of PEI-GOD and PEI-GA are independent of the sequence of assembly.

Nano-Sized Thin Films for Enzyme Reactors

397

Next, multienzyme ¢lms of GOD and GA were prepared on an ultra¢ltration membrane (Molcut II LC, limiting molecular weight 5000) under the conditions derived from the QCM experiment. The ¢lter membrane was ¢rst immersed in aqueous PEI solution and then immersed in aqueous PSS solution followed by washing. These procedures were repeated four times to produce precursor layers. Desired enzyme ¢lms were similarly prepared on the precursor layer. Film formation on the ¢lter was con¢rmed from elemental analysis by X-ray photoelectron spectroscopy (XPS). Only carbon and oxygen were detected on a bare ¢lter membrane which was made from cellulose. Elemental ratios obtained by XPS were C/N/O/S ¼ 70.9/7.3/19.1/2.8 and 68.1/ 11.3/20.2/0.3 for [¢lter þ (PEI/PSS)4 þ PEI] and [¢lter þ (PEI/PSS)4 þ (PEI/GA)2 þ (PEI/GOD)2 þ PEI], respectively. The calculated elemental ratio for PEI/PSS ¼ 1/1 (C10NO3S) is C/N/O/S ¼ 66.7/6.7/20.0/6.7. This is consistent with that of the precursor ¢lm if we take account of higher nitrogen and carbon values due to the outermost layer of PEI.The elemental ratio estimated for GOD from its amino acid sequence is C/N/O/S ¼ 62.9/17.2/ 19.6/0.3, which is fairly close to the latter experimental value of the GODsurface ¢lm except for the nitrogen content. These results support the progress of the alternate assembly even on an ultra¢lter. 2.3.2 Sequential Catalysis of GA-GOD Reactor on an Ultrafilter The experimental setup for the enzymatic reaction is depicted in Fig. 8. Hydrolysis of the glycoside bond in starch by GA produces glucose.Glucose is converted to gluconolactone by GOD with H2O2 as a co-product. An aqueous solution of water-soluble starch (2.0 wt%) in 0.1 M PIPES bu¡er (pH 7.0, 1 mL) was placed on enzyme-immobilized ultra¢lters in the upper cup. Filtration was started by applying pressure (3^4  105 Pa) to the upper cups with a syringe. The £ow rate was ca. 1 mL/h. The ¢ltrate (0.1 mL) was added to a mixed solution (0.1 M PIPES pH 7.0) of POD (5 mg/mL) and DA67 (1 mg/ mL), and the change in absorbance at 665 nm was monitored. This assay provides the H2O2 concentration in the ¢ltrate.The combined concentration of H2O2 and glucose was determined from the absorbance change at 665 nm when 0.1 mL of ¢ltrate was added to an aqueous mixture of GOD (5 mg/mL), POD (10 mg/mL), and DA67 (10 mg/mL) in PIPES (0.1 M, pH 7.0). The concentration of unreacted starch was assayed by the iodostarch reaction. Unreacted starch was not detected in the ¢ltrate at all, although the reaction conversion calculated from the glucose and H2O2 formed clearly indicated the presence of unreacted starch. Starch cannot pass through the ¢lter, allowing separation of substrate and products without additional procedures. One of the most pronounced advantages of the alternate layer-by-layer adsorption lies in the rich variety of the ¢lm organization. In order to

FIG. 8

Structure and sequential enzymatic reaction of multienzyme reactor composed of GA and GOD.

398 Ariga et al.

Nano-Sized Thin Films for Enzyme Reactors

399

investigate the e¡ect of ¢lm organization on the reaction e⁄ciency, the following enzyme ¢lms were prepared: Film A: Film B: Film C: Film D: Film E:

¢lter þ (PEI/PSS)4 þ (PEI/GOD)2 þ (PEI/PSS)2 þ (PEI/GA)2 þ PEI ¢lter þ (PEI/PSS)4 þ (PEI/GOD)2 þ (PEI/PSS)10 þ (PEI/GA)2 þ PEI ¢lter þ (PEI/PSS)4 þ (PEI/GA)2 þ (PEI/PSS)2 þ (PEI/ GOD)2 þ PEI ¢lter þ (PEI/PSS)4 þ (PEI/GA)2 þ (PEI/PSS)10 þ (PEI/GOD)2 þ PEI ¢lter þ (PEI/PSS) 4 þ (PEI/MIX)2 þ PEI

where MIX represents absorbed layers from an aqueous solution of an equimolar mixture of GOD and GA. Film A and ¢lm B have outer GA layers and inner GOD layers. In these ¢lms, the direction of reactant £ow and the sequence of enzyme layers are matched. In contrast, ¢lm C and ¢lm D have a reversed ordering of the enzyme layers. Comparison of the reaction e⁄ciency between these two kinds of ¢lms provides the e¡ect of the assembling sequence. Film A and ¢lm C have two PEI/PSS spacer layers between the GA and GOD layers, while 10 spacer layers of PEI/ PSS placed between the two enzyme layers in ¢lm B and ¢lm D. The reaction e⁄ciency of these ¢lms is related to the separation of the enzyme layers. The highest yield of the combined products (H2O2 and glucose) was obtained with ¢lm B, and ¢lm A showed the second highest yield. Because the sequential reaction requires that reaction with GA proceeds before the oxidation of glucose catalyzed by GOD, the order of the enzyme layers in these ¢lms agrees with the order of the sequential enzymatic reactions. In contrast,when the disposition of the two enzyme layers was reversed in ¢lms C and D, starch has to penetrate into the inner GA layer, and the resulting glucose has to di¡use to the outer GOD layer against the substrate £ow. A ¢lm structure unfavorable for the sequential reaction led to lower yields compared with those of the former ¢lms. These results indicate that the sequential enzymatic reaction is signi¢cantly a¡ected by the order of assembling of the enzyme layers. It is curious that ¢lm B provides higher yields of glucose and H2O2 than ¢lm A. This result suggests the important role of the spacer layer that separates the GA layer and GOD layer. This cannot be explained by the ease of substrate di¡usion. A plausible mechanism is related to the inhibition of GA activity by gluconolactone. A mix-type inhibition of GA by gluconolactone was reported [139,140]. Smaller molecules such as gluconolactone can di¡use backward and would reduce the activity of GA. The extent of this

400

Ariga et al.

inhibition must be larger in ¢lm A than in ¢lm B because of the thicker spacer layer of the latter. Film E has two mixed layers of GA and GOD and is used as control. This ¢lm showed the lowest yield. The reason for this result is not clear. If adsorption of GOD preferentially occurred from a mixed solution of GA and GOD, the lack of GA in the ¢lm would signi¢cantly suppress the reaction e⁄ciency. Coexistence of GA and GOD might result in the lowest activity due to the inhibition of GA by gluconolactone produced by GOD. These are no spacer layers between the two kinds of the enzymes, and considerable inhibition of the sequential reaction may result. In this section,we demonstrated that the multienzyme reactor of GOD and GA on an ultra¢lter catalyzed the sequential conversion of starch. The reaction e⁄ciency was a¡ected by the change in the ¢lm organization that can be readily obtained by the alternate layer-by-layer adsorption technique. We summarize the advantages of this method below. Thickness of the ¢lm can be controlled with large layering variation. Awide range of water-soluble enzymes can be assembled as a multilayer ¢lm without serious deactivation, because the preparative process is quite easy and mild. Another advantage of this method is the wide choice of solid supports. Charged surfaces of any kind can be used for this method. One can readily introduce charges by suitable methods, even if a speci¢c surface is not charged. The alternate layerby-layer adsorption does not require £atness of the solid supports, unlike the LB technique. This is important in constructing enzyme reactors on microporous supports, because e⁄cient reactors probably need large areas per unit mass. The alternate layer-by-layer adsorption is applicable to many charged materials other than proteins and conventional polyions. Polyion layers can be assembled in combination with redox proteins on electrode surfaces, leading to construction of electron-driven reactors [141]. Combined assembly of enzymes and semiconductor particles with nanometer-scale precision would also lead to development of a novel type of molecular reactor. 3

SWITCHABLE ENZYME REACTOR

In the previous sections, construction of multienzyme reactors is demonstrated. The alternate layer-by-layer adsorption method can immobilize enzymes in a layered structure with a desirable sequence. Therefore, molecular devices and machines which connect di¡erent enzyme functions would be constructed by this methodology. Addition of other functions to the enzyme reactors will open the way for further intelligent systems. For example, reactors which can be switched on and o¡ by an external signal would lead to controllable chemical transformation and information

Nano-Sized Thin Films for Enzyme Reactors

401

conversion. Such a switching mechanism can be commonly seen in biological systems, where many kinds of proteins cooperatively function and recognition of external stimuli triggers the speci¢c enzyme reactions. Connection of arti¢cially designed functions to naturally occurring enzymatic catalyses would provide a novel type of switchable reactor, because there are many kinds of synthetic molecules which can recognize a speci¢c guest and/ or are sensitive to physical stimuli such as light irradiation. Lipid assemblies provide a nice medium for immobilization of the arti¢cial molecules and enzymes. In the following sections, we demonstrate switchable enzyme reactors which connect receptor function and enzymatic catalysis on a lipid bilayer membrane [99^102]. We present only preliminary results, but this will open a way to developing a novel type of nano-sized switchable enzyme reactor. 3.1

Strategy for a Switchable Enzyme Reactor

3.1.1 Function of Naturally Occurring G-Protein System In order to develop a switchable enzyme reactor, we tried to mimic a signal transduction system in which an external signal controls an enzymatic activity. We especially paid attention to a G-protein-mediated signal transduction system [142^144]. Figure 9 brie£y summarizes a model of the Gprotein function. The G protein consists of three subunits (a, b, and g) with GTP(or GDP) and couples receptor function and enzyme (e¡ector, adenylate cyclase) activation. In its inactivated form, the G protein exists as a trimer with GDP bound to the a subunit (Fig. 9a). When a signal molecule (hormone) binds to a receptor, the G protein is activated through binding to the receptor ^ signal complex. In the activated form, the guanyl-nucleotidebinding site on the a subunit is altered, allowing GTP to bind in place of GDP (Fig. 9b). The binding of GTP is thought to dissociate the a subunit from the b and g subunits. The dissociated a subunit binds tightly to the enzyme (adenylate cyclase), which is activated to produce cyclic AMP (Fig. 9c). Within less than a minute, the a subunit hydrolyzes its bound GTP to GDP, causing the a subunit to dissociate from the enzyme (Fig. 9d). Reassociation of the a subunit and b, g subunits re-forms the inactivated form of the G protein. 3.1.2 Design of a Switchable Enzyme Reactor The above-mentioned G-protein function plays an important role in the signal transduction process. We were deeply interested in this mechanism, because the coupling mechanism of the receptor function and the enzyme activation is a good model of a switchable enzyme reactor. Our design of the switchable enzyme reactor is summarized in Fig. 10.We used a lipid bilayer

402

FIG. 9

Ariga et al.

Schematic model of G-protein-mediated signal transduction system.

vesicle as a support for the enzyme reactor. Arti¢cial receptors and enzymes are immobilized on the surface of the lipid bilayer. Because some metal ions can control enzyme activity, metal ions are potential candidates for a Gprotein mimic. What we have to do is to design an arti¢cial receptor which can trap and/or release the metal ions upon external signal application. Metal-ion binding is a common research target in host ^ guest chemistry, and a number of receptors for metal ions have been proposed in the history of supramolecular chemistry. The switching mechanism of the enzyme reactor is also proposed in Fig. 10. The enzyme (e¡ector) immobilized on the bilayer vesicle is inhibited by the metal ion in the initial state (OFFstate).When a suitable signal molecule is added to the system, a signal ^ receptor complex is formed through a speci¢c reaction. Because the signal ^ receptor complex has a higher a⁄nity for the metal ion than the enzyme, the metal ion is removed from the enzyme and it is activated (ONstate).This system is driven by the di¡erence in binding a⁄nity to the metal ion between the enzyme, the uncomplexed receptor, and the signal ^ receptor complex. The order of the a⁄nity for the metal ion must be an appropriate one (signal ^ receptor complex > enzymes > receptor). In order

Nano-Sized Thin Films for Enzyme Reactors

403

FIG. 10 Our strategy for a switchable enzyme reactor in which an artificial receptor and an enzyme are immobilized on a lipid bilayer vesicle and the functions of the receptor and enzyme are connected by a mediator.

to realize this design, we adopted an amphiphilic amine as a receptor and aldehyde compounds as a signal molecule (Fig. 11). Some aldehyde compounds easily form a Schi¡ ’s base with amines, and the Schi¡ ’s base has an incredibly high a⁄nity for metal ions.If we select the appropriate enzyme and metal ions, the desirable ion a⁄nity order (Schi¡ ’s base > enzyme > amine) can be achieved.We selected lactate dehydrogenase (LDH) and Cu2þ for this system. In the following section, we ¢rst examined the activity of the LDH immobilized on the lipid bilayer vesicle and then evaluated the interactions between the receptors,signal molecules, and the Cu2þ .The switchable ability of the enzyme reactor was ¢nally examined. 3.2

Enzyme Reactor Switched by Chemical Signal

3.2.1 Michaelis-Menten Analysis of Enzyme Immobilized on Lipid Bilayer Membrane In order to construct a nano-sized reactor, we used lipid assemblies. The reason we used the lipid assembly is based not only on the fact that a lipid assembly is biomembrane mimic but also on the necessity for stable

404

Ariga et al.

FIG. 11 Lipids, receptors, and signal molecules used for the switchable reactor developed in this study.

formation of the Schi¡ ’s base. The Schi¡ ’s base formation is usually disadvantageous in aqueous media. Therefore, a reaction ¢eld between the amine-type receptor and the aldehyde signal should be buried in hydrophobic media and located at the interfacial environment. Lipid assemblies

Nano-Sized Thin Films for Enzyme Reactors

405

are known to provide an environment suitable for this purpose. We used cationic lipid 1 as the vesicle-forming lipid. Because the LDH used (from pig heart, EC 1.1.1.27) has an isoelctric point at 5.2, the LDH is anionic at pH 7.0 and is strongly immobilized on the cationic vesicle through an electrostatic interaction [145,146]. Because micelles might provide a desirable environment, we ¢rst examined the cationic micelle of cetyltrimethylammonium bromide (CTAB) as a medium for enzyme immobilization. However, mixing of the CTAB micelle and the LDH in aqueous bu¡er caused serious activity loss of the LDH. As shown below, the presence of the cationic vesicle of 1 showed an opposite behavior, i.e., it induced an increase in the LDH activity. This difference originates in the stability of the assemblies. Micelles are a dynamically unstable assembly and are in fast equilibrium with a critical micellar concentration of unassociated species. The latter free surfactant molecules possibly denature the enzyme. In contrast, vesicles are a quite stable assembly and only a small amount of unassociated species coexists in the solution. High stability of the vesicle is a crucial point in maintaining the enzyme activity. Lipid 1 has a glycine moiety between a polar head and a dialkyl hydrophobic chain. Therefore, the vesicle from 1 is especially stabilized through hydrogen bonding formation at the glycine part [147,148]. Observation based on dynamic light scattering and electron micrography revealed the stable formation of a vesicle of 100 nm diameter. In order to determine the origin of the increased apparent activity of LDH on the assembled ¢lm of 1, the LDH activity was examined by the Michaelis-Menten analysis in the presence of an aqueous vesicle of 1. The apparent LDH activity was measured at given vesicle concentrations under varied pyruvate concentrations ([pyruvate] ¼ 0.025^0.500 mM) by keeping the other reaction parameters ([NADH] ¼ 0.25 mM, [LDH] ¼ 170 mg/L, pH 7.0, and 30 C) constant. The obtained data were ¢tted according to the following Lineweaver-Burk plot: 1 1 Km ¼ þ v0 Vmax Vmax ½pyruvate0

ð3Þ

where v0 and [pyruvate]0 are the initial reaction rate and pyruvate concentration, respectively. The Michaelis constant (km) and maximum initial rate (Vmax) obtained at various lipid concentrations are plotted as a function of the lipid concentration in Fig. 12. Because the Vmax value was almost constant (1.19^1.38 10 6 M/s), the intrinsic LDH activity was not signi¢cantly a¡ected by the presence of a cationic vesicle of 1.The LDH adsorbed on the vesicle of 1 retains its activity without serious conformational changes. In contrast, the Michealis constant K m changed signi¢cantly depending on the lipid concentration. Although the

406

Ariga et al.

FIG. 12 Michaelis constant (Km) and maximum initial rate (Vmax) of pyruvate reduction catalyzed by LDH in the presence of various concentrations of lipid 1 (30 C, pH 7, [LDH] ¼ 170 mg/L).

K m value is 34.8 10 6 M for the lipid-free LDH, this constant signi¢cantly decreased in the presence of the vesicle (3.10^40.2 106 M). This decreases leads to an increase in the apparent LDH activity, because a small K m corresponds to a high a⁄nity of the enzyme for the substrate. The decrease in the K m value is probably explained by the increase in the anionic pyruvate concentration in the vicinity of the positively charged surface of the vesicle. Due to the proximity e¡ect between the LDH and the pyruvate, the apparent a⁄nity between them becomes greater. In addition, we examined the e¡ect of metal ions (Cu2 þ , Mg2 þ , Zn2þ , 2þ Ni , and Ag þ ) on the activity of LDH immobilized on the vesicle of 1. The LDH activity was hardly a¡ected by the presence of Mg2þ , Zn2þ , and Ni2þ . Two ions (Cu2þ and Ag þ ) e¡ectively inhibited the LDH activity. MichaelisMenten analyses revealed that the Cu2þ inhibition seems to be competitive with pyruvate. In the following section, we mainly used Cu2þ as a mediator of the switchable enzyme reactor.

Nano-Sized Thin Films for Enzyme Reactors

407

3.2.2 Complexation of Receptor, Signal Molecules, and Mediator In the next step, we investigated the behavior of arti¢cial receptors. Amphiphilic amines and aldehyde compounds were used as the receptor and the signal, respectively. Schi¡ ’s base formed from the amine and the aldehyde would provide a suitable binding site for Cu2þ . Four kinds of the receptors were used in this study (Fig. 11). Receptor 2 is a cyclic cyclophane connected with four hydrophobic and rigid cholanic acids through a £exible lysine spacer. This receptor is called a steroid cyclophane. Amino groups in the side chain of the lysine spacers are receptor sites for the aldehyde signal. This type of the cyclophane ring is a good host for a naphthalene compound. The steroid cyclophanes are known to e¡ectively recognize several naphthalene guests in an aqueous solution, in a lipid bilayer, and at the air ^ water interface [149^157]. Therefore, the receptor 2 would selectively recognize naphthalene-type signal molecules. A simpli¢ed analog of 2 is receptor 3, in which one cholanic acid is connected to ethylenediamine. This simple receptor is useful for clarifying the detailed mechanisms of the switchable enzyme reactor. Therefore, receptor 3 was mainly used in this study. Receptor 4 is a monoacetylated derivate of ethylenediamine. Because this receptor does not have any hydrophobic part, it cannot be immobilized on the lipid bilayer. Comparison between 3 and 4 would demonstrate the importance of the receptor immobilization on the lipid bilayer membrane. Receptor 5 has an azobenzene moiety which shows cis ^ trans isomerization upon light irradiation. This receptor was used for a photoswitchable reactor. This system is described in a later section of this chapter. Schi¡ ’s base formation between receptor 3 and aldehyde signal 6 was easily monitored by electronic spectroscopy. A mixed vesicle of 3/1 (1/ 40 mol/mol) was prepared, and spectral changes were followed upon the addition of 6. The absorbance at 400^450 nm speci¢cally increased upon Schi¡ ’s base formation. A quantitative binding analysis provided a binding constant of 6.7  105 M1 between 3 and 6 at 30 C. A large binding constant was similarly observed for the other receptors. Binding of Cu2þ to the Schi¡ ’s base from 3 and 6 was also evaluated by electronic spectroscopy. Because the complexation of Cu2þ induced an absorbance decrease at 400^450 nm, the absorbance changes at 440 nm were measured at various Cu2þ concentrations. As a result, a huge binding constant was obtained between the 3/ 6 Schi¡ ’s base and Cu2þ . The binding constant was too large to be quantitatively determined by a curve-¢tting method, and it is apparently larger than the binding constant of Cu2þ to LDH. The binding of Cu2þ to receptor 3 itself is weak; therefore, a desirable order in a⁄nity for Cu2þ (3/6 Schi¡ ’s base > LDH > receptor 3) was

408

Ariga et al.

achieved. This a⁄nity order makes it possible to construct the switchable enzyme reactor proposed in Fig. 10. In the absence of the signal 6, Cu2þ e⁄ciently binds to LDH and inhibits the LDH activity, i.e., pyruvate reduction. When the signal was added to the system, the enzyme can be switched on through removal of Cu2þ to the formed Schi¡ ’s base. The Schi¡ ’s base formation and Cu2þ complexation induced an interesting change in the lipid-assembling state which was detected by di¡erential scanning calorimetry (DSC) (Fig. 13). A sharp endothermic peak based on gel-to-liquid crystalline transition was observed for a single-component vesicle of 1. Mixing of 3 with the vesicle 1 (3/1 ¼ 1/10 mol/mol) lowered the transition temperature and broadened the peak shape.This change indicates that receptor 3 dispersed in the lipid bilayer of 1 and disturbed the bilayer structure. Addition of the signal 6 to the mixed vesicle of 3/1 did not fundamentally change the peak shape. This means that the Schi¡ ’s base that was formed also dispersed in the lipid bilayer. However, the addition of Cu2þ to this mixture drastically changed the DSC peak, i.e., the transition temperature shifted to a higher temperature and the peak shape became sharp again. Because the transition temperature is very close to the single-component bilayer of 1, complexation of Cu2þ would induce some kind of a phase separation of the Schi¡ ’s base from the matrix component (1). This behavior suggests that the addition of Cu2þ induced assembly of the Schi¡ ’s bases. The binding stoichiometry between the 3/6 Schi¡ ’s base and Cu2þ is not 1:1.

FIG. 13 DSC thermogram of 1-containing aqueous dispersions: (a) 1 (10 mM); (b) 1 (10 mM) and 3 (1 mM); (c) 1 (10 mM), 3 (1 mM), and 6 (1 mM); (d) 1 (10 mM), 3 (1 mM), 6 (1 mM), and Cu(ClO4)2 (1 mM).

Nano-Sized Thin Films for Enzyme Reactors

409

In order to clarify the binding stoichiometry between the 3/6 Schi¡ ’s base and Cu2þ , electronic spectroscopy at various the Schi¡ ’s base/Cu2þ ratios was done while retaining the total concentration. The absorbance at 410 nm changes sensitively upon ratio variation, and this absorbance was plotted as a function of the Cu2þ content (Fig. 14). The plot apparently indicates the largest change at 0.33 Cu2þ content. This result means that the complexation most e⁄ciently occurred at a mixing ratio of 2:1 (Schi¡ ’s base: Cu2þ ), i.e., Cu2þ binds to two molecules of the Schi¡ ’s base.The motif of this complexation mode is illustrated in Fig. 15. Assembly of two chelate-type ligands with imine and naphthol OH groups forms an appropriate binding site for Cu2þ .This ¢nding is important to design reactor systems switched by other signals such as heat and light, because the assembling behavior of molecules in the lipid bilayer is sometimes controlled by physical stimuli.We applied this concept to produce a photoswitchable reactor which is described in a later section of this chapter. 3.2.3 Controlled Enzymatic Activity As the ¢nal step in construction of the chemically switched enzyme reactor, activity of the immobilized LDH was investigated under various conditions. We measured the pyruvate reduction activity of the LDH immobilized on the bilayer vesicle of 1 without and with 2 mM Cu2þ .The LDH activity increased

FIG. 14 Absorbance change at 410 nm upon addition of 3 and Cu(ClO4)2 at various ratios to an aqueous mixture of 1 and 6 (30 C, pH 7.0). Combined concentration of 3 and Cu(CIO4)2 were kept at 30 mM and the ratio between 1, 3, and 6 is 40:1:1.

410

Ariga et al.

FIG. 15 Mechanism of Cu ion trapping by Schiff’s base formed by 3 and 6 embedded in a lipid bilayer membrane of 1.

when it was immobilized on the lipid bilayer (see Fig. 12).The coexistence of the receptor and/or signal in the bilayer system did not signi¢cantly change the LDH activity in the absence of Cu2þ . The relative activity of the pyruvate reduction in the presence of Cu2þ compared with that under a Cu2þ -free condition is represented as enzymatic activity in percent (Fig. 16). The addition of 2 mM of Cu2þ decreased the LDH activity to 21% of the Cu2þ free enzyme activity when the LDH alone (170 mg/L) was immobilized on the lipid bilayer of 1 (1 mM) (Fig.16A). Existence of a signal molecule (6, 30 mM) or a receptor (3, 30 mM) did not change this situation (Figs 16B and 16C). However, coexistence of the signal and the receptor drastically increased the LDH activity, i.e., the LDH activity, increased to 82% of the Cu2 þ -free enzyme activity (Fig. 16D). This result indicates that the formation of the Schi¡ ’s base suppressed the enzymatic inhibition by removing the Cu2þ from the LDH. This can be regarded as conversion from molecular recognition to an enzymatic reaction (catalytic ampli¢cation). The basic concept illustrated in Fig. 10 was achieved. The e¡ects of a signal molecule, mediator, and receptor on the enzymatic activity were investigated (Fig. 17). Instead of 6, three kinds of signal molecules (7, 8, and 9) were used under the same condition ([1] ¼ 1 mM, [Cu2þ ] ¼ 2 mM, [3] ¼ 30 mM, and [signal] ¼ 30 mM). Signal 8 did not signi¢cantly increase the enzymatic activity, while signals 7 and 9 showed a superior e¡ect compared to 6.This di¡erence originates in the presence of an

Nano-Sized Thin Films for Enzyme Reactors

411

FIG. 16 Relative activity of LDH immobilized on lipid bilayer membrane (30 C, pH 7.0): (A) [1] ¼ 1 mM, [LDH] ¼ 0.17 mg/L, [Cu2þ ] ¼ 2 mM; (B) [1] ¼ 1mM, [LDH] ¼ 0.17 mg/L, [Cu2þ ] ¼ 2 mM, [6] ¼ 30 mM; (C) [1] ¼ 1 mM, [LDH] ¼ 0.17 mg/L, [Cu2þ ] ¼ 2 mM, [3] ¼ 30 mM; (D) [1] ¼ 1 mM, [LDH] ¼ 0.17 mg/L, [Cu2þ ] ¼ 2 mM, [3] ¼ 30 mM, [6] ¼ 30 mM. Enzymatic activity shown in the figure was measured relative to the activity in absence of Cu2þ under the corresponding conditions.

OH group on the signal molecules. The dissociated OH group plays an important role in the complexation of Cu2þ (see Fig.15).When Ag þ was used as a mediator, the enzymatic activity was signi¢cantly suppressed even in the presence of 3 and 6.The low a⁄nity of Ag þ for the signal ^ receptor complex (Schi¡ ’s base) did not block the Agþ inhibition of the enzymatic reaction. One of the most pronounced advantages of this system exists in the freedom of combination of the receptor and the enzyme.We investigated two other receptors (2 and 4) in the same system and evaluated the LDH activity. As mentioned above, the signal 9 showed better ability than 6 in the case of receptor 3. Interestingly, the system with receptor 2 showed the opposite tendency, i.e., signal 6 had a more e⁄cient e¡ect on the enzymatic activity than 9. The di¡erence in signal e⁄ciency between these two systems originated from the molecular recognition ability of the receptors.The receptor 2 has a superior recognition ability for naphthalene derivatives.Therefore, the naphthalene-type signal 6 worked better than the benzene-type 9 in the

412

Ariga et al.

FIG. 17 Relative LDH activity with various receptors, signal molecules, and mediators (30 C, pH 7.0). These values are relative to that measured under the corresponding Cu2þ free condition.

2-containing system. In contrast, formation of a small complex with the receptor 9 is probably advantageous for Schi¡ ’s base association in a 3-containing system. When water-soluble receptor 2 was used, a signi¢cant increase in the enzymatic activity was not observed. Schi¡ ’s base formation is not favorable in aqueous solution, and Cu2þ was not e⁄ciently trapped. This control experiment clearly demonstrates the importance of the receptor immobilization in the hydrophobic interfacial environment provided by the bilayer of 1. In order to construct a true switchable system, we sequentially added Cu2þ and the signal 6 to the LDH-immobilized mixed vesicle of 1 and 3. Addition of Cu2þ e⁄ciently inhibited the LDH activity, but the subsequent addition of signal 6 did not signi¢cantly increase the LDH activity. This unfortunate result might originate in irreversible denaturation of the LDH upon long-time exposure to Cu2þ .We believe that this disadvantage will be overcome by appropriate selection of mediator and enzyme. Now, e¡orts to construct a true switching reactor have been made in our laboratory. 3.3

Photoregulation of Enzyme Reactor

3.3.1 Photoswitching of Enzyme Reactor by Photoresponsible Receptor If the enzymatic activity was switched and/or controlled by multiple signals, logical information processing would be possible. Therefore, we started

Nano-Sized Thin Films for Enzyme Reactors

413

designing systems switched by another signal such as light and heat. These signals can be applied to the system without causing undesirable contamination.The shutdown of a photo signal is relatively easy,while complete removal of chemical signals from the solution is usually di⁄cult. In the following section, we describe a preliminary trial on photoregulation of the enzyme reactor. Receptor 5 was used as a photoswitchable receptor, because the central azobenzene moiety can be reversibly isomerizd by UV and visible-light irradiation. As the ¢rst step, the response to chemical signal application was investigated. Unfortunately, the receptor 5 was not well dispersed in the bilayer vesicle of 1. Therefore, zwitterionic dimyristoylphosphatidylcholine (DMPC) was used as a matrix lipid. Although electrostatic interaction as a major driving force to immobilize LDH to the lipid bilayer seems to be lost in this system, dynamic light-scattering data suggested successful immobilization of LDH on the DMPC vesicle. The Schi¡ ’s base formation of 5 (10 mM) in the DMPC bilayer (1 mM) was spectrophotometrically determined upon addition of the signal 6 at 40 C at pH 7.0.The absorbance change at 440 nm was well ¢t to a 1:1 binding isotherm with a binding constant of 8.1 105 M1. Complexation of Cu2þ to the Schi¡ ’s base formed was also monitored by electronic spectroscopy, indicating a 2:1 (Schi¡ ’s base:Cu2þ ) complex formation. Because the results obtained for the 5/DMPC system were basically the same as for the 3/1 system, enzymatic activity was similarly investigated under various conditions. Pyruvate reduction activity was completely suppressed when 4 mM of Cu2þ was added to the LDH (50 mg/L) immobilized to the DMPC vesicle (1 mM) at 40 C at pH 7.0.The addition of 5 (10 mM) or 6 (20 mM) did not change the situation, but the coaddition of 5 and 6 signi¢cantly increased the enzymatic activity to 53% of Cu2þ -free LDH (Fig. 18). The obtained results indicate that Cu2þ is e¡ectively trapped by the Schi¡ ’s base formation and association. Therefore, control of the Schi¡ ’s base association would lead to regulation of the enzymatic activity. The photoisomerization of the azobenzene moiety is expected to change the association behavior of the receptor molecules. Alternate irradiation of two kinds of light would switch the enzyme reactor (Fig. 19). We ¢rst examined the photoisomerization of 5 in the DMPC matrix upon the alternate irradiation of UVand visible light. A large spectral change was reproducibly and repeatedly observed at 290^390 nm, indicating that the receptor 5 tends to be a cis isomer and a trans isomer upon UV and visible-light irradiation, respectively, and that the isomerization can be freely repeated. The LDH activity was evaluated under the UV and visible-light irradiation at various Cu2þ concentrations. The LDH activity was almost 100% for the UV and

414

Ariga et al.

FIG. 18 Switchable reactor triggered by binding of 6 to receptor 5 (30 C, pH 7.0): (A) [DMPC] ¼ 1 mM, [LDH] ¼ 0.05 mg/L, [Cu2þ ] ¼ 4 mM, [5] ¼ 10 mM; (B) [DMPC] ¼ 1 mM, [LDH] ¼ 0.05 mg/L, [Cu2þ ] ¼ 4 mM, [5] ¼ 10 mM, [6] ¼ 20 mM. Relative LDH activity shown in the figure was measured relative to that under the corresponding Cu2þ free condition. Detailed conditions are described in the text.

visible-light conditions when the Cu2þ concentration was less than 2 mM. In contrast, the LDH activity was completely suppressed at high Cu2þ concentration ( > 4.5 mM) for both types of light irradiation. The higher LDH activity was observed for the system containing the trans receptor than that for the cis system at a limited range of Cu2þ concentration

FIG. 19

Concept of a photoswitchable enzyme reactor.

Nano-Sized Thin Films for Enzyme Reactors

415

FIG. 20 LDH activities at ON state and OFF state of the photoswitchable enzyme reactor. Detailed conditions are described in the text.

(2 mM < [Cu2þ ] < 4.5 mM). The most pronounced di¡erence was observed at 4 mM Cu2þ concentration. A relatively high activity (53%) was observed for the system containing the trans isomer (ON state), and the LDH activity was completely suppressed (6%) for the cis-isomer system (OFF state). As summarized in Fig. 20, the LDH activity was regulated by the UV and visible-light irradiation. Unfortunately, the reversible switching of the LDH activity by the alternate UV/visible irradiation failed, because long-time exposure of the LDH to Cu2þ caused irreversible inactivation of the LDH. Instead, we conducted a model experiment in which several solutions of a 5/DMPC/ Cu2þ mixture were prepared, and UV and visible light were irradiated on the solutions in various sequences. Freshly prepared LDH solution was added to every mixture and its activity was evaluated. As a result, the activities corresponding to the trans and cis systems were independently observed for the samples for various photoirradiation sequences. This result encouraged us to achieve our ¢nal goal, because appropriate

416

Ariga et al.

selection of a stable enzyme and a mediator system would lead to reversible photoregulation of the nano-sized enzyme reactor in the future. 3.3.2 Application as a Molecular Device In the latter part of this chapter, we describe the preliminary results on the switchable enzyme reactor. Molecular recognition by the arti¢cial receptors and catalytic reaction by the naturally occurring enzyme were successfully coupled on the lipid assembly with a diameter of ca. 100 nm. Instability of the enzyme causes di⁄culty in sequential switching, but it could be overcome by appropriate selection of the enzyme and the mediator (inhibitor). The results obtained lead to a nano-sized information converters, which are sometimes called a molecular device. Our system can be regarded as a switching device as illustrated in Fig. 21. In this system, a primary switch was activated by a chemical signal, and signal transduction was reversibly switched on and o¡ by a photo signal. Input of these signals activates an enzyme which can be regarded as an e¡ector. Because the enzyme can catalytically repeat the corresponding reaction, information can be ampli¢ed. One molecule and one photon can induce conversion of thousands of molecules.

FIG. 21 Molecular switching device based on a receptor^enzyme system on a lipid bilayer membrane.

Nano-Sized Thin Films for Enzyme Reactors

417

Figure 22 shows the application of the presented system as a logic device. Application of a chemical signal and a photo signal in appropriate combination can only activate the enzymatic reaction. The relation of signal application and enzymatic activity is summarized as a true ^ false table in Fig. 22. Enzymatic activity as the output was obtained only when both types of light irradiation and chemical application are true. Therefore, this system can be regarded as an AND-type logic gate. Other types of logic gates would be similarly constructed by appropriate design of receptors and selection of enzymes, because we can freely change the components of the system.

FIG. 22 Molecular logic device based on a receptor^enzyme system on a lipid bilayer membrane.

418

Ariga et al.

Appropriate combination of logic gates leads to some kinds of a nano-sized calculator. 4

FUTURE PERSPECTIVES

Development of nano-sized reactors or molecular devices is based on necessity in our future life. Accumulation of high function into nano-sized space leads to an incredibly low-energy process which can save limited amounts of energy sources and can suppress environmental pollution. Construction of the nano-sized functional reactor and device is di⁄cult, but nature has already overcome this problem long time ago; i.e., living creatures have many reactors and information converters in their body. Therefore, we have to use naturally occurring devices such as proteins and combine their function in a desirable way. Enzymes are a well-designed reactor which can catalyze the corresponding reactions with high e⁄ciency and selectivity. Several e¡orts have been made to create an arti¢cial enzyme on the basis of organic chemistry and/or biochemistry. Although many successful arti¢cial enzymes have already been developed, human e¡orts still cannot achieve the elegant design of naturally occurring enzymes. Therefore, arti¢cial organization of naturally occurring enzymes is a currently convenient way to nanosized reactors. Figure 23 schematically summarizes the enzyme reactors presented in this chapter. The ¢rst one is a nano-sized multiprocess reactor (Fig. 23A). This kind of reactor was prepared by the alternate layer-by-layer adsorption between polyions and enzymes. Multiple kinds of enzymes can be assembled in an ultrathin ¢lm (10^100 nm) in a desirable thickness and sequence. The choice of enzyme combination is almost free; therefore, multienzyme systems which do not exist in nature can be constructed. Preparation of such reactors by the alternate layer-by-layer adsorption is quite easy, mild, and inexpensive. The second example, illustrated in Fig. 23B, is a nano-sized switchable device in which an arti¢cial receptor and an enzyme are immobilized on a lipid membrane (5 nm) and their functions are connected. As shown in the corresponding model, the receptor and the enzyme play the roles of a switch and an ampli¢er, respectively. A signal spatially trnasmits from the switch (receptor) to the ampli¢er (enzyme) by di¡usion of a mediator material.This can be regarded as a wireless system. Therefore, development of the technique to spatially control receptor and enzyme dispositions will lead to great progress in this kind of wireless device. Several techniques such as twodimensional molecular patterning [158^160], molecular tiling [161,162], and two-dimensional dendrimers [163] have been currently developed to provide molecularly organized lipid assemblies. Appropriate immobilization of

Nano-Sized Thin Films for Enzyme Reactors

419

FIG. 23 Reactors and devices based on enzyme function immobilized nano-sized thin films: (A) nano-sized multiprocess reactor formed by alternate layer-by-layer adsorption; (B) nano-sized switchable device immobilized on lipid bilayer membrane.

functional molecules including enzymes onto such an organized lipid medium would lead to sophisticated wireless nano-sized devices. As mentioned above, the fabrication of the lipid assemblies would be one of the keys to designing nano-sized reactors. Amphiphilic molecules

420

Ariga et al.

such as lipids and surfactants usually provide micelles or vesicles, but many kinds of superbilayer structures have been reported. These superstructures are driven by hydrogen bonding, p^p stacking, and so on [164^166]. Complicated structures such as vesicle-containing nanotubes [167,168] have already been proposed.We have successfully modulated a superstructure in lipid cast ¢lms by simply changing the casting solvent. It was demonstrated that mesoscopic rods, thin needles with nanometer-size diameter, and regular patterns with nanometer-level thickness can be obtained from the same amphiphilic peptide [169^174]. Accumulation of knowledge to control the superstructures of the lipid assemblies would open the way to construction of supports for enzyme immobilization with nanometer-size precision. Connection of the nano-sized reactor and a semiconductor or metallic device is important when we read an output signal from the nano-sized system. For example, immobilization of the nano-sized enzyme reactor to a ¢eld-e¡ect transistor would provide a highly functional sensor device. In order to make a good connection between an organic system and an inorganic device, we have separately developed organic ^ inorganic hybrid lipid assemblies [175^181]. An amphiphile having a silanol head can form a bilayer vesicle and a planar monolayer which have a silica-life surface. The former vesicle structure has cell-membrane-like bilayer structure with a ceramiclike surface shell. This was called a ‘‘Cerasome.’’ The Cerasome can incorporate organic and biological components in the bilayer part and can form a covalent linkage with semiconductor devices at the surface. The Langmuir monolayer of the silanol amphiphile can be covalently immobilized onto a metal oxide electrode. Immobilization of the nano-sized enzyme reactor onto the organic ^ inorganic hybrid assembly would open the way to many practical applications. Support ¢lms composed of polyions have their own advantage, i.e., polyion-based ¢lms have higher permeability of substances for enzymatic reaction than a lipid-based ¢lm [154]. Therefore, systems for signal transduction along the z direction should be constructed of a polyion-based ¢lm. We already demonstrated the mixed layering of polyions and lipid bilayers by the alternate layer-by-layer adsorption. Therefore, combination of a polyion-based enzyme reactor and a lipid-based switchable reactor is highly possible. Our approach for the nano-sized enzyme reactor is based on supramolecular chemistry and molecular assembly technology. It provides huge freedom in the choice of enzymes and arti¢cial receptors, control of support structures, interfacing with arti¢cial devices, and combination of these systems. Many kinds of nano-sized reactors and devices can be £exibly constructed based on the £exible nature of supramolecular chemistry using our £exible thinking.

Nano-Sized Thin Films for Enzyme Reactors

421

ACKNOWLEDGMENTS Research explained in the ¢rst part (multienzyme reactor) was performed in the Supermolecules Project, Japan Science and Technology Corp. (JST), which was directed by Prof. Toyoki Kunitake. Research described in the latter part (switchable reactor) was supported in part by a Grant-in-Aid for Scienti¢c Research on a Priority Area (No. 404) from the Ministry of Education, Science, Sports, and Culture. REFERENCES 1. R Dagami. Chem Eng News 36, 2002. 2. http.//www.nano.gov/ 3. B Albert, D Bray, J Lewis, M Ra¡, K Roberts, JD Watson. Molecular Biology of the Cell. 2nd ed. New York: Garland, 1989. 4. DS Goodsell. Am Sci 88:230, 2000. 5. M Aizawa, N Damrongchai, E Kobatake. Supramolec Sci 3:149, 1996. 6. C Bourdillon, C Demaille, J Moriroux, JM Save¤nt. J Am Chem Soc 116:10328, 1994. 7. W Kong, X Zang, M Gao, H Zhou,W Li, JC Shen. Macromol Rapid Commun 15:405, 1994. 8. K Koyama, N Yamaguchi,T Miyasaka. Science 265:726, 1994. 9. W Mu«ller, H Ringsdorf, E Rump,G wildburg, X Zang, L Angermaier,W Knoll, M Liley, J Spinke. Science 262:1706, 1993. 10. L Boguslavsky, H Kalash, Z Xu, D Beckles, L Geng, T Skotheim, V Laurinavicius, HS Lee. Anal Chim Acta 311:15, 1995. 11. C Bullock. Sci Prog 78:119, 1995. 12. A Sibata,Y Iizuka, S Ueno,T Yamashita.Think Solid Films 284/285:549, 1996. 13. P Fromherz. In: W Baumeister, W Vogell, eds. Electron Microscopy at Molecular Dimension. Berlin, Heidelberg: Springer-Verlag, 1980, p 338. 14. N Damrongchai, E Kobatake, T Haruyama, Y Ikariyama, M Aizawa. Bioconjugate Chem 6:264, 1995. 15. TS Berzina,VI Troitsky, A Petrigliano, D Alliata, AYuTronin, C Nicolini.Thin Solid Films 284/285:757, 1996. 16. Y Okahata,T Tsuruta, K Ijiro, K Ariga. Langmuir 4:1373, 1988. 17. Y Okahata,T Tsuruta, K Ijiro, K Ariga. Thin Solid Films 180:65, 1989. 18. G Decher. Science 277:1232, 1997. 19. Y Lvov. In: Y Lvov, H Mo«hwald, eds. Protein Architecture
Interfacing Molecular Assemblies and Immobilization Biotechnology. New York: Marcel Dekker, 2000, p 125. 20. W Knoll. Curr. Opin. Colloids Interface 1:137, 1996. 21. RK Iller. J Colloid Interface Sci 21:569, 1966. 22. M Onda,Y Lvov, K Ariga,T Kunitake. Jpn J Appl. Phys 36:L1608, 1997. 23. G Decher,Y Lvov, J Schmitt. Thin Solid Films 244:772, 1994. 24. G Decher, JD Hong, J Schmitt. Thin Solid Films 210/211:831, 1992.

422

Ariga et al.

25.

D Korneev,Y Lvov, G Decher, J Schmitt, S Yaradaikin. Phidica B 213/214:954, 1995. Y Lvov, G Decher, H Mo«hwald. Langmuir 9:481, 1993. Y Lvov, K Ariga, M Onda, I Ichinose,T Kunitake.Colloid Surf A 146:337,1999. S Yamada, A Harada, T Matsuo, S Ohno, I Ichinose, T Kunitake. Jpn J Appl Phys 36:L1110, 1997. Y Lvov, S Yamada,T Kunitake. Thin Solid Films 300:107, 1997. I Ichinose, S Mizuki, S Ohno, H Shiraishi,T Kunitake. Polymer J 31:1065,1999. AC Fou, O Onitsuka, M Ferreira, MF Rubner, BR Hsieh. J Appl Phys 79:7501, 1996. SS Shiratori, MF Rubner. Macromolecules 33:4213, 2000. D Yoo, SS Shiratori, MF Rubner. Macromolecules 31:4309, 1998. WB Stockton, MF Rubner. Macromolecules 30:2717, 1997. M Yamada, SS Shiratori. Sensors Actuators B 64:124, 2000. SS Shiratori,Y Inami, M Kikuchi,T Yamada, M Yamada. Polymer Adv Technol 11:766, 2000. M Onoda, K Yoshino. Jpn J Appl Phys 34:L260, 1995. K Emoto, M Iijima,Y Nagasaki, K Kataoka. J Am Chem Soc 122:2653, 2000. A Toutianoush, F Saremi, B Tieke. Mater Sci Eng C 8^9:343, 1999. J Blaakmeer, MR Bo«hmer, MA Cohen, GJ Fleer. Macromolecules 23:2301, 1990. V Phuvanartnuruks,TJ McCarthy. Macromolecules 31:1906, 1998. MC Hsieh, RJ Farris,TJ McCarthy. Macromolecules 30:8453, 1997. X Jiang, P Hammond. Langmuir 16:8501, 2000. M Ginzburg, J Galloro, G Ja«kle, KN Power-Billard, S Yang, I Sokolov, CNC Lam, AW Neumann, I Manners, GA Ozin. Langmuir 16:9609, 2000. Y Lvov, K Ariga, I Ichinose,T Kunitake. J Am Chem Soc 117:6117, 1995. Y Lvov, K Ariga,T Kunitake. Chem Lett 2323, 1994. K Ariga, M Onda,Y Lvov,T Kunitake. Chem Lett 25, 1997. Y Lvov, K Ariga, I Ichinose,T Kunitake. Thin Solid Films 284/285:797, 1996. FCaruso,DN Furlong,K Ariga,IIchinose,TKunitake.Langmuir14:4559,1998. K Ariga, T Kunitake. In: Y Lvov, H Mo«hwald, eds. Protein Architecture
Interfacing Molecular Assemblies and Immobilization Biotechnology. New York: Marcel Dekker, 2000, p 169. M Onda,Y Lvov, K Ariga,T Kunitake. Biotech Bioeng 51:163, 1996. M Onda,Y Lvov, K Ariga,T Kunitake. J Ferment Bioeng 82:502, 1996. M Onda, K Ariga,T Kunitake. J Biosci Bioeng 87:69, 1999. K Ariga,Y Sasaki, H Horiguchi, N Horiuchi, J Kikuchi. In: RP Agarwala, ed. Soft Chemistry Leading to Novel Materials. Uetikon-Zuerich: Scitec Publications, 2001, p 35. Y Lvov, MOnda, K Ariga,T Kunitake.J Biomater Sci Polymer Edn 9:345,1998. Y Lvov, K Ariga, I Ichinose,T Kunitake.JChem Soc Chem Commun 278,1995. M Housk, E Brynda. J Colloid Interface Sci 188:243, 1997. J Hodak, R Etchenique, E Calvo, K Singhal, P Barlett. Langmuir 13:2708,1997. J.-A. He, L. Samuelson, L. Li, J. Kumar, S. Tripathy. Langmuir 14:1674, 1998.

26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50.

51. 52. 53. 54.

55. 56. 57. 58. 59.

Nano-Sized Thin Films for Enzyme Reactors

423

60. J-A He, L Samuelson, L Li, J Kumar, SK Tripathy. J Phys Chem B 102:7067, 1998. 61. M Li, B Li, L Jiang,T Tussila, N Tkachenko, H Lemmetyinen. Chem Lett 266, 2000. 62. F Caruso, K Niikura, N Furlong,Y Okahata. Langmuir 13:3427, 1997. 63. Y Lvov, Z Lu, X Zu, J Schenkman, J Rusling. J Am Chem Soc 120:4073, 1998. 64. J Lang, M Liu. J Phys Chem B 103:11393, 1999. 65. G Sukhorukov, H Mo«hwald, G Decher, Y Lvov. Thin Solid Films 284:220, 1996. 66. Y Lvov, H Haas, G Decher, H Mo«hwald. Langmuir 10:4232, 1994. 67. J Anzai,T Hoshi, N Nakamura. Langmuir 16:6306, 2000. 68. SV Rao, KWAnderson, LG Bachas. Biotech Bioeng 65:389, 1999. 69. J-D Hong, K Lowack, J Scmitt, G Decher. Prog Colloid Polymer Sci 93:98, 1993. 70. W Kong, X Zhang, ML Gao, H Zhou,W Li, JC Shen. Macromol Rapid Commun 15:405, 1994. 71. X Zhang,Y Sun, J Shen. In: Y Lvov, H Mo«hwald, eds. Protein Architecture
Interfacing Molecular Assemblies and Immobilization Biotechnology. New York: Marcel Dekker, 2000, p 229. 72. W Kong, LP Wang, ML Gao, H Zhou, X Zhang, W Li, JC Shen. J Chem Soc Chem Commun 1297, 1994. 73. YP Sun, X Zhang,CQ Sun, BWang, JShen.Macromol Chem Phys197:147,1996. 74. J-A He, R Valluzzi, K Yang, T Dolukhanyan, C Sung, J Kumar, SK Tripathy, L Samuelson, L Balogh, DA Tomalia. Chem Mater 11:3268, 1999. 75. Y Lvov, K Ariga, I Ichinose,T Kunitake. Langmuir 12:3038, 1996. 76. Y Lvov, K Ariga, M Onda, I Ichinose,T Kunitake. Langmuir 23:6195, 1997. 77. K Ariga,Y Lvov, M Onda, I Ichinose,T Kunitake. Chem Lett 125, 1997. 78. I Ichinose, H Tagawa, S Mizuki,Y Lvov,T Kunitake. Langmuir 14:187, 1998. 79. YM Lvov, JF Rusling, DL Thomsen, F Papadimitrakopoulos, T Kawakami, T Kunitake. Chem Commun 1229, 1998. 80. Y Lvov, B Munge, O Giraldo, I Ichinose, SL Suib, JF Rusling. Langmuir 16:8850, 2000. 81. K Ariga,Y Lvov, I Ichinose,T Kunitake. Appl Clay Sci 15:137, 1999. 82. I Pastoriza-Santos, DS Koktysh, AA Mamendov, M Giersig, NA Kotov, LM Liz-Marza¤n. Langmuir 16:2731, 2000. 83. FG Aliev, MA Correa-Duarte, A Mamedov, JW Ostrander, M Giersig, LM Liz-Marza¤n, NA Kotov. Adv Mater 11:1006, 1999. 84. NA Kotov, I Dekany, JH Fendler. J Phys Chem 99:13065, 1995. 85. T Cassagneau, JH Fendler,TE Mallouk. Langmuir 16:241, 2000. 86. T Cassagneau, JH Fendler,TE Mallouk. J Am Chem Soc 120:7848, 1998. 87. NI Kovtyukhova, PJ Oliver, BR Martin, TE Mallouk, SA Chizhik, EV Buzaneva, AD Gorchinskiy. Chem Mater 11:771, 1999. 88. M Gao, X Zhang, B Yang, F Li, J Shen. Think Solid Films 284/285:242, 1996. 89. K Ariga,Y Lvov,T Kunitake. J Am Chem Soc 119:2224, 1997. 90. C Tedeschi, F Caruso, H Mo«hwald, S Kristein. J Am Chem Soc 122:5841, 2000.

424

Ariga et al.

91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103.

