You are on page 1of 553

ISBN: 0-8247-0731-1p

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: 41-61-261-8482; fax: 41-61-261-8896
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 2002 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, microlming, 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

Electroanalytical Methods
for Biological Materials

edited by

Anna Brajter-Toth
University of Florida
Gainesville, Florida

James Q. Chambers
University of Tennessee
Knoxville, Tennessee

Marcel Dekker, Inc.

New York Basel

TM

Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

Preface

These are exciting times in science! The facile electronic communication between
laboratories and the increased access to information have accelerated the pace of
advancement of knowledge. More than ever before, science has become a collective effort of many individuals, with common and cross purposes, in a wide variety of laboratories throughout the world. Research areas and elds intersect, and
with the increased ability to communicate, cross-fertilization of ideas and
methodology has become easier and more prevalent than in the past. In these
times, succinct reviews of major advances in science, written by the researchers
who have led the advance, become more important and useful than ever before.
This change in the way science is done and the increased pace is certainly
evident in the area of science concerned with understanding electrical phenomena
in biological systems. Here is a eld of science that dramatically cuts across several disciplines, as reected in the way it is referred to by its practitioners: bioelectrochemistry by electrochemists, biophysics by physicists, bioelectroanalytical chemistry by electroanalytical chemists, and so forth. It is signicant that the
understanding of electrical phenomena originated with things biological, in the
discoveries of Galvani and Volta, more than two centuries ago. Although advances in the 19th and early 20th centuries were few, studies of electrical phenomena in nerves by Emil Du Bois-Reymond are a notable exception. Since then

iii

iv

Preface

this fascinating eld of science has been contributing to the understanding of such
diverse biological processes as energy production and DNA functions.
The impact of the advances in this eld of science concerned with electrical phenomena in biological processes can be documented by the Nobel prizes
awarded in this area. Among these are the prizes in medicine in 1936 to Dale and
Loewi, who proved the chemical basis of neurotransmitter release; in 1963 to
Hodgkin and Huxley for the sodium-potassium ion pump model of nerve impulses; in 1970 to Katz, von Euler, and Axelrod for mechanisms of humoral transmission in nerve cells; in 1978 to Mitchell for his chemiosmotic theory of the
electrochemical membrane gradient that drives ATP synthesis; in 1991 to Neher
and Sakmann for understanding of the function of single-ion channels; in 1997 in
chemistry to Skou for discovery of the ion-transporting enzyme Na+,K+-ATPase,
and continuing in 2000 in medicine, to Carlsson, Greengard, and Randel for signal transduction in the nervous system.
This timeline of accomplishment is continued in the chapters of the present
compilation. Mark Wightman and Andrew Ewing have a direct link to the most
important accomplishments, with their in vivo electrochemistry work described
in this book. This electroanalytical approach to the real time measurement of neurotransmitters in living cells of the brain, which was pioneered in the laboratory
of R. N. Adams, is a technique that works and is approaching a state of maturity.
It is this kind of methodology that will allow future scientists, regardless of their
mother eld, to continue to unravel Natures mysteries. In recent years charge
transport along the pi stack of the DNA double helix has been demonstrated,
notably in Bartons laboratory at Cal Tech, and its role in biology hotly debated.
Shana Kelley, who has played a major role in the work from Bartons laboratory,
reviews and puts this topic into perspective. Regardless of the importance of
DNA charge transport pathways, the development of electroanalytical polynucleotide hybridization sensors promises to be a foundation of future sensing technology. Joe Wang, who has pioneered in this area, reviews various approaches
and strategies for these sensors; and Willner and Katz summarize the newest
results from their laboratory in Jerusalem.
The use of electroanalytical methodology has yielded signicant insight
into the workings of redox enzymes. An important section of the present compilation is devoted to the voltammetry of redox enzymes from the laboratories of
Fraser Armstrong, Lo Gorton, Fred Hawkridge, and Jim Rusling, all major players in this area. These chapters describe the subtleties of the electrical and chemical requirements for facile communication between the redox centers of the
enzymes and the electrode transducers. The power of the simple voltammetric
experiment, in the hands of experts, is abundantly evident in these reviews. Especially interesting is the role of cell membrane mimic layers that point back to
a fundamental understanding of how electrical information is transferred in
biology.

Preface

Without the newest bioelectroanalytical methods, the solutions to problems


that have to be addressed to prevent and limit human disease may be too long in
coming. The chapters on bioelectroanalysis include descriptions of the most
advanced and successful methods, which in large part build on the fundamentals
described in the earlier chapters. The pioneering work on electrochemical immunoassays is described by its inventor, Bill Heineman. Werner Kuhr describes the
future of sensor technologies, Steve Weber summarizes the strategies that work
in the determinations of amino acids and peptides, many of which were developed
by his group, and Jim Cox summarizes the challenges in nding and developing
new catalysts for bioanalysis. The advances that have been made in complex biological determinations are apparent in the chapters by Adam Heller and his
coworkers on the development of glucose and other in vivo sensors, by Bob
Kennedy, and by the Lunte group, who describe powerful combinations of modern separations and newest electroanalytical methods in the analysis of extremely
small and complex biological samples. These last chapters face the wide horizons
of the future developments that seem so close to realization. We hope that you
will nd useful new information in this book, and we look forward to the new
developments that the work described in this book is likely to inspire.
Anna Brajter-Toth
James Q. Chambers

Contents

Preface
Contributors
Part I

iii
ix

Electrochemistry of DNA

1. Charge Migration Through the DNA Double Helix


Shana O. Kelley
2. Electrochemical DNA Biosensors
Joseph Wang
3. Amplied and Specic Electronic Transduction of DNA Sensing
Processes in Monolayer and Thin-Film Assemblies
Itamar Willner, Eugenii Katz, and Bilha Willner

1
27

43

Part II Protein Electrochemistry


4. Direct Electrochemistry of Proteins and Enzymes at Electrodes
James D. Burgess and Fred M. Hawkridge

109

5. Voltammetric Investigations of Iron-Sulfur Clusters in Proteins


Fraser A. Armstrong

143

vii

viii

6. Polyion and Surfactant Films on Electrodes


for Protein Electrochemistry
James F. Rusling and Zhe Zhang
7. Electrochemistry of Peroxidases
Tautgirdas Ruzgas, Annika Lindgren, Lo Gorton,
Hans-Jrgen Hecht, Joachim Reichelt, and Ursula Bilitewski

Contents

195
233

Part III In Vivo Electrochemistry


8. Mechanisms and Kinetics of Neurotransmission Measured
in Brain Slices with Cyclic Voltammetry
Joshua D. Joseph and R. Mark Wightman

255

9. Electrochemical Monitoring of Exocytosis from Individual


PC12 Cells in Culture
Leslie A. Sombers and Andrew G. Ewing

279

Part IV Bioelectroanalysis
10. Milestones of Electrochemical Immunoassay at Cincinnati
C. Ajith Wijayawardhana, H. Brian Halsall, and
William R. Heineman

329

11. Electrochemical Detection of Peptides


Eskil Sahlin, Amy T. Beisler, Stephen G. Weber,
and Mats Sandberg

367

12. Microfabrication of Electrode Surfaces for Biosensors


Steven E. Rosenwald and Werner G. Kuhr

399

13. Electrocatalytic Determination of Biochemical Compounds


James A. Cox and Long Cheng

417

Part V Biological Applications


14. Electrodes Based on the Electrical Wiring of Enzymes
Charles N. Campbell, Adam Heller, Daren J. Caruana,
and David W. Schmidtke
15. Capillary Electrophoresis/Electrochemistry: Instrument
Design and Bioanalytical Applications
Susan M. Lunte, R. Scott Martin, and Craig E. Lunte
16. Ultrahigh Sensitivity Analysis of Amino Acids and Peptides
by Capillary Liquid Chromatography with
Electrochemical Detection
Brendan W. Boyd and Robert T. Kennedy
Index

439

461

491

523

Contributors

Fraser A. Armstrong Inorganic Chemistry Laboratory, Oxford University,


Oxford, England
Amy T. Beisler Department of Chemistry, University of Pittsburgh, Pittsburgh,
Pennsylvania
Ursula Bilitewski Department of Biochemical Engineering, German Research
Center for Biotechnology (GBF), Braunschweig, Germany
Brendan W. Boyd Department of Chemistry, University of Florida, Gainesville, Florida
James D. Burgess Department of Chemistry, Case Western Reserve University, Cleveland, Ohio
Charles N. Campbell Department of Chemical Engineering and Texas Materials Institute, University of Texas, Austin, Texas
Daren J. Caruana Department of Chemistry, University College London,
London, England
ix

Contributors

Long Cheng Department of Chemistry, Miami University, Oxford, Ohio


James A. Cox Department of Chemistry, Miami University, Oxford, Ohio
Andrew G. Ewing Department of Chemistry, Pennsylvania State University,
University Park, Pennsylvania
Lo Gorton Department of Analytical Chemistry, Lund University, Lund,
Sweden
H. Brian Halsall Department of Chemistry, University of Cincinnati, Cincinnati, Ohio
Fred M. Hawkridge Department of Chemistry, Virginia Commonwealth University, Richmond, Virginia
Hans-Jrgen Hecht Department of Structural Research, German Research
Center for Biotechnology (GBF), Braunschweig, Germany
William R. Heineman Department of Chemistry, University of Cincinnati,
Cincinnati, Ohio
Adam Heller Department of Chemical Engineering and Texas Materials Institute, University of Texas, Austin, Texas
Joshua D. Joseph Department of Chemistry, University of North Carolina at
Chapel Hill, Chapel Hill, North Carolina
Eugenii Katz Department of Organic Chemistry, Hebrew University of Jerusalem, Jerusalem, Israel
Shana O. Kelley Department of Chemistry, Merkert Chemistry Center, Boston
College, Chestnut Hill, Massachusetts
Robert T. Kennedy Department of Chemistry, University of Florida, Gainesville, Florida
Werner G. Kuhr Department of Chemistry, University of California, Riverside, Riverside, California
Annika Lindgren Department of Analytical Chemistry, Lund University,
Lund, Sweden

Contributors

xi

Craig E. Lunte Department of Chemistry, University of Kansas, Lawrence,


Kansas
Susan M. Lunte Department of Pharmaceutical Chemistry, University of
Kansas, Lawrence, Kansas
R. Scott Martin Department of Chemistry, University of Kansas, Lawrence,
Kansas
Joachim Reichelt Department of Structural Research, German Research Center for Biotechnology (GBF), Braunschweig, Germany
Steven E. Rosenwald Department of Cell Biology and Neuroscience, University of California, Riverside, Riverside, California
James F. Rusling Departments of Chemistry and Pharmacology, University of
Connecticut, Storrs, Connecticut
Tautgirdas Ruzgas Department of Analytical Chemistry, Lund University,
Lund, Sweden
Eskil Sahlin Department of Chemistry, University of Pittsburgh, Pittsburgh,
Pennsylvania
Mats Sandberg Department of Chemistry, Gteborg University, Gteborg,
Sweden
David W. Schmidtke School of Chemical Engineering and Materials Science,
University of Oklahoma, Norman, Oklahoma
Leslie A. Sombers Department of Chemistry, Pennsylvania State University,
University Park, Pennsylvania
Joseph Wang Department of Chemistry, New Mexico State University, Las
Cruces, New Mexico
Stephen G. Weber Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania
R. Mark Wightman Department of Chemistry, University of North Carolina
at Chapel Hill, Chapel Hill, North Carolina

xii

Contributors

C. Ajith Wijayawardhana Department of Chemistry, University of Cincinnati, Cincinnati, Ohio


Bilha Willner Department of Organic Chemistry, Hebrew University of Jerusalem, Jerusalem, Israel
Itamar Willner Department of Organic Chemistry, Hebrew University of Jerusalem, Jerusalem, Israel
Zhe Zhang Department of Chemistry, University of Connecticut, Storrs, Connecticut

1
Charge Migration Through the
DNA Double Helix
Shana O. Kelley
Boston College, Chestnut Hill, Massachusetts

I.

INTRODUCTION

The charge-transport properties of the DNA double helix have intrigued chemists,
physicists, and biologists essentially since the structural features of this molecule
were revealed over 40 years ago [1]. The stack of aromatic heterocycles within
the double helix allows the readout of genomic information through the display
of functional groups within the grooves of this molecule. The striking similarity
of the -stacked array of DNA bases to -stacked solid-state conductors has
prompted the suggestion that DNA might efciently facilitate charge transport
[2]. This intriguing proposal, along with relevance of charge migration in DNA
to biological function and biosensing, has prompted the examination of this phenomenon from many different scientic perspectives.
Given the central role of DNA in cellular function, the dynamics and distance
dependence of electron-transfer reactions proceeding through this medium have
important biological ramications. Both damage to DNA bases [3,4] and repair of
base lesions [5] can result from the reaction of radicals with DNA. Therefore, the
extent of charge migration through DNA would determine whether only localized
reactions affect the integrity of genomic information or whether chemistry initiated
from a distance might also play a role in DNA damage and repair.
Moreover, with the recent progress in decoding the human genome and
identifying diseases that result from genetic errors, inexpensive and accurate
methods to read out DNA sequence information are needed so that genetic testing can become a more integral part of medical care. The characterization of the
1

Kelley

charge-transport properties of DNA provides the basis for the development of a


new class of biosensors. By monitoring DNA-mediated electron-transfer reactions sensitive to sequence or structural perturbations, genetic information could
be extracted from DNA samples using methods that may possess advantages over
conventional assays [6].
The aim of this chapter is to describe recent results that illustrate the salient
features of electron-transfer reactions mediated by DNA and, in particular, the
implications that these results have for DNA-based biosensing. The unique structural features of DNA strongly inuence electron-transfer reactions proceeding
through this medium. Moreover, structural perturbations drastically attenuate
reactivity, and, in particular, disruptions in base stacking within the DNA helix
signicantly decrease the efciency of electron transfer. The sensitivity of DNAmediated electron transfer to base stacking has been exploited in the development
of an electrochemical assay for point mutations that may have applications in
genetic testing.

II.

EVIDENCE FOR LONG-RANGE ELECTRON


TRANSFER IN DNA

Less than a decade after the structural features of the DNA double helix were elucidated, the rst experiments directed at understanding charge migration in DNA
were initiated [2,7,8]. In 1961, inspired by the structural resemblance of DNA to
conductive one-dimensional crystals, Eley and Spivey measured bulk conductivity of dry DNA and observed high levels of electron mobility [2]. These results
were both contradicted and conrmed by later experiments performed by different laboratories [7,8]. In the ensuing 40 years, the extent of DNA-mediated
charge migration has been studied by a variety of experimental methods, and
many different conclusions concerning the efciency of charge transfer through
DNA have been drawn. For a detailed account of the history of the eld of DNAmediated electron transfer, which has developed considerably over the past 10
years, the reader is referred to other reviews [911].
In the past ve years, a signicant number of experimental studies substantiate Eley and Spiveys original claim that the DNA base stack could promote
long-range charge migration [1227]. In well-dened assemblies containing
covalently constrained photoactive molecules associated with the DNA bases
through either intercalation or stacking interactions, a cohesive body of experimental results supports the notion that long-range redox reactions are facilitated
by the double helix of DNA. In addition, recent measurements of the electrical
conductivity of DNA molecules offer another line of evidence that this material
has conductive properties.

Charge Migration Through the DNA Double Helix

A.

Photophysical Studies of DNA-Mediated


Electron Transfer

The study of photoinduced electron transfer between molecular donors and


acceptors provides a means to assess the electronic coupling provided by the
DNA helix. Early applications of this method to DNA-mediated reactions utilized
reactants noncovalently bound to DNA [2830] These studies provided qualitative information concerning the efciency and distance dependence of electron
transfer, but the ambiguity associated with random distributions of reactants
along the DNA helix precluded a quantitative analysis. Once chemical methods
were developed for the covalent attachment of donors and acceptors to DNA [31],
the distance dependence of these reactions, as well as the effects of perturbations
within the intervening medium, could be systematically studied. The identication of unnatural DNA bases with appropriate photophysical and redox properties
has also helped to dene the extent of electronic coupling provided by the base
stack [16].

1.

Photoinduced Electron Transfer Between Intercalators

In DNA duplexes covalently derivatized with ethidium (Et), a classic organic


intercalator, and Rh(phi)2bpy3+, a metallointercalator, reductive electron-transfer
quenching (Figure 1A) was monitored as a function of distance and intervening
stacking [12]. The photoexcitation of ethidium generates a singlet excited state
with sufcient energy (Eo (Et2+/*Et) ~ 0.9 V vs. NHE) to reduce Rh(phi)2bpy3+,
which has a low-energy, ligand-centered reduction (Eo (Rh(phi)2bpy/Rh(phi)
(phi)bpy = 0.03 V vs. NHE). The loss of *Et uorescence in the presence of
Rh(phi)2bpy3+ was measured to monitor the efciency of electron transfer in the
intercalator-modied duplexes.
A series of DNA duplexes ranging from 10 to 14 base pairs in length were
prepared with the intercalators attached to opposite 5 ends through short aliphatic
linkers. Despite the long distances (1735 ) that separated the intercalators, uorescence quenching occurring on a subnanosecond timescale was detected [12].
The yield of this electron-transfer reaction decreased with increasing
donor-acceptor distance, but interestingly, the rate of the reaction was not greatly
attenuated. Thus, the distance dependence of this quenching reaction does
not solely reect changes in electron-transfer rate and is therefore not purely
a measurement of (the decay of electronic coupling) [32]. The distance dependencies of reactions exhibiting this behavior are represented by the symbol .
For the DNA-mediated reductive electron transfer reaction between *Et and
Rh(phi)2bpy3+, = 0.1 1.
To test the response of this long-range electron-transfer reaction to perturbations in the intervening medium, and conrm that the pathway for the reaction

Figure 1 Reactions used to study electron transfer in DNA with photophysical methods.
(A) Photoinduced reduction of Rh(phi)2bpy3+ by *Et [12]. (B) Triplet energy transfer (TET,
a double electron transfer) from *Ru(phen)(bpy)(Me2-dppz)2+ to Os(phen)(bpy) (Me2dppz)2+ [13]. (C) Photoinduced oxidation of Z by *Et [1415]. (D) Photoinduced oxidation
of G by adenine analogues (*Ax [16]).

Charge Migration Through the DNA Double Helix

did proceed through the base stack of the double helix, intercalator-derivatized
assemblies were engineered to include single-base mismatches. Base mismatches
of certain compositions cause subtle disruptions in base stacking without affecting the overall structure of DNA [33,34]. Indeed, the introduction of an intervening CA mismatch into an 11-base pair duplex signicantly attenuated the
electron-transfer reaction between *Et and Rh(phi)2bpy3+ (Figure 2A). An intervening GA mismatch, a pair that is well stacked due to its large aromatic surface
area and stable hydrogen bonding, slightly increased the efciency of the reaction. These experiments revealed the sensitivity of DNA-mediated electron transfer to base stacking and highlighted how the unique properties of a molecule dominated by noncovalent interactions modulate electronic coupling.
The inuence of base stacking on a DNA-mediated reaction was also considered in interpreting the electron-transfer distance dependence. The decreasing
yield of electron transfer over a distance range of 17 to 35 , without signicant
changes in rate, indicated that a process slower than the electron transfer might
gate the reaction. Given the identication of base stacking as a parameter
affecting the efciency of DNA-mediated electron transfer, base dynamics were
suggested as a source of this behavior. Because noncovalent forces stabilize
the double helix, breathing dynamics occur on time scales ranging from seconds
to picoseconds [35,36]. If these dynamics produced transient disruptions in stacking, the distance dependence observed could reect the increasing probability
of a stacking disruption resulting from the introduction of more intervening
bases. Therefore, stacking interactions within the DNA helix are not only critical
for long-range reactivity, but they could dictate the distance range of charge
migration.
Long-range triplet energy transfer between intercalators. Studies of triplet
energy transfer between two metallointercalators, Os(phen)(bpy)(Me2-dppz)2+
and Ru(phen)(bpy)(Me2-dppz)2+, also demonstrated the reliance of long-range
charge transfer in DNA on stacking (Figure 1B) [13]. In a mechanism that
involves a double electron transfer, emission from the Ru(II) complex is
quenched through the generation of an excited Os(II) complex. Luminescence
quenching was detected over distances up to 44 with = 0.1 1, and the presence of single-base bulges that interrupted base stacking signicantly attenuated
this quenching (Figure 2B). Systematic variation of the chirality of the donor and
acceptor revealed the importance of reactant intercalation. Stereoisomers with
right-handed chirality are complementary to the right-handed helix of DNA and
therefore intercalate more deeply into the base stack; indeed, reactions employing -enantiomers displayed more efcient energy transfer than those executed
with -isomers that did not intercalate as deeply. In addition, much lower reactions yields were observed when an analogue of Os(bpy)32+, which cannot intercalate, was used as an acceptor. These results illustrate that the stacking of reactants with the DNA bases is critical for efcient charge transfer, as is the stacking
of bases in the intervening pathway.

Figure 2 Experiments elucidating sensitivity of photoinduced DNA-mediated electron


transfer reactions to base stacking. (A) Photoinduced electron transfer between *Et and
Rh(phi)2bpy3+ is attenuated by the presence of an intervening CA mismatch [12]. (B)
Yields of triplet energy transfer between Os(phen)(bpy)(Me2-dppz)2+ and *Ru(phen)(bpy)
(Me2-dppz)2+ decrease in the presence of bulges [13]. (C) The efciency of the photooxidation of Z by *Et decreases when Z is incorporated in a mismatch (X = A,T, G) [14]. (D)
Base-base electron transfer is more efcient when initiated by *A2, an adenine analogue
that can stack within the double helix, compared to *A, an analogue that is sterically bulky
and not well stacked within DNA [16].

Charge Migration Through the DNA Double Helix

2. Photoinduced Electron Transfer Between an Intercalator


and a DNA Base
The guanine analogue 7-deazaguanine (Z) was used to investigate whether DNA
bases could participate in long-range reactions mediated by the -stack [14]. Z is
more easily oxidized than any of the other bases and therefore can selectively
reduce the singlet excited state of Et (Fig. 1C). To test the effect of distance and
reactant stacking on the photooxidation of Z by *Et, a series of DNA duplexes
were engineered that featured tethered Et and Z incorporated at a range of sites.
For reactant separations spanning 627 , subnanosecond electron transfer
occurred from Z to photoexcited Et. The distance dependence was affected by the
identity of the anking bases surrounding the electron donor. The incorporation
of Z between two purine bases resulted in a distance dependence ( = 0.2 1) that
was more shallow than that for a pyrimidine/purine anking sequence ( = 0.4
1). Stacking is generally enhanced by the larger purines; therefore, this effect
may originate from better stacking for Z within the purine/purine site.
The inuence of reactant stacking was evident in experiments varying the
identity of the base paired with Z. Incorporation of any base other than the natural mate, G, in the complementary strand signicantly decreased the efciency of
electron transfer (Figure 2C). When the ability of a reactant to stack within the
DNA helix is compromised, it appears that long-range reactivity is limited.
Recent studies of the dynamics of electron-transfer reaction between *Et
and Z conducted with femtosecond resolution [15] provided further evidence that
rates of DNA-mediated electron transfers are not greatly attenuated by distance.
Biphasic kinetics were detected with time constants of 5 ps and 75 ps. Only the
absolute amplitudes of the electron-transfer rates measured in this high-resolution
study changed as a function of distance, supporting the idea that the distance
dependencies of some DNA-mediated processes include contributions from
events (e.g., base dynamics) occurring on a slower timescale.
The biphasic kinetics were attributed to the existence of two *Et populations with subtly different orientations. Measurement of the orientation anisotropy of *Et revealed that the slower kinetic component corresponded to a population of molecules that underwent reorientational motion before reacting,
possibly because better stacking was required. These results illustrate that intricate and dynamic structural factors may control DNA-mediated reactions.
3.

Electron Transfer Between DNA Bases

To directly probe charge-transfer reactions between two DNA bases, two uorescent analogues of adenine, 2-aminopurine (A2) and 1,N6-ethenoadenine (A),
were employed as photooxidants (Figure 1D) [16]. In aqueous solution, both of
these bases are efciently quenched by deazaguanine (Z) and guanine (G), with
only small amounts of quenching observed with inosine (I).

Kelley

DNA-mediated photooxidations of G by *A2 and *A were examined in a


series of 12-bp DNA duplexes in which donor/acceptor separations ranged from
3.4 to 13.6 . The efciencies of the reactions of the two uorescent bases incorporated within DNA duplexes were signicantly different. Although both bases
were highly quenched by G at the short donor-acceptor distances, only *A2
showed electron-transfer quenching at separations >10 . In addition to exhibiting different distance dependences, the kinetics of the reactions were also distinct.
At 3.4 , where donor and acceptor are located at adjacent positions within the
DNA helix, *A2 undergoes fast electron transfer (kET 1010 s1), whereas *A
undergoes electron transfer on a slower timescale (kET ~ 109 s1). Electron transfer between *A and G was reected in measurable rate changes, thus the distance
dependence of the *A/G reaction provides a measure of ; based upon steadystate quantum yields, a value of 1.0(1) 1 was obtained for this parameter. The
quenching reaction between *A2 and G occurs on a timescale faster than that predicted by a Stern-Volmer analysis; thus, the distance dependence of this reaction
( = 0.4 1) is not purely a reection of and may include a contribution from
the destacking defects caused by base dynamics.
These two uorescent adenine analogues displayed comparable reactivity
in solution, but once incorporated into DNA, they were not equally reactive. An
analysis of existing structures of duplexes containing A2 and A revealed a difference between the two derivatives: stacking within the DNA helix. As A is
sterically bulky and does not form a base pair, it is poorly stacked within the double helix. A2 undergoes normal Watson-Crick pairing with thymine and is stacked
within the DNA helix quite similarly to the natural bases. These results provide
another example of the role of reactant stacking in facilitating coupling into the
base stack (Figure 2D).
B.

Long-Range Redox Chemistry Mediated


by the DNA Base Stack

In biological systems, radical and redox reactions can inict deleterious chemical
modication of natural bases [3,4]. Redox reactions are also important for the
repair of some base lesions [5]. The generation of such chemical charge-transfer
products by photoreactants separated spatially from the site of modication
allows the assessment of charge migration facilitated by the DNA double helix.
Indeed, studies monitoring long-range chemical reactions on DNA revealed that
charge migration can occur over signicant molecular distances, and that this
reactivity is sensitive to structural distortions in DNA.
1.

Long-Range Guanine Oxidation in DNA

One pathway for the formation of 8-oxoguanine, a mutagenic derivative of guanine, involves the oxidation of this base by one electron, followed by the reaction
of the guanine radical cation with water or O2. The presence of 8-oxo-G can be

Charge Migration Through the DNA Double Helix

visualized by the treatment of radiolabeled DNA with base followed by gel electrophoresis. The generation of 8-oxo-G at a site remote from a tethered reactant
was rst demonstrated in DNA duplex assemblies derivatized with Rh(phi)2bpy3+
[17]. Upon photoactivation, this metallointercalator generates lesions at the 5 G
of GG doublets; these base steps appear to be thermodynamically favored for oxidation because lowered potentials result from mixing of adjacent HOMOs [38].
The damage of guanine doublets proceeds over distances as long as 200 [19],
and, like the DNA-mediated reactions discussed above, it is sensitive to disruption of the base stack by bulges (Figure 3A) [18] and protein binding [37].
Similar results have been obtained with other reactants, including anthraquinone derivatives that also act as photooxidants [20,21]. The intercalation properties of these DNA-binders can be manipulated through the derivatization of the
ring structure. Anthraquinone species that intercalate poorly, as established by
NMR and other methods of physical characterization, do not affect long-range
guanine oxidation [21]. Derivatives of anthraquinones that do stack within the
DNA helix do oxidize guanine doublets [20]. These reactions have been found to
occur at distances up to 150 from the site of intercalation [11].

Figure 3 Sensitivity of DNA-mediated long-range chemical reactions to base stacking.


(A) The long-range oxidation of distal guanine doublets is decreased in the presence of
intervening bulges disrupting base stacking [18]. (B) Long-range repair of thymine dimers
is attenuated in the presence of bulges [22].

10

2.

Kelley

Long-Range Repair of Thymine Dimers in DNA

Thymine dimers are mutagenic photoproducts formed at contiguous thymines. In


bacterial systems they are repaired by reductive electron transfer in a reaction catalyzed by DNA photolyase [5]. Molecular photooxidants can also repair these
lesions, and can do so from remote positions within DNA assemblies. Longrange, oxidative repair of thymine dimers was investigated in synthetic DNA
duplexes containing a thymine dimer positioned systematically along the helix
and a photooxidant, Rh(phi)2bpy3+, tethered to one of the termini [22]. Following
irradiation of the modied duplex, the extent of T <> T repair was detected by
HPLC. These experiments showed that signicant amounts of T <> T repair
occurred over distances of as much as 25 . The ability of the DNA helix to mediate this reaction is contingent on a fully stacked pathway, as the introduction of
bulged regions into the intervening bases diminished reactivity (Figure 3B).
The oxidation of guanine and repair of thymine dimers over long molecular distances provide further evidence that the double helix, when its -stack is
accessed by species incorporated within the base stack, can efciently mediate
long-range charge migration. The distances over which these chemical reactions
can be detected are much longer than those over which the photoinduced reactions monitored in real time proceed. The mechanisms of charge transfer or transport (i.e., hopping versus direct superexchange-type reactions) may differ. Moreover, the timescales on which these reactions are monitored and the energetics of
the reactants used as probes with respect to the DNA bridge are signicantly different. Unifying mechanistic and theoretical descriptions of DNA-mediated
charge migration are being formulated [39] and will be essential in understanding
the different types of long-range reactions that proceed though this important biomolecule.
C.

Measurements of the Electrical Conductivity of DNA

Direct measurements of the electrical conductivity of DNA molecules allow comparisons with more conventional materials through which charge transport has
been well characterized. Early studies of DNA conductivity utilized bulk or dry
DNA samples and provided conicting assessments of the extent of electron
mobility [2,7,8]. However, recent studies of single molecules or well-dened
DNA lms have revealed signicant levels of conductivity [26,27].
In oriented DNA lms obtained by casting aligned DNA-lipid complexes,
anisotropic conductivity was detected [26]. When the DNA was aligned perpendicular to the electrodes (with the directionality dened by the helical axis), levels of conductivity measured were comparable to conducting polymers. However,
lms in which the DNA strands were aligned parallel to the electrodes supported
only small currents.
Electrical currents recently measured with a low-energy electron point
source microscope across DNA ropes 600 nm in length were also similar to

Charge Migration Through the DNA Double Helix

11

those expected for conducting polymers [27]. A separate study of electrical transport between two nanoelectrodes bridged by DNA molecules, however, revealed
much lower levels of electron mobility, prompting the comparison of the conduction properties of DNA to those of large-band gap semiconductors [40]. Thus,
there is not complete agreement on whether DNA is an effective conductor, but
many results do support the hypothesis that electrical current can be channeled
through DNA under certain conditions.
III.

PARAMETERS MODULATING ELECTRON-TRANSFER


REACTIVITY IN DNA

In a variety of systems, electron transfer between species stacked within the DNA
helix proceeds over exceptionally long distances. These reactions can also occur
on ultrafast timescales. Studies revealing the charge-transport properties of DNA
have also illustrated the importance of a collection of parameters that modulate
electron transfer in ways that are unique to this system. The effects of distance,
sequence, and base stacking that can be deduced from recent studies of DNAmediated charge transport are summarized below.
A.

Distance

In studies employing a wide range of reactants, distance dependencies for a variety of electron-transfer reactions proceeding through the DNA helix have been
obtained [1216,4143]. The most striking aspect of this collection of results is
that different distance dependencies are measured for every system studied, and
the connection between reaction kinetics and distance dependence has not been
straightforward. While it is expected that the measurement of electron transfer
through one medium should yield a single distance dependence, several have
now been measured for DNA, and they have differed by a full order of magnitude.
Now that parameters other than distance have been systematically varied, we nd
that different dependencies on donor/acceptor separation can be obtained in
chemically identical systems that differ only in the stacking or energetics of reactants [16]. Hence, it appears that the DNA helix can exhibit behavior ranging
from that of an insulator to that of a semiconductor. The main difference between
the systems yielding these distance dependencies is the structural and energetic
properties of the reactants. Thus, the properties of a molecular probe for studying
DNA-mediated electron transfer may inuence the behavior observed.
B.

Sequence

The presence of four different monomers within the -stack of DNA leads to
questions concerning how sensitive electron transfer through this medium is to
sequence composition. Although it is not known at present whether one can detect

12

Kelley

a difference in the dynamics for electron transfer through an A-T as compared to


a G-C base pair, anking sequence does appear to impact the thermodynamics of
a reactant incorporated within the base stack [14,38]. Indeed, studies of chemical
oxidation reactions of the DNA bases also elucidated that sequence context
strongly modulates the oxidation potential of a given site [38]. This effect highlights the interactions between adjacent base steps and may provide natural systems with a means of concentrating base damage at thermodynamic sinks to simplify repair mechanisms.
C.

Stacking

In many systems investigated to date, the behavior of the DNA helix as a bridge
and as a reactant in electron-transfer processes has been contingent upon stacking
interactions within this structure. Base mismatches and bulges markedly decrease
the efciency of reactions between donors and acceptors coupled through the
DNA -stack [1214,16,18,22]. The incorporation of reactants within base mismatches also causes dramatic decreases in electron-transfer yields [14,16]. The
observation of a proportion of unquenched reactants in many systems led to the
proposal that base dynamics within the double helix produce a nite population
of properly stacked duplexes or reactants that were able to undergo ultrafast electron transfer [12,14,15]; the remaining unquenched population was hypothesized
to be improperly congured for the reaction. The overall distance dependence of
DNA-mediated reactions appears to be strongly affected by reactant stacking.
Reactants within anking sequences composed of varying proportions of purines
and pyrimidines exhibit different distance trends [14]; a more shallow dependence is observed for a site where stronger stacking interactions would be predicted. Even more striking is the example of electron-transfer reactions in two
analogous sets of duplex assemblies differing only in the stacking of the acceptor
(as deduced from NMR structures) where distance dependencies differing by a
full order of magnitude are observed [16]. These results underscore that the stacking interactions of reactants with the DNA bases are critical for access to a pathway facilitating fast, long-range electron transfer. Moreover, studies in the literature that have reported different distance dependencies can be reconciled by
concluding that systems employing reactants that do not interact strongly with the
base stack tend to exhibit the most pronounced sensitivity to distance [42].
IV.

HARNESSING DNA-MEDIATED ELECTRON TRANSFER


FOR BIOSENSING

The development of sensitive, high-throughput DNA sensors is an essential technological goal for the utilization of newly acquired genomic information (see
other chapters in this volume for different approaches to electrochemical DNA

Charge Migration Through the DNA Double Helix

13

biosensing). The ability to locate single base mutations will facilitate the identication of genetic diseases, cancer, and polymorphisms that may inuence prescribed routes of medical treatment. To achieve this goal, most assays, including
those employing the DNA arrays available commercially, rely on the detection of
uorescence changes caused by differential hybridization of fully complementary
versus mutated DNA fragments [6]. The identication of point mutations in this
manner presents a difcult challenge; on an array of oligonucleotides of diverse
sequence composition, the difference in base-pairing energy among immobilized
sequences is often greater than that brought about by a point mutation within a
given sequence.
An approach to the detection of point mutations in genomic DNA based on
DNA-mediated electron transfer can be envisioned as an alternative to existing
methods. These long-range reactions are signicantly affected by the changes in
stacking brought about by single-base mismatches, so small perturbations in
sequence could potentially be detected as mismatches within hybridized duplexes.
A strategy of this nature might offer heightened sensitivity and would eliminate the
need to customize hybridization conditions to the specic gene or mutation of
interest.
To explore the utility of DNA-mediated electron-transfer events for biosensing applications, an electrochemical assay was developed [2325]. Electrochemical sensors are typically more portable and inexpensive than those using uorescence-based detection, hence a system suitable for monitoring electron
transfer through DNA duplexes immobilized on the surface of an electrode was
sought.
The construction of self-assembled monolayers of small molecules immobilized on solid surfaces has facilitated the study of electron transport through a variety of molecular systems [44]. In particular, the adsorption of aliphatic, thiolcontaining compounds on gold electrodes constitutes a powerful means of constructing high-density lms. The construction of DNA monolayers was approached with this precedent in mind. Despite the highly charged and bulky structure of DNA duplexes, molecules derivatized with a sulfur-containing linker form
densely packed monolayers on gold surfaces (Figure 4) [23,45]. This section
describes the characterization of these materials, the features of DNA-mediated
electron-transfer reactions through such lms, and strategies for the detection of
single-base mutations in this microenvironment.
A.

Fabrication and Characterization


of DNA-Modied Surfaces

DNA duplexes can be quantitatively modied with an alkanethiol linker at the 5


terminus through a combination of solid-phase and solution-phase methods, and
deposited on gold surfaces [45]. Electrochemical assays, radioactive tagging
experiments, and atomic force microscopy (AFM) all indicate that the derivatiza-

14

Kelley

Figure 4 Schematic illustration of an intercalator (dark gray) bound to a DNA monolayer immobilized on gold (top). Linkage between gold surface and DNA (bottom).

tion of gold surfaces with these DNA duplexes yield densely packed monolayers
oriented in an upright position with respect to the metal surface. Simple tests of
surface coverage as deduced from the electrochemical signals of anionic molecules (which would be repelled by negatively charged duplexes) at DNAmodied electrodes qualitatively indicated high levels of adsorption, as did the
analysis of gold surface waves of derivatized electrodes compared with underivatized electrodes [23]. More quantitative information was obtained in 32P tagging

Charge Migration Through the DNA Double Helix

15

experiments. Given the cross-sectional area of canonical B-form DNA (3.1 1012
cm2), and the number of radioactive molecules immobilized within a given area as
revealed by scintillation counting, a surface coverage of 4 1011 mol/cm2 was calculated. This value corresponds to a close-packed fractional coverage of 75%.
The surface morphology of these DNA lms was further characterized
using AFM [45]. Although the existence of long-range order within these lms in
aqueous solution could not be assayed, the monolayers appeared generally homogeneous and dense. In the absence of an electrochemical potential, height-contrast
measurements revealed a monolayer thickness of 45 . For a uniform lm, this
thickness suggests an orientation of the helical axis with respect to the gold surface of 45; this orientation would lead to packing consistent with that indicated
by radioactive tagging.
The application of an electrochemical potential generated reversible
changes in monolayer thickness that revealed features unique to these polyanionic
lms. At applied potentials inducing a negative surface charge, the monolayer
thickness increased to 65 . This value approaches that expected if the linkermodied DNA duplexes were oriented at a 90 angle with respect to the metal surface. Likewise, in monolayers with low surface coverage, the application of positive potentials caused compression of the lm to 20 . The agreement between this
observed thickness and the diameter of DNA (20 ) indicates that the duplexes lie
at under these conditions. Therefore, it appears that the surface morphology is
affected by electrostatic forces originating from the metal surface, but importantly,
in the presence of a negative surface potential, the DNA duplexes assume an
upright orientation and are not in direct contact with the electrode.
The features of the double-stranded DNA lms are strikingly different from
those of monolayers formed from single-stranded oligonucleotides, where the
individual DNA strands lie down at on the gold surface even in the absence of
an electrostatic potential [46]. The deposition of double-stranded DNA likely
decreases the adsorption of base functionalities to the electrode. The regular, helical structure of DNA may also promote dense packing of the monolayer resembling that found in crystals of this material. Therefore, the lms formed with
duplex DNA offer well-dened environments with which to explore DNA-mediated reactions; they also have the potential to be used for DNA biosensing.
B.