SL Clark, ES Handy, MF Rubner, PT Hammond. Adv Mater 11:1031, 1999. K Araki, MJ Wagner, MS Wrigton. Langmuir 12:5393, 1996. T Cooper, A Campbell, R Crane. Langmuir 11:2713, 1995. G Decher, J-D Hong. Ber Bunsenges Phys Chem 95:1430, 1991. Y Lvov, G Decher, H Mo«hwald. Langmuir 9:481, 1993. G Decher, J-D Hong. Makromol Chem Macromol Symp 46:321, 1991. I Ichinose, K Fujiyoshi, S Nizuki,Y Lvov,T Kunitake. Chem Lett 257, 1996. Y Lvov, F Essler, G Decher. J Phys Chem 97:13773, 1993. J Kikuchi, K Ariga,T Miyazaki, K Ikeda. Chem Lett 253, 1999. J Kikuchi, K Ariga, K Ikeda. Chem Commun 547, 1999. J Kikuchi, K Ariga,Y Sasaki, K Ikeda. J Mol Catal B, 11:977, 2001. K Fukuda,Y Sasaki, K Ariga, J Kikuchi. J Mol Catal B, 11:971, 2001. J Kikuchi, K Ariga,Y Sasaki. In: GW Gokel, ed. Advances in Supramolecular Chemistry,Vol. 8. South Miami: Cerberus Press, 2002, p 131. H Lee, LJ Kepley,H-GHong,SAkhter,TEMallouk.J Phys Chem 92:2597,1988. P Bernt, K Kurihara,T Kunitake. Langmuir 8:2486, 1992. SS Shiratori, M Yamada. Polymer AdvTechnol 11:810, 2000. Y Shimazaki, M Mitsuishi, S Ito, M Yamamoto. Langmuir 13:1385, 1997. G Decher. Nachr Chem Tech Lab 41:793, 1993. T Serizawa, K Hamada,T Kitayama, N Fujimoto, K Hatada, M Akashi. J Am Chem Soc 122:1891, 2000. T Serizawa, K Hamada, T Kitayama, K Katsukawa, K Hatada, M Akashi. Langmuir 16:7112, 2000. F Caruso, C Schu«ler. Langmuir 16:9595, 2002. F Caruso, D Trau, H Mo«hwald, R Renneberg. Langmuir 16:1485, 2000. S Leporatti, A Voigt, R Mitlo«hner, G Sukhorukov, E Donath, H Mo«hwald. Langmuir 16:4059, 2000. F Caruso, H Mo«hwald. J Am Chem Soc 121:6039, 1999. F Caruso, RA Caruso, H Mo«hwald. Science 282:1111, 1998. GB Sukhorukov, E Donath, S Davis, H Lichtenfeld, F Caruso, VI Popov, H Mo«hwald. Polymer AdvTechnol 9:759, 1998. E Donath,GB Sukhorukov, F Caruso, SA Davis, H Mo«hwald. Angew Chem Int Ed 37:2202, 1998. F Caruso, E Donath, H Mo«hwald. J Phys Chem B 102:2011, 1998. Y Lvov, AA Antipov, A Mamedov, H Mo«hwald, GB Sukhorukov. Nano Lett, 1:125, 2001. Y Lvov, RR Price. Colloid Surf B, 23:251, 2002. K Matsuura, K Ariga, K Endo,YAoyama,YOkahata.Chem Eur J 6:1750, 2000. K Ariga, K Endo,YAoyama,Y OkahataYoshio. Colloids Surf A 169:177, 2000. K Ariga, K Isoyama,OHayashida,YAoyama,YOkahata.Chem Lett1007,1998. K Ariga,Y Okahata. Langmuir 10:3255, 1994. Y Ebara, H Ebato, K Ariga,Y OkahataYoshio. Langmuir 10:2267, 1994. K Ariga,Y Okahata. Langmuir 10:2272, 1994. K Ariga,Y Okahata. Langmuir 10:3255, 1994. K Ariga,Y Okahata. J Colloid Interface Sci 167:275, 1994.

104. 105. 106. 107. 108. 109. 110. 111. 112. 113. 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128.

Nano-Sized Thin Films for Enzyme Reactors 129.

425

Y Okahata, K Ariga, K Tanaka. In: T Tsuruta, ed. New Functionality Materials Volume C. Amsterdam: Elsevier, 1993, p 489. 130. Y Okahata, K Kimura, K Ariga. J Am Chem Soc 111:9190, 1989. 131. Y Okahata, K Ariga. Thin Solid Films 178:465, 1989. 132. Y Okahata, K Ariga. Langmuir 5:1261, 1989. 133. Y Okahata, K Ariga. J Chem Soc Chem Commun 1535, 1987. 134. G Sauerbrey. Z Physik 155:206, 1959. 135. K Takagi, M Nakao,Y Ogura,T Nabeshima, A Kunii. Clin Chim Acta 226:67, 1994. 136. T Komori, N Muramatsu,T Kondo. J Microencapsulation 3:219, 1986. 137. A Pauliukonis, D Jankauskaite, A Dikciuviene, VG Malinovskii. Prikl Biokhim Mikrobiol 16:222, 1980. 138. NC Pace, BA Shirley, A Thomson. In: TE Creighton, ed. Protein Structure. Oxford, UK: IRL, 1990. 139. M Dixon, E-C Webb. In:Enzymes, 2nd ed. New York: Academic, 1964, p 315. 140. M Ohnish,T Yamashita, K Hiromi. J Biochem 79:1007, 1976. 141. JF Rusling. In: Y Lvov, H Mo«hwald, eds. Protein Architecture
Interfacing Molecular Assemblies and Immobilization Biotechnology. New York: Marcel Dekker, 2000, p 337. 142. M Rodbell. Angew Chem Int Ed 34:1420, 1995. 143. AG Gilman. Angew Chem Int Ed 34:1406, 1995. 144. CS Zuker, R Ranganathan. Science 283:650, 1999. 145. MA Perry, JN Benot, PR Kvietys, N Granger. Am J Physiol 245:G568, 1983. 146. J Kikuchi,Y Kamijyo, H Etoh,Y Murakami. Chem Lett 427, 1996. 147. Y Murakami, J Kikuchi. In: H Dugas, ed. Bioorganic Chemistry Frontiers. Vol. 2. Berlin: Springer-Verlag, 1991, p 73, and references cited therein. 148. S Kawanami,T Kosaka,T Abe, K Ariga, J Kikuchi. In: Y Iwasawa, N Oyama, H Kunieda, eds. Studies in Surface Science and Catalysis. Amsterdam: Elsevier, 2001, p 553. 149. K Ariga,YTerasaka,DSakai,HTsuji,J Kikuchi.JAmChem Soc112:7835, 2000. 150. K Ariga,DSakai,YTerasaka,HTsuji,JKikuchi.ThinSolidFilms,393:291,2001. 151. K Ariga, Y Terasaka, H Tsuji, D Sakai, J Kikuchi. In: Y Iwasawa, N Oyama, H Kunieda, eds. Studies in Surface Science and Catalysis. Amsterdam: Elsevier, 2001, p 443. 152. J Kikuchi,CMatsushima,K Suehiro,ROda,YMurakami.Chem Lett1807,1991. 153. J Kikuchi, C Matsushima, Y Tanaka, K Hie, K Suehiro, O Hayashida, Y Murakami. J Phys Org Chem 5:533, 1992. 154. J Kikuchi, M Inada, H Miura, K Suehiro, O Hayashida, Y Murakami. Recl Trav Chim Pays-Bas 113:216, 1994. 155. Y Murakami, J Kikuchi, O Hayashida. Top Curr Chem 175:133, 1995. 156. J Kikuchi,T Ogata, M Inada,Y Murakami. Chem Lett 771, 1996. 157. J Kikuchi, M Inada, Y Murakami, K Egami, K Suehiro. J Phys Org Chem 10:351, 1997. 158. Y Oishi, Y Torii, T Kato, M Kuramori, K Suehiro, K Ariga, K Taguchi, A Kamino, H Koyano, H Kunitake. Langmuir 13:519, 1997.

426 159.

Ariga et al.

Y Oishi, Y Torii, M Kuramori, K Suehiro, K Ariga, T Taguchi, A Kamino, T Kunitake. Chem Lett 411, 1996. 160. H Koyano, K Yoshihara, K Ariga, T Kunitake, Y Oishi, O Kawano, M Kuramori, K Suehiro. Chem Commun 1769, 1996. 161. P Bissel, M Onda, K Yoshihara, H Koyano, K Ariga, T Kunitake, Y Oishi, K Suehiro. Langmuir 15:1791, 1999. 162. K Ariga, N Takagi, R Tanaka, J Kikuchi. Mol Cryst Liq Cryst, 371:21, 2001. 163. K Ariga,T Urakawa, A Michiue,Y Sasaki, J Kikuchi. Langmuir 16:9147, 2000. 164. J-H Fuhrhop, J Ko«ning. In: Membranes and Molecular Assemblies: The Synkinetic Approach. Cambridge, UK: Royal Society of Chemistry, 1994. 165. J-H Fuhrhop,W Helfrich. Chem Rev 93:1656, 1993, and references therein. 166. T Kunitake. Angew Chem Int Ed 104:692, 1992, and references therein. 167. T Shimizu, S Ohnishi, M Kogiso. Angew Chem Int Ed 37:3260, 1998. 168. T Shimizu, M Kogiso, M Masuda. Nature 383:487, 1996. 169. K Ariga, J Kikuchi, M Naito, E Koyama, N Yamada. Langmuir 16:4929, 2000. 170. N Yamada, K Ariga. Synlett 575, 2000. 171. K Ariga, J Kikuchi, M Naito, N Yamada. Polymer AdvTechnol 11:856, 2000. 172. N Yamada, K Matsubara, K Narumi, Y Sato, E Koyama, K Ariga. Colloids Surf A 169:271, 2000. 173. K Ariga, J Kikuchi, K Narumi, E Koyama, N Yamada. Chem Lett 787, 1999. 174. N Yamada, K Ariga, M Naito, K Matsubara, E Koyama. J Am Chem Soc 120:12192, 1998. 175. K Katagiri, K Ariga, J Kikuchi. Chem Lett 661, 1999. 176. K Katagiri, K Ariga, J Kikuchi. In: Y Iwasawa, N Oyama, H Kunieda, eds. Studies in Surface Science and Catalysis. Amsterdam: Elsevier, 2001, p 599. 177. Y Okahata, M Yokobori,Y Ebara, H Ebato, K Ariga. Langmuir 6:1148, 1990. 178. K Ariga,Y Okahata. J Am Chem Soc 111:5618, 1989. 179. Y Okahata, K Ariga, H Nakahara, K Fukuda. J Chem Soc Chem Commun 1069, 1986. 180. E Ando,Y Goto, K Morimoto, K Ariga,Y Okahata. Thin Solid Films 180:287, 1989. 181. Y Okahata, K Ariga, O Shimizu. Langmuir 2:538, 1986.

9 Biomimetic Nanoparticle Synthesis Janos H. Fendler Clarkson University, Potsdam, New York, U.S.A.

1

INTRODUCTION

The highly desirable and often unique mechanical, chemical, electrical, optical, magnetic, magneto-optical, and electro-optical properties of nanoparticles have prompted their extensive investigation [1^3]. Indeed, a National Nanotechnology Initiative has been launched in the United States to ‘‘develop the capacity to create a¡ordable products with dramatically improved performance through gaining a basic understanding of ways to control and manipulate matter at the ultimate frontier
the nanometer
and through the incorporation of nanostructures and nanoprocesses into technological innovations. In addition to producing new technologies, the study of nanoscale systems also promises to lead to fundamentally new advances in our understanding of biological, environmental, and planetary systems’’ [4]. E¡orts similar to the U.S. National Natotechnology Initiative are underway in China, the European Community, Japan, Korea, and other countries. Nanoparticles, as their name implies, have diameters in the 1 to 100-nm range. Size quantization occurs when the electron ^ hole pair is con¢ned in particles whose sizes are smaller than the de Broglie electron, the mean free path of the exciton, or the wavelength of the phonons. Electron ^ hole con¢nement in nanosized spherical particles results in three-dimensional size quantization, i.e., in the formation of ‘‘quantum dots,’’ ‘‘quantum crystallites,’’ or zero-dimensional excitons.’’ Two-dimensional con¢nement of the size carriers results in the formation of ‘‘quantum wires’’ or 427

428

Fendler

‘‘one-dimensional excitons’’ (i.e., the exciton is provided with only onedimensional mobility). Finally, in one-dimensional size quantization, the exciton is permitted to move in two dimensions (‘‘two-dimensional excitons’’), with the resultant formation of ‘‘quantum wells.’’ Exploitation of the size-quantized nanoparticles and nanostructured materials is expected to lead to ‘‘breakthroughs in areas such as materials and manufacturing, nanoelectronics, medicine and healthcare, environment, energy, chemical, biotechnology, agriculture, information technology and national security [4]. There are two fundamentally di¡erent approaches to the synthesis of nanoparticles: ‘‘top-down’’ and ‘‘bottom-up’’ [5]. Exhaustive grinding of bulk materials down to submicrometer particles illustrates the ‘‘top-down’’ approach. Such industrial products as ferro£uids (liquid magnets) and components for photography are still fabricated by the ball milling of bulk materials. The ‘‘bottom-up’’ approach, based on the atom-by-atom or molecule-by-molecule assembly of nanoparticles, has the advantage of versatility and economy.Chemists, by vocation and de¢nition, are makers of molecules. Rather than ‘‘breaking down’’ materials, they ‘‘build them up’’ from their elements, often by innovative routes. Increasingly, chemists are turning their attention to the synthesis of molecular clusters and to the formation and stabilization of colloidal nanoparticles.Versatility and the relative ease of scale-up are the advantages of the chemical approach to nanoparticle preparations. Mother Nature routinely performs elegant and e⁄cient nanoparticle preparations in a more advanced manner than either physicists or chemists. Not only are the appropriate and monodisperse nanoparticles synthesized, they are processed into higher-level organizations in biomineralization [6]. Biomineralization involves the self-assembly of the reaction sites, molecular recognition and oriented site-speci¢c nucleation, vectorial regulation of the crystal growth, controlled nucleation and particle growth, hierarchial selfassembly, and self-replication [6]. Accomplishments of the humble magnetotactic bacterium illustrate well the power of biomineralization. The bacterium is capable of producing 20 to 25, 45 8-nm-diameter, spherical, single-domain Fe3O4 (magnetite) particles in the cytoplasmic membrane which are nicely aligned along its body (Fig. 1) [7]. The bacterium uses these monodisperse magnetites, in connection with the magnetic ¢eld of the earth, to navigate toward warmer waters. The mimicking of biology in general, and of biomineralization and the functioning of the biological membrane in particular, has led to the development and burgeoning of bioorganic (and bioinorganic) chemistry, of biomimetic materials chemistry [6], and of the membrane mimetic approach to advanced materials preparations [8]. The membrane mimetic approach relies on the construction of templates and/or compartments in which

Biomimetic Nanoparticle Synthesis

429

FIG. 1 Electron micrographs of a magnetic particle chain biomineralized within the cells of a magnetotactic bacterium. (From Ref. 5.)

nanoparticles are generated in situ, or into which they are incorporated. The templates and compartments are designed to imitate such essential functions of the biological membrane as organization and compartmentalization in distinct microenvironments. Zeolites and related molecular sieves, pillared clays and clay organocomplexes, porous glasses, graphite and metallic tubes, and polymeric membranes have been used as templates [9]. Monolayers, Langmuir-Blodgett (LB) ¢lms, self-assembled monolayers and

430

Fendler

multilayers, micelles, surfactant vesicles (liposomes), bilayer lipid membranes (BLMs), cast multibilayers, and polymers have been used as compartment [8].The terms ‘‘templates’’and ‘‘compartments’’are somewhat arbitrary and often used interchangeably. Similarly, membrane mimetic chemistry and biomimetic materials chemistry are closely related.The latter term has been used to describe biologically inspired advanced materials synthesis by molecular tectonics. The phrase molecular tectonics (Gk. tekton ¼ builder) has been coined by chemists to describe the construction of supramolecules which have integrated molecular synthesis and selfassembly into larger structures. The preparation of nanoparticles and the construction of nanostructured materials by the membrane mimetic approach is the long-term research objective in our laboratories. Advantage has been taken of membrane mimetic systems to provide chemical, spatial, and dimensionality control for the in-situ generation and stabilization of ultrasmall metallic, semiconducting, and magnetic particles and particulate ¢lms [8]. Here preparation of nanoparticles under monolayers, in reversed micelles, and in surfactant vesicles will be surveyed. Emphasis will be placed on providing examples for self-organization, self-assembly, and selfreplication. Inevitably, emphasis will be on work carried out in our own laboratories. 2

NANOPARTICLE SYNTHESIS UNDER MONOLAYERS

Surfactant monolayers, spread on aqueous solution surfaces in a ¢lm balance, can be considered to model the surface of biological membranes [10,11]. Accordingly, nanoparticles have been grown in situ under monolayers from their precursors [12]. Both chemical and electrochemical routes have been developed for the in-situ generation of nanoparticles under monolayers [8,12,13]. The experimental setup used for the chemical generation and in situ monitoring of nanocrystalline particulate ¢lms is illustrated in Fig. 2 [12]. Typically, a surfactant monolayer is spread on an aqueous solution of the metal salt precursor of the nanoparticles and crystallization is induced by the injection of the reactant gas into the closed system. Facilities are provided for determining surface pressure versus surface area and surface potential versus surface area isotherms in the ¢lm balance placed under the glass cover. Re£ectivities, angle-dependent re£ectivities, Brewster angle and £uorescence microscopies, and nonlinear optical parameters can also be monitored during the nanoparticle formation under the monolayer. Evolution of a nanocrystalline particulate ¢lm, as illustrated by the formation of sul¢de semiconductor particulate ¢lms (Fig. 3), has been

Biomimetic Nanoparticle Synthesis

431

FIG. 2 Schematic of the apparatus used for the preparation of nanoparticles and nanoparticulate films under monolayers. P is the polarizer and D the detector.

discussed in terms of the following steps [12]. 1. 2. 3. 4.

5. 6.

Formation of metal ^ sul¢de bonds at a large number of sites at the monolayer ^ aqueous interface Downward growth of well-separated nanocrystalline metal sul¢de particles Coalescence of clusters into interconnected arrays of semiconductor particles Formation of the ‘‘¢rst layer’’ of a porous sul¢de semiconductor particulate ¢lm composed of 20- to 40-—-thick, 30- to 80-—diameter particles Di¡usion of fresh metal ions to the monolayer head group area Formation of a ‘‘second layer’’of the porous sul¢de semiconductor particulate ¢lm (by using steps 1^3)

432

Fendler

FIG. 3 Schematics of the growth of CdS nanoparticles and nanoparticulate films under a monolayer. The structures from top to bottom are formed at progressively longer times subsequent to the injection of H2S.

7.

Buildup of ‘‘subsequent layers’’ of the sul¢de semiconductor particulate ¢lm (by using steps 1^3) up to a plateau thickness (ca. 300 — for CdS and ca. 3500 — for ZnS) beyond which the ¢lm cannot grow.

The presence of a monolayer with an appropriate surface charge is essential to sul¢de semiconductor particulate ¢lm formation. In the absence of a monolayer, infusion of H2S over an aqueous metal-ion solution results in the formation of large, irregular, and polydispersed metal sul¢de particles which precipitate in the bulk solution before settling to the bottom of the trough.

Biomimetic Nanoparticle Synthesis

433

The oriented growth requires the matching of the crystal lattice of the surfactants, constituting the monolayers, with that of the incipient nanocrystallites. Such epitaxial matching has been achieved by growing lead sul¢de [14,15], lead selenide [16], and cadmium sul¢de [17] under monolayers prepared from arachidic acid and from mixtures of arachidic acid and octadecylamine. The approach is illustrated here by the description of the epitaxial growth of well-oriented, relatively monodisperse equilateral triangular lead sul¢de nanocrystallites (Fig. 4). The mechanism of oriented crystal growth has been rationalized by comparing the structures of the arachidic acid monolayer and the lead sul¢de crystals grown under it. Synchrotron X-ray studies of arachidic acid monolayers in their solid states showed that they comprise fully extended molecules, bearing a planar zig zag conformation. The arachidic acid molecules are oriented approximately normal to the liquid surface in a hexagonal close-packed array and exhibit a lattice constant of a ¼ 4:85 —. An experimentally obtained lattice constant of arachidic acid monolayers on lead nitrate of a ¼ 4:81 —, as derived from surface pressure-versus-surface area isotherms, was considered to be in good agreement with the published data and was utilized in the analysis. Lead sul¢de possesses a NaCl-type cubic structure with a lattice constant of a ¼ 5:9458—. Epitaxial growth of lead sul¢de from the f1 1 1g face resulted from the geometric complementarity between the arachidic acid monolayer and the f1 1 1g lead sul¢de face (Fig. 5). The Pb ^ Pb and S ^ S interionic distances of 4:20 — in the lead sul¢de f1 1 1g plane geometrically matched the df100g spacing of 4:16 — for arachidic acid; the spatial mismatch between the crystals is only of the order of 1%. The investigations of epitaxial lead sul¢de growth were extended by doping the supporting arachidic acid monolayer with octadecylamine [18]. The size and orientation preference of lead sul¢de grown under mixed arachidic acid ^ octadecyl amine monolayers was shown to be profoundly in£uenced by the arachidic acid-to-octadecyl amine ratio and the applied surface pressure.The lead sul¢de growth habit was observed to change from [111] to [001] with a reduction in the arachidic acid to octadecyl amine ratio from 1:0 and 5:1 to 2:1. The II versus A isotherms were identical for these monolayer compositions, indicating a maintenance of the hexagonal close-packed structure [16]. The di¡erences in morphology between equilateral-triangular PbS-I, right-angle-triangular PbS-II (epitaxially grown under monolayers, prepared from AA:ODA ¼ 1:0 and AA:ODA ¼ 1:1), and disk-shaped PbS-III (nonepitaxially grown under monolayers, prepared from hexadecylphosphate), manifested themselves in di¡erent spectro electrochemical behavior [19].

434

Fendler

FIG. 4 Transmission electron micrographs of PbS generated under arachidic acid monolayers. The insert shows the diffraction pattern.

3

NANOPARTICLE SYNTHESIS IN REVERSED MICELLES

Surfactants in reversed micelles surround and con¢ne nano- to micrometersized water pools in an apolar organic solvent [8,20,21]. Reversed micelles can be considered, therefore, to mimic the aqueous regions of membrane

Biomimetic Nanoparticle Synthesis

435

FIG. 5 Schematics of the epitaxy of PbS grown under an arachidic acid monolayer. Empty circles correspond to the carboxyl groups of the arachidic acid. The solid circles represent Pb in PbS.

bound proteins as well as vesicles and/or cell interiors.The aqueous pools of reversed micelles have been fruitfully employed as media for the in-situ generation of nanoparticles [8,22^25]. Sodium bis(2-ethylhexyl)sulfosuccinate (Na-aerosol-OT, Na-AOT) has been the most frequently used surfactant, since it is able to solubilize vast amounts water ([H2O]/[AOT]  50) in such hydrocarbons as octane or hexane. The in situ generation of a nanoparticle in reversed micelles was ¢rst reported in 1984 [26]. Injection of stoichiometric amounts of H2S into a reversed micellar solution of AOT in isooctane, whose aqueous water pool contained CdCl2 or Cd(NO3)2, resulted in the formation of 5.0-nm-diameter cadmium sul¢de particles [26]. Arresting the growth of nanoparticles in the reverse micelles by capping them with thiols or with other nucleophilic reagents led to unprecedented size control [27,28]. Size control was achieved by controlling the nucleophile (thiophenol, for example) and the reagent (H2S, for example) ratios. Competition between the nucleophile and the reagent for the metal ion (Cd2 þ , for example) was believed to govern the

436

Fendler

growth of the semiconductor colloid (CdS, for example) [8].The process can be considered to be analogous to living polymerization. Nucleation and growth of nanoparticle clusters correspond to the initiation and propagation steps and the using up of the reagent terminates the growth, while the addition of extra reagent to the capped nanoparticles results in an additional size increase [29]. Reversed micelles have been advantageously employed as templates for the preparation of metallic nanoparticles and for their protection against unwanted oxidation. Reduction of functionalized copper bis(2-ethylhexyl) sulfosuccinate, Cu(AOT)2, resulted in the formation of copper nanoparticles; in contrast, reduction of aqueous copper sulfate solutions produced only oxidized copper [30]. Magnetic [31] and ferroelectric [32] nanoparticles have also been prepared in reversed micelles. Self-replication is an everyday occurrence in living systems [33]. Selfreplication of such surfactant-based membrane mimetic compartments as aqueous micelles [34], reversed micelles [35,36], and surfactant vesicles [37,38] has been described in the last decade by Luisi and his coworkers. In these entirely chemical systems, surfactant molecules were shown to be formed autocatalytically at the membrane mimetic compartment ^ solvent interface, and to aggregate spontaneously into new compartments. Selfreplicating membrane mimetic compartment were, in fact, considered to provide examples for the ‘‘minimal life’’ or autopoiesis (Greek auto ¼ self, poiesis ¼ formation). Importantly, the demonstrated self-replications were con¢ned within the boundaries of the membrane mimetic compartments and they could, therefore, be regarded as models for the primitive cell and for the chemical origin of life [39]. Three di¡erent self-replicating reversed micellar systems have been described. The ¢rst system was based on the oxidation of octanol (OL, 10% v/v), located in the bulk organic solvent (isooctane, IO:octanol, OL ¼ 85:15, v/v) by permanganate ions (97.2 mM), contained in the aqueous pools of reversed micelles formed from sodium octanoate, OA (50 mM) and water (1.5 M), i.e., w ¼ ½H2 O=½OA ¼ 30 [35]. The hydroxide ion-catalyzed reaction, believed to occur at the micellar interface, resulted in the formation of octanoate ions, OA,which then associated into additional reversed micelles. In a typical experiment, the micelle concentration increased from 0.3 to 1.23 mM with a concomitant decrease of the w value from 30 to 13.7 [35]. Oxidation of octanol was found, however, to be incomplete, indicating the occurrence of side reactions. The second system was based on the lipase (enzyme)-catalyzed hydrolysis of trioctanoyl glycerol (TOG, 100 mM) in reversed micelles formed from sodium octanoate (OA, 50 mM) and water (such that w ¼ ½H2 O=½OA ¼ 20) in isooctane:octylamine ¼ 85:15, v/v. The enzyme

Biomimetic Nanoparticle Synthesis

437

was located in the water pool of the reversed miceles [35]. Hydrolysis of the TOG, at the micellar interface, produced OA, which associated into additional reversed micelles, and glycerol, which became solubilized in the aqueous pools of the micelles. Although the micelle concentration was found to increase signi¢cantly (from 33 to 82 mM,with the concomitant decrease of w from 20 to 6), glycerol accumulation, aminolysis, and other side reactions detracted from the usefulness of this self-replication. The third system (Fig. 6) rested on the base-catalyzed hydrolysis of a surfactant-forming ester, octyloctanoate, O-OL. Sodium octanoate, OA reversed micelles in isooctane, IO:octanol, OL (9:1, v/v) were shown to undergo self-replication by placing excess O-OL in the organic solvent and lithium hydroxide (2.8 M) in the aqueous pool, w ¼ 9:23 [36,40]. Interfacial hydrolysis produced the reversed micelle-forming surfactant (octanoate ions, OA) and the co-surfactant/co-solvent (octanol, OL), which spontaneously associated into reversed micelles. Once again, selfreplication resulted in the formation of smaller reversed micelles than their progenitors [36]. It is this last reversed micellar self-replication which we modi¢ed for cadmium sul¢de nanoparticle production. Cadmium perchlorate (instead of lithium hydroxide) was solubilized in the water pools of the NaOA/OL

FIG. 6 Schematics of the self-replication of octanoate ion-octanol (OA-OL) reversed micelles. Interfacial hydrolysis of octyloctanoate (O-OL), present in the isooctane solution by lithium hydroxide, present in the aqueous water pools, results in the formation of octanoate ions (OA) and octanol (OL) which associate into reversed micelles. Notice that while the micelle concentration increases, the sizes of the replicated micelles decreases. (From Ref. 24.)

438

Fendler

reversed micelles and the hydrolysis of O-OL was mediated by hydrogen sul¢de, which also served to generate the CdS (Fig. 7). Hydrolysis of octyloctanoate, O-OL, in the environment of NaOA reversed micelle resulted in the formation of octanoic acid, HOA and octanol, OL [36,40]: CH3 ðCH2 Þ6 COOðCH2 Þ7 CH3 þ H2 O ¼ CH3 ðCH2 Þ6 COO Hþ þ HOðCH2 Þ7 CH3 O-OL OL HOA ð1Þ

The products of the O-OL hydrolysis are the micelle-forming surfactant (octanoate ion, OA) and the co-surfactant (and/or co-solvent, OL) whose assembly into additional reversed micelles constitute the self-replication. Initially, the OA formed in reaction (1) distributed itself into the existing reversed micelles and/or into the solvent. This is expected to result in an increased of RH . Indeed, such increases (5^10%) were observed. However, in the absence of added water, additional amounts of OA (partaking in the dynamic disassembly and assembly of reversed micelles with the concomitant exchange of their water pools) [41] could only be accommodated in smaller reversed micelles (with smaller w values) having smaller RH .

FIG. 7 Schematics of 3-mercaptopropionic acid capped (indicated by the solid circle) CdS nanoparticle formation in self-replicating octanoate ion (OA) reversed micelles in 85:15 ¼ isooctane:octanol (OL), containing w ¼ ½H2 O=½NaOA ¼ 40. Addition of octyloctanoate (O-OL) and H2S containing Cd2 þ in their water pools results in the formation of OA and OL (and hence additional reversed micelles), and CdS. In the absence of added water, the reversed micelles formed are necessarily smaller than their progenitors.

Biomimetic Nanoparticle Synthesis

439

Observation of the decrease of the size of the reversed micelles provides, therefore, evidence for self-replication. We have monitored the O-OLmediated self-replication of reversed micelles by dynamic light scattering. Even in the absence of any additives, the hydrolysis of O-OL and the concomitant reversed micelle self-replication proceeded at a reasonable rate. The observed base-catalyzed self-replication in the presence of LiOH was found to be similar to that reported previously for much smaller reversed micelles ðw ¼ 9Þ [40]. Importantly, the self-replication of NaOA reversed micelles is also catalyzed by the acid generated by H2S in its interaction with the reversed micelle solubilized water pool. The hydrodynamic radii of 0.05 M NaOA reversed micelles in 85:15 ¼ isooctane:octanol were found to decrease from 4.8 to 3.9 nm upon the addition of 0.025 M O-OL and di¡erent amounts of H2S.Concurrently, the intensity of the ester band of O-OL at1741.8 cm1 was found to decrease by FTIR (Fig. 8). The self-replication of reversed micelles

FIG. 8 FTIR spectra of the ester band of O-OL in self-replicating 0.05 M NaOA reversed micelles in 85:15 ¼ isooctane:octanol (OL), containing w ¼ ½H2 O=½NaOA ¼ 40 in the absence (spectrum 1) and in the presence of 0.025 M O-OL and 0.0001 M H2S, 15 min (spectrum 2), 2 h (spectrum 3), 2 days (spectrum 4), and 3 days (spectrum 5) subsequent to the introduction of O-OL and H2S. Path length of the FTIR cell ¼ 0.05 mm.

440

Fendler

was somewhat faster in the presence of H2S than in its absence, albeit not as fast as when catalyzed by LiOH.The acidity (‘‘pH’’) in reversed micelles is, of course, quite di¡erent than that in bulk water [41]. No signi¢cance should be attached, therefore, to the observed di¡erences in the rates of O-OL hydrolysis with di¡erent amounts of H2S. Addition of H2S to cadmium perchlorate containing NaOA reversed micelles resulted in the prompt appearance of a yellow coloration, due to the formation of CdS. The restricted volume of the water pools and the ratio of the precursors (Cd2 þ and H2S) have been shown to control the sizes of the CdS particles formed [8,42,43]. Since we were interested in maximizing the yield of the nanoparticles produced, we chose relatively large reversed micelles with w ¼ 40. Varying x (where x ¼ ½Cd2þ =½H2 S), from 0.25 to 10 provided, however, satisfactory size control of the CdS nanoparticles. Absorption spectra of CdS nanoparticles formed in NaOA reversed micelles at x ¼ 0:25, 0.50, 1.0, 1.5, 2.0, 4.0, and 10 are illustrated in Fig. 9. Increasing the ratio of cadmium ion to H2 SðxÞ is seen to shift the absorption threshold to higher energies. Using the relationship between the CdS absorption edge and the size of the particles [42], we estimated the mean diameters of the CdS nanoparticles shown in Fig. 9 to be 5.4 nm (x ¼ 0:25), 4.8 nm (x ¼ 0:50), 3.5 nm (x ¼ 1:0), 3.2 nm (x ¼ 1:5), 2.9 nm (x ¼ 2:0), 2.6 nm (x ¼ 4:0), and 1.8 nm (x ¼ 10).These values are somewhat larger than those reported previously for the formation of CdS particles in AOTreversed micelles at w ¼ 40 (2.7 nm, x ¼ 0:25; 3.04 nm, x ¼ 0:5; 4.5 nm, x ¼ 1; and 2.75 nm, x ¼ 2) [44]. The CdS nanoparticles prepared in the NaOA reversed micelles were capped by 3-mercaptopropionic acid immediately subsequent to their formation. Capping is known to arrest the growth of the nanoparticles and permits isolation of the capped nanoparticles [45]. Indeed, the precipitated 3-mercaptopropionic acid-coated CdS nanoparticles could be readily dispersed in water. Their absorption spectra corresponded well to those observed for freshly generated CdS particles in NaOA reversed micelles (Fig. 9). In particular, the mean diameters of the 3-mercaptopropionic acidcapped CdS nanoparticles in water were assessed to be 6.0 nm (x ¼ 0:25), 4.2 nm (x ¼ 0:50), 4.8 nm (x ¼ 1:0), 4.0 nm (x ¼ 1:5), 3.5 nm (x ¼ 2:0), 2.7 nm (x ¼ 4:0), and 1.9 nm (x ¼ 10) (Fig. 9). Furthermore, aqueous dispersions of the capped CdS nanoparticles could be repeatedly isolated (by rotary evaporation), dried, and redispersed in water to give nanoparticles with the same absorption threshold. Isolation of the capped CdS particles as solids permitted the determination of the overall yield of CdS formation in NaOA reversed micelles. It was quite satisfactory (80 8%). Additional CdS nanoparticles could be generated from the supernatant remaining subsequent to harvesting the capped CdS particles. Addition of

Biomimetic Nanoparticle Synthesis

441

FIG. 9 Absorption spectra of uncapped CdS nanoparticles in 0.10 M NaOA reversed micelles in 85:15 ¼ isooctane:octanol (OL), containing w ¼ ½H2 O= ½NaOA ¼ 40 (A) and 3-mercaptopropionic acid capped CdS nanoparticles in water, subsequent to their removal from the reversed micelles as described in the experimental section (B). Ratios of x ¼ ½Cd2þ =½H2 S were varied as x ¼ 0:25 (spectrum 1), x ¼ 0:50 (spectrum 2), x ¼ 1:0 (spectrum 3), x ¼ 1:5 (spectrum 4), x ¼ 2:0 (spectrum 5), x ¼ 4:0 (spectrum 6), and x ¼ 10 (spectrum 7).

442

Fendler

cadmium perchlorate (4:9  103 M), water (6.9 M) and H2S (in amounts to obtain the desired x) to the supernatant OC reversed micelles resulted in the formation of CdS particles. In this way we have demonstrated that reversed micelles can be recycled for additional CdS nanoparticle formation.

4

NANOPARTICLE SYNTHESIS IN SURFACTANT VESICLES

Closed bilayer aggregates, formed from phospholipids (liposomes) or surfactant (vesicles), have also provided compartments for nanoparticles [8]. Swelling of thin lipid (or surfactant) ¢lms in water results in the formation of onion-like100^800-nm-diameter multilamellar vesicles (MLVs). Sonication of the MLVs above the temperature at which they are transformed from a gel into a liquid (phase-transition temperature) leads to the formation of fairly uniform, small (30^60-nm-diameter) unilamellar vesicles (SUVs; Fig. 10). SUVs can also be prepared by injecting an alcohol solution of the

FIG. 10 Schematics of the formation of multilamellar surfactant vesicles (MLV) by the dispersion of a thin films of surfactant in water (by vortexing) and its subsequent transition to small single-bilayer unilamellar vesicles (SUV) by sonication.

Biomimetic Nanoparticle Synthesis

443

surfactant through a small-bore syringe into water, by detergent dialysis, by ultracentrifugation, and by gel, membrane, and ultra¢ltration [8]. Once formed, the SUVs, unlike aqueous micelles, do not break down upon dillution. Surfactant vesicles, depending on composition and temperature of storage, remain stable for weeks to months. SUVs can organize many guests in their compartments. Hydrophobic molecules may be distributed among the hydrocarbon bilayers and polar molecules may move relatively freely in the vesicle entrapped water pools, particularly if they are electrostatically repelled from the vesicle inner surface. The binding of small charged ions to the oppositely charged outer and/or inner vesicle surfaces is facile. Species with charges identical to those on the vesicle can be anchored onto the surfaces by long hydrocarbon tails. This versatility permits a large variety of di¡erent approaches to the in situ generation of nanoparticles in SUVs. Large (giant, ca.1-mm diameter) unilamellar surfactant vesicles (LUVs) have also been prepared. SUVs have been stabilized by polymerization.Vesicle-forming surfactants have been functionalized by vinyl, methacrylate, diacetylene, isocyano, and styrene groups in their hydrocarbon tails or head groups. Accordingly, SUVs could be polymerized in their bilayers or across their head groups. In the latter case, either the outer or the inner or both the outer and inner surfaces could be polymerized. Surfactant vesicles and polymerized vesicles have been extensively employed as compartments for the in-situ generation of nanoparticles [8,46,47]. The use of surfactant vesicles for the compartmentalization of semiconductor nanoparticles and the construction of a sacri¢cial water photoreduction apparatus is best illustrated by Fig.11. For the sake of comparison, Fig. 11 also includes a schematic of plant photosynthesis. Natural photosynthesis is best understood in terms of the zig zag or Z scheme. Brie£y, light is harvested by photosystem I (PSI) and photosystem II (PSII). These two systems operate in series: two photons are absorbed for every electron liberated from water. Light-induced charge separation in PSII leads to the formation of a strong oxidant and a weak reductant. Electrons £ow from a weak reductant, via a pool of plastoquinone and other complex carriers, to a weak oxidant, which is generated along with a strong reductant, X, in PSI. This electron £ow is coupled to the NADH þ ^ NADH cycle (via ferrodoxin). Protons enter the ATPase to react with ADP to form ATP; with the aid of ATP, X reduces CO2 to carbohydrates (these latter cycles are not shown in Fig. 11). In the arti¢cial system (Fig. 11b), a polymerized surfactant vesicle is substituted the thylakoid membrane. Energy is harvested by semiconductors, rather than by PSI and PSII. Electron transfer is rather simple. Water (rather than CO2) is reduced in the reduction half-cycle to hydrogen,

444

Fendler

FIG. 11 (a) Schematic representation of a portion of the thylakoid membrane, thought to be involved in photosynthesis. Photosystem I (PSI), upon light (hn), absorption, produces a strong reductant and a weak oxidant. Photosystem II (PSII),

Biomimetic Nanoparticle Synthesis

445

at the expense of benzyl alcohol. In spite of these di¡erences, the basic principles in plant and biomimetic photosynthesis are similar. Components of both systems are compartmentalized. The sequence of events is identical in both systems: energy harvesting,vectorial charge separation, and reduction. 5

SELF-ASSEMBLY AND PATTERNING

Self-assembly of alternating layers of oppositely charged polyelectrolytes and nanoparticles (or nanoplatelets) is deceptively simple (see Fig. 12). Selfassembly is governed by a delicate balance between adsorption and desorption equilibria. In the self-assembly of nanoparticles, for example, the e⁄cient adsorption of one (and only one) monoparticulate layer of nanoparticles onto the oppositely charged substrate surface is the objective of the immersion step. Desorption of nanoparticles forming a second and additional layers (and preventing the desorption of the ¢rst added layer) is the purpose of the rinsing process. The optimization of the self-assembly in terms of maximizing the adsorption of nanoparticles from their dispersions and minimizing their desorption on rinsing requires the judicious selection of stabilizer(s) and the careful control of the kinetics of the process. Forces between nanoparticles (or nanoplatelets) and binder nanolayers (polyions or dithiols, for example) govern the spontaneous layer-bylayer self-assembly of ultrathin ¢lms.These forces are primarily electrostatic and covalent (for self-assembled monolayers, SAMs, of dithiol derivatives onto metallic surfaces) in nature, but they can also involve hydrogen bonding, p-p interactions, van der Waals attractions, hydrophobic, and epitaxial or other types of interactions. It is important to recognize that polyionic binders must have displaceable counterions in order to electrostatically bind them to the oppositely charged surface. The use of dithiols is only relevant with building blocks which incorporate accessible metal atoms, M (Au and

!

upon light (hn) absorption, produces a strong oxidant and a weak reductant. Electron flow from the weak reluctant to the weak oxidant is coupled to phosphorylation, which converts adenosine diphosphate (ADP) and inorganic phosphate (P) to adenosine triphosphate (ATP). With the aid of ATP, the strong reductant produced by PSI reduces carbon dioxide to carbohydrate. The strong oxidant, produced by PSII, oxidizes water molecules to molecular oxygen and protons, released in the inner membrane. Protons can be transmitted through channels going through the bilayers (indicated by the dotted area). Cholesterol (indicated by shaded ovals) adds to the rigidity of the membrane and proteins (indicated by shaded circles) may be bound to the surface. (b) In the mimetic system, water (rather than CO2) is reduced in the reduction half-cycle to hydrogen at the expense of benzyl alcohol. The location of the CdS/Rh particles is based on chemical experiments. (From Ref. 8, p. 9.)

446

Fendler

FIG. 12 Schematics of spontaneous self-assembly of a nanostructured film. A well-cleaned substrate is immersed into a dilute aqueous cationic polyelectrolyte solution for a time optimized for the adsorption of a 2.0 0.5-thick polymer, rinsed and dried. Next the polycation-coated substrate is immersed into a dilute dispersion of negatively charged nanoparticles for a time optimized for adsorption of a monoparticulate layer, rinsed and dried to form an organic^inorganic sandwich unit. Subsequent sandwich units are deposited analogously. The method is amenable to the fabrication of more complex superlattices.

Ag nanoparticles, for example), or semiconducting nanoparticles (MS and MSe, for example, where M ¼ Cd, Zn, Pb) in which covalent M ^ S bonds can be formed. The properties of the self-assembled multilayers depend primarily on the choice of the building blocks used, their rational organization, and integration along the axis perpendicular to the substrate. Sequential adsorption of oppositely charged colloids was reported in a seminal paper [48]. Self-assembly was subsequently ‘‘rediscovered’’ and extended to the preparations of multilayers of polycations and phosphonate anions [49^52], as well as to the layering of polyelectrolytes [50]. Construction of electrodes coated by polyelectrolytes, clays, and other materials often involved self-assembly [53] albeit the method had not been

Biomimetic Nanoparticle Synthesis

447

called such. Self-assembly is now routinely employed for the fabrication of ultrathin ¢lms from charged nanoparticles (metallic, semiconducting, magnetic, ferroelectric, insulating, for example), nanoplatelets (clays or graphite platelets, for example), proteins, pigments, and other supramolecular species [52,54^56]. That any of these species in any order can be layer-by-layer adsorbed is the greatest advantage of self-assembly. The oppositely charged species are held together by strong ionic bonds and form long-lasting, uniform, and stable ¢lms which are often impervious to solvents. No special ¢lm balance is required for the self-assembly; indeed, the method has been referred to as a ‘‘molecular beaker epitaxy’’ [57]. Furthermore, self-assembly is economical (dilute solutions and dispersions are used, and the materials can be recovered) and readily amenable to scaling up for the fabrication of large-area defect-free devices on virtually any kind and shape of surfaces.

6

NANOPATTERNING

Substrate patterning to submicroscopic length scale by soft lithography (chemical nanopatterning) has been demonstrated to provide a viable alternative to the fabrication of electronic and memory storage devices [58,59]. Attachment of two-dimensional arrays of self-organized nanoparticles to substrate provide a valuable lithographic mask [60^62]. Spinodal dewetting of thin polymer ¢lm-mediated phase separation permits patterning without a template [63]. Simply rubbing the surface results in patterns of well-aligned polymer lines whose width is controlled by the thickness of the polymer ¢lm.

7

CONCLUSION

Substantial progress has been made in biologically inspired preparations of nanoparticles and in their organization and self-organization into twodimensional arrays and three-dimensional networks. The potential applications of these systems as electronic, optical, and electrooptical devices and sensors will ensure the rapid continued progress of biomimetic materials chemistry.

ACKNOWLEDGMENTS Support of this work by the National Science Foundation is gratefully acknowledged.

448

Fendler

REFERENCES 1. GA Ozin. Adv Mater 4:612^649, 1992. 2. H Weller. Adv Mater 5:88^95, 1993. 3. G Markovich, P Collier, SE Henrichs, F Remacle, RD Levine, JR Heath. Accts Chem Res 32:415^423, 1999. 4. http://www.nano-gov. National Nanotechnology Initiative: The initiative and its implementation plan. 5. JH Fendler. Nanoparticles and Nanostructured Films. Preparation, Characterization and Applications.Weinheim: Wiley-VCH, 1998. 6. S Mann. Biomimetic Materials Chemistry. New York: VCH, 1996. 7. DA Bazylinski, AJ Garratt-Reed, RB Frankel. Microsc Res Technique 27:389^ 401, 1994. 8. JH Fendler. Membrane-Mimetic Approach to Advanced Materials. Berlin: Springer-Verlag, 1994, pp 1^151. 9. GA Ozin. Adv Mater 6:71^76, 1994. 10. GL Gaines. Insoluble Monolayers at Liquid-Gas Interfaces. New York: Interscience, 1966. 11. A Ulman. An Introduction to Ultrathin Organic Film from Langmuir-Blodgett to Self-assembly. Boston: Academic, 1991. 12. JH Fendler. Isr J Chem 33:41^46, 1993. 13. NA Kotov, ED Zaniquelli, FC Meldrum, JH Fendler. Langmuir 9:3710^3716, 1993. 14. XK Zhao, J Yang, LD McCormick, JH Fendler. J Phys Chem 96:9933^9939, 1992. 15. JH Fendler. Supramol Chem 6:209^216, 1995. 16. J Yang, JH Fendler,TC Jao,T Laurion. Microsc ResTechnique 27:402^411,1994. 17. JP Yang, FC Meldrum, JH Fendler. J Phys Chem 99:5500^5504, 1995. 18. JP Yang, JH Fendler. J Phys Chem 99:5505^5511, 1995. 19. XK Zhao, LD McCormick, JH Fendler. Adv Mater 4:93^97, 1992. 20. GD Rees, BH Robinson. Adv Mater 5:608^619, 1993. 21. D Langevin. Accts Chem Res 21:255^260, 1988. 22. MP Pileni, I Lisiecki, L Motte, C Petit, J Cizeron, N Moumen, P Lixon. Progr Colloid Polymer Sci 93:1^9, 1993. 23. MP Pileni. J Phys Chem 97:6961^6973, 1993. 24. P Dutta, JH Fendler, Langmuir 18:314^318, 2002. 25. J B.Nagy, D Barette, A Fonseca, L Jeunieau, PH Monnoyer, P Piedigrosso, I Ravet-Bodart, J-P Verfaille, A Wathelet. In: JH Fendler, I De¤ka¤ny, eds. Nanoparticles in Solids and Solution. Dordreckt,The Netherlands: Kluwer, 1996: pp 71^129. 26. M Meyer, K Wallberg, K Kurihara, JH Fendler. J. Chem Soc Chem Commun 90^91, 1984. 27. ML Steigerwald. Mater Res Soc Symp Proc 272:15^20, 1992. 28. Y Wang, M Harmer, N Herron. Isr J Chem 33:31^39, 1993. 29. Y Wang, N Herron. Res Chem Intermed 15:17^29, 1991. 30. I Lisiecki, MP Pileni. J Phys Chem 99:5077^5082, 1995.

Biomimetic Nanoparticle Synthesis

449

31. MP Pileni, N Moumen, I Lisiecki, P Bonville, P Veillet. In: JH Fendler, I De¤ka¤ny, eds. Nanoparticles in Solids and Solution. Dordrecht,The Netherlands: Kluwer, 1996, pp 325^378. 32. S Schlag, H-F Eicke, D Mathys, R Guggenheim. Langmuir 10:3357^3361, 1994. 33. A Robertson, AJ Sinclair, D Philp. Chem Soc Rev 29:141^152, 2000. 34. PA Bachmann, PL Luisi, J Lang. Nature 357:57^59, 1992. 35. PA Bachmann, P Walde, PL Luisi, J Lang. J Am Chem Soc 113:8204^8209, 1991. 36. PA Bachmann, PL Luisi, J Lang. Chimia 45:266^268, 1991. 37. R Wick, P Walde, PL Luisi. J Am Chem Soc 117:1435^1436, 1995. 38. P Walde, R Wick, M Fresta, A Mangone, PL Luisi. J Am Chem Soc 116:11649^ 11654, 1994. 39. PL Luisi, FJ Varela. Self-Replicating Micelles
A Chemical Version of a Minimal Autopoietic System. Dordrecht, The Netherlands: Kluwer, 1989, pp. 633^ 643. 40. PA Bachmann, P Walde, PL Luisi, J Lang. J Am Chem Soc 112:8200^8201, 1990. 41. JH Fendler. Membrane Mimetic Chemistry, Characterizations and Applications of Micelles Microemulsions, Monolayers, Bilayers, Vesicles, Host-Guest Systems and Polyions. New York: Wiley, 1982. 42. A Henglein. Chem Rev 89:1861^1873, 1989. 43. L Motte, C Petit, L Boulanger, P Lixon, P, MP Pileni. Langmuir 8:1049^1053, 1992. 44. MP Pileni. Langmuir 13:3266^3276, 1997. 45. ML Steigerwald, AP Alivisatos, JM Gibson, TD Harris, R Kortan, AJ Muller, AM Thayer, TM Duncan, DC Douglass, LE Brus. J Am Chem Soc 110:3046^ 3050, 1988. 46. MT Kennedy, BA Korgel, HG Monbouquette, JA Zasadzinski. Chem Mater 10:2116^2119, 1998. 47. NM Correa, HG Zhang, ZA Schelly. J Am Chem Soc 122:6432^6434, 2000. 48. RK Iler. J Colloid Interface Sci 21:569^594, 1966. 49. H Lee, LJ Kepley, HG Hong,TE Mallouk. J Am Chem Soc 110:618^620, 1988. 50. G Decher, JR Hong. Ber Bunsen-Ges Phys Chem 11:1430^1434, 1991. 51. Y Lvov, H Haas, G Decher, H Mohwald, A Kalachev. J Phys Chem B 97:12835^ 12841, 1993. 52. JH Fendler. Chem Mater 8:1616^1624, 1996. 53. AJ Bard, TE Mallouk. In: RW Murray, ed. Molecular Design of Electrode Surfaces. New York: Wiley, 1992, pp 270^312. 54. TE Mallouk, HN Kim, PJ Ollivier, SW Keller. In: G Alterti, T Bein, eds. Comprehensive Supramolecular Chemistry: Solid State Supramolecular Chemistry: Two- and Three-Dimensional Inorganic Networks, Part I. Layered Solids and Their Intercalation Chemistry. Oxford, UK: Pergamon, 1996, 7:189. 55. MA Correa-Duarte, M Giersig, NA Kotov, LM Liz-Marzan. Langmuir14:6430^ 6435, 1998. 56. G Decher. In: JP Sauvage, ed. Comprehensive Supramolecular Chemistry. Oxford, UK: Pergamon, 1996, 9:507^528. 57. SW Keller, HN Kim,TE Mallouk. J Am Chem Soc 116:8817^8818, 1994.

450 58. 59. 60. 61. 62. 63.

Fendler YN Xia, GM Whitesides. Annu Rev Mater Sci 28:153^184, 1998. YN Xia, JA Rogers, KE Paul, GM Whitesides. Chem Rev 99:1823^1848, 1999. YN Xia, JA Rogers, KE Paul, GM Whitesides. Chem Rev 99:1823^1848, 1999. WA Hayes, C Shannon. Langmuir 14:1099^1102, 1998. H Brune, M Giovannini, K Bromann, K Kern. Nature 394:451^453, 1998. AM Higgins, RAL Jones. Nature 404:476^478, 2000.

10 Biosensors Anthony J. Killard Cambridge Diagnostics Ireland Ltd., Galway, and Dublin City University, Dublin, Ireland

Malcolm R. Smyth Dublin City University, Dublin, Ireland

1

INTRODUCTION

The purpose of a sensor is to provide information about its external environment. Sensors are common throughout our lives. Sensors such as thermometers and pressure gauges give us information about the physical environment. Sensors such as pH meters give us information about their chemical environment.Biosensors are a speci¢c class of sensing device that make use of the interaction of biological or biochemical species as the means of sensing the environment. Biological systems are characterized by very selective interactions between molecules
for example, enzymes and their substrates, antibodies and their antigens, cell receptors and their ligands. The other important characteristic of these interactions is their a⁄nity. These biochemical species interact strongly with one another, endowing great sensitivity. These two properties
selectivity and sensitivity
are the main characteristics that make biosensors such powerful analytical devices. To be useful, these biochemical interactions must be converted into a measurable signal. As data are all handled and processed electrically by modern electronic equipment and electronic computers, it is most useful to convert these into electrical signals that can be quanti¢ed and processed. 451



  

                                   

                                                      !              "                              

   $ +                                                                                      <                          >                     >        @               

             "  

                                                                                                

  



  

   $ +                \        <             

                    

              @              "           >              >                       \      "                               _         `         

  

         `  !      <       

   

           

    

                 

   | \  ~ +€  +‚ƒ     >                  >  

                         `                  `        `            `       …† ‰    `@ Šƒ  ƒ
  ƒ!  !ƒ Šƒ          `  | \  ~     >   `      `       ‹ >     `         `   `>          `@    ` 

Œ    Šƒ  !ƒ Š


     !ƒ Šƒ

"   `                     `_  >                $ ƒŽ     `        



  

             !                                                  "                 #                                    $ ' *+

` ‘   ’             >               

  

  ‘          

                                               

 `                     \     “\  !\      ””>  ••  +‚† ƒ€

     `           

        >   >   <      –  | \     “\  !\             @                Œ    —  ˜€ 

                       ™     

  



           >  ’                       !   >         “         `    @  

|Šš!ƒ ƒ  !ƒ Š
 |Šƒ  ƒš!˜ "  ‘     @ š!˜  !

 š! 
˜ 



 

 !            !      @  

  !   
!

         <               \  ! ‘        ‘ ›€                \   \                                           +†… œ€ ‹    \                    "     

           >   

 `     Ž !                  > "                  `          !   —                

     ‹ > ‚€          |  | Ž    >   

        ’      ` $ ˜Ž _             ‘      

   

 `    |    ’                       ‘       `    |    ’                        \         

                     ‹™Ž     "                 ’ "Ž



  

     <       <        $ +                        <   >                  

                "                                                                    $ ' ?+

                $ ›Ž             `              —             

>                            †€ _ ‹™        `                  \   ‹™  

             `          ’             ‹™                      

  ‹™   "   ‹™            ž€ Ÿ            –"   "   `      >                               >     "      >      

  

%

        <  <    !                          $       

+                  

 $*    @ +                                                                \                  <          *          

        

                                  !          

                        €                          

          ƒœ +…… \¦§ Ž  `                                      +  +…˜ | +…€ _                                      

 !" # $ #

   

       \      \         —       

'

  

           ‹   

                       >       –                ””   ••  ”” ••                                                –           ””¨    •• 

&    

           >      

   

   >     "     >     >    !  >         

    >                   "                 

    "                 \               \    @** _ #  ©>                         

      

    > ©Ž      ‹Ž    ™Ž        >      ©‹Ž@ ©  ™  ©‹
©  ‹
©>           `        <                       

 

  

   

 ‹             "  >           

   

             “  >                      ©>                       ‹  ++        > Ž ©>                         ‹

  

*

  >                 >                                    | \  ~ +€ !  ` >    `      >     Œ               

 @    ` 

Œ    Šƒ


 Œ    !ƒ Šƒ

"        `       

   `   !              ` >       `       `@  !ƒ Š  +ƒŠƒ ƒ!ƒ Šƒ
—       ` >         —          `             +ƒ€ ‹ `   >            

     + ©>         @   >   —     `    ‘  >       `     !                   \     ™     >     

         ©>         `               ˜+€              

                     >     ˜ƒ€ @*@ \   —    \       >           >   

        \ ‘  ‹              >  "              `  > \     >     >             !  \                         

  ` >         ‘           

   

  $  +

 

~   ~   ~   `   …     … …       

       

~ ~ ~ `  … … …  

_ # 

>€   

  _   €      

@ ~@ ~@

`  € ><       > "`€          €      > "`€       > "`€       _ €       ‰<< <  ‹   <   

|   #   

@ ~@ ~@ @ ~@ ~@ †‚ ƒ ~@ ~@ @ ~@ @ ~@ †‚ ƒ

` 

$+;"     _          

@@ @

@*

* *‚ *? *ƒ *‡ *ˆ *Š @ @Œ

'

<   

 Ž  ' **

    

       

       

   ‰    ‰   

    "   >               

           >  ! 
<      

                 _  < <  >€   ‹

     €   _ €        >€   

@ ~@ @ ~@

@ ~@ @ ~@

† ƒ    # "`  €          #         #  

@ ~@

   €     

@ ~@

@ŠŒ

@ˆ

@‡

@*

@ƒ

@?

@‚

  





  

@* ‘  "  `          

           —   

                >        Π               >   

    

        `         |Šƒ  !                _ 

  ‘                           >  

        \                                         ª    ƒŽ ˜˜€   
            `                         …  ƒœ…  ˜›€ $+;"  "   "  \    `     >      _  \     \  >    "~ !     †     †    “  *

\  

 ‹ 

'  $ +

'  $ ‹  +

                 

   

…       
              
               †       †      

~@

*Œ

*Œ’@ŒŒ

~@ ~@ ~@ ~@   ~@ ~@  ~@ ~@ ~@ ~@

*Œ *Œ *Œ *Œ Œ @ ? *Œ *? ? *Œ 

@’@Œ ’ƒŒ @’@? ?’@Œ *Œ’*Œ Œ’‡ *Œ’*ŒŒ *ŒŒ’?ŒŒ ’ƒŒ *Œ’*ŒŒ ŒŒ?’* Œ?*’@??

           

  

ƒŒ ƒŒ

*Œ? ’? *Œ ’*Œ@

 Ž  ' 

  



@*‚        

                   

      

                 

    > `                             |                     `    >  !

`  

       Š      `            

  –                          `           \              ˜œ€ 

+=   

@@*               •                      

   $                    Ž     \                                `      
   

 ””¨••              \          @         ++€            

         >             <     Ž              

                $ 

     >    >                       >            

>                   ‹  <                     >                                      Ž   

464

Killard and Smyth

FIG. 5 Basic antibody structure. An IgG antibody is composed of two light (L) and two heavy (H) chain proteins held together by a series of disulfide linkages. The IgG molecule is bivalent and has two antigen-binding sites. These regions possess extreme structural diversity (VL and VH) from one type of antibody to the next, thus allowing many different types of antibody to interact with many different antigens. The Fc region (CH2 and CH3) interacts with elements of the immune system. FR, framework region; CDR, complementarity determining region. (From Ref. 36.)

Antibody Structure. The basic structure of an antibody is illustrated in Fig. 5 [36]. It is composed of two identical pairs of peptides held together by disul¢de bonds.The Fc portion of the antibody interacts with the immune system, while the Fab region is responsible for binding to antigen, more speci¢cally at the antigen-binding site. Antibodies are either divalent, capable of binding antigen at two sites, as illustrated here, or decavalent, capable of binding at 10 sites, as is the case for IgM antibodies. These valence properties are important for increasing the overall binding characteristics (avidity) of the antibody to the antigen. Antibody Production. A signi¢cant drawback with the use of biological systems in measuring devices is the inherent variability associated with biologically derived materials. Antibodies are no exception. The production methods used have an important impact on the quality and reproducibility of the antibody material, and so also have a signi¢cant impact on the assays. The development of improved methodologies has led to better-de¢ned reagents for use in biosensors [11,36].

  



           >                 

 `   <                               

                   \                                     !         >   `      `  

      ‘           > —               `        !    

       \   – 

Ž          

    >           

   Ž        

          

               $ ‚Ž                     ‘        ˜…    !      \     $    

              ‹         

          

  Ž           "                              š              

                 Ž _                                   š        

         

     ‘                       "                      `  +…ž   "      

          

                         >       !     >                   – 

           





  

   \            "        #       "                        

 

                                     # _      $    +                                 $ ' ƒ+

                

    

            

              



  
   –    

    

         `                                       

>         ©

 

  

%

            

        Š  

            

         

                                             

    `                 >        

 

                 

             Ž                               

  ‘       `       ˜‚ ›+€

                  >    Ž " 

>                        "    >                                 ‹          

    Ž |                         —                  Ž

        >            >      `  

  

       ›ƒ€ "         >    

       —                                     !     \                                             ‹    `                !         

       \         >   ›˜ ›œ€ @@@ '  |

                       \   

'

  

          



            

   ’  



   |

               ` 

            

     ¨                                                  ++€ @@ †    ™                     +‚…                  "     Ÿš           `     ‹     

        

  `     š                 "                                              \  

  \   ¬ ¨                                   @@‚ "   ‰ 

              

           @                        "   >     >           _  

       Ž       ‘     ‘    ›‚€ ™ >           ` \                        \                 >                              —"™‹Ž  

                ‘ !        ¨    •         

  



*

 & ;">;+! &&  ; @+$ # $"[# \>"

        >               @ + ƒ ˜ ›

                            

–       ¨   \  >  !        ‘                          >          

   $ †       ˜      >    

  <            

                 Ž         "  >  

                               \                !    

    ‘             ›†€ 

+ ]

        >    !                 \

           _       ‹   >           ‘      !              ‘           "             >         <               \                                           ›ž€ !              

^ _

"                     "           >     

 

   

%<

  

  % "      #     $+        <

            ‘                $"+       `       

                  $+ <   "                          $+ _    "              $_+ _    "           

              —                   \          \                            

              ‹       ›€ –"   >   `  `      ‹™      

           >                œ…€ 



`{_

"   \     \         `               `  

_         

      “          #          

        #<

   

         >       *?< <@‚<   # 

    

< 

\

 

    # 

$+;"  |   #    

‚Š’?Œ



_ # 

                   \      <                       

     

               |          

`                !                              |    

|          \                   # 

$
   +

’

‚ˆ



'

"       

‘    |         `                  \          

  

`    †         

 

  

%

‰   `   † ` 

   ‰   ‰   ‰_

`  <    `<    !   <"    ‰      <  ‰ ‹>

_  

\   _ 

"      

\

_   

    # 

$+;" 
  

>                            <  #              '   

    #     

     

 

  #       #    |    <         

|          \                  

  \        

 

"       

`         #  

'    #       

‰     <    ‰ <                        

   



_ #     

|         _ #              `        

     …         <  <   

  

** ƒ*’ƒ

‚‚’‚? ??’ƒŒ

?‚

?*’?

'

%   

  

%



 —          \

             ‰             \                                            \    `          –‹ Ž    

"]

    `         ””••  ””   ••                       \                           \                 >            "                     !        

     "     

                      ‘    \ ! 

             ‘     `       >                           ‹             >            `

   œ+ œ˜€ !  



                >           

                  ””  ••  `                Ž     `      |              !     ! †        

                          $   `                  >    `        ””  •• œ+€ "          `             

 

        

      

\  ‹                      

474

3.5

Killard and Smyth

Encapsulation

Encapsulation is here de¢ned as the trapping of a biomolecule behind or between membrane layers, either to bring it into contact with the transducer or compartmentalize various components of the assay. Encapsulation simultaneously brings about localization of the biomolecule while allowing the introduction of perm-selective properties by controlling membrane characteristics such as porosity, charge, and hydrophobicity. Encapsulation is an extremely common feature of biosensor con¢gurations and can be used in conjunction with all other immobilization methodologies. A common encapsulation con¢guration is the three-layer device in which an enzyme layer is sandwiched between an outer membrane layer (often composed of polycarbonate) acting as a di¡usional barrier and an inner membrane close to the transducer surface (often composed of cellulose acetate) as an interference rejection layer. This con¢guration has been used in commercialized glucose biosensors such as that used by YSI, Inc. (Fig. 8). Its impact on the performance of a creatinine biosensor was also illustrated, where this arrangement reduced the sensitivity of the measurement at the expense of providing much better exclusion of the interferent, creatine [54]. Such trade-o¡s in performance are a frequent characteristic of biosensor designs.

FIG. 8 The YSI electrode. Initially, a polycarbonate membrane limits diffusion of the analyte to the immobilized enzyme layer, where it undergoes oxidation. Hydrogen peroxide diffuses to the platinum electrode surface via a cellulose acetate membrane which excludes interferences such as ascorbate. (Courtesy of YSI, Inc.)

  



%

&= $  |

                            

                `                         >            Ž   `      

   \      

ƒ* `  <‰      <     ‹                                "    

    \                            š

                                  `         >           "\                             "    



   \     "      \         |  \                                             \              

              Š       \   >           —   \                `     — `            Š  `                  >          š Ÿ        Ÿ!Ž œœ€@ Ÿ!

|!˜ |!ƒ Š!  š Ÿ
 |!˜ |!Š  š Ÿ!  !

%

  

š Ÿ!   `>      ›…… ‰     @  š Ÿ  !  ƒ
 š Ÿ!


§ |

           $ Ž       

                         § |       `     \   \   \        \  Ÿ!  š Ÿ  Ÿ© © ` ` 

                      \   + _©+Ž     `  –‹    Ÿ!    

  *    <             $+ ‰   $"+ <                                                               $    +     $  <  +                    $ ' ??+

  

%%

_©ƒ       $

      ` 



                 ƒ…  ž……    

         ƒ@ `<  \  —           >                              

      ~  \               \          ‹           >     œ‚€ ƒ !   <"    " ~–  ~–Ž           \  |+žŽ       ‘           > Ÿ              ~–                   > Ÿ              ~–            •      $ +…Ž ~–                  œ†€ ƒ‚ ‰  _  ™    \                 Š                   œž€      \                          `                         >                             ™          `                !        !    ¨   _                     ! ™             >        >               >        ›› ›œ œ€                        

%'

  

  <      !   <"           

        #       !        #            #                                       #  _             

   ‚…€ "                    ƒ? _    "   |                  $       `             `       `      –          `     ++€                    ¨       —              !      ¨        "     >                     

  

%*

      Š                                   ‚+€ ™  Œ          $               •                                                  

  ¨   ‚ƒ ‚˜€ — 

      >           

     ‹ ›         \     

 

$!+#}>$ # &"$[ } "|||

"                                           >      

     Ž  

  Ž                  |  Ÿ|          !                     

          `      



  … ‰ "               ‘    `              

   ‹ ++Ž   

                 \       >  —  > `     

                  ‚**    _ #  _  $                    \   

                     ˜…               

         —           >          >    `  

           | \•      ` š               

'<

  

    `                 ~   ¦   ¦   \        

  Š  “                              >          $ ++Ž $                  "        

 \             Ž               

    …     $!    | +                                                      #                                

      $  !    | +

Biosensors

481

de¢ne the electrode area. Then an ink containing the enzyme glucose oxidase and the mediator potassium hexacyanoferrate (III) (ferricyanide) is printed onto the electrodes, followed by a second insulation layer. Finally, lamination layers cover the top of the strip, leaving the edge of the strip open. To use, a drop of blood is drawn in by capillary action when touched against the side of the strip and ¢lls the compartment. Glucose is converted to gluconolactone as shown in Sec. 1.1. The electron is accepted by ferricyanide which is reduced to hexacyanoferrate (II) (ferrocyanide). The ferrocyanide is oxidized at 400 mV versus carbon reference electrode to ferricyanide, which results in a current £ow dependent on the following (Cottrell) equation: nFAD1=2 C p1=2 t 1=2 where n is the number of electrons taking part in the electron transfer, F is the Faraday constant, A is the electrode area, D is the di¡usion coe⁄cient of ferrocyanide, and C is its concentration. As ferrocyanide at the electrode surface is consumed, a di¡usion layer forms at the electrode surface. As more ferrocyanide is consumed near the electrode surface, the current decays with time. The glucose concentration can be calculated from this formula. Two working electrodes are employed to ensure that the compartment is adequately ¢lled with blood sample. i¼

4.1.2 Potentiometric Enzyme Biosensors Potentiometric biosensors are not as widely employed as their amperometric counterparts. Such biosensors must normally utilize the movement of small molecules such as gases or ions into or across permselective layers as was illustrated for the urea biosensor (Sec. 1.1). Liberation of ammonia has been exploited in potentiometric biosensors for the detection of creatinine. Creatinine is the hydrogenated form of creatine, the major energy source for skeletal muscle. Using the enzyme creatinine iminohydrolase, ammonia gas is liberated which can be detected by an ammonia-sensing ion-selective electrode. However, these electrodes su¡er interference from endogenous ammonium. Various methods of immobilization have been used in such potentiometric systems for creatinine. Guilbault and Coulet [64] employed covalent coupling (acyl ^ azide) to collagen ¢lms and glutaraldehyde cross-linking with pig intestine membranes.The resulting membranes were placed over an ammonia electrode. Winquist et al. [65] covalently immobilized creatinine iminohydrolase (CIH) to acrylic beads in a reaction column upstream from a solid-state ammonia electrode. However, since the biological event and the sensing event are separated from one another, such a device is not strictly

482

Killard and Smyth

considered a biosensor. CIH has also been entrapped within polypyrrol and coated onto a pH electrode [66].Shin et al. [67] have entrapped CIH within a hydrophilic polyurethane membrane which was then applied to an ammonia-selective polymethane membrane coated onto a solid-state electrode. Another nonprotein nitrogen substance that can be measured potentiometrically is uric acid, which can be converted to allantoin by uricase: uricase

Uric acid þ O2 þ 2H2 O ) * allantoin þ CO2 þ H2 O2 The CO2 generated can be detected using a Hþ electrode with a gas-permeable membrane such as polypropylene or Te£on containing a bicarbonate bu¡er. CO2 di¡using through the membrane forms carbonic acid which dissociates to bicarbonate and Hþ : þ * H2 CO3 ) * HCO COðaqÞ þ H2 O ) 3 þH 4.1.3 Electrochemical Immunosensors Enzyme-based biosensors restrict the range of analytes that can be studied to clinical species, or rely on nonspeci¢c enzyme-inhibition methods. It has been shown that immunoassays can be exploited by biosensors in much the same way as they are in immunoanalysis, opening up a very wide range of analytes that can be studied. Tests such as the ELISA use antibodies and enzymes in combination to bring about the optical recognition of the antibody ^ antigen interaction. The same principles can be used in electrochemical immunoassays by substituting for a colorimetric substrate with an electrochemical redox mediator. However, immunosensors have been poorly exploited due to the complexities of reaction schemes required, as reacted and unreacted species must be separated from one another and nonspeci¢c interactions excluded by washing, etc. Several groups have used the novel application of biomolecular ¢lms to bring about immunosensors with much greater simplicity [43]. Duan and Meyerho¡ [68] used a gold-coated microporous nylon membrane as an electrode on which to immobilize immunoreactants labeled with the enzyme horseradish peroxidase (HRP) (Fig. 12). The mediator, hydrogen peroxide, was applied from the opposite side of the membrane to the immunoreactants. As it passed through the membrane, reaction with immobilized materials would occur ¢rst and little substrate would di¡use into the bulk solution to interact with unbound material. Killard et al. have exploited electrodes modi¢ed with conducting polymers to enhance similar separation-free principles [44,45]. Conducting polymers such as polyaniline show some useful properties for immuno-

  

'

      <                         #                                       

        #                           $ ' ƒˆ+

                                    >            " 

>      \        

                               > $ +˜Ž ! >      

                      `  >  !™    >         

                 `             

              \   ’ <    

 –               $ +›Ž

'

  

      <                       #                 <   <      <  #          #      #           ~    # <'‰   

           $ ' ‚+



]| $  |

Š                            Š     `  `                                   "                                                             "  `        



                   Š             ™

                      ~              

           

Biosensors

485

FIG. 14 The real-time binding interaction of biotin-HRP with anti-biotin antibody on an electrochemical immunosensor. A bulk solution of different concentrations of biotin-HRP with hydrogen peroxide were applied to the antibody-polyaniline electrode via a flow cell. As more biotin-HRP became bound to the electrode surface and took part in the coupled electron transfer, a gradual increase in current was seen in a fashion similar to the real-time antibody-binding interactions seen with other systems. (From Ref. 44.)

£uorescence, chemiluminescence, and bioluminescence may all be used in such devices. Further to this, the interaction of the light with the sensing element can lead to several modes of operation, for example, transmittance or re£ectance spectroscopy or attenuation of the evanescent ¢eld. Optical systems can be further classi¢ed as direct or indirect. Direct techniques involve analytes or products that bring about direct spectral changes, or even direct changes in mass, both of which can be measured by optical systems. Alternatively, analyte analogues or additional labeling reagents can be employed to bring about the required optical response, e.g., enzyme or £uorescent labels, such as the example given in Sec. 1.1. Whatever optical systems are employed, they all need materials that guide and control the transmitting and receiving of optical signals. The most widely used device for achieving this is the optical ¢ber. Composed of

'

  

           

                

   $ ›               ’                       ‚@* “   !     —                                         !  `    !                >      ‚€  >         >  

    |™ŒŽ     `    ™©‹Ž |         >  |™Œ   ™©‹        `                  ’ 

$ +œŽ                    >           … …       ƒ +… !   ‚…ƒ  —                      !   ’  "       ’    >          Ÿ               …›  !               \                                                   †…€          

                        Ÿ© ©—Ž |           

                               ›‚…  ™\             +  +……… §~      ‚@@    $                             —   

          

        `           Ž  ƒŽ˜ƒ €    `  ’   `    

  ›†…   ‚+…     "       

  

'%

                    #               #        <               

    ‰                #         #                 ƒŒ@  $ ' ƒŠ+

`      ’                       `  >  

 `        

 @

 ` 

~  Šƒ


   !ƒ Šƒ

|            `   ‹                 

    \         †+€     \                 `         "        >       `        `       ’   

''

  

                        `    `   ""Ž› †  + +… Ž˜   Žƒ Ž€ †ƒ€ ‚@      ~           `               `         ›ƒœ                 `                   ‹ ˜˜› †˜€                     —       >     ‘ `  

 `  

                               —      ƒ      ‚@‚ "     —                    Š >         `             œ‚ƒ  ~                          ‹   

              Ÿš           Ÿš                                 `    |  

   >       `                         †›€            |              "   `               ‹         \       

 >                    \          ‚@? |   '   `   " ’                  ’             ’     " ‹ ++        ’   

  

'*

                                      ‹                    ‘              

                               "                               `                               

  ‹™   "               

                                                                          

         †œ€  ‹™  

 –"             –         >     ` 

   |—|Ž `  |—| `           

             |   

           

               

            

                               $ +‚Ž ‹      `     \ >           

~|| $  |

 >  ™©Ž ‘           

 
    >
 

             

                                        †‚€             >          >      ­|—Ž       ‹ _Ž  ­|—       ƒ……Ž >  \         $ +†Ž_                  

      

                

*<

  

   `‰'             #                #        …                       `‰'                              ‘                                   `‰'          

       \         |      

     +… —!> !                    

           ‹  @                                      §ƒ $                       

       

  

*

  % ‰ #   $+   #    $”\+     $ @ŒŒ< + #           $"+ `     $`‘+                            "       <           #      

_ ‹ _                        $ +† Ž ‹      +…… ƒ…… —!>

*

  

_   ™©       

 

                   Ž $    ™©                   ¨       

        !                     ©      >    

              —              ›ƒ€ !             

     ™©         ””••       ‰   >      €     

   

         ‰             `    ™©‹Ž               \           Š                    ™©  \                  ‘                

  \         — \                                  

    "            ††€       †ž€        ™©                            

   
                 † ž…€                \    ‹           ‘        

 ž+€ Š            

  ’ 

     \       \       žƒ€



| $  |

™                +…€ "          

            \  

  

*

  '   #      #         

                           `                      #       

              #                       $ ' *Œ+

       —     `  ©>  

       ž… \¦§            >              ¨                 `  ‹  

           ‘     +…œ ª       +……            + — ©>       \          

     "      

     `    $ +ž        >  

   >   

           

      >         \ |  

*

  

    + ~   >                  ž˜€ ‹                            @                                                       >         

         

 #+; # $"

–       >                                                                                    ©    

                                 >                            

             

   

!""!"#" + ƒ ˜ › œ ‚

~| | \ ¦ | ~   š®  ‹ +…ƒ@ƒ +‚ƒ ‹¦ “\ Œ™ !\  ¦ š ƒ+›@ž‚ +‚† ŒŒ Œ   ¦ —   | ‹ +@ƒ+‚› +‚ Ÿ— ™ —© —‘  | œ@ƒ˜›œ +ž† ¦‹ ‹ > Œ ‹  – – ‹ @‚œ +† ¦‹ ‹ > "@ ™$  " ª  Œ‹ _    –   @ $      Š` “ª@ Š` “  ™  ‚˜ž ‚œœ † –™! ‹‘ Š‹ _    "@ ~¦ –  ™ |    –   ™      š ®\ — Ÿ\\ ++  +‚˜ +› ž –Ÿ —| "@ Ÿ Ÿ  ™   |  –  ‹   š ®\@ _  +ž  +œ ƒ˜›  ¦! ~ ŒŒ Œ   "@ ~¦ –  ™ |    –   ™      š ®\@ — Ÿ\\ ++  +…† +˜ž

  

*

+… – Ÿ  "@ ~¦ –  ™ |    –   ™       š ®\@ — Ÿ\\ ++  ž˜ +…œ ++  —|\ Œ ª

ª

 –— —  Š•ª "@ Ÿ Ÿ  ™   |  –  ‹   š ®\@ _  +ž  +˜˜ +› +ƒ Œ –  $ ‹¯ ™ |  "@ ~¦ –  ™ |    –   ™      š ®\@ — Ÿ\\ ++  † ›œ +˜ | –  Œ >  ~ +@+œ+ +ž‚ +› — š   ŒŒ Œ   šŒ Ÿ Š   | œž@œƒ˜ +ž‚ +œ |!  ‰ Œ –  ™ |   ~ ƒ…@œ+˜ +ž† +‚ — š   ŒŒ Œ   Œš Ÿ Š   |  +ž+@ƒ+

+ž‚ +† ®® ª  —‰ ™ \ ‰‹ ~ ‹‹ \ ~ ‹  | ›…@+‚œƒ +žœ +ž Ÿ ª  $ ‹

 – Š  Œ ¦   |  +†+@˜›œ +žœ + " ª  " ª   ~ +@‚‚† +ž‚ ƒ… “ _

  — ª $ ‹

 ‰ Ÿ — ¦ – ++@˜…† +ž˜ ƒ+ — —  Ÿ —   |  +†@›˜ +ž‚ ƒƒ ~ |

 —   Œ Ÿ   —™ ‹ — ||  |  +‚@+†œ +ž† ƒ˜ ~ |

 —   Œ Ÿ   —™ ‹    ++˜@††

+žž ƒ› — —  ™ |   |  ƒ˜+@ƒ† +… ƒœ ¦~ — ª ‹ " ª   |  ƒƒž@› +… ƒ‚ ª —  ! ‹< " ª  ‹ ‹>\ – – ƒƒ@+…†+ +ž… ƒ†    Ÿ ™‘ $ ‹

 — ¦  |  +˜›@˜œ

+žƒ ƒž    ª ® |  | ƒ@œ+ +ž˜ ƒ — \ ¦ ª< š —  ‹ ›@˜‚+ +žž ˜… ¦ ª< š — — \  ‹ ›@›† +žž ˜+  – Ÿ ‹  |  ›…+@œ + ˜ƒ ¦ _ ‰– š  ‹ ª ª  — ‹ ~     ›˜@+…˜ +‚ ˜˜ " ª  "@ ™$  " ª  Œ‹ _    –   @ $     Š` “ª@ Š` “  ™ +ž†  +˜ ƒ ˜› — š  | — ‹   – œ@+…… + ˜œ ¦ _ —‹ ~  | ‚…@+œ›œ +žž ˜‚ ¦ ª

 – Ÿ   Š•ª — ‹   | +›@ƒœ† +œ ˜† š| $  ¦© $ —¦ Œ "@ ©Œ |  –   @ ™  Š` “ª@ "~ +…  † +ƒ› ˜ž ¦‰ –  "¦ ! ™$  "@ ~¦ –  ™ |    –   ™      š ®\@ — Ÿ\\ ++  ›† ‚ƒ ˜ — > "@ ~¦ –  ™ |    –   ™      š ®\@ — Ÿ\\ ++  ƒ› ƒ‚‚

*

  

›… – !\  |  ˜›†@+†† +† ›+ ~ Œ  ™   — _ \ © _ \  –   –  +˜@++˜ +ž ›ƒ ™ <  "@ ! – ™!  ª    ~    –   —   –  Š` “ª@ ©  ‰  °‰ ›˜ ¦ ª

 — ‹  ~ ˜˜@+›œ+ ƒ……… ›› ¦ ª

 ‹ ± ! ±  ¦ ©" " — ‹  |  ›……@+… + ›œ ¦ ª

 ~ —  ª Œ — $\ ‰ ª  Ÿ —  " ™ 

— ‹  |  ›ƒ†@+†˜ ƒ……+ ›‚ ª ® " ª    | +ž@+ + ›† – ©  –   @  " |  “ª@ _  +‚  ˜+ œ… ›ž ©¦ ª  ! ® ® © ª \ — >  |  ˜›@ƒƒœ + › © ! !

 –    —  ª “ª@ Š “  ™ +…

 ƒ˜ ƒ œ… “ ¦²  "@ $ ‹

 Ÿ ‹   –   @ $        Œ–$ — ‰  +† _@ ‰|! ‰ 

++  ›‚† ›†‚ œ+ " Œ

–

    – +ž@ƒžƒ ƒ……… œƒ Š ~ ± _ ‹ – ‰ Œ > —  ¦ Œ ~   ‹ ‹  | — @ƒ˜œ› +† œ˜ ¦ _  |  ˜@ƒ+ + œ› —– — "| ™  ‹ “ ™ –\  |  ˜+@˜˜œ +‚ œœ ¦ª ™\ !¦ ® ª‹ ~ _® ‹ —| ª ! ª ‹  |  ˜…@ž˜ + œ‚  _\ ‹¦  ±  –  _™  –\    +ƒƒ@›˜ +† œ† !  < ª —    ©    @+‚+ +† œž ‹–  < ŒŒ _

    +ƒ+@‚ +‚ œ ¦ ª

 — ‹ ª Œ ~ —  Œ ™

  – ‹ ƒž@ž+ ƒ……… ‚… ¦ Œ  ‹Ÿ ™ ¦ ©  | ƒ†ž@ƒž† +… ‚+ ~© –  ‰ ™ \ ° ~ ! ª~ ª ‘  ©  $ ‰ +›@›˜ +‚ ‚ƒ – ~ — ‹  Š•ª    +ƒ+@ƒ +‚ ‚˜ ¦ ­ ™ ™ – $>\ – — ™ Ÿ

 ‹ Ÿ   Š•ª

—  ! ~ — — ª ~ –   –  +›@œž† + ‚› ŒŒ Œ   ™ |   |  +œƒ@ƒƒ˜ +ž˜ ‚œ $ _  " ~ ² – Ÿ   | œž@+›œ +ž‚ ‚‚  Š \ ‹ ª 

 ¦ ©  ‹ +›œ@›…‚ +ž ‚† ¦! ‹ ‹® ® "¦ ® ‹! | ‹Ÿ ~ ! š Œ‹ | ‹    – œ…@+ +ž ‚ž | Ÿ —© —‘  | ‚‚@+˜‚ +› ‚ —™ ° – ‰

< —Ÿ —>  —| —– $ –  $  –   –  +›@žœ ƒ………

  

*%

†…  ~ ‹ — š ‹ ™  \ \ ‹! | ª ® " ª   | ‚@ƒ…+† +† †+ ‹Œ " ¦ $  Ÿ _  –   –  +‚@+… ƒ……+ †ƒ Š‹ _   " Š š ™\ \ " ª  –   – 

+œ@‚ ƒ……… †˜ —|  —| < š Ÿ> ‹   – †˜@†+ ƒ……+ †› – ™ \ © –  š ‹ – \ ‹ —\  ‹   – †›@+ž ƒ……+ †œ ¦Œ ­  Š•ª — ‹ ¦ —   $ ¦ " — ƒ…‚@ž†

+† †‚ ™_ _  —© –  — Š•$  – ‹ +@›› ++ †† ©© ! ™>          ™Ÿ    “   š Š  š Š  ~ +žž †ž ¦ šš  ™! $  ‹‹ ª ŒŒ Œ   ¦  | ‹ +…ž@œ›››

+ž‚ † © ™ \‹> \ ¦! ~ ŒŒ Œ   ©> —  +ƒ@+†˜ +… ž… ! —  ª ª< ©  " ª   |  +žž@ƒœ† +ž‚ ž+ ª Ÿ  ~  | ‚+@+ƒƒ† +ž žƒ | ‹  ™ ‹\ ¯  –   –  +ƒ@+ +† ž˜ – ° ª  – Ÿ    | +@˜›… ƒ………

11 Modified Electrodes for the Oxidation of NADH Evelyne Simon and Philip N. Bartlett University of Southampton, Southampton, United Kingdom

1

INTRODUCTION

Enzymes are very powerful catalysts. Many of them can turn over more than 10,000 reactant molecules every second. They are also very speci¢c in the type of reaction they catalyse and the reactant molecules (or substrates) with which they react. Redox enzymes (that is, enzymes which catalyse oxidation of reduction reactions) exist to catalyse both simple electron-transfer reactions and the transfer of atoms, or small groups of atoms, to or from a range of substrates. Enzyme-catalyzed reactions are speci¢c because the reactants are held in speci¢c orientations within the active site of the enzyme. This requires that the substrate can reach the active site, even though this is often deeply buried within the enzyme, and that it contains the right functional group or groups to interact with the protein. Most enzyme reactions occur via an intermediate complex in a ‘‘Michaelis-Menten’’ type of mechanism [1]. This can be represented as shown in Fig. 1. Some redox enzymes need a small molecule, called a co-enzyme, in order to be active. The co-enzyme acts as an acceptor or donor of small groups or atoms or electrons and provides the driving force for the reduction, or oxidation, of the substrate. In this case the enzyme provides the active site which allows the substrate to react speci¢cally with the co-enzyme. This

499

500

Simon and Bartlett

FIG. 1

The Michaelis-Menten mechanism.

active site contains complexing groups which interact with the co-enzyme and substrate to bring the two into the correct relative orientation. Dehydrogenase enzymes are redox enzymes which transfer hydrogen atoms and electrons from a substrate to an electron acceptor. The electron acceptor is the co-enzyme nicotinamide adenine dinucleotide (NAD). The overall reaction is SH2 þ NADðPÞþ ! S þ NADðPÞH þ Hþ where SH2 represents the reduced substrate and S the oxidized substrate, NAD þ and NADH represent the oxidized and reduced forms of b-nicotinamide adenine dinucleotide (Fig. 2), and NADPþ and NADPH represent the oxidized and reduced forms of b-nicotinamide adenine dinucleotide phosphate (Fig. 2). In cells, the NAD(P) þ/NAD(P)H co-enzymes act as electron carriers and participate in oxidation ^ reduction reactions. NAD(P) þ is reduced by two electrons and one proton, becoming NAD(P)H (Fig. 2). These molecules can also be considered as hydride carriers (transfer of 2e and 1H þ or H). The binding of the co-enzyme to the dehydrogenase protein (the apoenzyme) is relatively loose, so these co-enzymes are not considered as ¢xed prosthetic groups, as in the case of the £avin in £avoproteins, but rather as substrates. Generally, NAD(P) þ/NAD(P)H binds to and dissociates from the active site of the enzyme during the catalytic cycle (Fig. 3). In most cases the co-enzyme must bind ¢rst (to give a complex referred to as the haloenzyme) before the substrate can bind productively. Following reaction between the co-enzyme and substrate, the product is released ¢rst, followed by the co-enzyme. The NAD(P)H co-enzymes participate in many important biosynthetic reactions, and are widely dispersed throughout living tissues. The two co-enzymes NADH and NADPH have very similar structures (they di¡er only in the presence of a phosphate group attached to the ribose in NADPH) and practically identical redox potentials. Despite this similarity, NADH and NADPH have di¡erent roles in the cell. NADPH works with enzymes that catalyze anabolic reactions, supplying electrons needed to synthesise energy-rich biological molecules. In contrast, NADH plays an important role as an intermediate in the system that generates ATP through the oxidation of foodstu¡s (metabolism) in the

Modified Electrodes for the Oxidation of NADH

501

FIG. 2 The structures of NAD(P)þ and NAD(P)H.

mitochondria within the cell. Within the mitochondria, 42% of the energy released by the oxidation of foodstu¡s is stored in molecules of ATP. NADH is found in large amounts in mitochondria, where it is essential for the conversion of ADP to ATP. The oxidation of one molecule of NADH by oxygen releases enough energy to synthesize several molecules of ATP from ADP and inorganic phosphate. Molecules of ATP are used elsewhere in the cell to provide the free energy needed to drive the thermodynamically unfavorable reactions required to build essential cell components: ATP þ H2 O ! ADP þ P

DG 0 ¼ 31 kJ=mol

The hydrolysis of the terminal phosphate of ATP liberates 31 kJ/mol [2]. Dehydrogenase enzymes are speci¢c for the transfer of only one of the two enantiopic protons (on C4) of NADH, and deuterium-labeling studies

502

Simon and Bartlett

FIG. 3 The mechanism for oxidation of a substrate by NAD þ catalyzed by a dehydrogenase enzyme.

have shown that the hydrogen is transferred directly from the substrate to the nicotinamide ring. Thus, hydride transfer via a single state is the accepted mechanism [3]. The electrochemical oxidation of NADH has attracted a lot of attention because of its potential application in a wide range of biosensors. More than 300 deydrogenase enzymes are known, each of them speci¢c to one substrate or group of substrates. An e¡ective system for the oxidation of NADH to enzymatically active NAD þ is the ¢rst step in exploiting these enzymes and their speci¢city in amperometric biosensors. In this chapter we will review the di¡erent chemically modi¢ed electrodes that have been shown to catalyze the electrochemical oxidation of NADH. 2

THE ELECTROCHEMICAL OXIDATION OF NAD(P)H

The electrochemistry of the NAD(P) þ /NAD(P)H couple has been extensively studied, in both aqueous and nonaqueous solutions. In aqueous solution the E0 values for NADH and NADPH at pH 7.0 are 0.561 and 0.565 V versus SCE, respectively [4], and both shift by 0.030 V for each unit change in pH as expected for 2e, 1H þ reactions. The electrochemistry of both couples is highly irreversible, and the oxidation of NADH at bare electrodes occurs only at high overpotentials. At low concentrations (from a few micromolar up to 10 mMÞ, Blaedel and co-workers did not observe fouling of the electrodes during steady-state voltammetry [5,6]. For concentrations of NADH >0.5 mM, the electrochemical oxidation occurs via the formation of

Modified Electrodes for the Oxidation of NADH

503

NADH þ radicals that can be absorbed at the electrode surface,the product, NAD þ , can also be adsorbed at the electrode [7^11]. These phenomena are not reversible and lead to the fouling of the electrode. Even if we set aside this problem, if this system were used to analyze biological samples, the high overpotentials required for the direct oxidation of NAD(P)H would result in signi¢cant problems of interference caused by the oxidation of other components in the sample (uric acid, ascorbic acid, proteins, etc.). Consequently, for practical applications it is necessary to ¢nd electrode surfaces, or mediators, which are catalytic for the oxidation of NAD(P)H. 2.1

Soluble Mediators

The use of an electron-transfer mediator can help to overcome the problems observed duirng the direct electrochemical oxidation of NAD(P)H. A mediator is a small molecule which shuttles electrons between the substrate and the electrode, thus catalyzing the electrode reaction. In the ¢rst step the mediator must react rapidly with NAD(P)H to give NAD(P) þ , in the second step the mediator must be reoxidized at the electrode surface at some potential less positive than that required for the direct oxidation of NADH, yet at the same time positive of E0 for the NAD(P)H/NAD(P) þ couple. The net e¡ect is then to generate an electrocatalytic current (Fig. 4). In e¡ect, the mediator reacts via an EC0 mechanism in which reaction with NADH in solution regenerates the reduced form of the mediator. Miller and coworkers [12^17] studied the oxidation of NADH by a wide range of oxidized molecules in aqueous solution in considerable detail and investigated the mechanism of the reaction. They studied the oxidation of NADH by a range of substituted ferricinium compounds and showed that these one-electron redox couples oxidized NADH via the unstable NADH þ radical at high overpotentials. More interesting mediators are the ortho- and para-quinone species because they combine fast reaction kinetics for NADH oxidation with low overpotentials. Ortho- and para-diaminobenzenes and diaminopyrimidines are also suitable candidates as mediators for NADH oxidation, for the same reason. Based on the linear free-energy relationship for the reaction and comparison with the results for the substituted ferrocenes, Miller et al. concluded that for the ortho- and para- quinones and diaminobenzenes the oxidation of NADH proceeded via either hydride transfer or by hydrogen atom transfer followed by electron transfer [18]. Based on the insight gained from such studies, it is possible to conclude that suitable mediators will be compounds for which hydride or hydrogen atom transfer is the preferred pathway (so that the high-energy radical intermediate is avoided, Fig. 5) and which show facile electrochemical reoxidation. Almost all of the mediators

504

Simon and Bartlett

FIG. 4

Oxidation of NADH by a homogeneous mediator.

used successfully in modi¢ed electrodes for NADH oxidation have these properties. The disadvantage of using soluble redox mediators is that they can di¡use away from the electrode surface and are thus no longer available to catalyze the electrode reaction. To avoid this it is desirable to con¢ne the mediator species in some way at the electrode surface. 2.2

Modified Electrodes

The simplest way to con¢ne the mediator species at the electrode surface is if it is part of the electrode surface itself. For example, it has been shown that a pretreatment of the electrode can lead to the mediation of NADH oxidation. Cenas et al. [19] studied NADH oxidation at bare glassy carbon electrodes pretreated in pH 7.0 phosphate bu¡er by cycling between high cathodic and anodic potentials.They showed that stable, redox-active groups were present on the electrode surface and they suggested that these groups were quinoidal in structure, based on the observed redox potentials and their pH dependence. In addition, it was already known that quinones can catalyze the oxidation of NADH (see above).

FIG. 5

A possible mechanism for the facile oxidation of NADH.

Modified Electrodes for the Oxidation of NADH

505

FIG. 6 The general scheme for the oxidation of NADH at a modified electrode.

Deliberate chemical modi¢cation of electrodes is an interesting, and more £exible, way to catalyze NADH oxidation. Chemically modi¢ed electrodes are widely used for catalysis of NADH oxidation. In operation, the mediator immobilized at the electrode surface passes the electrons from NADH to the electrode (Fig. 6). After oxidation of NADH by the immobilized catalyst, the oxidized form of the mediator must be regenerated at the electrode surface. Thus the catalytic cycle should proceed as described in Fig. 7, where the precise order in which the electron- and proton-exchange steps, required to reoxidize the mediator, occur may vary from mediator to mediator and with the experimental conditions (such as pH) [20]. In order to design this type of modi¢ed electrode, it is necessary to identify suitable mediators. One way to do this is to choose molecules which

FIG. 7 The catalytic cycle for the oxidation of NADH by an immobilized mediator species. (Reprinted from P. N. Bartlett and E. N. K. Wallace, ‘‘The oxidation of betanicotinamide adenine dinucleotide (NADH) at poly(aniline)-coated electrodes. Part II. Kinetics of reaction at poly(aniline)-poly(styrenesulfonate) composites,’’ J. Electroanal. Chem. 486, p. 23, Copyright 2000, with permission from Elsevier Science.)

506

Simon and Bartlett

oxidize NADH e⁄ciently in homogeneous solution (such as the quinones or diaminobenzenes discussed above) and which show reversible, or nearly reversible, electrode reactions. One then needs to tether these molecules in some way at the electrode surface. It is also important to study, and understand, the mechanism and kinetics, including the e¡ects of mass and charge transport within the mediator ¢lm, if one wishes to optimize the design of such modi¢ed electrodes. A wide variety of ways to immobilize mediator species at electrode surfaces have been described in the literature. Broadly, they can be divided into those methods which produce monolayers and those which produce multilayers. 2.2.1 Monolayers Mediators can be adsorbed onto bare electrode surfaces (graphite, glassy carbon, gold, etc.). Such adsorption is particularly e¡ective if the mediator contains groups that adsorb strongly, such as thiol groups on gold. Another way to form monolayers is to covalently attach the mediator to the electrode surface. To do this the electrode surface is ¢rst functionalized by generation of groups that will permit the subsequent covalent attachment of the mediator, for example, carboxylate groups formed on carbon electrode surfaces by oxidation or surface oxide ¢lms on metal electrodes. Alternatively, the surface may be functionalized by a strongly adsorbed species, such as a thiol compund on gold (Fig. 8), which has an amino or carboxylic group at the other end. In this case the sulfur strongly adsorbs onto the gold electrode surface, giving an electrode surface functionalized with NH2 or COOH groups. The mediator species themselves can then be covalently attached to the electrode by forming a chemical bond to the NH2 or COOH groups via a suitable functional group within the mediator, using a coupling agent [such as 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDAC)].

FIG. 8

Formation of monolayers on Au surfaces by covalent attachment.

Modified Electrodes for the Oxidation of NADH

507

Clearly, the concentration of mediator groups at the electrode surface when using the monolayer approach is limited by steric packing constraints. Formation of multilayers can overcome this problem. 2.2.2 Multilayers Polymeric ¢lms can be maintained at the electrode surface by chemical grafting, by chemisorption, or because of the low solubility of the polymer in the electrolyte solution. There are several di¡erent ways to deposit polymeric ¢lms onto electrode surfaces. A ¢lm can be cast from a solution of the polymer; in this case the polymer has to be soluble in an organic solvent but insoluble in the solvent/electrolyte medium used for the electrochemical experiments. Alternatively, the polymer ¢lm can be formed in situ by thermal, plasma, electrochemical, or photochemical polymerization. The latter approaches have the disadvantage that the polymer ¢lm is less well characterised. Using these methods, thick ¢lms, containing many monolayer equivalents, can be obtained. The most commonly used polymers for this purpose are poly(vinylpyridine), Na¢on, poly(styrene), and conducting polymers such as poly(pyrrole) or poly(aniline). Multilayers of adsorbed inorganic materials such as clays and zeolites can also be built up on electrode surfaces. The polymer ¢lm can act as the catalyst itself, or it can be a matrix within which to entrap or immobilize the mediator. In addition, the mediator may undergo polymerization itself, or be covalently linked to a monomer prior to polymerization, or can be co-immobilized during the polymerization process. In all cases, the ¢lm must be su⁄ciently porous to allow the substrate from the solution to di¡use through the layer in order to undergo catalytic reaction with the mediator. Another way to localize a mediator at an electrode surface is to use a carbon paste electrode [21]. In this case the mediator is mixed with carbon power and a binder (typically para⁄n oil) before being packed into a cavity to form the electrode.The main drawback of this method is the leakage of the mediator from the paste into the electrolyte solution. 2.2.3 Comparison of Monolayer and Multilayer Modified Electrodes The multilayer modi¢ed electrode has the advantage over the monolayer modi¢ed electrodes that there are many more mediator species present at the electrode surface (higher coverage). Thus, if there is any slow loss of mediator activity,this will be less severe for the multilayer modi¢ed electrode and consequently it should be more robust. In addition, the fact that there are more mediator species present at the surface of the multilayer modi¢ed electrode means that, to support the same total catalytic current, the rate of

508

Simon and Bartlett

reaction of the mediator with the substrate can be slower in the case of the multilayer modi¢ed electrode. Albery and Hillman [22] have analyzed this situation, and they conclude that for e⁄cient mediation in the monolayer case the second-order rate constant for the reaction of the mediator and the substrate should be greater than 10 4 dm3 mol1s1. In contrast, for the multilayer electrode the second-order rate constant need only be greater than 10 dm3 mol1s1. This is a signi¢cant bene¢t. However, to realize this bene¢t it is essential that the mediator/substrate reaction is the rate-limiting step. This means that the substrate must be able to di¡use freely within the polymer ¢lm in order to reach the mediator sites throughout the ¢lm. In addition, the charge transport between the meditor sites within the ¢lm must be fast, otherwise the mediator sites at the outside of the ¢lm will not be in e⁄cient communication with the electrode and therefore will be unable to react with the substrate. 3

MODIFIED ELECTRODES FOR NADH OXIDATION

Numerous techniques for the preparation of chemically modi¢ed electrodes have been widely used to make modi¢ed electrodes for NADH oxidation. In Table 1, the various modi¢ed electrodes reported in the literature since 1990 have been collected. For a review of earlier work, the reader should consult [23]. The use of ortho-quinones as mediators for NADH oxidation has been extensively studied. Pariente et al. have used, 3, 4-dihydroxybenzaldehyde as a mediator for NADH oxidation [24,25]. This compund was deposited at the surface of glassy carbon electrodes by electropolymerization (Fig. 9). The modi¢ed electrodes prepared, after an initial loss of activity, exhibit a good stability during hours of continuous potential cycling. In addition, these electrodes show electrocatalytic activity toward NADH oxidation, and the loss of activity is less than 5% in the presence of NADH. These authors also studied the mechanism of NADH and ascorbate oxidation at poly(3,4dihydroxybenzaldehyde) modi¢ed electrodes [26] and derived rate constants for the reactions using the model proposed by Gorton [18,27]. The same group have also used derivatives of 3,4-dihydroxybenzaldehyde [28]. They electropolymerized these derivaties on glassy carbon electrodes and showed that two of the derivatives had electrocatalytic activity toward NADH oxidation. These poly(3,4-dihydroxybenzaldehyde) modi¢ed electrodes have been used to make an aldehyde biosensor [29,30], and this polymer has also been deposited on carbon felt composite electrodes [31]. Monolayers of an imine derived from 3,4-dihydroxybenzaldehyde and 4-aminopyridine have been immobilized on polycrystalline platinum

3,4-Dihydroxybenzaldehyde derivatives

3,4-Dihydroxybenzaldehyde

Mediator

0.4 V vs SSCE

Self-assembly

32

25, 28, 30

 0.2 V vs SSCE

On carbon felt/epoxy composite

Electropolymerization

Ref.

24, 26 29 31

Comments

0.20 V vs SSCE 0.15 V vs SSCE aldehyde biosensor 0.2 V vs SSCE

Electropolymerization on GCE

Quinones

Imobilization method

TABLE 1 Modified Electrodes for NADH Oxidation

Modified Electrodes for the Oxidation of NADH 509

Adsorbed on gold

 0.1 V vs Ag/AgCl

75

68^70

0.05 V vs Ag/AgCl

Monolayers

Nitro-fluorenone derivatives

10-(30 -Methylthiopropyl)isoalloxazinyl-7-carboxylic acid

38

Ref.

 0.25 V vs SCE

Comments

Adsorption on GCE

Imobilization method

Pyrocatechol sulfonephthalein

Mediator

TABLE 1 Continued

510 Simon and Bartlett

0 V vs SCE

0.22 V vs SCE Lactate sensor

Modified Ag electrode

Monolayers on Au electrode

Graphite paste

R ¼ CH3 (Methyl) or CH2-C6H5 (benzyl) Histidine

PQQ

Tetracyanoquinodimethane

73

74

76

49, 108

0.2 to 0.2 V vs SCE

Adsorption

Methyl-/and benzylviologen

0.25 V vs SCE

71

0.05 V vs SSCE

Adsorption on Au

(5,50 -Dithiobis(2-nitrobenzoic acid)

Modified Electrodes for the Oxidation of NADH 511

Naphthol green

Adorption on carbon electrode

Nile blue

Electropolymerization

Adsorbed on paraffin-impregnated graphite carbon Zirconium phosphate Carbon paste and zirconium phosphate Electropolymerized

Silica gel Graphite-epoxy Siloxane polymer, covalently bound Sol-gel-derived carbon composite electrode

Redox dyes

Imobilization method

Meldola blue

Mediator

TABLE 1 Continued

50 51 52 54 53 35 41 48 49 55, 56 67

0.2 V vs Ag/AgCl 0.1 V vs Ag/AgCl 0.0 V vs Ag/AgCl Glyceride sensor 0.32 V vs Ag/Ag/Cl 0.05 V vs SCE 0.2 to þ 0.2 V vs SCE 0.0 V vs SCE  0.1 V vs SCE

Ref.