Electrochemistry of Molecules Bound to DNA Films

The rst studies of redox reactions occurring within lms composed of DNA
duplexes were rst conducted with methylene blue (MB) [23], an intercalating
probe that exhibits a low-energy, reversible reduction at 0.25 V vs. SCE. Pronounced electrochemical signals exhibiting the features expected for a surfacebound species were observed at low intercalator concentrations, indicating that
MB bound with high afnity to the modied electrode surface. Quantitation of the

16

Kelley

binding of MB using chronocoulometry yielded a micromolar binding constant


consistent with that observed in solution, but evaluation of the number of sites
available to this probe revealed a unique feature of the immobilized DNA. While
nearest-neighbor exclusion would dictate that 7 MB molecules could bind per 15
base pair duplex, only ~1.5 intercalators were bound to each immobilized DNA
molecule. This nding indicated that the dense packing of the monolayer prohibited diffusion of MB into the interior of the lm and constrained binding to the
most solvent accessible sites. When the duplexes were presaturated with MB
before deposition, much higher currents were measured, conrming that adsorption of the intercalator to a preformed DNA lm limited the number of available
binding sites to those most solvent accessible at the periphery of the lm.
The electron-transfer kinetics corresponding to the reduction of MB bound
to the DNA lm were surprisingly fast. The rate observed corresponded to that
expected for tunneling through the aliphatic tether linking the DNA molecules to
the metal surface; the rate did not reect a signicant contribution for electron
transfer through DNA. Given the hypothesis that MB occupied the sites within
the monolayer farthest from the electrode, the suggestion was made that efcient
electron transfer through the DNA base stack was operative in this system. However, the lack of unambiguous information concerning the binding site of the
intercalator prevented strong conclusions from being drawn.

C.

Long-Range Electron Transfer Through DNA Films

In order to systematically study the long-range reduction of an intercalator bound


to DNA at an electrode surface, experiments employing a cross-linked probe were
initiated [24]. Daunomycin, a redox-active intercalator, can be cross-linked to
adjacent GC base steps, and thus the binding site of this probe was rationally varied through manipulation of the sequence of immobilized duplexes (Figure 5). A
series of lms containing DM cross-linked to different sites resulting in intercalator-electrode separations that spanned 30 did not exhibit detectable changes
in the electron-transfer rate. The morphology of the various lms was again characterized by AFM, and the duplexes appeared to assume an upright orientation
with respect to the metal surface.
Although not signicantly distance dependent, electron transfer through
DNA duplexes immobilized on an electrode surface is highly sensitive to the
presence of base mismatches. Indeed, the change of one base, which introduced
a CA mispair between the electrode and the site of intercalation, switched off the
electrochemical response entirely (Figure 6). This striking sensitivity to basepairing conrmed that interior of the DNA helix constituted the electron-transfer
pathway. Moreover, this experiment served as proof-of-principle that single base
substitutions could be detected in an assay based on DNA-mediated charge transfer at an electrode surface.

Charge Migration Through the DNA Double Helix

17

Figure 5 Series of DM-modied duplexes constructed to evaluate long-range electron


transfer through DNA lms. (Adapted from Ref. 24.)

Figure 6 The presence of a CA mismatch between electrode and intercalator prohibits


the electrochemical reduction of crosslinked DM (Ref. 24.)

18

D.

Kelley

Development of an Electrochemical Assay


for DNA Point Mutations

In order to exploit the efciency and sensitivity of DNA-mediated redox reactions


in biosensing applications, further improvements in the methodology were
required to demonstrate the feasibility of this approach. A practical adaptation of
the system would have three features: (1) the ability to use a non-cross-linked
redox probe, (2) the ability to detect point mutations within oligonucleotides of
varying sequence composition, and (3) the ability to achieve in situ hybridization
at the electrode surface. The success obtained in each of these areas [25] indicates
that this approach may hold promising applications.
1.

Variation of DNA-Binding Redox-Active Probe

The evaluation of electrochemical signals obtained for a variety of redox probes


revealed that strong intercalation was an important property of a probe that
reported on the presence of point mutations [25]. Four different redox-active molecules, MB, DM, Ir(bpy)(phen)(phi)3+, and Ru(NH3)5Cl2+, were bound noncovalently to DNA lms, and currents measured at electrodes derivatized with fully
base-paired duplexes were compared to currents measured at electrodes derivatized with duplexes containing a single mutation producing a CA mismatch. Each
of these probes, with the exception of Ru(NH3)5Cl2+, binds to DNA by intercalation. The electrochemical signals obtained with the intercalators accurately
reported the presence of a CA mispair: the currents measured decreased at least
threefold (Figure 7) when the immobilized duplex contained a mispair. However,
the electrochemical signals observed for the groove-bound probe, Ru(NH3)5Cl2+,
were identical at both types of lms. It appears that direct access to the base stack
is an important requirement for reporting perturbations within this structure.
2.

Variation of Base Composition and Mismatch Position

The detection of a base mismatch by the intercalating probe DM was not highly
sensitive to the position of the mismatch within the lm or to the composition of
the DNA duplex composing the lm [25]. Sequences featuring mismatches at different locations and mismatched within different sequence contexts were evaluated; accurate discrimination of fully paired versus mismatched sequences was
achieved in all cases, although the incorporation of the mismatch at a position
deep within the interior of the monolayer yielded the highest TA/CA current
ratios.
The presence of point mutations giving rise to CA mismatches was detected
in sequences containing from 13% to 100% GC base pairs under identical conditions [25]. This feature highlights an important advantage of this approach over
hybridization-based methods. Because the detection strategy described here

Figure 7 The electrochemical response of DM noncovalently bound to a DNA lm decreases when the immobilized duplexes contain a CA mismatch. (Adapted from Ref. 25.)

Charge Migration Through the DNA Double Helix


19

20

Kelley

allows intact duplexes to be assayed for base substitutions, the inherent thermodynamic properties of sequences of interest do not affect the results obtained.
This feature should be of utility in the adaptation of an approach based on DNAmediated electron transfer to a sequence array.
3.

Variation of Mismatch

The range of mismatches that would result from genetic mutations requires an
assay that is sensitive to mispairs of varying compositions. The electrochemical
response of DM noncovalently bound to DNA-modied electrodes reported the
presence of almost all types of mismatches [25]. In general, purine-pyrimidine
and pyrimidine-pyrimidine mispairs (CA, TT, CC, GT, CT) caused the most pronounced attenuations in the current generated by the reduction of DM. The one
purine-purine mispair studied, GA, could not be detected. Photophysical measurements of the effects of GA mispairs on long-range electron transfer through
DNA also revealed an insensitivity to this sequence perturbation, likely due to the
preservation of stacking interactions by this pair with increased aromatic surface
area.
The differences in the thermodynamic destabilization introduced by various mismatches did not dictate the electrochemical response obtained at the corresponding lms. For example, a GT mispair that caused a 6 decrease in the Tm
of a 15-mer duplex gave the same attenuation in integrated current for the reduction of DM as a CA mispair that caused a 12 decrease in the Tm of the same
duplex. Both perturbations in the orientations of bases involved in mispairs and
increased base dynamics for wobble base pairs may disrupt stacking and attenuate rates of electron transfer. This attenuation may allow for the detection of mismatches with varying thermodynamic stabilities.
4.

In Situ Hybridization of Sequences Containing Point Mutations

The detection of point mutations in an oligonucleotide array generated by the deposition of double-stranded sequences would require a way to generate singlestranded probes and a way to hybridize target molecules to single-stranded DNA
on the electrode. The DNA lms described here are amenable to this manipulation [25]. Experiments monitoring the reduction of DM revealed that the heat
denaturation of the lms produced signals distinct from duplex-containing lms
(signals for surfaces presenting single-stranded DNA were less reversible and
broadened). Upon hybridization of a sequence containing a point mutation to the
single strand on the electrode, the electrochemical currents were attenuated to levels comparable to those generated with the direct deposition of mismatched
duplexes. Electrodes could be cycled through the hybridization process repeatedly and reproducibly.

Charge Migration Through the DNA Double Helix

21

The observation of distinguishable voltammograms for lms composed of


double-stranded DNA versus lms composed of single-stranded DNA offers further conrmation that the approach described does not measure hybridization but
instead reports on the presence of perturbations within intact duplexes. Moreover,
the achievement of in situ hybridization demonstrates that double-stranded
duplexes can be initially deposited to ensure high levels of surface coverage without precluding the generation of surfaces presenting single-stranded DNA that
provide the capability to assay samples of interest.

5.

Electrocatalysis at DNA-Modied Electrodes

To increase the inherent sensitivity of the mismatch detection assay, the direct
through-lm charge transport to noncovalent intercalators was coupled with an
electrocatalytic cycle involving a non-intercalating substrate in solution (Figure
8) [25]. The resulting signals imparted enhanced selectivity and sensitivity to this
assay.
The chemical reduction of Fe(CN)63 was identied as a reaction that would
be catalyzed by electrochemically reduced MB intercalated within a DNA lm.
This electrocatalytic process is thermodynamically favored by 0.6 eV. Indeed, in
the absence of MB, no signal is obtained for the anionic probe at a DNAmodied electrode, presumably because of electrostatic repulsion. In addition,
the association of MB with the DNA lm produces currents corresponding to the
reduction of 1.4 molecules per immobilized duplex (Figure 8). However, when
both MB and Fe(CN)63 are present in solution, irreversible electrochemical signals corresponding to multiple turnovers of the intercalated catalyst are observed.
Only low levels of electrocatalysis were obtained when daunomycin was used as
the intercalated species; this probe binds DNA with higher afnity and may have
slower exchange dynamics or solvent accessibility.
The incorporation of an electrocatalytic event affords better discrimination
of point mutations. Electrocatalytic signals detected with lms containing mismatched duplexes relative to fully paired duplexes reported the presence of the
base substitution with much larger signal differentials than signals obtained with
direct electrochemistry (Figure 9). Moreover, the use of chronocoulometry allows
the variation in signals obtained in the presence of matched versus mismatched
samples to be amplied over time, thereby greatly increasing the sensitivity of the
assay.
Electrocatalytic detection of base mismatches also requires an intercalated
species to shuttle electrons. Ru(NH3)5Cl2+ effectively promoted the catalytic
reduction of ferricyanide, but the electrocatalysis was not sensitive to the presence of a mismatched base pair. Therefore, access to the base stack again determines the ability of a reporter molecule to diagnose sequence perturbations.

(B)

Figure 8 (A) Electrocatalytic reduction of Fe(CN6)3 by MB intercalated within a DNA lm. (B) Comparison of signals obtained at
DNA lms for direct electrochemistry of MB (dotted line) versus electrocatalysis (solid line). (Adapted from Ref. 25.)

(A)

22
Kelley

Charge Migration Through the DNA Double Helix

23

Figure 9 Detection of a CA mismatch with electrocatalysis at a DNA lm. (Adapted


from Ref. 25.)

6.

Other Approaches and Future Challenges

Other methods have been developed that allow the detection of point mutations
in non-hybridization based assays. An effective alternative electrochemical
method harnesses differences in the kinetics of the reaction of Ru(bpy)32+ with
guanine in the context of base mismatches to report base substitutions [47]. In
addition, altered patterns of chemical reactivity have been detected in RNA-DNA
hybrids containing 2-NH2 modications in an RNA complement at mispaired
positions [48]. Extension of this approach to an immobilized system has not yet
been demonstrated, but it may hold promise for any applications where alterations
in chemical, rather than electrochemical, reactivity are required.
A biosensing approach based on attenuated electrochemical signals brought
about by perturbations in base stacking requires more exploration of the parameters essential in a practical assay. The detection of very low levels of a target
sequence obtained from a biological sample must be demonstrated. This approach
must be extended to the interrogation of an array of sequences immobilized on a
DNA chip with electrochemical capabilities. Moreover, the sensitivity of this
assay must be tested in samples of broadly heterogeneous genetic composition.
The results obtained thus far are promising, and indicate that irrespective of the

24

Kelley

debate surrounding the efciency of DNA-mediated charge transport, fundamental studies of this phenomenon have elucidated unique features of reactions mediated by the base stack that may be exploited in a new class of DNA diagnostics.
ACKNOWLEDGMENTS
Many of the systems described in this chapter were developed during the authors
thesis work conducted in the laboratory of Prof. Jacqueline Barton. The development of DNA lms as a tool for the investigation of DNA-mediated electron
transfer and for novel DNA biosensors is an ongoing collaboration effort of the
research group of Prof. Michael G. Hill at Occidental College and the Barton
Group at the California Institute of Technology. The author thanks Dr. R. Erik
Holmlin and Dr. Karla Ewalt for reading the manuscript of this chapter.
ABBREVIATIONS
Et
Phi
Phen
Bpy
Dppz
Z
I
A2
A
AFM
MB
DM

ethidium
9,10-phenanthrene-quinone diimine
1,10-phenanthroline
bipyridine3
dipyridophenazine
7-deazaguanine
inosine
2-aminopurine
1,N6-ethenoadenine
atomic force microscopy
methylene blue
daunomycin

REFERENCES
1.
2.
3.
4.

5.
6.

J. Watson, F. Crick, Nature, 171, 737 (1953).


D. D. Eley, D. I. Spivey, Trans. Faraday Soc., 58, 411 (1961).
S. Steenken, Chem. Rev., 89, 503 (1989).
J. Cadet in DNA Adducts: Identication and Signicance (K. Hemminki, A. Dipple,
D. E. G. Shuker, F. F. Kadlubar, D. Segerback, and H. Bartsch, eds.), IARC Publications, Lyon (1994).
A. Sancar and G. B. Sancar, Ann. Rev. Biochem., 57, 29 (1988).
M. Chee, R. Yang, A Hubbel Berno, X. C. Huang, D. Stern, J. Winkler, D. J. Lockhart, M. S. Morris, S. P. A. Fodor. Science, 274, 610 (1996).

Charge Migration Through the DNA Double Helix


7.
8.
9.
10.
11.
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.

25

C. Y. Liang, E. G. Scalo, J. Chem. Phys., 40, 919 (1964).


R. S. Snart, Biopolymers, 6, 293 (1968).
R. E. Holmlin, P. J. Dandliker, J. K. Barton, Angew. Chem. Int. Ed., 36, 2714 (1997).
S. O. Kelley, J. K. Barton in Metal Ions in Biological Systems, 36 (Eds.: A. Sigel, H.
Sigel), Marcel Dekker, New York, 211 (1999).
G. B. Schuster, Acc. Chem. Res., 33, 253 (2000).
S. O. Kelley, R. E. Holmlin, E. D. A. Stemp, and J. K. Barton, J. Am. Chem. Soc.,
119, 9861 (1997).
R. E. Holmlin, R. T. Tong, J. K. Barton, J. Am. Chem. Soc., 120, 9724 (1998).
S. O. Kelley, J. K. Barton, Chem. & Biol., 8, 413 (1997).
C. Wan, T. Fiebig, S. O. Kelley, C. R. Treadway, J. K. Barton, A. Zewail, Proc. Natl.
Acad. Sci. U.S.A., 96, 6014 (1999).
S. O. Kelley, J. K. Barton, Science, 283, 375 (1999).
D. B. Hall, R. E. Holmlin, and J. K. Barton, Nature, 382, 731 (1996).
D. B. Hall and J. K. Barton, J. Am. Chem. Soc., 119, 5045 (1997).
M. E. Nunez, D. B. Hall, J. K. Barton, Chem. Biol., 6, 85 (1998).
D. Ly, Y. Kan, B. Armitage, G. B. Schuster. J. Am. Chem. Soc., 118, 8747 (1996).
D. T. Breslin, C. Yu, D. Ly, G. B. Schuster, Biochemistry, 36, 10463 (1997).
P. J. Dandliker, R. E. Holmlin, and J. K. Barton, Science, 275, 1465 (1997).
S. O. Kelley, J. K. Barton, N. M. Jackson, and M. G. Hill, Bioconj. Chem. 8, 31
(1997).
S. O. Kelley, N. M. Jackson, M. G. Hill, J. K. Barton, Angew. Chem., 38, 941 (1999).
S. O. Kelley, E. M. Boon, J. K. Barton, N. M. Jackson, M. G. Hill, Nucl. Acids Res.,
27, 4830 (1999).
Y. Okahata, T. Kobayashi, K. Tanaka, M. Shimomura, J. Am. Chem. Soc., 120, 6165
(1998).
H.-W. Fink, C. Schenenberger, Nature, 398, 407 (1999).
D. Purugganan, C. V. Kumar, N. J. Turro, and J. K. Barton, Science, 241, 1645
(1986).
L. M. Davis, J. D. Harvey, and B. C. Baguley, Chem.-Biol. Interactions, 62, 45
(1987).
M. R. Arkin, E. D. A. Stemp, R. E. Holmlin, J. K. Barton, A. Hrmann, E. J. C.
Olson, and P. A. Barbara, Science, 273, 475 (1996).
R. E. Holmlin, P. J. Dandliker, J. K. Barton, Bioconj. Chem., 10, 1122 (1999).
R. A. Marcus and N. Sutin, Biochim. Biophys. Acta, 811, 265 (1985).
D. J. Patel, S. A. Kozlowski, S. Ikuta, K. Itakura, FASEB, 11, 2664 (1984).
T. Brown, W. N. Hunter, G. Kneale, O. Kennard, Proc. Natl. Acad. Sci. U.S.A., 83,
2402 (1986).
B. H. Robinson, C. Mailer, G. Drobny, Ann. Rev. Biophys. Biomol. Sruct., 26, 629
(1997).
D. P. Millar, R. J. Roberts, A. H. Zewail, Proc. Natl. Acad. Sci. U.S.A., 77, 5593
(1980).
S. R. Rajski, S. Kumar, R. J. Roberts, J. K. Barton, J. Am. Chem. Soc., 121, 5615
(1999).
H. Sugiyama, I. Saito J. Am. Chem. Soc., 119, 617 (1996).

26

Kelley

39.

J. Jortner, M. Bixon, T. Langenbacher, M. E. Michel-Beyerle, Proc. Natl. Acad. Sci.,


U.S.A., 95, 12759 (1998).
D. Porath, A. Bezryadin, S. deVries, C. Dekker, Nature, 403, 635 (2000).
F. D. Lewis, T. Wu, Y. Zhang, R. L. Letsinger, S. R. Greeneld, and M. R.
Wasielewski, Science, 277, 673 (1997).
T. J. Meade and J. F. Kayyem, Angew. Chem., 34, 352 (1995).
K. Fukui and K. Tanaka, Angew. Chem., 37, 158 (1998).
H. O. Finklea, Electroanal. Chem., 19, 109 (1996).
S. O. Kelley, J. K. Barton, N. M. Jackson, L. McPherson, A. Potter, E. M. Spain,
M. J. Allen, M. G. Hill, Langmuir, 14, 6781 (1998).
R. Leviky, T. M. Herne, M. J. Tarlov, S. K. Satija, J. Am. Chem. Soc., 120, 9787
(1998).
M. E. Napier, C. R. Loomis, M. F. Sistare, J. Kim., A. E. Eckhardt, H. H. Thorp, Bioconj. Chem., 8, 906 (1997).
D. M. John, K. M. Weeks, Chem. Biol., 7, 405 (2000).

40.
41.
42.
43.
44.
45.
46.
47.
48.

2
Electrochemical DNA Biosensors
Joseph Wang
New Mexico State University, Las Cruces, New Mexico

I.

INTRODUCTION

Sequence-specic DNA detection has been a topic of tremendous interest because


of its application to the screening of genetic and infectious diseases, for ensuring
our food safety, for criminal investigations, and eld testing of microbial and viral
pathogens. For example, the detection of genetic mutations and pathogens creates
the possibility of identifying genetic or pathogenic diseases before any symptoms
appear. Wide-scale DNA testing requires the development of fast, easy-to-use,
inexpensive, tiny analytical devices. Traditional methods for studying DNA
hybridization are too slow and labor intensive. Biosensors offer a promising
alternative for faster, cheaper, and simpler hybridization assays.
Biosensors are small devices employing biochemical molecular recognition properties as the basis for a selective analysis [1]. The major processes
involved in any biosensor system are analyte recognition, signal transduction, and
readout (Figure 1). The use of nucleic acids recognition layers adds new and
unique dimensions to the arsenal of modern biosensors. DNA hybridization
biosensors commonly rely on the immobilization of a single-stranded oligonucleotide probe onto a transducer surface to recognize (by hybridization) its complementary sequence. Such devices thus belong to bioafnity-based biosensors
that rely on the selective binding of the target analyte to a surface-conned ligand
partner. In DNA biosensors, binding of the surface-conned probe and its complementary target strand is translated into a useful analytical signal. Transducing
elements reported in the literature have included optical [2], electrochemical [3],
and microgravimetric [4] devices. The two major requirements for a successful
27

Figure 1 Major processes involved in any biosensor system: analyte recognition, signal transduction, and readout.

28
Wang

Electrochemical DNA Sensors

29

operation of a DNA biosensor are high specicity (including observation of a


change in a single nucleotide) and high sensitivity. Even though nucleic acids are
relatively simple molecules, nding the sequence that contains the desired information is a very challenging task.
Electrochemical devices have traditionally received the major share of the
attention in biosensor technology [3,5,6]. The high sensitivity of electrochemical
transducers, coupled to their compatibility with modern miniaturization/microfabrication technologies, low cost, minimal power requirements, and independence of sample turbidity, make them excellent candidates for DNA diagnostics.
In addition, electrochemistry offers innovative routes for interfacing, at the
molecular level, the nucleic acid recognition system with the signal-generating
element. Direct electrical reading of DNA interactions thus offers great promise
for developing simple, rapid, and user-friendly DNA sensing devices (in a manner analogous to miniaturized blood-glucose meters). Such opportunities and
electrochemical avenues for generating the hybridization signal are the subject of
the present chapter.

II.

ELECTROCHEMICAL DNA HYBRIDIZATION BIOSENSORS

Electrochemical detection of DNA hybridization usually involves monitoring of


a current response (resulting from the recognition event) under controlled potential conditions [6]. The basis for the recognition event is the Watson-Crick base
pairing [i.e., the high afnity pairing of guanine with cytosine (G-C) and adenine
with thymine (A-T)]. Polynucleotide probes thus provide one of the most specic
biological recognition systems. Under stringent conditions, a single base mutation may be detected by hybridization. The probe-modied electrode is commonly immersed into a solution of a target DNA whose nucleotide sequence is to
be tested. When the target DNA contains a sequence that exactly matches that of
the immobilized oligonucleotide probe DNA, a hybrid duplex DNA is formed at
the electrode surface (Figure 1). Such hybridization event is commonly detected
via the increased current signal of an electroactive indicator (that preferentially
binds to the DNA duplex) or from other hybridization-induced changes in electrochemical parameters (e.g., conductivity or capacitance).
In the following sections we will focus on the major steps involved in electrochemical DNA hybridization biosensors, namely the formation of the DNA
recognition layer, the actual hybridization event, and the transformation of the
hybridization event into an electrical signal (Figure 2). As will be illustrated below,
the success of such devices requires a right combination of synthetic-organic and
surface chemistries, DNA recognition, and electrochemical detection schemes.

30

Wang

Figure 2 Major processes involved in electrochemical DNA biosensors based on the


use of redox indicators (Ox).

A.

Interfacial Immobilization

The probe immobilization step plays a major role in the overall performance of
electrochemical DNA biosensors. The achievement of high sensitivity and selectivity requires maximization of the hybridization efciency and minimization of
nonspecic adsorption events, respectively. The probes are typically short olig-

Electrochemical DNA Sensors

31

onucleotides (2040 mers) that are capable of hybridizing with specic and unique
regions of the target nucleotide sequence. Control of the surface chemistry and
coverage is essential for assuring high reactivity, orientation/accessibility, and stability of the surface-bound probe, as well as for avoiding nonspecic binding/
adsorption events. For example, it was demonstrated recently that the density of
immobilized ssDNA can inuence the thermodynamics of hybridization and
hence the selectivity of DNA biosensors [7]. Greater understanding of the relationship between the surface environment of biosensors and the resulting analytical performance is desired. This is particularly important as the physical environment of hybrids at solid/solution interface can differ greatly from that of hybrids
formed in the bulk solution [7]. Several useful schemes for attaching nucleic acid
probes onto electrode surfaces have thus been developed. The exact immobilization protocol often depends on the electrode material used for signal transduction.
Common probe immobilization schemes include self-assembly of organized monolayers of thiol functionalized probes onto gold transducers, carbodiimide covalent binding to an activated surface, attachment of biotin-functionalized probes to avidin-coated surfaces, as well as adsorptive accumulation onto
carbon-paste or disposable strip electrodes. The use of alkanethiol self-assembly
methods has been particularly attractive for fabricating reproducible probe-modied surfaces with high hybridization activity [8]. For this purpose, the DNA is
commonly immobilized on gold by forming mixed monolayers of thiol-derivatized single-stranded oligonucloetide and 6-mercapto-1-hexanol (Figure 3). The

Figure 3 Schematic of a mixed thiol-derivatized single-stranded oligonucloetide/6mercapto-1-hexanol monolayer in a solution containing the target DNA. (From Ref. 8.)

32

Wang

thiolated probe is put upright as a result of such coassembly with a short-chain


alkanethiol. A hydrophilic linker (between the thiol group and DNA) is often used
for minimizing nonspecic adsorption effects.
Despite impressive progress, there are many fundamental questions concerning surface orientation and accessibility and the nature of the interfacial
molecular interactions. Surface characterization techniques (e.g., XPS, ellipsometry, reectance IR) are often used to obtain useful insights into the surface coverage and organization.
B.

The Hybridization Event

The development of DNA biosensors requires proper attention to experimental


variables affecting the hybridization event at the transducer-solution interface.
These include the salt concentration, temperature, the presence of accelerating
agents, viscosity, contacting time, base composition (%G + C), and length of
probe sequence. Careful control of the hybridization event is thus required. The
stability of duplexes formed between strands with mismatched bases is decreased
according to the number and location of the mismatches. Elevated temperatures
can thus be used for discriminating among oligonucleotide hybrids. Control of the
hybridization time can be used for tuning the linear dynamic range, with shorter
time offering a wide range at the cost of lower sensitivity. Detection limits from
the nanomolar to the picomolar concentration range can thus be achieved in connection to 5 and 60 min hybridization times.
We have demonstrated that signicantly enhanced selectivity can be
achieved by the use of peptide nucleic acid (PNA) probes [9]. PNA is a structural
DNA analogue, with an uncharged pseudopeptide backbone (replacing the charged
phosphate-sugar one). Because of their neutral backbone, PNA probes offer greater
afnity in binding to complementary DNA, and improved distinction between
closely related sequences (including differentiating between point mutations).
Such mismatch discrimination is of particular importance in the detection of disease-related mutations, in connection to genetic screening and therapy.
Attention should be given also to the reusability of the DNA biosensors (i.e.,
to the regeneration of the surface-bound single-stranded probe after each assay).
Both thermal and chemical (urea, sodium hydroxide) regeneration schemes have
been shown useful for removing the bound target in connection with different
DNA biosensor formats. Even more elegant is the use of controlled electric elds
for facilitating the denaturation of the duplex [10]. Such electronic control has
been used also (by Nanogen Inc.) for discriminating among oligonucleotide
hybrids. Mechanically renewed electrodes, including polishable biocomposites
and graphite pencils, have also been reported for regenerating a fresh probe layer
[11,12]. Alternatively, one can use single-use (disposable) screen-printed elec-

Electrochemical DNA Sensors

33

trodes (similar to those use for self-testing of glucose), and hence obviate the need
for regeneration [13].

C.

Electrochemical Transduction of DNA Hybridization

The hybridization event is commonly detected via the increased current signal of
a redox indicator (that associates with the newly formed surface hybrid), and from
changes in electrochemical parameters (e.g., conductivity or capacitance) or in
the redox activity of the nucleic acid resulted from the duplex formation.

1.

Indicator-Based Detection

Hybridization indicators are small redox-active DNA-intercalating or groovebinding substances that possess a much higher afnity for the resulting duplex
compared with the single-stranded probe. Accordingly, the concentration of the
indicator at the electrode surface increases when hybridization occurs, resulting in
increased electrochemical response. Besides effective discrimination between
ss- and ds-DNA, the indicator should possess a well-dened, low-potential,
voltammetric response. These properties of hybridization indicators are essential
for attaining high sensitivity and selectivity. Both linear-scan or square-wave
voltammetric modes [14] or constant-current chronopotentiometry [15] can be
used to detect the association of the redox indicator with the surface duplex.
Mikkelsens group, which pioneered the use of redox indicators, demonstrated their utility for detecting the cystic brosis F508 deletion sequence associated with 70% of cystic brosis patients [16]. A detection limit of 1.8 fmol was
demonstrated for the 4000-base DNA fragment in connection to a Co(bpy)33+
marker. High selectivity toward the disease sequence (but not to the normal DNA)
was achieved by performing the hybridization at an elevated (43C) temperature.
Such use of the electrochemical transduction mode requires that proper attention
be given to the choice of the indicator and its detection scheme. Our laboratory
demonstrated the use of the Co(phen)3+3 indicator, in connection to a carbon-paste
chronopotentiometric transducer, for detecting a point mutation in the p53 gene
[17]. Other common and useful redox indicators include anthracycline antibiotics
such as daunomycin [18] or bisbenzimide dyes such as Hoecht 33258 [19]. The
daunomoycin-based chronopotentiometric biosensor was combined with PCR
amplication of DNA extracted from whole blood for the genetic detection of
apolipoprotein E polymorphism [20].
New redox indicators, offering greater discrimination between ss- and dsDNA are being developed for attaining higher sensitivity. Very successful has
been the use of a threading intercalator ferrocenyl naphthalene diimide (FND)

34

Wang

[21] that binds to the DNA hybrid more tightly than usual intercalators and displays a negligible afnity to the single-stranded probe. This threading indicator
resulted in a detection limit of 10 zmol in connection to differential pulse voltammetric monitoring of the hybridization event (Figure 4). The oligonucleotide
probe was chemisorbed onto gold electrodes through a thiol anchor. Table 1 summarizes common redox indicators used for the biosensing of DNA hybridization.
2.

Use of Enzyme Labels for Detecting DNA Hybridization

Enzyme labels have been widely used in bioafnity sensors, particularly in


immunosensors. The use of enzyme labels offers also great promise for electrochemical detection of DNA hybridization. Hellers group [22] demonstrated that
a direct low-potential sensitive amperometric monitoring of the hybridization

Figure 4 Differential pulse voltammograms for the ferrocenyl naphthalene diimide indicator at the dT20-modied electrode before (a) and after (b) hybridization with dA20. Also
shown, the chemical structure of the indicator. (From Ref. 21.)

Electrochemical DNA Sensors

35

Table 1 Electroactive indicators commonly used for the biosensing of DNA


hybridization
Indicator
Co(bpy)33+
Co(phen)3+3
Daunomycin
Hoechst 33258
Ferrocenyl
naphthalene

Detection mode
Cyclic voltammetry
Chronopotentiometry
Chronopotentiometry
Pulse voltammetry
Pulse voltammetry
diimide

Electrode
transducer

Ep,a (vs.
Ag/AgCl), V

Ref.

Carbon paste
Carbon paste
Screen-printed
Gold
Gold

0.15
0.15
0.45
0.58
0.50

16
15
18
19
21

event could be achieved in connection to the use of horseradish-peroxidase (HRP)


labeled target and an electron-conducting redox polymer. In this system, the
hybridization of enzyme-labeled oligo(dA)25 target with oligo(dT)25 probe, covalently attached to electron-conducting redox hydrogel, resulted in the wiring of
the enzyme to the transducer and in a continuous hydrogen-peroxide electroreduction current. A single-base mismatch in an 18-base oligonucleotide was thus
detected using a 7 m-diameter carbon ber transducer. Enzyme (HRP) labels
have been combined by Willners group with a biocatalytic precipitative accumulation of the reaction product to achieve multiple amplications and hence
extremely low detection limits [23]. Applicability for the detection of mutations
relevant to the Tay-Sachs genetic disorder was demonstrated. Such amplication
routes are described in detail in Chapter 3. The same enzyme label was employed
by AndCare Inc. for quantitative pulse amperometric monitoring of PCR amplication [24].

3.

Label-Free Electrochemical Biosensing of DNA Hybridization

Increased attention has been given recently to new indicator-free electrochemical


detection schemes that greatly simplify the sensing protocol (as the need for the
indicator addition/association/detection steps is eliminated) and offer an instantaneous detection of the duplex formation. Such direct, in-situ detection can be
accomplished by monitoring changes in the intrinsic electrochemical properties
of the interface or changes in the redox activity of the nucleic acid target or probe.
For example, it is possible to exploit changes in the intrinsic electroactivity of
DNA accrued from the hybridization event [25,26]. Among the four nucleic acids
bases, the guanine moiety is most easily oxidized and is most suitable for such

36

Wang

label-free hybridization detection. To overcome the limitations of the probe


sequences (absence of G), guanines in the probe sequence were substituted by
inosine residues (pairing with Cs) and the hybridization was detected through the
target DNA guanine signal [25]. A greatly amplied guanine signal, and hence
hybridization response, can be obtained by using the electrocatalytic action of a
Ru(bpy)3+2 redox mediator [26]. This involves the following catalytic cycle:
Ru(bpy)3+2 B Ru(bpy)3+3 + e

(1)

Ru(bpy)3+3 + G B Ru(bpy)3+2 + G+

(2)

Direct, label-free, electrical detection of DNA hybridization has also been


accomplished by monitoring changes in the conductivity of conducting polymer
molecular interfaces, e.g., using DNA-substituted or doped polypyrrole lms
[27,28]. For example, Garniers group [27] has demonstrated that a 13-mer
oligonucleotide substituted polypyrrole lm displays a decreased current
response during the duplex formation (Figure 5). Such change in the electronic
properties of polypyrrole has been attributed to bulky conformational changes
along the polymer backbone due to its higher rigidity following the hybridization.
Eventually, it would be possible to eliminate these polymeric interfaces and to
exploit different rates of electron-transfer through ss- and ds-DNA for probing
hybridization (including mutation detection via the perturbation in charge transfer through DNA). Recent activity in exploiting charge transport through DNA
lms for mismatch detection is very encouraging [29]. Such new electrochemical
assays based on charge migration through dsDNA are described in detail in Chapter 1. A related and very attractive protocol, developed by Clinical Micro Sensors
Inc., employs DNA-label complexes that are connected to the electrode by
phenylacetylene molecular wires that are embedded in a self-assembled monolayer of alkane thiols [30]. Such a packed layer also protects the surface (against
nonspecic adsorption and electroactive interferences), hence facilitating the
analysis of complex biological samples.
New routes for generating the hybridization signal are currently being
explored in several laboratories. Siontorou et al. [31] reported on the use of selfassembled bilayer lipid membranes (BLMs) for the direct monitoring of DNA
hybridization. A decrease in the ion conductivity across the lipid membrane surface, containing the single-stranded probe, was observed during the formation of
the duplex. This was attributed to alterations in the ion permeation associated
with structural changes in the BLM accrued by the desorption of the dsDNA.
Umezawas group described a novel ion-channel protocol for the indirect
biosensing of DNA hybridization [32]. The system relied on the electrostatic
repulsion of the ferrocyanide redox marker, accrued from the hybridization of the
negatively charged target DNA and the neutral PNA probe (Figure 6). High specicity (toward one-point mutations) was demonstrated.

Electrochemical DNA Sensors

37

Figure 5 Voltammetric hybridization response of a biosensor based on a electropolymerizable oligonucleotide-substituted polypyrrole, to increasing levels of the DNA target:
0(a,b), 66(c), 165(d), and 500(e) nmol. (From Ref. 27.)

Johanssons group demonstrated recently that changes in the capacitance of


a thiolated-oligonucleotide modied gold electrode, provoked by hybridization to
the complementary strand (and the corresponding displacement of solvent molecules from the surface), can be used for rapid and sensitive detection of DNA
sequences [33].
The electrochemical response of the G nucleobase is also very sensitive to
the DNA structure and can thus be used for probing DNA damage or interactions.
Changes in the guanine oxidation, and of other intrinsic DNA redox signals have
thus been used for detecting small damage to DNA induced by various chemical
agents, enzymatic digestion, or ionizing radiation [34,35]. Similarly, electrochemical biosensors based on immobilized dsDNA can be used for studying the rate and
strength of the interaction of drugs or carcinogens with DNA and for shading useful insights into the pharmaceutical or toxic action of such compounds [36].

Figure 6 An ion-channel sensor based on a PNA probe immobilized on gold electrode, and detection of the hybridization based on the electrostatic repulsion of a negatively charged redox marker (shown as an octahedron). (From Ref. 32.)

38
Wang

Electrochemical DNA Sensors

IV.

39

CONCLUSIONS

Over the past 10 years, we have witnessed a tremendous progress toward the
development of electrochemical DNA biosensors. Such devices are of considerable recent interest due to their extraordinary promise for obtaining sequence-specic information in a faster, simpler, and cheaper manner compared to traditional
hybridization assays. In addition to excellent economic prospects, such devices
offer innovative routes for interfacing (at the molecular level) the DNA-recognition and signal-transduction elements, i.e., an exciting opportunity for basic
research. The realization of instant on-site (clinical, forensic, or environmental)
DNA testing would require additional developmental work. Particular attention
should be given to the integration of various processes, including sample collection, DNA extraction and amplication, with the actual hybridization detection,
on a single microfabricated chip. By performing all the steps of the biological
assay on a microchip platform, we expect signicant advantages in terms of cost,
speed, simplicity, and automation. The integration of multiple biosensors in connection to DNA microarrays should lead to the simultaneous analysis of multiple
DNA sequences, and hence to the generation of characteristic hybridization patterns and acquisition of expression information. Screening of DNA-protein or
DNA-drug interactions would also benet from such DNA microarrays. New
DNA biosensor technologies are anticipated in the near future in response to the
above opportunities.

ACKNOWLEDGMENT
The author gratefully acknowledges nancial support from the US Army Medical
Research (Award No. DAMD17-00-1-0366) and the National Institutes of Health
(Grant No. R01 14549-02).

ABBREVIATIONS
BLM
DNA
FND
HRP
PCR
PNA
XPS

Bilayer lipid membrane


Deoxyribonucleic acid
Ferrocenyl naphthalene di-imide
Horseradish peroxidase
Polymerase chain reaction
Peptide nucleic acid
X-ray photoelectron spectroscopy

40

Wang

REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.

Taylor, R. F., Schultz, J. S. (1996) Handbook of Chemical and Biological Sensors,


Institute of Physics Publishing, Bristol.
Piunno, P. A. E., Krull, U. J., Hudson, R. H. E., Dahma, M., Cohen, H. (1995) Anal.
Chem. 67, 26352643.
Wang, J. (1999) Chem. Eur. J. 5, 16811684.
Okahata, Y., Matsunobo, Y., Ijiro, K., Mukae, M., Murkami, A., Makino, K. (1992)
J. Am. Chem. Soc. 114, 82998300.
Wang, J. (1999) Anal. Chem. 70, 328332.
Mikklesen, S. R. (1996) Electroanalysis 8, 1519.
Watterson, J. H., Piunno, P. A. E., Wust, C. C., Krull, U. J. (2000) Langmuir 16,
49844992.
Steel, A. B., Herne, T. M., Tarlov, M. J. (1998) Anal. Chem. 70, 46704677.
Wang, J., Palecek, E., Nielsen, P., Rivas, G., Cai, X., Shiraishi, H., Dontha, N., Luo,
D., Farias, P. (1996) J. Am. Chem. Soc. 118, 76677671.
Cheng, J., Sheldon, E. L., Wu, L., Uribe, A., Gerrue, L. O., Carrino, J., Heller, M. J.,
OConnell, J. P. (1998) Nature Biotech. 16, 541545.
Wang, J., Fernandes, J., Kubota, L. (1998) Anal. Chem. 70, 36993702.
Wang, J., Kawde, A., Sahlin, E. (2000) Analyst 125, 57.
Wang, J., Cai, X., Tian, B., Shiraishi, H. (1996) Analyst 121, 965970.
Millan, K. M., Mikkelsen, S. R. (1993) Anal. Chem. 65, 23172321.
Wang, J., Cai, X., Rivas, G., Shiraishi, H. (1996) Anal. Chim. Acta 326, 141147.
Millan, K. M., Saraulo, A. Mikklesen, S. R. (1994) Anal. Chem. 66, 29432948.
Wang, J., Rivas, G., Cai, X., Chicharro, M., Parrado, C., Dontha, N., Begleiter, A.,
Mowat, M., Palecek, E, Nielsen, P. E. (1997). Anal. Chim. Acta 344, 111118.
Marrazza, G., Chianella, I., Mascini, M. (1999) Anal Chim. Acta 387, 297307.
Hashimoto, K., Ito, K., Ishimori, Y. (1994) Anal. Chem. 66, 38303833.
Marrazza, G., Chiti, G., Mascini, M., Anichini, M. (2000) Clin. Chem. 46, 3137.
Takenaka, S., Yamashita, K., Takagi, M., Uto, Y., Kondo, H. (2000) Anal. Chem.,
72, 13341341.
de Lumley Woodyear, T. Campbell, C. N., Heller, A. (1996) J. Am. Chem. Soc. 118,
55045508.
Patolsky, F., Katz, E., Bardea, A., Willner, I. Langmuir 15(1999)3703.
Wojciechowski, M., Sundseth, R., Moreno, M., Henkens, R. (1999) Clin. Chem. 45,
16901693.
Wang, J., Rivas, G., Fernandes, J., Paz, J. L., Jiang, M., Waymire, R. (1998) Anal.
Chim. Acta 375, 197203.
Johnston, D. H., Glasgow, K., Thorp, H. H. (1995) J. Am. Chem. Soc. 117, 8933
8938.
Korri-Youssou, H., Garnier, F., Srivtava, P., Godillot, P., Yassar, A. (1997) J. Am.
Chem. Soc. 119, 73887391.
Wang, J., Jiang, M., Fortes, A., Mukherjee, B. (1999) Anal. Chim. Acta 402, 712.
Kelley, S. O., Boon, E. M., Barton, J. K., Jackson, N. M., Hill, M. G. (1999) Nucleic
Acids Res. 27, 48304837.

Electrochemical DNA Sensors


30.

31.
32.
33.
34.
35.
36.

41

Creager, S., Yu, C. J., Bamdad, C., OConnor, S., MacLean, T., Lam, E., Chong, Y.,
Olsen, G. T., Luo, J. Y., Gozin, M., Kayyem, J. F. (1999) J. Am. Chem. Soc. 121,
10591064.
Siontorou, C. G., Nikolelis, D. P., Piunnu, P. A. E., Krull, U. J. (1997), Electroanalysis, 9, 10671072.
Aoki, H., Buhlmann, P., Umezawa, Y. (2000) Electroanalysis, 12, 12721276.
Berggren, C., Stalhandske, P., Brundell, J., Johansson, G. (1999) Electroanalysis,
11, 156160.
Palecek, E. (1996) Electroanalysis, 8, 714.
Wang, J., Rivas, G., Ozsoz, M., Grant, D., Cai, X., Parrado, C. (1997) Anal. Chem.
69, 14571460.
Palecek, E., Fojta, M., Tomschik, M., Wang, J. (1998) Biosensors Bioelect. 13,
621628.

3
Amplied and Specic Electronic
Transduction of DNA Sensing
Processes in Monolayer and
Thin-Film Assemblies
Itamar Willner, Eugenii Katz, and Bilha Willner
Hebrew University of Jerusalem, Jerusalem, Israel

I.

INTRODUCTION

The development of DNA sensors attracts recent research efforts directed to gene
analysis, the detection of genetic disorders, tissue matching, forensic applications,
and the detection of viral infections [13]. Optical detection of DNA was accomplished by the application of uorescence-labeled oligonucleotides [4] and the use
of surface plasmon resonance spectroscopy (SPR) [5]. Recent optical detection of
DNA was accomplished by the use of Au-nanoparticles as photonic probes [6,7].
Electronic transduction of oligonucleotide-DNA recognition events, and specically the quantitative assay of DNA, are major challenges in DNA-based bioelectronics [8]. Electrochemical DNA sensors based on the amperometric or
voltammetric transduction of the formation of double-stranded (ds) oligonucleotide-DNA complexes were reported by following the direct electrical response
of the ds-assembly [9], the examination of the effect of the ds-assembly on the
voltammetric wave of conductive polymers [10], and the electrical response of
transition metal complexes [11] or dyes [12] that are intercalated or electrostatically attracted to the double-stranded assembly. Microgravimetric quartz-crystalmicrobalance (QCM) analyses were also applied to sense the formation of dsoligonucleotide-DNA complexes on surfaces [13]. Two fundamental problems
43

44

Willner et al.

that need to be addressed while developing electronic DNA sensors relate to the
sensitivity and specicity of the devices. PCR provides a means for the amplication of the DNA content, but the method is limited to quantitative and parallel high
throughput analyses of DNA [14]. Thus, the development of novel amplication
means for the quantitative DNA sensing events is essential. Furthermore, the
specicity of the DNA recognition is important, and the feasibility of discriminating mutants from the normal base-sequence, with the optimal capability to distinguish single-base mismatches or mutations, is a challenging goal.
Furthermore, the development of electronic transduction means that probe
the DNA sensing events on the transducers is essential. It is also desirable to
develop sensing protocols that could later adapt a chip conguration that enables
the screening of multiple sensing processes.
This chapter will address recent advances in the development of electronic
DNA-sensors. The aspects that will be considered include:
Description of the electronic transduction means that enable the characterization of the sensing interfaces and the sensing processes occurring on
the transducers.
The development of amplication routes for the DNA sensing events.
The electronic transduction of biocatalytic transformations occurring on
surfaces such as replication, ligation, specic scission, etc., and the utility of these processes for amplied and specic DNA analyses.
Biosensors based on biorecognition events occurring in monolayer or thinlm assemblies on electronic transducers represent an important recent advance
in bioelectronics [15]. The nano-architecture of the sensing interface in a monolayer, multilayer, or thin-lm structures precludes diffusion barriers, and hence
the rapid response-times of the sensing devices are achieved. Enzyme-electrodes
[16,17], immunosensors [18,19], or DNA sensors [13] were developed by tailoring nanoscale sensing interfaces on the transducer. We will specically address
the development of DNA sensors in monolayer and thin-lm congurations.

II. ELECTRONIC TRANSDUCTION OF DNA SENSING EVENTS


ON SURFACES
The sensing events on the electronic transducers are designed to alter the interfacial properties at the transducer/solution interface. Changes in the interfacial
charge, capacitance, resistance, mass, thickness, and so forth would occur upon
the hybridization of the analyte DNA with a probe nucleic acid, and upon stimulating the amplication route. Thus, any electronic transduction method that follows such interfacial changes could be employed to detect the analysis of the
DNA.

Electronic Transduction of DNA Sensing Processes

45

Different electronic transduction methods to probe DNA recognition events


on surfaces will be addressed in the present account. These include electrochemical transduction means such as amperometry, faradaic impedance spectroscopy
or chronopotentiometry, and microgravimetric, quartz-crystal-microbalance
(QCM) measurements.
Impedance spectroscopy is an effective method for probing the features of
surface-modied electrodes [20,21]. The complex impedance can be presented as
the sum of the real, Zre(), and imaginary, Zim() components that originate
mainly from the resistance and capacitance of the cell, respectively. The general
electronic equivalent scheme (Randles and Ershler model [21]), Figure 1(A),
includes the ohmic resistance of the electrolyte solution, Rs, of the Warburg
impedance, Zw, resulting from the diffusion of ions from the bulk electrolyte to
the electrode interface, the double layer capacitance, Cdl, and electron-transfer
resistance, Ret, that exists if a redox probe is present in the electrolyte solution.
The two components of the electronic scheme, Rs and Zw, represent bulk properties of the electrolyte solution and diffusion features of the redox probe in solution, respectively. Therefore, these parameters are not affected by chemical transformations occurring at the electrode surface. The other two components in the
scheme, Cdl and Ret, depend on the dielectric and insulating features at the electrode/electrolyte interface. The double-layer capacitance consists of the constant
capacitance of an unmodied electrode (e.g., for an Au electrode, CAu 4060
Fcm2, depending on the applied potential [22]) and a variable capacitance originating from an electrode surface modier, Cmod. In case the modied layer has a
dielectric constant smaller than that of water, the double-layer capacitance is
expressed by Eq. 1., where CAu and Cmod are the capacitances of the nonmodied
Au-electrode and the variable component introduced by the modier, respectively. Ret controls the electron-transfer kinetics of the redox probe at the electrode interface.
1
1
1
= +
Cdl
CAu
Cmod

(1)

The modication of the metallic surface with a biomaterial or an organic


layer decreases the double-layer capacitance and retards the interfacial electrontransfer kinetics. The electron-transfer resistance at the electrode is given by Eq.
2, where RAu and Rmod are the electron-transfer resistance of the nonmodied
electrode and the
Ret = RAu + Rmod

(2)

variable electron-transfer resistance introduced by the modier, in the presence of


the solubilized redox probe, respectively. A typical shape of a faradaic impedance
spectrum (presented in the form of a Nyquist plot), Figure 1(B), curve a, includes

46

Willner et al.

Figure 1
(A) Equivalent circuit corresponding to the impedance features of a DNAmodied electrode interfaces in the presence of a redox probe. (B) Schematic faradaic
impedance spectra presented in the form of a Nyquist plot for: (a) A modied electrode
where the impedance is controlled by diffusion of the redox probe (low frequencies)
and by the interfacial electron transfer (high frequencies). (b) A modied electrode where
the impedance is controlled by diffusion of the redox probe. (c) A modied electrode where
the impedance is controlled by the interfacial electron transfer within the entire range of the
applied frequencies.

Electronic Transduction of DNA Sensing Processes

47

a semicircle region lying on the Zre-axis followed by a straight line. The semicircle portion, observed at higher frequencies, corresponds to the electron-transferlimited process, whereas the linear part is characteristic of the lower frequencies
range and represents the diffusional-limited electron-transfer process. In the case
of very fast electron-transfer processes, the impedance spectrum could include
only the linear part, Figure 1(B), curve b, whereas a very slow electron-transfer
step results in a big semicircle region that is not accompanied by a straight line,
Figure 1(B), curve c. The electron-transfer kinetics and diffusional characteristics
can be extracted from the spectra. The semicircle diameter equals Ret. The intercept of the semicircle with the Zre-axis at high frequencies ( B ) is equal to Rs.
Extrapolation of the circle to lower frequencies yields an intercept corresponding
to Rs + Ret. The characteristic frequency, 0, given by Eq. 3 has the meaning of
the reciprocal of the time constant of the equivalent circuit. The maximum value
of the imaginary impedance in the semicircle part corresponds to Zim = Ret/2 and
is achieved at the characteristic frequency, 0.
0 = (CdlRet)1

(3)

Chronopotentiometry provides a further electrochemical means to characterize resistance changes at an electrode surface. Chronopotentiometry [20,23] is
an electrochemical technique that applies a constant and controlled current
between the working and auxiliary electrode, while the potential between the
working electrode and the reference electrode is altered to retain the desired current value. In the presence of a reversible redox-probe in the solution, a nernstian
electrochemical process occurs upon the application of the constant current value,
and the electrode potential is shifted to the characteristic potential of the redoxprobe in solution. The potential of the electrode is constantly altered according to
the Rox/Rred ratio of the redox-label at the electrode surface, Figure 2, curve a.
Provided that a cathodic current is driven through the solution, after a transition
time, , the concentration of the oxidized redox species, Rox, at the electrode,
drops to zero. Under these conditions, and in the absence of any other redox probe
in solution, the potential on the electrode will be sharply shifted to negative values, corresponding to the cathodic discharge of the electrolyte (or the reduction
of oxygen), in order to retain the passage of the desired current value. The biocatalyzed precipitation of an insulating layer on the electrode support is anticipated to inhibit the interfacial electron-transfer rate constant. Thus, in order to
retain the set current value in the cell, the application of an overpotential, , on
the electrode is required, Figure 2, curve b. The overpotential on the electrode will
relate to the change in the electrode resistance, R, as a result of modication (formation of the insoluble precipitate) as given by Eq. 4, where I is the set constant
current, and Rmod and RAu correspond to the resistances of the modied electrode
and the bare electrode, respectively. The slope of the E-t curve in the presence of

48

Willner et al.

Figure 2
Schematic chronopotentiometric transients in the presence of a diffusional
redox probe for: (a) An electrode of low resistance. (b) A modied electrode exhibiting
high resistance.

a modier (or a precipitate) on the electrode is different from the respective curve
corresponding to reversible, nernstian, behavior of the redox-probe because the
electron-transfer rate is slower in the presence of the modier. Thus, for different
experiments, the -values should be monitored at identical time-intervals of the
chronopotentiometric pulse.
R = /I = Rmod RAu

(4)

Several precautions and limitations should be mentioned upon the application of chronopotentiometry as an electrochemical method to analyze interface
properties of layered-modied-electrodes, and, specically, monolayer-functionalized-electrodes: (1) The chronopotentiometric pulse results in a potential shift
on the electrode after the transition-time, , resulting in the discharge of the electrolyte. This potential shift often ruins the chemically functionalized layer on the
electrode (e.g., thiolated monolayers on Au-electrodes). Thus, it is essential to terminate the chronopotentiometric pulse at shorter time-intervals than the transition
time, , in order to eliminate the destruction of the functionalized electrode. (2)
The electrode resistance, R, values derived from the chronopotentiometric exper-

Electronic Transduction of DNA Sensing Processes

49

iments do not coincide with the electron-transfer resistances, Ret, obtained from
the faradaic impedance spectra, and the comparison of these values needs to be
made with caution. While the electrode resistances, R, correspond to the entire
current ux at the electrode, the electron-transfer resistances, Ret, correspond only
to the faradaic current at the electrode interface. Thus, in the chronopotentiometric experiment, the non-faradaic current originating from the double-layer charging always affects the electrode resistance. At high concentrations of the redoxprobe (>103 M) the double-layer charging current is negligible as compared to
the faradaic current [23]. Under these conditions, it is expected that R Ret [24].
Furthermore, the double-layer charging current increases with the potential
applied onto the electrode. Thus, at high overpotential values, resulting at certain
modications of the electrode, deviations between the total electrode resistances
derived from chronopotentiometry and the electron-transfer resistances determined by faradaic impedance spectroscopy, may be observed, even at high concentrations of the redox-probe.
The frequency responses of piezoelectric crystals (i.e., quartz crystals) may
be employed as an electronic transduction means of biorecognition events. Specifically, microgravimetric quartz-crystal-microbalance, QCM, measurements can
probe mass changes, m, that occur on the crystal as a result of adsorption, binding
or precipitation. The relation between the crystal frequency-change, f, as a result
of a mass change on the crystal, m, is given by the Sauerbrey equation, Eq. 5,
where f0 is the fundamental frequency of the quartz crystal, A is the piezoelectric
active area, q is the density of quartz (2.648 gcm3, and q is the shear modulus
(2.947 1011 dyncm2 for AT-cut quartz) [25]. The application of the Sauerbrey
equation to follow surface modication steps in solvents suffers from limitations
due to viscoelastic effects of the solution on the crystal. Nonetheless, this relation
proved to be an adequate rst-approximation for the mass changes occurring on the
crystal.
m
f = 2f20
A(qq)1/2

(5)

III. DEPOSITION OF NUCLEIC ACID PROBES


ON ELECTRONIC TRANSDUCERS
Chemical modication of nucleic acids with functionalized linkers, such as
amine-functionalized alkyl chains tethered to nucleic acids, has been employed to
modify surfaces with DNA probes [26]. Conductive supports and particularly Ausurfaces were functionalized with nucleic acid labeled with an oligothymine thiophosphate [(Ts )n] [27] or with a thiol-functionalized linker tethered to the probe

50

Willner et al.

DNA [28]. An electrochemical method for the quantitative assay of the surface
coverage of the nucleic acids assembled on the conductive transducers was developed by Tarlov [29].
The DNA coverage at the electrode surface was calculated from the number of cationic redox molecules electrostatically associated with the anionic DNA
backbone. Cations provide charge compensation for the anionic phosphate
groups in DNA. In solution these cations are labile and readily exchange with
other cations. The association constant between cations and DNA phosphate
groups increases with the cation charge. When an electrode modied with DNA
is placed in a low ionic strength electrolyte containing a multivalent redox cation,
the redox cation exchanges with the original charge compensation cation (usually
K+) and becomes electrostatically trapped at the interface. The validity of this
approach is based on the following three assumptions:
1. The redox marker associates with DNA strictly through electrostatic
interactions and the intercalation of the redox marker into the hydrophobic region of DNA should be avoided.
2. The amount of electrostatically trapped redox marker can be determined
precisely, i.e, all redox ions should be electrochemically contacted.
3. The charge compensation for the DNA phosphate groups is complete
and provided by the redox marker only.
The amount of cationic redox marker was then measured using chronocoulometry, a current integration technique, under equilibrium conditions. The
advantage of chronocoulometry is that the nondiffusional and diffusional electrochemical processes can be easily discriminated by this technique. Thus, chronocoulometry allows the measurements of the charge originating from the electrochemical process of the surface-conned redox marker associated with the
DNA-monolayer in the presence of a redox marker in the solution. The doublelayer charge resulting from the charging of the interfacial capacitance can be differentiated from the charge originating from the faradaic process as well. Thus,
quantitative analysis of the surface-conned redox probe can be performed in the
presence of solution redox probe being under equilibrium with the probe associated with the DNA-monolayer. The charge, Q, measured as the function of time, t,
in a chronocoulometric experiment is given by the integrated Cottrell equation,
Eq. 6:
2nFAD01/2C0 1/2
Q =
t + Qdl + nFA0
1/2

(6)

where n is the number of electrons per molecule for the redox probe reduction, F
is the Faraday constant (C/equiv), A is the electrode area (cm2), D0 is the diffu-

Electronic Transduction of DNA Sensing Processes

51

sion coefcient (cm2 s1), C0 is the bulk concentration (mol cm3), Qdl is the double-layer charging (C), and nFA0 is the charge corresponding to the reduction of
0 (mol cm2) of the surface-conned redox probe. The chronocoulometric intercept at t = 0 is then the sum of the double-layer charging, Qdl, and the faradaic
charge, nFA0, of the surface-conned redox marker. The parameter 0 can be
translated to the surface concentration of the DNA molecules, DNA, if the loading of the DNA with the redox marker is known, Eq. 7:
DNA = 0 (z/m) NA

(7)

where DNA is the DNA surface density in molecules per cm2, m is the number of
bases in the DNA molecule, z is the charge of the redox marker ion, and NA is
Avogadros number.
The assembly of the probe nucleic acid on the Au-support alters the mass
associated with the transducer as well. In a recent account [30], the surface coverage of the thiolated nucleic acid (5-TCTATCCTACGCT-(CH2)6-SH-3), (1)
on a gold surface was examined by comparing Tarlovs method to the microgravimetric quartz-crystal-microbalance measurements. An Au-electrode was
functionalized with the thiol-functionalized oligonucleotide (1). Figure 3(A)
shows the chronocoulometric assay of the stepwise modication of the electrode
with (1), according to Tarlovs method [29]. The redox-label Ru(NH3)63+ is used
to probe the content of the oligonucleotide (1) on the conductive support. Figure
3(A) shows the chronocoulometry transients that correspond to the bare Au-electrode, curve (a), and the transients upon the modication of the electrode with (1)
for different time-intervals, curves (b)(d). The double-layer charge, Qdl, and the
charge resulting from the Faradaic process of the DNA-associated redox marker,
nFA0, are extracted from the intercepts of the lines extrapolating the experimental curves, taking into account Eq. 6. As the time of modication is longer,
the charge associated with the redox process of Ru(NH3)63+ is higher. The DNAprobe surface density, DNA, is calculated by using Eq. 7. The modication of the
Au-surface with (1) was also followed by microgravimetric quartz-crystalmicrobalance measurements. Au-functionalized-quartz crystals (9 MHz, AT-cut)
were modied with (1), and the crystal frequency changes were monitored in air
at time-intervals of modication. The DNA surface-coverage on the electrode at
different time-intervals of modication with (1), was calculated using the data
obtained by Tarlovs method and the data obtained by QCM, Figure 3(B). We see
that the values of the surface coverage of (1) derived by the two methods are very
similar and the values determined by chronocoulometry are slightly higher.
The two methods were also applied to identify the surface-coverage of the
thymine thiophosphate-tagged nucleic acid, (3TSTSTSTSTSTCGCATCCTAT
CT5), (2), on Au-supports. The saturated surface coverage was found [31] to be ca.
4.5 1011 molecm2.

52

Willner et al.

Figure 3
(A) Chronocoulometric transients measured in the presence of 50 M
Ru(NH3)63+ in 0.01 M Tris-buffer, pH 7.3, for: (a) A bare Au-electrode; (b), (c), and (d)
After modication of the electrode in the presence of (1), 5 M, for 60, 90 and 180 min,
respectively. (B) Surface coverage of (1) on the electrode derived by chronocoulometry
(), and by microgravimetric experiments (), at different time intervals of modication
with (1).

Electronic Transduction of DNA Sensing Processes

IV.

53

AMPLIFIED DNA SENSING ON ELECTRONIC TRANSDUCERS

The concepts for the amplied detection of DNA are outlined in Fig. 4. The association of the analyte DNA to the probe oligonucleotide is followed by the coupling of an enzyme to the resulting double-stranded assembly, Fig. 4(A). The
enzyme activates a biocatalytic process that amplies a single recognition event
between the probe and the target-DNA by transforming numerous substrate mol-

Figure 4 Schematic amplied analysis of a target DNA using: (A) An oligonucleotideenzyme conjugate and a biocatalyzed transformation as amplication route. (B) An
oligonucleotide-functionalized particle (liposome or nanoparticle) as an amplifying unit.

54

Willner et al.

ecules to the product with a specied time-interval. Alternatively, the primary


probe-DNA recognition event is accompanied by the coupling of a macromolecular component, a particulate or colloid system, or a bio-membrane/membrane
analogue that alters the interfacial properties of the transducer, Fig. 4(B). The two
methods may be further combined into a bifunctional amplication route where
the particulate macromolecular amplier is conjugated to a biocatalytic amplier.
The enzyme amplier may stimulate bioelectrocatalytic transformations [32] or
may generate a product resulting from a bioelectrocatalytic process [33], thus
enabling the electrochemical transduction of the amplied DNA detection
process. Alternatively, the biocatalyst may generate an insoluble product or a
polymer lm on the transducer thus altering the interfacial and mass features of
the transducer surface. These latter biocatalyzed processes may be recorded and
transduced by chronopotentiometry, faradaic impedance or microgravimetric
quartz-crystal-microbalance measurements [34].
A.

Amplied DNA Sensing Through Bioelectrocatalysis

Electrical contacting of redox-enzymes with electrode supports was extensively


studied to develop amperometric biosensor and other bioelectronic devices [35].
Electrobiocatalytic transformations coupled to DNA recognition events were
used to amplify nucleic acid detection process [32,36]. The hemoprotein
cytochrome c lacks direct electrical contact with the bare Au-electrode. Immobilization of the thiolated primer (5-ACGGATGCTCC-(CH2)3-SH-3) onto the
electrode followed by the formation of the double-stranded assembly with the
complementary nucleic acid resulted in the quasi-reversible redox-response of
cytochrome c [36], Figure 5. This was attributed to the electrostatic attraction of
the hemoprotein by the negatively charged double-stranded DNA that facilitates
interfacial electron transfer. Although the DNA-stimulated electrical contacting
of cytochrome c with the electrode allows the secondary activation of redoxenzymes, and thus the amplication of sensing events, the differentiation of single-stranded versus double-stranded DNA on the electrical contact effectiveness
of cytochrome c needs further investigations.
Hybridization of a model oligonucleotide was transduced electrochemically
using bioelectrocatalytic activity of horseradish peroxidase [32]. The primary
oligonucleotide, poly(deoxythymidine)-5-phosphate, (T)2530, (3), was covalently linked to a hydrogel containing redox groups providing electron propagation
throughout the polymer. The polymer composed of polyacrylamide-hyrazide
functionalized with [Os(dmebpy)2Cl]+/2+ (dmebpy = 4,4-dimethyl-2,2-bipyridine), (4), redox units was crosslinked with a diepoxy-crosslinker, (5), on a vitreous carbon electrode as a conductive support (Figure 6). The complementary
oligonucleotide, poly(deoxyadenosine)-5-phosphate, (A)2530, labeled with the
horseradish peroxidase, HRP, (6), has interacted with the primary oligonucleotide
to yield the double-stranded oligonucleotide. The HRP-label immobilized at the

Electronic Transduction of DNA Sensing Processes

55

Figure 5 Cyclic voltammogram of cytochrome c at the Au-electrode functionalized with


the primer 5-ACGGATGCTCC-(CH2)3SH that was hybridized with the complementary
oligonucleotide. Electrochemical experiment recorded in Tris-buffer, pH 7.5; potential
scan rate, 50 mVs1. (Adapted from Ref. 36 with permission.)

electrode surface as the result of the double-stranded oligonucleotide formation


was electrically wired via the redox active hydrogel layer. The electrochemical
reduction of H2O2 biocatalyzed by the wired HRP-label resulted in the current
ow that allowed the electronic transduction of the biorecognition event (Figure
7). The binding of the complementary oligonucleotide (A)2530 resulted in much
higher electrocatalytic current than the current observed in case of the electrode
treatment with non-complementary (G)2530 oligonucleotide also labeled with the
HRP. Thus, the technique allows the efcient electrochemical transduction of the
double-stranded oligonucleotide formation that can be differentiated from nonspecic adsorption of oligonucleotides.
B. Amplied and Specic DNA Analysis by the Biocatalyzed
Precipitation of an Insoluble Product on the Transducer
Various biocatalyzed transformations induce the polymerization of a thin lm
[37], or the precipitation of an insoluble product [38,39] on the transducer, leading
to the electrode support insulation (increase of electrode resistance and decrease
of interfacial electron-transfer) or to an increase in the mass associated with a

56

Willner et al.

Figure 6 Enzyme-amplied analysis of DNA in a redox hydrogel, (4), crosslinked with (5).

piezoelectric crystal. Figure 8 exemplies several biocatalyzed transformations


that lead to the precipitation of an insoluble product on the transducer. Peroxidasemediated oxidation of 4-chloro-1-naphthol, (7), or 3,3,5,5-tetramethylbenzidine,
TMB, (9) by H2O2 to form the insoluble products, (8) and (10), respectively, or the
alkaline phosphatase oxidative hydrolysis of 5-bromo-4-chloro-3-indoylphosphate, (11), that forms the indigo derivative, (12), represent biocatalytic transformations that precipitate an insoluble product on the surface [38,39]. Similarly,
horseradish peroxidase, HRP, mediated polymerization of phenol yields a polymer

Electronic Transduction of DNA Sensing Processes

57

Figure 7 Amperometric responses of the (T)2530 functionalized redox-hydrogel, (4),


associated with the electrode upon (a) hybridization with (G)2530-HRP, (b) hybridization
with (A)2530-HRP: amperometric responses originate from the electrobiocatalyzed reduction of H2O2 by the oligonucleotide-HRP conjugate. (Adapted from Ref. 32 with permission.)

lm on surfaces [37]. The insoluble product generated on the electrode insulates


the conductive support and introduces a barrier for interfacial electron transfer.
Faradaic impedance spectroscopy was found to be a sensitive transduction method
to probe the formation of the insoluble product on the conductive support [34].
Similarly, microgravimetric, quartz-crystal-microbalance measurements were
used to probe the formation of the precipitate on an Au-quartz piezoelectric crystal
[34]. These amplication routes and the respective electronic transduction methods were used to develop enzyme-sensing electrodes [34a,b], and immunosensors
for the amplied detection of an antibody [34c].
Figure 9 outlines the method for the amplied sensing of DNA by the biocatalyzed precipitation of an insoluble product on the transducer [33]. A primer
oligonucleotide (13) monolayer, complementary to a part of the target-analyteDNA, is assembled on the electrode and acts as the sensing interface. Interaction
with the analyte-DNA, (14), reacted with the biotinylated oligonucleotide, (15),
results in the double-stranded (ds) assembly. This surface-bound oligonucleotideassembly is further reacted with the avidin-HRP biocatalytic conjugate, (16).

Figure 8 Biocatalyzed transformations leading to the precipitation of insoluble products


on the transducer interfaces.

Electronic Transduction of DNA Sensing Processes

59

Figure 9 Stepwise assembly of a DNA-sensing electrode by the functionalization of the


conductive support with the sensing oligonucleotide (13), interaction with the target DNA
(14) pre-hybridized with the biotinylated oligonucleotide (15), and the association of the
avidin-HRP conjugate (16). The biocatalyzed precipitation of the insoluble product (8) is
used as an amplication route for the electrochemical transduction of the biorecognition
event.

Note that the enzyme conjugate binds to the interface only if the primary target
DNA was linked to the sensing interface. In the presence of H2O2 and 4-chloro1-naphthol, (7), the biocatalyzed precipitation of (8) amplies the primary sensing of DNA. That is, a single recognition event of the target DNA results in a biocatalytic amplication cascade of the insoluble product (8). One expects that in
the presence of Fe(CN)63/Fe(CN)64 as a redox label, the assembly of the dscomplex between the sensing interface and the analyte (14) will increase the electron transfer resistance at the electrode, due to the electrostatic repulsion of the

60

Willner et al.

redox label. Further binding of the biotinylated oligonucleotide (15) and then the
avidin-HRP conjugate (16) are anticipated to increase the electron transfer resistance due to the further electrostatic repulsion of the redox-label and the
hydrophobic insulation of the electrode interface, respectively. The biocatalyzed
precipitation of (8) on the electrode is expected to form an insulating layer on the
electrode that perturbs the interfacial electron transfer and results in an increase
in the electron transfer resistance. The latter process is time-dependent, and the
electron transfer resistance at the electrode will increase as precipitation proceeds. The formation of the precipitate on the transducer may be sensed also by
microgravimetric quartz-crystal-microbalance analyses. It should be noted that
the amount of ds-assembly formed on the respective transducers, and consequently, the amount of associated avidin-HRP conjugate and the resulting insoluble precipitate, are a function of the target-DNA concentration in the analyte
sample. Thus, monitoring the electron transfer resistance at the electrode on the
biocatalyzed precipitation of (8) for a xed time-interval in the presence of different concentrations of (14), may provide a quantitative measure for the targetDNA in the sample. It should be noted that the gene (14) corresponds to one of
the characteristic mutants for the Tay-Sachs genetic disorder [40]. Thus, the
development of a sensing interface for this genetic disorder could represent a
generic methodology for other Tay-Sachs mutants or genetic disorders.
The thiophosphate-thymine-tagged oligonucleotide (13) was assembled on
an Au-electrode. The surface-coverage of the sensing oligonucleotide monolayer
was determined by microgravimetric measurements as well as by chronocoulometry [33] using Ru(NH3)63+ as a redox-label to be 1.5 1011 molecm2. Figure
10(A) shows the impedance features of the electrode interface upon the buildup of
the double-stranded assembly on the conductive support, using Fe(CN)63/
Fe(CN)64 as a redox-probe. While the bare electrode shows the impedance spectrum depicted in curve (a), the assembly of the sensing interface modied with
(13), and then the formation of the ds-complex with the analyte (14) and the
biotinylated oligonucleotide (15), yield the impedance spectra shown in curves (b)
and (c), respectively. The respective semicircle diameters correspond to the interfacial electron-transfer resistances Ret. It can be seen that the electron-transfer
resistance increases upon the buildup of the biotinylated oligonucleotide-DNA
assembly. For example, for the (13)-functionalized electrode Ret = 1.1 k, Ret
increases to approximately 2 k upon the association of the complex between (14)
and the biotinylated oligonucleotide (15). These results are consistent with the fact
that the negative charge associated with the phosphate groups of the different
oligonucleotides increases upon the two-step organization of the assembly. This
results in the enhanced electrostatic repulsion of the redox probe and introduces
higher interfacial electron-transfer resistances. Figure 10B shows the impedance
spectra of the bifunctional double-stranded assembly consisting of the target DNA,
(14), linked to the sensing interface and the biotinylated oligonucleotide, (15),

Figure 10 Faradaic impedance spectra corresponding to: (A) The stepwise assembly of
the bifunctional double-stranded oligonucleotide-DNA assembly: (a) a bare Au-electrode;
(b) after deposition of the sensing oligonucleotide (13) on the interface; (c) after the formation of the complex (2 h interaction) between the sensing interface and the target DNA analyte, (14), 5.8 107 mgmL1, pre-hybridized with the biotinylated oligonucleotide (15),
2.6 106 mgmL1. (B) Stepwise deposition of the avidin-HRP conjugate and biocatalyzed
precipitation of (8) onto the electrode: (c) Bifunctional ds-assembly formed between the
complex (14)/(15) and the sensing interface; (d) After the deposition of the avidin-HRP,
(16), 1 108 mgmL1, for 3 h; (e) After the precipitation of (8) onto the electrode in the
presence of (7), 5 103 M, and H2O2, 5 103 M, for 20 min. Inset: Electron transfer resistance observed at the sensing electrode upon the precipitation of the insoluble product (8)
where the ds-oligonucleotide interface was constructed using different bulk concentrations
of the DNA-analyte (14). All measurements were performed in a 0.1 M phosphate buffer,
pH 7.2, in the presence of [Fe(CN)63/4] (5 103 M; 1:1) as a redox probe.