0.0 V vs SCE (pH 7.4) 0.0 V vs Ag/AgCl 0.0 V vs SCE

Comments

512 Simon and Bartlett

Thionine

Azure I

Toluidine blue O

Monolayer electrode coated with 3,30 -dithio bis(succinimidylypropionate) Monolayers on cysteamine Electropolymerization

Electropolymerisation Electropolymerization on glassy carbon

Carbon fiber microcylinder

Adsorbed on paraffinimpregnated graphite carbon Carbon paste toluidine blue O covalently bound through an amide linkage and an aqueous insoluble polymer Carbon paste

Graphite electrode

46 47

42, 43

57, 58 64

39

40 65, 66

0.1 V vs Ag/AgCl Ethanol sensor

0.05 V vs Ag/AgCl Ethanol sensor 0.2 V vs SCE Lactate sensor 0.1 V vs SCE 0.1 V vs SCE

0.24 V vs Ag/AgCl

0.15 V vs SCE  0.10.2 V vs Ag/AgCl

37

112 35 41

Glucose sensor Glyceride sensor 0.16 V vs Ag/AgCl

Modified Electrodes for the Oxidation of NADH 513

Immobilization on zirconium phosphate and incorporation in carbon paste Electropolymerization Thick film Laponite gel-poly(methylene blue) Electropolymerization pyrrole þ methylene blue Adsorption on graphite and incorporation in carbon paste Carbon paste Immobilized in carbon paste in presence of diaphorase Electropolymerization

Methylene green

Imobilization method

Methylene blue

Mediator

TABLE 1 Continued

60 61 62

0.0 to 0.1 V vs SCE 0.2 V vs SCE Glucose biosensors 0V vs SCE Lactate or alcohol biosensors  0.1 V vs SCE

44 45 97 63

0.2 V vs Ag/AgCl  0V vs SCE 0^0.25 V vs SCE  0.2 V vs SCE

83

48

Ref.

0.25 V vs SCE

Comments

514 Simon and Bartlett

Co-polymers of pyrrole and pyrrole derivatives substituted by quinone moities Co-polymer pyrrole and flavin reductase-amphiphilic pyrrole Co-polymerization of pyrrole and a pyrrole substituted by an isoalloxazine ring of riboflavin Poly(aniline)-poly(anion) composite films electropolymerized on glassy carbon electrodes

Poly(aniline)

Conducting polymers

Graphite modified electrodes

Poly(pyrrole)

Phenothiazine derivatives

79, 80

81 82

20, 77, 78

0.1 V vs SCE 0.1 V vs SCE

0.1^0.05 V vs SCE

36

 0.2 V vs SCE

þ 0.32 V vs Ag/AgCl

Modified Electrodes for the Oxidation of NADH 515

Osmium phenanthrolinedione

Carbon paste electrodes Adsorption on graphite

0.15 V versus Ag/AgCl 0.05 V vs Ag/AgCl

87 85

86 84

30

Ru: 0.05 V; Cr: 0.01 V; Co and Ni: 0.02 V; Fe: 0.05 V; Re 0.0 V vs SSCE Re and Fe 0.0 V vs SSCE Ru, Cr, Co, Fe, Ni, and Re 0.01 to þ 0.05 V vs SSCE

Electrodeposition

Transition metal complexes of 1, 10-phenanthroline-5,6-dione (phendione) Carbon paste electrodes Adsorption on glassy carbon electrode

90

Ref.

Co:  0.1 V vs SCE

Comments

Electropolymerization

Metal complexes

Imobilization method

Zn, Ni, and Co tetraminophtalocyanine

Mediator

TABLE 1 Continued

516 Simon and Bartlett

Highly boron-doped diamond

Ruthenium Titanium or zirconium

Nickel hexacyanoferrate

Cobalt hexacycanoferrate

[Os(bpy)2(PVI)10]Cl bpy ¼ 2,20 -bipyridine and PVI ¼ poly(vinylimidazole) polymer Catechol-pendant terpyridine complexes Co, Cr, Fe, Ni, Ru and Os

Dispersed carbon paste Sol-gel carbon composite electrode Microwave plasma-assisted chemical vapor deposition

 0.6 V vs SCE No fouling of the electrode

0.58 V vs SCE

92 93

0.52 V vs SCE 0.55 V L-lactate sensor Alcohol sensor  0.5 V vs Ag/AgCl, (pH 8.1) þ 0.2 V vs Ag/AgCl (pH 7.4)

95, 96

113 94 53

91

Co: 0.3 V

Modified electrode

0.48 V vs SCE

89

Co: 0.25 V vs Ag/AgCl

Electrodeposition of thin film on microband gold electrode

88

þ 0.1 to 0.5 V vs Ag/AgCl pH ¼ 7.4

Carbon fiber microelectrodes Os-polymer electrode (drop-coating) In solution

Modified Electrodes for the Oxidation of NADH 517

Diaphorase and ferricyanide or 2,6-dichloroindophenol

Diaphorase and methylene green or Meldola blue Diaphorase and methylene blue

Mediator

TABLE 1 Continued

Immobilized in carbon paste in presence of diaphorase Diaphorase immobilized in laponite gel and methylene blue electropolymerized on the laponite-diaphorase electrode Covalently attached to a tin(IV) oxide electrode

Enzyme þ Mediator

Imobilization method

97 98

99

0 V vs SCE

þ 0.35 V vs Ag/AgCl

Ref.

0^0.25 V vs SCE

Comments

518 Simon and Bartlett

Modified Electrodes for the Oxidation of NADH

519

FIG. 9 The electropolymerization of dihydroxybenzaldehyde derivatives. (Reprinted from F. Pariente, F. Tobalina, M. Darder, E. Lorenzo, and H. D. Abruna, ‘‘Electrodeposition of redox-active films of dihydroxybenzaldehydes and related analogs and their electrocatalytic activity toward NADH oxidation,’’Anal. Chem. 68, pp. 3135^3142, Copyright 1996, with permission from The American Chemical Society.)

electrodes, and the resulting modi¢ed electrode lowers the overpotential for NADH oxidation by 300 mV [32]. Although quinones have attracted a lot of attention, in general, they lose activity over a longer or shorter-period depending on the experimental conditions. According to Jaegfelt and co-workers [33], this loss in activity is mainly the result of side reactions. Jaegfelt et al. suggest that the formation of a chemical bond between the bulky NADH molecule and a quinone site on the electrode leads to the ‘‘blocking’’of several adjacent catalytic sites at the electrode surface in such a way that, although the adjacent quinone sites are still electroactive, they are no longer accessible to other molecules of NADH. Another group of compounds that has been widely used for the mediation of the electrochemical oxidation of NADH is the redox dyes,

520

Simon and Bartlett

phenoxazines and phenothiazines. All of these compounds contain the structural features required for e⁄cient mediation of NADH oxidation described in Fig. 4.Various immobilization methods have been used to prepare electrodes modi¢ed by these compounds. A number of dyes have been immobilized by adsorption on graphite electrodes; examples include meldola blue, nile blue, toluidine blue, methylene blue [34,35], and phenothiazine derivatives [36]. Some have also been immobilized on carbon ¢bers [37]. Pyrocatechol sulfonephtalein (catechol violet) has been adsorbed on a glassy carbon electrode and has shown catalytic activity toward NADH oxidation [38]. This method of immobilization has the disadvantage that the adsorption is reversible and not especially strong, and so desorption and a rapid loss of activity occur. This problem can be overcome by ¢rst depositing a compound that adsorbs strongly at the electrode surface and by then covalently attaching the dye to this compound. For example, cysteamine and dithiobis(succinimidylpropionate) have both been used to attach monolayers of thionine at gold electrode surfaces [39,40]. In order to improve adsorption, the dyes can also be immobilized onto graphite impregnated with para⁄n oil [41]. Alternatively, the compounds can be mixed with carbon and oil and used to form a carbon paste electrode [42^45]. The main problem with these carbon paste electrodes is the steady leakage of the mediator from the electrode. Di¡erent approaches have been tried to avoid this leakage. Before incorporation into the carbon paste, the dye can be covalently linked to a polymer,which is then incorporated in the paste [46,47].The dye can be ¢rst adsorbed on zirconium phosphate and then incorporated into the carbon paste [48], or the carbon paste can be formed by mixing carbon, mediator, zirconium phosphate, and para⁄n oil [49]. In the latter case,the catalytic currents obtained were higher than those for the carbon paste without ziconium phosphate. Titanium phosphate has also been used in the same way with meldola blue [50]; in this case the modi¢ed electrodes showed good stability, with 500 measurements being carried out with the same electrode. N-methyl-phenazinium, 1-methoxy-N-methylphenazinium, and meldola blue have all been immobilized as tetrarhodonato-diammine-chromates (Reineckates) in graphite-epoxy composite electrodes. These modi¢ed electrodes are very stable, allow more than 9500 measurements, and have been used in an alcohol biosensor [51]. A siloxane polymer containing covalently attached meldola blue deposited on graphite electrodes has also shown catalytic activity toward NADH oxidation. These modi¢ed electrodes operate at low overpotential and show good stability [52]. Meldola blue has also been immobilized in sol-gel-derived composite electrodes [53,54]. Another method to immobilize redox dyes at an electrode surface is by the electropolymerization of the dye itself from a solution of the

Modified Electrodes for the Oxidation of NADH

521

monomer. Poly(nile blue) [55,56], poly(toluidine blue) [57^59], poly(methylene blue) [60^62], poly(methylene green) [63], poly(azure I) [64], poly(thionine) [65,66], and poly(naphthol green) [67] have been successfully deposited and show catalytic activity toward NADH oxidation, although the chemical structures of the deposites are often hard to establish. Various other organic compounds have been used as mediators for NADH oxidation. Recently it has been shown that monolayers of nitro£uorenone derivatives, following reduction, catalyze NADH oxidation at very low potential (around 0.15V versus SCE) [68^70]. Dithiobis(2nitrobenzoic acid) [71], TCNQ [72,73], PQQ [74], methyl- and benzylviologen [49], isoalloxazine derivatives [75], histidine [76], PQQ [74], and TCNQ [72,73] are also catalysts for the electrochemical oxidation of NADH. Conducting polymers are an attractive way to immobilize catalysts on electrode surfaces. They can be deposited electrochemically, allowing the localization and thickness of the deposit to be controlled
a useful feature if one wishes to build arrays of selectively modi¢ed, yet closely spaced, microelectrodes. In addition, it may be possible to use the conducting polymer backbone as a conduit to transfer charge from the underlying electrode surface to the individual mediator sites. In our group we have shown that poly(aniline) is a catalyst for NADH oxidation at pH 7.0. When poly(aniline) is doped with poly(anions) such as poly(vinylsulfonate), poly(styrenesulfonate) or poly(acrylate), it remains electroactive at pH 7 and can catalyze NADH oxidation at 0.1 V versus SCE [20,77,78]. Conducting polymers can also be used as a host matrix for the mediator. There are di¡erent ways to incorporate mediators within the polymer: as a counter-ion during the electropolymerization process or covalently attached to the monomer and then electrochemically polymerized in the presence or absence of nonsubstituted monomer. All of these methods have been used with pyrrole to prepare modi¢ed electrodes for NADH oxidation. Figure 10 illustrates the di¡erent approaches. Co-polymers of pyrrole and pyrrole derivatives substituted by quinone moieties [79,80] or derivatives substituted by £avin reductase [81] or isoalloxazine [82] all catalyze NADH oxidation. Methylene blue has been immobilized in poly(pyrrole), and the ¢lm obtained catalyzes NADH oxidation at about þ 0.1 V versus SCE [83]. Metal complexes form another group of interesting mediators for the electrochemical oxidation of NADH. Ligands containing a quinone moity (such as 1,10-phenantroline-5,6-dione and its derivatives) have been used in several di¡erent metal complexes. These metal complexes have been electrodeposited [30], adsorbed [84,85], or incorporated in carbon paste [86,87], and all catalyze NADH oxidation around 0 V versus SSCE. An osmium complex, [Os(bpy)2(PVI)], where bpy is 2,20 -bipyridine and PVI is a poly

522

FIG. 10

Simon and Bartlett

Methods for the immobilization mediators in poly(pyrrole) films.

(vinylimidazole) polymer, has been coordianted to poly(vinylimidazole) and deposited (by drop-coating) on carbon ¢ber electrodes. These modi¢ed electrodes can mediate NADH odixation between 0.1 and 0.5 V versus Ag/ AgCl at pH 7.4 [88]. A catechol-terpyridine complexed to cobalt and deposited on an electrode can also catalyze NADH oxidation at 0.3 V versus SCE. Catechol-pendant terpyridine complexes (Co, Cr, Fe, Ni, Ru, and Os) have been electrodeposited onto glassy carbon electrodes. The cobalt complex catalyzed NADh oxidation around 0.3 V versus Ag/AgCl [89]. Cobalt tetraminophtalocyanine has been electropolymerized on glassy carbon electrodes. The ¢lm catalyzes NADH oxidation at around 0.1 V versus SCE [90]. Cobalt hexacyanoferrate and nickel hexacyanoferrate have also been used as catalysts, but they operate at much higher potentials (around þ 0.5 V versus SCE) [91^93], consistent with the observation that one-electron redox couples are less e⁄cient oxidants for NADH because the reaction must pass through the unstable NADH radical cation intermediate. Ruthenium dispersed in graphite oxidizes NADH at þ 0.6 V versus Ag/AgCl at pH 8.1 [94]. Tetraethyl orthotitanate or zirconium isopropoxide immobilized in sol-gel carbon composites catalyzes NADH oxidation at about 0.2 V versus Ag/AgCl at pH 7.5 [53]. Recently, diamond electrodes have also been shown to exhibit catalytic properties toward NADH oxidation [95,96]. Diaphorase is an enzyme that catalyzes NADH oxidation by a broad range of redox compounds, and the enzyme has been used to prepare modi¢ed electrodes for NADH oxidation. For example, diaphorase has been immobilized into a carbon paste electrode with methylene green or meldola

Modified Electrodes for the Oxidation of NADH

523

blue as redox mediators [97]. The oxidation of NADH is observed at 0^ 0.25 V versus SCE at these modi¢ed electrodes. Carbon paste electrodes containing methylene green or meldola blue alone also catalyze NADH oxidation at the same potential; the advantage of incorporating the diaphorase is that the e⁄ciency of the modi¢ed electrode is increased. Diaphorase has also been immobilized into laponite gel and methylene blue and subsequently electropolymerized onto the laponite-diaphorase electrode [98]. This modi¢ed electrode catalyzes NAD þ oxidation at 0 V versus SCE at pH 7 and is e⁄cient at NADH concentrations below 100 mM. Diaphorase has also been covalently immobilized in the form of a monolayer on a tin(IV) oxide electrode, and then used for NADH oxidation at þ 0.35 V versus Ag/AgCl in the presence of ferricyanide or 2,6-dichloroindophenol [99]. 4

KINETIC MODELS

Kinetic models of chemically modi¢ed electrodes have attracted a lot of attention over the last two decades, because of the important potential applications of these electrodes in analytical chemistry. A common assumption in these kinetic models is that the rate of electron transfer at the electrode surface to, or from, the mediator molecules in the ¢lm in the layer adjacent to the electrode surface is fast. Anson has shown that this step is unlikely to be rate-limiting, based on consideration of the Marcus theory [100]. The most commonly suggested model for the oxidation of NADH is a two-step reaction mechanism, similar to the classical Michaelis-Menten mechanism. In this mechanism, NADH ¢rst forms a complex with the oxidized form of the mediator; this is followed by a second step in which the complex breaks down to give NAD þ and the reduced form of the mediator (Fig. 11). Tanaka et al. have demonstrated the formation of a complex during chemical oxidation of an NADH model compound (1-benzyl-1,4-dihydronicotinamide) by p-benzoquinone derivatives [101] and Gorton and coworkers have shown the formation of a charge-transfer complex between meldola blue and NADH at pH 7 [18].

FIG. 11

The commonly proposed reaction mechanism for NADH oxidation.

524

Simon and Bartlett

Gorton et al. have studied NADH oxidation at electrodes modi¢ed by dyes [18,27] and proposed an equation for the catalytic currents based on the Koutecky-Levich equation:   1 1 KM 1 1 ¼ þ þ 0 ð1Þ J kþ2 G kþ2 G kD S1 where KM

  k1 þ kþ2 ¼ kþ1

ð2Þ

and S1 is the bulk concentation of NADH, G the coverage of mediator sites, k0 D is the mass transfer rate constant for the rotating disk electrode (k0 D ¼ 1.554 D2/3n 1/6 W1/2 where D is the di¡usion coe⁄cient, n is the kinematic viscosity,W is the rotation speed in hertz [102]), and j ¼ i=nFA is the steadystate £ux. Equation (1) has been used to determine kinetic parameters for NADH oxidation at electrodes modi¢ed by dyes electropolymerized at the electrode surface [56,58]. Some of the corresponding data are collected inTable 2.This treatment is strictly applicable only at low concentrations of NADH, because the derivation of the Koutecky-Levich equation assumes that the surface reactions are ¢rst-order in substrate. This is true only for the Michaelis-Menten type of reaction mechanism given in Fig.11 at low NADH concentration ([NADH]k1/k1), because at high NADH concentration the reaction becomes zero-order in NADH. The derivation of Eq. (1) also TABLE 2 Kinetic Data for Some Modified Electrodes for the Electrochemical Oxidation of NADH kþ2 k1 ð½NADH¼0Þ (dm3mol1s1) (s1) Meldola blue (adsorbed) 1,2-Benzophenoxazine-7-one (adsorbed) PQQ Poly(3,4-dihydroxybenzaldehyde) Poly (nile blue) Poly (toluidine blue o) a

(2.7  104)a (1.1  103)a

(2.3  102)a (4.3  103)a

Calculated from k þ 2/KM.

KM (mM)

Conditions Model

Ref.

1.1

pH 7

Gorton

18

0.21

pH 7

Gorton

27

25 109 0.70 0.14

pH 7 pH 7

Gorton Gorton

74 26

31.8 0.22

6.3  102





1.8  103





pH 6.8

Koutecky- 56 Levich Koutecky- 57 Levich

Modified Electrodes for the Oxidation of NADH

525

FIG. 12 Reaction scheme proposed by Lyons et al. (Reprinted from M. E. G. Lyons, C. H. Lyons, A. Michas, and P. N. Bartlett, ‘‘Heterogeneous redox catalysis at hydrated oxide layers,’’ J. Electroanal. Chem. 351, pp. 245^258, Copyright 1993, with permission from Elsevier Science.)

assume that there is no concentration gradient for NADH within the ¢lm and is therefore restricted to monolayers or, at best, thin ¢lms. Lyons et al. have given a full analysis for the mechanism proposed in Fig. 12 [103]. This analysis makes several assumptions: that the reaction occurs in a thin layer at the modi¢ed electrode/solution interface so that concentration polarization within the layer can be neglected; that electron transfer between the electrode and the mediator ¢lm is fast; and that there is no product inhibition. The e¡ect of concentration polarization in the electrolyte solution is explicitly included in the theoretical analysis. The overall reaction involves the following steps: 0 kD

S1 ! S0 k1

B0 þ S0 ! ½BS0 k1

k2

½BS0  ! ½AP0 k2 k3

½AP0  ! A0 þ P0 k3 kE0

A0 ! B0 0 kE

P0 ! P1 0 kD

526

Simon and Bartlett

where S0 and S1, P0 and P1 represent the substrate and product at the ¢lm/ solution interface and in the bulk of the solution, respectively, and B0 and A0 are the oxidized and reduced mediators, respectively. If the total concentration of the mediator in the ¢lm is cS , it can be related to the surface coverage, G, by G ¼ cS L, where L is the thickness of the mediator layer. Solution for the steady-state £ux, j, gives

(reprinted from Ref. 103 with permission from Elsevier Science)

where 1 1 k2 1 ¼ þ þ kc k1 k2 k3 k3 and

KM ¼ kc

1 k1 k1 k2 þ þ k1 k1 k2 k1 k2 k3

ð4Þ  ð5Þ

Usually the assumption made is that the ¢rst term is equal to 1. Then Eq. (3) becomes 1 1 1 KM 1  0 þ þ þ 0 ð6Þ j kE G kc G kc GS1 kD S1 This assumption is valid only if the currents are far lower than the solution mass transport limited current
in other words, this means that the assumption is valid only if the electrocatalysis is ine⁄cient. If we further assume that the kE0 electrode kinetic term is not rate-limiting, Eq. (6) reduces to Eq. (1) used by Gorton et al. to study NADH oxidation at phenoxazine or phenothiazine modi¢ed electrodes. The models discussed above are restricted to monolayers, or at best thin mediator ¢lms, because they neglect concentration polarization within

Modified Electrodes for the Oxidation of NADH

527

the ¢lm at the electrode surface. In our group, we have been using poly(aniline)-poly(anion) composite ¢lms for the mediation of the electrochemical oxidation of NADH [20,77,78]. These ¢lms can be much thicker, and our experimental data show that NADH oxidation occurs throughout the whole ¢lm and is not restricted to a thin layer at the polymer/solution interface. Thus the models described above are inappropriate even though the mechanism proposed for the reaction between the NADH and mediator is the same. To take account of the reaction occurring throughout the whole ¢lm, as well as the possibility of product inhibition, a new model was developed. This model is described below. 0 kD;S

NADHbulk )* NADH0

solution mass transport

NADH0 )* NADHfilm

partition into the film

KS

KM

NADHfilm þ fsiteg )*fNADHg kcat

þ

þ

fNADHg ! fNAD g þ H þ 2e þ

Ki

fNAD g )* NADþ film þ fsiteg Kp þ * NADþ film ) NAD0 0 kD;P þ * NADþ 0 ) NADbulk

adsorption to a site within the film 

oxidation of bound NADH desorption of product partition into the film solution mass transport

where the subscripts bulk, 0, and ¢lm refer to the bulk solution, the interface between the polymer and the solution, and the inside of the ¢lm, respec0 tively; kD,X is the mass transport coe⁄cient for the species X at the rotating

FIG. 13 The scheme proposed for oxidation of NADH at a chemically modified electrode.

528

Simon and Bartlett

0 disk electrode (kD,X ¼ 1.554DX2/3n1/6W1/2; where DX is the di¡usion coef¢cient, n is the kinematic viscosity, and W is the rotation speed in Hertz [102]); K S and K P are the partition coe⁄cients for NADH and NAD þ into the ¢lm, respectively; K M is the equilibrium constant for adsorption and kcat is the rate constant for oxidation of the adsorbed NADH; site represents a catalytic site within the ¢lm; and K i is the inhibitor constant for reversible inhibition by NAD þ . According to this mechanism the current for NADH oxidation is given by nFAKM DS;film y ð7Þ i¼ L with ! e1=2 a 1=2 y ¼ f2e½a  lnð1 þ aÞg tanh ð8Þ ð1 þ aÞf2½a  lnð1 þ aÞg1=2

e¼

L2 ½sitekcat DS;film KM

a¼

KS ½NADH0  KM

ð9Þ

and ð10Þ

where L is the ¢lm thickness and [site] is the concentration of active catalytic sites within the ¢lm. Assuming that the ¢lm thickness is related to the charge passed during the ¢lm growth process,Qgrowth, L ¼ sgrowth Qgrowth

ð11Þ

where sgrowth is a constant. In these equations [NADH]0 refers to the concentration of the NADH at the outside of the polymer ¢lm. For the rotating disk electrode this concentration is related to the bulk concentration by i ½NADH0  ¼ ½NADHbulk   ð12Þ 0 nFAkD;S The e¡ects of NAD þ inhibition may be taken into account [77] by replacing K M and kcat by 0 ¼ kcat

kcat 1  KM =Ki

and 0 KM ¼ KM

1 þ ðKP ½NADþ 0  þ KS ½NADH0 Þ=Ki ð1  KM =Ki Þ

ð13Þ  ð14Þ

Modified Electrodes for the Oxidation of NADH

529

respectively,where þ ½NADþ 0  ¼ ½NADbulk  þ

i 0 nFAkD;P

ð15Þ

The expressions for the current, obtained from Eqs. (7)^(10), corresponding to the di¡erent limiting cases are summarized in the case diagram given in Fig. 14. Case I: no concentration polarization in the polymer layer. If di¡usion within the polymer layer is fast, the concentration of NADH is uniform throughout the ¢lm. This is valid when the ¢lm is thin ðe < 1Þ and the concentration of NADH is not su⁄cient to saturate the ¢lm ða < 1Þ. Under these circumstances the reaction is ¢rst-order in NADH and occurs throughout the whole ¢lm. Case II: unsaturated kinetics, low concentration of substrate. If the concentration of NADH is not high enough to saturate the catalytic

FIG. 14 Case diagram derived from Eqs. (7)^(10). The different equations represent the different limiting expressions for the current in each case and across the boundaries between each pair of cases. (Reprinted from P. N. Bartlett and E. Simon, ‘‘Poly(aniline)-poly(acrylate) composite films as modified electrodes for the oxidation of NADH,’’ Phys. Chem. Chem. Phys. 2, pp. 2599^2606, Copyright 2000, with permission from The Royal Society of Chemistry.)

530

Simon and Bartlett

sites ða < 1Þ but the ¢lm is thick ðe > 1Þ, all of the NADH will be consumed in a ¢rst-order reaction layer, thickness XK , at the outside of the ¢lm. The current is then independent of the ¢lm thickness and ¢rst-order in the concentration of NADH. Case III: saturated kinetics, high concentration of substrate. If the concentration of NADH is su⁄cient to saturate the ¢lm ða > 1Þ and the ¢lm is thin ðe < 2aÞ, then the reaction of NADH occurs with zero-order kinetics throughout the whole ¢lm. The current is now dependent on ¢lm thickness, the concentration of catalytic sites, and independent of the NADH concentration. Case IV: partially saturated kinetics, intermediate substrate concentration. If the concentration of NADH is su⁄cient to saturate the catalytic sites at the outside of the ¢lm ða > 1Þ yet the ¢lm is thick ðe > 2aÞ, then, since the concentration of NADH falls as it is di¡uses into the ¢lm and reacts, at the outside of the ¢lm the reaction will be zero-order in NADH concentration, but farther into the ¢lm it will become ¢rst-order in NADH concentration. The current is then independent of the ¢lm thickness and turns out overall to be halforder with respect to the concentration of catalytic sites and the concentration of NADH. We can use the expressions given in the case diagram to ¢t the amperometric response to NADH. First, the appropriate initial and ¢nal cases for a given set of experimental data are identi¢ed by considering the dependence of the current on NADH concentration, [NADH]0, and ¢lm thickness, L. Next the appropriate expression for the current is selected from Fig. 14. The experimental data are then ¢tted to this expression using a commercial nonlinear least-squares ¢tting routine. For example, in our group we have used poly(aniline)-poly(acrylate) (PANi-PAA) composite ¢lms as catalysts for NADH oxidation [78]. The ¢lms were deposited on a glassy carbon rotating disk electrode.The modi¢ed electrode was than transferred to an electrochemical cell containing 0.1 M citrate/phosphate bu¡er pH 7.0. Aliquots of a concentrated NADH solution were added and the amperometric response at a given potential was recorded with the electrode rotated at a constant speed.We studied the in£uence of the potential, the ¢lm thickness, and the rotation speed on the current. Kinetic data were obtained from the ¢tting of the experimental data, using the expressions given in Fig. 14. The results obtained for six di¡erent potentials and for six di¡erent ¢lms of the same nominal thickness showed that the potential has an in£uence on the amperometric response (Fig. 15). At potentials less than 0.05 V versus SCE the response is fast, but with increasing potential the response

Modified Electrodes for the Oxidation of NADH

531

FIG. 15 Currents for the oxidation of NADH recorded at six different potentials at a poly(aniline)-poly(acrylate) modified glassy carbon electrode (deposition charge 90 mC, geometric area ¼ 0.38 cm2) plotted as a function of NADH concentration recorded at rotation speed of 9 Hz, in 0.1 M citrate/phosphate buffer pH 7, under argon: * 30 mV; ^ 0 mV; 25 mV; m 50 mV; & 75 mV; 100 mV. The solid lines represent the best fits of the experimental data to the expression for the case I/III boundary; the resulting kinetic parameters are given in Table 3. (Reprinted from P. N. Bartlett and E. Simon, ‘‘Poly(aniline)-poly(acrylate) composite films as modified electrodes for the oxidation of NADH,’’Phys. Chem. Chem. Phys. 2, pp. 2599^2606, Copyright 2000, with permission from The Royal Society of Chemistry.)

time of the ¢lm to changes in NADH concentration becomes much longer, due to the increasing resistance of the ¢lm even though the measured steadystate current is greater due to a higher concentration of reactive catalytic sites within the ¢lm. In these experiments, thin ¢lms were used, so that we start in case I, where the current depends on [NADH]0. With increasing NADH concentration we move into case III.Therefore the expression for the current across the case I/III boundary,  i ¼ nFA½siteL

KS ½NADH0 KS ½NADH0 þ KM

 ð16Þ

532

Simon and Bartlett

TABLE 3 Best-Fit Parameters from the Analysis of the Currents for NADH Oxidation at Poly(aniline)-poly(acrylate) Modified Glassy Carbon Electrode, at Different Potentials E vs. SCE(V) 0.030 0.000 þ 0.025 þ 0.050 þ 0.075 þ 0.100

kcat[site]s(mol/cm2mC1s1)

KM/KS (mM)

(0.86 0.04)  106 (1.56 0.03)  106 (2.06 0.02)  106 (3.69 0.07)  106 (3.88 0.07)  106 (4.24 0.05)  106

1.055 0.071 0.558 0.195 0.635 0.138 0.775 0.025 0.685 0.021 0.569 0.012

Source: Reprinted from P. N. Bartlett and E. Simon, ‘‘Poly(aniline)-poly(acrylate) composite films as modified electrodes for the oxidation of NADH,’’ Phys. Chem. Chem. Phys. 2, pp. 2599^2606, Copyright 2000, with permission from The Royal Society of Chemistry.

was used to ¢t the data and gave good results. In carrying out this ¢tting, the surface concentration of NADH, [NADH]0, was calculated from the bulk value, [NADH]1, using Eq. (12) for the rotating disk electrode. The data show that the parameter, kcat[site]DSK S depends on the applied potential,whereas K M/K S is practically independent of the potential (Table 3). We also studied the in£uence of the ¢lm thickness (Fig. 16), and observed that the current increased with increasing ¢lm thickness and NADH concentration. This shows that the reaction occurs throughout the whole ¢lm: the thicker the ¢lm, the greater is the number of sites that participate in the electrocatalysis. The data in Fig. 16 were ¢tted to the equation for the I/II boundary (thin ¢lm to thick ¢lm under unsaturated conditions); the results are given inTable 4.These ¢ts were performed without taking into account inhibition by the product, NAD þ . The e¡ects of NAD þ inhibition were studied by adding NAD þ to the bulk solution. It is clear from Fig. 17 that NAD þ does inhibit the reaction. However, the inhibition is weak, so its e¡ects in the experiments above where the only NAD þ is that generated by the electrode reaction itself are negligible. From this type of detailed analysis we get an intimate picture of the catalysis of the reaction at the modi¢ed electrode, which can be used to compare the performance of di¡erent polymer ¢lms and can help in designing electrodes for application in biosensors and biofuel cells. 5

APPLICATIONS

The high selectivity of enzymes, coupled with the relative simplicity, portability, and low cost of amperometric electrochemical measurement, has

Modified Electrodes for the Oxidation of NADH

533

FIG. 16 Currents for the oxidation of NADH at poly(aniline)-poly(acrylate) modified glassy carbon electrodes, geometric area 0.38 cm2, coated with films of different thickness plotted as a function of the deposition charge. Results for eight different NADH concentrations are shown, recorded at þ 0.05 V at a rotation speed of 9 Hz, in 0.1 M citrate/phosphate buffer pH 7, under argon: * 0.05 mM; ! 0.1 mM; 0.15 mM; ^ 0.2 mM; * 0.3 mM; ! 0.45 mM; & 0.5 mM; ^ 0.6 mM. The solid lines represent the best fits of the experimental data to the expression for the case I/III boundary, the resulting kinetic parameters are given in Table 4. (Reprinted from Phys. P. N. Bartlett and E. Simon, ‘‘Poly(aniline)-poly(acrylate) composite films as modified electrodes for the oxidation of NADH,’’ Phys. Chem. Chem. Phys. 2, pp. 2599^2606, Copyright 2000, with permission from The Royal Society of Chemistry.)

meant that there is a continued interest in developing amperometric enzyme biosensors for environmental, medical diagnostic, and food applications. In an amperometric enzyme electrode, the current should be a direct measure of the substrate concentration within the sample. This means that for NADH- dependent dehydrogenase-based electrodes the rate of oxidation of NADH should not be rate-limiting and the chemically modi¢ed electrode should be stable in the sample medium. Many of the modi¢ed electrodes described in the preceding paragraph have been used as the basis on which to develop amperometric biosensors. Examples of amperometric biosensors using NADH-dependent dehydrogenase enzymes have been reviewed [21], although this paper concentrates mainly on direct electrocatalytic methods of NAD þ recycling as

534

Simon and Bartlett

TABLE 4 Best-Fit Parameters from the Analysis of the Currents for NADH Oxidation at Poly(aniline)-poly(acrylate) Modified Glassy Carbon Electrode, for Different Film Thicknesses [NADH]0 (mM) 0.05 0.10 0.15 0.20 0.30 0.45 0.50 0.60

(kcat[site]s2/ DSKM)1/2(C1)

(kcat[site]KS2DS/ KM)1/2(cm/s)

0.14 0.04 0.21 0.03 0.20 0.02 0.25 0.02 0.22 0.01 0.23 0.02 0.22 0.02 0.24 0.02

(1.44 0.03)  103 (9.53 0.07)  104 (9.63 0.04)  104 (8.26 0.03)  104 (7.79 0.02)  104 (7.11 0.02)  104 (7.06 0.03)  104 (6.35 0.02)  104

Source: Reprinted from P. N. Bartlett and E. Simon, ‘‘Poly(aniline)-poly(acrylate) composite films as modified electrodes for the oxidation of NADH,’’ Phys. Chem. Chem. Phys. 2, pp. 2599^2606, Copyright 2000, with permission from The Royal Society of Chemistry.

opposed to enzymatic regeneration of the co-enzyme. In their review, Katakis and Dominguez point out that fused aromatic dyes seem to be promising mediators for NADH oxidation. They also conclude that polymeric mediators, electropolymerization techniques, and NADH-oxidizing enzymes provide the most promising basis for the development of biosensors and bioelectronics. In an earlier review in the same journal, Trijanowicz et al. [104] describe biosensors for 13 di¡erent substrates based on dehydrogenase enzymes immobilized within, or on top of, electropolymerized ¢lms. Among the di¡erent types of NADH-dependent dehydrogenases, the most widely applied in biosensors are alcohol dehydrogenase (E.C. 1.1.1.1) and lactate dehydrogenase (E.C. 1.1.1.27). To give some idea of the scope of the literature in this area, we have gathered together examples of biosensors based on NADH-dependent dehydrogenases in Table 5. Fuel cells convert chemical energy into electrical energy as long as the fuel is supplied (Fig. 18). To be an e⁄cient converter of chemical energy into electrical energy, both fuel cell electrodes must be able to carry out their respective electrode reactions at low overpotentials and high rates
in other words, they must be good electrocatalysts.This is a key limitation, since most fuels are not ideal as electrode reactants, and catalysis of the electrode reactions is a limiting factor. Biocatalysts (either whole microorganisms or enzymes) can be used as one possible approach to overcome this problem. NADH-dependent dehydrogenase enzyme-based electrodes are attractive

Modified Electrodes for the Oxidation of NADH

535

FIG. 17 Currents for the oxidation of NADH at poly(aniline)-poly(acrylate) modified glassy carbon electrodes (deposition charge 90 mC, geometric area ¼ 0.38 cm2) ploted as a function of NADH concentration, recorded at þ0.05 V at a rotation speed of 9 Hz in 0.1 M citrate/phosphate buffer pH 7, under argon. Results recorded both in the absense (*) and in the presence (*) of 2.1 mM NAD þ are shown. (Reprinted from P. N. Bartlett and E. Simon, ‘‘Poly(aniline)poly(acrylate) composite films as modified electrodes for the oxidation of NADH,’’ Phys. Chem. Chem. Phys. 2, pp. 2599^2606, Copyright 2000, with permission from The Royal Society of Chemistry.)

in this respect because they can be developed for a wide range of fuels, they are renewable, and they operate at moderate temperatures and pH. Bioelectrochemical fuel cells can be used, not only to generate electricity but also, at the same time, to produce chemicals of potential commercial interest. Microbiological and enzyme fuel cells were described in the literature in the 1980s, and the di¡erent systems used are summarized in a review [105]. The anodic compartment of a microbiological fuel cell is constructed by placing a solid electrode in a suspension of microbial cells (i.e., Microcus ceri¢cans, Saccharomyces cerevisiae, etc.). The source of hydrocarbon used in these fuel cells is typically n-hexane or glucose. This technique does not require puri¢ed enzymes or cofactors. However, the role of the microorganisms and the mechanism of such bioelectrochemical fuel cell reactions are not well established. It is notable that many of the redox mediators used in these microbial systems are the same as, or closely related to, the types of redox dye and mediators used to catalyze NADH oxidation, and

Glutamic pyruvic transaminase and 1-lactate dehydrogenase Glyceride dehydrogenase

Film Carbon paste Carbon paste

D-Lactate dehydrogenase L-Lactate dehydrogenase

Carbon electrode

Carbon paste

Gold microband Screen-printed Carbon fiber microcylinder Pyruvate

Carbon paste

Technique

Glucose dehydrogenase

Enzyme

Adsorbed Meldola blue, nile blue, or toluidine blue O

Osmium phenanthrolinedione Toluidine Blue O Poly(toluidine blue) Poly(o-phenylenediamine) TCNQ Carbon paste electrodes coated with poly(o-phenylenediamine) (PPD) and poly(o-aminophenol) conducting films Nickel(II) hexacyanoferrate Meldola blue Toluidine blue O (covalently attached) Poly(3-methylthiophene)/poly (phenol red) electrode poly(o-phenylenediamine)

Mediator

TABLE 5 Examples of Amperometric Enzyme Electrodes based on NADH-Dependent Dehydrogenases

35

118

117

93 116 37

87 112 59 114 73 115

Ref.

536 Simon and Bartlett

Alcohol dehydrogenase

Film

Monolayers

Carbon paste

polymer-toluidine Blue O Toluidine Blue O Methylene green (adsorbed) Poly(o-aminophenol)-poly(o-phenylenediamine) Phenoxazine, phenothiazines Poly(o-phenylenediamine) [Re(phen-dione)(CO)3Cl and [Fe (phendione)3](PF6)2 Ruthenium Metal complexes containing 1, 10phenanthroline-5,6-dione ligands Films derived from 3,4-dihydroxy benzaldehyde 3,4-Dihydroxybenzaldehyde on carbon felt/epoxy 31

26, 30

94 84

42 119 86

47 43 44 115

Modified Electrodes for the Oxidation of NADH 537

538

FIG. 18

Simon and Bartlett

The principle of the fuel cell.

one can speculate that the coupling into the electron-transport chain within the microorganism is an important part of the operation of these systems. In NADH-dependent dehydrogenase-based fuel cells, in contrast, puri¢ed enzyme and co-enzyme are placed in the anodic compartment. As for the biosensor application, the enzyme is often immobilized onto the electrode surface, and a redox mediator is employed to shuttle charge between the electrode and the NAD(P)H co-enzyme.Thus, once again, an understanding of the principles of design of NADH electrodes is important for the development of this type of biofuel cell. Since the biofuel needs to operate at, or close to, neutral pH, there is also interest in using enzymes to catalyze the cathode reaction. In order to achieve a high current density, four-electron transfer reactions are interesting.Two substrates have been used, methanol and dioxygen. Methanol is enzymatically oxidized to formate in a four-electron process, and it has been used in the anodic compartment of fuel cells. Dioxygen is reduced in a four-electron reaction to hydroxide ions in the presence of laccase, and this system has been used in a biocathode [105]. Abiofuelanodebasedon D-glucosedehydrogenasehasbeendeveloped, using a graphite electrode modi¢ed by adsorption of Meldola blue [106]. In these experiments a simulated cathode was used rather than an oxygen electrode. Using the simulated cathode the bioanode gave a current density of 0.2 mA/cm2 at a cell voltage of  0:8 V, and the cell ran for more than 8 h. More recently, Willner and coworkers described a biofuel cell based on pyroquinoline quinone (PQQ) and microperoxidase-11 monolayer

Modified Electrodes for the Oxidation of NADH

539

functionalized electrodes [107]. PQQ and microperoxidase-11 were immobilized in monolayers by covalent bonding onto gold electrodes ¢rst modi¢ed by cysteine. The oxidizing agent supplied to the cathodic compartment was H2O2 which was catalytically reduced at the gold electrode in the presence of microperoxidase-11. The fuel used in the anodic compartment was NADH which was oxidized at an electrode coated with PQQ. In these experiments the two compartments of the fuel cell were separated by a porous membrane. The open-circuit voltage obtained for this biofuel cell was  320 mV, and the short-circuit current density was 30 mA=cm2 , with a maximum power density of 8 mW at an external load of 3 kO. Palmore and co-workers recently developed a methanol/dioxygen biofuel cell using dehydrogenases [108]. In this cell, diaphorase (D) is used to catalyze the oxidation of NADH to NAD þ using benzylviologen (BV) as the electron acceptor; BV þ is then oxidized to BV2 þ at a graphite anode. This is coupled to a platinum cathode used for oxygen reduction. The NAD þ is used in the enzymatically catalyzed oxidation of methanol to CO2 using alcohol (ADH), aldehyde (AldDH), and formate (FDH) dehydrogenases. The two compartments are separated by a Na¢on membrane, which allows cations to migrate from the anodic compartment to the cathodic compartment. The biofuel cell is shown schematically in Fig. 19. Although quite a number of di¡erent systems have been used in the anodic compartment of biofuel cells, only a few systems have been used in the cathodic compartment, notably microperoxidase-11 [107,109,110] and laccase [111]. Biochemical fuel cells are potentially interesting for use in energy conversion in a number of applications. However, the practical problems in achieving useful devices should not be underestimated. The biofuel cell application is considerably more demanding than the biosensor application since; although the problems of interference from other species in the sample and the need for selectivity are removed, these advantages are more than made up for by the constraints imposed by the need to operate at low overpotentials (to achieve reasonable energy conversion e⁄ciencies) at high current densities (to achieve reasonable power to volume ratios) and for extended periods of time (to be economical). Although the studies above show that some progress can be made in these areas, for biofuel cells there is still a long way to go. 6

CONCLUSION

Direct electrochemical oxidation of NADH requires a signi¢cant overpotential at bare metal or carbon electrode surfaces. Consequently it is essential, if we wish to exploit NAD(P)H-dependent dehydrogenases in

540

Simon and Bartlett

FIG. 19 Scheme of methanol/dioxygen biofuel cell. NAD þ -dependent dehydrogenases oxidize CH3OH to CO2; diaphorase (D) catalyzes the oxidation of NADH in NAD þ using benzylviologen (BV) as the electron acceptor; BV þ is oxidized to BV2 þ at a graphite anode and the electrons released flow around the external circuit to the platinum cathode where dioxygen is reduced. (Reprinted from G. T. R. Palmore, H. Bertschy, S. H. Bergens, and G. M. Whitesides, ‘‘A methanol/dioxygen biofuel cell that uses NAD þ -dependent dyhydrogenases as catalysts: Application of an electro-enzymatic method to regenerate nicotinamide adenine dinucleotide at low overpotentials,’’ J. Electroanal. Chem. 443, p. 159, Copyright 1998, with permission from Elsevier Science.)

biosensors, biofuel cells, or bioelectronics, that we seek electrocatalytic surfaces for NAD(P)H oxidation. At the same time, the oxidation of NAD provides an excellent model system for the development, design, and understanding of modi¢ed electrodes, because for NADH oxidation we have a very good idea of what is required of an e⁄cient meditor from knowledge of the biochemical role of NADH and studies of its reactions in homogeneous solution. Reviewing the examples in the literature for NADH mediation suggests that the vast majority are consistent with a model in which the mediator is reduced by hydride transfer from NADH and is then reoxidized by the electrode in sequential one-electron and proton transfers. In this way the mediator bridges the two-electron oxidation/reduction world of NADH and the one-electron oxidation/reduction world of the electrode.

Modified Electrodes for the Oxidation of NADH

541

ACKNOWLEDGMENT This work was supported by the U.S. O⁄ce of Naval Research. REFERENCES 1. A Fersht. Enzyme, Structure and Mechanism. Reading, UK: Freeman, 1985. 2. CK Mathews, KE Van Holde. Biochemistry. Menlo Park, CA: Benjamin/ Cummings, 1996. 3. T Bugg. An Introduction to Enzyme and Coenzyme Chemistry. Oxford, UK: Blackwell Science, 1997. 4. AL Lehninger. Biochemistry. New York: Worth, 1975. 5. WJ Blaedel, RA Jenkins. Anal Chem 46:1952, 1974. 6. WJ Blaedel, RA Jenkins. Anal Chem 47:1337, 1975. 7. RD Braun, KSV Santhanam, PJ Elving. J Am Chem Soc 97:2591, 1975. 8. J Moiroux, PJ Elving. Anal Chem 50:1056, 1978. 9. J Moiroux, PJ Elving. J Am Chem Soc 102:6533, 1980. 10. Z Samec, PJ Elving. J Electroanal Chem 144:217, 1983. 11. PJ Elving,WT Bresnahan, J Moiroux, Z Samec. Bioelectrochem Bioener 9:365, 1982. 12. LL Miller, JR Valentine. J Am Chem Soc 110:2982, 1988. 13. BW Carlson, LL Miller. J Am Chem Soc 105:7453, 1983. 14. BW Carlson, LL Miller, P Neta, J Grodkowski. J Am Chem Soc 106:7233, 1984. 15. BW Carlson, LL Miller. J Am Chem Soc 107:479, 1985. 16. A Kitani, LL Miller. J Am Chem Soc 103:3595, 1981. 17. A Kitani,Y.-H So, LL Miller. J Am Chem Soc 103:7636, 1981. 18. L Gorton, A Torstensson, H Jaegfelt, G Johansson. J Electroanal Chem 161:103, 1984. 19. N Cenas, J Rozgaite, A Pocius, JJ Kulys. J Electroanal Chem 189:163, 1985. 20. PN Bartlett, ENK Wallace. J Electroanal Chem 486:23, 2000. 21. I Katakis, E Dominguez. Mikrochim Acta 126:11, 1997. 22. WJ Albery, AR Hillman. J Electroanal Chem 170:27, 1984. 23. PN Bartlett, P Tebbutt, RG Whitaker. Prog Reaction Kinetics 16:55, 1991. 24. F Pariente, E Lorenzo, HD Abruna. Anal Chem 66:4337, 1994. 25. G Moreno, F Pariente, E Lorenzo. Anal Chim Acta 420:29, 2000. 26. F Pariente, F Tobalina, G Moreno, L Hernandez, E Lorenzo, HD Abruna. Anal Chem 69:4065, 1997. 27. L Gorton, G Johansson, A Tortensson. J Electroanal Chem 196:81, 1985. 28. F Pariente, F Tobalina, M Darder, E Lorenzo, HD Abruna. Anal Chem 68:3135, 1996. 29. F. Pariente, E. Lorenzo, F. Tobalina, H. D. Abruna. Anal. Chem. 67:3936, 1995. 30. E Lorenzo, F Pariente, L Hernandez, F Tobalina, M Darder, Q Wu, M Maskus, HD Abruna. Biosensors Bioelectron 13:319, 1998. 31. F Tobalina, F Pariente, L Hernandez, HD Abruna, E Lorenzo. Anal Chim Acta 358:15, 1998.

542

Simon and Bartlett

32. E Lorenzo, L Sanchez, F Pariente, J Tirado, HD Abruna. Anal Chim Acta 309:79, 1995. 33. H Jaegfelt,T Kuwana, G Johansson. J Am Chem Soc 105:1805, 1983. 34. H Huck. Fresenius J Anal Chem 350:599, 1994. 35. V Laurinavicius, B Kurtinaitiene, V Gureviciene, L Boguslavsky, L Geng, T Skotheim. Anal Chim Acta 330:159, 1996. 36. D Dicu, L Muresan, IC Popescu, C Cristea, IA Silberg, P Brouant. Electrochim Acta 45:3951, 2000. 37. HX Ju, L Dong, HY Chen. Talanta 43:1171, 1996. 38. CX Cai, LH Yin, KH Xue. J Mol Cat A, Chem 152:179, 2000. 39. M Ohtani, S Kuwabata, H Yoneyama. J Electroanal Chem 422:45, 1997. 40. HY Chen, DM Zhou, JJ Xu, HQ Fang. J Electroanal Chem 422:21, 1997. 41. QJ Chi, SJ Dong. J Mol Cat A, Chem 105:193, 1996. 42. MJ Lobo, AJ Miranda, P Tunon. Electroanalysis 8:591, 1996. 43. MJ Lobo, AJ Miranda, P Tunon. Electroanalysis 8:932, 1996. 44. Q Chi, S Dong. Anal Chim Acta 285:125, 1994. 45. HY Chen, AM Yu, JL Han,YZ Mi. Anal Lett 28:1579, 1995. 46. E Dominguez, HL Lan,Y Okamoto, PD Hale,TA Skotheim, L Gorton. Biosensors Bioelectron 8:167, 1993. 47. E Dominguez, HL Lan, Y Okamoto, PD Hale, TA Skotheim, L Gorton, B HahnHagerdal. Biosensors Bioelectron 8:229, 1993. 48. CA Pessoa, Y Gushikem, LT Kubota, L Gorton. J Electroanal Chem 431:23, 1997. 49. A Malinauskas,T Ruzgas, L Gorton. J Colloid Interface Sci 224:325, 2000. 50. LT Kubota, F Gouvea, AN Andrade, BG Milagres, G DeOliveiraNeto. Electrochim Acta 41:1465, 1996. 51. B Gurendig, G Wittstock, U Ruedel, B Strehlitz. J Electroanal Chem 395:143, 1995. 52. PD Hale, HungSui Lee,Y Okamoto. Anal Lett 26:1073, 1993. 53. J Wang, PVA Pamidi, Mian Jiang. Anal Chim Acta 360:171, 1998. 54. S Sampath, O Lev. J Electroanal Chem 446:57, 1998. 55. F Ni, H Feng, L Gorton,TM Cotton. Langmuir 6:66, 1990. 56. CX Cai, KH Xue. Anal Chim Acta 343:69, 1997. 57. CX Cai, KH Xue. Talanta 47:1107, 1998. 58. CX Cai, KH Xue. Microchem J 64:131, 2000. 59. DM Zhou, JJ Sun, HY Chen, HQ Fang. Electrochim Acta 43:1803, 1998. 60. AA Karyakin, EE Karyakina,W Schuhmann, HL Schmidt, SD Varfolomeyev. Electroanalysis 6:821, 1994. 61. A Silber, N Hampp,W Schuhmann. Biosensors Bioelectron 11:215, 1996. 62. S Cosnier, K LeLous. J Electroanal Chem 406:243, 1996. 63. DM Zhou, HQ Fang, HY Chen, HX Ju,Y Wang. Anal Chim Acta 329:41, 1996. 64. CX Cai, KH Xue. J Electroanal Chem 427:147, 1997. 65. T Ohsaka, K Tanaka, K Tokuda. J Chem Soc Chem Commun 222, 1993. 66. K Tanaka, S Ikeda, N Oyama, K Tokuda,T Ohsaka. Anal Sci 9:783, 1993. 67. CX Cai, KH Xue. Microchem J 58:197, 1998.

Modified Electrodes for the Oxidation of NADH 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102.