62

Willner et al.

before (curve c) and after (curve d) interaction with the avidin-HRP conjugate,
(16). Upon the association of the avidin-HRP biocatalytic conjugate to the layer, a
considerable increase in the electron-transfer resistance is observed due to the partial insulation of the electrode by the proteins. In the presence of H2O2 and the substrate (7), biocatalytic precipitation of the product (8) onto the electrode occurs.
This insulates the conductive support, resulting in a very high increase in the electron-transfer resistance (curve e; Ret = 17 k). Note the difference in the scales of
the Zre and Zim axes of parts A and B of Figure 10. The association of the avidinHRP conjugate to the oligonucleotide-DNA assembly, and the precipitation of the
product, induce an approximately 10-fold increase in the interfacial electron-transfer resistance as compared to that for the changes that occurred upon the formation
of the ds-assembly between the sensing interface and the complex between the
DNA analyte (14) and the biotin-labeled oligonucleotide (15). It should be noted
that the two parameters controlling the sensitivity of the DNA-sensing devices are
the time of incubation of the (13)-functionalized monolayer electrode with the
complex between the analyte DNA (14) and the biotinylated oligonucleotide (15),
and, more important, the time interval used to precipitate the product (8) by the
avidin-HRP biocatalytic conjugate. Figure 10(B) (inset) shows the electron-transfer resistance at the sensing interface upon precipitation of the insoluble product at
different concentrations of the analyte DNA (14). It is evident that as the bulk concentration of the DNA is lowered, the observed electron-transfer resistance
decreases as a result of the precipitation of the insoluble product. This is consistent
with the fact that lower bulk concentrations of (14) yield a lower coverage of the
(14) and biotin-labeled oligonucleotide complex on the sensing interface. This
results in lower coverage of the interface with avidin-HRP (16), and consequently,
a decreased efciency in the deposition of the insoluble product (8) is observed.
Using this conguration, and upon precipitation of (8) for 40 min, we were able to
sense (14) at a concentration of 2 108 mgmL1, Ret = 7.9 k. It should be noted
that upon the application of longer precipitation time intervals, the sensitivity of
the analysis could be enhanced.
Control experiments show that the oligonucleotide-sensing assembly
reveals high specicity and selectivity. Treatment of the (13)-functionalized electrode with the HRP conjugate (16), but without the interaction with the DNA analyte (14), yields only a minute change in the electron-transfer resistance. Also, the
sensing interface, for example, the (13)-functionalized monolayer electrode, was
interacted with a solution that included the DNA fragment (14a) and the biotinlabeled oligonucleotide (15). The oligonucleotide (14a) corresponds to the normal gene sequence in which the 7-base mutation leads to the Tay-Sachs genetic
disorder. After treatment of the sensing interface with the complex between (14a)
and (15), the system was subjected to the biocatalytic precipitation process using

Electronic Transduction of DNA Sensing Processes

63

the avidin-HRP conjugate (16). No noticeable changes in the electron-transfer


resistances at the electrode were observed, implying that the lack of formation of
the ds-oligonucleotide-DNA (analyte) complex with the sensing interface prevented the subsequent formation of the precipitate layer on the electrode.
The amplied sensing of the target DNA (14) through the biocatalyzed precipitation of (8) can also be transduced by microgravimetric quartz-crystalmicrobalance measurements, Figure 11. An Au-quartz crystal (9 MHz) modied
with the (13)-oligonucleotide sensing monolayer, was interacted with the target
DNA, (14), 5.8 107 mgmL1, that was pre-hybridized with the biotinylated
oligonucleotide, (15), 2.6 106 mgmL1. The resulting tri-component ds-assembly on the Au-quartz crystal was then reacted with the avidin-HRP conjugate
(16). Figure 11, curve (a), shows the crystal frequency changes resulting upon the
biocatalyzed precipitation of (8). The crystal frequency decreases by approximately 200 Hz upon precipitation of (8), for a time-interval of approximately 35
minutes, implying a mass increase on the crystal. Figure 11, curve (b), shows an
identical experiment where the sensing interface was interacted with the normal
gene (14a), pre-hybridized with (15) and the avidin-HRP (16). No changes in the
crystal frequency are observed upon an attempt to induce the biocatalyzed precipitation of (8). These results clearly indicate that the complex (14a)/(15) does
not bind to the sensing interface and that the amplied sensing of (14) is specic.
C.

Amplied Electronic Transduction of DNA Sensing Using


Functionalized Liposomes

Liposomes exhibiting average sizes of 200 nm in diameter represent micromembrane systems. Negatively charged liposomes associated with a surface generate a
charged interface that repels negatively charged redox-species and thus increase
the interfacial electron transfer resistance. Similarly, the association of the highmolecular-weight liposomes to a piezoelectric crystal is anticipated to increase the
weight of the crystal, and thus induce a substantial decrease in the crystal frequency. Figure 12(A) outlines one conguration for the amplied sensing of DNA
using functionalized liposomes [41,42]. The primer oligonucleotide (17) is assembled on the electrode or the Au-quartz crystal. The target analyte (18) binds to the
sensing interface and forms the ds-assembly. The secondary association of the
(19)-oligonucleotide-functionalized negatively charged liposome yields the threecomponent ds-complex with the liposomes linked to the interface. Electrostatic
repulsion of a negatively charged redox-probe (Fe(CN)63/Fe(CN)64) will introduce an increase in the interfacial electron-transfer resistance, Ret, in the impedance spectrum. Alternatively, the organization of the ds-assembly on an Au-quartz
crystal will enable the microgravimetric transduction of the association of the

64

Willner et al.

Figure 11 Time-dependent frequency changes resulting upon the biocatalyzed precipitation of (8) onto an Au-quartz crystal modied with the (13)-oligonucleotide sensing
monolayer, which was interacted with the target DNA, (14), 5.8 107 mgmL1 that was
pre-hybridized with the biotinylated oligonucleotide, (15), 2.6 106 mgmL1. The resulting tri-component ds-assembly on the Au-quartz crystal was then reacted with the avidinHRP conjugate, (16). Curve (a) shows the frequency changes upon the precipitation of (8)
in the presence of (7), 5 103 M, and H2O2, 5 103 M. Curve (b) shows the frequency in
the similar experiment where the sensing interface was interacted with the normal gene
(14a) pre-hybridized with (15) and interacted with avidin-HRP, (16).

membrane-mimetic units to the sensing interface. Figure 12(B) shows a dendritictype amplication of DNA sensing using functionalized liposomes. The primer
(17) is assembled on the Au-electrode or the Au-quartz crystal, and the sensing
interfaces are interacted with the target DNA (18), pre-hybridized with the biotinylated oligonucleotide, (20). The formation of the three-component ds-assembly is
then amplied by the association of avidin and the biotin-functionalized liposomes, (21). The rst amplication cycle can then be amplied by a dendritic-type
amplication using the avidin and the biotin-functionalized liposomes.
The amplied electrical transduction of the sensing of the target DNA (18) by
the (19)-functionalized liposomes, according to Fig. 12(A), was studied by faradaic

Figure 12 Amplied assay of a target DNA by: (A) An oligonucleotide-functionalized liposome. (B) An avidin/biotin-functionalized liposome.

66

Willner et al.

impedance spectroscopy [41]. The electrode was functionalized with the primer
oligonucleotide, (17), with a surface coverage corresponding to 1 1011
molecm2. The resulting monolayer-functionalized electrode was interacted with
the analyte DNA, (18), to yield the double-stranded assembly on the electrode surface. The resulting electrode interface was then treated with the oligonucleotide(17)-labeled liposome. The oligonucleotide (19) is complementary to the residual
base-sequence of the analyte (18). Thus, a liposome-linked three-component double-stranded assembly, consisting of the primer, (17), the analyte, (18), and the liposome tagged with (19), is generated on the electrode support. The (19)-labeled liposomes are negatively charged in order to eliminate nonspecic adsorption of the
liposomes onto the sensing interface. The oligonucleotide-functionalized liposomes were prepared by the assembly of liposomes that are composed of phosphatidic acid, phosphatidyl choline, maleimide-phosphatidylethanolamine, cholesterol (marked with 3H-cholesterol, 45 Cimole1) at a ratio of 79:20:1:0.1, that were
modied with (19) (4C, 20 h) and were puried by chromatography (Sephadex
G-75). The surface coverage of the liposomes with (19), (5060 oligonucleotide
units per liposome) was determined by reacting the resulting liposomes with Oligreen (Molecular Probes) and following the uorescence intensity of the resulting
liposomes suspension at ex = 480 nm. The size of the liposomes was determined by
dynamic light-scattering and corresponded to 220 20 nm. The liposomes associated with the electrode support represent giant negatively charged interfaces
(negatively charged micromembranes) that electrostatically repel a negatively
charged redox-probe solubilized in the electrolyte solution. That is, the biorecognition event between the primer (17) and the analyte DNA, (18), is amplied upon
complexing the (19)-functionalized liposomes by the generation of a highly
charged microenvironment that repels the electroactive probe, Fe(CN)63/4, in
solution. This electrostatic repulsion of the redox-probe introduces a barrier for
interfacial electron transfer, and results in an interfacial electron transfer resistance
that can be assayed by faradaic impedance spectroscopy. Figure 13(A), p. 67,
shows the impedance spectra of the (17)-oligonucleotide-functionalized electrode,
curve (a), after hybridization with the analyte DNA, (18), at a bulk concentration
corresponding to 5 106 M, curve (b), and after interaction with the probing
oligonucleotide-(17)-functionalized liposomes, curve (c). A bare Au-electrode
exhibits an electron transfer resistance of 0.5 k, but the association of the primer
(17) onto the conducting support increases the electron transfer resistance to 3 k.
This is attributed to the electrostatic repulsion of the redox probe, Fe(CN)63/4, that
results in a barrier for the interfacial electron transfer. The formation of the doublestranded assembly with the analyte DNA, (18), increases the electron-transfer
resistance to Ret = 4.5 k. This is consistent with the fact that the higher negative
charge formed on the surface as a result of hybridization enhances the electrostatic

Electronic Transduction of DNA Sensing Processes

67

Figure 13 (A) Faradaic impedance spectra of (a) the (17)-functionalized Au electrode,


(b) after interaction of the sensing electrode with (18) (5 106 M, 15 min, 25C), and (c)
after interaction with the (19)-functionalized liposomes. Inset: Faradaic impedance spectra of (a) the (17)-modied electrode, (d) after interaction with (18a) (5 106 M), (e) after
treatment with (19)-functionalized liposomes. All measurements were performed in a 0.1
M phosphate buffer (pH 7.2) in the presence of [Fe(CN)6]3/[Fe(CN)6]4 (5 103 M, 1:1)
as a redox-probe. (B) Changes in the electron transfer resistance of the (17)-functionalized
electrode upon treatment with the analyte DNA, (18), at different concentrations and upon
the secondary amplication with the (19)-functionalized liposomes corresponding to the
difference in the electron transfer resistance, Ret, after amplication with the (19)-functionalized liposomes and the resistance of the (17)-functionalized electrode.

68

Willner et al.

repulsion of the electroactive species in solution. Binding of the (19)-modied liposomes introduces a very high electron transfer resistance corresponding to 15 k.
This result is attributed to the formation of a negatively charged micromembrane
upon the association of the liposomes to the ds-assembly. The resulting charged
interface strongly repels the redox-label from the electrode interface, resulting in a
high electron transfer resistance. A control experiment, where only (19) binds to the
ds-assembly of the primer oligonucleotide and the analyte-DNA introduces only a
small increase in the electron transfer resistance, Ret = 4.7 k, indicates that the
negatively charged liposome indeed amplies the electrostatic repulsion of the
redox label. A further control experiment, involving an attempt to sense the DNA
(18a), included a 6-base mutation relative to the analyte-DNA, (18). Figure 13(A),
curve (d), shows the impedance spectrum of the functionalized-electrode after its
treatment with the mutant (18a), and curve (e) shows the impedance spectrum of the
resulting electrode after treatment with the (19)-functionalized liposomes. The
interfacial electron transfer resistances are almost unchanged, implying that the
sensing interface is selective for the analyses of (18). The results also indicate that
no nonspecic association of (18a) or the (19)-functionalized liposomes on the
electrode takes place. This is attributed to the electrostatic repulsions existing
between these components and the sensing interface. The extent of increase in the
electron transfer resistances upon the binding of the analyte-DNA, and the secondary association of the (19)-modied liposomes, is controlled by the bulk concentration of the analyte DNA, (18), Figure 13(B). The lower limit for analyzing the analyte DNA is 1.2 1012M.
The amplied sensing of (18) by the (19)-labeled liposomes was also transduced by microgravimetric quartz-crystal-microbalance measurements [42]. The
sensing interface of oligonucleotide (17) was assembled on an Au-quartz crystal (9
MHz). Figure 14(A) exemplies the crystal frequency changes upon the amplied
sensing of (18). Interaction of the sensing interface with the analyte (18), 5 106 M
(step a) results in a frequency change of f = 17 Hz, implying a surface coverage of
the analyte as a result of hybridization that corresponds to 1.2 1011 molecm2.
Reaction of the resulting interface with the (19)-labeled liposomes (step b), yields a
frequency change that corresponds to f = 120 Hz. Figure 14(A) shows also the
control experiment, where the sensing interface is interacted with the non-complementary DNA, (18a), (step c) and then with the (19)-labeled liposomes (step d). The
crystal frequency is almost unaltered, implying that the sensing protocol reveals
high selectivity. The extent of the association of the (19)-labeled liposomes to the
sensing array is controlled by the amount of hybridized (18) that forms the dsassembly. In turn, the amount of (18) linked to the surface is dependent on the bulk
concentration of the analyte DNA in the sample. For example, when the bulk DNA
concentration is 5 109 M, the association of the (19)-tagged liposomes to the

Electronic Transduction of DNA Sensing Processes

69

interface results in a frequency change of f = 70 Hz, Figure 14(A) (inset). The


lower sensitivity limit for the piezoelectric transduction of the amplied sensing of
(18) by the (19)-labeled liposomes was estimated to be 5 1012 M, where a frequency change of f = 20 Hz was observed after the amplication step.
The electrochemical transduction (impedance spectroscopy) of the dendritic
type amplication of the sensing of the analyte DNA (18), using the negatively
charged biotin-labeled liposomes, (21), is depicted in Figure 15(A). The (17)-functionalized interface exhibits an electron transfer resistance corresponding to 3 k,
curve (a), and upon the formation of the ds-assembly with the analyte DNA (18),
complexed to the biotinylated oligonucleotide, (20), the electron transfer resistance increases to Ret = 4.8 k, curve (b). Association of avidin to the interface further increases the electron transfer resistance to 7.6 k, as a result of the hydrophobic, insulating features of the protein, curve (c). Association of the biotin-labeled
liposome, (21), to the surface, substantially increases the electron transfer resistance, Ret = 14.8 k, curve (d). The enhanced electron transfer resistance is due to
the electrostatic repulsion of the electroactive species in solution by the charged
membrane interface. The sensing of the target-DNA can be further amplied by
the application of a second step of association (dendritic amplication) of the
avidin-biotinylated liposomes that enhance the electron transfer resistances to 17
k and 20 k, curves (e) and (f), respectively. In a control experiment, the sensing
interface was interacted with the non-complementary DNA, (18a), 5 106 M that
was pretreated with (20), and subsequently treated with avidin and the biotinylated
liposome, (21). A minute increase in the electron-transfer resistance corresponding to Ret = 3.4 k was observed, attributed to the nonspecic adsorption of avidin
to the sensing interface. The increase in the electron-transfer resistances at the electrode upon binding of avidin and the biotin-labeled liposome, (21), are controlled
by the bulk concentration of the analyte-DNA, (18), in the sample, Figure 15(B).
Using a double-step avidin/biotin-labeled-liposome amplication pathway, analyte-DNA concentrations as low as 5 1014 M could be detected.
Similarly, the dendritic-type amplication of the analyte DNA, (18), by the
biotin-labeled liposome can be transduced by microgravimetric quartz-crystal
microbalance measurements. Figure 14(B) shows the two-step amplied sensing
of (18), 5 106 M. Association of the (18)-analyte/20 ds-system to the sensing
interface results in a frequency decrease of approximately 25 Hz (step a). Binding of avidin to the biotinylated assembly yields a frequency change of f ~ 50
Hz (step b). Linkage of the biotin-tagged liposome to the system amplies the primary association of (18), and a very high frequency change, f ~ 500 Hz, is
observed (step c). Additional treatment of the interface with avidin, f ~ 50 Hz
(step d), and then with the biotin-labeled liposome (step e) results in a second
amplication corresponding to f = 690 Hz. Note that the amplication in the

70

Willner et al.

Electronic Transduction of DNA Sensing Processes

71

second step is higher than that in the rst step, due to the multiligation afnity of
avidin for the biotinylated liposome. The sensing of (18) is specic, Figure 14(B).
Treatment of the sensing interface with the noncomplementary DNA, 18a/20
complex, does not yield any signicant frequency change (step f) and subsequent
interaction of the resulting assembly with avidin and the biotin-tagged liposome,
(21), results in a frequency change of only approximately 30 Hz, (steps g and h,
respectively) that is attributed to the nonspecic association of the liposome to the
interface. Using the dendritic amplication route, the lower sensitivity limit for
the sensing of (18) is 1 1013 M (or 1 1016 molmL1). Note that by additional
binding steps of the avidin-biotinylated liposome, the sensitivity of the analysis
could be further enhanced.

D.

Amplied DNA Analysis by Dendritic Nanoparticle Arrays

The plasmon absorbance of Au-nanoparticles has been employed as an optical


probe for analyzing DNA [43]. Oligonucleotide-functionalized-Au-nanoparticles
exhibit a red color due to the plasmon excitation (for example, particles with an
average diameter of 12 1 nm reveal the plasmon absorbance at = 526 nm).
Hybridization of two-oligonucleotide fragments functionalized by Au-nanoparticles with the analyte DNA led to a red-shifted coupled interparticle plasmon
absorbance and the appearance of a blue color.
Oligonucleotide-functionalized Au-nanoparticles allow the formation of
double-stranded oligonucleotide-DNA-crosslinked multilayers on surfaces
[44,45]. This enabled the development of DNA sensing paths on piezoelectric

Figure 14 (A) Time-dependent frequency changes of the (17)-functionalized Au-quartzcrystal upon: (a) Interaction with (18), 5 106 M. (b) After interaction of the resulting electrode with the (19)-functionalized liposomes. (c) Treatment of the sensing crystal with (18a),
5 106 M. (d) Treatment of the resulting crystal with the (19)-labeled liposomes. Inset:
Time-dependent frequency changes resulting (e) Upon treatment of the (17)-functionalized
Au-quartz-crystal with (18), 5 109 M. (f) Subsequent treatment of the resulting electrode
with the (19)-tagged liposomes. (B) Time-dependent frequency changes of the Au-quartz
crystal upon: (a) Interaction of the sensing interface with (18), 5 106 M, and (18)/(20)-double-stranded complex. (b) As a result of the reaction of the resulting interface with avidin, 2.5
gmL1. (c) Upon reacting the resulting assembly with the biotin-labeled liposomes, (21).
(d) Step (b) repeated; (e) step (c) repeated. (f) Treatment of the sensing interface with the
(18a)/(20) double-stranded complex. (g) Interaction of the resulting interface with avidin;
and (h) reacting the resulting interface with biotin-labeled liposomes, (21).

72

Willner et al.

Electronic Transduction of DNA Sensing Processes

73

crystals with the dendritic-type-amplied microgravimetric transduction of DNA


analysis [46]. Figure 16 shows the method for the microgravimetric analysis of
the DNA, (23). The primer (22) is assembled on an Au-quartz crystal and two
kinds of Au-functionalized nanoparticles were prepared. One kind of Aunanoparticle is functionalized with the 3-terminated oligonucleotide (24), which
is complementary to the 3-end of the analyte DNA, whereas the second kind of
Au-nanoparticle is functionalized with the 5-terminated oligonucleotide (22),
complementary to the 5-end of the analyte DNA. The double-stranded primeranalyte complex on the surface is interacted with the 3-terminated Au-nanoparticle, leading to the core layer of the nanoparticles. Further interaction of the functionalized surface with the analyte DNA, pretreated with the 5-terminated
Au-nanoparticles, yields the second generation of the nanoparticles. Figure
17(A), curve (a), shows the frequency changes of the (22)-functionalized Auquartz crystal (9 MHz, AT-cut) upon interaction with the DNA (23), 2 108 M.
A frequency decrease of approximately 9 Hz is observed. Treatment of the interface with the (24)-functionalized Au-nanoparticles leads to a further decrease of
the crystal frequency, approximately 60 Hz, Figure 17(A), curve (b). The binding
of the Au-nanoparticles represent the rst amplication step for analyzing (23).
The mutant (23a) is effectively differentiated by the (22)-functionalized Auquartz crystal, as evident in curves (c) and (d), Figure 17(A), that reveal no frequency changes of the crystal upon reaction with (23a) and then with the (22)functionalized-Au-nanoparticles, respectively.
The dendritic-type amplication of the primary Au-nanoparticle-amplied
sensing of (23) is depicted in Figure 17(B), and its inset. The primary amplication path enables the sensing of (23) in the concentration range of 2 108 M to
1 1010 M, where no noticeable frequency changes are observed upon the

Figure 15 (A) Faradaic impedance spectra of (a) The (17)-functionalized Au electrode.


(b) After interaction of the sensing electrode with (18) (5 106 M), which was pretreated
with (20) (1 105 M, 30 min, 25C). (c) After treatment of the resulting electrode with
avidin (2.5 gmL1). (d) After interaction with the biotinylated liposomes, (21). (e) After
treatment of the interface for a second time with avidin (2.5 gmL1), (f) After interaction
of the interface for a second time with the biotinylated liposomes, (21). Data were recorded
in 0.1 M phosphate buffer, pH = 7.2, in the presence of [Fe(CN)6]3/4, (1:1), as redoxprobe. (B) Calibration curve corresponding to the changes in the electron transfer resistances of the sensing electrode upon interaction with the analyte DNA, (18), at different
concentrations and enhancement of the sensing process by a double-step avidin/ biotinylated liposome amplication path. Ret corresponds to the difference in the electron transfer resistance after a double-step avidin/biotinylated liposome amplication and the electron resistance of the (17)-functionalized electrode.

Figure 16 Dendritic amplication of DNA-sensing by oligonucleotide-functionalized Au-nanoparticles.

74
Willner et al.

Electronic Transduction of DNA Sensing Processes

75

Figure 17 (A) Time-dependent frequency changes of a (22)-functionalized Au/quartz


crystal upon interaction with: (a) The analyte DNA, (23), 2 108 M. (b) After treatment
of the double-stranded assembly of (22) and (23) with the (24)-functionalized Au-nanoparticles. (c) After treatment of the sensing interface with (23a), 5 106 M. (d) After treatment of the resulting surface with the (24)-functionalized Au-nanoparticles. (B) Frequency
changes of the (22)-functionalized Au/quartz crystal upon the dendritic amplied sensing
of different concentrations of the analyte DNA, (23): (a) Upon the association of the analyte (23) with the sensing (22)-interface; (b) Upon the amplication of the primary doublestranded assembly of (22) and (23) with (24)-functionalized Au-nanoparticles; (c) Upon the
dendritic amplication of the primary (22)-(23)-(24)-Au-nanoparticle array with the (23)(22)-Au-nanoparticle probe.

76

Willner et al.

hybridization of (23). The formation of the second generation of Au-nanoparticles reveal a dendritic-type, nonlinear amplication of the sensing of (23), Figure
17(B), inset. For example, while the primary association of the (24)-functionalized Au-nanoparticles to the (22)-modied Au-quartz crystal treated with 2
108, 1 108, 1 109 and 1 1010 M of (23) yields frequency changes of 70,
40, 12, and 8 Hz, respectively (Figure 17(B), curve b), the formation of the
second generation of the (22)-Au-nanoparticle yields a secondary amplication
characterized by a nonlinear change in the crystal frequency that corresponds to
250, 115, 45, 20 Hz, respectively (curve c). Thus, for a concentration of 2
108 M of (23), the association of the core-generation of the Au-nanoparticles
induces an eight-fold amplication in the transduced signal, whereas the formation of the second Au-nanoparticle generation stimulates a 28-fold enhancement
in the transduced signal.

E. Two-Step Amplication of DNA-Sensing Using Biotin/HRP


Labeled Liposomes
The ability to amplify DNA sensing events by the application of an enzyme-conjugate that biocatalyzes the precipitation of an insoluble product on the transducer, and the second means for the amplication of DNA sensing by the use of
charged labeled liposomes, suggests that enhanced amplication could be accomplished by using functionalized enzyme-liposome conjugates [31]. This is outlined in Fig. 18, where a biotin-labeled horseradish peroxidase liposome conjugate, HRP-functionalized liposome, (25), is used as the amplifying probe. The
primer is organized on the electrode as the sensing interface, and the analyte
forms a double-stranded assembly on the transducer. Secondary binding of a
biotin-labeled oligonucleotide, complementary to the other end of the analyte
DNA, enables the association of avidin. This enables the rst amplication of the
sensing event through the binding of the biotin-tagged HRP-functionalized, negatively charged liposome. The secondary amplication process is stimulated by
the biocatalyzed HRP-stimulated oxidation of 4-chloro-1-naphthol, (7), by H2O2
to yield the insoluble product (8) on the transducer. The two-phase amplication
process for sensing an analyte DNA was applied to sense one of the Tay-Sachs
genetic disorder mutants, (27). A primer, (26), consisting of a thiophosphate
thymine tag linked to an oligonucleotide that is partially complementary to the
analyte-mutant, (27), is used to generate the sensing interface. The thiophosphate
thymine tag is used to link the oligonucleotide onto the Au-electrode surface.
Figure 19(A) shows the faradaic impedance spectra resulting upon the buildup of
the (26)-functionalized sensing interface (curve b); formation of the double-

Figure 18 Amplied detection of a target DNA by biotin-tagged-HRP-functionalized liposomes and the biocatalyzed precipitation of the
insoluble product, (8), on the electrode.

Electronic Transduction of DNA Sensing Processes


77

78

Willner et al.

stranded assembly between the analyte-DNA (27) and the sensing interface (26)
(curve c); the association of the biotinylated oligonucleotide, (28), complementary to the analyte DNA (curve d); the linked avidin (curve e); and, nally, the
association of the biotin/HRP-functionalized liposome, (25), (curve f). The interfacial electron transfer resistance increases constantly from the value Ret = 0.2 k
for the (26)-functionalized electrode to the value Ret = 1.5 k, resulting in the
binding of the liposome, (25). This is consistent with the fact that the binding of
the liposomes repels the negatively charged redox-label in the electrolyte solution, resulting in an enhanced interfacial electron transfer resistance. Figure 19(B)
shows the faradaic impedance spectra of the liposome-functionalized electrode
upon the biocatalyzed precipitation of (8) for different time-intervals. While the
electron transfer resistance increases to 6.5 k and 22 k after the precipitation
of (8) for 5 and 10 minutes, respectively, it reaches a saturation value of Ret 145
k after the precipitation of (8) for 20 minutes. It can be seen that the amplication of the sensing process through the biocatalyzed precipitation of the insoluble
product (8) proceeds nonlinearly with time. This is attributed to the fact that at the
initial phases of precipitation, the electrode surface still includes substantial base
domains that enable electron transfer with the redox-label in solution. As precipitation proceeds, the electrode is fully blocked toward the interfacial electron
transfer.
Using the HRP-functionalized liposomes, the analyte (27) could be sensed
without any attempt to optimize the process, with a sensitivity that corresponds to
1 1013 M. The sensing process is very specic, and the interaction of the sensing interface with the normal gene, (27a), at a concentration that corresponds to
1.3 108 M, followed by a sequence of amplications with the HRP-functionalized liposomes, and the biocatalyzed precipitation of (8) results in an increase in
the electron transfer resistance of only 0.2 k (Note that the same protocol for the
analyte (27), 6.5 1012 M, results in an increase in the electron transfer resistance of 2.7 k.) Thus, the system is free from the binding of the noncomplementary normal gene, (27a), and from the nonspecic binding of the labeled liposomes. This might be attributed to the electrostatic repulsion of the liposomes by
the negatively charged oligonucleotide sensing interface.
Figure 20(A) shows the E-t chronopotentiometric transients upon the
buildup of the layered assembly shown in Fig. 18. The assembly of the primer,
(26), on the surface is accompanied with an overpotential for the reduction of the
redox-probe as a result of its electrostatic repulsion. The hybridization with the
target DNA, (27), and the subsequent coupling of the biotinylated-oligonucleotide, (28), further repel the redox-probe [Fe(CN)6]3/4, thus enhancing the
overpotential for the electron transfer at the electrode. The buildup of avidin and
the biotinylated liposomes insulates the electrode interface and results in an addi-

Figure 19 (A) Faradaic impedance spectra corresponding to: (a) A bare Au electrode.
(b) The (26)-functionalized electrode. (c) After interaction of the sensing electrode with
the analyte DNA, (27), 6.5 1012 M. (d) After treatment of the sensing electrode with the
(29)-biotinylated-oligonucleotide, (28), 7 108 M. (e) After interaction with avidin, 200
ngmL1. (f) After interaction of the interface with the biotin-HRP-labeled liposomes, 1.46
1011 M. (B) Faradaic impedance spectra corresponding to the electrode modied
according to steps (a)(f) in part A, and upon the biocatalyzed precipitation of (8) for different time intervals: (a) Before precipitation; (b), (c), and (d) After 5, 10, and 20 minutes
of the precipitation of (8), respectively. Inset: curves (a)(c) enlarged. In all measurements
[Fe(CN)6]3/4, 10 mM, is used as a redox probe.

80

Willner et al.

tional increase in the overpotential. Figure 20(B) shows the plot of electron transfer resistances, Ret, of the electrode upon the buildup of the DNA-liposomes
assembly, derived from the faradaic impedance measurements, and the electrode
resistances, R, at the different steps of formation of the assembly, derived from
the chronopotentiometric experiments. A good correlation between the two values exists.

V. BIOCATALYZED TRANSFORMATIONS ON NUCLEIC


ACIDFUNCTIONALIZED SURFACES
Different biocatalyzed transformations proceed with single-stranded or doublestranded DNA assemblies. The ligation of nucleic acids, replication of doublestranded assemblies, or the sequence-specied scission of the double-stranded
system by specic endonucleases are a few representative examples. Bioelectrocatalytic transformations occurring on surfaces provide additional nanometric
tools to manipulate and nanoengineer the nucleic acid surfaces and to amplify the
sensing processes by the replication route. The different electronic transduction
means that follow the surface functionalization by DNA may then be applied to
follow the biocatalyzed transformations occurring on the nucleic acidfunctionalized supports.

A.

Electronic Transduction of Biocatalytic Transformations


on Oligonucleotide-Functionalized Transducers

A series of biocatalyzed transformations involving nucleic acids that include the


surface-stimulated ligation, replication, and specic scission of nucleic acids by
a restriction enzyme were transduced electronically using faradaic impedance and
QCM transduction tools [47].
The 18-mer oligonucleotide, (29), was assembled on an Au-electrode or on
an Au/quartz piezoelectric crystal (9 MHz, AT-cut) by the association of the
thymine thiophosphate-tag to the gold supports [13,40]. The surface coverage of
the oligonucleotide (29) was estimated to be 4.5 1011 molecm2. The resulting
(29)-functionalized surfaces were reacted with polynucleotide kinase, PNK, in
the presence of ATP to phosphorylate the 5 termini of the oligonucleotide-monolayer. The resulting interface was reacted with the oligonucleotide (30) in the
presence of ligase (Figure 21, see color plate), to induce the ligation of (30) to the
base oligonucleotide associated with the surfaces. Figure 22 shows the faradaic
impedance spectra observed upon performing the biocatalyzed transformations

Figure 20 (A) Chronopotentiometric transients in the presence of [Fe(CN)6]3/4, 10


mM, as a redox-probe corresponding to (a) The Au-bare electrode. (b) The oligonucleotide
(26)-functionalized electrode. (c) After interaction of the sensing electrode with the analyte
DNA, (27), 6.5 1012 M. (d) After treatment of the sensing electrode with the biotinylated-oligonucleotide, (28), 7 108 M. (e) After interaction with avidin, 200 ngmL1. (f)
After interaction of the interface with the biotin-HRP-labeled liposomes, 1.46 1011 M.
(B) The electron transfer resistances, Ret, derived from the Faradaic impedance spectra
shown in Figure 10(A) () and the electrode resistance, R, derived from the chronopotentiometric measurements shown in Figure 11(A) () measured at the different modication
steps of the electrode (a) to (f).

82

Willner et al.

on the nucleic acids associated with the electrode. The ligation of (30) results in
an increase in the interfacial electron transfer resistance from Ret = 0.44 k to Ret
= 1.33 k (Figure 22, curve (b)). This is consistent with the fact that the increase
of the negative charge associated with the electrode, as a result of ligation,
enhances the electrostatic repulsion of the redox-label, [Fe(CN)6]3/4, thus
increasing the interfacial electron transfer resistance. The resulting nucleic acid
associated with the interface was hybridized with the oligonucleotide (31), which
is complementary to a part of the nucleic acid associated with the solid supports.
The interfacial electron transfer resistance increases as a result of the hybridization of (31), Ret = 1.9 k, Figure 22, curve (c), consistent with the increase of the
negative charge associated with the electrode. The incomplete hybridization is
due to steric constraints on the electrode support that eliminate the formation of
ds-DNA with all of the nucleic acid components. Control experiments revealed
that no ligation occurred if the base oligonucleotide was not phosphorylated with
PNK prior to the ligation process.
The resulting assembly was then reacted with the mixture of phosphorylated bases, dNTP, in the presence of polymerase (Klenow fragment, DNA polymerase I). This yields an increase in the interfacial electron transfer resistance, Ret
= 3.1 k, as a result of the higher negative charge associated with the interface,
Figure 22, curve (d). Reaction of the assembly with the endonuclease restriction
enzyme Dra I that stimulates the specic scission of 5TTT/AAA3 sequence does
not yield any change in the impedance spectrum of the assembly. Reaction of the
resulting assembly with the endonuclease restriction enzyme Cfo I (Hha I) that
induces the specic scission of the 5GCG/C3 sequence results, however, in the
cleavage of the ds-assembly (Figure 21, see color plate). The resulting faradaic
impedance spectrum is shown in Figure 22, curve (e). The interfacial electron
transfer resistance decreases to Ret = 0.9 k. This is consistent with the fact that
removal of a major part of the ds-assembly and the negative charge associated
with it, by the endonuclease activity, reduces the barrier for electron transfer
between the redox-label and the electrode. The scission of the double-stranded
DNA yields a 5-phosphorylated primer on the electrode. Note that the interfacial
electron transfer resistance of the resulting electrode is higher than the electron
transfer resistance of the (29)-functionalized electrode despite the fact that the
endonuclease cleavage generates a shorter oligonucleotide than (29) on the electrode. Complementary microgravimetric quartz-crystal-microgravimetric experiments reveal that the Cfo I cleavage process proceeds with a yield of only 32%.
Thus, the higher interfacial electron transfer resistance after the treatment of the
surface with Cfo I is attributed to residual ds-replicated DNA on the electrode
support. The resulting interface was then reacted with the oligonucleotide (32) in
the presence of ligase to yield the original surface, Figure 22, curve (f), exhibiting an electron transfer resistance of Ret = 2.2 k. Further hybridization of (31)

Electronic Transduction of DNA Sensing Processes

83

Figure 22 Faradaic impedance spectra corresponding to the biocatalytic transformations


of nucleic acidfunctionalized electrodes: (a) The (29)-functionalized electrode; (b) after
ligation of (30), 1 104 M, with the (29)-functionalized electrode in the presence of ligase, 20 units, 37C, 30 min; (c) after hybridization of the resulting electrode with (31), 1
104 M, 2 h; (d) after replication of the double-stranded assembly in the presence of dNTP,
1 103 M, and polymerase, 3 units, 37C, 30 min; (e) after scission of the resulting assembly with endonuclease, Cfo I, 10 units, 37C, 1 h; (f) after ligation of the resulting interface
with (32), 6.5 105 M, in the presence of ligase, 20 units, 37C, 30 min; (g) after
hybridization of the assembly with (31), 1 104 M, for 2 h.

84

Willner et al.

with the ligated interface results in an additional increase in the electron transfer
resistance to Ret = 2.6 k, curve (g), Figure 22. The ligation of (32) to the interface and the hybridization of (31) with the interface yield, however, higher interfacial electron-transfer resistances than those observed for the originally functionalized electrodes, curves (b) and (c), respectively. This is consistent with the
fact that endonuclease-induced scission proceeds with a 32% yield, and thus the
secondary ligation and hybridization occurs on an interface that includes a partial
coverage of the polymerase-induced replicated double-stranded DNA. The negative charge associated with these latter components introduces the higher interfacial electron transfer resistances observed in the second cycle of the biocatalytic
transformations.
It should be noted that the biocatalytic reactions performed directly on the
DNA-sensing interface followed by the electronically transduced signals can be
applied for cyclic usage of the sensing interface with regeneration between the
usage steps. The electronically transduced processes allow control of the reactions to complete the desired biocatalytic process.

B. Amplied Detection of Viral DNA/RNA by Biocatalyzed


Replication and Precipitation of an Insoluble Product
on the Transducers
The ability to biocatalyze the replication of nucleic acids on surfaces (see Sec. V.
A) and the amplication of DNA recognition events on electronic transducers by
the biocatalyzed precipitation of an insoluble product on the transducer in the
presence of an enzyme conjugate (see Sec. IV. B), suggest that amplied detection of DNA on surfaces may be achieved by the replication of DNA while tagging. Indeed, the amplied detection of genomic DNA or RNA was accomplished
by polymerase-induced replication of the biotin-tagged DNA or RNA, followed
by the biocatalyzed precipitation of an insoluble product on the transducer [48].
The method for analysis of the target DNA or RNA is depicted in Fig. 23.
The primer thiolated oligonucleotide (33), complementary to a segment of the target M13 mp8 DNA, is assembled on an Au-electrode or an Au-quartz crystal
through a thiol functional group [30]. The sensing interface is then reacted with
the analyte DNA of M13 mp8 (+) strand, (34), and the resulting complex on the
transducer is interacted with dATP, dGTP, dTTP, dCTP and biotinylated-dCTP
(ratio 1:1:1:2/3:1/3, nucleotides concentration of 1 mM) in the presence of DNA
polymerase I, Klenow fragment. Replication of the target DNA provides the rst
amplication step for the analyte-gene. Polymerase introduces biotin tags to the
double-stranded assembly, thus providing a high number of docking sites for the
binding of the avidinalkaline phosphatase conjugate. The associated enzyme

Electronic Transduction of DNA Sensing Processes

85

biocatalyzes the oxidative hydrolysis of 5-bromo-4-chloro-3-indolylphosphate


(11) to form the insoluble indigo product (12), which precipitates on the transducer, thus providing a second amplication step for the analysis of the target
DNA [30]. The synthesized strand on the electrode is anticipated to attract a positively charged redox-label that can be assayed by chronocoulometry [24]. This
enables the replication process to be monitored on the surface continuously. The
negatively charged double-stranded replicated assembly is also anticipated to
repel a negatively charged redox-probe, i.e. Fe(CN)63/Fe(CN)64, and to enhance
the electron-transfer resistance on the transducers surface. The barrier for electron-transfer to a negatively charged redox-label in solution is then assayed by
faradaic impedance spectroscopy [24]. Furthermore, hybridization, formation of
the double-stranded assembly, polymerization and precipitation of the insoluble
product (12) alter the mass on the transducer. Thus, the detection of the target
DNA can be assayed by microgravimetric analysis of the frequency change of a
piezoelectric crystal.
The coverage of the probe oligonucleotide on the transducer was determined by microgravimetric quartz-crystal-micro balance analysis and chronocoulometric experiments, using Ru(NH3)63+ as a redox-label [29] (see Sec. III).
Upon the interaction of the Au-electrode with the primer (33), 4.2 106 M, for
60 minutes, a surface with optimal surface coverage, corresponding to (6.3 0.3)
1011 molecm2, for the sensing of M13 was generated.
Figure 24 shows the chronocoulometric transients, in the presence of
Ru(NH3)63+, of the (33)-probe-functionalized-electrode, curve (b), and of the
sensing interface after hybridization with the analyte DNA (34) for periods of 1.5
and 4 hours, curves (c) and (d), respectively. After 4 hours of hybridization, the
charge associated with the linked redox-probe was estimated to be 54 C. Assuming that all of the Ru(NH3)63+ units linked to the hybridized analyte DNA communicate electrically with the electrode, the surface coverage of the M13 mp8
DNA on the surface is ca. 9.0 1013 molecm2. Thus, only 1.5% of the sensing
oligonucleotide units underwent hybridization with the viral DNA. This value of
surface coverage of the hybridized DNA is further supported by microgravimetric quartz-crystal-microbalance (QCM) measurements that reveal a frequency
change of f = 445 Hz upon the binding of the viral DNA to the surface. This
frequency change translates to a surface coverage of the hybridized DNA of 1.1
1012 molecm2, which correspond to hybridization to ca. 1.7% of the sensing
interface.
The increase in the charge associated with Ru(NH3)63+ linked to the doublestranded assembly as a result of the polymerase-induced replication of the analyte
DNA is shown in Figure 25. The charge increases with time, implying that polymerization occurs on the surface, and it levels off after ca. 60 minutes of polymerization. Note that the charge associated with the analyte DNA is 29.2 C, sub-

Figure 23 Amplied electronic transduction of viral DNA/RNA by the polymerase-induced or reverse transcriptionstimulated replication of DNA or RNA, respectively, and the biocatalyzed precipitation of an insoluble product on the transducers surface.

86
Willner et al.

Electronic Transduction of DNA Sensing Processes

87

Figure 24 Chronocoulometric transients for (a) A bare Au-electrode. (b) The oligonucleotide (33)-modied Au-electrode. (c) The (33)-modied Au-electrode after hybridization with M13, 2.3 109 M, for 1.5 hours. (d) Hybridization with M13, 2.3 109 M,
for 4 hours. All transients were recorded in the presence of Ru(NH3)63+, 5 105 M, in 10
mM Tris-buffer, pH = 7.4. Hybridizations were conducted in 0.1 M phosphate buffer, pH
= 7.5, that included 30% formamide, at room temperature.

stantially lower than the theoretical value of 54 C for full replication. Thus, the
replication led only to 54% of formation of the double-stranded assembly (on
average 3900 bases were incorporated over each analyte DNA). This has been
attributed to steric constraints for the formation of the fully replicated doublestranded assembly on the surface or to the interruption of the Klenow fragmentinduced polymerization that is known to occur at specic sites during M13 replication [49]. The partial polymerization on the surface is further reected by QCM
experiments that indicate that polymerization yields a frequency change of f =
195 Hz, whereas the attachment of the analyte-DNA to the surface results in a
frequency change of f = 445 Hz.
Figure 26 shows the faradaic impedance spectra (in the form of Nyquist

88

Willner et al.

Figure 25 The time-dependent changes of the charge associated with the polymeraseinduced replication of the double-stranded assembly on the M13 DNA. The changes in the
charge are measured by chronocoulometry using Ru(NH3)63+, 5 105 M, as a redox probe
in 10 mM Tris-buffer, pH = 7.4.

plots) of the oligonucleotide-functionalized electrode, curve (a); after its


hybridization with the analyte DNA, 2.3 109 M, curve (b); the polymeraseinduced formation of the double-stranded assembly, curve (c); after the association of the avidinalkaline phosphatase conjugate, curve (d); and the subsequent
biocatalyzed precipitation of (12) on the surface for 20 minutes, curve (e). The
electron transfer resistance, Ret, increases upon the binding of the virus DNA
from 3 k to 55 k. This is consistent with the fact that binding of the highmolecular-weight DNA electrostatically repels the negatively charged redoxlabel, Fe(CN)63/Fe(CN)64 from the electrode surface. The polymerase-mediated
replication of the complementary strand further increases the electron-transfer
resistance to ca. ca. Ret = 33 k. The binding of the conjugate avidinalkaline
phosphatase, and the subsequent biocatalyzed precipitation of (12) on the electrode, results in an insulating layer that introduces a barrier for the interfacial electron transfer, and the electron transfer resistance increases to Ret = 55 k. In the
electrochemical transduction, a change in the electron transfer resistance of Ret

Figure 26 Faradaic impedance spectra corresponding to (a) The (33)-modied electrode. (b) After hybridization with
M13 DNA, 2.3 109 M. (c) After the polymerase-induced replication and formation of the double-stranded assembly for
45 minutes. (d) After the binding of the avidin-alkaline phosphatase conjugate to the surface. (e) After the biocatalyzed precipitation of (12) for 20 minutes in the presence of (11), 2 103 M, in 0.1 M Tris-buffer, pH = 7.2.

Electronic Transduction of DNA Sensing Processes


89

90

Willner et al.

= 2.8 k is observed upon the analysis of the DNA, 2.0 1016 M, as a result of
the precipitation of (12). Control experiments revealed that the analysis of the target DNA of M13 phage is specic and foreign DNA did not yield any apparent
interference signal.
A similar approach [48] was used for the amplied sensing of the 11161
base RNA of vesicular stomatitis virus (VSV) [49], using reverse transcriptase
(enhanced avian reverse transcriptase) as the replication biocatalyst. The oligonucleotide, (35), was immobilized as the primer sensing interface on an Au-electrode or an Au-quartz crystal, 1.4 1011 molecm2, Fig. 23 [48]. Figure 27(A)
shows the Faradaic impedance spectra of the (35)-functionalized electrode, curve
(a); after the hybridization with the respective 1 1012 M, RNA, curve (b); after
the reverse transcription of the RNA in the presence of dATP, dGTP, dCTP,
dTTP, and biotinylated-dUTP (ratio 1:1:1:2/3:1/3, base concentration 1 mM),
curve (c); after the binding of the avidinalkaline phosphatase conjugate, curve
(d); and upon the biocatalyzed precipitation of (12) on the transducer, curve (e).
Each of these steps increases, as expected, the electron transfer resistance at the
electrode surface. For example, upon the analysis of 1 1012 M VSV-RNA, the
reverse transcription increases the interfacial electron transfer resistance by Ret
= 4.5 k relative to the previous modication step, and the precipitation of the
insoluble product (12) on the electrode increases the electron transfer resistance
by Ret = 14.0 k. The viral RNA could be analyzed with a detection limit that
corresponds to 1 1017 M. At this concentration, the hybridization process and
the reverse transcription of the VSV-RNA were invisible, yet the alkaline phosphatase precipitation of (12) on the electrode resulted in an amplication path and
the interfacial electron transfer resistance increased by Ret = 2.2 k. Control

Figure 27 (A) Faradaic impedance spectra corresponding to the amplied sensing of


the vesicular stomatitis virus (VSV) RNA: (a) The (35]-functionalized electrode. (b) After
hybridization with the VSV-RNA, 1 1012 M, in a solution consisting of 40 mM PIPES,
1 mM EDTA, 400 mM NaCl, 80% formamide, pH = 7.5. (c) After the reverse transcription for 45 minutes (80 UmL1) in the presence of dGTP, dATP, dCTP, dTTP and biotinylated dUTP (1:1:2/3:1:1/3 each base 1 mM) in a solution consisting of 50 mM Tris-buffer,
40 mM KCl, 8 mM MgCl2, pH = 8.3, at room temperature. (d) After the association of the
avidinalkaline phosphatase conjugate, bulk concentration 1 mM. (e) After the biocatalyzed precipitation of (12] for 20 minutes in the presence of (11], 2 103 M in Trisbuffer, pH = 7.5. (B) Time-dependent frequency changes of an Au/quartz crystal upon the
analysis of the VSV-RNA, 1 1012 M, (f) Frequency changes as a result of the replication of the bound RNA in the presence of reverse transcriptase and dGTP, dATP, dCTP,
dTTP and biotinylated-dUTP; (g) Upon the association of the avidinalkaline phosphatase
conjugate; (h) Upon the biocatalyzed precipitation of (12]. The conditions for the polymerization and biocatalyzed precipitation (12] are detailed in Part (A). Inset: Magnication of curve (f).

Electronic Transduction of DNA Sensing Processes

91

92

Willner et al.

experiments revealed that the interaction of the sensing interface with a foreign
RNA (yeast RNA of heterogeneous length of 2 kb7 kb), 1 109 M, followed by
an attempt to stimulate the reverse-transcription and the biocatalyzed precipitation of (12) resulted in a minute change in the electron transfer resistance at the
electrode, Ret = 0.3 k, indicating that the amplied detection of the VSV-RNA
is selective. Figure 27(B) shows the microgravimetric, quartz-crystal-microbalance analysis of the RNA. Hybridization with the VSV RNA, 1 1012 M, results
in a frequency change of f = 72 Hz that indicates a surface coverage of ca. 1%
of the sensing interface. The replication of the viral RNA in the presence of
dATP, dGTP, dTTP, dCTP, and biotinylated-dCTP (1:1:1:2/3:1/3 base concentration 1 mM) in the presence of reverse transcriptase, 80 UmL1, results in a frequency change of 26 Hz, curve (f), Figure 27(B). This frequency decrease translates to an average replication of the surface associated analyte RNA of 36%.
Binding of the avidin-alkaline phosphatase conjugate onto the surface is shown
in curve (g), f = 52 Hz, and the biocatalyzed precipitation of (12) results in a
signicant change in the crystal frequency that corresponds to ca. f = 400 Hz,
curve (h), Figure 27(B).

VI.

SPECIFICITY IN DNA DETECTION

The specic detection of DNA and the ability to detect single-base mismatches in
DNA is an important, challenging topic in DNA bioelectronics. In the different
systems described in sections IV and V, specicity was accomplished by tailoring the probe nucleic acid. The probe nucleic acid sequence usually includes the
minimal number of complementary bases to the target DNA to form a single double-stranded helical structure. Thus the respective mutants are unable to form a
complementary double-stranded assembly. Consequently, the substantially different thermodynamic stability of a double-stranded helical structure compared
with the hydrogen-bonded base pairs enables the selective differentiation of the
target DNA. The detection of a single-base mismatch in the target DNA is, however, more complex. The effect of temperature on the hybridization and melting
of double-stranded DNA systems and more complex biocatalytic transformations
were employed to detect single-base mismatches in DNA.

A.

Temperature-Controlled Detection of Base Mismatches in DNA

The different thermodynamic stabilities of fully matched and partially matched


double-stranded hybridized DNA were utilized to detect mismatches in the analyzed DNA, by controlling the hybridization temperature [50]. The 30-base primer
(36) was covalently linked to a polyacrylamide-acrylhydrazide and 1-vinylimida-

Electronic Transduction of DNA Sensing Processes

93

zole, complexed to [Os(4,4-dimethyl-2,2-bipyridine)Cl]+/2+ copolymer hydrogel


that was associated with a 7-m diameter carbon-ber electrode, Fig. 28. The complementary target DNA (37), the mutant that includes the four-base mismatch,
(37a), or the mutant with the single-base mismatch, (37b), were covalently linked,
through a NH2-12-methylene spacer tethered to the 5-termini of the oligonucleotides (37), (37a), (37b) to soy bean peroxidase, SBP. The covalent bonds
between the amino-alkyl-tethered-oligonucleotides and SBP were generated by the
oxidation of the glycoprotein with NaIO4, the subsequent formation of the respective Schiff-bases between the aldehyde-functionalized protein and the amino-functionalities of the oligonucleotide, followed by the reduction of the Schiff base units
with NaBH4. The redox-active hydrogel acts as an electrical contacting matrix
between the peroxidase and the electrode support (see Sec. IV.A). The electrobiocatalyzed reduction of H2O2 by SBP provides an amplication path for the formation of the double-stranded DNA between (36) and the oligonucleotide-functionalized SBP on the electrode support, Fig. 28. Hybridization of the electrode
functionalized with the primer (36) with (37), (37a), or (37b) at 25C results in
almost identical amperometric responses, indicating that at this hybridization temperature the target DNA (37) cannot be differentiated from the mutants (37a) or
(37b). At a hybridization temperature corresponding to 45C, the amperometric
response of the functionalized electrode treated with (37a) is very low and easily
differentiated from the amperometric responses of the (36)-modied electrode
treated with (37) or (37b), Figure 29(A). At a higher temperature corresponding to
57C, the amperometric responses of the four-base mutant (37a) and the singlebase mutant (37b) are low, whereas the current transduced by the sensing electrode
in the presence of the target DNA, (37), is high, Figure 29(B), indicating effective
hybridization. The temperature-controlled differentiation between the target DNA
and the four-base and single-base mutants (37a) and (37b), respectively, were
attributed to the different melting temperatures of the hybridized DNA. The calculated melting temperature of the four-base mismatched double-stranded assembly
between (36)/(37a) and (36)/(37) are 52.554.5 and 59.5C, respectively. Thus, the
delicate control of the hybridization temperature enables the selective hybridization of the target DNA (37) to the sensing interface and the amplied electronic
transduction of a single-base mismatch in the target DNA.
The experimental amperometric responses of the (36)-functionalized electrode were theoretically modeled. The time-dependence of the current evolution
has been attributed to the kinetics of the binding of the oligonucleotide-labeled
SBP that follows the diffusion-limited Langmuir equation, Eq. 8,
= max[1 exp(kD t1/2)]

(8)

where kD is the diffusionally limited rate of hybridization of the labeled SBP, and
max is the maximal surface concentration of the labeled enzyme as a result of

94

Willner et al.

Figure 28 Amperometric transduction of the formation of a double-stranded complementary DNA complex using a SBP-DNA conjugate as an electrobiocatalytic amplier.

hybridization. For a thin hydrogel redox-active lm, and neglecting diffusion barriers for the SBP substrate (H2O2), the electrocatalytic saturation current is given
by Eq. 9,
Icat = 2k[Os2+][S]FA

(9)

where k is the rate-constant for the reduction of H2O2 by the reduced biocatalytic
assembly, [Os2+] is the concentration of the reduced redox-active hydrogel, [S] is

Electronic Transduction of DNA Sensing Processes

95

Figure 29 Chronoamperometric responses of the (36)-functionalized redox-hydrogelmodied electrode upon the hybridization and the sensing of DNA, (37), and its mutants,
(37a) and (37b), at variable temperatures: (a) the analyte, (37); (b) the single-base mismatched DNA mutant, (37b); (c) the four-bases mismatched DNA mutant, (37a). (A) At
45C. (B) At 57C. The amperometric responses originate from the electrobiocatalyzed
reduction of H2O2, 1 102 M, E = 0.06 V vs. Ag/AgCl, by the DNA-SBP conjugates. The
dashed lines correspond to the optimized ts of the experimental data to Eq. 10. (Adapted
from Ref. 50 with permission.)

96

Willner et al.

Table 1 Derived Best-Fit Parameters to Eq. 10 for the Target


DNA, (37), and Its Mutants, (37a) with Four Mismatches and (37b)
with One Mismatch, at Different Hybridization Temperatures
Temperature (C)
45

57

Oligonucleotide

b (pA)

kD(s1)

(37)
(37a)
(37b)
(37)
(37a)
(37b)

27.7
3.44
30.2
44.6
5.9
11.8

0.117
0.077
0.086
0.082
0.079
0.133

From Ref. 50.

the substrate concentration, F is the Faraday constant, A is the electrode surface


area, and is the surface coverage of the hybridized enzyme. Combining Eq. 8
and Eq. 9, where b = 2k[Os2+][S]FA yields Eq. 10, which represents the dependence of the catalytic current on the hybridization rate. The dashed lines in Figure 29 represent the best-tted curves of Eq. 10 to the experimental data. Table 1
summarizes the extracted
Icat = b[1 exp(kD t1/2)]

(10)

values of b and kD (rate constant for hybridization for the different systems). The
analysis of the results clearly reveals that the temperature-controlled hybridization of the analyte (37) and its mutants (37a) or (37b) allows their selective analysis at the electrode interface. The method suffers, however, from the limitation
that the mutants yield residual amperometric responses due to partial hybridization of the mutants with the sensing interface. Thus, it would be difcult to differentiate low concentrations of the target DNA in the presence of high concentrations of its mutants.
B.

Amplied Transduction of Single-Base Mismatches in DNA

Figure 30 outlines the method for the identication of a single base mutation in
an analyte DNA [51]. The sensing of the 41-base oligonucleotide, (39), which
includes a G-mutation, as compared with the normal gene (40), is exemplied.
The thiolated oligonucleotide (38) that is complementary to the oligonucleotide
fragment of (39) or (40) up to the point of mutation is used as the oligonucleotide
probe. The assembly of the probe (38) onto the transducer (e.g., Au-electrode or
Au-quartz crystal,) yields the sensing interface. Interaction of the sensing interface with the mutant (39) or the normal gene (40) generates the respective dou-

Figure 30 Scheme for the electronic transduction of a single-base mutation in an analyte DNA using the biocatalytic precipitation of an insoluble product on the transducer as an amplication route.

Electronic Transduction of DNA Sensing Processes


97

98

Willner et al.

ble-stranded assembly on the transducer. The resulting interface is then reacted


with the biotinylated base complementary to the mismatch site (e.g., biotinylated
cytosine triphosphate, b-dCTP) in the presence of polymerase, Klenow fragment.
In the presence of the double-stranded assembly that includes the mutant, surface
coupling of b-dCTP to the probe oligonucleotide proceeds. The resulting assembly is then interacted with an enzyme-linked avidin conjugate, where the enzyme
catalyzes the precipitation of an insoluble product on the electrode support. In the
present study, alkaline phosphataseavidin, (41), is used as the conjugate, and its
association to the sensing interface catalyzes the oxidative hydrolysis of (11) to
the insoluble indigo derivative, (12). Note that precipitation occurs only if the single-base mutant is linked to the sensing interface. The biocatalyzed precipitation
of (12) provides a means to amplify the sensing process, and the extent of precipitate formed on the transducer is controlled by the amount of DNA-mutant
associated with the sensing interface and the time interval employed for the biocatalyzed precipitation.
The probe oligonucleotide (38) was assembled on the Au-electrode, or the
Au-quartz crystal, to yield a surface coverage of ca. 2.3 1011 molecm2 of the
sensing probe. Figure 31(A) shows the faradaic impedance spectra (displayed as
Nyquist plots) of the sensing interface, curve (a); the sensing surface after interaction with the mutant, (39), 3 109 molemL1, curve (b); the sensing interface
that includes the double-stranded probe-oligonucleotide/mutant (39) assembly
after treatment with polymerase (Klenow fragment) and b-dCTP, curve (c); the
resulting surface after treatment with the alkaline phosphatase avidin conjugate,
(41), curve (d); and upon the biocatalyzed precipitation of the insoluble product
(12) for 10 minutes and 40 minutes, curves (e) and (f), respectively. The semicircle diameters of the different impedance spectra correspond to the electron transfer resistances at the electrode interface at the different phases of analysis. Using
[Fe(CN)6]34 as a redox probe, the electron transfer resistance increases from 1.6
k to ca. 4.1 k upon the formation of the double-stranded assembly between the
(38)-probe oligonucleotide and (39). This is consistent with the fact that the formation of the double-stranded assembly electrostatically repels the redox label
from the electrode interface, thereby introducing a barrier for interfacial electron
transfer. While the treatment of the electrode with polymerase and b-dCTP does
not affect the interfacial electron transfer resistance, the association of the
hydrophobic alkaline phosphatase-avidin conjugate introduces a barrier for electron transfer, Ret 6.2 k. Biocatalyzed precipitation of (12) onto the electrode
insulates the electrode, a process that increases the electron transfer resistance at
the electrode. The electron transfer resistance increases to 8.4 k and 16 k upon
the precipitation of (12) for 10 and 40 minutes, respectively. Figure 31(B) shows
the faradaic impedance spectra corresponding to similar experiments executed
with the normal gene (40). It is evident that after the formation of the double-

Figure 31 (A) Faradaic impedance spectra corresponding to (a) the (38)-modied electrode; (b) the (38)-modied electrode upon interaction with (39), 3 109 molemL1; (c)
the double-stranded (38)/(39)-functionalized electrode after interaction with the Klenow
fragment, 20 UmL1 in a Tris-buffer, pH = 7.8. (d) Upon the interaction of the biotinlabeled double-stranded assembly with avidinalkaline phosphatase conjugate, (41), 100
nmolemL1 in 0.1 M phosphate buffer solution for 15 minutes; (e) After the interaction of
the avidinalkaline phosphatase labeled assembly with (11), 2 102 M, in 0.1 M Trisbuffer solution, pH = 7.6, for 10 minutes; (f) After the interaction of the avidinalkaline
phosphatase labeled assembly with (11), in 0.1 M Tris buffer solution, pH = 7.6, for 40 minutes. (B) Faradaic impedance spectra (Zim vs. Zre) corresponding to (a) the (38)-functionalized electrode; (b) upon the interaction of the (38)-modied electrode with (40), 3 109
molemL1; (c)(f) Repetition of the steps outlined in (A).

100

Willner et al.

stranded assembly, Ret = 3.9 k, no increase in the electron transfer resistance is


observed upon treatment of the surface with alkaline phosphataseavidin or an
attempt to precipitate (12). Thus, the successful analysis of (39) is attributed to the
specic polymerase-mediated coupling of b-dCTP to the mutant assembly, resulting in the biocatalytic precipitation of (12). Using this method, the mutant (39)
was detected with a sensitivity limit that corresponded to 1.0 1014 molemL1.
The analysis of the single-base mutant by the protocol outlined in Fig. 30 is
also transduced by microgravimetric quartz-crystal-microbalance assay of the
biocatalyzed formation of the precipitate. An Au-quartz crystal (9 MHz, AT-cut)
was modied with the thiolated oligonucleotide probe (38). The resulting
oligonucleotide-functionalized crystals were then treated with a high concentration, 3 109 molemL1, of the mutant (39), low concentration of the mutant,
(39), 1 1012 molemL1, or a high concentration, 3 109 molemL1, of the
normal gene sequence (40). The resulting crystals were then reacted with b-dCTP
and polymerase Klenow fragment to couple the biotinylated base to the assemblies that included the single-base mutation. Figure 32 shows the time-dependent
frequency changes as a result of the association of the avidinalkaline phosphatase conjugate to the respective interfaces, curves (a), (b), and (c), and upon
the biocatalyzed precipitation of the insoluble product, (12), to the resulting
assemblies, curves (d), (e), and (f). Treatment of the (38)-probe oligonucleotide/(39)-mutant-functionalized crystal with alkaline phosphataseavidin,
(41), results in a frequency change of f = 47 Hz (curve a), indicating the association of the conjugate to the coupled b-dCTP base. The surface-associated
enzyme-conjugate biocatalyzes the precipitation of (12), and a further change in
the crystal frequency, ca. f = 70 Hz, occurs within 30 minutes (curve d). Treatment of the (38)-probe oligonucleotide/(40)-functionalized crystal with alkaline
phosphataseavidin, (41), does not yield any frequency change (curve c), and no
precipitation is observed upon the addition of (11) (curve f). The fact that b-dCTP
is not coupled to the (38)-probe-oligonucleotide/(40) assembly eliminates the
association of the enzyme conjugate and the subsequent precipitation of (12).
Treatment of the sensing interface with the lower concentration of the mutant
results in lower frequency changes upon interaction with the alkaline phosphataseavidin conjugate and the subsequent precipitation of the insoluble product, curves (b) and (e), respectively. These results originate from the fact that at
low bulk concentration of (39), low surface coverage of the mutant associated
with the sensing interface is obtained.
This method was applied to detect the gene corresponding to the Tay-Sachs
genetic disorder [51]. Tay-Sachs disease [52] is caused by hexoseaminidase deciency of the enzyme that degrades the GM2 ganglioside to GM3. It appears at
approximately l6 months of age and is fatal usually in early childhood. Affected
children become blind and physically and mentally regressed. The disease originates from a genetic disorder that is frequent in Jews of Eastern European descent.

Electronic Transduction of DNA Sensing Processes

101

Figure 32 Time-dependent frequency changes upon the association of avidinalkaline


phosphatase and the biocatalyzed precipitation of (12) in the presence of the oligonucleotide/DNA assemblies: (a) and (b) correspond to the out of cell interaction of the (38)functionalized electrode with (39), 3 109 molemL1 and 1 1012 molemL1, respectively, followed by the reaction of the double-stranded functionalized assemblies with
Klenow fragment and biotinylated dCTP, and in cell recording of the frequency changes of
the resulting functionalized electrodes upon interaction with avidinalkaline phosphatase,
100 nmolemL1. (c) Out-of-cell interaction of the (38)-functionalized Au-quartz crystal
with (40), 3 109 molemL1, followed by the reaction of the interface with the Klenow
fragment and biotinylated dCTP, and in-cell monitoring of frequency changes upon the
interaction of the interface with avidinalkaline phosphatase, 100 nmolemL1. (d), (e), and
(f) correspond to the time-dependent frequency changes of the respective interfaces formed
in (a), (b), and (c) in the presence of (11), 20 mM in 0.1 M Tris-buffer, pH = 7.6, upon the
biocatalyzed precipitation of (12).

In a recent survey [53] it was reported that 1 of 30 Jews of Eastern European


descent in the United States and Canada is a carrier of the respective defective
gene. Although several mutations in the respective gene were reported [53], the
most frequent mutation includes the four-base (ATAG) insertion into the G-G
base of the normal gene (P), to yield the mutant (Q). The primer (T) that is complementary to the normal gene as well as to the mutant till one base prior to the
mutation site was immobilized on Au-electrodes. The human DNA was isolated
from 0.5 mL blood samples of individuals that carry the heterocygotic gene
(genetic disorder carrier), the homocygotic gene (carrier of the disease), and the

102

Willner et al.

normal gene. The DNA was denaturized and hydrolyzed to yield smaller fragments. The mixtures of the respective DNAs were interacted with the T-functionalized electrodes, with no pre-PCR amplication. The resulting electrodes
were then interacted with polymerase (Klenow Fragment) in the presence of
biotinylated-dUTP. As dUTP complements the A base, attachment of the U-base
by polymerase would occur only on the electrodes carrying the double-stranded
assembly with the heterocygotic gene and homocygotic gene, whereas no polymerization occurs in the presence of the normal gene. Table 2 summarizes the
changes in the interfacial electron transfer resistances, Ret, ([Fe(CN)6]3/4 is
used as a redox-probe) as a result of the association of avidinalkaline phosphatase to the biotinylated label and the precipitation of the insoluble product (12)
on the electrode support. Only the samples carrying the heterocygotic or homocygotic mutants lead to the insulation of the electrode, Ret = 1.8 and 2.3 k,
respectively. Thus, a single measurement enables us to trace the individuals carrying the disease (homocygotic gene) or the carriers of the genetic disorders (heterocygotic gene). By a second measurement, the individuals carrying the homocygotic or heterocygotic genes could be differentiated by the reaction of the
sensing electrodes, which includes the primer (T) hybridized with the respective
genes, with polymerase in the presence of biotinylated-dCTP. In the heterocygotic gene sample, the normal gene, (P), and the mutant, (Q), are included in alleles and the biotinylated C-base is attached to the double-stranded assembly with
the normal gene, Q. This enables the binding of the avidinalkaline phosphatase
conjugate and the subsequent biocatalyzed precipitation of (12) on the electrode

Table 2 Interfacial Electron Transfer Resistances Observed as a Result of the


Interaction of the (T)-Functionalized Electrode Hybridized with the Normal
Gene, (P), or the Tay-Sachs Mutant, (Q), with Polymerase and Biotinylated
Bases and the Subsequent Alkaline Phosphatase Precipitation of (12)

Gene sample
Heterocygotic gene
(P) + (Q)
Homocygotic gene
(Q) + (Q)
Normal gene
(P) + (P)

Reaction with
biotinylated-dUTP
followed by the binding
of avidin-alkaline and
precipitation of (12).
Ret (k)

Reaction with
biotinylated-dCTP
followed by the binding
of avidinalkaline
phosphatase and
precipitation of (12).
Ret (k)

1.8 6 0.1

1.6 6 0.1

2.3 6 0.1
<0.1

2.1 6 0.1

Electronic Transduction of DNA Sensing Processes

103

support that leads to the observation of an interfacial electron transfer resistance


of 1.6 k in the presence of the heterocygotic gene (Table 2).
On the other hand, no polymerization of the biotinylated C-base can occur
on the homocygotic gene hybridized to the sensing interface, and thus no increase
in the interfacial electron-transfer resistance is observed upon interaction with
polymerase/biotinylated dCTP and an attempt to precipitate (12).

VII.

CONCLUSIONS AND PERSPECTIVES

The present account has addressed novel methods for the amplication of DNA
sensing processes and the electronic transduction of the DNA recognition events.
Three different approaches to the design of amplication routes have been discussed. The rst method includes the selective association of an enzyme-conjugate
to the resulting nucleic-acid/DNA recognition complex. The secondary biocatalyzed precipitation of an insoluble product on the transducers or an electrobiocatalyzed transformation provide the amplication path. The second approach
involves the polymerase-induced replication of DNA in the presence of a labeled
base. The secondary binding of an enzyme conjugate to the replicated DNA provides the amplication route. The third approach includes the use of labeled
nanoparticles or nano-membranes as amplifying units. A dendritic-type amplication process for the analyte DNA was described by the stepwise assembly of multilayered generations of the analyzed complex.
Different electronic transduction means were discussed, including faradaic
impedance spectroscopy, chronopotentiometry, chronocoulometry, and microgravimetric quartz-crystal-microbalance measurements. The different methods
reveal high sensitivities that allow the detection of 5002000 copies of DNA in
1050 L of samples. These rapid advances in electronic DNA detection open the
way to construct electrode chips for the parallel or consecutive analyses of a collection of nucleic acids. The technologies to construct such electrode arrays are
available, and robotic methods for the site-specic spotting of such arrays are
known. This suggests that the different approaches are sufciently ripe to be harnessed and optimized for practical targets.
The scientic accomplishments suggest, however, that other methodologies may be envisaged to expand the area of DNA electronics. The construction
of dendritic arrays that include semiconductor nanoparticles, similar to the assembly of the Au nanoparticles, may be envisaged. The photoactivity of the particles
may be used to adapt photoelectrochemistry (photocurrents) as the transduction
means. The high stability of semiconductor nanoparticles and their intense uorescence may then be used for the optical detection of DNA. The polymeraseinduced replication of the DNA may be performed with redox-functionalized
phosphorylated bases. The amperometric responses of the replicated DNAs and

104

Willner et al.

the potential to activate secondary enzyme-cascades by the redox units pave the
way for additional DNA amplication and detection routes.
Although the present account has emphasized the methods for DNA analysis, one important aspect of this study is the ability to assemble organized and predesigned nano-architectures of DNA on surfaces. We realize that a variety of
tools are available to immobilize, hybridize, polymerize, and specically
hydrolyze DNA on surfaces. Methods to functionalize nanoparticles, proteins, or
liposomes with nucleic acids are available, and other nanoaggregates, such as
nucleic acidfunctionalized nanotubes, may be envisaged. This provides a unique
battery of components for the nano-engineering of complex DNA/nano-aggregates and scaffolds. These nanostructures may then be used to design complex
circuits or nanoparticle-linked wires.
The second topic addressed in this chapter relates to the specic detection
of DNA. Here we demonstrated not only the possibility of identifying a singlebase mismatch but of identifying heterocygotic and homocygotic polymorphism
of the analyte DNA. The method is based on the selection of a primer that enables
the secondary polymerase-induced coupling of a labeled-base complementary to
the mutation site on the primer-DNA double-stranded assembly. The subsequent
association of an amplifying conjugate on the label provides a means to amplify
the single-base analysis. The method is not free of limitations, as it requires the
primary knowledge of the mutants structure. For most genetic disorders, the
characteristic structures of the different mutants are known. Thus, by the appropriate selection of primers and the design of an appropriate set of reactions of
polymerase-induced coupling of labeled phosphorylated bases, the development
of electrode-arrays that are capable of identifying target genetic disorders, is feasible. The electronic transduction of a sequencing process and the identication
of unknown mutants remains a challenging topic in DNA-electronics.

ACKNOWLEDGMENT
Our research on DNA electronics is supported by the Max Planck Research
Award for International Cooperation and by The Israel Ministry of Science as an
Infrastructure Project (Biomicroelectronics).

REFERENCES
1.
2.

(a) Wilson, E. K., Chem. Eng. News 76 (21), 47 (1998).


(b) Mikkelsen, S. R., Electroanalysis 8, 15 (1996).
(a) Bard, A. J., Rodriguez, M., Anal. Chem. 62, 2658 (1990).
(b) Millan, K. M., Mikkelsen, S. R., Anal. Chem. 65, 2317 (1993).

Electronic Transduction of DNA Sensing Processes

3.
4.
5.
6.
7.
8.
9.
10.
11.

12.
13.

14.
15.

16.

17.

18.

19.

20.
21.

105

(c) Yang, M. S., McGovern, M. E., Thompson, M., Anal. Chim. Acta 346, 259
(1997).
Thompson, M., Furtado, M., The Analyst 124, 1133 (1999).
Piunno, P. A. E., Krull, U. J., Hudson, R. H. E., Damha, M. J., Cohen, H., Anal.
Chim. Acta 288, 205 (1994).
Liedberg, B., Nylander, C., Lundstrm, I., Sens. Actuator 4, 299 (1983).
Elghanian, R., Storhoff, J. J., Mucic, R. C., Letsinger, R. L., Mirkin, C. A., Science
277, 1078 (1997).
Storhoff, J. J., Elghanian, R., Mucic, R. C., Mirkin, C. A., Letsinger, R. L., J. Am.
Chem. Soc. 120, 1959 (1998).
Wang, J., Anal. Chem. 71, 328R (1999).
Wang, J., Palecek, E., Nielsen, P. E., Rivas, G., Cai, X. H., Shiraishi, H., Doutha, N.,
Luo, D. B., Farias, P. A. M., J. Am. Chem. Soc. 118, 7667 (1996).
Korri-Yousou, H., Garnier, F., Srivastava, P., Godillot, P., Yassar, A., J. Am. Chem.
Soc. 119, 7388 (1997).
(a) Hashimoto, K., Ito, K., Ishimori, Y., Anal. Chem. 66, 3830 (1994).
(b) Takenaka, S., Yamashita, K., Takagi, M., Uto, Y., Kondo, H., Anal. Chem. 72,
1334 (2000).
Millan, K. M., Saraullo, A., Mikkelsen, S. R., Anal. Chem. 66, 2943 (1994).
(a) Bardea, A., Dagan, A., Ben-Dov, I., Amit, B., Willner, I., Chem. Commun. 839
(1998).
(b) Okahata, Y., Kawase, M., Niikura, K., Ohtake, F., Furusawa, H., Ebara, Y., Anal.
Chem. 70, 1288 (1998).
Landegren, U., Curr. Opin. Biotechnol. 7, 95 (1996).
(a) Rickert, J., Brecht, A., Gpel, W., Biosens. Bioelectron. 12, 567 (1997).
(b) Bunde, R. L., Jarvi, E. J., Rosentreter, J. J., Talanta 46, 1223 (1998).
(c) Rogers, K. R., Mol. Biotechnol. 14, 109 (2000).
(a) Willner, I., Riklin, A., Shoham, B., Rivenzon, D., Katz, E., Adv. Mater. 5, 912
(1993).
(b) Willner, I., Heleg-Shabtai, V., Blonder, R., Katz, E., Tao, G., Bckmann, A. F.,
Heller, A., J. Am. Chem. Soc. 118, 10321 (1996).
(a) Willner, I., Lapipdot, N., Riklin, A., Kasher, R., Zahavy, E., Katz, E., J. Am.
Chem. Soc. 116, 1428 (1994).
(b) Bardea, A., Katz, E., Bckmann, A. F., Willner, I., J. Am. Chem. Soc. 119, 9114
(1997).
(a) Blonder, R., Levi, S., Tao, G., Ben-Dov, I., Willner, I., J. Am. Chem. Soc. 119,
10467 (1997).
(b) Blonder, R., Katz, E., Cohen, Y., Itzhak, N., Riklin, A., Willner, I., Anal. Chem.
68, 3151 (1996).
(a) Katz, E., Willner, I., J. Electroanal. Chem. 418, 67 (1996).
(b) Willner, I., Katz, E., Willner, B. In: Sensors Updates, Baltes, H., W. Gpel, W.,
Hesse, J. (Eds.), 1999, Vol. 5, Chapter 2, pp. 45102.
Bard, A. J., Faulkner, L. R., Electrochemical Methods: Fundamentals and Applications, Wiley, New York, 1980.
Stoynov, Z. B., Grafov, B. M., Savova-Stoynov, B. S., Elkin, V. V., Electrochemical Impedance, Nauka, Moscow, 1991.

106
22.
23.
24.

25.

26.
27.
28.
29.
30.
31.
32.
33.
34.

35.
36.
37.
38.
39.

40.
41.
42.
43.

44.

45.