543

N Mano, A Kuhn. J Electroanal Chem 477:79, 1999. N Mano, A Kuhn. J Chem Soc Chem Commun 497, 1999. N Mano, A Kuhn. J Electroanal Chem 58, 2001. E Casero, M Darder, K Takada, HD Abruna, F Pariente, E Lorenzo. Langmuir 15:127, 1999. ASN Murthy, Anita. Bioelectrochem Bioenerget 33:71, 1994. PC Pandey,V Pandey, S Mehta. Biosensors Bioelectron 9:365, 1994. E Katz,T Loetzbeyer, DD Schlereth,W Schuhmann, HL Schmidt. J Electroanal Chem 373:189, 1994. KJ Stine, DM Andrauskas, AR Khan, P Forgo,VT Dsouza, Jingyue Liu, RM Friedman. J Electroanal chem 472:147, 1999. LYiTao, Z Jie, C HongYuan. Anal Lett 30:2691, 1997. PN Bartlett, PR Birkin, ENK Wallace. J Chem Soc FaradayTrans 93:1951,1997. PN Bartlett, E Simon, PCCP. Phys. Chem. Chem. Phys. 2:2599, 2000. W Schuhmann, R Lammert, M Hammerle, HL Schmidt. Biosensors Bioelectron 6:689, 1991. W Schuhmann, J Huber, H Wohlschlager, B Strehlitz, B Grundig. J Biotech 27:129, 1993. S Cosnier, M Fontecave, D Limosin,V Niviere. Anal Chem 69:3095, 1997. S Cosnier, JL De¤court, C Frier, C Innocent. Electroanalysis 10:521, 1998. Q Chi, S Dong. Analyst 119:1063, 1994. Q Wu, M Maskus, F Pariente, F Tobalina, VM Fernandez, E Lorenzo, HD Abruna. Anal Chem 68:3688, 1996. IC Popescu, E Dominguez, A Narvaez, V Pavlov, I Katakis. J Electroanal Chem 464:208, 1999. F Tobalina, F Pariente, L Hernandez, HDAbruna, E Lorenzo. Anal Chim Acta 395:17, 1999. M Hedenmo, A Narvaez, E Dominguez, I Katakis. Analyst 121:1891, 1996. H Ju, D Leech. Anal Chim Acta 345:51, 1997. GD Storrier, K Takade, HD Abruna. Inorg Chem 38:559, 1999. F Xu, H Li, SJ Cross,TF Guarr. J Electroanal Chem 368:221, 1994. C-X Cai, H-X Ju, H-Y Chen. J Electroanal Chem 397:185, 1995. C-X Cai, H-X Ju, H-Y Chen. Anal Chem Acta 310:145, 1995. C-X Cai, H-X Ju, H-Y Chen. Anal Lett 28:809, 1995. J Wang, E GonzalezRomero, AJ Reviejo. J Electroanal Chem 353:113, 1993. A Fujishima,TN Rao, E Popa, BV Sarada, I Yagi, DATryk. J Electroanal Chem 473:179, 1999. TN Rao, I Yagi,T Miwa, DA Tryk, A Fujishima. Anal Chem 71:2506, 1999. J Kulys, G Gleixner,W Schuhmann, HL Schmidt. Electroanalysis 5:201, 1993. S Cosnier, K LeLous. Talanta 43:331, 1996. T Tatsuma,T Watanabe. J Electroanal Chem 310:149, 1991. FC Anson. J Phys Chem 84:3336, 1980. S Fukuzumi, M Nishizawa,T Tanaka. J Org Chem 49:3571, 1984. WJ Albery. Electrode Kinetics. Oxford, UK: Oxford University Press, 1975.

544

Simon and Bartlett

103. MEG Lyons, CH Lyons, A Michas, PN Bartlett. J Electroanal Chem 351:245, 1993. 104. M Trijanowicz, T Krawczynski, K Vel Krawczyk. Mikrochim Acta 121:167, 1995. 105. LBJ Wingard, CH Shaw, JF Castner. Enzyme Microbial Technol 4:137, 1982. 106. B Persson, L Gorton, G Johansson, ATorstensson. Enzyme Microbial Technol 7:549, 1985. 107. I Willner, G Arad, E Katz. Bioelectrochem Bioenerget 44:209, 1998. 108. GTR Palmore, H Bertschy, SH Bergens, GM Whitesides. J Electroanal Chem 443:155, 1998. 109. I Willner, E Katz, F Patolsky, AF Bu«ckmann. J Chem Soc FaradayTrans 2:1817, 1998. 110. E Katz, B Filanovsky, I Willner. New J Chem 481, 1999. 111. GTR Palmore, HH Kim. J Electroanal Chem 464:110, 1999. 112. LI Boguslavsky, L Geng, IP Kovalev, SK Sahni, Z Xu, TA Skotheim, V Laurinavicius, B Persson, L Gorton. Biosensors Bioelectron 10:693, 1995. 113. CX Cai, KH Xue,YM Zhou, H Yang. Talanta 44:339, 1997. 114. HC Shu, L Gorton, B Persson, B Mattiasson. Biotechnology Bioeng 46:280, 1995. 115. MJ Lobo, AJ Miranda, JM LopezFonseca, P Tunon. Anal Chim Acta 325:33, 1996. 116. SD Sprules, JP Hart, SA Wring, R Pittson. Anal Chim Acta 304:17, 1995. 117. K Warriner, S Higson, P Vadgama. Materials Sci Eng C, Biomimetic Mater Sensors Systems 5:91, 1997. 118. MJ LoboCastanon, AJ MirandaOrdieres, P TunonBlanco. Anal Chim Acta 346:165, 1997. 119. MJ LoboCastanon, AJ MirandaOrdieres, P TunonBlanco. Biosensors Bioelectron 12:511, 1997.

12 DNA-Based Sensors Michael J. Tarlov National Institute of Standards and Technology, Gaithersburg, Maryland, U.S.A.

Adam B. Steel MetriGenix, Inc., Gaithersburg, Maryland, U.S.A.

1

INTRODUCTION AND OVERVIEW

The sequencing of the human genome ranks as one of the great milestones of biomedical research [1,2]. It has the potential to provide new insights into battling disease, comprehending the e¡ect of environmental factors on health, and understanding the origin and evolution of the human species. Ironically, while the human genome project represents the greatest tour de force of DNA sequencing to date, it only increases the biomedical community’s appetite for faster DNA sequencing and diagnostic methods. One of the keys to unlocking the secrets of the human genome will be to compile the variations in genome sequence over statistically meaningful populations. It is these minute disparities that underlie di¡erences in an individual’s susceptibility to disease, the severity of illness, and response to medical treatments [3]. This genetic variability often takes the form of single DNA base changes, so-called single-nucleotide polymorphisms (SNPs). Many predict a future in which doctors will routinely test patients for particular SNPs and then tailor drug treatments according an individual’s genetic makeup [4]. Making this dream a reality will require vast improvements in DNA diagnostics and measurements. Putting this into perspective, consider 545

546

Tarlov and Steel

the requirements for cataloging SNPs, a necessary ¢rst step in understanding disease susceptibility. It is estimated that thousands of SNPs from hundreds of thousands of individuals will have to be analyzed, a ¢gure approaching billions of assays. Given the enormous number of assays, paramount considerations for genetic test methods will be accuracy, speed, and cost. We believe that DNA-based sensors are a potential linchpin technology for providing the DNA sequence information necessary to usher in the age of genetic medicine. In typical DNA-based sensors, a DNA probe sequence recognizes a complementary nucleic acid sequence of the analyte, or so-called target, with high speci¢city, and binds or hybridizes with that sequence. Eventually this binding event is converted into an electrical response that signals the presence of a particular nucleic acid sequence. Consequently, these sensing platforms can potentially provide genetic assays that are accurate, rapid, and economical. Moreover, they will almost certainly ¢nd use in a variety of other applications in which DNA sequence information is desired. These areas include fundamental biomedical research such as elucidating regulatory and signaling pathways [5], drug discovery [6,7], toxicology [8], agriculture [9], forensics, and detection of pathogens in consumer products and food. In this chapter, we attempt to provide a broad review of the current state of the art of DNA-based sensors. Our goal is to review basic sensing principles, including descriptions of the design, construction, and transduction methods. In addition to providing recent examples of DNA-based sensors in genetic test situations,we will also describe applications in which these platforms detect analytes other than nucleic acids. Because the surface structure of DNA sensors can strongly a¡ect their sensing performance, we also review studies that have attempted to gain a better understanding of the interfacial molecular architecture of DNA sensors. Several outstanding monographs and reviews of DNA-based sensors have appeared previously [10^14], and we will report on recent developments in the ¢eld that are not covered in these works.While we aim for comprehensive coverage, the pace of study of DNA-based sensors is rapid. We regret any omissions and emphasize that in no way is this a re£ection of the quality of the work.We will not review the area of DNA microarrays, powerful tools for studying gene expression in many di¡erent organisms [15,16]. DNA microarrays, closely related to DNA-based sensors, are ¢nding widespread application in a variety of areas related to molecular biology and human health, including the diagnosis, classi¢cation, and treatment of cancers [17,18]. The reader is referred to a number of excellent books, monographs, and reviews that describe the basic concepts and applications of this revolutionary technology [19^24]. Another exploding area of research that we will not cover in this review is that of micro£uidic devices, also known as ‘‘lab-on-a-chip.’’ These

DNA-Based Sensors

547

miniaturized analysis systems have been developed for nearly all aspects of nucleic acid analysis, including sample preparation, ampli¢cation [25,26], separation [27,28], and diagnostics [29]. We refer readers to several monographs and books for overviews of this exciting new technology [30^36]. 2 2.1

BACKGROUND Structure of DNA

DNA is one of nature’s most elegant structures and where its sensing properties originate. We provide a brief review here of DNA structure, but more detailed descriptions can be found elsewhere [37]. Nucleic acids are polymers in which the monomeric unit is the nucleotide. A nucleotide consists of a nitrogenous heterocyclic base, a ribose sugar, and a phosphate group (Fig. 1). In the DNA molecule, bases carry the genetic information whereas the sugar and phosphate groups play structural roles. DNA is comprised of four bases: cytosine (C), thymine (T), adenine (A), and guanine (G). It is the sequence of the four bases in a DNA chain that determines the information content for RNA and proteins. There are two structural classes of nucleic acids, ribonucleic acid (RNA) and deoxyribonucleic acid (DNA), which di¡er in the composition of the ribose sugar present in the monomer. This subtle di¡erence in structure causes signi¢cant di¡erences in the properties and functions of the two biopolymers. DNA contains genetic code that is carefully reproduced and conserved in successive generations of cells, whereas RNA is the working copy of the code used to produce proteins. Messenger RNA (mRNA), synthesized from a DNA template, serves as the template for protein synthesis in a process called translation. A current dogma in molecular biology is that the concentration of a mRNA in a particular cell is directly correlated with the metabolic activity of a particular gene. Gene expression monitoring refers to the measurement of levels of hundreds to thousands of mRNAs in cells or tissues.Through comparison of the complex pattern of expression levels from normal and diseased cells, it is possible to generate a diagnostic ¢ngerprint of a particular disease. Di¡erences in function between DNA and RNA are re£ected in respective cellular half-life: the cell does not intentionally destroy DNA, but RNA is constantly regulated by transcription from DNA and enzymatic degradation in response to environmental conditions. Because of the greater stability of DNA, most genetic sensors have used DNA rather than RNA. 2.2

Hybridization and Melting

Because all DNA-based sensors use hybridization as the basis for recognizing the presence of nucleic acid targets, we brie£y review fundamental

548

Tarlov and Steel

FIG. 1 Schematic diagram of the structure of DNA. In the model on the right the helixes and horizontal bars represent the sugar^phosphate backbone and base pairs, respectively. (Reprinted with permission from the National Health Museum, Washington, D.C., and the Access Excellence Forum at http://www.accessexcellence.org.)

concepts of hybridization. More detailed descriptions of nucleic acid hybridization are found elsewhere [38^40]. Hybridization occurs when two di¡erent single-stranded DNA (ssDNA) chains hydrogen-bond to each other to form a double helix. The helix, or duplex, is 2.0 nm in diameter and the structure repeats at intervals of 3.4 nm along the axis of the helix. The formation of double-standard DNA (dsDNA), or duplex DNA, requires proper alignment of complementary bases between two single-stranded

DNA-Based Sensors

549

polynucleotide chains. The hydrogen bonding occurs between the two strands such that A is hydrogen-bonded toT and C to G. In the double-helix structure, bases are compactly stacked in the middle to exclude water and promote hydrogen bonding. The phosphate groups are located on the outer surface and are ionized, and thus, negatively charged, under physiological conditions. The formation of dsDNA is reversible. The helix can be unwound, or denatured, by heating. The unwinding of the helix is called melting, because it occurs abruptly at a certain temperature. The melting temperature, Tm, is typically de¢ned as the temperature at which half of the helical structure is lost and is a measure of the stability of the duplex. The Tm depends strongly on the base composition, sequence length, and solution conditions such as ionic strength, pH, and bu¡er composition. For example, a DNA duplex rich in GC base pairs has a higher Tm because G:C bases are held together by three hydrogen bonds compared to the two hydrogen bonds of A:Tpairs. For short sequences, a general rule of thumb is that Tm increases by 2 C with the addition of an A:T base, and 4 C for a G:C pair [41]. Higher ionic strength also stabilizes duplexes because of the screening of the electrostatic repulsion of the negatively charged phosphate groups. ‘‘Stringency’’ is another important consideration when performing hybridization. Stringency refers to the manipulation of temperature or bu¡er conditions for the purpose of discriminating between perfectly complementary and mismatched strands. High stringency conditions favor the formation and stabilization of perfect complements, while under low stringency, base-pair mismatches can often be tolerated in a duplex. 3

BASIC ANATOMY OF DNA-BASED SENSORS

The operation and components of three di¡erent classes of DNA-based sensors are represented schematically in Fig. 2, in which recognition of three analytes is illustrated
nucleic acids, proteins, and small molecules. As for any biosensor, the heart of the device is a molecular recognition element. Nearly all DNA sensors use immobilized nucleic acids, or probes, as recognition elements. For genetic analysis, DNA probes are typically short sequences of ssDNA, 8^70 bases in length. The surface-con¢ned probes are designed to hybridize speci¢cally with complementary ssDNA targets for genetic assay applications. In other sensing con¢gurations, such as those measuring protein or small-molecule interactions, immobilized dsDNA is often used as the molecular recognition element. The hybridization or binding event causes a change in the interfacial chemical/physical properties, such as optical, mass, or electrochemical parameters, and the change is converted into a measurable analytical signal. As with any sensor, the

550

Tarlov and Steel

FIG. 2 Cartoon illustrating three DNA-based sensing formats: genetic assay, protein^DNA binding, and small molecule^DNA interaction.

principal criteria used to judge performance are sensitivity, selectivity, speed, stability, and cost,with the relative importance of each of these factors varying according to the application. In a strict de¢nition of a chemical sensor, a device must be capable of measuring the concentration of an analyte in real time. While we would not narrowly de¢ne a DNA-based sensor in this manner, most of the DNA-based sensors described in this review share the following attributes. The device contains immobilized nucleic acid probes that act as molecular recognition elements for complementary single-stranded nucleic acid targets, proteins, or other molecules. The solid substrate and/or the DNA probe often play an active role in transducing the molecular recognition event into an electrical signal. The sensor system is capable of being miniaturized for multiplexed detection. Some DNA-based sensors can be adapted for real-time measurements of target concentration, a feature useful in applications such as monitoring of PCR products. An additional feature of real-time measurements is the ability to collect kinetic data of duplex formation or melting. The presence of base mismatches in a target sequence is often revealed by comparing the kinetic pro¢les of perfectly complementary and mismatched strands. In addition, as alluded to above, DNA-based sensors are more likely to be used for diagnosis or screening, such as genotyping of SNPs. In these

DNA-Based Sensors

551

applications, the sequence of a target is known and typically the desired information is a determination of whether that target is present and/or sequence variations (mutations) in the target. DNA-based sensors are not designed for the purpose of de-novo sequencing of DNA, i.e., the sequencing of unknown, genomic DNA. There is no clear distinction between DNAbased sensors and DNA microarrays. Indeed, some DNA-based sensors are touted as potential platforms for gene expression assays. Like DNA-based sensors, DNA microarrays also rely on the use of immobilized DNA probes. However, much higher DNA probe sequence densities, as high as 10 6 probe sequences/cm2, are typically found in these devices that enable the simultaneous monitoring of the expression of thousands of genes.

4 4.1

FABRICATION OF DNA-BASED SENSORS Probe Synthesis

Nearly all DNA-based sensors use presynthesized and puri¢ed nucleic acid probes. The advent of automated, solid-phase DNA synthesizers has enabled the widespread use of nucleic acids in sensing devices. DNA probes of almost any base sequence and length up to 150 bases can be made quickly, at high yield (1 mmole), of high purity, and at relatively low cost. Automated nucleic acid synthesizers use phosphite triester chemistries employing stable phophoramidite monomers to construct the biopolymers of nucleic acids. A wide variety of modi¢ed bases or labeling groups can be incorporated into the synthesis of an oligonucleotide. End-labels such as biotin, amines, thiols, or disul¢des are readily incorporated to anchor nucleic acids to surfaces.When end-labeling, a long £exible spacer or tether is frequently added to provide ample accessibility for surface attachment and for interaction with target molecules. Hydrocarbon linkers and nucleotide spacers are frequently used for these purposes. Puri¢cation is typically accomplished using either gel electrophoresis or high-performance liquid chromatography (HPLC).The reader is referred to the review by Ellington and Pollard and the citations therein for a complete description of automated synthesis, labeling, and puri¢cation of nucleic acids [42]. For fabricating high-density probe arrays used for gene expression analysis, an elegant photolithographic method has been developed to synthesize DNA probes directly on glass surfaces. Similar to methods used in microelectronics fabrication, light-directed synthesis of probes of up to 25 bases in length with micrometer or less lateral resolution is possible through the use of photolabile protecting groups [20,43]. The selection of sensor probe sequence(s) will, of course, hinge primarily on the sequence of the target. The speci¢c application also in£uences

552

Tarlov and Steel

the choice of probe length. For example, in applications requiring the discrimination of single-base-pair mismatches as in the analysis of SNPs, shorter probes are more likely to be used. Here a single point mutation is more likely to disrupt duplex formation and, thus, be detected. Longer probe strands are more often used for gene expression analysis. A ¢nal consideration includes the potential for probe secondary structure due to probe selfcomplementarity, which can hinder facile hybridization. 4.2

Immobilization of DNA Probes

To maximize the analytical signal, immobilization of DNA to the sensor surface should result in relatively high coverages of chemically stable, bound probes that are all accessible for hybridization with complementary targets. Covalent end-tethering of DNA is generally thought to be the best approach to ensure robust attachment and adequate conformational freedom for unhindered hybridization [44]. In addition, careful consideration must often be given to the surface properties of the substrate to limit the nonspeci¢c adsorption of interfering targets or analytes. The chemical strategy used for attachment is strongly in£uenced by the substrate used for the sensor. In turn, the choice of substrate is often determined by the transduction method selected for detecting hybridization events. For example, in surface plasmon resonance detection, gold thin (50 nm) ¢lms on glass are typically used because surface plasmon modes can readily be excited,while in £uorescence detection schemes, glass if often used because of relatively low background £uorescence. There are a plethora of approaches for immobilizing DNA probes on a multitude of surfaces.The chemical strategies for immobilizing nucleic acids on a variety of solid substrates were recently described in several excellent, thorough reviews [11,12,14].We provide a brief survey of the more commonly used methods for end-tethering DNA probes to surfaces and describe some recently developed protocols. The methods for attaching presynthesized DNA probes to glass or quartz surfaces are numerous and well developed [45^49]. The general approach is ¢rst to silanize the glass or quartz surface and then to link an end-derivatized DNA probe to the organosilane monolayer using crosslinking reagents. The DNA probes are attached by a robust covalent linkage and can withstand temperature cycling to 80 C. DNA probe coverages of up to 1013 probes/cm2 are possible, with hybridization e⁄ciencies ranging from approximately 10% to 75%, depending on the sequence [48]. Similar methods have been developed to attach DNA probes to gold surfaces. In the case of gold, however, the well-known thiol ^ gold interaction

DNA-Based Sensors

553

[50,51] is exploited to form a self-assembled monolayer (SAM) on the gold surface, yielding terminal functionalities of the SAM for reaction. A crosslinking reagent is then used to covalently attach end-modi¢ed oligonucleotides to the SAM surface [52^57]. DNA surface densities of 1.5 1012 probes/cm2 with hybridization e⁄ciencies of  60% have been reported [53]. Direct linkage to gold using thiol-derivatized DNA is also a common method of immobilization, in which DNA monolayer formation resembles self-assembly of alkanethiols [58^62]. The DNA probe molecules are endmodi¢ed with a thiol or disul¢de that chemisorbs to the gold surface. After DNA self-assembly, a second SAM is formed using another alkanethiol molecule that prevents strong interactions between the probe and the gold surface and leaves probes in a largely end-tethered con¢guration, fully accessible for hybridization. In addition, the alkanethiol SAM surface limits nonspeci¢c adsorption of nucleic acid targets that are not complementary to probes [59]. At coverages of 5 1012 probes/cm2, hybridization e⁄ciencies of nearly 100% have been reported [60,63]. A disadvantage of thiol-tethering of probes is that thermal stability above 75 C is limited [64]. Biotin ^ avidin coupling is another commonly used method to attach DNA probes to a variety of surfaces. In this approach the strong natural a⁄nity of biotin for the protein avidin [65] is used as a molecular ‘‘glue’’ to secure terminally biotinylated probes to surfaces [66]. This route o¡ers £exibility because avidin adsorbs strongly to a variety of surfaces, including bare gold [66], and preformed monolayers on gold [67] or glass [68]. Another strategy ¢nding increasing use is the immobilization of DNA probes in three-dimensional polymer or inorganic matrixes. The principle advantage of using 3-D porous sca¡olds is that they o¡er higher e¡ective surface areas and, thus, higher probe loadings can be realized resulting in greater signal intensities. The formation of DNA co-polymers has been reported by the co-polymerization of allyl- [69,70] or acrylamide-modi¢ed [71] DNA probes in a polyacrylamide co-polymer. Covalent attachment to polymer matrices formed by electropolymerization and biotin ^ avidin coupling of probes in acrylamide gel pads [72] have also been reported [73^75]. Other approaches investigated for increasing probe coverages include the use of colloidal silica supports [76], probe attachment to immobilized multifunctional dendrimer molecules [77], and DNA dendrimers [78^80]. A potential concern with high-surface-area substrates such as the above, however, is reduced mass transfer of DNA targets. Reports of methods for attaching DNA probes to planar and nanoparticle semiconductor surfaces are also on the rise. By exploiting the reaction of carbon double bonds with bare silicon surfaces, monolayers of long-chain alkenes terminated with reactive functional groups have been prepared on the (100) and (111) surfaces of silicon [81,82]. Probes are then

554

Tarlov and Steel

attached using a heterobifunctional cross-linker. Linkage of thiolderivatized probes to CdS [83,84] and amine-terminated probes to hydroxylated CdSe [85] surfaces has also been reported. 5

CHARACTERIZATION OF DNA-BASED SENSOR INTERFACES

The molecular-scale structural features of surface-con¢ned nucleic acid probes and their in£uence on the interaction with targets in solution are critical factors in the hybridization process. Linking one end of a DNA probe to a surface is expected to in£uence its ability to form a duplex or interact with other analytes. Ideally, surface-con¢ned DNA probes should have adequate conformational freedom such that their behavior approaches that of DNA probes in solution. In addition to being tethered on one end, bound probes also di¡er from solution-phase probes in two other ways. First, the local concentration of bound probes is typically much higher than solutionphase probes; and, second, bound probes are located at a solid ^ liquid interface at which there is an abrupt discontinuity in the dielectric medium [86]. Numerous groups have explored the nature of the DNA probe/substrate interface and structural factors that a¡ect hybridization performance. Steric crowding as a function of probe surface coverage and the length of the linker used to tether probes to surfaces have been the two most studied interfacial parameters. To address these issues, many surface-sensitive methods have been applied to examine DNA probe interfaces directly. Measurements of the surface coverage of DNA probes and the target captured by hybridization are most informative. From these data, the hybridization e⁄ciency can be determined. The dispersion and orientation of probes and duplexes then can be inferred as well. The techniques include X-ray photoelectron spectroscopy (XPS) [59,87,88], ellipsometry [48,88,89], neutron re£ectivity [60], UV-visible spectroscopy [90], infrared (IR) spectroscopy [53,91], surface plasmon resonance (SPR) [92^95], scanning tunneling microscopy (STM) [96^98], atomic force microscopy (AFM) [56,62,88], £uorescence [47,49,99,100], and radiolabeling [47,101^103]. Electrochemical methods have also been used extensively to characterize the sensor interface, and many of these studies are cited in the section below on electrochemical transduction. We also discuss recent SPR studies of DNA hybridization kinetics in the SPR sensing section. 5.1

Experimental Studies of DNA Probe Surface Structure

Many of the issues related to DNA surface structure and hybridization have been explored by Southern and coworkers [41,89,102,104^106]. They

DNA-Based Sensors

555

extensively examined di¡erent spacer properties, including length, charge, hydrophobicity, and solvation, and found that length is the most important factor in in£uencing hybridization [102]. The spacers were built on aminefunctionalized polypropylene surfaces from a variety of monomeric units using phosphoramidite chemistry. With radiolabeling, Southern’s team determined that hybridization yields can be increased up to two orders of magnitude by introducing spacers between the oligonucleotides and the surface. An optimal spacer length of 40 atoms was found beyond which the hybridization yield decreases, presumably because the single-stranded probe ‘‘dissolves’’ in itself and becomes less accessible to targets. Guo has found similar enhancements in hybridization using poly-dT spacers for probes immobilized on glass supports [47]. In contrast to these studies, other groups have found that high hybridization e⁄ciencies can be attained using relatively short spacer groups. Using an electrochemical method [63] and neutron re£ection [60] to quantitate probe and target coverages, nearly 100% hybridization e⁄ciencies were achieved for probes linked to gold by a six-methylene spacer. The reason for the discrepancy between these studies is unclear; however, di¡erences in the substrates and the oligonucleotide coupling chemistries may play a role. The e¡ect of probe surface coverage on hybridization e⁄ciency has been examined by several groups. Most experimentally determined values for optimal probe coverage, i.e., those yielding the highest hybridization e⁄ciency, generally fall in the range of 1012^1013 probes/cm2. Putting this into perspective, the packing density of n-alkanethiol self-assembled monolayers (SAMs) on gold is 4.6  1014/cm2, or two orders of magnitude higher than that of DNA probes [107]. Krull’s group has studied in detail how the surface density of DNA probes in£uences steric and electrostatic interactions between bound probes and subsequent hybridization reactions [99,108]. In their experiments, ssDNA probes (dT20) were covalently bound to the surfaces of fused silica optical ¢bers using £exible hexaethylene glycol linkers attached via organosilane chemistry. Melting temperatures of duplexes containing a £uorescein-labeled complement (A20) were determined using total internal re£ection £uorescence. Melting temperatures and enthalpy changes calculated from melting pro¢les suggest that the thermodynamic stability of duplexes is dependent on the surface density of immobilized DNA with the highest surface density (4.6  1012 probes/cm2) showing the greatest destabilization. In addition, a greater depression of the Tm arising from a centrally located single base-pair mismatch was found for the surface immobilized duplex relative to the duplex in solution. Krull et al. have also examined whether nonspeci¢c binding of noncomplementary targets can alter the packing and charge density of probe surfaces, thereby adversely a¡ecting hybridization [100].The rate of nonspeci¢c adsorption of

556

Tarlov and Steel

noncomplementary targets was found to be faster than hybridization of complementary targets to the surface. Interestingly, the presence of nonspeci¢cally adsorbed targets or high concentrations of the noncomplementary target in solution did not seem to interfere signi¢cantly with hybridization of complementary targets. Evidence is beginning to emerge that the behavior of surface-con¢ned DNA is in many ways analogous to that of human-made polyelectrolytes. First, in work by Steel and colleagues, it was found that the random-coil nature of DNA probes directly in£uences their packing density [103]. In this study, the surface coverage of thiol-derivatized probes on gold surfaces was determined as a function of probe length using radiolabeling (see Fig. 3). A maximum surface packing density for nucleotides was found that indicates a transition in surface structure of the ssDNA as probe length increases. Probes shorter than 24 bases tend to organize in end-tethered, highly extended con¢gurations,while longer strands likely assume a more randomcoil, polymer-like con¢guration occupying a larger footprint. Second, as for the adsorption of polyelectrolyte brushes on surfaces, the surface coverage of bound DNA probes can also be dramatically in£uenced by ionic strength. It is well known that bu¡er composition and concentration have a large e¡ect on solid-phase hybridization yields [105]. A sharp onset in the adsorption of thiol-derivatized DNA on gold surfaces has been observed with increasing bu¡er concentration [59]. Presumably, intermolecular electrostatic repulsion between neighboring probe strands is minimized and higher surfaces coverages are attained under high-ionic-strength conditions. Similar observations have been reported for the adsorption of hydrophobically modi¢ed polyelectrolyte brushes, in which salt concentration acts as a virtual switch for adsorption [109]. Several studies have used surface and thin-¢lm characterization techniques that provide direct information concerning the interfacial structure of surface-con¢ned DNA probes. Levicky et al. used neutron re£ectivity to determine in situ concentration pro¢les of thiol-tethered DNA monolayers on gold [60].The results demonstrate that the conformation of ssDNA can be controlled by self-assembly methods to enhance hybridization. Chemisorption of a second alkanethiol molecule following probe immobilization was found to passivate the surface to nonspeci¢c adsorption and produce endtethered probes that are fully accessible for hybridization. Hybridization e⁄ciencies close to 100% were observed. Duplexes were also observed to orient near normal to the surface plane for coverages of 6  1012/cm2. Levicky et al. suggested that the relatively close packing of duplexes was responsible for their nearly perpendicular orientation on the gold surfaces. Scanning probe microscopy also has been used for direct visualization of DNA-modi¢ed surfaces; however, the success of scanning probe methods

DNA-Based Sensors

557

FIG. 3 Nucleotide surface density expressed in nucleotides per square centimeter as a function of the number of nucleotides in a given DNA probe for probes adsorbed on gold surfaces. Surface densities of nucleotides were determined by first measuring the probe density with phosphor imaging of 32P-labeled probes and then multiplying the probe density by the number of nucleotides in the probe. Data are from two types of probes: those with (closed circles, HS-DNA) and those without (open circles, Non-HS-DNA) a HS-(CH2)6-linker on the 50 end. The probe sequence was a repeated non-self-complementary 4-base unit, ACTG. Cartoons above plot illustrate two approximate structural regimes of thiol-derivatized probes in which shorter probes (<25 nucleotides) assume a primarily extended configuration, while longer probes (>25 nucleotides) assume a more random-coil-like configuration.

in determining the structure of relatively short immobilized DNA probes has been limited. Near-atomic resolution imaging of long dsDNA (>550 base pairs) on metallic surfaces has been achieved with STM [110,111]. On the other hand, Rekesh et al. report that short DNA probes (less than 33 bases)

558

Tarlov and Steel

must be tethered at both ends to be imaged, otherwise STM images are ‘‘blob-like’’ [97]. Recently Huang et al. used AFM to examine dispersion of thiol-derivatized ssDNA on gold surfaces [62].They found that DNA duplex formation occurs preferentially at grain boundaries of the gold substrate, suggesting that steric hindrance plays an important role in hybridization. Presumably, probes located in the vicinity of grain boundaries are more accessible to targets. Other groups have used AFM to determine the topography of DNA duplexes immobilized on gold surfaces [112,113]. 5.2

Theoretical Studies of DNA Probe Surface Structure

There has been a paucity of theoretical or modeling studies of DNA surface interactions.Graves and colleagues have developed a mathematical model of heterogeneous DNA hybridization on solution/solid interfaces [114]. In their model, hybridization occurs either by direct hybridization from solution or through nonspeci¢c, but reversible, adsorption of target to the surface followed by two-dimensional surface di¡usion and hybridization.Their results predict that for rapid and e⁄cient hybridization, probe surfaces should be designed to promote reversible adsorption and 2-D di¡usion of target strands. Wong and Pettitt recently reported the ¢rst all-atom molecular dynamics simulation of DNA tethered to a surface [115]. DNA duplexes were found to tilt spontaneously toward nearest neighbors and settle in a leaning position with an interaxial distance of 2.2 nm (see Fig. 4). This close packing of the DNAs, which a¡ects both in-situ synthesis and deposition of probes on microarray surfaces, was explained by salt-induced, colloid-like DNA ^ DNA attractions. It was postulated that the colloidal structures may form before DNA probe ^ strand attachment to the surface, implying that probes may adsorb as aggregates on the surface. In light of this result, probe attachment protocols may have to be revised to limit solution aggregate formation, a phenomenon that may result in deleterious steric crowding of probes. Clearly, the development of models and theories to describe the behavior of surface-con¢ned DNA systems is a fertile area for exploration. As demonstrated above, much progress has been made in elucidating the structure of surface-con¢ned DNA probes and the nature of their interactions with DNA targets. However, many of the details of the molecularscale architecture of these systems are still poorly understood and warrant further study. Although many sophisticated surface analytical tools have been brought to bear on surface-con¢ned DNA, these complex interfaces present a signi¢cant challenge in interfacial characterization. Our understanding of these structures will improve with the application of techniques capable of providing interfacial structural information in real time and in situ. With DNA emerging as a potential structural material for the

DNA-Based Sensors

559

FIG. 4 Snapshot of the molecular dynamics simulation at 6.3 ns. (a) and (b) are the views from two perpendicular directions. Each figure displays two duplexes. (a) In this view, the DNAs are pointing out of the page. (b) This view shows the DNA tilted toward its periodic image. Note the narrow space between the DNAs. (Reprinted with permission from Ref. 115. Copyright 2001 Springer-Verlag.)

560

Tarlov and Steel

fabrication of surface-con¢ned nanostructures, we anticipate that more attention will be focused on improving our fundamental understanding of DNA modi¢ed surfaces in the coming years. 6

DNA-BASED SENSORS FOR GENETIC ASSAY APPLICATIONS

The majority of DNA-based sensors have been developed for genetic assay applications. As described above, assays generally involve the detection of a known target sequence (or sequences) and frequently, a determination of its concentration. In addition, assays for the detection of mutations such as SNPs are becoming increasingly important.We outline below the numerous sensors developed for genetic assays, their operation, performance, and speci¢c applications. The overview is structured according to the transduction principle used: electrochemical, optical, mass, or scanning probe and microcantilever sensors. 6.1

Electrochemical Sensors

Electrochemistry is by far the most widely reported transduction method used in DNA-based sensors, as judged by the number of related publications. Several excellent reviews of electrochemical sensors for DNA sequence detection have already appeared [10,13]. The popularity of electrochemical methods stems from their demonstrated sensitivity and selectivity, as well as the relative ease and low cost of implementation. Furthermore, electrochemical methods are easily miniaturized and inherently sensitive to molecular species immobilized on electrode surfaces, properties that are exceptionally advantageous for developing DNA diagnostics based on surface-con¢ned hybridization for molecular recognition. Because of their structure and composition, nucleic acids are particularly amenable to electrochemical measurements. Numerous electrochemical detection strategies have been devised to exploit these unique properties. Direct electron transfer between DNA and certain electrode surfaces is possible; some components of DNA are electroactive within aqueous voltammetric limits, including the nucleobases and the ribose sugar. Electrochemical communication with nucleic acids is also possible through the use of redox mediators. Another approach is the use of redoxactive reporter molecules that selectively associate with DNA. For example, cationic electroactive molecules can interact electrostatically with the anionic ribose-phosphate backbone. The grooves of the DNA double helix may also serve as binding sites for a wide variety of electroactive molecular species. There are also examples of incorporating an electrochemically

DNA-Based Sensors

561

active label in DNA targets to facilitate detection. In addition, electrochemical techniques can often complement spectroscopic methods for the fundamental study of interactions between metal complexes and DNA. For example, small molecules that are not amenable to spectroscopic study, either because of weak adsorption bands or because of an overlap of electronic transitions with those of the DNA molecule, potentially can be studied using voltammetric techniques. In the following sections, DNA sensors are classi¢ed according to the strategy used for electrochemical transduction: direct electrochemistry, mediated electrochemistry, polymer electrode systems, redox indicators or electroactive intercalators, ampli¢ed systems, and engineered nanopores. Examples of electrochemical DNA-based biosensors used to detect analytes other than DNA are included in Sec. 7. 6.1.1 Direct Electrochemistry with DNA The roots of electrochemical detection methods for DNA can be traced to the seminal work of Palecek, who ¢rst noted the electroactivity of DNA using oscillographic polarography at a mercury electrode [116]. By using adsorptive stripping voltammetry, Palecek distinguished single-, and doublestranded DNA and created electrodes for the study of the interaction of immobilized DNA with substances in solution [117]. The purine bases of DNA, guanine and adenine, and their derivatives can be oxidized under a range of voltammetric conditions, and their electrochemical behavior has been examined at mercury, carbon, gold, indium tin oxide, and polymer electrodes [118]. The one-electron oxidation of guanosine occurs at 1.34 V (versus the normal hydrogen electrode), adenosine at1.79 V, and thymine and cytosine at much higher potentials [119]. Although nanomolar (nM) sensitivity levels have been reported for the oxidation of nucleobases [120], signi¢cant background currents must be contended with at the positive potentials required to e¡ect oxidation. Background-correction schemes have been investigated by Wang et al. that greatly improve the sensitivity of direct oxidation of guanine at carbon electrodes [121].The ribose sugar in the DNA backbone can also be oxidized, but this process destroys the nucleic acid and is thus better suited to a £ow-through detector rather than a DNAmodi¢ed electrode [118]. 6.1.2 Mediated Electrochemistry with DNA The oxidation of nucleic acids has been intently studied because of its potential role in initiating cancers and the natural aging process [122].Of the DNA bases, guanine is the most reactive with oxidants, alkylating agents, hydroxyl radicals, singlet oxygen, and transition-metal complexes [123]. Electrochemical oxidation of guanine in nucleic acids is the basis of a method developed by Thorp and colleagues for detecting the hybridization of DNA

562

Tarlov and Steel

targets. Thorp et al. have investigated the use of transition-metal complexes for mediating the oxidation of guanine in dsDNA and ssDNA, and have found ruthenium trisbipyridyl, Ru(bpy)32þ , to be an e¡ective oxidation catalyst for guanine. In this two-step mechanism, addition of DNA to a solution of Ru(bpy)32þ leads to a catalytic enhancement of the oxidation current, RuðbpyÞ32þ ! RuðbpyÞ33þ þ e RuðbpyÞ3 3þ þ DNA ! DNAox þ RuðbpyÞ3 2þ where DNAox is a DNA molecule containing a guanine that has been oxidized by Ru(bpy)33þ, or Ru(III), generated at an electrode surface. The reaction mechanism is catalytic in that the Ru(II) mediator is regenerated; the guanine is consumed through the formation of a radical cation that deprotonates rapidly and undergoes further reaction. The catalytic scheme was ¢rst investigated in solution, in which the rate constant for oxidation of guanine in freely di¡using calf thymus dsDNA by Ru(III) was reported as 9.0  103 M1s1 [124]. The oxidation rate was found to depend on the accessibility of the guanine. Guanines in ssDNA were more reactive than mismatched guanines in dsDNA, and properly matched base pairs of guanine (G to C) had the lowest reactivity of all, 200 times less than ssDNA. Thorp and co-workers also demonstrated the application of a similar electrocatalytic strategy to surface-immobilized nucleic acids, thereby allowing the detection of unlabeled targets. Several DNA-modi¢ed electrode schemes were developed, including covalent coupling of nucleic acids to dicarboxylic acid SAMs [125], covalent attachment to microporous polymer membranes [126], and irreversible adsorption from dimethylformamide/ acetate solutions [127]. In all cases, tin-doped indium oxide (ITO) electrodes were used; the electrolysis of water is minimal at the potentials required for the electrocatalytic reaction, and nonspeci¢c adsorption of targets is negligible at ITO electrodes. The catalytic oxidation of immobilized guanine by Ru(III) was detected using cyclic voltammetry or chronoamperometry (see Fig. 5). In a clever approach to limit background current originating from guanine bases contained in the immobilized probes, inosine bases were substituted for guanine in all of the synthesized probes. Inosine, the nucleoside derivative of hypoxanthine, has similar base-pairing properties to guanine but has much lower electrochemical activity than guanine, resulting in signi¢cantly lower background currents. A notable, but subtle, feature of the electrocatalytic scheme is that the measurement distinguishes between probe and target, not between ssDNA and dsDNA as in many DNA-sensing con¢gurations. Thus, an advantage of the approach is that the assay is capable of detecting single- or double-stranded DNA targets, as well as RNA for gene expression applications. In addition, because the mediator shuttles

DNA-Based Sensors

563

FIG. 5 Cyclic voltammograms (25 mV/s) of 200 mM Ru(bpy)32þ at a (A, dashed) DDCA-modified electrode to which poly[dC] has been attached and (B, dotted) after exposure of the electrode in (A) to poly[G]. Curve (C, solid) shows the voltammogram obtained at the hybridized electrode from (B) in the absence of Ru(bpy)32þ . Buffer and cell conditions are as in Fig. 1. (Reprinted with permission from Ref. 125. Copyright 1997 American Chemical Society.)

current to the electrode, a great variety of immobilization schemes for probes can be entertained to optimize conformation and recognition properties [126]. The utility of the three immobilization approaches has been demonstrated with various assays. The probe-derivatized polymer membrane electrodes were able to detect PCR-ampli¢ed genomic DNA from herpes simplex virus type II, genomic DNA from Clostridium perfringens, and genomic RNA from human immunode¢ciency virus [126]. For DNA irreversibly adsorbed to the electrode surface, a sensitivity of 44 amol/cm2, or  3 109 DNA molecules/cm2, was reported for the 1497 -bp PCR product from the HER-2 gene, a gene correlated with certain types of breast cancer [127]. 6.1.3 Polymer-Based Electrodes Several groups have investigated DNA/polymeric systems for electrochemical detection of DNA hybridization. The potential advantages of incorporating DNA in polymer systems include increased capacity and

564

Tarlov and Steel

greater stability relative to single monolayers of DNA probes. One of the ¢rst reports of a DNA-polymer modi¢ed electrode was by Korri-Youssou¢ et al., in which an electroactive co-polymer pyrrole ¢lm was electropolymerized on a Pt electrode [75,128]. DNA probes were covalently bound to the polypyrrole ¢lm by reacting amine-modi¢ed probes with ester groups contained in one of the pyrrole moieties. Upon exposure of the ¢lm to complementary target DNA, a decrease in electroactivity of the polypyrrole ¢lm was observed by cyclic voltammetry, as manifested by a lower oxidation current and a positive shift of the oxidation wave. The decrease in electrochemical activity was attributed to conformational changes occurring along the conjugated polymer backbone as a result of duplex formation. A sensitivity of 2 nM was reported for a 14-mer target. Wang and co-workers have also investigated the fabrication of composite ¢lms of polypyrrole and singlestranded DNA [129]. In this approach, the DNA serves as a chargecompensating counter ion that is incorporated during the electropolymerization of the polypyrrole ¢lm. The DNA is presumably electrostatically bound to the oxidized polypyrrole ¢lm. Experiments using an electrochemical quartz crystal microbalance suggest that the DNA is di⁄cult to expel from the polymer ¢lm when the polypyrrole is electrochemically reduced. Thorp et al. have also described using redox polymer ¢lms for electrocatalytic detection of singlestranded DNA [130]. Similar to previous work from this group, the detection strategy also relies on electrocatalytic oxidation of guanine residues by a Ru complex, but in the later study, the mediator is covalently linked to a polymer matrix. The polymer-modi¢ed electrodes were prepared by electrodeposition of a copolymer of p-vinylbenzoic acid and polypyridyl complexes of Ru(II) on glassy carbon or Pt surfaces. Amine-modi¢ed DNA probes were then covalently attached via carbodiimide reaction to the carboxylate groups. Using cyclic voltammetry, the electrocatalytic oxidation of a covalently bound 20-mer of poly(dG) at a coverage of 7 1012/cm2 was detected with an estimated current e⁄ciency of 65%. 6.1.4 Redox Indicators In general, there are four properties of redox indicators that a¡ect sensor performance in DNA-based biosensors: formal potential, a⁄nity for DNA, selectivity of dsDNA over ssDNA, and low nonspeci¢c binding to the electrode. The formal potential should be readily accessible within aqueous voltammetric limits. The redox indicator should also bind preferentially to dsDNA versus ssDNA when trying to detect duplex DNA. Likewise, the redox indicator should exhibit little nonspeci¢c binding to the electrode surface so as to minimize background currents. Conjugation of the redox indicator to dsDNA can be achieved through a covalent bond or through

DNA-Based Sensors

565

adduct formation. An example of covalent attachment of a redox indicator to DNA for measurement was given by Palecek and Hung, who determined nanogram quantities of DNA by labeling the nucleic acid with osmium oxide in the presence of pyridine [131]. The osmium label incorporated into dsDNA to a lesser extent than ssDNA and was thereby used as a probe of DNA structure. Additional examples of direct incorporation of redox indicators into DNA has been reported; however, the most documented of the electrochemical DNA-based biosensors involve redox indicators that are selective for the hybridized complex over the immobilized probes. Two primary modes of interaction are available for redox indicators with DNA: electrostatic interactions with the charged backbone or hydrophobic/intercalative interactions inside the grooves and the base stack of the helix. Conjugate formation is often an interplay of these two types of interaction, with signi¢cant ionic strength dependence [132]. Millan and Mikkelsen developed a voltammetric method to monitor DNA hybridization based on the electrostatic interaction of cobalt(III) bipyridyl (CoBpy) with the anionic backbone of DNA [133]. Carbon-paste electrodes were modi¢ed with probes using carbodiimide chemistry. Electrodes modi¢ed with calf thymus ssDNA and dsDNA, and synthetic ssDNA homopolymers, were examined using cyclic voltammetry of roughly 100 mM cobalt tris(bipyridyl) in 5 mM tris bu¡er with 20 mM NaCl. The peak current at the poly(dG) homopolymer was signi¢cantly larger than the other electrodes, indicating that the guanine residue is critical to the immobilization procedure. At a hybridized electrode, both the faradaic and charging currents increased and the CoBPY formal potential shifted negatively by 31 mV. The negative shift in formal potential is indicative of the 3þ cobalt complex being stabilized by the DNA layer on the electrode. In a prototype sequence-selective sensor for polyA-containing targets, Millan and Mikkelsen were able to show that immobilized DNA probes provided recognition through hybridization that could be detected voltammetrically by preconcentration of the CoBPY complex at the dsDNA electrode surface layer. In an extension of this work, Millan et al. reported a voltammetric DNA biosensor for cystic ¢brosis [134]. Steel et al. have developed an approach to quantitate DNA at gold electrodes via electrostatic trapping of cationic redox molecules in the polyanionic DNA layer at electrode surface [63]. Using thiol-driven self-assembly, DNA-modi¢ed gold electrodes were prepared with variable loadings of probe [59]. Accessibility of immobilized probes to complementary target sequences was enhanced by treating the surface with a small-molecule blocking agent, for example, 6-mercapto-1-hexanol. The thiol group of the blocking agent displaced weaker adsorptive contacts between DNA nucleotides and the substrate, leaving the probes tethered primarily through

566

Tarlov and Steel

the thiol end groups. Quantitation was made on the basis of three assumptions: (1) the redox marker associated with DNA strictly through electrostatic interactions; (2) all redox molecules in the DNA surface layer were electrochemically accessible; and (3) charge compensation for the DNA phosphate functionalities was provided solely by the redox marker. Purely electrostatic interaction was assumed for ruthenium(III) hexamine in lowionic-strength bu¡er based on literature reports. In the low-salt bu¡er, the redox indicator exchanged for labile, native counterions associated with the phosphate residues of the probe was determined using chronocoulometry. Typical chronocoulometric responses for RuHex at a control, ssDNA, and dsDNA electrode are given in Fig. 6.The data for the control electrode shows negligible nonspeci¢c adsorption of the redox indicator at the electrode surface in the absence of DNA. The increase in surface charge by 1.18 mC at the DNA-modi¢ed electrode corresponds to a ssDNA density of 3.9  1012 probes/cm2. The accumulated charge nearly doubled at the hybridized dsDNA 25-mer electrode, indicating almost complete hybridization of the initial ssDNA probes. From the reproducibility of the double-layer charge determination, a detection limit of 1 1011 DNA strands/cm2 was estimated. Using this quantitation method, Steel et al. observed ideal hybridization behavior for 25-mer probe densities below  4  1012 probes/cm2 and decreasing hybridization e⁄ciency with increasing probe density. Background currents often limit the sensitivity of electrochemical DNA hybridization sensors. DNA-modi¢ed electrodes often have poorly characterized probe morphology, and the electrochemical response of these systems is often marked by signi¢cant nonfaradaic currents. For example, gold electrodes modi¢ed with denatured calf thymus DNA showed signi¢cant nonfaradaic currents, as did carbon electrodes [134,135]. Several novel methods have been reported to di¡erentiate the current generated in response to the hybridization event from nonspeci¢c currents. Background correction strategies have been e¡ectively developed by Wang and coworkers, including constant-current chronopotentiometry and baselinecorrected adsorptive stripping square voltammetry, improvements that have led to femtomole detection limits for nucleic acids [121,131]. Another background-reduction method that has been particularly successful in improving DNA sensing is the use ‘‘electrochemically silent’’ probes. As mentioned above, Thorp et al. have detected hybridization via oxidation of guanine bases, and limited the background current by substituting ‘‘electrochemically silent’’ bases in place of guanine in their probes. Another example comes from Wang’s group, where hybridization sensors with near-zero background were prepared by using peptide nucleic acid (PNA) probes in place of ssDNA [136]. PNA contains the same nucleotide bases DNA but has an electrically neutral peptide-like backbone. In these

DNA-Based Sensors

567

FIG. 6 Chronocoulometric traces at electrodes with no DNA (squares), ssDNA (circles), dsDNA (diamonds) in the absence (open symbols) and presence (closed symbols) of an electrochemical reporter, ruthenium(III) hexaamine at 50 mM. The lines represent best fits to the data at long times following the potential pulse, t > 0:2 s supra 1/2. The slope of the line is proportional to the electrochemical reporter concentration and is the same for all three traces. The intercept is equal to the sum of the double-layer charge and the charge due to electrochemical reporter absorbed at the electrode surface. The intercepts for the no-DNA measurements are the same, indicating minimal nonspecific adsorption of the electrochemical reporter in the absence of DNA. The intercept for the dsDNA electrode is nearly twice that of the ssDNA electrode, indicating nearly complete hybridization efficiency.

studies, a redox indicator, Co(phen)33þ , which associates with DNA primarily through electrostatic interactions,was used. Because PNA is neutral, association of a positively charged redox indicator with PNA probes is expected to be signi¢cantly reduced, and, indeed, a much lower background chronopotentiometric response was observed. In addition, PNA probes provided higher speci¢city and faster hybridization kinetics than ssDNA probes. Using a 17-mer PNA probe, Wang et al. were able to detect point

568

Tarlov and Steel

mutations in the p53 gene, SNPs thought to be correlated with the ability to control the development of certain types of cancers [137]. The selectivity of the redox indicator for dsDNA over ssDNA and the nonspeci¢c binding properties with the electrode material are also critical to reducing background currents. Electrochemical reporters based on selective conjugation with dsDNA can also provide limited background currents. For example, Hashimoto and colleagues used the electroactive dye and heteroaromatic DNA minor groove binder, Hoescht 33258, for sequence-speci¢c gene detection at a DNA-modi¢ed gold electrode [138]. An end-tethered geometry was e¡ected by incorporating a mercaptohexyl group at the 50 phosphate end of synthetic probes. The dye, which is irreversibly oxidized at 550 mV (versus Ag/AgCl), was found to bind roughly two times more strongly to dsDNA than to ssDNA probes. The selectivity of Hoescht 33258 was su⁄cient to permit discrimination between two targets, one of which was complementary to the ssDNA probe. Sensitivity was limited by nonspeci¢c adsorption of the dye to the gold electrode. In subsequent work, Hashimoto et al. described a microfabricated, disposable DNA sensor for the detection of the hepatitis B virus with a reported sensitivity of 10,000 copies of the virus plasmid per milliliter [139]. Takenaka and coworkers synthesized a ‘‘threading intercalator,’’ a DNA ligand containing a major substituent group attached to the intercalating moiety that must thread through adjacent DNA base pairs of duplex DNA when the ligand becomes bound. The threading intercalator used in these studies, a ferrocenylnaphthalene diimide derivative, also contained two redox-active ferrocenyl functionalities for electrochemical reporting [140]. The ligand was thought to interact simultaneously with the major and minor grooves of dsDNA, and the resulting intercalator/DNA complex exhibited enhanced thermodynamic and kinetic stability. Spectrophotometric measurements indicated stronger binding to dsDNA than to ssDNA by a factor of 4 and a dissociation rate that is 80 times slower for dsDNA than ssDNA.Thiol-modi¢ed DNA probes were immobilized on gold electrodes and cyclic and di¡erential pulse voltammetry (DPV) were used to monitor the binding of the threading intercalator electrochemically [141]. By lowering the density of adsorbed probes and passivating the surface with a SAM of mercaptoethanol to minimize nonspeci¢c adsorption of targets, greater sensitivity was achieved. Using DPV to monitor the capture of a homopolymer dT20 target by a dA20 probe-modi¢ed electrode surface, a sensitivity of 10 zmol was claimed for solution volumes of 1 mL (1014 M). A roughly logarithmic dependence of the DPV signal on target concentration was also observed from 1014^1012 M. Several other electroactive intercalators have been investigated. Using the redox-active, DNA-intercalating, anticancer drug daunomycin,

DNA-Based Sensors

569

Marrazza and coworkers developed a disposable DNA electrochemical sensor on screen-printed graphite electrodes [142^144]. The drug accumulated in the DNA layer, where the oxidation potential of daunomycin shifted relative to that for free daunomycin, indicating the increased stability of the conjugate. The groups of Barton and Hill have developed a novel electrochemical strategy for the detection of single-base mismatches based on the use of electroactive intercalators [145,146]. Hybridization assays with direct electrochemical reports can be limited in sensitivity for single-base mismatches because the duplex structure for fully complementary and mutated sequences have similar stability. The approach by Barton and Hill relies on variations in electrochemical communication with a redox intercalator residing at the end of the DNA duplex located farther away from the electrode surface. Modi¢ed electrodes were prepared by the self-assembly on gold surfaces of 15-bp duplexes derivatized at the 50 -end with a thiol-terminated aliphatic linker [45]. The assembly procedure was thought to produce electrode surfaces presenting densely packed duplexes [145,146]. The electrochemistry of the aromatic heterocyclic compound methylene blue (MB), a known electroactive DNA intercalator, was reported. The MB-binding constant K = 3.8 10 6 M1 at the surface compared favorably with several literature reports for MB ^ DNA adduct formation. However, the extent of binding at saturation was signi¢cantly less than anticipated, with a binding site size of 10 base pairs at the electrode relative to 2 base pairs for solution DNA. The subsaturation association was thought to result from the densely packed nature of the duplex-covered surface, whereby MB can access only a limited number of sites at the end of each duplex. Similar behavior was also reported for daunomycin, another electroactive intercalator that was used to signal electrochemically the presence of mismatches. A decrease in current e⁄ciency by roughly a factor of 2 was observed for the reduction of daunomycin when intercalated in a duplex containing a C ^ A mismatch relative to the fully base-paired complement (T ^ A) [146]. It was suggested that the presence of mismatches disrupts charge transport through the duplex, thereby attenuating the electrochemical response of intercalators located at the periphery of the duplex monolayer. An electrocatalytic process was also developed to amplify the intercalator signal. In this method, MB acted as the intercalated catalyst while potassium ferricyanide was the solution substrate. Using this strategy,the C ^ A mismatch displayed a sixfold lower current than the perfectly matched complement T ^ A. It should be noted that Willner and colleagues have also developed a sensitive electrochemical and microgravimetric method for the detection of single-base mismatches [147]. The strategy is described in greater detail in Sec. 6.3.

570

Tarlov and Steel

6.1.5 Amplified Detection To increase assay sensitivity, several groups have developed strategies to amplify the electrochemical signals associated with the hybridization event. One such example is by Bard and colleagues, who have developed a DNA detection system based on electrogenerated chemiluminescent detection [148,149]. Gold electrodes were modi¢ed with DNAvia a self-assembled ¢lm that presented cationic metal centers for electrostatic binding to the phosphate groups in the DNA backbone. DNA adsorbed to the aluminum(III) alkanebisphosphonate ¢lm was detected by electrogenerated chemiluminescence of ruthenium(II) phenanthroline (RuPHEN). RuPHEN emits light when oxidized electrochemically in the presence of a suitable coreactant. RuPHEN was shown to associate with dsDNA by intercalation and, thus, used to detect solution-phase DNA. The electrostatic-based assembly method was used to immobilize ssDNA including homopolymers and short, 8 and 30 bases, synthetic oligonucleotides. The immobilized ssDNA probes recognized complementary strands as determined by electrogenerated chemiluminescence detection. A di¡erent class of electrochemical ampli¢cation has been reported by Heller’s group, who developed a ‘‘wired’’ redox polymer for amperometric detection and enzymatic ampli¢cation of target DNA [73,74]. A two-step process is used for forming the DNA probe containing redox ¢lms on carbon microelectrodes. In the ¢rst step, a thin ¢lm of an electron-conducting, acrylamide-based, redox polymer is deposited electrophoretically on the microelectrodes. In the second step, carbodiimide-activated DNA probes are electrophoretically deposited and covalently attached to the redox polymer ¢lm.Target DNA is labeled at its 50 end with a soybean peroxidase (SBP) enzyme. Accompanying hybridization, electrochemical communication is established between the electrode and the SBP enzyme through osmium redox centers located in the acrylamide ¢lm. A current due to the electroreduction of H2O2 to water is then measured amperometrically. Using this method, it was possible to measure a current corresponding to 4  10 4 SBPlabeled 18-mer targets using a 7-mm-diameter carbon microelectrode. In addition, because of the relatively good thermal stability of SBP, melting programs could be used to increase stringency for detection of mismatches. By monitoring currents at 25, 45, and 57 C, single-base-pair mismatch was detected in an 18-mer duplex. Willner and colleagues have also developed an electrochemically based ampli¢cation scheme based on coupling of biotin/avidin and functionalized liposomes [147]. Hybridization events induce the formation of a network of negatively charged liposomes that act as a barrier to electron transfer between the electrode and a negatively charged redox couple in solution.

DNA-Based Sensors

571

Using faradaic impedance spectroscopy to measure the change in interfacial electron transfer properties, a DNA target sensitivity of 1013 M was achieved. Transduction was also demonstrated with a quartz crystal microbalance. This strategy is discussed in greater detail in Sec. 6.3. 6.1.6 Engineered Nanopores A relatively new development in electrochemical-based DNA sensors is the measurement of ionic conductivity through engineered nanopores. The promise of the method lies in its extraordinary sensitivity and speed for characterizing nucleic acids. Indeed, microsecond sequencing of single DNA molecules may not be unrealistic. In this nonfaradaic approach pioneered by Kasianowicz and coworkers, a single pore-forming protein molecule, typically a-hemolysin (aHL), is incorporated in a lipid bilayer that forms an electrically insulating barrier between two chambers ¢lled with electrolyte solution [150]. Application of a voltage across the membrane causes ions to £ow through the protein pore and a current on the order of 100 pA is measured. The diameter of the nanopore is such that only individual molecules of ssRNA or ssDNA can ¢t through the pore.When a nucleic acid molecule traverses the membrane, the ion channel becomes partially blocked. The passage of each molecule is signaled by a transient decrease in ionic current for which the duration of the transient is proportional to sequence length. Channel blockades can thus be used to measure sequence length. In addition, the repetition rate of channel-blocking events is directly proportional to the concentration of the molecule in solution [151,152]. To gain a better understanding of the mechanism by which single molecules are driven through nanometer-scale pores, Kasianowicz and colleagues recently studied some of the factors that in£uence the interaction between DNA and aHL ion channels [152].They examined the transport of a 30-nucleotide-long homopolymer of deoxycytidylic acid biotinylated at the 50 end, bT-poly(dC)30, through a single aHL ion channel in a planar lipid bilayer of diphytanoyl phophatidylcholine. The blockade frequency was found to be proportional to the polymer concentration and to depend exponentially on applied potential (see Fig. 7A). The blockade rate was described well by a van’t Ho¡-Arrhenius or transition-state relationship and a barrier height of 8kT was estimated. An interesting asymmetry was observed in the £ux of DNA depending on the direction of the applied potential, i.e., which side of the protein pore the DNA is driven through. Considering the crystal structure of the channel protein, Kasianowicz et al. postulated several possible mechanisms for the observed asymmetry in the blockade rate: (1) electrostatic repulsion and attraction e¡ects resulting from negatively and positively charged amino acid residues, respectively, on opposite sides of

572

Tarlov and Steel

FIG. 7 (A) Voltage and concentration dependence of the rate of blocking of the current by a polynucleotide. The number of blockades per minute depends markedly on the magnitude and direction of the applied potential V and on which side polynucleotide is added. (Main graph) The rate versus the magnitude jVj when 400 nM bT poly(dC)30 is added to the cis side (open squares), and when 800 nM is added to the trans side (solid squares). The polarity of V in each case is such that negative ions are driven from that side into the pore. The solid lines are

DNA-Based Sensors

573

the aHL; and (2) lower and higher entropic barriers on opposing sides of the aHL, resulting from asymmetrically shaped pore openings (see Fig. 7B). An extension of this method was recently reported by Howorka et al., in which sequence selectivity was introduced by modi¢cation of the pore entrance with a DNA probe. Here, a single ssDNA probe was attached via a disul¢de bond to a cysteine residue, introduced by mutagenesis, located near the mouth of the aHL pore. The hybridization of target ssDNA to the tethered probe was detected as a decrease in the ionic current,which lasted up to tens of milliseconds. Based on di¡erences in lifetimes of DNA duplexes, the DNA-nanopores were able to distinguish between perfectly complementary targets and those with single-base mismatches. Detection of a single base mutation of a 30-mer that confers drug resistance in the reverse transcriptase gene of HIV was demonstrated by analysis of the distribution of the channel-blocking event lifetimes and amplitudes. An alternative strategy to modifying ion channels was also recently demonstrated. In this approach, a molecular recognition element is attached to a polymer that threads through the nanopore [153]. Binding of an analyte molecule to the recognition element alters the ability of the polymer to thread through the pore which is manifested as a change in ionic current. Proof of concept was demonstrated using biotinylated ssDNA as the porepermeant polymer and streptavidin as the analyte. The advantages of this approach are that the sensor can be used to detect multiple analytes simultaneously, the range of analytes that can be detected is greatly expanded, and analyte molecules do not need to ¢t inside the pore in order to be detected. For the nanopore method to realize its full potential for DNA sequence applications, a number of challenges must be overcome, including developing a better understanding of the £ow of biopolymers through nanopores,

3 least-squares fits of a van’t Hoff-Arrhenius or transition-state relation (see text) to the data. (Inset) The blockade rate is proportional to the concentration of polymer added to the cis side when V ¼ 120 mV or to that added to the trans side when V ¼ þ120 mV. At a given concentration, polymer added to the cis side is about six times more likely to transiently block the pore than when it is added to the trans side. (B) Candidate locations for polynucleotide^pore interactions. Polynucleotide entry from the cis side into the pore may be favored over that from the trans side because of the relatively large vestibule (the entropic barrier is lower for larger confinement volumes) and/or electrostatic attraction (or repulsion) caused by charged amino acid side chains in key regions. A major barrier for polynucleotide entry may also be the physical constriction that is shown schematically in this representation of the channel’s crystal structure. (Reprinted with permission from Ref. 152. Copyright 2000 American Physical Society.)

574

Tarlov and Steel

improving the robustness of nanopores, and devising strategies for constructing arrays of individual electronically addressable nanopores. Regarding the issue of pore robustness,work is ongoing to fabricate arti¢cial nanopore structures using silicon micro- and nanomachining techniques. 6.2

Optical Sensors

Potential advantages of optical transduction of DNA-based sensors are high sensitivity and relative simplicity because sensor readout does not require individual electronic accessibility to each probe sequence. The lack of hardwired connections also can save valuable surface real estate, facilitating the fabrication of high-density arrays. Fluorescence labeling of DNA targets (and other proposed optical labels [154,155]) is the most commonly used method to read out DNA microarrays; however, potential disadvantages include the extra labor-intensive step of target labeling and disparate coupling e⁄ciencies to targets. We therefore restrict our review primarily to novel optical strategies that obviate the need for target labeling. 6.2.1 Surface Plasmon Resonance (SPR) Other than £uorescence, surface plasmon resonance (SPR) is the most widely used optical method for detection in DNA-based sensing and DNA microarrays. SPR is a surface-sensitive optical technique that can detect changes in refractive index resulting from the adsorption of molecules at thin ( 50-nm) metal surfaces, most commonly gold [156,157]. SPR has been applied to the study of alkanethiol SAMs [158] and biological a⁄nity reactions [159^161] at surfaces.The principal advantages of SPR for DNA-based sensing are that the labeling of DNA targets is not required, hybridization data can be obtained in real time, and it is easily adapted for in-situ detection. In addition, SPR can be con¢gured in an imaging mode for simultaneous detection of spatially localized adsorption events with a spatial resolution of 50 mm or less [162]. We note that optical waveguiding methods for DNA detection, an approach similar to SPR for measuring refractive index changes, have also been reported [163]. Numerous SPR studies of DNA probe interfaces have demonstrated unequivocally that sensitivity is adequate to detect hybridization of unlabeled targets, in real time, and that an imaging con¢guration can detect spatially separated hybridization events [52,92,93,164,165]. Almost all of these studies examined end-tethered DNA probe monolayers immobilized on thin gold ¢lms by direct adsorption through a thiol linker or by covalent attachment of derivatized probes to preformed SAMs. When using SPR in the scanning mode, a detection limit for DNA probe or target sequences was reported to be approximately 1011 molecules/cm2, which corresponds

DNA-Based Sensors

575

to 2^5% of a typical DNA probe monolayer [165]. Using near-infrared SPR imaging [166,167], a similar sensitivity was recently demonstrated by Corn and colleagues for the hybridization of both DNA and RNA. This surface sensitivity corresponds to a detection limit of 10 nM for DNA and RNA 18-mers,while that for 1500-base ribosomal RNAwas 2 nM (see Fig. 8) [54]. Using SPR and other techniques as tools for characterization, improved probe attachment chemistries have been developed that enhance DNA probe layer robustness and hybridization e⁄ciencies [53,168]. These

FIG. 8 SPR image showing hybridization adsorption of 50 nM DNA 18-mer oligonucleotides onto a DNA-modified surface array. Hybridization adsorption onto the array is indicated by a change in the percent reflectivity of incident light. The pattern used for immobilization of single-stranded DNA probe sequences A and B is shown in (a). Hybridization adsorption occurs at perfect match spots after exposure to a 50 nM solutin of DNA complement A0 for 30 min, as shown in (b). The surface is briefly denatured with 8 M urea, and exposed for 30 min to a 50 nM solution of DNA complement B0 , resulting in the image seen in (c). Arrays were denatured and hybridized up to 25 cycles without a significant loss in signal or specificity. Plot profiles (below) taken across the dotted line in the fifth row show the excellent specificity and signal-to-noise ratio for both interactions. The resulting change in reflectivity (%R) upon hybridization for both arrays is 0.6%. (Reprinted with permission from Ref. 54. Copyright 2001 American Chemical Society.)

576

Tarlov and Steel

attachment protocols also facilitate the fabrication of DNA probe arrays that are compatible with SPR imaging. While SPR sensitivity is adequate for fundamental studies of surfacecon¢ned hybridization, researchers have sought to increase sensitivity for DNA sensing through ampli¢cation strategies. One of the ¢rst examples of ampli¢cation was by Jordan et al., who showed that subsequent attachment of streptavidin to biotinylated complements results in a fourfold improvement in the hybridization detection limit in SPR imaging [165]. Recently, Keating and colleagues reported even greater enhancements in sensitivity using another sandwich-type assay [55]. Keating et al. used Au nanoparticles to amplify the SPR response by 1000-fold. The sandwich ampli¢cation assay format is represented schematically in Fig. 9a. DNA probes (S1), covalently linked to an alkanethiol SAM on thin ¢lm gold, are complementary to half of a target strand (S2), a 24-mer. SPR ampli¢cation results from the binding of Au-nanoparticles that are tagged with a DNA probe (S3) that is complementary to the remaining unhybridized half of the target sequence (see Fig. 9b). Reliable detection of 24-mer targets down to 10 pM was reported corresponding to target coverages of 8 108 molecules/cm2, a detection limit approaching that for £uorescence methods ( 107 molecules/cm2). They also suggest that lower detection limits may be possible by reducing nonspeci¢c adsorption of the DNA-derivatized Au nanoparticles to the probe-modi¢ed Au surface. Three factors are thought to be responsible for the large enhancement: (1) greatly increased surface mass, (2) the high dielectric constant of the Au nanoparticles, and (3) electromagnetic coupling between the Au nanoparticles and the Au ¢lm. Keating et al. also demonstrate that use of the Au nanoparticle ampli¢cation method is compatible with SPR imaging to detect hybridization events on DNA arrays. 6.2.2

Fundamental Studies of Surface Hybridization with SPR

SPR has also provided fundamental insights into the hybridization and structure of DNA-sensing monolayers. Exploiting the real-time monitoring capabilities of SPR, thermodynamic and kinetic studies of duplex formation have revealed the complex behavior of DNA probe interfaces. Georgiadis and co-workers have used two-color SPR for quantitative measurements of surface probe and target coverages [93,95]. By acquiring surface plasmon curves at two di¡erent wavelengths, both the thickness and dielectric constant for an unknown adsorbed ¢lm can be determined unambiguously [169]. For thiol-derivatized DNA probes self-assembled on gold at coverages of 5 1012/cm2, hybridization has been found to conform to a di¡usion-limited Langmuir adsorption model suggesting limited lateral interaction between probes and duplexes formed [93]. Langmuir-like hybridization behavior for both DNA and RNA 18-mers was also recently reported by others [54].

DNA-Based Sensors

577

FIG. 9 (a) SPR surface assembly. (b) SPR curves of surfaces prepared in sequential steps as illustrated in Scheme 1: a MHA-coated Au film modified with a 12-mer oligonucleotide S1(A), after hybridization with its complementary 24-mer target S2 (B), and followed by introduction of S3: Au conjugate (C) to the surface. (Inset) surface plasmon reflectance changes at 53.2 for the oligonucleotide-coated Au film measured duringa 60-minexposure to S3:Au conjugates. (Reprinted with permission from Ref. 55. Copyright 2001 American Chemical Society.)

578

Tarlov and Steel

Noteworthy is the use by Knoll and colleagues of a derivative of SPR, surface-plasmon (SP) ¢eld-enhanced £uorescence spectroscopy, for realtime studies of hybridization of surface-tethered DNA probes [170]. This approach likewise exploits the resonant excitation of an evanescent SP mode, but in this case the strong optical ¢elds obtained at resonance are used to excite £uorescently labeled DNA targets. Although target labeling is required, the method o¡ers high sensitivity ( 1010 oligonucleotides/cm2) for fundamental studies of surface hybridization reactions. Moreover, the problem of background £uorescence from bulk solution is largely circumvented because only £uorophores within the evanescent ¢eld are excited by the SP mode. In Knoll’s studies, biotinylated 15-mer probes were anchored via biotin/streptavidin coupling to biotinylated alkanethiol SAMs on gold surfaces. By plotting equilibrium £uorescence intensities as a function of solution concentration of labeled 15-mer targets, adsorption isotherms were generated and Langmuir-like behavior was observed. A single mismatch was found to reduce the hybridization a⁄nity constant by two orders of magnitude, while a second mismatch resulted in a ¢ve-order-ofmagnitude reduction relative to the perfect complement. The kinetics of hybridization have been found to be sensitive to both the presence of mismatches and to the position of hybridization along the probe relative to the surface [94]. Faster hybridization was observed for a complementary 25-mer versus a 25-mer with two base-pair mismatches, behavior rationalized by the di¡erence in thermal stability of the two duplexes. In experiments exploring the position of the mismatch, the rate of hybridization was measured for two di¡erent 18-mers complementary to a thiol-modi¢ed 25-mer probe immobilized on gold. One 18-mer, 18low, was complementary to the ¢rst 18 bases closest to the surface, while the other, 18high, was complementary to the 18 bases farthest away. The hybridization of 18high was found to be faster than that of 18low even though the thermodynamic stabilities of the two duplexes measured in solution are equivalent. The observed behavior was attributed to a kinetic e¡ect whereby, according to postulated models of hybridization, duplex formation starts with a transient nucleation event involving the bonding of few base pairs followed by zippering [41,171]. Thus, for the 18low target to hybridize, it must penetrate deeper into the DNA probe monolayer than the 18high target, a kinetically slower process. Recent SPR studies by Georgiadis and coworkers point to an interesting ¢nding that electric ¢elds can be tuned at DNA probe/metal interfaces to enhance rates of hybridization and to better discriminate base-pair mismatches, albeit a form of electronic stringency [172]. In these studies the gold substrate to which DNA probes were end-tethered was used as an electrode and hybridization rates were recorded with SPR as a function of applied

DNA-Based Sensors

579

potential. Relatively low potentials were applied, i.e., only those resulting in nonfaradaic, charging currents. Hybridization rates were observed to increase and decrease with charging of the electrode positively and negatively, respectively (see Fig. 10). Presumably a positively charged electrode surface attracts negatively charged DNA targets, while a negative surface repels targets. In addition, application of repulsive potentials was found to preferentially denature duplexes with base-pair mismatches relative to those with perfect complementarity. The term ‘‘electronic stringency’’ was ¢rst coined to describe hybridization enhancement e¡ects observed in

FIG. 10 Comparison of the kinetics of hybridization at open circuit (squares) and under electrochemical control at an applied potential of þ 300 mV (circles) and 300 mV (triangles). Data are shown for the interaction of the fully complementary target (solid symbols) for the 2-bp mismatched target (open symbols). For each electrochemical experiment, hybridization proceeds for 3 min before the selected potential is applied (see text for details). Here only the first 30 min are shown, although each hybridization proceeds for at least 5 h. The magnitude of hybridization is compared with the steady-state value reached after 14 h of unassisted (open-circuit) hybridization in an unstirred cell. All experiments were carried out in 1 M NaCl solution containing 1mM oligonucleotide probe on the same immobilized ssDNA thiol probe film. The probe coverage remains constant (within 10%) when regenerated by rinsing with hot water between runs. (Reprinted with permission from Ref. 172. Copyright 2001 National Academy of Sciences.)

580

Tarlov and Steel

microelectronic chips using electrophoresis [172,173,174]. However, it is not clear that the phenomena observed in these devices is the same as that observed by Georgiadis and coworkers. In the microelectronic chips, DNA probes are immobilized in hydrogels signi¢cantly removed from electrode surfaces. Applied potentials also are high enough to cause hydrolysis of solutions. More closely related to the electrostatic e¡ects reported by Heaton et al. is a study by Kelley et al. demonstrating with AFM that the application of electric ¢elds causes reorientation of thiol-tethered duplexes on gold electrodes [112]. In a similar context, electrostatic e¡ects have also been invoked to explain pH-tunable hybridization rates observed for streptavidin-modi¢ed DNA microarray surfaces [175]. A recently developed thermodynamic theory also supports the contention of electronic control of interfacial hybridization reactions [176]. Further experimental and theoretical studies are required to understand fully the mechanism of these promising electrostatic e¡ects. 6.2.3 Other Optical Methods Tan et al. have exploited £uorescent molecular beacons (MBs) for DNAbased sensing [177]. A MB is a ssDNA probe that contains a looplike structure and a stem (see Fig. 11) [178].The loop is formed by the pairing of four to six complementary bases located at opposite ends of the strand, while the paired section becomes the stem. On opposite ends of the strand are a £uorophore and quencher that are brought into close proximity from the formation of the stem, thus quenching £uorescence. Upon encountering a complementary target, hybridization occurs and a rigid duplex is formed. This forces the stem apart and spatially separates the quencher and £uorophore, thereby restoring £uorescence. The MBs have been immobilized on planar silica and ¢ber-optic surfaces using organosilane and biotin ^ avidin coupling. An evanescent wave geometry is employed for £uorescence detection and detection limits in the1-nM range have been reported for DNA and RNA targets [68,177,179]. MBs also have been incorporated in a randomly ordered ¢ber-optic microarray for detection of cystic ¢brosis-related targets [180]. The microarrays were prepared by randomly distributing MB-derivatized microspheres in an array of wells formed by the etching of bunched optical ¢bres. The optical properties of porous silicon have been exploited for sensitive DNA detection via interferometric sensing [181]. Here, transduction is based on induced wavelength shifts in the Fabry-Perot fringes in the visiblelight re£ection spectrum of appropriately derivatized thin ¢lms of porous silicon. Hybridization of targets to DNA probes immobilized on the porous silicon surface causes a change in the refractive index of the nanocrystalline surface that shifts the wavelength of the fringe pattern. The thin (1^5 mm)

DNA-Based Sensors

FIG. 11

581

Cartoon describing the operation of a DNA molecular beacon.

layer of porous Si is prepared by electrochemical etching of single crystal silicon. Cavities of up to 200 nm in diameter provide a large surface area to immobilize probes. The DNA probes are immobilized on the porous silicon by covalent linkage to an organosilane monolayer. For 16-mer targets, a detection limit of approximately 100 fM was reported, corresponding to a target surface coverage of 108/cm2. The reasons behind the exceptional sensitivity are not fully understood, but the sensing mechanism may involve a double-layer-induced dielectric change in the porous silicon layer. Particularly promising approaches for simple diagnostic assays involve visual detection by the human eye. Several novel strategies for visual detection have recently appeared. The ¢rst method, developed by the Mirkin research group, exploits the unique optical properties of colloidal gold nanoparticles 10^15 nm in diameter [182,183]. This approach relies on two di¡erent Au nanoparticles of identical size, but modi¢ed with di¡erent DNA probes. On one Au nanoparticle, 30 alkylthiol-modi¢ed ssDNA probes are chemisorbed. On the other, similarly derivatized 50 probes are attached. When a target that is complementary to both probe sequences is introduced to a solution containing both nanoparticle probes, a cross-linked gold particle/DNA aggregate is formed (see Fig. 12). The formation of the aggregate results in a color change from red to purple that is due to red shift in the

582

Tarlov and Steel

FIG. 12 Schematic representation of an aggregate of Au nanoparticles linked by DNA duplexes. The DNA and nanoparticles are not drawn to scale. (Reprinted with permission from Ref. 183. Copyright 1997 American Chemical Society.)

surface plasmon resonance of the Au nanoparticles. The hybridized gold nanoparticle/DNA aggregates also exhibit unusually sharp melting transitions that are thought to arise from the nature of the multiple DNA duplex cross-links holding the assemblies together.The sharp melting pro¢les allow mutations such as mismatches or deletions to be di¡erentiated from perfect complements. The color changes exhibited at a given temperature during a melting pro¢le can be captured by spotting the solutions on thin-layer chromatography plates. Detection limits of 10 nM have been reported. A related sandwich-type assay also using probe-labeled Au nanoparticles together with conventionally DNA probe-derivatized glass surfaces was recently reported [184]. An ampli¢cation step following the sandwich assay by electrode less deposition of silver on Au nanoparticle surfaces results in a signal enhancement of 105. With ampli¢cation a conventional £at-bed scanner could detect 27-base targets at concentrations as low as 50 fM. Another ampli¢cation-aided visual detection method recently developed is based on changes in the interference patterns of re£ected light from a substrate with appropriately designed thin-¢lm optical coatings [185,186]. A change in the optical thickness, such as that caused by the binding of target

DNA-Based Sensors

583

molecules to the surface, results in attenuation of certain wavelengths from destructive interference and a visible color change. The refractive index and thickness of the ¢lms are carefully adjusted to maximize color changes. One such optical system consisted of a single-crystal silicon wafer coated with a 47.5-nm silicon nitride thin ¢lm and a spin-coated 13.5-nm polymer ¢lm of aminoalkyl-derivatized polydimethyl siloxane to which DNA probes were covalently attached with a bifunctional cross-linker. Following the binding of targets with surface-con¢ned probes, a biotin-labeled detection probe was hybridized to the target. An anti-biotin antibody conjugated to horseradish peroxidase (HRP) was introduced that bound to the surface-immobilized complex. HRP enzymatically catalyzed the precipitation of an insoluble product on the surface, enhancing the change in optical thickness. With enzymatic ampli¢cation using a CCD camera to distinguish color changes, a detection limit of 10 fM was reported for 38-mer targets. 6.3

Mass Sensors

DNA sensors based on measurement of changes in mass have most frequently used the quartz crystal microbalance (QCM) for transduction. QCMs are piezoelectric acoustic wave devices capable of monitoring changes in surface mass by measuring shifts in resonance frequency. Detailed descriptions of the principles and applications of quartz crystal microbalances are found elsewhere [187,188]. Advantages of QCMs for DNAbased sensing include adequate sensitivity for detecting surface hybridization and the ability to operate in aqueous solution, allowing real-time monitoring of binding events. QCM devices possess some drawbacks, though. For example, the interpretation of frequency changes of QCMs operated in solution is complicated by the di⁄culty in deconvoluting mass, density, and viscosity e¡ects [11, 189]. Also, miniaturization of these devices for highthroughput, microarray-type assays presents challenges. The application of quartz crystal microbalances to in situ DNA/RNA hybridization studies was recently reviewed, and we will concentrate primarily on some of the latest developments [11,190,191]. Earlier QCM studies [58,66,101,192^196] have established that: (1) the hybridization of DNA and RNA can be monitored in real time, (2) detection of single-bp mismatches with PNA probes is feasible [193], and (3) the detection limit for solution DNA targets (  25-mers) is in the range of 5^10 nM [197]. Recent work has examined various strategies for increasing sensitivity of QCM measurements. One approach ¢nding modest success has been to increase the e¡ective surface area of devices through the assembly of biotinylated DNA/avidin/polyelectrolyte multilayers [194] or by immobilization of gold colloids attached to gold-coated QCMs via dithiol SAMs

584

Tarlov and Steel

[198]. A more promising strategy for increasing the sensitivity of QCM measurements is by mass amplication of hybridization events.One of the ¢rst reports of QCM ampli¢cation used the binding of anti-dsDNA antibody and goat anti-mouse Fc antibody to enhance mass gain associated with hybridization events [199]. Willner’s group has extended this ampli¢cation approach using tagged liposomes or denditric-like liposome ^ biotin/avidin networks as shown schematically in Fig.13. Probes were immobilized by the adsorption of thiolmodi¢ed oligonucleotides on Au QCM surfaces, with optimal coverages of

FIG. 13 (A) The amplified sensing of a target DNA with oligonucleotidefunctionalized liposomes. (B) Sensing of a target DNA with a biotinylated oligonucleotide, avidin and liposomes labeled with biotin as an amplification conjugate. (Reprinted with permission from Ref. 147. Copyright 2001 American Chemical Society.)

DNA-Based Sensors

585

3.6  1013 probes/cm2 reported [147,197]. Upon capture of target DNA by surface-bound probes, overhanging bases of the target can undergo hybridization with DNA-tagged liposomes (scheme A). Even greater ampli¢cation is achieved by hybridization of the surface duplex with a biotinylated-DNA probe which serves as a point of attachment for the liposome ^ biotin/avidin dendritic networks (scheme B). The negative charge of the liposomes inhibits their nonspeci¢c adsorption to the negatively charged DNA probe surface. The changes in the interfacial properties of the network were sensed by measuring the change in mass with a QCM. Due to the dimensions of the liposomes and the complex interfacial viscosity properties of the bound structures, the QCM data were used in a qualitative fashion to measure liposome association. Nonetheless, QCM detection limits reported for schemes A and B were 1 and 0.1 pM, respectively, for 27-mer targets (see Fig. 14). Similar detection limits were obtained by measuring the electron transfer properties of a redox probe in solution with impedance spectroscopy. The biotin ^ avidin ^ liposome assemblies were also applied similarly for the detection and ampli¢cation of single-base mismatches. Here, a probe is immobilized on the electrode that is complementary to the target up to the base before the single point mutation, for example, G. A biotinylated C base can then be coupled to the duplex with polymerase whereby the coupled C nucleotide serves as an attachment point for the biotin ^avidin/coupled liposomes. Both QCM and impedance measurements were able to discriminate mutant targets present at 5 1013 M in the presence of perfectly matched targets at 1 nm. 6.4

Scanning Probe and Microcantilever Sensors

Scanning probe microscopies have been used for imaging DNA on surfaces for well over a decade [200]. The transduction of DNA hybridization with scanning probe measurements or microfabricated cantilevers, however, is still relatively young. Essentially three approaches have thus far been devised for detecting hybridization: (1) AFM imaging of probes with topographic labels, (2) changes in AFM tip ^ surface interaction forces resulting from hybridization, and (3) nanomechanical transduction involving minute de£ections of microcantilever beams modi¢ed with DNA probes. A relatively new development, DNA sensing using nanomechanical transduction with microfabricated cantilevers (see Fig. 15I), was ¢rst described by Fritz et al. [201].The strategy requires no labeling of targets and holds potential for high-throughput applications via semiconductor processing methods (see Fig 15II). One side of an array of microcantilevers was coated with thin-¢lm gold and then each cantilever was modi¢ed by adsorption of di¡erent sequence thiol-modi¢ed probes. The bending of the

586

Tarlov and Steel

FIG. 14 (A) Time-dependent frequency changes of the 1-functionalized Au-quartz crystal upon (a) interaction with 2, 5  106 M, (b) after interaction of the resulting electrode with the 3-functionalized liposomes (lipid concentration 0.2 mM), (c) treatment of the 1-functionalized Au-quartz crystal with 2, 5  109 M, (d) subsequent treatment of the resulting electrode with the 3-tagged liposomes (lipid concentration 0.2 mM), (e) treatment of the 1-functionalized-Au/ quartz crystal with 2a, 5  106 M, and (f) treatment of the hybridized interface with (3)-labeled liposomes. (B) Quartz-crystal frequency changes as a result of (closed circles) interaction of the 1-modified electrode with different concentrations of the complementary target DNA (2) and (open triangles) as a result of the association of the 3-functionalized liposomes on the ds assembly resulting from the interaction of the sensing interface with different concentrations of 2. (Reprinted with permission from Ref. 147. Copyright 2001 American Chemical Society.)

DNA-Based Sensors

587

FIG. 15 (I) Scanning electron micrograph of a section of a microfabricated silicon cantilever array (eight cantilevers, each 1 mm thick, 500 mm long, and 100 mm wide, with a pitch of 250 mm, spring constant 0.02 N/m; Micro- and Nanomechanics

588

Tarlov and Steel

cantilevers was monitored in situ using a beam de£ection technique. By measuring the de£ection of two microcantilevers, physically identical except for di¡erent immobilized probe sequences, e¡ects such as nonspeci¢c adsorption and thermal drift can be subtracted out (see Fig. 16). Exposure to target DNA results in a di¡erential signal from greater bending of the microcantilever coated with the complementary probe. The molecular origin of the nanomechanical motion was attributed to alteration of surface stress resulting from changes in surface electrostatic, steric, and hydrophobic interactions. A detection limit of 10 nM was reported for 12-mer targets. In addition, the detection of single-base mismatches was demonstrated using two cantilevers di¡ering in only one base of their 12-mer probes. Another study explored in greater detail the mechanism(s) responsible for cantilever bending [202]. Upon hybridization, upward de£ection of cantilevers was observed in all cases, indicating that hybridization relieves compressive stress created during immobilization of thiolated probes. The stress relief is thought to be due to the reduction in con¢gurational entropy of dsDNA versus ssDNA. Duplex DNA assumes a rodlike con¢guration, while ssDNA assumes a coil-like shape owing to its much lower persistence length [203]. The relatively large footprint of the coil-like ssDNA leads to a more convex cantilever surface, allowing each chain to occupy a larger area. Conversely, the smaller footprint of rodlike dsDNA results in less cantilever bending, i.e., upward de£ection. A more recent study comparing microcantilever de£ections for various mismatched and complementary targets bolsters this mechanistic interpretation [204]. It was found that the magnitude and direction of cantilever de£ection could be used to discern the number and location of mismatches in 10-mer targets. Force measurements for DNA detection can be traced to seminal experiments by Lee and colleagues in which AFM was used to pull single DNA molecules in order to measure the forces required to disrupt the double helix [205]. By covalently attaching ssDNA to both the AFM tip and surface, adhesive forces between complementary strands were measured. More

3 Group, IBM Zurich Research Laboratory, Switzerland). (II) Scheme illustrating the hybridization experiment. Each cantilever is functionalized on one side with a different oligonucleotide base sequence. (A) The differential signal is set to zero. (B) After injection of the first complementary oligonucleotide, hybridization occurs on the cantilever (left) that provides the matching sequence, increasing the differential signal Dx. (C) Injection of the second complementary oligonucleotide causes the cantilever (right) functionalized with the second oligonucleotide to bend. (Reprinted with permission from Ref. 201. Copyright 2000 American Association for the Advancement of Science.)

DNA-Based Sensors

589

FIG. 16 Hybridization experiment using two cantilevers functionalized with the sequences 50 -TGCACTAGACCT-30 (12-mer oligonucleotide), and 50 -TAGCCGATATGCGCAT-30 (16-mer oligonucleotide). After taking a baseline (interval I), the complementary 16-mer oligonucleotide (1 mL, 400 nM in HB) was injected (interval II). The liquid cell was purged 20 min later with 3 mL of HB. Then, the complementary 12-mer oligonucleotide (1 mL, 400 nM in HB) was injected (interval III). The liquid cell was again purged 20 min later with 3 mL of HB. (A) Absolute deflection versus time of two individual cantilevers covered with the 16mer (gray) and the 12-mer (black) oligonucleotide. (B) Corresponding differential signal. (Reprinted with permission from Ref. 201. Copyright 2000 American Association for the Advancement of Science.)

590

Tarlov and Steel

recently, AFM has been used to monitor the adsorption and subsequent hybridization of ssDNA attached to positively charged ¢lms of Al(III) bound to mercaptopropanoic acid SAMs on Au [206,207]. Force measurements were performed in situ with a negatively charged silica tip. Analysis of changes in AFM force ^ distance curves yielded information concerning duplex formation. Hybridization of adsorbed 10-mer probes with 10-mer targets resulted in an increase in repulsive force, suggesting an increase in surface charge and potential. Single-base mismatches in 10-mers were also distinguishable from analysis of force ^ distance curves. Another approach related to AFM force measurements worth noting is the use of magnetic microbeads to detect DNA targets. Here, microfabricated magnetoresistive transducers on the substrate indicate the presence of targets whether or not beads are removed when pulled by magnetic forces. In addition, the method can also distinguish between background arising from nonspeci¢cally adsorbed species and signal from targets using magnetic force measurements [208,209]. An example of the use of topographic labels is work by Woolley et al., in which AFM with carbon nanotube probes was used for the multiplexed detection of SNPs [210]. DNA probes were designed such that under the proper hybridization conditions, binding to a target strand did not occur in the presence of single-base mismatches. The probes were labeled with streptavidin or a £uorophore that can be distinguished from one another on the basis of size using single-wall carbon nanotube tips. Woolley et al., demonstrated the AFM imaging of labeled sites with 10-kb DNA fragments that were PCR-ampli¢ed and immobilized on mica surfaces. Another example of AFM detection with topographic labels is by Moller et al., using 30-nm colloidal gold particles [211]. DNA probes were coupled via organosilane monolayers to silicon wafers and, following a sandwich-type assay, AFM images revealed individual gold nanoparticles on the surface. 7

DNA-BASED SENSORS FOR ANALYTES OTHER THAN NUCLEIC ACIDS

Some of the most exciting prospective applications for DNA-based sensors involve the detection of analytes other than nucleic acids. In the majority of these sensors, immobilized DNA or RNA serves as a selective molecular recognition ligand for analytes other than DNA. These devices are being used to explore protein ^ DNA interactions, drug ^ DNA interactions, and DNA damage, as well as to detect a broad range of species including metal ions, and small organic and biological molecules. A diverse breed of sensors, they may eventually serve as platforms for environmental monitoring, drug discovery, controlled drug release, and understanding fundamental

DNA-Based Sensors

591

processes in molecular biology. We describe below some of the more recent work involving DNA-based sensors that is directed toward these promising opportunities. 7.1

Protein–DNA Binding

The sequence-speci¢c binding of DNA by proteins controls gene expression, replication, restriction, and recombination; however, our understanding of the mechanisms of sequence-speci¢c recognition by proteins is still limited. DNA-based sensors and diagnostics of many types are being developed as research tools to elucidate the nature of protein ^ DNA interactions. Of particular interest is the in£uence of sequence mutations on protein ^ DNA interactions and identi¢cation of the amino acid residues associated with binding. Moreover, dsDNA microarray formats are being developed to allow the e⁄cient study of sequence variation in a highly parallel fashion [212]. Advancements in this area are expected to assist in drug discovery and the design of new therapies for ¢ghting disease. An example of an SPR imaging array-based platform for the study of protein ^ DNA interactions has been reported by Corn and co-workers. A novel feature of the array fabrication procedure was the use of a reversible amine-protecting group to control surface hydrophobicity and to attach a protein-resistant poly(ethyleneglycol) (PEG) group to gold surfaces modi¢ed with alkanethiol SAMs [213]. Control of surface wetting allowed the spatial patterning of DNA probes on the surface,while the modi¢cation with PEG limited protein nonspeci¢c adsorption that could give rise to background interference. To demonstrate the utility of this protocol, the speci¢c interaction of single-stranded DNA-binding protein (SSB) with a checkerboard array of ssDNA and dsDNA was monitored using SPR imaging. Signi¢cant binding of SSB was observed only at points of the array where ssDNAwas located,while little binding was observed to array locations with dsDNA. The use of a di¡erent protecting group together with SPR imaging to measure the adsorption of a mismatch binding protein has also been reported [168]. AFM imaging has also been used to monitor protein binding with dsDNA in work by Porter and colleagues [61,112,113]. In these studies, disul¢de-modi¢ed dsDNA was attached to gold surfaces following the photopatterning of a £uorinated alkanethiol SAM. The immobilized dsDNA contained recognition sequences for either ECoR1 or HaeIII, both restriction enzymes for dsDNA. By measuring the height di¡erence between the patterned areas containing dsDNA and the £uorinated SAM, the successful enzymatic cleavage of dsDNA was demonstrated (see Fig. 17) [112,113]. Before enzyme treatment, regular topographical features 8.8 nm in height

592

Tarlov and Steel

FIG. 17 In-situ AFM topographic images (80  80 mm) of a dsDNA microarray containing the recognition sequence specific for EcoR1 (Fig. 1A). The images were collected before (A) and after (B) digestion with EcoR1. The cross-sectional contours below each image reflect the average of the individual scan lines contained in the area of a single row of the array. The in-situ images were obtained in 10 mM THAM (pH 7.4) at a scan rate of 1 Hz. (Reprinted with permission from Ref. 113. Copyright 2000 American Chemical Society.)

were measured that were consistent with the expected length of the immobilized 26-mer dsDNA. Following enzyme digestion the features decreased to approximately 4.3 nm, a value close to that expected for the point of cleavage along the dsDNA. Enzymatic cleavage was also con¢rmed by monitoring the £uorescence decrease associated with the loss of a £uorescein label that was attached to the cleaved dsDNA fragments. Reportedly, control of the duplex packing density and orientation was important to ensure adequate access for enzymatic cleavage. Several studies of realtime monitoring of protein interactions with immobilized nucleic acids using the QCM should also be noted [196,214^217].

DNA-Based Sensors

7.2

593

Drug–DNA Interactions

DNA-based sensors have also been used as research tools to study the interaction between anticancer drugs and nucleic acids. One of the ¢rst examples is from Thompson’s group, who studied the binding of cis- and trans-platin with calf thymus dsDNA immobilized on palladium electrodes of quartz crystal acoustic wave devices [218].The binding of both drugs was evidenced by a decrease in series resonant frequency as a function of exposure time, with the shapes of the transients indicating two distinct kinetic processes. The kinetic behavior was attributed to the interaction of the hydrolysis products of cis-platin and trans-platin with the bound DNA. Recently, Burgess and co-workers examined the interaction of a hydrolyzed form of cis-platin using gold QCM electrodes modi¢ed with ssDNA. Their results indicate preassociation of the complex with DNA, presumably through electrostatic interactions, prior to covalent binding [219]. The binding of another anticancer-related compound, doxorubicin, to thiol-modi¢ed dsDNA (718 bp) was also examined using the QCM [220]. To enhance the sensitivity of kinetic measurements, the drug was conjugated to a soluble dextran polymer. Electrochemical methods have also been used to study the binding of several anticancer-related drugs with DNA [45,146,221^223].Wang and coworkers examined the interaction of daunomycin (DM) with calf thymus dsDNA adsorbed on carbon-paste electrodes using cyclic voltammetry and constant-current chronopotentiometric stripping analysis. Changes in the oxidation current associated with guanine residues were interpreted to result from alteration of the DNA structure, possibly bending, caused by the intercalation of DM with DNA. The interaction of DM with immobilized DNA was also thought to di¡er signi¢cantly from that of DM with solutionphase DNA [221]. As discussed above in Sec. 6.1, the groups of Barton and Hill have also examined the electrochemical behavior of DM intercalated into thiol-tethered dsDNA on gold surfaces [146,222]. These groups report that the electrochemical response of DM as measured by cyclic voltammetry is dramatically decreased by the presence of single-base mismatches. Related to studies of drug/DNA interactions are several reports in which DNA itself serves as a therapeutic agent. In these studies, platforms closely related to DNA-based electrochemical sensors have been developed that allow the controlled release of DNA for applications such as gene therapy [224]. One such protocol is based on the reductive desorption of thiolate-modi¢ed DNA monolayers from gold electrodes. In these studies the controlled release of 350-bp dsDNA and 25-mer ssDNA from gold microelectrodes was investigated as a function of applied electrode potential and time. The electrochemical desorption is thought to occur by the

594

Tarlov and Steel

reductive cleavage of the sulfur ^ gold bond, as has been observed for alkanethiol SAMs [225,226]. The extent of desorption of DNA from the gold electrodes was assessed using cyclic voltammetry, XPS, and electrochemical QCM measurements. Poising the electrode potential at 1.3V (versus Ag/ AgCl) in pH 7.4 phosphate bu¡er for 25 min was found to be su⁄cient for complete removal of the thiolated DNA monolayers. Several protocols for releasing DNA and DNA ^ lipid complexes from gold and carbon microelectrodes under potential control have also been developed based on electrostatic interactions [227,228]. 7.3

Environmental Sensing

Wang and co-workers have developed three strategies for electrochemical detection of toxic pollutants that rely on analyte ^ DNA interactions [229]. In the ¢rst approach, dsDNA is used as an a⁄nity layer to concentrate electroactive compounds at electrode surfaces for direct electrochemical detection. The detection of electroactive aromatic amines in groundwater at carbonpaste electrodes with nanomolar sensitivity was demonstrated with this method [230]. In the second approach, the interaction of certain analytes with dsDNA can cause chemical, structural, or conformational changes in the surface-con¢ned DNA that can be detected by variations in the oxidative signal associated with guanine. For example, the interaction of hydrazine compounds with dsDNA-modi¢ed electrodes was manifested by a decrease in the oxidation current of guanine residues as measured by chronopotentiometry [231]. In the third strategy, changes in the guanine electroactivity can also be used to indicate DNA damage caused by radiation or chemical interactions [121,232^234]. This approach builds on the earlier work of Palecek and colleagues [117,136,235^238], which has since been extended [239,240]. Square-wave voltammetry of dsDNA in cast polymer ¢lms or dsDNA attached directly to carbon electrodes has also been used to detect damage caused by reaction with styrene oxide [241]. 7.4

Catalytic DNA and RNA

One of the most exciting developments to emerge recently in DNA sensing is the use of catalytically active RNA (ribozymes) and DNA (deoxyribozymes). In this approach, DNA and RNA have been engineered to behave as allosteric enzymes that act as ligand-speci¢c ‘‘molecular switches’’ to signal the presence of analyte molecules [242^244]. Allosteric enzymes contain an e¡ector-binding (allosteric) site and a separate active, catalytic site. The binding of a ligand, the analyte, to the allosteric site induces a conformational change that either enhances or inhibits the catalytic site. Using a combinatorial-like process called in-vitro selection [245] for screening random,

DNA-Based Sensors

595

candidate sequences of DNA and RNA, deoxyribozymes, and ribozymes have been discovered that respond speci¢cally to small organic molecules and metal ions. One example of such work is the recent report by Li and Lu of a biosensing strategy for lead ions that uses a deoxyribozyme derived from in-vitro selection techniques [246]. Illustrated schematically in Fig. 18, the deoxyribozyme (17E) is capable of cleaving a single RNA linkage contained in a complementary DNA substrate (17DS). The enzymatic rate of cleavage is highly dependent on the solution activity of Pb2þ , the e¡ector that triggers the allosteric deoxyribozyme. To signal the cleavage event, a molecular beacon-like approach was used whereby the 50 -end of the substrate, 17DS, was labeled with a £uorophore and the 30 -end of the deoxyribozyme, 17E, was labeled with a quencher. When the substrate is hybridized with deoxyribozyme, £uorescence is quenched. Addition of Pb2þ to the solution switches on the catalytic cleavage reaction. The resulting product strand, end-labeled with the £uorophore, £uoresces when released. The allosteric deoxyribozyme is highly selective; the £uorescence response rate for Pb2þ is >80 times higher than that of eight other divalent metal ions examined (see Fig. 19). The detection limit of Pb2þ was estimated to be approximately 10 nM.While only solution-phase operation of the sensor was demonstrated, Li and Lu point out that attachment of DNA enzymes to optical ¢bers or other surfaces would be relatively straightforward. In addition, they suggest that sensitivity and selectivity could be further enhanced by careful selection of the £uorophore and tuning of the metal binding domain with additional cycles of in-vitro selection. Another powerful demonstration of biosensing using molecular switches of nucleic acids was recently reported by Breaker and co-workers [247]. In this work, a prototype biosensing array was constructed from allosteric ribozymes that were engineered by in-vitro selection. The RNA switches were based on a class of hammerhead ribozyme that normally undergoes self-cleavage in the presence of Mg2þ . By engineering the allosteric domains, seven di¡erent hammerhead ribozymes were generated, each activated by a speci¢c e¡ector, i.e., an analyte, including Co2þ , cyclic adenosine monophosphate (cAMP), and cyclic guanosine monophosphate (cGMP). Further highlighting the versatility and selectivity of allosteric engineering, the triggering of one RNA switch required the simultaneous presence of two e¡ectors, theophylline and £avin mononucleotide. The cleavage fragments were radiolabeled with 32P so that the cleavage reactions could be monitored as a function of time. The cyclic nucleotide monophosphate-dependent switches were able to detect analyte concentrations in the range of 1 to 1000 mM. The ribozyme switch speci¢c for cobalt exhibited a detection limit near 1 mM, with a dynamic range of greater than 10 4. To

596

Tarlov and Steel

FIG. 18 (a) Sequence and proposed secondary structure of the deoxyribozyme/ substrate complex. The cleavable substrate (Rh-17DS) is a DNA/RNA chimera in which rA represents a ribonucleotide adenosine. (b) Steady-state fluorescence spectra of the substrate (Rh-17DS) alone (I), after annealing to the deoxyribozyme (17E-Dy) (II), and 15 min after adding 500 nM Pb(OAc)2 (III). (Reprinted with permission from Ref. 246. Copyright 2000 American Chemical Society.)

demonstrate array-based sensing, the seven RNA switches were immobilized on gold thin ¢lms via a 50 -thiophosphate functionality. By monitoring levels of cAMP secreted into culture media, the array was able to determine phenotype of mutant Escherichia coli strains. The number of analytes that can be sensed with catalytic DNA and RNA is still quite limited; nonetheless, some researchers suggest that this

DNA-Based Sensors

597

FIG. 19 Fluorescence response rate (vfluo) of Rh-17EDS-Dy in the presence of 500 nM of different divalent metal ions in 50 mM HEPES (pH 7.5). The inset shows the change of fluorescence intensity at 580 nm in response to the addition of M2þ . The curve with dramatic change was collected in Pb2þ ; the other curves were collected in one of the other eight divalent metal ions. (Reprinted with permission from Ref. 246. Copyright 2000 American Chemical Society.)

relatively young technology may some day supplant antibodies in array formats for proteomics and metabolomics. Further bolstering this contention is the recent identi¢cation of protein-dependent ribozymes [248]. As pointed out by Ellington and colleagues [249], ribozymes possess several advantages over antibodies for microarray applications. First, allosteric ribozymes can be rapidly screened and identi¢ed using in-vitro selection methods. Second, the engineering of ribozymes is simpler than that of antibodies because the secondary structure of ribozymes is generally far less complex than the tertiary structure of antibodies. Third, as a result of their relative structural simplicity, ribozymes are generally more robust than antibodies, and thus are more likely to retain activity when attached to surfaces.