Willner et al.
Champagne, G. Y., Belanger, D., Fortier, G., Bioelectrochem. Bioenerg. 22, 159
(1989).
Bond, A. M., Modern Polarography Methods in Analytical Chemistry, Marcel
Dekker, New York, 1980.
(a) Alfonta, L., Bardea, A., Khersonsky, O., Katz, E., Willner, I., Biosens. Bioelectron. 76, 134 (2001).
(b) Katz, E., Alfonta, L., Willner, I., Sensors and Actuators B 76, 134 (2001).
(a) Buttry, D. A., Ward, M. D., Chem. Rev. 92, 1355 (1992).
(b) Buttry, D. A. in: Electrochemical Chemistry, Bard, A. J., Ed., Vol. 17, Marcel
Dekker, New York, 1991, p. 1.
Maskos, U., Southern, E. M. Nucleic Acids Res. 20, 1679 (1992).
Bardea, A., Dagan, A., Ben-Dov, I., Amit, B., Willner, I., Chem. Commun. 839,
(1998).
Herne, T. M., Tarlov, M. J., J. Am. Chem. Soc. 119, 8916 (1997).
Steel, A. B., Herne, T. M., Tarlov, M. J., Anal. Chem. 70, 4670 (1998).
Patolsky, F., Lichtenstein, A., Willner, I., J. Am. Chem. Soc. 123, 5194 (2001).
Alfonta, L., Singh, A. K., Willner, I., Anal. Chem. 73, 91 (2001).
de Lumley-Woodyear, T., Campbell, C. N., Heller, A. J. Am. Chem. Soc. 118, 5504
(1996).
Patolsky, F., Katz, E., Bardea, A., Willner, I., Langmuir 15, 3703 (1999).
(a) Patolsky, F., Zayats, M., Katz, E., Willner, I., Anal. Chem. 71, 3171 (1999).
(b) Alfonta, L., Katz, E., Willner, I., Anal. Chem. 72, 927 (2000).
(c) Bardea, A., Katz, E., Willner, I. Electroanalysis 12, 1097 (2000).
(a) Willner, I., Katz, E., Angew. Chem. Int. Ed. 39, 1180 (2000).
(b) Willner, I., Katz, E., Willner, B., Electroanalysis 9, 965 (1997).
Lisdat, F., Ge, B., Scheller, F. W. Electrochemistry Commun. 1, 65 (1999).
Courteix, A., Bergel, A., Comtat, M. J. Appl. Electrochem. 25, 508 (1995).
Ebersole, R. C., Ward, M. D., J. Am. Chem. Soc. 110, 8623 (1988).
(a) Reddy, S. M., Jones, J. P., Lewis, T. J., Vadgama, P. M., Anal. Chim. Acta 363,
203 (1998).
(b) Abad, J. M., Pariente, F., Hernandez, L., Lorenzo, E., Anal. Chim. Acta 368, 183
(1998).
Bardea, A., Patolsky, F., Dagan, A., Willner, I., Chem. Commun. 21 (1999).
Patolsky, F., Lichtenstein, A., Willner, I., Angew. Chem. Int. Ed. 39, 940 (2000).
Patolsky, F., Lichtenstein, A., Willner, I., J. Am. Chem. Soc. 122, 418 (2000).
(a) Alivisatos, A. P., Johnsson, K. P., Peng, X. G., Wilson, T. E., Loweth, C. J.,
Bruchez, M. P., Schultz, P. G., Nature 382, 609 (1996).
(b) Mirkin, C. A., Letsinger, R. L., Mucic, R. C., Storhoff, J. J., Nature 382, 607
(1996).
(a) Taton, T. A., Mucic, R. C., Mirkin, C. A., Letsinger, R. L., J. Am. Chem. Soc. 122,
6305 (2000).
(b) Storhoff, J. J., Lazarides, A. A., Mucic, R. C., Mirkin, C. A., Letsinger, R. L.,
Schatz, G. C., J. Am. Chem. Soc. 1 22, 4640 (2000).
Zhou, X. C., OShea, S. J., Li, S. F. Y., Chem. Commun. 953 (2000).

Electronic Transduction of DNA Sensing Processes


46.
47.
48.
49.
50.
51.
52.

53.

107

Patolsky, F., Ranjit, K. T., Lichtenstein, A., Willner, I., Chem. Commun. 1025
(2000).
Alfonta, L., Willner, I., Chem. Commun. 1492 (2001).
Patolsky, F., Lichtenstein, A., Kotler, M., Willner, I., Angew. Chem. Int. Ed. 40,
2261 (2001).
Blanco, L., Bernard, A., Lazaro, J. M., Martin, G., Garmendia, C., Salas, M., J. Biol.
Chem. 264, 8935 (1989).
Caruana, D. J., Heller, A., J. Am. Chem. Soc. 121, 769 (1999).
Patolsky, F., Lichtenstein, A., Willner, I., Nature Biotechnol. 19, 253 (2001).
Gravel, F. A., Clarke, J. T. R., Kaback, M. M., Mahuran, D., Sandhoff, K., Suzuki,
K., in: The Metabolic and Molecular Basis of Inherited Diseases, Vol. 2; Scriver, C.
R., Blander, A. L., Sly, W. S., Vale, D. (Eds.), McGraw-Hill, New York, pp.
28392879 (1995).
Myerowitz, R., Costigan, F. C., J. Biol. Chem. 263, 18587 (1988).

4
Direct Electrochemistry
of Proteins and Enzymes
at Electrodes
James D. Burgess
Case Western Reserve University, Cleveland, Ohio

Fred M. Hawkridge
Virginia Commonwealth University, Richmond, Virginia

I.

INTRODUCTION

The use of electrochemical methods to study protein and enzyme electron transfer reaction kinetics, thermodynamics, and mechanisms directly with electrodes
is becoming a mature eld. Twenty years ago such studies were rarely conducted
outside of laboratories with substantial experience in electrochemistry. Now scientists in diverse elds have taken up cyclic voltammetry, square wave voltammetry, and other electrochemical methods to study biological systems. Clearly
much has been learned about how to conduct reliable electrochemical experiments on complex biological samples using direct electron transfer at electrodes.
Progress in this eld was slow, and some background is provided to put the current state of this eld in context.
The problem of slow heterogeneous electron transfer between proteins/
enzymes and electrodes was understood fty years ago. In the earliest attempts to
characterize the thermodynamics of respiratory electron transfer proteins by
potentiometric titrations, the low exchange currents between platinum and gold
potentiometric indicator electrodes and electron transfer proteins were evident
[14]. Irreversible potential measurements made this work problematic. Small
109

110

Burgess and Hawkridge

molecule redox couples that were chemically reversible and that exhibited
reversible heterogeneous electron transfer at platinum and gold electrodes (i.e.,
high exchange currents) were added to protein sample solutions at low concentration so that the redox state of the protein could be indirectly measured potentiometrically. A series of mediators was used that spanned the range of potentials
about what was thought to be the approximate value of the formal reduction
potential of the electron transfer protein. Tabulations of mediators, together with
their formal reduction potentials and electron transfer stoichiometry that were
used for this purpose, have been published [5,6].
Mediators were adopted by Theodore Kuwana for use in faradaic electrochemical studies of electron transfer proteins. The electron transfer reactions of a
protein/enzyme were coupled to the potential applied to an electrode by having
the appropriate mediators present in solution. Initial experiments involved using
mediators to conduct indirect coulometric titrations of proteins/enzymes, often
using optical absorption spectroscopy at optically transparent electrodes to simultaneously monitor the titration progress. The reaction scheme in its simplest form
is illustrated with the equations:
Electrode reaction: Mo + ne B Mr
Solution reaction:
Mo

Mr

Mr

BRMo

Mo

(1)
+

BRMr

(2)

and
are the oxidized and reduced forms of the redox mediator,
where
respectively, and BRMo and BRMr are the oxidized and reduced forms of the biological redox molecule, respectively. In this example, Mo catalytically reduces the
BRMo at a controlled rate via the solution reaction, Equation (2). By including
two redox mediators, one with a formal reduction potential sufciently negative
of the BRM formal reduction potential, and the other positive, repetitive reductive and oxidative mediated coulometric titrations can be conducted. This makes
it convenient to assess both the chemical reversibility of the BRM and to obtain
statistically reliable measures of the electron transfer stoichiometry and the formal reduction potential of the BRM. With this approach, it is easy to accurately
introduce nanocoulombs of charge and to spectrally monitor absorbance changes
after each titration increment during the titration. Even reductive titrations are
simplied using an electrochemical cell from which oxygen could be easily
removed and excluded. The spectral titration data were just simply beautiful using
this experimental method.
The earliest work in Ted Kuwanas laboratory using the mediated approach
focused on studies of a complex enzyme/substrate system with the goal of characterizing the kinetics of this reaction as opposed to the thermodynamics. In this
work, methyl viologen was used as a reductive redox mediator to couple ferredoxin-NADP-reductase and its substrate, NADP, to the applied electrode potential [7]. As noted in this paper, the approach had its origins in work done in the
group of Herb Silverman, who at that time was at TRW Systems.

Direct Electrochemistry

111

This approach was subsequently developed in Ted Kuwanas laboratories to


accurately quantify electron transfer stoichiometry (n values) and formal reduction
potentials in studies of simpler systems involving more readily accessible heme
proteins such as myoglobin and cytochrome c. The method of indirect coulometric
titration followed by optical absorption spectroscopy was developed [8] such that
formal reduction potentials and the electron transfer stoichiometry could be easily
measured using repetitive reductive and oxidative titrations on the same sample.
An advantage of this method was that no exogenous titrants were necessary, so
sample volumes remained constant during titrations. Soon this approach was being
applied to complex detergent-solubilized samples of cytochrome c oxidase, which
exhibits complex electron transfer thermodynamics and electron transfer stoichiometry, i.e., four different active metalloenzyme sites and uncertainties regarding serial one electron transfer and serial two electron transfer reaction mechanisms [9,10]. The inuence of physiologically relevant ligands, such as carbon
monoxide, on the electron transfer thermodynamics, kinetics, and mechanism of
cytochrome c oxidase was subsequently probed [1114].
Bill Heinemans group developed an elegant indirect titration method in the
1970s [15]. Indirect coulometric titrations and optically transparent thin-layer
electrochemical cells were combined to provide a simple and quick means of
making formal potential measurements on electron transfer proteins. Moreover,
the amount of sample required for this measurement was quite small. This is now
a routine method for measuring the formal potential of electron transfer proteins,
which is used by a wide range of non-electrochemical scientists. Use of this
method in our laboratory, greatly facilitated by help from the Heineman laboratory, led to our later efforts to develop direct electrochemical methods for protein
and enzyme studies.
This background is provided to show that work done in this laboratory initially on mediated electron protein studies led to the study of direct electron transfer reactions of proteins and enzyme at electrodes. It was a not a large intellectual
step to consider direct studies of electron transfer proteins at electrodes, given this
background and the work being done in other laboratories in the area of chemically modied electrodes at that time. Still, it was the unexpected that led to actually thinking that an electron transfer protein might react reversibly or quasireversibly at an electrode, as will be described later.
An earlier review attempted to cover the large body of work on electron
transfer proteins that exhibited irreversible heterogeneous electron transfer kinetics, particularly at dropping mercury electrodes, and the emerging area of direct
electrochemistry of proteins [16]. Other reviews have treated more recent work
on electron transfer proteins that exhibit reversible to quasi-reversible heterogeneous electron transfer kinetics [1720]. Some overlap will occur here, but the
emphasis will be on the progression that was followed in this laboratory. It is
worth stating again that in the mid-1970s, the paradigm was that electron transfer

112

Burgess and Hawkridge

proteins would never be shown to exhibit quasi-reversible to reversible heterogeneous electron transfer kinetics at solid electrodes, and in some quarters this
notion persists today.

II.

MODIFIED GOLD ELECTRODES

Our initial research focus began in the area of photosynthetic electron transfer
proteins, following on work that Kuwana had initiated in the early 1970s [7]. This
work, which was started by two graduate students, Lyman H. Rickard and
H. Lynn Landrum, involved work on the electron transfer protein spinach ferredoxin, a small protein of 11,000 Daltons with a quite negative formal reduction
potential, 0.43 V vs NHE. An important advantage of the indirect coulometric
titration method is the ability to use redox mediators with very negative formal
potentials to drive the reduction of electron transfer proteins with negative formal
potentials, a point recognized initially by Kuwana. A chemical reductive redox
titration cannot drive the solution potential more negative than E = 0.0591 (pH)
[21]. This explains why most redox titrations of photosystem I photosynthetic
samples are conducted at solution pH at or above values of 10.0 where these samples are probably not in their native structures. A family of mediators that has
been widely used for their negative formal potentials in titrations of photosynthetic samples is the viologens, or 4,4- or 2,2-bipyridinium substituted compounds with methyl viologen (1,1-dimethyl-4,4-bipyridinium dichloride) being
the most commonly used example. This family of compounds was developed for
a number of diverse applications including its efcacy as herbicides [22]. Our rst
paper using methyl viologen in an indirect coulometric titration of spinach ferredoxin was published in 1978 [23]. This work was an extension of earlier work by
Ito and Kuwana described above [7].
Earlier, during the summer of 1975, the opportunity to work for Bacon Ke
at the Charles F. Kettering Laboratory in Yellow Springs, Ohio, arose. Ke had
asked Ted Kuwana, a short distance away in Columbus, Ohio to collaborate, but
Ted suggested that he contact us. This is another example of Kuwanas generous
support of his students. Bacon Ke, who was doing cutting edge work on both Photosystem I and Photsystem II, was interested in doing mediated (using viologens)
indirect coulometric titrations of these systems monitored by circular dichroism
spectroscopy, cryogenic uorescence yield measurements, and light-induced
absorbance and electron paramagnetic resonance measurements. This was
extremely exciting work that resulted in the most negative titration of Photosystem I at that time. These experiments provided a direct measurement of the formal reduction potential for the primary electron acceptor in Photosystem I [24].
The implementation of optically transparent thin-layer electrochemical cells for
cryogenic experiments that summer was facilitated immensely by information

Direct Electrochemistry

113

gained during visits to the nearby Heineman laboratory in Cincinnati. Lightinduced uorescence yield titrations of Photosystem II [25] and a technique paper
demonstrating the power of circular dichroism for monitoring indirect coulometric titrations of proteins came from that summers work [26].
Bacon Ke provided us with an ample amount of highly puried spinach
ferredoxin for study using indirect coulometric titrations. In preparing to work
with this precious sample, Lynn Landrum was learning to do experiments using
Heinemans optically transparent thin-layer electrochemical (OTTLE) cell using
only methyl viologen, the mediator. The rst reduction of methyl viologen produces the bright blue-violet color of the cation radical, making it ideal for optical
monitoring. Several initial experiments with methyl viologen in the OTTLE
failed to show any evidence of the cation radical. These were very simple experiments: position the cell in a spectrophotometer, apply a reducing potential to the
gold minigrid working electrode to reduce the parent dication, and acquire the
absorption spectrum. Adding a small amount of solid sodium dithionite to these
solutions immediately produces the bright blue-violet color. After reproducing
this unexpected result many times, Lynn began trying to determine why the spectrum for the cation radical was not observed in these experiments, including looking for evidence of the well-known disproportionation reaction of the cation radical. Ultimately, he learned that an electrochemically initiated polymerization
was occurring to produce a lm on the gold minigrid working electrode that itself
was not electroactive. Scanning electron microscopy showed the physical presence of this polymer lm. As is often the case, a literature survey revealed a large
history for the formation of electropolymerized viologens as electrochromic display devices. Near this time, several groundbreaking papers on chemically modied electrodes, including work from Royce Murrays group [27,28], were
appearing. The idea to see if this modied surface would directly transfer electrons with ferredoxin was a logical step, and the experiment worked [29]. The
cyclic voltammetry of spinach ferredoxin taken in an OTTLE with a modied
gold minigrid electrode is shown in Figure 1. Although the voltammetry was not
ideal, at the time this was a really exciting result. Later that year, papers on the
direct electron transfer of cytochrome c at indium oxide by Yeh and Kuwana [30]
and at bipyridyl modied gold by Eddowes and Hill appeared [31]. These papers,
together with work by Niki on the reactivity of the special four-heme cytochrome
c3 at mercury electrodes [32], prompted others to search for electrode materials
and chemically modied electrodes that exhibited direct electron transfer reactivity with electron transfer proteins (see reviews [1620]).
On moving to Virginia Commonwealth University an undergraduate, Joyce
F. Stargardt, expressed interest in doing research with a clinical slant. A literature
search revealed that analysis for myoglobin in serum was used as a diagnostic for
myocardial infarction, but no simple assay was available because of the need for
detecting low concentration levels. Earlier experience with myoglobin [8] led to

114

Burgess and Hawkridge

Figure 1 Cyclic voltammograms at a surface-modied gold minigrid electrode: ( )


0.30 M spinach ferredoxin in 0.1 M Tris, 0.1 M NaCl, pH 7.0; ( --- ) same electrode after
rinsing cell and introducing buffer alone. Scan rate 5 mV/s.

experiments using an OTTLE cell using the viologen/polymer modied working


electrode. The heterogeneous electron transfer kinetics were found to be too slow
to make direct electrochemical measurements feasible, but OTTLE experiments
clearly showed that myoglobin could be reversibly reduced and oxidized [33].
At that point a method for quantifying the heterogeneous electron transfer
rate for the reactions of electron transfer proteins at this modied gold minigrid
electrode surface was not available. Purely electrochemical methods could not be
used due to the low values of the heterogenous electron transfer rate constants.
Chronocoulometry required use of large overpotential step values with accompanying large background signals. Cyclic voltammetry of myoglobin and cytochrome c at the modied gold minigrid working electrode was barely distinguishable from scans on buffer/electrolyte alone. To learn what factors contribute
to the rates of heterogeneous electron transfer of electron transfer proteins at elec-

Direct Electrochemistry

115

trodes, a way to quantify the kinetics of these reactions was needed. It should be
noted that during the development of the spectroelectrochemical kinetic methods
described below, a purely electrochemical method using channel ow hydrodynamic conditions was reported by Weber and Purdy [34]. Indeed, work in this
group by James F. Castner used this method and obtained a heterogeneous electron transfer rate constant for myoglobin of 8.9 (61.5) 105 cm/s [35], in good
agreement with methods described below.

III. SPECTROELECTROCHEMICAL HETEROGENEOUS


ELECTRON TRANSFER KINETIC METHODS
Our work then entered into a lengthy and productive collaboration with Henry
Blounts group at the University of Delaware. Several sensitive and selective spectroelectrochemical methods for measuring heterogeneous electron transfer rate
constants were developed by the Blount group. The appeal of these spectroelectrochemical methods for studying heterogeneous electron transfer kinetics was in part
derived from the availability of intense changes in the optical absorbance spectra
that accompanied electron transfer for heme proteins, especially for myoglobin
and cytochrome c. Difference molar absorptivities of 50,000 M1 cm1 permitted
studies on protein samples at concentrations in the 100 M range.
Douglas E. Albertson rst developed a potential step chronoabsorptometry
method for the irreversible heterogeneous electron transfer kinetic case [36]. This
method was analogous to the earlier chronocoulmetry approach developed by
Christie, Lauer, and Osteryoung [37]. Use of large overpotential steps did not
inuence the optical absorption signal in contrast to the electrochemical current/
charge signal where the signal-to-noise ratio was too low to use. This was followed
by the challenging development of a general potential step chronoabsorptometry
method for the quasi-reversible case by Eric E. Bancroft [38]. At the same time,
Bancraft developed an elegant spectroelectrochemical method, derivative cyclic
voltabsorptometry (DCVA), the optical analog of cyclic voltammetry [39]. This
method was especially critical as the factors affecting the rates of heterogeneous
electron transfer for proteins were being sorted out as described later.
Edmund F. Bowden conducted potential step chronoabsorptometry experiments on the reaction of myoglobin at modied gold minigrid electrodes [40].
Although these experiments were very reproducible, the heterogeneous electron
transfer kinetic parameters raised concerns, namely, the rate constant was very
low (ko = 3.9 1011 cm/s) and alpha was high ( = 0.88). These issues became
muted as the work progressed, as will be discussed below.
At the same time, these newly developed spectroelectrochemical methods
were being applied to the reaction of cytochrome c. The heterogeneous electron

116

Burgess and Hawkridge

transfer kinetic parameters for cytochrome c were more reasonable than for myoglobin at the modied gold minigrid electrode as well as at tin oxide and indium
oxide optically transparent electrodes. Potential step chronoabsorptometry was
used to measure heterogeneous electron transfer rate constants at these three electrodes and values near 1 105 cm/s with variable values of the transfer coefcient were obtained [41]. When derivative cyclic voltabsorptometry was used for
the rst time to characterize the reaction of cytochrome c at tin oxide optically
transparent electrodes, the unexpected result shown in Figure 2 was obtained
[42]. The solid line is the experimental optical signal; the dots are the calculated
response using heterogeneous electron transfer kinetic parameters obtained from
potential step chronoabsorptometry experiments [41]. The marked discrepancy
during the reverse, oxidative scan held much information about this reaction that

Figure 2 DCVA response of cytochrome c at a tin oxide optically transparent electrode


and simulated results. [Cyt c] = 89 M, 0.07 M phosphate buffer at pH = 7.0, 0.1 M NaCl,
scan rate = 4.0 mV/s. Solid circles are simulated responses using ko = 2.2 106 cm/s, =
0.31, and Do = DR = 1.1 106 cm2/s. Solid line is experimental result.

Direct Electrochemistry

117

would be explained in subsequent work. Equally disconcerting was the fact that
this response shape was stable over time and suggested that the electrode reaction
could involve unexplained reaction mechanism steps.
The picture became clearer when experiments were conducted in which
cyclic voltammograms were obtained immediately on introducing a solution of
cytochrome c into a cell with an indium oxide working electrode. There was a
marked initial time dependence with a nal stable response exhibiting electron
transfer kinetics that was much more irreversible [43]. Following the suggestion of
a collaborator, Jan F. Chlebowski, the now-obvious move to chromatographically
purify the cytochrome c sample was made with profound results. Figure 3 shows a
set of cyclic voltammograms and derivative cyclic voltabsorptammograms for
cytochrome c at indium oxide. The Tris/cacodylic acid buffer was used in this work
because there is minimal anion binding to the positively charged lysine residues on
the reaction surface of cytochrome c [43]. These responses show no time-dependent loss of response and are stable for hours at indium oxide when the sample is
puried chromatographically. The optical results are shown compared with calculated responses, and the agreement is excellent. As will be described later,
lyophilized samples contain oligomeric forms of cytochrome c that foul solid electrode surfaces, and the samples used in this work also contained deamidated forms
of the protein. A very small amount of impurity is adequate to completely foul a
solid electrode, a point commonly known in the electrochemical community, but
this is the rst report in which importance of this simple principle is described for
studies of proteins. Nevertheless, much work continues on protein samples of
unknown impurity, accompanied in some cases by convoluted mechanistic explanations that aim to explain results, when the results are simply corrupted by fouled
electrodes and have nothing to do with the protein under study.
A further important factor contributing to artifactual electrochemical
responses for cytochrome c was revealed in work done by David E. Reed at silver
electrodes [44]. Reed was aiming to use surface-enhanced resonance Raman spectroelectrochemistry with Song-Chen Sun, a postdoctoral in this group, in collaboration with James Terner of this department. This experiment required the use of
electrochemically roughened silver at that time. Again Reed used the derivative
cyclic voltabsorptometry method to show that even puried samples of
cytochrome c exhibit irreversible electrochemistry at silver working electrodes,
much as reported earlier [42]. Single potential step chronoabsorptometry experiments, not shown here, also showed good agreement between experiment and calculated responses. When samples of cytochrome c were taken directly from the
chromatographic purication column, the electrochemistry at silver was quasireversible and stable as shown in Figure 4. The complementary nature of the scan
rate dependence of DCVA compared with cyclic voltammetry is nicely illustrated
in this gure. In DCVA the peak derivative optical response is linear with the
inverse square root of the scan rate, whereas in cyclic voltammetry the peak current

118

Burgess and Hawkridge

Direct Electrochemistry

119

response is linear with the square root of the scan rate. Interestingly, if the optical
absorbance is displayed as a derivative with respect to time, rather than potential,
the peak scan rate dependence is the same as in cyclic voltammetry [39]. This work
conrmed that lyophilization of cytochrome c solutions reintroduces oligomeric
forms that strongly adsorbed on silver, fouling the surface. Indeed, it was then
shown that when fresh samples of cytochrome c are studied straight from a chromatographic column, the electrochemical response at platinum and gold is also
quasi-reversible and stable for extended periods of study [45].
An earlier extensive study by Edmund Bowden had demonstrated the
importance of using electrodes with hydrophilic surfaces in electrochemical studies of cytochrome c [46]. At that time the cyclic voltammetry of cytochrome c at
gold was found to decay with time, but the response could be restored on cleaning the electrode in a soft hydrogen ame that rendered the surface hydrophilic.
It is now clear that the lack of stability was due to these samples being chromatographically puried, lyophilized, and then stored at 4C for later use. This produced oligomeric forms of cytochrome c that fouled metal electrode surfaces.
Metal surfaces are more active to adsorption of protein impurities than tin oxide
and indium oxide and more easily fouled by protein adsorption.
The importance of electrode surface hydrophilicity [47] and the impact of
cytochrome c impurities [48] on electrochemical responses of cytochrome c at
solid electrodes have been quantitatively assayed by Isao Taniguchis group. His
scheme for modifying electrodes chemically with bis(4-pyridyl)disulde
(PySSPy) to impart resistance to adsorption has proven superior to alternative
approaches. Indeed, his use of gold hiol self-assembly chemistry in his initial
paper in 1982 [49,50] was a harbinger of the explosion in interest in thiol selfassembly chemistry at gold that erupted in the mid-1980s.
At this point, the direct electrochemistry of cytochrome c at a host of solid
electrodes had become well controlled, stable, and quasi-reversible. This group
began to then use this platform to study properties of cytochrome c using direct
electrochemical methods, often coupled with optical probes. Studies of the temperature dependence of the formal reduction potential and the heterogeneous
electron transfer kinetics were subsequently reported, and reaction center entropy
values were shown to agree well with earlier reports in work by Kent B. Koller

Figure 3 (A) Cyclic voltammetry of cytochrome c a tin doped indium oxide optically
transparent electrode after purication. Solutions contained [Cyt c] = 73 mM, 0.21 M Tris
and 0.24M cacodylic acid, pH 7.0, electrode area = 0.71 cm2. Potential scan rates in mV/s
are (a) 100; (b) 50; (c) 20; (d) 10; (e) 5; (d) 2. (B) DCVA experiments conducted under the
same conditions. Circles are calculated responses using parameters above and Eo = 0.260
V vs NHE, n = 1.0, DO = DR = 1.2 106 cm2/s, = 57,000 M1 cm1, ko = 1.0 103
cm/s, a 0.5.

Figure 4 (A) DCVA and (B) cyclic voltammetry of cytochrome c at silver. [Cyt c] = 199 M, 0.05 M Na2SO4, electrode
area 1.23 cm2, scan rates in mV/s are (a) 10.4; (b) 5.2; (c) 2.0; (d) 1.0. Circles are calculated responses for ko = 1.5 103
cm2/s, = 0.55, Eo = 0.260 V vs NHE.

120
Burgess and Hawkridge

Direct Electrochemistry

121

[51]. However, interesting temperature-dependent kinetic effects were observed


with the largest rates of electron transfer occurring at temperatures corresponding
to the physiological temperature of the organism. A broader study including pH
followed [52]. Yuan Xiaoling extended this work to comparisons between
cytochrome c from different species with differing amino acid sequences [53] and
good correlations were found with the differences in the thermodynamics and
kinetics of electron transfer. Yuan also conducted some very difcult measurements using potential step chronoelliptometry to probe the dynamics of protein
conformational change that occurs in concert with the reductive and oxidative
electron transfer reactions [54]. These experiments required long periods of signal averaging, and the optimization of signal to noise in Circular Dichroism and
Magnetically Induced Circular Dichroism spectroscopy was challenging. However, the dynamics of the conformational changes that occur on going from the
reduced to the oxidized form of cytochrome c were distinguishably slower than
for the reverse reaction, as shown in Figure 5. In trace (a), the cytochrome c sample is in the oxidized form and the initial forward reduction step shows no difference in the rate of CD change compared with the rate of electron transfer (the
smooth solid line). The back oxidation step shows the CD change to be slower
than the rate of electron transfer. The same situation is shown in Figure 5 trace (b)
in reverse order where the cytochrome c sample is present in the reduced form
rather than the oxidized form. The rate constant for the change on conformation
from oxidized to reduced was 50 s1 whereas the rate constant for reduced to
oxidized was estimated to be 10 s1.
I was most fortunate to have Dr. Song-Chen Sun, from Ron Birkes group at
City College of New York, Queens College, join the group as a postdoctoral at
the same time that a graduate student, David Reed, was in my group. Dr. Sun
brought extensive experience in Surface Enhance Resonance Raman Spectroscopy (SERRS) to my group from work with Birke, and Reed had worked with
James Terner, an expert in time-resolved Raman spectroscopy of heme proteins
and enzymes, as an undergraduate student. With the collaboration of James Terner,
an effort to study the time-resolved structure of the oxidized to reduced and the
reverse reaction was mounted. Reeds work, mentioned above on the voltammetry
of cytochrome c at silver [44], provided the foundation for this research thrust.
However, in this work spin-state marker bands in the SERRS spectra showed that
the heme iron was in a denatured ve coordinate, low-spin state. The native heme
structure for cytochrome c is six coordinate high-spin. The electrochemically
roughened silver surfaces used to enhance the Raman signal apparently denatured
the cytochrome c present at the electrode surface. Attempts to reproduce the one
report in the literature that has described the native conformation of cytochrome c
in both redox states at roughened silver surfaces failed in this group. However, during the course of this work Dave Reed found a new effect as described abovethe
formation of cytochrome c oligomers on lyophilizationthat remains an underappreciated point in the literature.

122

Burgess and Hawkridge

Figure 5 Double potential step chronoelliptometry experiments of cytochrome c.


Results are signal-averaged 1000 times at 417 nm: (a) ferricytochrome c at 275 mM; (b)
ferrocytochrome c at 310 M. Solution is pH 7.0 Tris/cacodylic acid buffer.

Direct Electrochemistry

123

IV. MYOGLOBIN ELECTRON TRANSFER


AND LIGAND-BINDING REACTIONS
Myoglobin is an intriguing molecule because it has the physiological function of
storing dioxygen and that function is tied to the redox state of its heme ironi.e.,
this function depends on the iron being in the reduced, +2 state. The ability to
electrochemically control the redox state of the heme iron in myoglobin and
thereby initiate ligand binding and dissociation reactions has provided a way to
study these reactions. Work in Kuwanas laboratory used mediated indirect
coulometric titrations to probe the formal reduction potential of myoglobin due to
irreversible heterogeneous electron transfer kinetics [8]. As mentioned in an earlier section we became involved with myoglobin due to the clinical interests of
Joyce Stargardt [33]. However, the direct electrochemical reaction was highly
irreversible as seen in both a purely electrochemical study and by spectroelectrochemical kinetic methods [35,40].
As the importance of preparing clean protein solutions became increasingly
evident in work on cytochrome c, similar purication efforts with myoglobin also
had a big effect on improving the heterogeneous electron transfer kinetics in work
done by Bertha C. King [55]. As mentioned earlier, Taniguchis group later quantied the relationship between the hydrophilicity of an electrode surface with the
heterogeneous electron transfer rate constant for myoglobin [47]. The direct electron transfer reaction of myoglobin at a number of solid electrodes now falls in
the quasi-reversible heterogeneous electron transfer kinetic regime.
An exciting collaboration subsequently developed with Professor Brian M.
Hoffman at Northwestern University in work done on the cyanometmyoglobin
complex. The initial question was to determine if the formal reduction potential for
cyanometmyoglobin could be determined by direct electrochemical methods.
Bertha King took on this project and found some very interesting results using simple cyclic voltammetric experiments. The problem is essentially the following:
Electrode: Mb(III)CN + e B Mb(II)CN
Solution:

Mb(II)(CN)

Mb(III)CN,

7 Mb(II) +

Mb(II)CN,

CN

(3)
(4)

and Mb(II) are the cyanometmyoglobin comwhere


plex, the cyanomyoglobin complex, and reduced myoglobin (e.g., deoxymyoglobin). When conducting a conventional potentiometric titration of cyanometmyoglobin to determine the formal reduction potential of Equation (3), the reaction is
not chemically reversible. However, the hope was that by using variable potential
scan cyclic voltammetry it would be possible to observe the reversible or quasireversible reaction of Equation (3) under conditions where it was not inuenced
by the dissociation of the cyanide ligand as shown in Equation (4). This was possible by using a molar excess in cyanide concentration and using modestly fast
potential scan cyclic voltammetry [57].

124

Burgess and Hawkridge

An even more interesting nding in this work was the nearly two-order of
magnitude increase in the heterogeneous electron transfer rate constant for the
cyanometmyoglobin complex compared with the free metmyoglobin case. A
detailed study of this system under conditions in which the concentration of the
cyanide ligand as well as the potential scan rate was varied followed [58]. A good
digital simulation t was obtained that described the thermodynamics, kinetics,
and reaction mechanism of this system. An example of the profound effect of
cyanide on the voltammetry of myoglobin is shown in the two frames in Figure 6
together with the ability to easily control the observation of cyanide dissociation
by simply changing the voltammetric scan rate in Figure 6B. Two conclusions
drawn from this work were the importance of the spin-state of the heme iron in
controlling the heterogeneous electron transfer rate constant and the role of the
ligand eld strength in controlling this factor. Other studies have been conducted
by David J. Cohen, who used different ligands, to further elucidate this overall
reaction mechanism [59]. Lucy Duah-Williams, who extended this work and conducted a careful study of the temperature dependence of these electrode reaction
induced ligand dissociation reactions, was able to determine reaction center
entropy changes [60]. The long-term goal of this work is to better understand how
myoglobin releases oxygen in vivo, a seemingly simple problem, about which no
direct experimental studies have been found. Moreover, the release of oxygen
from oxymyoglobin in vivo is driven by the need to sustain energy production
during respiration. The oxygen that is released is then consumed at the terminal
enzyme in mammalian respiration, cytochrome c oxidase, where it is converted
into water. This of course further requires that the oxygen be transported from the
cytosol region between the inner and outer mitochondrial membranes to the
cytochrome c oxidase oxygen-binding site on the inner surface of the inner mitochondrial membrane.

V. CYTOCHROME C AS MEDIATOR IN INDIRECT COULOMETRIC


TITRATIONS OF CYTOCHROME C OXIDASE
When the electron transfer reaction of cytochrome c became stable and quasireversible at a variety of electrodes through careful control of sample integrity
and electrode surface hydrophilicity, the idea to implement mediated studies of
detergent-solubilized cytochrome c oxidase logically arose. The aim was not to
use a small redox couple as the mediator, but instead cytochrome c, which is the
in vivo reaction partner of this enzyme. Moreover, this system had some features
that mimicked the in vivo case in the following sense. The potential applied to the
electrode to reduce cytochrome c, that then transferred its electron to detergentsolubilized cytochrome c oxidase, was not sufcient to directly reduce dioxygen
(the direct reduction of dioxygen has a large overpotential to direct reduction at

Direct Electrochemistry

125

Figure 6 Background-subtracted cyclic voltammetry of metmyoglobin: (A) 86 M metmyoglobin, pH 7.0; (B) 67 mM metmyoglobin, pH 7.0, with excess cyanide to give
[Mb(III)H2O]:[CN] of 1:300. Scan rates 20, 50, 100, 200 mV/s.

126

Burgess and Hawkridge

electrodes). Additionally, native cytochrome c has a high activation energy for


the direct reduction of dioxygen although this reaction is very favorable thermodynamically. This meant that electrons supplied by the electrode were only
received by cytochrome c, which in turn only reduced cytochrome c oxidase that
nally could reduce dioxygen in a serial reaction that mimics the in vivo case. The
optical absorption data taken during a representative indirect coulometric titration
are shown in Figure 7 for anaerobic conditions [61]. These data were t to calculated responses to determine the formal potential of the heme a3 site without using
exogenous titrants.
The electron transfer properties of this reaction system provided interesting
temperature dependencies, and coupled with differential scanning calorimetry, a
model for the denaturation mechanism was proposed to include the bound cyto-

Figure 7 Difference absorption spectra for an indirect coulometric titration of cytochrome c oxidase. Solution contained 150 M cytochrome c, 11 M cytochrome c oxidase,
and less than 1M oxygen. Charge injected in rst two reductive increments 5 nmol, 10
nmol thereafter. Optical path length 2.3 mm, solution 0.21 M Tris, 0.24 M cacodylic acid
(0.20 M ionic strength), pH 7.0, 0.5% (v/v) Tween 20.

Direct Electrochemistry

127

chrome c/cytochrome c oxidase reaction complex [62]. Reminiscent of the temperature dependence of the electron transfer kinetics and thermodynamics of
cytochrome c, an optimum temperature near the physiological temperature of the
organisms from which the protein and enzyme were isolated was observed. Above
45C, the reaction system irreversibly denatured. All of this work was made possible by cytochrome c oxidase samples provided by Charles R. Hartzell of the Alfred
I. DuPont Institute, who was also a valued collaborator in this research.
Missing in this scheme was the in vivo orientation of the cytochrome c oxidase imparted by the inner mitochondrial membrane in which it resides. This concern led to the work described in the following section, in which a mimic of this
architecture was constructed on an electrode surface.