8

CHALLENGES AND OUTLOOK

Signi¢cant progress has been made in developing DNA-based sensors aimed at a variety of applications involving the detection of nucleic acids, proteins, or other small molecules. The past decade has witnessed steady improvements in the sensitivity and selectivity of DNA-based sensors along with the development of innovative transduction schemes for sensor readout.

598

Tarlov and Steel

Nevertheless, in order for DNA-based sensors to transition from novel laboratory devices to real-world measurement and diagnostics tools, it is clear that advances in several aspects of sensor performance are essential. We anticipate future research activities on DNA-based sensors to be directed toward overcoming some of the following challenges and technical barriers. Improvements in sensitivity. Improving the sensitivity of DNA-based sensors will enable many new applications.The‘‘holy grail’’ in DNAbased sensing is single-copy detection. Further improvements in sensitivity may allow the circumvention of DNA ampli¢cation. PCR, molecular biology’s gold standard of ampli¢cation, is often an expensive and time-consuming bottleneck in high-throughput genetic screening [4]. Pushing detection limits lower will likely require the development of new transduction strategies that involve ampli¢cation. Indeed, as discussed in this review, several groups have developed transduction schemes that exploit ampli¢cation for detection. Nonetheless, the search will undoubtedly continue for new, more sensitive, and discriminating transduction methods that include ampli¢cation. Increased miniaturization. Increased miniaturization of DNA-based sensors, together with advances in sensor sensitivity, will permit greater assay throughput and lower sample volumes. An even more exciting prospect of miniaturization is the potential for in-situ realtime monitoring of gene transcripts in living organisms. The key to realizing this goal may lie in the ability to manipulate DNA with near-nanometer precision and assemble nanoscale DNA hybrid structures. Toward this end, several groups have recently reported the construction of geometric DNA assemblies [250,251], submicrometer patterning of DNA on solid surfaces, the directed assembly of nanometer-scale metal semiconductor, and organic colloidal particles decorated with DNA probes [252^261]. Work in these nanotechnology-related areas is increasing and may have farreaching implications for areas such as molecular-based electronic devices, as well as DNA-based sensors. Integration with microanalytical systems. Successfully mating DNAbased sensors with MEMS- and micro£uidic-based systems will enable complete, integrated microanalysis systems. Progress is being made in developing micro£uidic devices that serve as a front end to perform sample preparation and delivery functions, such as cell lysis, DNA extraction, transcription, and ampli¢cation [25,26]. A potential role for miniaturized DNA sensors in microchannels is that of a multiplexing detector. Another bene¢t of micro£uidic

DNA-Based Sensors

599

delivery of sample to DNA sensing elements is the enhancement of mass transfer, ensuring that the sensor ‘‘sees’’all of the analyte. Acceptance of DNA-based sensors. Ultimately, the acceptance of a particular DNA-based sensor will hinge on its a¡ordability, simplicity, and, most important, the ability to deliver reliable, accurate results. These criteria will ¢gure most prominently in applications involving human health.Genetic diagnostic devices for human clinical use will have to be FDA-approved, and thus their performance will be rigorously scrutinized and validated. Therefore, it is likely that DNAbased sensors will ¢rst ¢nd application as research tools to answer biological questions, followed by commercial applications in areas such as the monitoring of genetically modi¢ed organisms in agriculture or detection of pathogens in food products. Once DNAbased sensors establish a track record in these areas, applications to human health problems will likely follow. As in any ¢eld, the continual re¢nement of measurement tools and the development of new ones often create new areas of research, which, in turn, enable radically new technologies. It is not di⁄cult to envision such scenarios with DNA-based sensors. For example, as the sensitivity of DNAbased sensors increases and their size decreases, quantitative, real-time monitoring of gene expression in tissues or even single cells may be realized. Such an advance would pave the way to establishing a molecular understanding of biological processes and human disease. We are con¢dent that DNA-based sensors will play a prominent role in advancing biomedical research. ACKNOWLEDGMENTS MJT is grateful to the NIST Advanced Technology Program, and particularly to Dr. Stanley Abramowitz, for their support of this and past work. He also thanks Dawn Hurley for help in collating references. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8.

ES Lander. Nature 409:860^921, 2001. JC Venter. Science 291:1304^1351, 2001. A Chakravarti. Nature 409:822^823, 2001. M Chicurel. Nature 412:580^582, 2001. LM Staudt. Trends Immunol 22:35^40, 2001. MJ Cunningham. J Pharmacol Toxicol Meth 44:291^300, 2000. C Debouck, PN Goodfellow. Nature Genet 21:48^50, 1999. K Olden, J Guthrie. Mutat Res ^ Fundam Mol Mech Mutagen 473:3^10, 2001.

600

Tarlov and Steel

9. 10. 11. 12. 13. 14. 15.

P Reymond. Plant Physiol Biochem 39:313^321, 2001. SR Mikkelsen. Electroanalysis 8:15^19, 1996. MS Yang, ME McGovern, M Thompson. Anal Chim Acta 346:259^275, 1997. L Henke, UJ Krull. Can J Anal Sci Spectrosc 44:61^70, 1999. J Wang. Nucleic Acids Res 28:3011^3016, 2000. SL Beaucage. Curr Med Chem 8:1213^1244, 2001. DJ Duggan, M Bittner,YD Chen, P Meltzer, JM Trent. Nature Genet 21:10^14, 1999. DDL Bowtell. Nat Genet 21:25^32, 1999. CM Perou,T Sorlie, MB Eisen, M van de Rijn, SS Je¡rey, CA Rees, JR Pollack, DT Ross, H Johnsen, LA Aksien, O Fluge, A Pergamenschikov, C Williams, SX Zhu, PE Lonning, AL Borresen-Dale, PO Brown, D Botstein. Nature 406:747^752, 2000. L Liotta, E Petricoin. Nat Rev Genet 1:48^56, 2000. Microarray Biochip Technology. Natick, MA: Eaton Publishing, Biotechniques Books Division, 2000. MC Pirrung. Chem Rev 97:473^488, 1997. M Schena, RA Hellar, TP Theriault, K Konrad, E Lachenmeier, RW Davis. Trends Biotechnol 16:301^306, 1998. CM Niemeyer, D Blohm, Angew Chem Int Edit 38:2865^2869, 1999. S Abramowitz. J Biomed Microdevices 1:107^112, 1999. CC Xiang,YD Chen. Biotechnol Adv 18:35^46, 2000. AT Woolley, D Hadley, P Landre, AJ deMello, RA Mathies, MA Northrup. Anal Chem 68:4081^4086, 1996. P Belgrader, S Young, B Yuan, M Primeau, LA Christel, F Pourahmadi, MA Northrup. Anal Chem 73:286^289, 2001. YN Shi, PC Simpson, JR Scherer, D Wexler, C Skibola, MT Smith, RA Mathies. Anal Chem 71:5354^5361, 1999. BB Haab, RA Mathies. Anal Chem 71:5137^5145, 1999. MB Esch, LE Locascio, MJ Tarlov, RA Durst. Anal Chem 73:2952^2958, 2001. SC Jakeway, AJ de Mello, EL Russell. Fresenius J Anal chem 366:525^539, 2000. DD Cunningham. Anal Chim Acta 429:1^18, 2001. RC McGlennen. Clin Chem 47:393^402, 2001. T Vo-Dinh, B Cullum. Fresenius J Anal Chem 366:540^551, 2000. T Vo-Dinh, BM Cullum, DL Stokes. Sensors Actuator B Chem 74:2^11, 2001. DJ Harrison, A van den Berg. Micro Total Analysis Systems ’98. Dordrecht, The Netherlands: Kluwer, 1998. A van den Berg, W Olthuis, P Bergveld. Micro Total Analysis Systems 2000. Dordrecht,The Netherlands: Kluwer, 2000. W Saenger. Principles of Nucleic Acid Structure. New York: Springer-Verlag, 1984. JG Wetmur. Annu Rev Biophys Bioeng 5:337^361, 1976. GM Wahl, SL Berger, AR Kimmel. Meth Enzymol 152:399^407, 1987. JG Wetmur. Crit Rev Biochem Mol Biol 26:227^259, 1991.

16. 17.

18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40.

DNA-Based Sensors

601

41. E Southern, K Mir, M Shchepinov. Nat Genet 21:5^9, 1999. 42. A Ellington, JD Pollard. In: Current Protocols in Molecular Biology. New York: Wiley, 1998, pp 2.11.11^12.11.25. 43. AC Pease, D Solas, EJ Sullivan, MT Cronin, CP Holmes, SPA Fodor. Proc Natl Acad Sci USA 91:5022^5026, 1994. 44. E Southern, K Mir, M Shchepinov. Nat Genet Suppl 21:5^9, 1999. 45. M Eggers, M Hogan, RK Reich, J Lamture, D Ehrlich, M Hollis, B Kosicki,T Powdrill, K Beattie, S Smith, R Varma, R Gangadharan, A Mallik, B Burke, D Wallace. Biotechniques 17:516^524, 1994. 46. JB Lamture, KL Beattie, BE Burke, MD Eggers, DJ Ehrlich, R Fowler, MA Hollis, BB Kosicki, RK Reich, SR Smith, RS Varma, ME Hogan. Nucleic Acids Res 22:2121^2125, 1994. 47. Z Guo, RA Guilfoyle, AJ Thiel, RF Wang, LM Smith. Nucleic Acids Res 22:5456^5465, 1994. 48. LA Chrisey, GU Lee, CE Oferrall. Nucleic Acids Res 24:3031^3039, 1996. 49. JH Watterson, PAE Piunno, CC Wust, UJ Krull. Sensors Actuators B Chem 74:27^36, 2001. 50. LH Dubois, RG Nuzzo. Annu Rev Phys Chem 43:437^463, 1992. 51. AR Bishop, RG Nuzzo. Curr Opin Colloid Interface Sci 1:127^136, 1996. 52. AJ Thiel, AG Frutos, CE Jordan, RM Corn, LM Smith. Anal Chem 69:4948^ 4956, 1997. 53. EA Smith, MJ Wanat, YF Cheng, SVP Barreira, AG Frutos, RM Corn. Langmuir 17:2502^2507, 2001. 54. BP Nelson, TE Grimsrud, MR Liles, RM Goodman, RM Corn. Anal Chem 73:1^7, 2001. 55. L He, MD Musick, SR Nicewarner, FG Salinas, SJ Benkovic, MJ Natan, CD Keating. J Am Chem Soc 122:9071^9077, 2000. 56. E Huang, FM Zhou, L Deng. Langmuir 16:3272^3280, 2000. 57. C Bamdad. Biophys J 75:1997^2003, 1998. 58. Y Okahata, Y Matsunobu, K Ijiro, M Mukae, A Murakami, KJ Makino. Am Chem Soc 114:8299^8300, 1992. 59. TM Herne, MJ Tarlov. J Am Chem Soc 119:8916^8920, 1997. 60. R Levicky, TM Herne, MJ Tarlov, SK Satija. J Am Chem Soc 120:9787^9792, 1998. 61. JC O’Brien, JT Stickney, MD Porter. J Am Chem Soc 122:5004^5005, 2000. 62. E Huang, M Satjapipat, SB Han, FM Zhou. Langmuir 17:1215^1224, 2001. 63. AB Steel,TM Herne, MJ Tarlov. Anal Chem 70:4670^4677, 1998. 64. MJ Tarlov,TH Huang. Unpublished results. 65. M Wilchek, EA Bayer. Anal Biochem 171:1^32, 1998. 66. RC Ebersole, JA Miller, JR Moran, MD Ward. J Am Chem Soc 112:3239^3241, 1990. 67. E Kai, S Sawata, K Ikebukuro,T Iida,T Honda, I Karube. Anal Chem 71:796^ 800, 1999. 68. XJ Liu,WH Tan. Anal Chem 71:5054^5059, 1999.

602

Tarlov and Steel

69.

AV Vasiliskov, EN Timofeev, SA Surzhikov, AL Drobyshev, VV Shick, AD Mirzabekov. Biotechniques 27:592, 1999. VAVasiliskov, EN Timofeev, SA Surzhikov, AL Drobyshev,VV Shik, AD Mirzabekov. Mol Biol 32:780^782, 1998. FN Rehman, M Audeh, ES Abrams, PW Hammond, M Kenney, TC Boles. Nucleic Acids Res 27:649^655, 1999. MJ Heller, AH Forster, E Tu. Electrophoresis 21:157^164, 2000. DJ Caruana, A Heller. J Am Chem Soc 121:4728^4728, 1999. T de Lumley-Woodyear, DJ Caruana, CN Campbell, A Heller. Anal Chem 71:394^398, 1999. H KorriYoussou¢, F Garnier, P Srivastava, P Godillot, A Yassar. J Am Chem Soc 119:7388^7389, 1997. J Fidanza, M Glazer, D Mutnick, G McGall, C Frank. Nucleosides and Nucleotides 20:533^538, 2001. M Beier, JD Hoheisel. Nucleic Acids Res 27:1970^1977, 1999. TW Nilsen, J Grayzel,W Prensky. J Theor Biol 187:273^284, 1997. MS Shchepinov, KU Mir, JK Elder, MD Frank-Kamenetskii, EM Southern. Nucleic Acids Res 27:3035^3041, 1999. J Wang, G Rivas, JR Fernandes, M Jiang, JLL Paz, R Waymire, TW Nielsen, RC Getts. Electroanalysis 10:553^556, 1998. T Strother, RJ Hamers, LM Smith. Nucleic Acids Res 28:3535^3541, 2000. T Strother,W Cai, XS Zhao, RJ Hamers, LM Smith. J Am Chem Soc 122:1205^ 1209, 2000. JR Lakowicz, I Gryczynski, Z Gryczynski, K Nowaczyk, CJ Murphy. Anal Biochem 280:128^136, 2000. I Willner, F Patolsky, J Wasserman. Angew Chem Int Edit 40:1861^1864, 2001. S Pathak, SK Choi, N Arnheim, ME Thompson, J Am Chem Soc 123:4103^ 4104, 2001. JE Forman, ID Walton, D Stern, RP Rava, MO Trulson. In: NB Leontis, J SantaLucia, eds. Molecular Modeling of Nucleic Acids. Washington, DC: American Chemical Society, 1998, pp 206^228. AJ Leavitt, LA Wenzler, JM Williams,TP Beebe, J Phys Chem 98:8742^8746, 1994. R Lenigk, M Carles, NY Ip, NJ Sucher. Langmuir 17:2497^2501, 2001. DE Gray, SC CaseGreen, TS Fell, PJ Dobson, EM Southern. Langmuir 13:2833^2842, 1997. LA Chrisey, PM Roberts, VI Benezra, WJ Dressick, CS Dulcey, JM Calvert. Mater Res Soc 330:179^184, 1994. M Boncheva, L Scheibler, P Lincoln, H Vogel, B Akerman, Langmuir 15:4317^ 4320, 1999. D Piscevic, R Lawall, M Veith, M Liley,Y Okahata,W Knoll. Appl Surface Sci 90:425^436, 1995. KA Peterlinz, RM Georgiadis, TM Herne, MJ Tarlov. J Am Chem Soc 119:3401^3402, 1997.

70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86.

87. 88. 89. 90. 91. 92. 93.

DNA-Based Sensors

603

94. AW Peterson, RJ Heaton, R Georgiadis. JAm Chem Soc122:7837^7838, 2000. 95. R Georgiadis, KP Peterlinz, AW Peterson, J Am Chem Soc 122:3166^3173, 2000. 96. TW Jing, AM Je¡rey, JA Derose, YL Lyubchenko, LS Shlyakhtenko, RE Harrington, E Appella, J Larsen, AVaught, D Rekesh, FX Lu, SM Lindsay. Proc Natl Acad Sci USA 90:8934^8938, 1993. 97. D Rekesh, Y Lyubchenko, LS Shlyakhtenko, SM Lindsay. Biophys J 71:1079^ 1086, 1996. 98. M Hegner, M Dreier, P Wagner, G Semenza, HJ Guntherodt. J Vacuum Sci Technol B 14:1418^1421, 1996. 99. JH Watterson, PAE Piunno, CC Wust, UJ Krull. Langmuir 16:4984^4992, 2000. 100. JH Watterson, PAE Piunno, CC Wust, S Raha, UJ Krull. Fresenius J Anal Chem 369:601^608, 2001. 101. HB Su, KMR Kallury, M Thompson, A Roach. Anal Chem 66:769^777, 1994. 102. MS Shchepinov, SC CaseGreen, EM Sourthern. Nucleic Acids Res 25:1155^ 1161, 1997. 103. AB Steel, RL Levicky,TM Herne, MJ Tarlov. Biophys J 79:975^981, 2000. 104. U Maskos, EM Southern. Nucleic Acids Res 20:1675^1678, 1992. 105. U Maskos, EM Southern. Nucleic Acids Res 21:4663^4669, 1993. 106. KU Mir, EM Southern. Nat Biotechnol 18:1209^1209, 2000. 107. GE Poirier. Chem Rev 97:1117^1127, 1997. 108. PAE Piunno, J Watterson, CC Wust, UJ Krull. Anal Chim Acta 400:73^89, 1999. 109. YY Zhang, M Tirrell, JW Mays. Macromolecules 29:7299^7301, 1996. 110. MG Youngquist, RJ Driscoll, TR Coley, WA Goddard, JD Baldeschwieler. J Vacuum Sci Technol B 9:1304^1308, 1991. 111. LA Bottomley, JN Haseltine, DP Allison, RJ Warmack, T Thundat, RA Sachleben, GM Brown, RP Woychik, KB Jacobson, TL Ferrell. J Vacuum Sci Technol A
Vacuum Surface Films 10:591^595, 1992. 112. SO Kelley, JK Barton, NM Jackson, LD McPherson, AB Potter, EM Spain, MJ Allen, MG Hill. Langmuir 14:6781^6784, 1998. 113. JC O’Brien, JT Stickney, MD Porter. Langmuir 16:9559^9567, 2000. 114. V Chan, DJ Graves, SE McKenzie. Biophys J 69:2243^2255, 1995. 115. KY Wong, BM Pettitt. Theor Chem Accts 106:233^235, 2001. 116. E Palecek. Nature 188:656^657, 1960. 117. E Palecek. Anal Biochem 170:421^431, 1988. 118. P Singhal,WG Kuhr. Anal Chem 69:3552^3557, 1997. 119. HH Thorp. Trends Biotechnol 16:117^121, 1998. 120. E Palecek. Anal Biochem 108:129^138, 1980. 121. J Wang, S Bollo, JLL Paz, E Sahlin, B Mukherjee. Anal Chem 71:1910^1913, 1999. 122. BS Strauss. Cancer Res 52:249^253, 1992. 123. HH Thorp. In: AG Sykes, ed. Advances in Inorganic Chemistry. San Diego, CA: Academic, 1995, vol 43, pp 127^177.

604

Tarlov and Steel

124. DH Johnston, KC Glasgow, HH Thorp. J Am Chem Soc 117:8933^8938, 1995. 125. ME Napier, HH Thorp. Langmuir 13:6342^6344, 1997. 126. ME Napier, CR Loomis, MF Sistare, J Kim, AE Eckhardt, HH Thorp. Bioconjugate Chem 8:906^913, 1997. 127. PM Armistead, HH Thorp, Anal. Chem 72:3764^3770, 2000. 128. F Garnier, H Korri-Yousssou¢, P Srivastava, B Mandrand,T Delair. Synth Met 100:89^94, 1999. 129. J Wang, M Jiang. Langmuir 16:2269^2274, 2000. 130. AC Ontko, PM Armistead, SR Kircus, HH Thorp. Inorg Chem 38:1842^1846, 1999. 131. E Palecek, MA Hung. Anal Biochem 132:236^242, 1983. 132. FA Tanious, TC Jenkins, S Neidle, WD Wilson. Biochemistry 31:11632^11640, 1992. 133. KM Millan, SR Mikkelsen. Anal Chem 65:2317^2323, 1993. 134. KM Millan, A Saraullo, SR Mikkelsen. Anal Chem 66:2943^2948, 1994. 135. YD Zhao, DW Pang, ZL Wang, JK Cheng, YP Qi. J Electroanal. Chem 431:203^209, 1997. 136. E Palecek, M Fojta, M Tomschik, J Wang. Biosensors Bioelectron 13:621^628, 1998. 137. J Wang, E Palecek, PE Nielsen, G Rivas, XH Cai, H Shiraishi, N Dontha, DB Luo, PAM Farias. J Am Chem Soc 118:7667^7670, 1996. 138. K Hashimoto, K Ito,Y Ishimori. Anal Chem 66:3830^3833, 1994. 139. K Hashimoto, K Ito,Y Ishimori. Sensors Actuator B Chem 46:220^225, 1998. 140. S Takenaka. Bunseki Kagaku 48:1095^1105, 1999. 141. S Takenaka, K Yamashita, M Takagi, Y Uto, H Kondo. Anal Chem 72:1334^ 1341, 2000. 142. G Marrazza, I Chianella, M Mascini. Anal Chim Acta 387:297^307, 1999. 143. G Marrazza, I Chianella, M Mascini. Biosensors Bioelectron 14:43^51, 1999. 144. M Mascini, I Palchetti, G Marrazza. Fresenius J Anal Chem 369:15^22, 2001. 145. SO Kelley, JK Barton, NM Jackson, MG Hill. Bioconjugate Chem 8:31^37, 1997. 146. SO Kelley, EM Boon, JK Barton, NM Jackson, MG Hill. Nucleic Acids Res 27:4830^4837, 1999. 147. F Patolsky, A Lichtenstein, I Willner. J Am Chem Soc 123:5194^5205, 2001. 148. XH Xu, HC Yang,TE Mallouk, AJ Bard. J Am Chem Soc 116:8386^8387, 1994. 149. XH Xu, AJ Bard. J Am Chem Soc 117:2627^2631, 1995. 150. JJ Kasianowicz, E Brandin, D Branton, DW Deamer. Proc Natl Acad Sci USA 93:13770^13773, 1996. 151. M Akeson, D Branton, JJ Kasianowicz, E Brandin, DW Deamer. Biophys J 77:3227^3233, 1999. 152. SE Henrickson, M Misakian, B Robertson, JJ Kasianowicz. Phys Rev Lett 85:3057^3060, 2000. 153. JJ Kasianowicz, SE Henrickson, HH Weetall, B Robertson. Anal Chem 73:2268^2272, 2001. 154. T Vodinh, K Houck, DL Stokes. Anal Chem 66:3379^3383, 1994.

DNA-Based Sensors 155. 156. 157. 158. 159. 160. 161. 162. 163. 164. 165. 166. 167. 168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180. 181. 182. 183. 184. 185. 186. 187. 188.

605

NR Isola, DL Stokes,T Vo-Dinh. Anal Chem 70:1352^1356, 1998. H Raether. In: Physics of Thin Films. New York: Wiley, 1977, vol 9, p 145. W Knoll. MRS Bull 16:29^39, 1991. KA Peterlinz, R Georgiadis. Langmuir 12:4731^4740, 1996. L Haussling, H Ringsdorf, FJ Schmitt,W Knoll. Langmuir 7:1837^1840, 1991. V Silin, A Plant. Trends Biotechnol 15:353^359, 1997. JM Brockman, BP Nelson, RM Corn. Annu Rev Phys Chem 51:41^63, 2000. B Rothenhausler,W Knoll. Nature 332:615^617, 1988. MA Karymov, AA Kruchinin, YA Tarantov, IA Balova, LA Remisova, YG Vlasov. Sensors Actuators B Chem 29:324^327, 1995. SJ Wood. Microchem J 47:330^337, 1993. CE Jordan, AG Frutos, AJ Thiel, RM Corn. Anal Chem 69:4939^4947, 1997. BP Nelson, AG Frutos, JM Brockman, RM Corn. Anal Chem 71:3928^3934, 1999. AG Frutos, SC Weibel, RM Corn. Anal Chem 71:3935^3940, 1999. AG Frutos, JM Brockman, RM Corn. Langmuir 16:2192^2197, 2000. KA Peterlinz, R Georgiadis. Opt Commun 130:260^266, 1996. T Liebermann, W Knoll, P Sluka, R Herrmann. Colloid Surfaces A
Physicochem Eng Asp 169:337^350, 2000. CR Cantor, CL Smith. Genomics. New York: Wiley, 1999. RJ Heaton, AW Peterson, RM Georgiadis. Proc Natl Acad Sci USA 98:3701^3704, 2001. RG Sosnowski, E Tu, WF Butler, JP Oconnell, MJ Heller. Proc Natl Acad Sci USA 94:1119^1123, 1997. CF Edman, DE Raymond, DJ Wu, EG Tu, RG Sosnowski, WF Butler, M Nerenberg, MJ Heller. Nucleic Acids Res 25:4907^4914, 1997. Y Belosludtsev, I Belosludtsev, B Iverson, S Lemeshko, R Wiese, M Hogan, T Powdrill. Biochem Biophys Res Commun 282:1263^1267, 2001. AVainrub, BM Pettitt. Chem Phys Lett 323:160^166, 2000. XH Fang, XJ Liu, S Schuster,WH Tan. J Am Chem Soc 121:2921^2922, 1999. S Tyagi, FR Kramer. Nat Biotechnol 14:303^308, 1996. XH Fang, JJ Li,WH Tan Anal Chem 72:3280^3285, 2000. FJ Steemers, JA Ferguson, DR Walt. Nat Biotechnol 18:91^94, 2000. VSY Lin, K Motesharei, KPS Dancil, MJ Sailor, MR Ghadiri, Science 278:840^843, 1997. JJ Storho¡, R Elghanian, RC Mucic, CA Mirkin, RL Letsinger. J Am Chem Soc 120:1959^1964, 1998. JJ Storho¡, CA Mirkin. Chem Rev 99:1849^1862, 1999. TA Taton, CA Mirkin, RL Letsinger. Science 289:1757^1760, 2000. RM Ostro¡, D Hopkins, AB Haeberli, W Baouchi, B Polisky. Clin Chem 45:1659^1664, 1999. R Jenison, S Yang, A Haeberli, B Polisky. Nat Biotechnol 19:62^65, 2001. AW Czanderna, C Lu, Applications of Piezoelectric Quartz Crystal Microbalances, Methods and Phenomena. New York: Elsevier,1984, vol 7. DA Buttry, MD Ward. Chem Rev 92:1355^1379, 1992.

606

Tarlov and Steel

189. 190. 191. 192.

NC Fawcett, RD Craven, P Zhang, JA Evans. Anal Chem 70:2876^2880, 1998. BA Cavic, GL Hayward, M Thompson. Analyst 124:1405^1420, 1999. A Jansho¡, HJ Galla, C Steinem. Angew Chem Int Ed. 39:4004^4032, 2000. S Yamaguchi,T Shimomura,T Tatsuma, N Oyama. Anal Chem 65:1925^1927, 1993. J Wang, PE Nielsen, M Jiang, XH Cai, JR Fernandes, DH Grant, M Ozsoz, A Beglieter, M Mowat. Anal Chem 69:5200^5202, 1997. F Caruso, E Rodda, DF Furlong, K Niikura, Y Okahata. Anal Chem 69:2043^2049, 1997. Y Okahata, M Kawase, K Niikura, F Ohtake, H Furusawa, Y Ebara. Anal Chem 70:1288^1296, 1998. K Niikura, H Matsuno,Y Okahata. J Am Chem Soc 120:8537^8538, 1998. F Patolsky, A Lichtenstein, I Willner. J Am Chem Soc 122:418^419, 2000. L Lin, HQ Zhao, JR Li, JA Tang, MX Duan, L Jiang. Biochem Biophys Res Commun 274:817^820, 2000. A Bardea, A Dagan, I Willner. Anal Chim Acta 385:33^43, 1999. HG Hansma. Annu Rev Phys Chem 52:71^92, 2001. J Fritz, MK Baller, HP Lang, H Rothuizen, P Vettiger, E Meyer, HJ Guntherodt, C Gerber, JK Gimzewski. Science 288:316^318, 2000. GH Wu, HF Ji, K Hansen, T Thundat, R Datar, R Cote, MF Hagan, AK Chakraborty, A Majumdar. Proc Natl Acad Sci USA 98:1560^1564, 2001. B Tinland, A Pluen, J Sturm, G Weill. Macromolecules 30:5763^5765, 1997. KM Hansen, HF Ji, GH Wu, R Datar, R Cote, A Majumdar,T Thundat. Anal Chem 73:1567^1571, 2001. GU Lee, LA Chrisey, RJ Colton. Science 266:771^773, 1994. K Hu, R Pyati, AJ Bard. Anal Chem 70:2870^2875, 1998. J Wang, AJ Bard. Anal Chem 73:2207^2212, 2001. DR Baselt, GU Lee, M Natesan, SW Metzger, PE Sheehan, RJ Colton. Biosensors Bioelectron 13:731^739, 1998. MM Miller, PE Sheehan, RL Edelstein, CR Tamanaha, L Zhong, S Bounnak, LJ Whitman, RJ Colton. J Magn Magn Mater 225:138^144, 2001. AT Woolley, C Guillemette, CL Cheung, DE Housman, CM Lieber. Nat Biotechnol 18:760^763, 2000. R Moller, A Csaki, JM Kohler,W Fritzsche. Nucleic Acids Res 28:e91, 2000. ML Bulyk, E Gentalen, DJ Lockhart, GM Church. Nat Biotechnol 17:573^577, 1999. JM Brockman, AG Frutos, RMJ Corn. Am Chem Soc 121:8044^8051, 1999. K Niikura, K Nagata,Y Okahata. Chem Lett 863^864, 1996. Y Okahata, K Niikura, Y Sugiura, M Sawada, T Morii. Biochemistry 37:5666^5672, 1998. LM Furtado, HB Su, M Thompson, DP Mack, GL Hayward. Anal Chem 71:1167^1175, 1999. J Wang, M Jiang, E Palecek. Bioelectrochem Bioenerget 48:477^480, 1999. HB Su, P Williams, M Thompson. Anal Chem 67:1010^1013, 1995. YJ Wang, N Farrell, JD Burgess. J Am Chem Soc 123:5576^5577, 2001.

193. 194. 195. 196. 197. 198. 199. 200. 201. 202. 203. 204. 205. 206. 207. 208. 209. 210. 211. 212. 213. 214. 215. 216. 217. 218. 219.

DNA-Based Sensors

607

220. MS Yang, HCM Yau, HL Chan. Langmuir 14:6121^6129, 1998. 221. J Wang, M Ozsoz, XH Cai, G Rivas, H Shiraishi, DH Grant, M Chicharro, J Fernandes, E Palecek. Bioelectrochem Bioenerget 45:33^40, 1998. 222. SO Kelly, NM Jackson, MG Hill, JK Barton. Angew Chem Int Ed 38:941^945, 1999. 223. A Erdem, M Ozsoz. Anal Chim Acta 437:107^114, 2001. 224. J Wang, G Rivas, MA Jiang, XJ Zhang. Langmuir 15:6541^6545, 1999. 225. CA Widrig, C Chung, MD Porter. J Electroanal Chem 310:335^359, 1991. 226. CJ Zhong, MD Porter. J Electroanal Chem 425:147^153, 1997. 227. J Wang, M Jiang, B Mukherjee. Bioelectrochemistry 52:111^114, 2000. 228. J Wang. Electroanalysis 13:635^638, 2001. 229. J Wang, G Rivas, X Cai, E Palecek, P Nielsen, H Shiraishi, N Dontha, D Luo, C Parrado, M Chicharro, PAM Farias, FS Valera, DH Grant, M Ozsoz, MN Flair. Anal Chim Acta 347:1^8, 1997. 230. J Wang, G Rivas, DB Luo, XH Cai, FS Valera, N Dontha. Anal Chem 68:4365^ 4369, 1996. 231. J Wang, M Chicharro, G Rivas, XH Cai, N Dontha, PAM Farias, H Shiraishi. Anal Chem 68:2251^2254, 1996. 232. XH Cai, G Rivas, PAM Farias, H Shiraishi, J Wang, M Fojta, E Palecek. Bioelectrochem Bioenerget 40:41^47, 1996. 233. J Wang, G Rivas, M Ozsos, DH Grant, XH Cai, C Parrado. Anal Chem 69:1457^1460, 1997. 234. J Wang, M Jiang, AN Kawde. Electroanalysis 13:537^540, 2001. 235. M Vorlickova, E Palecek. Biochim Biophys Acta 517:308^318, 1978. 236. P Boublikova, M Vojtiskova, E Palecek. Anal Lett 20:275^291, 1987. 237. F Jelen, M Fojta, E Palecek. J Electroanal Chem 427:49^56, 1997. 238. F Jelen, M Tomschik, E Palecek. J Electroanal Chem 423:141^148, 1997. 239. T Kubicarova, M Fojta, J Vidic, L Havran, E Palecek. Electroanalysis12:1422^ 1425, 2000. 240. E Palecek, M Fojta. Anal Chem 73:74A ^83A, 2001. 241. J Mbindyo, LP Zhou, Z Zhang, JD Stuart, JF Rusling. Anal Chem 72:2059^ 2065, 2000. 242. GA Soukup, RR Breaker. Proc Natl Acad Sci USA 96:3584^3589, 1999. 243. GA Soukup, RR Breaker. Trends Biotechnol 17:469^476, 1999. 244. GA Soukup, RR Breaker. Curr Opin Struct Biol 10:318^325, 2000. 245. RR Breaker. Chem Rev 97:371^390, 1997. 246. J Li,Y Lu. J Am Chem Soc 122:10466^10467, 2000. 247. S Seetharaman, M Zivarts, N Sudarsan, RR Breaker. Nat Biotechnol 19:336^ 341, 2001. 248. MP Robertson, AD Ellington. Nat Biotechnol 19:650^655, 2001. 249. D Ho¡man, J Hesselberth, AD Ellington. Nat Biotechnol 19:313^314, 2001. 250. NC Seeman. Angew Chem Int Ed 37:3220^3238, 1998. 251. NC Seeman, H Wang, XP Yang, FR Liu, CD Mao, WQ Sun, L Wenzler, ZY Shen, RJ Sha, H Yan, MH Wong, P Sa-Ardyen, B Liu, HX Qiu, XJ Li, J Qi, SM

608

252. 253. 254. 255. 256. 257. 258. 259. 260. 261.

Tarlov and Steel Du,YW Zhang, JE Mueller,TJ Fu,YLWang, JH Chen. Nanotechnology 9:257^ 273, 1998. AP Alivisatos, KP Johnsson, XG Peng, TE Wilson, CJ Loweth, MP Bruchez, PG Schultz. Nature 382:609^611, 1996. CJ Loweth,WB Caldwell, XG Peng, AP Alivisatos, PG Schultz. Angew Chem Int Ed 38:1808^1812, 1999. KJ Watson, SJ Park, JH Im, ST Nguyen,CA Mirkin. JAm Chem Soc123:5592^ 5593, 2001. JKN Mbindyo, BD Reiss, BR Martin, CD Keating, MJ Natan, TE Mallouk. Adv Mater 13:249^254, 2001. J Richter, M Mertig, W Pompe, I Monch, HK Schackert. Appl Phys Lett 78:536^538, 2001. J Richter, R Seidel, R Kirsch, M Mertig,W Pompe, J Plaschke, HK Schackert. Adv Mater 12:507^510, 2000. T Torimoto, M Yamashita, S Kuwabata,T Sakata, H Mori, H Yoneyama. J Phys Chem B 103:8799^8803, 1999. CM Niemeyer. Appl Phys A
Mater Sci Process 68:119^124, 1999. E Braun,Y Eichen, U Sivan, G Ben-Yoseph. Nature 391:775^778, 1998. SM Waybright, CP Singleton, K Wachter, CJ Murphy, UHF Bunz. J Am Chem Soc 123:1828^1833, 2001.

Index An italic page number indicates that the entry appears in a figure; an italic t following a page number indicates that the entry appears in a table.

absolute quality factor, 354 absorption: of adsorbed layers, 194 of cytochrome c, 237 and fluorescence, 196–201 molecular, 193 optical, 192–193 protein films, 6t and SECM, 260 Stark effect, 236 absorption spectroscopy, 232–236, 240–241 acetylcholine, 29 acetylcholine receptor, 149–150 acetylcholinesterase, 486, 487 N-acetylcysteine, 90 acidity: See pH acoustic impedance, 363, 365, 369 acoustic impedance spectroscopy, 120–121 actin, 173–174 acyl, 141

adenine, 52–53, 547 adenosine phosphosulfate reductase, 184 adenosine triphosphatase, 145, 152 adenosine triphosphate, 149 admittance/impedance analysis, 351–356 adsorbed films, 25–27, 34 adsorption: of alkanethiol, 69–70, 74, 77–78 alternate, 44–45 and antibodies, 469 of avidin–biotin, 361 and biosensors, 469 of BSA, 189 of catalase, 189 of cytochrome c, 26–27, 83 of cytochrome P450, 205 and dehydrogenase enzymes, 506–507, 520 of DNA, 556, 558 of enzymes, 469 609

610 [adsorption] of glucose oxidase, 365–366, 372–373 of human growth hormone, 197 in hybrid bilayers, 109, 110–111, 128 and hydrophilicity, 197 layer-by-layer, 337–339, 366–367, 383–393, 397–400 of lipids, 128 of nanoparticles, 445 nonspecific, 166–167, 178–180 of phospholipids, 154 of proteins, 40–41, 167, 172, 178–180, 203, 342, 356–358, 361–363 and SECM, 260 and SPR, 218 timing, 48 aerosols, 462 affinity-type biosensors, 463–468, 492 AFM: See atomic force microscopy agglutinin, 193 air–water interface, 34, 167–169, 196, 286–296 alamethicin, 302 alanine, 170, 203–204 albumin, 340, 357, 361 See also bovine serum albumin (BSA) alcohols, 223, 539 aliphatic, 289–291 aldehydes, 223, 508, 539 algae, 318–319, 462 alkane chains, 137 alkanethiol: and adsorbed films, 27 and bilayer lipid membranes, 35–36 and capacitance, 114, 145 chain length, 81 and DNA biosensors, 556 in hybrid bilayers, 104–105, 107, 111–112, 114, 115–116, 120, 122–124, 126, 135, 145, 153–154 melittin effect, 137 and plasma proteins, 203

Index [alkanethiol] and QCM, 128–129 and self-assembly, 27, 66, 87–90, 119, 128–129, 172–173, 180, 230–232 and SPR, 230–232 alkyl chain length, 78 alkythiol, 6t, 68–69 alternate adsorption, 44–45 alternating direction implicit finitedifference method (ADIF-DM), 269–270 aluminum, 590 American convention, 8–9 amino groups, 506 a-amino isobutyric acid, 203–204 aminopeptidase N, 144 aminopropyltriethoxysilane, 281 ammonium, 280 amperometric enzyme electrodes, 479–481, 536–537t amperometry, 454, 460–461t, 467, 477–478 chrono-, 260, 276, 290–291 See also current amphiphiles, 291–292 amplitude reflectivity, 174 analytes, 452 antibiotics, 151 antibodies: and adsorption, 469 anti-atrazine, 483 anti-C4, 183 antigen binding: and biosensors, 456–457, 463, 482–483 and ellipsometry, 175 and glucose oxidase, 343 kinetics, 224 and layer-by-layer, 44 and piezoelectric transducers, 492 and polyelectrolytes, 340–342 and SECM, 279–280 and spectroscopy, 489 as biosensors, 456–457, 463–467

Index [antibodies] and biotin–avidin interaction, 478–479 blood group-specific, 489 and cell membrane hybrids, 150 and DNA microarrays, 597 and enzymes, 44, 463, 467, 482–484 Fc-binding, 479 and fluorescence, 173–174 immobilization, 170, 342, 477, 478–479 labeled, 173–174 and liposomes, 186, 341–342 monoclonal, 180, 224, 465, 466 and polyelectrolyte layers, 340–341 and QCM, 183–185, 186, 359–361, 374 recombinant, 465–467 versus ribozymes, 597 and SECM, 279–280 See also Fab0 ; immunoglobulins; immunosensors antigens: See antibodies, antigen binding applied electric field, 312 A protein, 479 arachidic acid, 83, 196, 433 ascorbate, 508 ascorbic acid, 145 atomic force microscopy (AFM): and biomolecular assemblies, 189–191, 207 of cytochrome c, 26–27 and cytochrome c oxidase, 131, 149 and DNA biosensors, 558, 580, 588–589 and enzymes, 345, 366 and hybrid bilayers, 129–133, 149, 154 and layer-by-layer growth, 51 and myoglobin inserts, 40 and protein-DNA binding, 591–592 with SECM, 323–327 with SPR, 230

611 [atomic force microscopy (AFM)] tapping mode, 189–191, 207 atoms, 201 atrazine, 483 attenuated total reflection (ATR), 203, 204, 220 avidin–biotin: adsorption, 361 and antibodies, 478–479 and DNA probes, 553, 570–571, 584–585 and glucose-oxidase, 344 and Langmuir-Blodgett, 167–168 and protein biomolecular assemblies, 167–168, 179–180, 183 and protein films, 338–339, 361–363, 371 and SECM, 281–283 and SPR, 179–180, 223 azurin, 27, 91 bacteria: biomineralization, 428 as biosensors, 462 cell wall, 151 channel gating, 147–149, 151–152 Desulfovibrio dusulfuricans, 26, 85 Rhodobacter sphaeroides, 1, 42–43 S. typhimurium, 492 sulfur-reducing, 31–32 See also Escherichia coli bananatrodes, 463 basal-plane graphite, 26–27, 30, 35 basicity: See pH beads, paramagnetic, 283 benzylviologen, 539 BIAcore, 470, 489 bi-cell detection, 221–222, 244–247 bilayer lipid membranes (BLMs): advantages, 6t electrodes, 32–36, 301–302 and iodine, 302–303 permeability, 259, 301–302, 305–310, 315–317

612 [bilayer lipid membranes (BLMs)] and potassium, 303–305 solute flux, 310 topographical imaging, 297–300, 315–322, 327 transport processes, 300–310 unstirred layers, 305–310 bilayer membranes: See hybrid bilayers biohazards, 43, 594 biological membranes: models, 100–104 structure/function, 99–100, 165 See also biomolecular assemblies; hybrid bilayers bioluminescence, 488 biomimetic materials, 430 biomineralization, 428 biomolecular assemblies: absorption measurement, 192–196 advantages, 167–171 and AFM, 189–191, 207 and ellipsometry, 174–175, 176, 206 and fluorescence, 196–201 mechanical properties, 183–192 metal nanoparticles, 296 and microscopy, 172–174 and SECM, 207, 276–283 and SPR, 175–182, 185–187 vibrational spectroscopy, 202–206 and XPS, 201–204, 206–207 biomolecular films: as biosensors, 454, 482 (dis)advantages, 6–7t and fluorescence, 196–197 future applications, 56–57 history, 3–4 immunosensing, 482 biosensors: affinity type, 463–468, 492 aldehyde, 508 antibodies as, 456–457, 463–467 applications, 56–57, 451–453 bacteria as, 462 and biological interactions, 478–479

Index [biosensors] biomolecular films as, 454, 482 blood glucose level, 480–481 and catalysis, 458–459, 487 cells as, 459–463, 467–468 dehydrogenase enzymes as, 532–534, 535, 538 DNA-based, 546 see also biosensors, DNA-based fiber-optic, 455–456, 486, 488 history of, 453–457 hybrid bilayers as, 143–150 and hydrophilic redox species, 145 immobilization techniques, 469–479 metabolism type, 458–463, 492 nonenzymatic, 29 optical, 455–456 performance, 549–550 piezoelectric, 456–457 potentiometric enzyme, 581–582 reusability, 478–479 and SPR, 182, 223, 455–456, 457, 489 and temperature, 457, 459, 463, 467 transducers, 475–479 see also transduction methods See also biosensors, DNA-based; immunosensors biosensors, DNA-based: and AFM, 558, 580, 588–589 applications, 549–551, 590–597 catalytic, 594–597 and charge, 554–555, 579, 590 and concentration, 550 covalence, 552, 553, 564–565 DNA structure, 547 and drugs, 593–594 and environmental toxins, 594 fabrication, 551–554 for genetic assays (probes), 549, 560–574 hybridization: critical factors, 554 description, 547–549

Index [biosensors, DNA-based] detecting, 563–564, 565, 570, 574–581, 585–590 enhancing, 555–560, 567–568, 578–580 immobilization, 549, 552–553, 562–563, 570 interface characterization, 554–560, 585 mass sensors, 583–585 microcantilever, 585–590 miniaturization, 598 mismatches, 569, 578, 582, 590, 593 single-base, 573, 590, 593 optical strategies, 574–583 performance, 564–571, 597–599 probe length, 551–552, 556 and protein–DNA binding, 591–592 selectivity: duplex versus ss, 568, 588 sequence, 573 and thickness, 582–583 See also deoxyribonucleic acid (DNA); electrochemistry biotin: and DNA probes, 551, 576, 578, 583 and horseradish peroxidase, 359–360, 583 See also avidin–biotin biotin-streptavidin: and biomimetic membranes, 144 and biomolecular assemblies, 167–168, 175, 180, 183 and DNA probes, 576, 578 and ellipsometry, 175 and enzyme reactors, 387–388 and fluorescence, 198 and Langmuir-Blodgett, 167–168 and protein films, 357 and QCM, 357 and SPR, 144 4,40 -bipyridine, 80, 84 4,40 -bipyridyl, 85–86 blood, 145 glucose in, 480–481

613 [blood] hemoglobin, 31, 340, 358 blood cells, 150, 152, 306–307 blood groups, 489, 490 bone cells, 173–174, 201, 322 bovine serum albumin (BSA): adsorption, 189 and AFM, 190 and Fab0 , 180 and fluorescence, 199–201 and immobilization, 473 and impedance, 43 and QCM, 357, 361 selective adsorption, 189 brain cells, 173 breast cells, 321–322 bromine, 290–291 Butler-Volmer theory, 15–18 Butterworth-Van Dyke (BVD) circuit, 347–348, 351, 352, 355 cadmium perchorate, 437–442 cadmium selenide, 446, 554 cadmium sulfide, 433, 440–442, 446, 554 calcium, 228, 307–309, 322 calcium oxalate, 314 calculators: See enzyme reactors calmodulin, 185, 228 calorimetric transduction, 492–494 cancer, 172–173, 321–322 Candida albicans, 492 cantilevers, 189–191, 585–590 capacitance, 112–114, 132–133, 152–154 Helmholtz, 75–76, 117 capillary electrophoresis, 32, 52 carbon fiber electrodes, 521–522 carbonic anhydrase, 179 carbon paste electrodes: and DNA hybridization, 565 drawbacks, 507, 520 and groundwater, 594 and NADH oxidation, 520, 521, 522–523

614 carbon surfaces, 281–283 carbonyl groups, 302 carboxyanhydride coupling, 170, 203 carboxyl groups, 25, 28–29 carboxylic acid: and cytochrome c, 85–86 and SAMs, 76–77, 80–81, 85–86, 87–90 and SECM, 278, 300–301 and SPR, 179–180 carboxylic groups, 506 cardiolipin, 227 cartilage, 314–315 cast films: advantages, 6t, 7t and DNA damage, 52 DNA-ionomer, 32 lipid, 40 multilayer, 193 polymeric, 29–32, 507 protein–polyion, 29–32 surfactant, 36–42, 48 and temperature, 40 casting, 36 catalase, 189 catalysis: and biosensors, 458–459, 487, 594–597 and covalent bonding, 27–29 and DNA, 594–597 on electrodes, 19–24 in enzyme reactors, 391–393, 395–400 in nanoparticles, 436 and polymers, 507 and redox enzymes, 499–500 and self-replication, 436–442 and square wave voltammetry, 52 of TCA reduction, 31 See also dehydrogenase enzymes; enzymes cathodes, 538, 539 cation probe, 53–55 cation-selectivity, 302

Index cells: adhesion, 172, 201 as biosensors, 459–463, 467–468 blood, 150, 152, 306–307 bone, 173–174, 201, 322 brain, 173 breast, 321–322 cancer, 172–173, 321–322 COS, 150 dehydrogenase path, 500–502 endothelial, 201 heart, 173 kidney, 173 of leaf, 318 membranes, 150, 459, 487 mitochondria, 31–32, 227, 321, 500–502 nerve, 173 and proton hopping, 293–296 and SECM, 268–269, 317–322 and semiconductors, 173 spreading, 173–174 umbilical, 201 wall of, 151 Cerasomes, 420 cetyltrimethylammonium bromide (CTAB), 405 chain length and hybrid bilayers, 115, 120–122 and self assembly, 71, 73, 74, 76, 77–79, 81, 87–89 and silane layers, 194 of surfactants, 33 chain orientation, 125–127 chain thickness, 137 channel gating, 147–149, 151–152 channels, 103 charge: control of, 474 distribution, 52 and DNA biosensors, 554–555, 579, 590 and polyelectrolytes, 342 surface, 215

Index [charge] and thickness, 528 charge-coupled devices, 221 chemicals, commercial, 535 chemical toxicity, 4, 52–55 chemiluminescense, 197–198, 488 chemokine receptors, 150 chemotherapy agents, 488, 568–569, 593 chloramphenicol-methyl red, 486 chlorine, 20 chlorophenol red, 486, 487 cholera, 143–144 cholesterol, 125, 131, 306–307 chronoamperometry, 260, 276, 290–291 a-chymotrysinogen A, 357–358 circular dichroism, 196 cis-platin, 593 clay platelets, 43 coagulants, 144, 203 cobalt, 53–55, 522, 566–567, 595–596 cocaine, 492 co-enzymes, 499–500 colloidal particles, 388, 428, 446–447 colloidal supports, 553, 558 complex phase elements, 114 composite films, 7t computers: See enzyme reactors concanavalin A, 193, 387, 455 concentration: and current, 21 and DNA-based biosensors, 550 and SECM, 255, 259 surface, 12, 109 concentration polarization, 526–532 contrast variation, 134, 138 convection, 268, 312–314 conventions, 8–9 copper: and enzyme reactors, 406, 407–416 and nanoparticles, 436 problems, 169 and self-assembly, 73, 169 and SPR, 223, 228

615 copper phthalocyanine films, 223 p-coumaric acid, 488 counterion transport, 10–11 covalence: coupling agents, 506 and dehydrogenase enzymes, 506–507, 523 and DNA probes, 552, 553, 564–565 to electrode surface, 477 to gold, 7t, 27, 29 and immobilization, 144, 469–470 with metal supports, 103 and nanoparticles, 445–446 proteins to electrodes, 27–29, 82 and SAMs, 27–29, 82 creatinine, 469, 474, 481–482 cross-linking, 470–473, 581–582 CrylAc, 151 cryovoltammetry, 26 current: background, 562–563, 565, 566–567, 568 and biosensors, 454, 467, 477–478 DNA-based, 562–563, 565, 566–567, 568, 570, 573 and concentration, 21 and hybrid bilayers, 118–121, 152 and NADH oxidation, 526–532, 533, 538 peak, 9, 12, 13 pinhole redox, 152 reversible, 9 and time, 275–276 tip (SECM), 271, 311–313 cyclic adenosine monophosphate (cAMP), 595–596 cyclic guanosine monophosphate (cGMP), 595–596 cyclic voltammetry, 5–14 diffusion-controlled, 12, 39 and metalloproteins, 37 of protein–surfactant films, 39–40 and reversible, 37 with SPR, 229–232 cyclophanes, 407

616 cyctochrome c, and bipyridine, 84 cystamine, 366 cysteamine, 520 cysteine, 27, 83–84 cytochrome b5: and cytochrome c, 226 and prosthetic groups, 83–84 and Raman scattering, 205–206 and SPR, 226 cytochrome c: absorption, 237 absorption spectroscopy, 233–235 adsorption, 26–27, 83 AFM, 26–27 at bare electrodes, 79–80 and bilayer lipid membranes, 35 and carboxylic acid, 85–86 and cardiolipin, 227 and cytochrome b5, 226 and dichroism, 196 and Eastman AQ ionomers, 31–32 heme group orientation, 83 and hybrid bilayers, 110 and Nafion, 30 orientation, 194–196 and polyelectrolyte layers, 340 and Raman scattering, 205 redox-induced changes, 228, 232, 235–236 and SAMs, 80–86, 86–91, 194–196 and silver electrodes, 84–86 and SPR, 226, 230–232 cytochrome c oxidase (COX): and hybrid bilayers, 131, 149, 152 resting state, 152 and SPR, 230–232 cytochrome P450: and cast surfactant films, 37–38 charge distribution, 52 DNA damage, 4 and layering, 47, 51–52 and Raman scattering, 205 and square wave voltammetry, 16–17

Index daunomycin, 568–569, 593 decane, 34 dechlorination, 20 defects: and electroactivity, 145–146 and hybrid bilayers, 118–119, 120–121, 145–146 and SAMs, 119 dehydrogenase enzymes: applications, 532–540 as biosensors, 532–534, 535, 538 covalence, 506–507, 523 and electrodes, 502–508 amperometric, 536–537t modifications, 504–523 kinetic models, 523–532 D-glucose dehydrogenase, 538 in metabolism, 500–502 NAD(H)-dependent, 232 NADH-linked reductases, 321 dendrimers, 553 density, 367, 371–373 and acoustic impedance, 363 scattering length, 134–138, 140–142 of surface charge, 215 dentine, 312–314 deoxyribonucleic acid (DNA): and bioluminescence, 488 in cast films, 32 catalytic, 594–597 damage screening, 4, 32, 52–55 and drugs, 593–594 duplex (ds), 547–549, 550, 558, 590 versus ss, 568, 588 and hybrid bilayers, 144 hybridization: critical factors, 554 description, 547–549 detecting, 563–564, 565, 570, 574–581, 585–590 enhancing, 555–560, 567–568, 578–580 and ionomers, 32 and layering, 49, 50–55

Index [deoxyribonucleic acid (DNA)] melting, 549, 550, 555 microarrays, 546, 551, 597 microfluidic delivery, 546–547, 598–599 and oxidation, 24–25, 32, 52–53, 561–563 and potential, 564–565, 578–579, 590, 593–594 protein binding, 591–592 and QCM, 564, 570–571, 583–585, 586 single-base mismatches, 573, 590, 593 single-nucleotide polymorphisms, 545–546, 550–551, 567–568 stringency, 549, 579–580 structure, 547 surface-confined, 556, 558 and temperature, 549, 552, 553, 555 as therapeutic agent, 593–594 See also guanine deoxyribozymes, 594–597 deprotonation, 76–77 desorption, 218, 445, 593–594 Desulfovibrio dusulfuricans, 26, 85 detergent, 35–36, 149, 150 deuterium, 134–136, 141 diaphorase, 279, 282, 283, 522–523, 539 dichroism: circular, 196 linear, 194 dichromism, 83–84 didodecyldimethylammonium bromide, 18 dielectric–dielectric interface, 197 dielectric–metal interface, 177–178, 215–217, 242–244 dielectric response, 174 Diels-Alder, 180 diethylaminoethyl methacrylate (DEAEM), 486

617 differential scanning calorimetry (DSC), 408 diffusion: across monolayer, 287–291 back, 264, 306 lateral, 101–104 amphiphilic, 291–293 proton, 293–296 and SECM, 259, 268, 272, 275–276 and unstirred layers, 305–310 diffusion coefficient, 90 diffusion-controlled voltammetry, 12–13, 39 diffusion layer, 305–310 dihydrofolate reductase, 228 3,4-dihydroxybenzaldehyde, 508 dimethylaminopyridine (DMAP), 194 dimyristoylphosphatidylcholine (DMPC): acyl chains, 141 and cyclic voltammetry, 10 as electrode coating, 21 in hybrid bilayers, 114, 115, 123, 126–127, 135 and neutron reflectivity, 135–137 and nonlinear optical spectroscopy, 141 and photoswitching, 413 dioctadecyldimethylammonium bromide, 43 dioxygen, 538 dipalmitoyl-L-a-phosphatidic acid, 204 4,40 -dipyridyl disulfide, 75, 80 dissipation energy, 128, 354, 358 distillation, 223 dodecylamine, 35 double potential step chronoamperometry (DPSC), 260, 276, 290–291 doxorubicin, 593 drugs, 488, 568–569, 593 transdermal delivery, 312

618 drying, 48, 371 dual-mediator imaging, 315–317 dyes: and dehydrogenase enzymes, 519–524, 535, 538 and DNA-based biosensors, 568 dynamic reorganization, 103 Eastman AQ ionomers, 31–32, 52 edge-plane pyrolytic graphite, 6t, 25 Einstein equation, 13 Elasticity: See viscoelasticity electroactive protein, 8, 50–51 electrochemistry: and DNA-based sensors, 561–563 adduct formation, 569 intercalators, 568–569, 593 nanopores, 571–574 polymeric electrodes, 563–564 redox indicators, 564–569 of drug-DNA reactions, 593–594 thin-film, 5–25, 9 electrode–film interface, 42 electrodes: activated, in base, 519 amperometric enzyme, 479–481, 536–537t bare, 38, 79–80, 86, 506 and bilayer membranes, 32–36, 301–302 carbon fiber, 521–522 carbon paste: see carbon paste electrodes cathodes, 538, 539 composition, 42 covalent bonding with, 27–29, 82 and dehydrogenase enzymes, 502–508 modifications, 504–523 kinetic models, 523–532 design advances, 475–479 DNA-modified, 562, 569 polymeric, 563–564 and enzymes: amperometric, 454–455, 479–482

Index [electrodes] catalysis, 19–24, 82 chemically modified, 343 electron transfer, 2 monolayered, 506–507 multilayered, 341, 507–508 fuel cell, 534–535 impurities on, 40–41 ion-selective micropipette, 303–305 lipids and surfactants on, 32–36 micro-, 300–305, 307–310, 326 see also ultramicroelectrodes monolayer versus multilayer, 506–508 oscillation amplitude, 327 potentiometric, 481–482 rotating disk, 22, 528–529 roughened, 45–48 SAM-modified, 73–76 and SECM, 255, 264–265, 267, 300–305 submarine, 261, 264, 268–269, 295 submicrometer, 264 surface-modification, 65–66 See also gold; graphite; silver electrolyte–metal interface, 217–219 electromagnetic waves, 215 electron hopping, 19, 39, 47 electron transfer: and adsorbed films, 25 in bilayer lipid membranes, 300–310 in biomimetic photosynthesis, 443–445 direct: and detergent, 35 DNA and electrodes, 560 enzymes and electrodes, 2 and polymers, 29–32 rate, 19 distance of, 19 four-electron, 538 in hybrid bilayers, 118–121, 129, 146 kinetics, 17–18, 86–91 Marcus theory, 17–19 measurement, 87

Index [electron transfer] and melittin, 152 and metal, 26 at modified electrodes, 86–91 in NADH oxidation, 525–526, 538, 540 outer sphere, 17 polymer–protein films, 29–32 rate, 86 reversible, 180 and SECM, 265–267, 272–273, 284–286, 292–293, 300–310 and self assembly, 73–76, 86–91 self-exchange, 19, 39, 47 between solutions, 265–267 and temperature, 86, 91, 286 and thick membranes, 310–317 vectorial, 28 and viscosity, 86–89 electrophoresis, 32, 52 electrostatic binding, 445 electrostatic repulsion, 302, 556 electrostatic self-assembly, 170 ellipsometry: and antibodies, 175 description, 174–175, 176 and DNA biosensors, 554 ex situ, 128 and hydration, 367, 373 and QCM, 371–373 in situ, 127–128 and thickness, 206, 367, 369–371, 373 embryos, 322 emission anisotropy, 194–195, 197 encapsulation, 474 endgroups, 76–77, 79 endothelial cells, 201 energy: dehydrogenase enzymes, 534–539 dissipation, 128, 354, 358 fuel cells, 534–539 reorganization, 90, 91 and SPR, 178 surface free, 154

619 enthalpy, 492–494, 555 entrapment, 473, 477 sol-gel, 488 enzyme films, 342–345, 356, 367, 371–374 enzyme loading, 10 enzyme reactors: activity enhancement, 395, 396 applications, 418–420 catalysis, 391–393, 395–400 and layer-by-layer adsorption, 383–393, 397–400 methodology, 388–390 and lipids, 387, 401, 403–416, 419–420 and logic gates, 417–418 multienzyme, 395–400 polyion-based, 420 and QCM, 390–391, 395, 396 stability, 393–395, 416 switchable, 400–412, 420 by photoregulation, 412–418 and temperature, 393–395 enzymes: active site, 499–500 adsorption of, 469 and AFM, 345, 366 allosteric, 594 amperometric electrodes, 479–481, 536–537t and antibodies, 44, 463, 467, 482–484 artificial: see enzyme reactors and bilayer lipid membranes, 32 as biosensors, 453–459, 463, 467, 474–475, 481–482 co-enzymes, 499–500 and electrodes, 19–24, 82–83, 341, 343, 454–455, 479–482 and fluorescence, 486–487 and hybrid bilayers, 144, 145 immobilization of, 383–384, 400, 405, 418–420, 474, 477 and layer-by-layer growth, 44 and luminescence, 488

620 [enzymes] and myoglobin, 20, 22, 193 and NAD, 28–29 and polyelectrolyte layers, 340 and protein–DNA binding, 591–592 and QCM, 356, 371–374 redox, 19–24, 26, 28, 82, 499–500 and SECM, 255, 276–280, 321 and self-assembly, 81–83, 340, 356, 367 and signal transmission, 383 See also catalysis; dehydrogenase enzymes; specific enzymes enzyme–substrate complex, 20–21 enzyme thermistors, 493–494 erythrocytes, 150, 152, 306–307 Escherichia coli: and bioluminescence, 488 dihydrofolate reductase, 228 and fluorescence, 487 and genetic engineering, 466, 596 lectin binding, 193 ethylenediamine, 407 ethylene oxide, 135–137, 146–147, 148, 149 evanescent wave, 455–456, 457 exfoliated clay platelets, 43 Fab0 : and antibodies, 180, 189–191, 341–342 assembly, and SPR, 180 and ellipsometry, 175 and fluorescence, 199 fabrication: of DNA-based biosensors, 551–554 of hybrid bilayers, 104, 145, 150, 153–154 layer-by-layer, 7t, 44–55, 203, 343, 361 of memory storage devices, 447 at nanometer level, 381–382, 428, 430–433 faradaic currents, 118–120 Faraday’s law, 8

Index Fenton reaction, 281 ferredoxin, 30–31, 35 ferric ammonium citrate, 31–32 ferrocenecarboxylic acid, 300–302 ferrocene (Fc) binding, 344, 364–365, 479 ferrocene monocarboxylic acid, 278 ferrocyanide: and glucose biosensor, 481 and NADH oxidation, 523 and SECM, 285, 298, 302, 312–313, 318–319 and SPR, 229–230 fiber-optic biosensors, 455–456, 486, 488 fibrinogen, 203 finite-difference methods, 269 finite-element methods (FEM), 269 fixed angle method, 221 flavin reductase, 521 flavor, 223 fluorescence: and absorption, 196–201 and biomolecular assemblies, 196–201 and DNA-based biosensors, 554, 576, 578, 580, 595 and emission anisotropy, 194–195 and enzymes, 486–487 from labeled antibodies, 173–174 and oxidase enzymes, 486–488 fluorescence microscopy, 172 fluorescence recovery after photobleaching (FRAP), 199–201 fluorescence spectroscopy, 165, 196–201 fluorescent molecular beacons, 580 formate dehydrogenase, 535 Fourier transform-infrared spectroscopy, 194 Fresnel equations, 127, 174, 175–177, 371 fructose dehydrogenase, 145 fruit extracts, 145

Index fuel cells, 534–539 fullerenes, 388 gases, 492 Gaussian distribution, 18 genes: damage screens, 4, 52–55 expression monitoring, 547–549, 551, 562 hazard detection, 488 metabolic activity, 547 genetic assays, 549, 560–574 genetic engineering, 185, 193–194, 465–467, 545–546 glass, 101, 131, 264 glucoamylase, 388–390, 395–400 glucose: blood levels, 2, 480–481 oxidation of, 29 D-glucose dehydrogenase, 538 glucose oxidase: adsorption, 365–366, 372–373 and AFM, 189 and antibody–antigen binding, 343 and biosensors, 453–454, 459, 479–481, 488 enzyme reactor, 388–395 and fluorescence, 488 and QCM, 342–343, 350, 356, 361–366 with ellipsometry, 371–373 and SAMs, 81–82 and SECM, 278–279, 281 glutathione reductase, 82 glutathione-s-transferase, 185 glycopeptides, 151 glycoproteins, 193 glycosylphosphatidylinositol, 144 gold: and AFM, 131–132 and antibodies, 341, 342 and bilayer lipid membranes, 34–36 and covalent bonding, 7t, 27, 29 and dehydrogenase enzymes, 506 (dis)advantages, 6t, 7t

621 [gold] and DNA: biosensors, 552–553, 555, 558, 565, 568, 576, 577, 580, 585–588 drug interaction, 593 nanoparticles, 576, 581–582 protein-DNA binding, 591 therapeutic use, 593–594 and dyes, 520 and horseradish peroxidase, 29, 197–198 and hybrid bilayers, 102, 105–106, 117–119, 121–126, 144 and hydrophilic redox species, 145 and infrared spectroscopy, 203 and layering, 44, 45, 49 and luminescence, 197–198 as membrane coating, 482 nanoparticles, 296, 445–446 and QCM, 128–129, 183, 364–365 and redox species, 145–146 and self-assembly, 27, 87, 90, 169, 203 and SPR, 105–106, 223 substrate reuse, 131–132 and sulfur, 102 with thiols, 506, 552–553, 565–566, 568, 576, 580, 585–588, 593 GPI-linked aminopeptidase M, 151 G-protein, 401–412, 479 G-protein-coupled receptors (GPCR), 150, 152 gramicidin, 146, 148, 204, 305–310 graphite: basal plane, 26–27, 30, 35 and DNA, 568–569 and dyes, 520 and immunoglobulins, 189 and NADH oxidation, 538, 539 pyrolytic, 18, 35 edge-plane, 6t, 25, 29–30 growth, 44–52 layer-by-layer, 7t, 44–55, 203, 343, 361

622 growth hormone, 197 guanine: and cast films, 32 in DNA, 547 and environmental sensing, 594 and oxidation, 24–25, 52–53 heart muscle, 173 helical peptides, 203–204 Helmholtz capacitance, 75–76, 117 heme: and circular dichroism, 196 and covalent bonding, 30 and cytochrome c, 27 and electron hopping, 19 and hydrogen peroxide, 22 and myoglobin, 27 and prosthetic groups, 83 and trichloroacetic acid, 20 See also myoglobin hemoglobin, 31, 340, 358 a-hemolysin, 147–149 hexanethiol, 91 high performance liquid chromatography (HPLC), 226 high-spatial-resolution, 79 hippocampus, 173 hormones, 197 horseradish peroxidase (HRP): and chemiluminescence, 197–198 and DNA probes, 583 and enzyme reactor, 388–390 and fluorescence microscopy, 172 and glucose oxidation, 29 and immunosensors, 482 and luminescence, 488 and QCM, 359–360 human growth hormone, 197 human serum albumin (HSA), 340, 357 humidity, 223 hybrid bilayers: adsorption, 109, 110–111, 128 and AFM, 129–133, 149, 154 and binding kinetics, 150–153 as biosensors, 143–150

Index [hybrid bilayers] capacitance, 112–114, 132–133, 152–154 and chain length, 115, 120–122 composition, 137–138 and cytochrome c, 110 defects, 118–119, 120–121, 145–146 description, 102–104 electroactive, 145–146 electrochemical analysis, 118–121 electron transfer, 118–121, 129, 146 ellipsometry, 127–128 and enzymes, 144, 145 fabrication, 104, 145, 150, 153–154 formation kinetics, 107–110 impedance, 111–118, 120–121, 132–133, 146, 154 infrared spectroscopy, 121–127, 146 and lipids, 109–110, 122, 125–126, 128, 132–133, 143 molecular layers, 119 molecular structure, 121–127 neutron reflectivity, 134–138, 146 and scattering length density, 140–142 and SPR, 105–111, 141–142, 143–144, 154 and temperature, 123 and thickness, 115, 125–126, 132–133 time factors, 109 and transmembrane proteins, 149–150, 153–155 hydrogel matrix, 180 hydrogen, isotopes of, 134–136, 141 hydrogenase, 26 hydrogen peroxide, 22–24, 482, 483, 488 hydrogen sulfide, 440 hydrophilicity: and adsorption, 197 and cancer cells, 321–322 and gold, 145 and SAMs, 197, 203 and SECM, 321–322

Index hydrophobicity: and AFM, 189 control of, 474 and DNA biosensors, 554–555 in hybrid bilayers, 103, 104–105, 107 and immobilization, 91 and nanoparticles, 443, 445 and protein-binding, 227 and protein-DNA binding, 591 and stability, 38 and vesicles, 128–129 hydroquinone, 180 hydrostatic pressure, 268, 346 imaging, topographical, 298–300, 315–322, 327 immiscible electrolyte solutions (ITIES), 265–267, 272–273, 286 immobilization: of antibodies, 170, 342, 477, 478–479 IgG, 171 and covalence, 144, 469–470 of DNA probes, 549, 552–553, 562–563, 570 of enzymes, 383–384, 400, 405, 418–420, 474, 477 and hydrophobicity, 91 of mediators, 506–523 methods, 144, 469–479, 477, 492 and pH, 182, 340, 342 of protein, 80–82, 182, 340, 342 immunoglobulins: and AFM, 189–191 and ellipsometry, 174–175 and fluorescence, 197, 199–200 immobilization, 171 immobilized, 206–207, 340, 341, 342 and Langmuir-Blodgett, 169 and QCM, 183, 359 See also antibodies immunosensors, 340–341, 463–467, 482–483 separation-free, 482–483, 483, 484 impedance: acoustic, 120–121, 363, 365, 369

623 [impedance] admittance/impedance, 351–356 and cast films, 43 and DNA-based biosensors, 585 in hybrid bilayers, 111–118, 120–121, 132–133, 146, 154 quartz crystal, 349–356 of self-assembled proteins, 367 and thickness, 352, 367 implants, 57, 172 indium tin oxide (ITO), 31, 34, 35 information technology, 382–383, 417–418, 442 infrared spectroscopy, 121–127, 146, 202–204 Fourier transform-, 179, 194, 196 reflection-absorption, 121–127, 203 inosine, 562 insulation, 102–103 insulin, 361 intensity at fixed angle, 221 interfaces: air–water, 34, 167–169, 196, 286–296 dielectric–dielectric, 197 dielectric–metal, 177–178, 215–217, 242–244 and DNA-based biosensors, 554–560, 585 electrode–film, 42 electrolyte–metal, 217–219 liquid–liquid, 265–267, 268, 284–286 liquid–sensor, 346 metal–electrolyte, 217–219 oil–water, 284–286 optical frequency dielectric response, 174 planar, 193–194, 202–203, 208 and SECM, 256–257, 259–260, 263 solid–liquid, 263 and SPR, 177–181 and surfactant, 288–289 internal reflectance spectroscopy, 488–489 iodine, 302–303 p-iodophenol, 488

624 ionic strength, 86, 89 ionomers, 31–32, 52 ion pairs, 346–347 ion transfer, 265–267, 268 See also electron transfer; protons ion transport, counter, 10–11 isoalloxazine, 521 isocyanate, 201–202 isotopes, 134–136, 141 IUPAC convention, 8–9 Kanazawa-Bruckenstein equation, 353 kidney cells, 173 kinetics: of chemically modified electrodes, 523–532 and DNA, 550, 579 electron transfer, and SAMs, 86–91 of hybrid bilayers, 107–110 at interfaces, 259 of protein binding, 150–153, 224–226 of protein uptake, 356–358 and thickness, 526–532 See also Michaelis-Menton kinetics Koutecky-Levich equation, 524 Kramers-Kronig relation, 234–238, 239 Kretschmann geometry: description, 215–216 and SPR, 177–178, 219–220, 241, 243–244 lactate oxidase, 81, 487 Langmuir adsorption, 69, 576–577, 578 Langmuir-Blodgett (LB) films: and cytochrome c, 83 description, 167–169, 278, 477, 478 and enzyme reactors, 383, 387 and gramicidin, 204 and lipid bilayers, 34 and protein films, 338 Langmuir trough, 268, 286–287, 295 large surfactant vesicles, 443 laser beams, 139–142, 199–201 lateral amphiphile diffusion, 291–293

Index lateral proton hopping, 293–296 lauric acid, 35 layer-by-layer adsorption, 339–340, 366–367 and enzyme reactors, 383–393, 397–400 layer-by-layer growth, 44–55 and avidin–biotin poly(amines), 361 (dis)advantages, 7t and enzyme films, 343 and poly(alanine), 203 and thin films, 343 lead, 433, 445–446, 595 leaves, 318 lecithin, 35 lectins, 193 length scale, 165 ligand binding, 178 light: luminescence, 197–198, 488 reflected, 582–583 visible, 486 See also fluorescence linear diode array, 221 linear-sweep voltammetry, 300–302 lipase, 436–437 lipid bilayer membranes, 32–36, 401 lipids: adsorption, 128 and biological membranes, 99–100 charge transport, 40 on electrodes, 32–36 and enzyme reactors, 387, 401, 403–416, 419–420 and Fab0 , 191, 341 in hybrid bilayers, 109–110, 122, 125–126, 128, 132–133, 143 and IgG, 191 and myelin basic protein, 191 organic–inorganic hybrids, 420 and proteins, 227 and SPR, 227 lipopeptides, 144 liposomes: and antibodies, 186, 341–342

Index [liposomes] and DNA-based biosensors, 570–571, 584–585 as membrane model, 101 and nanoparticles, 442 liquid–liquid interfaces, 265–267, 268, 284–286 liquids: Maxwellian, 369 and piezoelectric transducers, 492 QCM, 346, 348–350, 352, 354 liquid–sensor interface, 346 liquid–solid interfaces, 263 logic, 447 See also enzyme reactors Lorentz-Lorenz relation, 217 luciferase, 488 luciferin, 197 luminescence, 197–198, 488 lysine, 89 lysozymes, 340 magnesium, 228, 406, 595 magnetic field, transverse, 215 magnetic microbeads, 590 manganese oxide, 51 Marcus theory, 17–19 mass balance, 456–457, 463, 469–470, 492 mass sensors, 583–585 Maxwell equations, 215–216 Maxwellian fluids, 365, 369 Mediators: alternate, 2 description, 503–508 dual, 315–317 for fuel cells, 535–538 immobilization of, 506–523 lipid-soluble, 321–322 metal complexes, 25–26, 521–523 see also specific metals for NADH oxidation, 503–523, 541 and nucleic acids, 560 redox, 318, 503–523, 560 soluble, 503–504

625 [Mediators] toxic, 318 See also ferrocyanide; ruthenium mediator–substrate reaction, 508 melittin, 137, 152–153 membrane mimetic chemistry, 430 membranes: cellular, 150, 459, 487 gold-coated microporous, 482, 483 thick, 310–317 see also thickness memory storage, 447 menadione, 322 mercaptoalkanoic acids, 205 mercaptoethanol, 118, 131, 154, 568 16-mercaptohexadecanoic acid, 90 mercaptopropane sulfonate (MPS), 45, 49, 350 mercaptopropanoic acid, 590 3-mercaptopropionic acid (3MPA), 78, 359, 438, 440 mercaptopyridine, 75, 80 11-mercaptoundecanoic acid, 85–86, 150 mercury, 26, 561 metabolism, 500–508, 547 metabolism-type biosensors, 458–463, 492 metal complexes, 26, 521–523, 562 metal–dielectric interfaces, 177–178, 215–217, 242–244 metal–electrolyte interface, 217–219 metalloproteins, 36–38, 193–194, 204–205 metal oxides, 49–51 metals: and electron transfer, 26 and enzyme reactors, 402–403, 406–412 and nanoparticles, 296, 431–432, 435–436, 445–446 and SPR, 223 See also specific metals metal–sulfide bonds, 431

626 metal supports, 102–104 and surface plasmons, 104–111 metal wire, 6t, 34 methanol, 538, 539 methemoglobin, 358 methylene blue, 569 micelles, 33, 405, 419–420 reversed, 434–442 Michaelis-Menton kinetics and enzyme reactors, 403–406 and enzymes, 22, 279, 523–524 equation, 500 and glucose, 279 microcantilever sensors, 585–590 microelectrodes, 300–305, 307–310, 326 See also ultramicroelectrodes microfluidic devices, 546–547, 598–599 microorganisms, 535–538 See also bacteria; viruses; specific microorganisms micropatterning, 153–154, 170, 198–199 with SECM, 280–283 microperoxidase-II, 538–539 micropipette electrodes, 303–304 mitochondria, 31–32, 227, 321, 500–502 mitomycin C, 488 molar enthalpy, 492–494 molecules: absorption, 193 aggregation-prone, 387 detection, 164–165 displacement, 13 grafting, 170 neutral, 312, 346–347 orientation, 181–182, 196, 341–342 recognition, 102–103 reorganization, 107 signal, 410–411 small, 2, 387, 481, 503, 549 structure, 140 tectonics, 430 transfer rate, 287–291

Index monoclonal antibodies, 180, 224, 465, 466 monolayers: and AFM, 26–27 and dehydrogenase enzymes, 506–507 kinetic models, 523–526 limitations, 507 nanoparticles, 430–433 and QCM, 346–347 and SECM, 283–296 voltammetry, 26 See also self-assembled monolayers (SAMs) 11-MUDA, 85–86 muscle, 173 mussel-adhesive protein, 358 Mycobacterium tuberculosis, 2, 10 myelin basic protein (MBP), 191 myoglobin: adsorbed films, 26–28 and AFM, 40 and cast surfactant films, 36, 38–43 and covalent binding, 29 and Eastman AQ ionomers, 31 and enzymes, 20, 22, 193 and layering, 47, 49–51 and Marcus transfer, 18 pH, 38, 49 in polyelectrolyte layers, 340 as redox enzyme, 193 and self assembly, 84 NADH-linked reductases, 321 NAD(P)H: See dehydrogenase enzymes Nafion, 30, 43, 52, 539 nanometer level: and DNA-based biosensors, 598 electrodes, 264–265, 267 fabrication, 381–382, 428, 430–433 optical properties, 327–328 resolution, 337 See also enzyme reactors Nanoparticles: adsorption, 445

Index [Nanoparticles] defined, 427 and DNA, 576, 581–582 ferroelectric, 436 and information storage, 447 magnetic, 436 metallic, 49–51, 296, 431–432, 435–436, 436, 445–446 quantum wires, 427–428 self-assembly, 445–447 and semiconductors, 430–432, 443– 446 size control, 435, 440 synthesis: approaches, 428 under monolayers, 430–433 in reversed micelles, 434–442 in surfactant vesicles, 442–445 templates/compartments, 428–429, 436 nanopores, 571–574 naphthalene, 411–412 2-naphthol, 488 napthaquinone, 322 nerve cells, 173 network analysis, 357–358 neutron reflectivity, 134–138, 146, 153, 555 neutron reflectometry, 339 nickel, 406, 522 nickel phthalocyanine, 223, 236–240 nicotinamide adenine nucleotide (NAD), 28–29, 500–502 See also dehydrogenase enzymes nicotinic acetylcholine receptor (NAR), 149–150 nonlinear optical spectroscopy, 139–142 normal pulse voltammetry (NPV), 40 nuclear magnetic resonance (NMR), 202–203, 207–208 nucleic acid films, 4 nucleic acids: and bioluminescence, 488 as biosensors, 468 and DNA

627 [nucleic acids] probe synthesis, 551 versus RNA, 547 immobilization, 552 measurement, 560 oxidation, 24–25 peptide probes, 566–567 recognition, 547–549 See also deoxyribonucleic acid (DNA) octadecanethiol, 133, 135–136, 154 octadecylamine, 433 oil–water interfaces, 284–286 OmpF, 147–149, 151–152 optical biosensors, 455–456, 574–583 optical imaging, 327–328 optical measurement, 373 optical microscopy, 172–174, 188–189 optical properties, 327–328, 371–372 optical spectroscopy, nonlinear, 139–142 optical transducers, 484–489 organosilane, 193–194, 201 oscillation, 354, 489 osmium, 369, 521–522, 565 See also poly(allylamine) osmotic pressure, 268 osteoblasts, 173–174, 201 osteoclasts, 322 osteosarcoma, 172–173 Otto configuration, 220, 241 Overton’s rule, 302 Oxidation: and acoustic impedance, 369 ascorbate, 508 of carbohydrates, 342 and DNA, 32, 52–53, 561–563 guanine, 24–25 and enzyme catalysis, 22–24 of glucose, 29 and gold versus silver, 169 of NADH: applications, 533–540 electrodes, 504–523 kinetic models, 523–532

628 [Oxidation] mediators, 503–523, 541 and SECM, 278–279 and self-assembly, 75, 169 and self-replication, 436 oxygen: and algae, 318–319 and embryos, 322 and SECM, 287–290, 305, 318–319, 322 oxygen transport, 287–290, 305 PAH: See poly(allylamine) paraffin oil, 520 paramagnetic beads, 283 parathion, 492 patch clamp technique, 100–101 patterning, 153–154, 170, 198–199 with SECM, 280–283 pegylation, 199–201 peptide nucleic acid probes, 566–567 peptides, 201 helical, 203–204 See also biomolecular assemblies perchloroethylene, 462 permeability, 259, 301–302, 305–310, 315–317 peroxidase, 340 pertussis, 144 pesticides, 459, 483, 484, 486, 487, 492 pH: and biomolecular films, 38, 49–50 and biosensors, 455, 486 and cast surfactant films, 38 and enzymes, 344–345 and immobilization, 182, 340, 342 and LBL adsorption, 340 and modified electrodes, 86, 88–89 and myoglobin, 38, 49 and NADH oxidation, 521 and phospholipids, 141–142 of phospholipids, 295–296 and polyelectrolytes, 342 and polymeric electrodes, 477 and protein films, 357–358

Index [pH] and protein immobilization, 182, 340, 342 and SAMs, 74, 76–77, 78 and SECM, 267 and SPR, 228 phase separation, 78–79 phenothiazine, 526 phenoxazine, 526 pheochromocytoma, 321 phosphatidylcholine, 40–42, 110–111, 286 and SECM, 295–296, 305 phospholipase, 153 phospholipids: adsorption, 154 buffering, 141–142 and gramicidin, 204 in hybrid bilayers, 102, 104–105, 107, 110–111, 115, 120, 122–125, 141–142, 143, 146, 153–154 hydrogenated, 122 pH of, 295–296 and QCM, 128–129 and SECM, 285, 286 and SPR, 180 photobleaching, 199–201 photochemical reactions, 170 photons, 139–142, 177–178 photoregulation, 412–418 photosynthesis, 318–319, 443 biomimetic, 443–445 phthalocyanine, 223 piezoelectric biosensors, 456–457, 489–492 piezoelectric effects, 218, 346, 349 piezoelectric transducers, 342, 489–492 and liquids, 346, 348–350, 352, 354, 492 See also quartz crystal microbalance (QCM) pinholes, 152 See also defects platinum, 132–133, 325

Index polarization, 526–532 polyacrylamide, 388, 553 poly(acrylate), 521 poly(alanine), 203–204 poly(allylamine) (PAH): and acoustic impedance, 356 and ellipsometry, 371 and glucose oxidase, 344, 350, 371 as model, 339 QCM studies, 359–366, 371 poly(amines), 361 poly(aniline), 477, 482–483, 521 polyanions, 361 polyelectrolytes: and antibodies, 359 and cytochrome c, 340 and enzyme reactors, 383–384 LBL adsorption, 337–342, 366–367 and pH, 342 and proteins, 30 and resistance, 363 and shear modulus, 367–368 sinusoidal perturbation, 369 poly(ethylene glycol), 223, 591 poly(ethylene oxide), 388 poly(ethylenimine) (PEI), 50–52, 388–399 polyion films: and enzymatic reaction, 420 polyion–protein, 6t, 7t, 44–52 polyion–surfactant, 7t, 43 stability, 48 poly(lysine), 30–31, 179–180, 388 polymeric films, 29–32, 351, 507–508, 529 polymers: as biosensors, 468, 482–483 and BVD model, 351 conducting, 477–478, 482–483, 521–522 co-polymers, 521, 553, 564 and DNA, 553, 563–564, 570, 573 and dyes, 520–521 and electron transfer, 29–32 and enzyme reactors, 388–390

629 [polymers] hairy, 361–363 for mediators, 507–508 and nanoparticles, 443, 447 and protein films, 4, 340–342, 362–363, 369 and quartz crystal impedance, 349–351 poly(methyl methacrylate), 388 polynitrostyrene, 351 polypeptide residues, 201 poly(pyrroles): See pyrroles polysaccharides, 387 poly(styrenesulfonate) (PSS): and antibodies, 43, 359 and electroactive protein, 50–51 and enzyme reactors, 388–393, 397–399 and layer-by-layer, 339 and NADH oxidation, 521 poly(vinyl alcohol), 201–202, 388 poly(vinylimidazole), 521–522 poly(vinylpyrrolidone), 388 poly(vinylsulfonate), 521 porosity, 474, 507 nanopores, 571–574 potassium: and bilayer membranes, 303–305, 307–309 and hybrid bilayers, 145, 149, 152 and SECM, 303–305, 307–309 potential: and biosensors, 454–455, 481–482 and cyctochrome c oxidase, 231–232 and DNA, 564–565, 578–579, 590, 593–594 in hybrid bilayers, 112–114 increasing, 17 and NADH oxidation, 530–532, 539–540 in SECM, 259, 260, 261–262, 267, 305 and SPR, 218 potentiometry, 454–455, 481–482 pressure, surface, 168, 288

630 prisms, 177–178, 216–217, 220 rotating, 220–221 promastigote protease, 144 prosthetic groups, 83–84 Protein A, 479 protein–DNA binding, 591–592 protein films: absorbed, 6t cast, polyion, 29–32 history, 3–4 multilayer, 337–342 and QCM, 356–371 self-assembled, 337–342 and ellipsopmetry, 367, 369–371, 371–376 QCM, 356–371 protein G, 401–412, 479 G-protein-coupled receptors, 150, 152 protein receptors, 143–144 proteins: adsorption of, 40–41, 167, 172, 178–180, 203, 356–358, 361–363 binding kinetics, 150–153, 224–226 in biological membranes, 99–100 and cast films, 29–32 cell membrane, 150 charged, 357–358 coagulant, 203 complex, 42–43 conformation changes, 227–228 covalent bonding, 27–29, 82 diffusion, 13 electroactive, 8, 50–51 genetically engineered, 185, 197 GPI-linked, 144 hinge-bending, 29 in hybrid bilayers, 134–138 immobilization, 80–82, 182, 340, 342 in lipid bilayers, 37 and lipids, 227 measurement, 549 membrane-bound, 150–153 mussel-adhesive, 358 and pH, 182 plasma, 203

Index [proteins] reaction center, 1, 42–43 redox, 9, 18 RNA role, 547 and SPR, 227–228 transmembrane, 149–150, 153–155 protein–surfactant films, 33, 39–40 proteoglycans, 193 proteoliposomes, 101 protons: hopping, 293–296 reduction, 296 transfer, 26 4-pyridinethiol, 80, 81 pyroquinoline quinone (PQQ), 538–539 pyrroles: and density, 372 and NADH, 521 and NADH oxidation, 507 and patterning, 280 and polymer electrodes, 478, 564 pyrroloquinoline quinone (PQQ), 28–29 pyruvate oxidase, 145 quality factor, 354 quartz crystal impedance, 349–356 quartz crystal microbalance (QCM): admittance/impedance analysis, 351–356 and antibodies, 183–185, 186, 359–361, 374 and biomolecular assemblies, 183–187 description, 489–490, 491 and DNA, 564, 570–571, 583–585, 586 electrochemical (EQCM), 345 and ellipsometry, 371–373 and energy dissipation, 354, 358 enhancements, 185 and enzyme reactors, 390–391, 395, 396 and enzymes, 356, 371–374 gravimetric error, 374 interfaces, 346

Index [quartz crystal microbalance (QCM)] and layered films, 48–51 and hybrid bilayers, 128–129, 144 of protein films, 356–358 self-assembled, 358–371 with ellipsometry, 371–376 shear-mode resonators, 345–356 and viscoelasticity, 342, 349, 351 quinoid-like forms, 180 quinines: and covalent bonding, 28–29 and dehydrogenase enzymes, 519, 521, 538–539 and polymer electrodes, 478 and SECM, 322 Raman scattering, 204–206 rate, 13–14, 88–89, 508 Butler-Volmer kinetics, 17–18 Rayleigh scatter, 178 reaction centers, 1, 42–43 redox currents, pinhole, 152 redox dyes, 519–524, 535, 538 redox enzymes, 19–24, 26, 28, 82, 499–500 See also dehydrogenase enzymes; specific enzymes redox indicators, 564–569 redox proteins, 9, 18 redox reactions: and cytochrome c, 228, 232, 235–236 efficiency, of protein, 9 and SECM, 256–259, 271 steps, 91 redox species: charged, 145–146 hydrophilic, 145 and melittin, 152 and SECM, 272–273 reduction: and heme enzymes, 22 peaks, 5–9 sequential, 20 two-electron, 20

631 reflectance spectroscopy, 87 reflection, attentuated total (ATR), 203 reflection absorption infrared spectroscopy (RAIRS), 121–123, 203 polarization-modulation, 123–127 reflection interference contrast, 173 reflectivity, 174, 217, 240–241 neutron, 134–138, 146, 153, 555 refractive indices, 177–178, 371–373 reorganization, dynamic, 102–103 reorganization energies, 90, 91 resistance, 361–363, 366, 369 resonance angle, 217–218 respiration, 318–319, 321 Rhodobacter sphaeroides, 1, 42–43 rhodopsin, 150, 152, 227–228 ribonucleic acid (RNA), 547, 594–597 ribozymes, 594–597 rotating-disk, 22, 528–529 rotating prism, 220–221 roughness: of electrodes, 45–48 and neutron reflectivity, 135–136 root-mean-square, 131 of surfaces, 154, 194, 204–205 ruthenium: and DNA, 25, 52–53, 562, 567, 570 and fluorescence, 486–488 and NADH oxidation, 522 pinhole currents, 152–153 and SECM, 292, 301–302, 314, 321–322 Saccharomyces cerevisiae, 535 Salmonella typhimurium, 492 SAMs: See self-assembled monolayers sapphire, 137–138 Sauerbrey equation: and QCM, 356, 357, 365 admittance/impedance, 353 and shear mode resonators, 346 and thickness, 373 scanning angle reflectometry (SAR), 338, 339

632 scanning electrochemical microscopy (SECM): and adsorption, 260 with AFM, 323–327 alternating direction implicit finitedifference method (ADIF-DM), 269–270 and antibodies, 279–280 applications, 254, 259, 276–283, 322–328 bilayer studies, 297–310 and biomolecular assemblies, 207, 276–283 of cells, 268–269, 317–322 and concentration, 255, 259 convolution, 264 double potential step chronamperometry, 260, 276, 290–291 dual mediators, 315–317 and electrodes, 255, 264–265, 267, 300–305 and electron transfer, 265–267, 272–273, 284–286, 292–293, 300–310 and enzymes, 255, 276–280, 321 feedback modes, 255–259, 271–273, 277–280, 283, 321–322 and ferrocyanide, 285, 298, 302, 312–313, 318–319 general-collection, 277–278 induced transfer (SECMIT), 259, 273–276, 314–315 and interfaces, 256–257, 259–260, 263 and micropatterning, 280–283 monolayer studies, 284–296 optical imaging, 327–328 probes, 264–268, 270 sensitivity, 278 theory, 269–271 and thick membranes, 310–317 time factors, 275–276, 291–292 tips: collection imaging, 278

Index [scanning electrochemical microscopy (SECM)] control of, 260–262 and oxidation, 278 and substrate, 191, 268, 271, 275, 300, 305, 315–317, 323–327 uses, 255–256, 280–283 topographical imaging, 298–300, 315–322, 327 two-phase, 270 scanning electron microscopy, 207, 322, 390, 587 scanning near field optical microscopy (SNOM), 327–328 scanning probe microscopies, 165, 187–192 and DNA-based biosensors, 556–557, 585–590 See also atomic force microscopy scanning tunneling microscopy (STM), 253–254, 557–558 scattering length density (SLD), 134–138, 140–142 Schiff’s base, 403–412, 413 screen printing, 475, 480, 569 SECM: See scanning electrochemical microscopy selenium, 433, 446 self-assembled enzyme films, 342–345, 356 self-assembled monolayers (SAMs): and AFM, 189 amino-terminated, 201 and attenuated total reflection, 204 and carboxylic acid, 76–77, 80–81 and cells, 172–173 and chain length, 71, 73, 74, 76, 77–79, 81, 87–89 and covalent bonding, 27–29, 82 defect sites, 119 description, 477 and DNA: biosensors, 552–553, 555, 562, 568, 574, 590 therapeutic use, 594

Index [self-assembled monolayers (SAMs)] electrochemistry, 73–76 and electron transfer, 86–91 electrostatically assembled, 185 endgroup modification, 76–77 and enzyme reactors, 383 film structure, 71–73 formation, 67–71 functional groups, 203 and hydrophilicity, 197, 203 hydroxyl- versus methyl-terminated, 172–173 and infrared spectroscopy, 203–204 mixed monolayers, 77–79 organosilane, 193–194 and osteosarcoma, 172–173 porosity, 78 and protein–DNA binding, 591–592 and Raman scattering, 205 and SPR, 230–232 structural studies, 83–86 uses, 65–68 voltammetric studies, 79–82 self-assembled protein films, 337–342 and ellipsometry, 367, 369–371, 371–376 and QCM, 356–371 self-assembly: advantages, 169, 447 and DNA, 556, 569 electrostatic, 170 and enzymes, 81–83, 340, 356, 367 and nanoparticles, 445–447 of surfactants, 36, 67 and thickness, 71, 75–76 self-exchange, 19, 39, 47 self-replication, 436–442 semiconductors: and cells, 173 and DNA probes, 553–554, 598 and enzyme reactors, 420 and nanoparticles, 430–432, 443–446 See also silicon sensor–liquid interface, 346 serine, 27

633 shear force topographical imaging, 327 shear-mode resonators, 345–356 shear wave, 342, 357, 362–363, 373 signal-to-noise ratio, 52 signal transduction: See biosensors; enzyme reactors silane, 194 silicon, 553–554, 580–581, 583, 585–590 silicon crystal, 125, 137–138 silicon oxide, 49–51, 128–129, 169, 193–194 silver: and cytochrome c, 84–86 and DNA, 582 and enzyme reactors, 406, 411 and NADH oxidation, 522 and oxidation, 169 and Raman scattering, 205 and SECM, 267, 301, 304, 310 See also surface plasmon resonance single-nucleotide polymorphisms (SNPs), 545–546, 550–551, 567–568 skin, 312 sodium, 145, 152, 307 solid–liquid interfaces, 263 solutes: and SECM, 259, 311–317 transport, 310–317 uncharged, 305–310 solution-borne components, 178 solutions, immiscible electrolyte (ITIES), 265–267, 272–273, 286 solvents: drag, 310 fluxes of, 346–347 and self-assembly, 69 spatial ordering, 342–343, 383 spatial resolution, 79 SPR: See surface plasmon resonance squalane/butanol, 34 square wave voltammetry: description, 14–17 and DNA damage, 52, 54

634 [square wave voltammetry] and Eastman AQ ionomers, 31 and layering, 47–48 sensitivity, 52 stability: of biosensors, 469–470, 473 and carboxyl groups, 25 of cast surfactant films, 38, 48 of COX, 149 and DNA-based biosensors, 563–564 of enzyme reactors, 393–395, 416 in hybrid bilayers, 103, 141 and hydrophobicity, 38 layered films, 48 and SAM-modified electrodes, 74 of surfactant vesicles, 443 stagnant-point models, 305–310 Stark effect, 218, 236–240 stearic acid, 294–295 steroid cyclophanes, 407 streaming potential measurement (SPM), 339 streptavidin, 144, 590 See also biotin-streptavidin stringency, 549, 579–580 structure/function, 99–100, 165 styrine oxide, 32, 52, 54–55 submarine electrodes, 261, 264, 268–269, 295 submicrometer probes, 264–265 substrate–enzyme complex, 20–21 substrate–mediator reaction, 508 sulfide semiconductor films, 430–433 sulfur, 26, 31–32, 102, 341 sum frequency generation, 139–142 surface acoustic wave (SAW), 363, 491 surface concentration, 12, 109 surface-enhanced Raman scattering (SERS), 205–206 surface-enhanced resonance (SERRS), 84, 85–86 surface free energy, 154 surface plasmon resonance (SPR): absorption spectroscopy, 240–241

Index [surface plasmon resonance (SPR)] and adsorption, 218 with AFM, 230 analyte gradient-, 226 applications, 213–228, 232–236, 241–242 bi-cell detection, 221–222, 244–247 and biomolecular assemblies, 175–182, 185–187 and biosensors, 182, 223, 455–456, 457, 489 DNA-based, 552, 574–580 with cyclic voltammetry, 229–232 and cytochrome c, 226, 230–232 electrochemical, 217–219, 228–232, 241–242 multiwavelength, 232–236, 241 experimental setups, 219–223, 244–247 and ferrocyanide, 229–230 field-enhanced fluorescence, 578 fundamentals, 214–217, 242–244 grating-based, 220 and hybrid layers, 105–111, 141–142, 143–144, 154 and interfaces, 177–181 Kretschmann geometry, 177–178, 219–220, 241, 243–244 and pH, 228 and protein–DNA binding, 591 sensitivity enhancement, 182 Stark spectroscopy, 236–240 and thickness, 217–219 surface pressure, 168, 288 surfaces: carbon, 281–283 and cell spreading, 173–174 and DNA biosensors, 552–560, 590 hydrophobic, 189 micropatterned, 153–154, 170, 198–199, 280–283 optical properties, 327–328 roughened, 154, 194, 204–205 SECM substrates, 271–272, 315–317 surfactant–protein films, 33, 39–40

Index surfactants: casting, 36 chain length, 33 double-chain, 33, 36 on electrodes, 32–36 head group charge, 42 and nanoparticles, 435, 438, 442–445 and oxygen transfer, 288–289 packing parameter, 33 and self-assembly, 36, 67 uses, 7t, 9 volume, 33 surfactant vesicles, 442–445 synchrotron X-ray, 433 synergy, 153 temperature: and biosensors, 457, 459, 463, 467 calorimetric transduction, 492–494 and cast lipid films, 40 cryovoltammetry, 26 and DNA, 549, 552, 553, 555 and electron transfer, 86, 91, 286 and enzyme reactors, 393–395 gel-to-liquid, 122–123 and hybrid bilayers, 123 and reorganization energy, 91 terpyridine complexes, 522 tetanus, 144 tetraethyl orthotitanate, 522 THEO-C18, 135–137 thermal reactions, 170 thermistors, 493–494 thick-film technology, 475–477 thick membranes, 310–317 thickness: acoustic, 187 of alkane chains, 137 calculating, 174 and charge, 528 and current, 13 and diffusion control, 12–13 and DNA-based biosensors, 582–583 effective, 373 and electron transfer, 310–317

635 [thickness] and ellipsometry, 206, 367, 369–371, 373 and enzyme films, 367–368 and frequency shift, 390–391 and hybrid bilayers, 115, 125–126, 132–133 and impedance, 352, 367 kinetic models, 526–532 layering, 44–52, 507–508 and LBL adsorption, 339–340 measurement, 128, 276, 371–373 and NADH oxidation, 525–532, 534t and oxidation, 369–371 of phosphatidylcholine films, 40 of protein films, 338 and resistance, 369 Sauerbrey mass, 373 of self-assembled proteins, 367–369 and self-assembly, 71, 75–76 shear-mode resonators, 345–356 and SPR, 217–219 ultrathin, 337, 446–447 of unstirred layers, 307 and viscoelasticity, 368–369 and water, 363, 369, 371, 373 thickness-shear mode (TSM) devices, 357 thiol: alkyl-, 6t, 68–69 and DNA-based sensors, 551 hexane-, 91 octadecane-, 133, 135–136, 154 pyridine-, 80, 81 trithiolamine, 170 See also alkanethiol; gold thiophenol, 75 thioredoxins, 35 thiotic acid, 340–341, 364 time factors: adsorbed saturation, 48 and current, 275–276 and hybrid bilayers, 109 incubation, 2, 52, 55

636 [time factors] and SECM, 275–276, 291–292 tin, 523, 562 indium oxide, 31, 34, 35 titanium, 201, 520 toluene, 52, 55, 223, 303–304 topographical imaging, 298–300, 315–322 shear force, 327 topographic labels, 590 total internal reflection fluorescence (TIRF), 167, 197 toxicity screens, 4, 52–55, 143–150, 594 toxins, 43, 144, 488, 594 transducers, modified, 474–479 transduction methods: calorimetric, 492–494 electrochemical, 479–483 optical, 484–489 piezoelectric, 484–489 see also liquids transfer function method, 354–356 transmission line model (TLM), 351 transport systems, 102–103 transverse magnetic field, 242–243 trichloroacetic acid, 20–21, 31 trimethylchlorosilane, 303–304 trimethylsiloxanes, 199 trioctanoyl glycerol (TOG), 436–437 triple potential step chronoamperometry (TPSC), 260 trithiolamine, 170 tryptophan, 197 ultrafiltration, 395–400 ultramicroelectrodes (UMEs): and amperometry, 261, 264 and amphiphiles, 291–292 C-fiber-based, 321 liquid membrane ion-selective, 267 nanometer-scale, 265, 267 for optical properties, 327–328 positioning, 263 role of, 253, 256, 260

Index [ultramicroelectrodes (UMEs)] with small radii, 312 submarine, 261, 268–269, 295 TMPD-generating, 321 ultrathin films, 337, 446–447 unstirred layers, 305–310 urate oxidase, 145 urea, 280, 308, 455 urine, 145 van der Waals forces, 141, 445 van’t Hoff-Arrhenius relation, 572–573 vapor detection, 223 vesicles, 128–129, 141–142, 154, 442–445 vibrational bands, 139–142 vibrational spectroscopy, 202–206 vinculin, 173–174 viruses, 193, 224 viscoelasticity: and avidin–biotin, 361 and dissipation energy, 128 interfacial, 48–49 and PAH, 364 and piezoelectric devices, 349 of polymer films, 351 and QCM admittance/impedance, 351, 352–356 and Sauerbrey equation, 357 and shear-wave, 342 and thickness, 368–369 viscosity: and acoustic impedance, 363 and electron transfer, 86–89 and QCM, 351–353, 585 voltammetry: cryovoltammetry, 26 cyclic: see cyclic voltammetry diffusion-controlled, 12–13, 39 and DNA, 561, 562, 564, 565 history of, 3 of hybrid bilayers, 119–120 linear-sweep, 300–302 monolayer, 26 normal pulse, 40 rotating-disk, 22

Index [voltammetry] of SAM/protein films, 79–86 square wave: see square wave voltammetry and thin monolayers, 26 wagging progressions, 122–123 water: content of, 30, 371 and enzyme reactors, 388, 393, 404 and hybrid bilayers, 134–138 and nanoparticles, 434–435, 438, 440, 443–445 in protein films, 340, 357, 358, 361–363, 366–367, 371 and shear wave, 373 and solvent drag, 310 and thickness, 369, 371, 373

637 water–air interface, 34, 167–169, 196, 286–296 water–oil interface, 284–286 waveguide linear dichroism, 194 wavelength modulation, 222–223 wheat germ agglutinin, 193 width, peak, 10 wire, cleaved, 6t, 34 X-ray, synchrotron, 433 X-ray photoelectron spectroscopy (XPS), 201–201, 206–207, 554 x–y degenerate oscillator, 194, 195 zinc, 228, 284–286, 406, 445–446 zirconium, 520, 522