VI. ELECTRODE SUPPORTED BILAYERS CONTAINING


CYTOCHROME C OXIDASE
The goal was to immobilize cytochrome c oxidase in an electrode-supported lipid
bilayer membrane so that it would undergo direct electron transfer with the electrode and mediate electron transfer from cytochrome c in solution to the electrode. This system was envisioned to better mimic the in vivo conditions of the
reaction between cytochrome c and the oxidase compared to experiments with
detergent-solubilized oxidase [62,63] as described above. Also, this system was
designed to provide a more direct means of probing this reaction couple compared
to experiments with the oxidase inserted into the bilayer membrane of vesicles in
solution [64]. In other experiments probing this reaction couple, the electrons
involved in the catalytic oxidation of reduced cytochrome c by cytochrome c oxidase ultimately reduce another species also present in solution (e.g., molecular
oxygen). Direct electron transfer between the electrode and the cytochrome c oxidase would make it possible to probe the kinetics of the reaction between cytochrome c in solution and the immobilized cytochrome c oxidase. The current generated by enzyme-mediated electron transfer from cytochrome c in solution,
through the cytochrome c oxidase to the electrode, could then be controlled by the
applied electrode potential.
Electron microscopy of the cytochrome c oxidase embedded within bilayer
structures (e.g., vesicles) showed subunit II, which contains the cytochrome c
binding site, extending 80 from the membrane surface [65]. The microscopy
also showed the portion of the enzyme containing the oxygen binding site (heme
a3-Cub) inserted into the membrane as is the case in the inner mitochondrial membrane. The asymmetric distribution of hydrophobic and hydrophilic residues on
the surface of the enzyme (i.e., one end of the enzyme is hydrophilic and the other
end is largely hydrophobic) allows the enzyme to be unidirectionally immobilized
in a bilayer membrane on a gold electrode with the oxygen binding site near to the

128

Burgess and Hawkridge

Figure 8 Model of the cytochrome c oxidase immobilized electrode surface. The octadecyl mercaptan anchor molecules contain sulfur, biological amphiphiles have two tails, and
Cyt is cytochrome c.

electrode surface and the cytochrome c binding site exposed to solution. The
components of the idealized model, Figure 8 [66], will be discussed below.
The initial idea for constructing electrode-supported bilayers containing
cytochrome c oxidase involved the use of Langmuir-Blodgett procedures. John K.
Cullison started this work about ten years ago, but this approach failed to produce
stable responses. Cullison sought alternative methods for achieving stable
responses and devised an approach that combined self-assembly chemistry with
deoxycholate dialysis used to prepare vesicles containing cytochrome c oxidase.
Cullison used this cholate dialysis procedure to prepare stable bilayer membranes
containing the oxidase on gold electrodes functionalized with submonolayers of
octadecyl mercaptan [66]. The octadecyl mercaptan submonolayer, which is covalently bonded to the gold surface through thiol chemistry, anchors the bilayer to the
electrode by becoming incorporated into the lipid membrane. Voltammetric data
showed direct electron transfer between the oxidase and the gold electrode, and
spectroelectrochemical data veried that the enzyme could both reduce and oxi-

Direct Electrochemistry

129

dize cytochrome c in solution [66]. However, only about one in eight electrodes
showed this behavior, as reported in this paper. We now attribute this to a lack of
control of the surface coverage of octadecyl mercaptan on gold. Although thiol
monolayer structures can be reliably prepared with substantial reaction times (i.e.,
12 hours [67]), consistently forming submonolayer coverages of octadecyl mercaptan on gold quartz crystal microbalance (QCM) electrodes from dilute solution
was not possible in this laboratory (see below). As discussed below, the thiol selfassembly reaction can be better controlled at gold QCM electrodes that have been
coated with 1.6 monolayers of electrodeposited silver [68]. The electrodeposited
silver masks chromium at the surface of the gold QCM electrodes. Chromium is
used as an adhesion layer between the quartz and gold. Controlling the surface coverage of octadecyl mercaptan was found to be a critical step in the reliable implementation of John Cullisons method. The dialysis step for constructing bilayer
membranes containing cytochrome c oxidase was also rened. The original dialysis conditions were taken from the literature for reconstituting cytochrome c oxidase into vesicles in solution [64]. Bilayers were formed by placing gold electrodes
modied with about 0.5 monolayer of thiol into dialysis tubing for an electrode
preparation time of ve to six days. In this work, a simple, small-volume, dualchambered cell was constructed, and dialysis was complete in 12 hours. This cell
design, which is shown in Figure 9 [69], is just another variation on the design
developed in Ted Kuwanas laboratory thirty years ago [8]. Additionally, the electrochemical behavior of the immobilized oxidase can be evaluated using this procedure without further handling of the modied electrode.
Dr. Zoia Nikolaeva came to this lab from the group of Professor Mikahail
Smirnov of St. Petersburg University, Russia, to assist with the isolation of
cytochrome c oxidase from fresh beef heart. She isolated about 100 mg of
cytochrome c oxidase, more sample than had ever been available for this work in
this laboratory. This was a major contribution, as we had been struggling with the
task of consistently isolating oxidase with high activity for ve years. With this
supply of sample in hand, in work with another graduate student in the group,
Melissa C. Rhoten, it was demonstrated that the oxidase undergoes the resting to
pulsed state kinetic transition within the electrode supported bilayer [69,70]. Also,
the immobilized oxidase exhibits biphasic kinetics with respect to cytochrome c
concentration [71]. Both of these characteristics are discussed below with respect
to mechanisms of oxidase-mediated electron transfer from protein in solution to
the electrode.
Formation of a submonolayer of octadecyl mercaptan on the electrode is the
rst step in the procedure for forming lipid bilayer membranes containing
cytochrome c oxidase on gold QCM electrodes. Construction of a QCM in our
laboratory was greatly facilitated by the generous advice and guidance of Stanley
Bruckenstein, who published the original circuit diagram for this instrument [72].
Initially several cleaning procedures were applied to the gold QCM electrodes,

130

Burgess and Hawkridge

Figure 9 Cell used in preparing and characterizing cytochrome c oxidase in the electrode-supported bilayer membrane. Cell dimensions are 2 5 5 cm. The QCM electrode
and the two cell parts are clamped together.

including an oxygen plasma, a UV surface cleaner, and aggressive chemical treatments, to reproduce the rate of octadecyl mercaptan self-assembly. Despite these
efforts, the measured rates of thiol self-assembly varied widely (i.e., several
orders of magnitude in time) from dilute thiol solution. It had been shown that
thiol monolayers on silver are oriented more normal to the surface plane [73] and
that the sulfur silver bond is stronger than the sulfur gold bond [73,74]. This led
to experiments using electrodeposited silver on gold QCM electrodes as the substrate for bilayer formation [68]. Using the QCM to monitor the self-assembly
kinetics of octadecyl mercaptan at the silver-coated gold QCM electrode surface,
a selected submonolayer coverage could be prepared in no more than about ten
minutes. This change in procedure proved to be very important in this work.
Silver is electrodeposited without added supporting electrolyte to avoid
anion discharge problems. The contribution of migration to mass transfer results

Direct Electrochemistry

131

in nearly linear rates of deposition of silver that depend on the applied potential.
Electrodes coated with 1.6 monolayers of silver were found to work best for
reproducing octadecyl mercaptan submonolayer coverages and can be routinely
prepared by double-potential step experiments [68].
Submonolayers of octadecyl mercaptan are formed on silver-coated gold
QCM electrodes from dilute solution (e.g., 2 M), and gentle nitrogen bubbling
provides enough convection to maintain mass transfer. Exact control of mass
transfer using nitrogen bubbling is not possible but reaction times were generally
about ten minutes. The data shown in Figure 10 are representative for (a) the most
commonly observed rate, (b) a slower mass transfer example, and (c) the stability of the QCM resonant frequency [68]. Irrespective of the differences in mass
transfer rate, the reaction is stopped after a 20 Hz shift in frequency, which indicates one half monolayer coverage.
The structures of alkane thiol monolayers and submonolayers on gold [75]
and silver [76] have been studied using STM. In general, the structure and mobility of thiol molecules on gold and silver changes at threshold surface coverages.

Figure 10 Representative QCM responses for (a) a typical OM deposition rate, (b) a
slow OM deposition rate, and (c) the baseline of the QCM on injecting only an ethanol sample. The electrode area is 0.2 cm2 and the frequency integration time is 850 ms.

132

Burgess and Hawkridge

At low coverages, thiol molecules are mobile on gold and silver. With increasing
surface coverage, islands with the alkane carbon chain tails parallel to the surface
plane form and nucleate to cover the surface. Islands with the alkane tails normal
to the surface plane consequently form and nucleate to form a complete monolayer structure. If thiol islands exist on the substrates used in this work prior to
lipid deposition, lateral hydrophobic interactions between alkane carbon tails and
the lipids, deoxycholate, and cytochrome c oxidase may disperse the islands,
leading to a thiol submonolayer with the thiol molecules more evenly spaced over
the electrode surface.
Using the silver-coated gold QCM electrodes functionalized with a controlled submonolayer surface coverage of octadecyl mercaptan and the dual chamber dialysis cell for bilayer deposition, over 95% of the electrodes worki.e., they
are successfully modied with functional cytochrome c oxidase [69]. Figure 11
shows cyclic voltammograms conducted at three different representative oxidasemodied electrodes and a control experiment for a lipid bilayer membrane containing no immobilized oxidase [69]. Some difference in peak current and wave

Figure 11 Cyclic voltammograms of three different cytochrome c oxidase modied


electrodes (unlabeled). Scan (a) is the control of a lipid bilayer modied electrode containing no cytochrome c oxidase. The scan rate is 20 mV/s, solution contains 0.1 M phosphate,
pH 7.4, and the electrode area is 0.2 cm2.

Direct Electrochemistry

133

shape is apparent between oxidase-modied electrodes. However, the oxidative


wave corresponds well with the formal potential (350 mV) of the oxygen-binding
site (heme a3 and Cub), indicating electron transfer between this site and the electrode. The formal potential of the primary electron acceptor (Cua) from
cytochrome c in solution is 190 mV [13]. The slow reduction of the cytochrome c
oxidase relative to the oxidation rate, as indicated by the voltammetric response,
may be due to the cytochrome c oxidase being unidirectionally inserted into the
bilayer, as shown in Figure 8. Oxidative scans initiate electron transfer through the
cytochrome c oxidase in the direction of electron ow that occurs in vivo. John
Cullisons spectroelectrochemical data showed that the immobilized oxidase
could both oxidize and reduce cytochrome c in solution [66].
With a method in hand for routinely constructing cytochrome c oxidase
modied electrodes that exhibited direct electron transfer between the electrode
and the oxidase, amperometry was used to detect reduced cytochrome c in solution
at the oxidase-modied electrodes in a ow injection analysis format [69]. The
dialysis cell was equipped with a wall jet inlet to direct cytochrome c solution past
the oxidase-modied electrodes. Figure 12 shows the current response for three
sequential reduced cytochrome c injections. Control experiments conducted at
bilayer modied electrodes containing no oxidase showed current responses that
are about 2% of those shown in Figure 12. This response may be due to changes in
electrode capacitance and/or cytochrome c reacting at bilayer defect sites on the
electrode. QCM measurements showed that no cytochrome c incorporated into the
bilayer. However, cytochrome c was electrostatically held at the surface of the bilayer membrane at lower ionic strength [69].
The concentration dependence of cytochrome c binding to a cytochrome c
oxidase modied electrode under lower ionic strength conditions (0.010 M phosphate buffer) is shown in Figure 13 (Traces (bd)) [69]. As shown in Trace (a),
no binding is observed at 0.10 M phosphate, reecting diminished electrostatic
attraction between the highly charged cytochrome c molecules and the polar head
groups of the lipid bilayer membrane.
Controlled potential amperometry of anaerobic solutions of reduced
cytochrome c show a kinetic transition of the bilayer-conned cytochrome c oxidase from its resting to pulsed kinetic state [70]. The current responses show a
kinetic transition involving an increase in the rate of oxidase mediated electron
transfer from cytochrome c to the electrode. A turnover rate of about one electron
per second is required for transition to the pulsed kinetic state. A key advantage
of studying the kinetics of reaction between cytochrome c and the immobilized
cytochrome c oxidase in these experiments is that the turnover rate can be controlled by simple variation of the concentration of reduced cytochrome c. Figure
14 shows that at lower turnover rates (5 M cytochrome c) the measured current
is constant. At higher turnover rates (10 M cytochrome c), the measured current
undergoes a transition over time. These data show that the distribution of the

134

Burgess and Hawkridge

Figure 12 Flow injection analysis of reduced cytochrome c reacting at a cytochrome c


oxidase modied electrode. Three sequential 83 L injections of 100 M reduced
cytochrome c. Applied potential is 472 mV vs NHE, solution contains 0.1 M phosphate
buffer, pH 7.4, and the electrode area is 0.2 cm2.

cytochrome c oxidase between its resting and pulsed states is governed by its
turnoverratei.e., the concentration of cytochrome c that reacts with the immobilized oxidase. As turnover rate increases, the cytochrome c oxidase becomes
more fully reduced and the pulsed state is favored. At the minimum turnover rate
required to induce transition from the resting to the pulsed state (0.5 electrons/s),
about 100 four-electron oxidation/reduction cycles of the oxidase occur before
transition to the pulsed state begins. However, the number of turnovers required
to induce the pulsed state depends on turnover rate. At higher turnover rates, the
number of turnovers required to induce the resting to pulsed state kinetic transition decreases. This behavior was rst hypothesized two decades ago by Antonini

Direct Electrochemistry

135

Figure 13 QCM-FIA data of oxidized cytochrome c at a lipid bilayer modied electrode


containing no cytochrome c oxidase: (a) 10 M cytochrome c in 0.10 M phosphate buffer; (b)
2.5 M cytochrome c in 0.010 M phosphate buffer; (c) 5 M cytochrome c in 0.010 M phosphate buffer; and (d) 10 M cytochrome c in 0.010 M phosphate buffer. Buffer pH is 7.4 and
the electrode area is 0.2 cm2. Experiments were conducted under open circuit conditions.

et al. [63], who originally reported the kinetic transition and introduced the notion
of resting and pulsed states for cytochrome c oxidase.
In evaluating the kinetics of this reaction under varied cytochrome c concentrations, a model has been assumed in which the rate-limiting step is not electron transfer from the cytochrome c oxidase to the electrode. The rate-limiting
step is either electron transfer though the cytochrome c oxidase or between the
cytochrome c in solution and the cytochrome c oxidase, depending on the experimental conditions. The biphasic kinetic response with increasing cytochrome c
concentration is ascribed to the cytochrome c oxidase cycling between two conformations [77]. The two conformations form when electron transfer from protein
in solution to the primary electron-accepting redox site of the oxidase (Cua)
occurs, and when electron transfer through the oxidase from Cua to the oxygenbinding site (Cub-heme a3) occurs. Oscillation between these two conformations
also involves coupling of proton translocation with electron transfer through the

136

Burgess and Hawkridge

Figure 14 Anaerobic controlled potential amperometric analysis of reduced cytochrome


c at an oxidase modied electrode: (a) 5 M and (b) 10 M cytochrome c. The sample ow
rate is 0.5 mL/min and the applied potential is 472 mV vs NHE.

oxidase. As the cytochrome c concentration is increased above 10 M, the rate of


electron transfer from cytochrome c in solution to the electrode becomes limited
by electron transfer through the cytochrome c oxidase. This is due to the rate of
mediated electron transfer becoming limited by the rate of the conformation
change of the oxidase from the electron injection conformation to the conformation required for electron transfer through the oxidase.
The resting to pulsed state kinetic transition involves an increase in the rate
of intramolecular electron transfer through the oxidase (i.e., from Cua to CubHeme a3) [63]. Unlike the conformations involved with proton pumping, the resting and pulsed conformational states persist over minutes. Observation of the
resting to pulsed state kinetic transition indicates that electron transfer through the

Direct Electrochemistry

137

cytochrome c oxidase is rate limiting. Both current plateaus (corresponding to the


resting and pulsed states) shown in Figure 14b for the 10 M injection are less
than twice that observed at 5 M. As discussed above, this is the concentration
range where electron transfer through the oxidase becomes limiting. The transition of the cytochrome c oxidase from its resting to pulsed state apparently
requires that electron transfer through the oxidase be rate limiting. This model is
consistent with the fact that the pulsed state is favored when the cytochrome c oxidase is fully reduced [63].
Cytochrome c derived from tuna sh reacts slower with the immobilized
oxidase compared to cytochrome c derived from horse at the same concentration.
This is due to a smaller protein/enzyme dissociation constant for tuna protein
compared to horse protein. These data show that electron transfer from tuna
cytochrome c to the enzyme is rate limiting even at concentrations of 10 M and
above.
The topology of the cytochrome c oxidase modied electrodes was characterized using tapping mode scanning force microscopy (TM-SFM) in Marc
Porters laboratory [78]. Images of the electrode-supported bilayer membranes
containing cytochrome c oxidase and electrode-supported lipid bilayer membranes devoid of the enzyme were collected. These images gave the rst direct
verication of the structural model. TM-SFM revealed the cytochrome c oxidase
embedded in the bilayer membrane with a portion of the enzyme extending from
the bilayer. The topographical image in Figure 15a shows several cone-shaped
features, which protrude by as much as 2 nm from the surrounding bilayer, that
are not present in images of bilayer membranes devoid of oxidase. The phasecontrast images also reveal the presence of immobilized oxidase as shown in Figure 15b. Small domains of large phase shift are surrounded by larger regions of
signicantly lower phase shift. The locations of the regions of large phase shift
coincide with those of the cone-shaped features in the topographical images.
Phase-contrast images of the bilayer membranes with no cytochrome c oxidase
show only a low phase shift over the imaged region. Therefore, the features present in the cytochrome c oxidase containing bilayer-modied electrodes are
assigned to monomers, dimer, and larger aggregates of immobilized oxidase. The
size of the smallest feature (14 nm diameter) is comparable to the expected 8 nm
diameter of an individual cytochrome c oxidase molecule on taking into account
the magnication that occurs because of the size and shape of the probe tip. The
probe tip was calibrated by imaging 10 nm and 30 nm gold particles, which
appear to have more than doubled in size upon imaging. Images over much larger
areas (e.g., 2 m2) allow an estimate of the cytochrome c oxidase surface coverage (20%) on the electrodes. The estimate of oxidase surface coverage was
required for calculating the turnover rates and turnover numbers discussed above
for reaction of cytochrome c in solution with the immobilized oxidase. This visual

Figure 15 Tapping mode-scanning force microscopy images of cytochrome c oxidase immobilized in a lipid bilayer membrane on an
electrode. (a) topographic image, (b) phase contrast image.

138
Burgess and Hawkridge

Direct Electrochemistry

139

picture of the electrode-supported bilayer containing cytochrome c oxidase is


consistent with the cartoon model depicted in Figure 8.
Progress in preparing and characterizing electrode-supported bilayers containing cytochrome c oxidase has been slow over the past ten years. Now the
potential for extending this structure to other membrane integral enzymes that
exhibit direct electron transfer at electrodes holds promise in several areas of
interest to our research group. Although it is premature to speculate at this point,
this may prove to be a general procedure for constructing this architecture as other
oxidases and other membrane enzymes are studied.

ACKNOWLEDGMENTS
The authors gratefully acknowledge the support of this work by the National Science Foundation. Critical collaborations with Charles R. Hartzell of the A. I.
duPont Institute, Isao Taniguchi of Kumamoto University, and Dr. Zoia Nikolaeva and Professor Mikhail Smirnoff of St. Petersburg University are gratefully
acknowledged.

REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.

W. M. Clark, Oxidation-Reduction Potentials of Organic Systems, The Williams &


Wilkins Company, Baltimore, 1960.
H. Theorell and . kesson, J. Am. Chem. Soc. 63:1818 (1941).
J. F. Taylor and V. E. Morgan, J. Biol. Chem. 217:15 (1942).
F. L. Rodkey and E. G. Ball, J. Biol. Chem. 182:17 (1950).
R. Szentrimay, P. Yeh, and T. Kuwana, ACS Symposium Series, Vol. 38, American
Chemical Society, Washington, D. C., 1977, pp. 143169.
M. L. Fultz and R. A. Durst, Anal. Chim. Acta 140:1 (1982).
M. Ito and T. Kuwana, J. Electroanal. Chem. 32:415 (1971).
F. M. Hawkridge and T. Kuwana, Anal. Chem. 45:1021 (1973).
W. R. Heineman, T. Kuwana, and C. R. Hartzell, Biochem. Biophys. Res. Commun.
49:1 (1972).
W. R. Heineman, T. Kuwana, and C. R. Hartzell, Biochem. Biophys. Res. Commun.
50:892 (1973).
L. N. Mackey, T. Kuwana, and C. R. Hartzell, FEBS Lett. 36:326 (1973).
Y. Fujihira, T. Kuwana, and C. R. Hartzell, Biochem. Biophys. Res. Commun. 61:538
(1974).
J. L. Anderson, T. Kuwana, and C. R. Hartzell, Biochemistry 15:3847 (1976).
R. Szentrimay and T. Kuwana, Anal. Chem. 59:1879 (1978).
W. R. Heineman, B. J. Norris, and J. F. Goelz, Anal. Chem. 47:79 (1975).

140
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.

Burgess and Hawkridge


E. F. Bowden, F. M. Hawkridge, and H. N. Blount, in Comprehensive Treatise of
Electrochemistry, Vol. 10, Plenum Press, New York, 1985, pp. 297346.
F. A. Armstrong, H. A. O. Hill, and N. J. Walton, Acc. Chem. Res. 21:407 (1988).
P. Bianco and J. Haladjian, Biochimie 76:605 (1994).
F. M. Hawkridge and I. Taniguchi, Comments Inorg. Chem. 17:163 (1995).
F. M. Hawkridge and I. Taniguchi, in The Porphyrin Handbook, K. M. Kadish, K.
M. Smith, and R. Guilard (Eds.), Academic Press, New York, 2000, Vol 8, pp.
191202.
B. Ke, R. E. Hansen, and H. Beinert, Proc. Natl. Acad. Sci. USA 70:2491 (1973).
L. A. Summers, The Bipyridinium Herbicides, Academic Press, Inc., New York,
1980.
L. H. Rickard, H. L. Landrum, and F. M. Hawkridge, Bioelectrochem. Bioenerg.
5:686 (1978).
B. Ke, E. Dolan, K. Sugahara, F. M. Hawkridge, S. Demeter, and E. R. Shaw, Plant
Cell Physiol. 187 (1977).
B. Ke, F. M. Hawkridge, and S. Sahu, Proc. Natl. Acad. Sci. USA, 73:2211 (1976).
F. M. Hawkridge and B. Ke, Anal. Biochem. 78:76 (1977).
P. R. Moses, L. Wier, and R. W. Murray, Anal. Chem. 47:1882 (1975).
C. M. Elliott and R. W. Murray, Anal. Chem. 48:1247 (1976).
H. L. Landrum, R. T. Salmon, and F. M. Hawkridge, J. Am. Chem. Soc. 99:3154
(1977).
P. Yeh and T. Kuwana, Chem. Lett. (Japan) 1145 (1977).
M. J. Eddowes and H. A. O. Hill, J. C. S. Chem. Commun. 771 (1977).
K. Niki, T. Yagi, H. Inokuchi, and K. Kimura, J. Electrochem. Soc. 124:1890 (1977).
J. F. Stargardt, F. M. Hawkridge, and H. L. Landrum, Anal. Chem. 50:930 (1978).
S. G. Weber and W. C. Purdy, J. Electroanal. Chem. 115:1975 (1980).
J. F. Castner and F. M. Hawkridge, J. Electroanal. Chem. 143:217 (1983).
D. E. Albertson, H. N. Blount, and F. M. Hawkridge, Anal. Chem. 51:556 (1979).
J. H. Christie, G. Lauer, and R. A. Osteryoung, J. Electroanal. Chem. 7:60 (1964).
E. E. Bancroft, H. N. Blount, and F. M. Hawkridge, Anal. Chem. 53:1862 (1981).
E. E. Bancroft, J. S. Sidwell, and H. N. Blount, Anal. Chem. 53:1390 (1981).
E. F. Bowden, F. M. Hawkridge, and H. N. Blount, Bioelectrochem. Bioenerget.
7:447 (1980).
E. F. Bowden, F. M. Hawkridge and H. N. Blount, Adv. Chem. Series, K. M. Kadish
(Ed), American Chemical Society, Vol. 201, 1982, pp. 159171.
E. E. Bancroft, H. N. Blount, and F. M. Hawkridge, Biochem. Biophys. Res. Commun. 101:1331 (1981).
E. F. Bowden, F. M. Hawkridge, J. F. Chlebowski, E. E. Bancroft, C. Thorpe, and H.
N. Blount, J. Am. Chem. Soc. 104:7641 (1982).
D. E. Reed and F. M. Hawkridge, Anal. Chem. 59:2334 (1987).
S-C. Sun, D. E. Reed, J. K. Cullison, L. H. Rickard, and F. M. Hawkridge,
Mikrochim. Acta 3:97 (1988).
E. F. Bowden, F. M. Hawkridge, and H. N. Blount, J. Electroanal. Chem. 161:355
(1984).

Direct Electrochemistry
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.
72.
73.

141

M. Tominaga, T. Kumagai, S. Takita, and I. Taniguchi, Chem. Lett. (Japan) 1771


(1993).
I. Taniguchi, H. Kurihara, K. Yoshida, M. Tominaga, and F. M. Hawkridge, Denki
Kagaku 66:1043 (1992).
I. Taniguchi, K. Toyosawa, H. Yamaguchi, and K. Yasukouchi, J. Chem. Soc. Chem.
Commun. 1032 (1982).
I. Taniguchi, K. Toyosawa, H. Yamaguchi, and K. Yasukouchi, J. Electroanal.
Chem. 140:187 (1982).
K. B. Koller and F. M. Hawkridge, J. Am. Chem. Soc. 107:7412 (1985).
T. Ikeshoji, I. Taniguchi, and F. M. Hawkridge, J. Electroanal. Chem. 270:297
(1989).
X. Yuan, F. M. Hawkridge, and J. F. Chlebowski, J. Electroanal. Chem. 350:29
(1993).
X. Yuan, S-C. Sun, F. M. Hawkridge, J. F. Chlebowski, and I. Taniguchi, J. Am.
Chem. Soc. 112:5380 (1990).
B. C. King and F. M. Hawkridge, J. Electroanal. Chem. 237:81 (1987).
B. C. King and F. M. Hawkridge, Talanta 36:331 (1989).
M. J. Natan, D. Kuila, W. W. Baxter, B. C. King, F. M. Hawkridge, and B. M. Hoffman, J. Am. Chem. Soc. 112:4081 (1990).
B. C. King, F. M. Hawkridge, and B. M. Hoffman, J. Am. Chem. Soc. 114:10603
(1992).
D. J. Cohen, B. C. King, and F. M. Hawkridge, J. Electroanal. Chem. 447:53 (1998).
L. Duah-Williams and F. M. Hawkridge, J. Electroanal. Chem. 466:177 (1999).
R. C. Long, F. M. Hawkridge, and C. R. Hartzell, J. Electroanal. Chem. 198:89
(1986).
R. C. Long, F. M. Hawkridge, J. F. Chlebowski, and C. R. Hartzell, J. Electroanal.
Chem. 256:111 (1988).
E. Antonini, M. Brunori, A. Colosimo, C. Greenwood, and M. T. Wilson, Proc. Natl.
Acad. Sci. USA 74:3128 (1977).
E. Racker, J. Membr. Biol. 10:221 (1972).
R. Henderson, R. A. Capaldi, and J. S. Leigh, J. Mol. Biol. 112:631 (1977).
J. K. Cullison, F. M. Hawkridge, N. Nakashima, and S. Yoshikawa, Langmuir,
10:877 (1994).
C. D. Bain, E. B. Troughton, Y. Tao, J. Evall, G. M. Whitesides, and R. G. Nuzzo,
J. Am. Chem. Soc. 111:321 (1989).
J. D. Burgess, and F. M. Hawkridge, Langmuir 13:3781 (1997).
J. D. Burgess, M. C. Rhoten, and F. M. Hawkridge, Langmuir 14:2467 (1998).
J. D. Burgess, M. C. Rhoten, and F. M. Hawkridge, J. Am. Chem. Soc. 120:4488
(1998).
M. C. Rhoten, J. D. Burgess, and F. M. Hawkridge, Electrochim. Acta 45:2855
(2000).
S. Bruckenstein, M. Michaliski, A. Fensore, Z. Li, and A. R. Hillman, Anal. Chem.
66:1847 (1994).
M. A. Bryant and J. E. Pemberton, J. Am. Chem. Soc. 113:8284 (1991).

142
74.
75.
76.
77.
78.

Burgess and Hawkridge


J. B. Schlenoff, M. Li, and H. Ly, J. Am. Chem. Soc. 117:12528 (1995).
G. E. Poirier, Langmuir 13:2019 (1997).
A. Dhirani, M. A. Hines, A. J. Fisher, O. Ismail, and P. Guyot-Sionnest, Langmuir,
11:2609 (1995).
G. T. Babcock and M. Wikstrm, Nature 356:301 (1992).
J. D. Burgess, V. W. Jones, M. D. Porter, M. C. Rhoten, and F. M. Hawkridge, Langmuir 14:6628 (1998).

5
Voltammetric Investigations
of Iron-Sulfur Clusters in Proteins
Fraser A. Armstrong
Oxford University, Oxford, England

I. INTRODUCTION: DIFFICULTIES IN STUDYING CLUSTER


REACTIONS IN PROTEINS
Iron-sulfur clusters are found throughout Biology as active sites of proteins [14].
As shown in Figure 1, they exist in a variety of forms2Fe, 3Fe, 4Fe, and higher
nuclearitiescomprising bridging inorganic suldes, usually (but not always)
coordinated to the protein by the side-chain thiolates (RS) of cysteine ligands,
and very occasionally containing a metal other than Fe. These ligands often lie in
a recognizable sequence of amino acids (the binding motif) that is characteristic of a particular cluster and enables its presence in a protein to be predicted from
the gene. Iron-sulfur clusters function in a wide range of biological processes:
Table I shows that aside from their long-established place among electron-transfer centers, known roles (to the present century) include redox catalysis (nitrogenase), acid-base catalysis (e.g., aconitase), atom-transfer catalysis (biotin synthase), transcriptional regulation (e.g., the fumarate-nitrate regulatory protein
FNRa transcription factor that senses the oxygen tension in bacterial cells in
order to switch between aerobic and anaerobic respiration), translational regulation (e.g., the iron-regulatory protein IRP that senses Fe levels to control Fe
uptake and storage by eukaryotic cells), and oxygen-sensitive enzymatic regulation [111]. Aside from these activities, iron-sulfur clusters are important in stabilizing protein conformation, possibly serving a role like that of disulde bridges
or structural Zn sites. Special proteins coordinating clusters are also implicated in
cluster biosynthesis [12]
143

Figure 1 Structures of the major classes of Fe-S clusters. The larger clusters are found in nitrogenase: [8Fe7S]ox and [8Fe-7S]red are oxidized and reduced forms of the P-cluster, while [Mo7Fe-8S] is believed to be
the site at which N2 is reduced.

144
Armstrong

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

145

Table 1 Functions of Fe-S Clusters


Examples
Ferredoxins
and multicentered
ET enzymes
Nitrogenase
Aconitase
Biotin synthase
Iron regulatory protein

Cluster types

2Fe, 3Fe, 4Fe


[Mo7Fe-8S]
[8Fe7S]
4Fe (active)
3Fe (inactive)
4Fe/2Fe
4Fe/3Fe/apo

Fumarate-nitrate
regulator

4Fe/2Fe

SoxR

2Fe

NifU

2Fe

Function

Electron transfer
Redox catalysis
(N2 B NH3)
Dehydratase
Radical/atom transfer
Translation regulator
(Fe sensor)
Transcription factor
(Oxygen sensor)
Transcription factor
(Oxygen sensor)
Biological assembly
of Fe-S clusters

Ref.

15
5,6
4,7,8
9
8
10
11
12

This diversity directs us to examine and understand cluster reactivities that


transcend their long-established capability to serve in fast single-electron transfer.
A direct role in catalysis requires that substrate molecules bind at one (or more) Fe
subsites and are then activated and chemically transformed in processes that may
involve bond polarization and transfer of electrons, protons, atoms, and radicals.
Regulatory functions require that the cluster is sensitive to some changing aspect of
the environment, such as (trace) levels of O2, superoxide, NO, and Fe (or perhaps
other metals)the point being that these agents may alter the structure and properties of the cluster and hence translate into a change in stability or conformation of
the protein and consequential modulation of DNA or RNA binding. We are also
reminded that these clusters are tiny nano-chunks of mineral-like inorganic material; questions are therefore raised about how they are assembled and transferred
from one protein to another [12] and how they are degraded (disassembled). Ironsulfur clusters are thermodynamically unstable with respect to oxide formation and
tend to be O2-sensitive even in higher oxidation levels; consequently, their very
existence under oxidizing conditions lies in critical balance. Figure 2 illustrates
how (at least in principle) clusters may exist in many different oxidation levels covering a wide potential range; in practice, however, access to these levels is restricted
by the protein environment. Through these changes in oxidation level, the chemical
properties of clusters become linked to the local electrochemical potential: for
example, as we will discuss below, certain reactivities of [3Fe-4S] clusters, such as

Figure 2 The oxidation levels that are possible for 2Fe, 3Fe, and 4Fe clusters. Oxidation levels having an odd number of d-electrons are
usually easy to detect by EPR.

146
Armstrong

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

147

the ability to bind metal ions or protons, change dramatically upon reduction from
the 1+ to the 0 level.
Our understanding of the fundamental chemistry of Fe-S clusters owes much
to studies of small analogue compounds. Compounds have been prepared and
characterized that are analogues of [2Fe-2S]2+/+ (i.e., [Fe2S2((S)2Ar)2]2/3), [4Fe4S]2+/+ (i.e., [Fe4S4(SR)4]2/3), and [3Fe-4S] ([Fe3S4(SR)3]2/3); clusters of higher
nuclearity, perhaps having some as-yet unrealized biological relevance, have also
been synthesized, including [6Fe-6S] [1,13]. Further efforts have led also to the
synthesis of heterometal cubanes and sub-site differentiated clusters such as the
thiolate-bridged [4Fe-4S]Fe-porphyrin active site of sulte reductase [1,13]. In all
these studies, the practical emphasis has been on identifying and examining clusters
that are sufciently stable to allow their characterization as isolated species. However, we are certain that the protein plays a major role in determining the properties
of Fe-S clusters; importantly, unstable, reactive intermediates such as fragments,
weakly bound adducts, and species generated under extremes of electrochemical
potential may be difcult to detect and characterize. Furthermore, clusters may vacillate between species differing in ligation or nuclearity, perhaps in rapid response
to changes in external conditions that are so subtle as to give the impression of
chaotic behavior. Consequently, the goal should be to study reactions of clusters in
their natural protein environment. We must therefore be able to (a) generate and
examine species formed at extremes of reduction potential, (b) keep track of vacillatory cluster systemsthose lacking sufcient stability in any single form to be
easily studied, and (c) resolve the complex reactions that are coupled to electron
transfer. These are each characteristics of the catalysts and sensors, mentioned
above, that have become a major focus of interest.
In principle, UV-visible spectrophotometry provides the simplest and most
convenient way to monitor reactions of metalloproteins, and this is much used for
heme proteins: however, the absorption spectra of Fe-S clusters are typically broad
and featureless because they comprise numerous overlapping charge-transfer transitions. Although techniques such as NMR are being used with increasing success
to study small Fe-S proteins under ambient conditions, the established Fe-S detection methods such as EPR (electron paramagnetic resonance) require samples to be
frozen and subsequently examined at cryogenic temperatures [14]. Moreover, only
those species with noninteger spin (Figure 2) are usually detectable by conventional EPR methods. Another factor hindering studies is that for systems with very
negative reduction potentials (i.e., <550 mV), it becomes very difcult to generate reduced species by chemical titrants such as sodium dithionite.
My intention with this chapter is to illustrate how dynamic electrochemical
methods can be used to investigate Fe-S clusters in ways that provide a perspective different from those of conventional methods. We will not be concerned with
studies in which the sole aim is to determine reduction potentials, except where
these measurements pose a problem; for example, where highly reducing or unstable species are involved. Instead, we will emphasize how these methods can

148

Armstrong

provide the all-important handles on active speciesserving both as qualitative


trailblazers (i.e., to discover new reactivities or to visualize complex behavior)
and as quantitative tools for determining thermodynamic and kinetic constants
that are difcult to measure by other means. The important complementarity with
the more established spectroscopic methods will become obvious.

II.

THE CASE FOR VOLTAMMETRIC METHODS

A.

Why Voltammetry?

If voltammetric methods were used only to measure reduction potentials of stable,


chemically reversible electron-transfer systems, then our interest would arise
solely from their relative speed and convenience. Indeed, at least in principle,
voltammetry is easier, less time-consuming and is more economical on sample
quantity than redox potentiometry: for Fe-S clusters, this usually involves measuring EPR spectra of numerous samples frozen at intervals throughout a slow titration
in the presence of mediators! However, errors may arise if the redox equilibrium
shifts during the freezing processa possibility that has been raised in relation to
measuring how reduction potentials of proteins from hyperthermophiles vary at
high temperatures [15]. The capability for making in situ measurements also means
that with a suitable cell, it is possible to determine reduction potentials at very high
pressures, thus revealing how compressibility varies with oxidation state [16].
Another advantage of direct electrochemical methods is that it is possible to generate species at potentials well outside the thermodynamic limits of aqueous solution
(0.8 to 0.4V vs. SHE at pH 7)the point being that even if such species are kinetically stable, it may be impractical to generate them using chemical titrants that may
themselves decompose in water. Moreover, voltammetric methods have further
capabilities: one of these is that the voltammetric response can provide an identication tag in the same way as a spectroscopic signal; another is that voltammetric
methods are dynamic, in the sense that they also reveal kinetics and permit detection
of transient species in which the reaction itself is induced and controlled by varying
the potential applied to the electrode. Themes that we will see appearing throughout
this chapter are the coupled electrochemical-chemical (EC) reaction and the representation of coupling in terms of square schemes [17]: these are important aspects
of both electrochemistry and biological redox chemistry, and they make voltammetry a rst-choice method for untangling reaction complexities.
Proteins present a special problem for electrochemistry because their interaction with most common types of electrode surface does not lead to an electrochemical response. Two commonly quoted reasons are (1) that the protein shell
acts as an insulator (although a recent view is that electron transfer is unlikely to be
a problem if the donor-acceptor separation is less than 14 ) [18]; and (2) that the
protein denatures upon contact with the electrode and there is strong adsorption of
the resulting nonnative and electroinactive material [19]. In many cases these

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

149

problems and misconceptions have been overcome, and voltammetric investigations of increasing intricacy are being directed at subjects ranging from heme proteins to blue copper proteins, and from small electron carriers to large complex
enzymes with molecular mass > 100 kDa [20]. Successes have resulted largely
from the use of appropriate electrode materials, specic types of surface modication, optimization of interfacial electrostatics, and care in sample preparation. It
has long been known that protein-protein interactions are important in biological
electron transfer; consequently, efforts have been made to modify the electrode
surface and solution interface in such a manner as to mimic physiological environments. For example, oxide surfaces such as indium oxide, polished carbon (edgeoriented pyrolytic graphitePGE, or glassy carbon) or noble metals (gold or silver) modied with various organic adsorbates or surfactants, have each proved
successful for different systems [2123]. A working hypothesis is that these surfaces project chemically stable groups that provide the appropriate properties of
charge and hydrophobic/hydrophilic balance for interaction with protein molecules, without inducing denaturation. In many cases, the strength of adsorption can
be adjusted easily by varying the electrode preparation and electrolyte conditions.
Most of this chapter describes studies on ferredoxinssmall Fe-S proteins
that usually possess a signicant excess of negatively charged amino acids; therefore, their interaction with an electrode usually requires that the electrode surface
or solution interface contains positively charged groups. One way of achieving
this has been to add certain polycations to the electrolyte. Agents such as Mg2+,
Cr(NH3)63+, and complex organic amines such as neomycin or polymyxin, each
cause ferredoxins to interact increasingly strongly with oxide surfaces, probably
by binding simultaneously to protein and electrode via non-covalent cross-linkages (salt bridges) [24]. As reported recently, these cations are not necessarily
innocuous and they may alter the reduction potential; so this possibility should be
considered when comparing data derived from voltammetry and potentiometry
[25]. Another way to make the electrode interactive is to derivatize it with covalently attached functionalities that project positively charged groups toward the
solution. This approach has been described by Taniguchi and coworkers, who
have achieved electrochemistry of chloroplast [2Fe-2S] ferredoxin using an
indium oxide electrode modied by positively charged aminosilane groups [26].
In the same way, Vilker and coworkers have used a gold electrode modied with
mercaptoethylamine (HSCH2-NH3+) to obtain cyclic voltammetry of putidaredoxin, another small [2Fe-2S] protein [27].
B.

Voltammetry of Dissolved Proteins

As indicated in Figure 3A, the conventional electrochemical experiment involves


dissolved redox-active molecules, which are free to diffuse to and from a planar
electrode surface. The cyclic voltammogram that is expected for a reversible nelectron system consists of a pair of peak-type waves for reduction and oxidation,

150

Armstrong

separated (at 25 C) by 59/n mV, with equal current amplitudes that vary with the
square root of the potential scan rate [28,29]. For proteins, this situation can be
achieved with an electrode at which adsorption is weak; however, it has often
been found that the waves appear attenuated and rounded off, with an apparent
increase in peak separation. Bond and coworkers proposed [30,31] that this nonclassical electrochemical response is nonetheless still due to a reversible electrode reaction, in which the poorly dened waveform is a consequence of the proteins inability to transfer electrons at all but isolated sites on the electrode (Figure
3B). The essence of the microscopic model, as it is termed, is that since the electron-transfer activity depends so critically on how the protein interacts with the
electrode, a heterogeneous electrode surface may present relatively few productive interaction sites; consequently, the diffusion trajectory becomes radial (hemispherical) rather than linear. Radial diffusion gives rise to a sigmoidal (steady
state) waveform, provided the individual diffusion shells remain small enough so
that overlap does not occur. The effect is similar to that of a partially blocked electrode [32] (i.e., like a microelectrode array), except that the active zones on the
electrode are not usually of uniform size or separation; therefore, the resulting
range of diffusion geometries leads to a hybrid waveform that is difcult to analyze and may be interpreted incorrectly as reecting a slow electrode reaction.
Other techniques for studying protein molecules in solution are less inuenced by these microscopic effects. Square-wave voltammetry is widely used due
to its great sensitivity, and even a low density of productive sites on the electrode
may give rise to a sharp and analyzable response [28]. The electrode may also be
rotated to achieve forced convection and hydrodynamic control of solution redox
species, while amperometric (and coulometric) measurementswhere the current (or charge) is recorded following a potential stepenable the time and potential domains to be deconvoluted [28,29]. These options complement each other to
provide a detailed picture of the thermodynamics and kinetics of redox processes.
Finally, bulk electrolytic methods enable samples of a particular redox state to be
prepared quantitatively for spectroscopic examination, at precise electrode potentials that may lie outside the range of conventional chemical titrants.
C.

Voltammograms of Proteins Immobilized on an Electrode

Another way of conguring the sample involves inducing strong adsorption of protein molecules at the electrode surface, or immobilization within a layer of lipid or
other surfactant. As indicated in Figure 3C, the ideal conguration for studying a
Figure 3 Interactions of protein molecules with an electrode. Voltammetric responses
expected for (A) diffusion to an electrode densely packed with productive sites; (B) diffusion to an electrode having such a low population of productive sites that they behave as
micro-electrodes; (C) a homogeneous adsorbed layer of noninteracting protein molecules.

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

151

152

Armstrong

proteins redox properties is a stable mono-/submonolayer lm of molecules, each


oriented for facile electron transfer [20]. To be an effective tool, native structural
and reactivity characteristics must be retained; furthermore, the redox centers
should act independently, homogeneously and independently of each other,
remaining fully accessible to ions and small reagents in the contacting electrolyte.
The relatively large size of protein molecules means that some of these requirements are easily satised, i.e., active sites are well isolated from those in neighboring molecules and the resulting layer is usually porous to small ions and molecules.
Regarding retention of native structure and reactivity, there are many examples for
which it is certain that no signicant changes have occurred to the active site.
Adsorbed proteins can often be examined by surface-enhanced Raman spectroscopy (although this requires that the electrode is a metal (usually bare or SAMmodied Ag) and by FTIR reectance methods [22,33,34]. In favorable cases,
light absorption or reectance has been used: heme proteins show an intense electronic transition in the near-UV (the Soret band) which is sensitive to the spin-state
of the Fe, and this property has been exploited to study proteins such as cytochrome
c or myoglobin, either adsorbed directly on transparent or reective electrode surfaces or conned within a surfactant lm, in order to ascertain the structural
integrity of the active site [34,35]. With Fe-S proteins, such diagnostically useful
UV-visible bands are absent; however, in several interesting cases it has been
established that specic chemical reactivities known in the dissolved proteins,
such as metal replacement and proton-transfer, are retained in the adsorbed state.
There are several advantages in protein lm voltammetry, as well as a number of cautionary points to be aware of. In an ideal case, the following aspects are
realized.
1.

The Redox Status of the Entire Sample Can Be Fine-Tuned

The oxidation states of the entire population of centers under investigation are
directly and immediately controlled by the applied potential. This achieves a
much greater degree of control over redox-state dependent reactivities compared
to conventional experiments that address freely diffusing molecules. It is also
possible to apply potential pulses of precise value and duration to the sample: this
exposes the active sites to oxidative or reductive stress, either of which can induce
further reactions that may continue well after the pulse [36]. As will be described
in Section IIIE, cyclic voltammetry experiments utilizing short, high-potential
pulses have been used to study oxidative damage to [4Fe-4S] clusters.
2.

Waveform Denition

In cyclic voltammetry, an ideal (homogeneous) immobilized layer of molecules


undergoing a reversible electron transfer that is uncomplicated by coupled chemical reactions gives a pair of reduction and oxidation peaks that are compact at all
scan rates (the half-height width for a simple n-electron transfer is approximately

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

153

83/n mV at 0C) [37]. The number of electroactive centers is easily determined


by integrating the peak envelope. In real situations, the homogeneity of the sample of protein molecules is assessed by inspecting the peakshapes: typically there
may be some broadening (dispersion) due to inhomogeneity within the layer (i.e.,
nonuniform electron-transfer rates or reduction potentials) [38]. Also, the peaks
for oxidation and reduction separate as the scan rate is increased. From this separation, the kinetics of electron transfer can be analyzed [see Edition (6) below]
and it is usually assumed that the reduction potential is the average of oxidation
and reduction peak values.
3.

Efcient Screening for Reactivities

The protein-coated electrode can be transferred sequentially between different


electrolyte solutions, thus subjecting the redox centers to a range of chemical
environments and opportunities for reaction. Conditions that can be examined
even include extremes of pH, or other media such as cryosolventsin which the
protein would normally denature or undergo some other irreversible change during the time required for manipulations and measurements in conventional experiments [39]. By contrast, a typical cyclic voltammogram can be measured very
quickly, thus allowing a snapshot to be taken before the sample deteriorates; in
practice this delay need be little more than a few seconds after the electrode has
contacted the cell solution.
4.

Sample Economy

Theoretically, the amount of sample required is no more than that needed to form
up to a monolayer on the electrode surfacei.e., about 1012 to 1011 mole per
square centimeter. (A typical EPR spectrum requires about 103 to 104 times more
material!)
5.

Sensitivity and Stoichiometry

The minuscule sample size (the local concentration of which is extremely high)
enables the detection and measurement of coupled reactions that may occur with
very low (trace) levels of agents in the contacting electrolyte.
6.

Fast Reactions

Because the voltammetric waveform and current are not limited by sluggish diffusion, the kinetics of electron transfer between the electrode and the proteins
active site(s) come into better focus. As a rough guide, a standard electron-transfer rate constant of 500 s1 or higher enables coupled (electrochemical-chemical,
EC) reactions with half-lives below a millisecond to be addressed quite easily. As
will be described in the following sections, we have been able to resolve several

154

Armstrong

interesting examples of coupled redox reactions by analyzing how the waveforms


and peak positions in each direction vary with scan rate.

III. INVESTIGATIONS BY VOLTAMMETRY


A.

Observing Fe-S Clusters by Voltammetry: Accessing


Low-Potential Redox Couples and Generating Highly
Reducing Species

Iron-sulfur clusters commonly exhibit very negative reduction potentials, a property that poses problems both for their spectroscopic detection and for quantitative generation and handling of reduced samples. For [2Fe-2S] and most [4Fe-4S]
clusters, it is only the reduced (1+) level that is EPR detectable; thus, unless this
state can be accessed, a cluster (particularly if this is one of many in a complex
enzyme) may remain undetected. The standard way to reduce an Fe-S cluster is
to add sodium dithionite: however, for practical purposes, the reduction potential
of this titrant lies in the region of 450 mV [40], so it is ineffective at generating
reduced species if the reduction potential of the target is much more negative than
500 mV. Beyond this limit, the main chemical alternatives are deazoavin/light
or titanium citrate [41]. The potential range for direct electrochemical methods is
restricted by the onset of solvent breakdown (hydrogen or oxygen formation) but
at many electrode materials, such as graphite, breakdown is signicant only upon
application of large overpotentials. This extension of the potential window has
enabled the characterization of several highly reducing species, including [4Fe4S]+ clusters with reduction potentials as low as 720 mV, and the hyperreduced all-Fe(II) species [3Fe-4S]2, each of which will be discussed later.
The 7Fe ferredoxin from Azotobacter vinelandii (A.v. FdI), which is shown
in Figure 4 (see color plate), is perhaps the best characterized of all iron-sulfur proteins: it has been extensively studied by different spectroscopic techniques and its
x-ray structure has been solved for various redox and pH states, as well as for certain mutant forms. [4246]. Both the [4Fe-4S] and the [3Fe-4S] clusters have
interesting properties that have been revealed by voltammetric methods. A voltammogram of a solution of A.v. FdI obtained at a PGE electrode in the presence of
neomycin is shown in Figure 5A. Two diffusion-controlled redox processes (A and
B) are observed as well-dened peaks with currents that depend on the square root
of the scan rate. Without neomycin or analogous coadsorbates, no signicant
cyclic voltammetry response is observed; with intermediate levels, the response is
unstable and deteriorates rapidly to a poorly dened sigmoidal wave. This behavior can be explained on the basis of the microscopic model, in which there are few
sites on the electrode suited for protein interaction: these sites require the participation of polycations that enable the negatively charged surface of ferredoxins to

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

155

Figure 5 Voltammograms of solutions of 7Fe ferredoxins at a PGE electrode: (A) Azotobacter vinelandii, (B) Sulfolobus acidocaldarius. In each case, proteins are 80 M in 20
mM Mes, 0.1 M NaCl, 2mM neomycin, 0.1 mM EGTA, pH 6.4. Scan rate 5 mV s1, temperature 0C.

156

Armstrong

bind to oxidized functionalities on the graphite [24]. On the other hand, squarewave voltammetry produces well-dened signals even at quite low neomycin concentrations [48]
Signals A and B were assigned, respectively, as the [3Fe-4S]+/0 and [4Fe2+/+
4S]
couples, based on EPR spectra of samples prepared by controlled potential
bulk electrolysis [47]. What was striking at the time the rst experiments were carried out in the similar protein from Azotobacter chroococcum (1988) was the revelation that the [4Fe-4S]2+/1+ cluster has an unusually negative reduction potential,
645 mV at pH 8.3 [47]. The cluster is coordinated within the classic binding
motif, i.e., Cys-Xaa-Xaa-Cys-Xaa-Xaa-Cys- . . . . . . Cys-Pro that ligates other
[4Fe-4S]2+/+ clusters; however, it had previously been proposed, incorrectly, that
this cluster shuttles between 3+ and 2+ oxidation levels (see Figure 2). The paramagnetic 1+ cluster is too reducing to be accessed by chemical titration with
dithionite; yet it is quite unreactive and does not reduce water over the course of
several hours. Samples of [4Fe-4S]1+ for EPR spectroscopy could be prepared easily by bulk electrolysis at 850 mV, using a cell incorporating a high surface area
PGE electrode. It was thereby established that the reduced form has the typical spin
state S = 1/2 that is well established for most other 1+ clusters: as expected, there
is also a magnetic interaction with the [3Fe-4S]0 cluster (S = 2) [47].
Since that time, a number of other proteins have been found to contain Fe-S
clusters with very negative potentials. Azotobacter vinelandii produces several
other ferredoxins, one of which (FdIII) contains two [4Fe-4S] clusters; of these,
one has the normal binding motif given above, whereas the other has an extended
linker in the sequence Cys-Xaa-Xaa-Cys-Xaa79-Cys . . . Cys-Pro [49]. Voltammetric studies on FdIII have shown that unlike most 8Fe ferredoxins (for example,
the one from Clostridium pasteurianum mentioned later), the two clusters have
different reduction potentials, 486 and 644 mV at pH 7.0. The 8Fe ferredoxin
from the photosynthetic bacterium Allochromatium vinosum also has two [4Fe4S]2+/+ clusters with very different reduction potentials, 460 and 655 mV: here,
by measuring the voltammetry of variants for which residues close to either cluster
had been altered by site-directed mutagenesis, it was determined that the cluster
having the unusually low potential is actually the one that is coordinated by the
classical binding motif [50]. Recently an 8Fe ferredoxin from the denitrifying bacterium Thauera aromatica has been studied [51]. The two [4Fe-4S] clusters have
reduction potentials of 431 and 587 mV at pH 7.0, and there is an interesting
complication in that the cluster having the higher potential exhibits EPR spectra
that arise from the higher spin states 3/2, 5/2, and even 7/2. The EPR spectra
obtained during a potentiometric titration are difcult to analyze without the quantitative information on the low-potential center obtained from voltammetry.
Clusters having reduction potentials this negative may be more common than
once believed, especially when one considers that for [2Fe-2S] and [4Fe-4S] clusters, only the reduced (1+) form (generally S = 1/2) is revealed by EPR. Although

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

157

such centers have usually been proposed to serve a structural role, this may be a mistaken assumption; so they may yet turn out to be involved in electron transfer, most
obviously in systems where very reducing conditions are generatedat least for
transient periods. Determining the factors that inuence the reduction potentials of
clusterseven explaining why systems such as A.v. FdI have Eo values that are
some 200300 mV more negative than othersis not straightforward [52]. To help
solve this problem, voltammetric studies have been made on several mutant forms
of A.v. FdI that have crystallographically dened alterations in the amino acid
sequence. Surprisingly, it has been difcult to achieve large shifts in Eo even by
altering residues close to, or which coordinate, the cluster [53]
The solution cyclic voltammogram of another 7Fe ferredoxin, from Sulfolobus acidocaldarius (Figure 5B) provides another surprise; a third redox couple
with a higher peak current than the other two is clearly visible at even more negative
potentials [54,55]. As with the other ferredoxins, couples A and B were assigned,
respectively, as [3Fe-4S]+/0 and [4Fe-4S]2+/+ by spectroscopic examination of bulkreduced samples [54]. Bulk reduction across the third couple (C) showed that two
more electrons are taken up, and spectroscopic examination of the extensively
bleached reduced form (UV-visible, EPR and MCD) showed that the [3Fe-4S]
cluster had been altered [55]. Re-oxidation restored the normal spectral features of
[3Fe-4S]+ thus conrming that this reaction (hyper-reduction) is reversible. It is
now certain that the redox transition associated with couple C produces a form of
the [3Fe-4S] cluster in which, formally, all the Fe atoms are in the 2+ oxidation
state. The reaction is complex and involves uptake of at least two protons, and, as
discussed later, provides a good example of the improved resolution that can be
obtained by conning the protein sample as a lm on the electrode surface.
Figure 6 shows protein lm cyclic voltammograms of some 7Fe ferredoxins.
The lms were obtained by painting a solution of the protein onto a PGE electrode surface in the presence of polymyxin as co-adsorbate, then transferring the
electrode immediately to the cell [5557]. The voltammograms are stable over
quite long periods of time (hours) and reveal more chemistry than those measured
for the solution samples in which proteins react by transient, diffusional encounter.
In all cases, the two signals at most positive potential (A and Bthe prime
denotes that the protein is in the adsorbed state) are assigned to the transitions [3Fe4S]1+/0 and [4Fe-4S]2+/1+ respectively, based on correspondence with bulk solution
voltammetry and spectroscopic (EPR and MCD) examination of samples prepared
by controlled potential electrolysis. The electrochemical responses are often surprisingly close to theoretical expectations. For example, at slow scan rates, signal
A of A.v. FdI has half-height peak widths of approximately 90 mV for both reduction and oxidation directions (the theoretical value at 0C is 83 mV), thus showing
that the [3Fe-4S] clusters in the lm behave in a reasonably uniform manner. Signals A and B each integrate to the same charge passed (i.e., one electron), whereas
signal C integrates as a two-electron process and the narrow peak width suggests

158

Armstrong

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

159

that there is signicant cooperativity. Following the assignment made using solution voltammetry and spectroscopy, signal C is due to the couple [3Fe-4S]0/2, and
it is clear that the intermediate 1 oxidation level is unstable relative to 0 and
2 (see Fig. 2). The absolute integrations for each active site are consistent with
a coverage of approximately one protein monolayer. As will be discussed below,
these redox transitions, so clearly displayed in the lm voltammograms, can each
be probed further to reveal novel chemistry.
B.

Kinetics of Proton Transfer at the [3Fe-4S] Cluster


in Azotobacter vinelandi Ferredoxin I

Provided interfacial electron exchange is fast, it is possible to make detailed studies


of the kinetics and energetics of coupled reactions, and to detect and analyze how
electron transfer in a protein may be gated by other processes such as proton transfer. Whereas an electron can easily tunnel through 10 , the equivalent tunneling
distance of a proton at the same energy is less than 0.25 [58]. This means that
long-range proton transport through proteins, which is of great importance in the
conservation of energy, depends on the existence of a chain of closely spaced
donors and acceptors (these can be water molecules or amino-acid side chains)
whose separation and pK values control the rate. Although redox-driven proton
pumps are large membrane-bound enzymes, the use of voltammetry to study proton
transfer is illustrated well in recent studies on Azotobacter vinelandii ferredoxin I.
The one-electron-reduced [3Fe-4S]0 cluster of A.v. FdI (see Figure 4) exists
in two pH-interconvertable forms (pK 7.8), that are distinguished by circular
dichroism (CD) and magnetic circular dichroism (MCD) spectroscopies [4248].
Solution and lm voltammetry at slow scan rates each show that the [3Fe-4S]0 cluster has a pH-dependent reduction potential; the slope above pH 7.8 approaches
zero, while the slope below pH 7 increases to a value of about 55 mV (0C) [47,48].
From extensive investigations, including crystallography, site-directed mutagenesis and voltammetry over a wide pH range, it has been established that the [3Fe-4S]0
cluster binds a single proton, probably at one of the 2 suldo groups. Cluster protonation of this type is not unique to A.v. FdI: a similar result is found for the [3Fe-4S]
cluster of S.a. Fd, although the pK is much lower, at 5.8 [54]; for the 3Fe ferredoxin
from Pyrococcus furiosus, the pK is below 3 [59]. The system is described by a simple square scheme for coupled electrochemical/chemical (EC) reactions (Figure 7).

Figure 6 Cyclic voltammograms of 7Fe ferredoxins immobilized at a PGE electrode with


polymyxin as co-adsorbate, pH 7.6, temperature 0C. Conditions (not optimized to reveal
couples in clearest detail) are as follows. Top: Azotobacter vinelandii FdI; scan rate 100 mV
s1; Center: Desulfovibrio africanus FdIII; scan rate 10 mV s1; Bottom: Sulfolobus acidocalarius Fd; scan rate 100 mV s1. For A.v. FdI, signals B interfere signicantly with C at
this scan rate. (Reproduced with permission from J. Am. Chem. Soc. 1996, 118, 85938603.
Copyright 1996 American Chemical Society).

Figure 7 Square scheme for proton binding to the [3Fe-4S] cluster of A.v. FdI, showing parameters giving best t to data obtained for
studies of the D15N mutant, in which proton transfer is retarded. There is no evidence for participation of the species [3Fe-4S]1+-H+, thus
only two sides of the square are used. (Adapted with permission from J. Am. Chem. Soc. 1998, 120, 70857094. Copyright 1998 American Chemical Society.)

160
Armstrong

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

161

The [3Fe-4S] cluster in A.v. FdI is buried approximately 8 below the protein surface and the intervening space contains no water molecules (or spaces to
accommodate them). These structural properties make A.v. FdI a valuable model
system for studying the mechanism of discrete proton-transfer events inside a protein. Extensive studies have been made using protein lm voltammetryexploiting the fact that interfacial electron exchange is sufciently facile (k0 is typically
about 500 s1) so that scan rates exceeding 100 V s1 are able to reveal the coupled
kinetics [6063]. Figure 8 (top; see color plate) shows the structure in the vicinity
of the [3Fe-4S] cluster. A clue to how protons are passed between the cluster and
solvent was the knowledge that the carboxylate group from aspartate-15 (D15),
which is salt-bridged with the NH4+ from lysine (K)-84, lies some 5 from the
cluster, and that this group moves away slightly when the cluster is reduced at pH
values above 8 [42]. Aspartate (and glutamate) carboxylates are well known as
proton-transfer agents in proteins, as is clear from recent studies of the light-driven
proton pump bacteriorhodopsin [64].
To elucidate the role of aspartate-15 in the proton-coupled redox reaction
of [3Fe-4S]1+/0, several mutant forms were prepared: in some cases, aspartate-15
was replaced by other amino acids; in other cases, residues apart from aspartate15 that might be important in transferring the proton were changed [62]. For each
variant it was ascertained by CD that the [3Fe-4S]0 cluster is still protonated, and
lm voltammetry at slow scan rates was used to determine how the reduction
potential varies with pH. For example, with the mutant D15N in which aspartate15 is replaced by asparagine (the carboxylate is replaced by carbamide) the pK of
the [3Fe-4S]1+/0 couple is lowered to 6.9 and the limiting Eo value (at high pH
where no proton is transferred) increases from 430 mV to 409 mV [63]. This is
consistent with the removal of a negatively charged group (pK < 6) at a crystallographically dened short distance from the cluster. In the pH region sufciently
lower than respective pK values, the native and mutant forms have very similar
reduction potentials; this reects the fact that the redox reaction is electroneutral
(e/H+) and thus little inuenced by the presence of nearby charged residues.
The mechanism of proton transfer in A.v. FdI was determined by analyzing
the lm voltammetry of the [3Fe-4S]0 redox couple (Signal A) in the native protein and mutant forms [6063]. These studies showed that the different variants
fall into two categories; those that display slow proton transfer and those for
which proton transfer is fast. Figure 9 shows examples of the cyclic voltammograms, started from an oxidizing potential, that are observed in either category,
i.e., D15N (slow) and native (fast).
At high pH (pH > pK) electron transfer is not coupled to protonation; so,
except for variation in Eo, the voltammograms of the [3Fe-4S]1+/0 couple in different variants are quite similar in appearance and reversible even at 100 V s1
(although a little asymmetry, which is pH independent, is evident at these high scan
rates). By contrast, in the lower pH range (pH < pK) where protonation of [3Fe-

162

Armstrong

Figure 9 Voltammograms of Azotobacter vinelandii 7Fe ferredoxin D15N and native


forms, each commenced from the oxidized state, showing how the oxidation reaction is
gated in different time domains (*) by proton transfer. Proton transfer is retarded in the
D15N mutant, thus re-oxidation is not observed at 1 V s1 but is observed again at 100 V
s1 because the proton does not arrive at the cluster during the cycle. For the native protein,
re-oxidation is not observed at 100 V s1.

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

163

4S]0 occurs, the appearance of voltammograms depends critically on scan rate. At


slowest scan rates, the species involved in Fig. 7 equilibrate with the electrode
potential and the voltammograms appear reversible, with Eo increasing as the pH
is raised. More interestingly, as the scan rate is increased, the kinetics of the coupled proton transfer are revealed. First we note that certain voltammograms show
the reoxidation process to be gated by proton transfer (i.e., the oxidation peak vanishes, as indicated by asterisks *). The slow mutant D15N shows gating at 1 V s1,
whereas with the native protein, gating does not become apparent until much
higher rates (e.g., 100 V s1) are used. Second, the oxidation peak for D15N and
other slow variants reappears as the scan rate is increased above 10 V s1, and we
note that the voltammograms at 100 V s1 are identical regardless of pH.
The data were analyzed by plotting peak positions (oxidation and reduction) against scan rate, and modeling in terms of kinetic schemes incorporating a
number of parameters. We have called these trumpet plots because for simple,
uncoupled electron transfer, the data separate symmetrically and give a trumpetlike shape. In addition to the electron-transfer rate constant, the additional parameters include pH dependent proton-transfer rate constants, participation of a base
that relays the proton between solvent and cluster, and variations of the pK of both
cluster and base during the reaction. Representative plots for D15N and native
FdI are shown in Figure 10; in each case, the voltammograms were commenced
from the oxidative limit. At high pH the plots are indeed simple, but note how
their forms change dramatically as the pH is lowered. At the left (i.e., using low
scan rates), the data correspond to those that might be measured in a potentiometric experimenti.e., the system is able to equilibrate with the electrode potential. By contrast, at the right, the potential is cycled so fast that the reduction
potential that is measured (average of oxidation and reduction peaks) corresponds
to a snapshot of the system, recorded, in this case, on the millisecond timescale.
Rapid sweeps to very positive potentials produced no evidence for any new
signal corresponding to the protonated redox couple H+ -[3Fe-4S]1+/0. Therefore
the reaction pathway follows only two sides of the square shown in Fig. 7; i.e.,
oxidation of [3Fe-4S]0-H+ to [3Fe-4S]1+ involves deprotonation followed by electron loss, and not the other way round. The forms of the voltammograms, i.e., the
obvious gating at certain scan rates, and the asymmetry of the resulting trumpet
plots, show that the electron and proton transfers occur stepwise and not in a concerted manner. The standard rate constants for electron-transfer lie in the region
of 500 s1 for all mutants; thus, if the scan rate is fast enough, the [3Fe-4S] cluster can gain and release its electron before proton transfer can occur. The apparent reduction potential shifts to more negative values as the scan rate is increased;
this is because in the limiting situation, the process being observed is the uncoupled [3Fe-4S]1+/0 electron transfer, even under acidic conditions. The data for different pH values are expected to converge at high scan rates, and this is indeed
observed for the slow variants.

164

Armstrong

Figure 10 Representative Trumpet Plots for D15N and native A.v. FdI. Those for
D15N also show the ts based on koff = 2.5 s. Note the region (pH 5.50) in which the oxidative peak is not observed. Open symbols refer to data obtained at pH = 8.34.

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

165

These results have been corroborated by stopped-ow studies under selective conditions, using Ru(NH3)63+ or Fe(CN)63 as oxidants [61]. The rate of oxidation of the protonated [3Fe-4S]0 cluster in the D15N variant is always slow and
pH independent; whereas with native FdI, oxidation is fast and occurs at a rate
constant that decreases as the pH is lowered. In all cases the kinetics are independent of the nature of the oxidant, and replacement of H2O by D2O reveals only
a small isotope effect, showing that proton tunneling is not rate limiting.
The mechanism of proton transfer that emerges is shown in Fig. 11, which
extends the simple thermodynamic cycle shown in Fig. 7. As summarized in
Table 2, the results for different mutant forms show the importance of aspartate15 as a base (B); importantly, the aspartate can neither be replaced by asparagine
(D15N, which removes the acid-base properties) nor by glutamate (the D15E
mutant, which places an additional CH2 group into the side chain). Furthermore, the salt-bridge that the D15-carboxylate forms with the NH3+ of lysine-84
appears to play little role. A satisfactory t to the trumpet plots for native and fast
mutants requires that immediately after the electron has transferred to the cluster,
the pK of the carboxylate increases (pK1 >> pKox) to promote proton capture
from solvent. Molecular dynamics calculations reveal that the resulting neutral
COOH group is very mobile and penetrates the protein with high-frequency
excursions (timescale 1011 sec), approaching to within hydrogen-bonding distance of one of the S atoms [62]. The carboxylate thus serves as a proton courier,
enabling the buried cluster to receive a proton at rates that approach the diffusion
controlled limit. Electron transfer to the cluster drives proton transfer, whereas
electron transfer off the cluster is gated by proton transfer.

Figure 11 Mechanism of proton transfer between bulk water and the [3Fe-4S] cluster
in A.v. FdI. Fast proton transfer to [3Fe-4S]0 is pH dependent, and protonation constants of
Asp15 are sensitive to whether cluster is protonated or unprotonated. At low pH, Asp15 reprotonates (K2), thus inhibiting proton transfer off the cluster. For native A.v. FdI, pKOX =
5.4. Rate expressions are shown in legend to Table 2. (Reproduced by permission from
Nature (Ref. 62). Copyright 2000 Macmillan Magazines Ltd.)

8.4 0.1
8.1 0.1

0.464
0.476
0.443

T14C

K84Q

Native (D2O)

6.6 0.1
6.7 0.1

0.397
0.388

D15K-K84D

D15E

6.5 0.1

6.6 0.1

7.1 0.1

6.7 0.1

6.5 0.1

pKcluster
low pH

3.0 0.1
4.5 0.2

1.2 107
2.0 107

222

6.0 109

2.5 0.1

232

9.0 109

2.0 107

207

6.6 109

koff (s1)

207

4.8 109

kon(M1s1)

308

Koff(s1)*
(pH 7.0)

7.9 109

kon (M1s1)*
(pH 7.0)

7.2 0.1

7.4 0.1

8.0 0.1

7.1 0.1

7.2 0.1

pK1

5.9 0.1

5.9 0.1

6.7 0.1

6.1 0.1

5.9 0.1

pK2

970 100

1252 100

720 70

910 90

1294 100

khopon(s1)

240 25

250 25

310 25

230 25

332 25

hop

k off (s1)

All terms are as dened in Figures 7 and 11.


*For the fast reactions, interpreted in terms of Figure 11, kon = khopon/([H+] + K1) and koff = khopoff K2/([H+] + K2). The effective bimolecular rate constants kon and koff
take consideration of the fact that interaction with Asp15 causes the pK of the cluster to differ at high and low pH. In all cases k0, the standard rst-order electrochemical rate constant for electron exchange (at the reduction potential) is > 200 s1; thus, electron transfer is never rate-limiting. (Reprinted by permission from Nature
(Ref. 62). Copyright 2000 Macmillan Magazines Ltd.)

6.9 0.1

0.408

pKcluster

D15N

SLOW

7.7 0.1

0.453

E18Q

7.8 0.1

7.8 0.1

pKcluster
high pH

0.443

Ealk (V)

Native

FAST

Protein

Table 2 Thermodynamic and Kinetic Parameters for Redox-Coupled Proton Transfer at A.v. FdI and Mutant Forms, Extracted from
Protein Film Voltammetry Experiments

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

167

Interaction between the cluster and the aspartate carboxylate is an important part of the electron-proton coupling mechanism. Once the cluster is protonated, the pK of the D15 carboxylate decreases to a value approaching that
observed when the cluster is oxidized (pK1 >> pK2 ~ pK ox). Simultaneously, the
pK of the cluster is sensitive to the protonation state of the aspartate.
C.

Generation and Properties of the Hyperreduced All-Fe(II)


[3Fe-4S] Cluster

It was apparent in Figures 5B and 6 that proteins containing a [3Fe-4S] cluster


exhibit a two-electron transfer (signal C) at electrode potentials in the region of
700 mV (at pH 7). This is a widespread observation and it thus appears that [3Fe4S] clusters may be unique in having the ability to span the wide range from
all-Fe(III) to all-Fe(II) states shown in Fig. 2 [55]. This was evident early on from
studies on Da FdIII, for which other potentially redox-active groups (such as nonligating cysteines) are absent, and it was noted that signal C vanishes when the [3Fe4S] cluster is transformed into a [M3Fe-4S] cubane (see next section) [65]. This
section describes what we know so far about the hyperreduced state [3Fe-4S]2.
The reason why the [3Fe-4S] cluster displays this unusual property may be
because addition of two more electrons to the normal one-electron reduced state
[3Fe-4S]0 is balanced by the accompanying transfer of two or three protons.
These probably bind at the electron-rich 2 sulfur atoms, thus rendering the
process electrically neutral and minimizing disturbance of the surrounding protein structure. The role of proton transfer is evident from the fact that the C signal is less readily observed in the slow mutants of A.v. FdI described above
[63]. The overall redox reaction is certainly slow, as evidenced from the fact that
peak separations increase rapidly with scan rate; notably, apart from S.a. Fd and
the ferredoxin from Pyrococcus furiosus, the reaction has not been observed for
proteins in solution. This is very likely due to the multistep transformation being
dependent on an extended encounter time at the electrode, i.e., of long enough
duration to commit the overall reaction to product formation.
As discussed below, the novelty of a Fe-S cluster undergoing a cooperative
two-electron reaction suggests some direct involvement of the sulfur atoms, possibly with some rearrangement of the bonding. The [3Fe-4S]2 cluster is remarkably
inert, as demonstrated by the persistence of the reoxidation wave even after several
minutes in the hyperreduced state, and the lack of any signicant level of electrocatalytic H2 production [55]. High spin iron(II) is not expected to bind as strongly
as Fe(III) to donor ligands, and thus all-Fe(II) clusters should be relatively unstable as well as highly reducing. One example that has recently been studied is the
Fe-protein of nitrogenase, which can be reduced to the [4Fe-4S]0 level with Ti citrate as reductant [66]. The voltammetry of this protein would be very interesting,
because it might reveal how ATP is involved in redox coupling: so far, however,

168

Armstrong

direct electrochemical experiments with this protein (MW 63 kDa) have not been
successful. A further relevant aspect of these highly reduced states is that cluster
assembly in a protein may at some early stage involve transient species that have
captured Fe in its more labile 2+ state.
The hyperreduced [3Fe-4S] cluster offers another surprise. When a lm of
S.a Fd or A.v. FdI is cycled rapidly at negative potentials, a much faster redox couple appears at potentials somwhat positive of the original C signal [67]. This result
is shown in Figure 12. The pH dependence and integration of this new signal shows
it to be a two-electron two-proton reaction and it is chemically reversible, provided
the scan rate is sufciently fast; moreover, if the protein lm is cycled in D2O
instead of H2O, the new signal appears at 0.1 V s1 instead of 1 V s1.
Fast and cooperative two-electrontwo-proton transfers have not previously been observed for Fe-S clusters, and the result suggests that the hyperreduced 3Fe cluster might be capable of undertaking sulfur-based oxidation. This
idea is depicted in Fig. 13. Reductive cleavage of -S-S- bonds is normally
observed at higher potentials (for comparison, the reduction potential of glu-

Figure 12 Voltammetry of a lm of Sulfolobus acidocaldarius ferredoxin showing the


formation of a new redox couple at low potential. (Reproduced with permission from J. Am.
Chem. Soc. 1998, 120, 1199411999. Copyright 1998 American Chemical Society.)

Figure 13 Proposed formation of a rapid intermediate in the oxidation of the hyper-reduced [3Fe-4S]2 cluster. (Adapted with
permission from J. Am. Chem. Soc. 1998, 120, 1199411999. Copyright 1998 American Chemical Society.)

Voltammetric Investigations of Iron-Sulfur Clusters in Protein


169

170

Armstrong

tathione is 0.25V) but reduction of a constrained -S-S- bond is likely to be


entropically less favorable [68]. The appearance of the new couple only at higher
scan rates can be explained in terms of the putative -S-S- bridged species (the
rapid oxidation intermediate) rearranging to the normal Fe-oxidized [3Fe-4S]0
product unless it can be quickly re-reduced. This relaxation process can be
retarded by running the voltammetry in a cryosolvent, so that the new redox couple is retained at slow scan rates [39]. The slower relaxation observed in the presence of D2O vs H2O suggests further that there is extensive rearrangement of
hydrogen bonds around the cluster. Whether this chemistry could have any physiological function remains to be seen, but the results do indicate quite a remarkable capability for [3Fe-4S] clusters to undertake the fast coupled transfer of two
electrons and two protonsa property that has previously gone undetected.
D.

Reversible Metallation of [3Fe-4S] Clusters

Despite the abundance of [3Fe-4S] clusters, there has been considerable debate
concerning their biological signicance. One example is the citric acid cycle
enzyme aconitase (Table 1), which is catalytically inactive in the 3Fe form that is
isolated under aerobic conditions, but is activated by addition of Fe under reducing conditions [8]. The reason for this is that the [3Fe-4S] cluster readily takes up
a fourth iron to rebuild an active [4Fe-4S] cubane, in a reversible process that
depends on the level of available Fe in the surrounding medium and on the ambient electrochemical potential. The Fe that is incorporated is the site of binding of
the interconverting substrates citrate, aconitate, and iso-citrate, hence the inactivity of the [3Fe-4S] form. Similar Fe-uptake reactions occur in certain ferredoxins,
and metals (M) apart from Fe may be taken up to give heteronuclear clusters
[M3Fe-4S] [59,6980]. However, this behavior is by no means universal, and in
certain proteins, notably A.v. FdI, the [3Fe-4S] cluster is resistant to cubane formation; furthermore, whole-cell EPR spectra conrm that the [3Fe-4S] cluster is
present in vivo [81]. We will now outline some voltammetric studies of reversible
[3Fe-4S] / [M3Fe-4S] transformations and address the specicity for Fe compared to other metal ions. The reactions that we will discuss are shown in Fig. 14.
Questions have been raised about the possible wider occurrence of heterometal clusters such as the [Mo7Fe-8S] cofactor of nitrogenase [5,6]; yet,
whereas the [4Fe-4S] cluster is familiar to all biological chemists, no direct evidence exists to suggest that simple heterometal replacements might occur naturally. Two issues are (1) the spectroscopic characteristics that might be useful for
detecting these species and (2) the chemical factors that inuence and usually
ensure the delity of assembly of all-Fe clusters. Under biological control (i.e.,
in the organism), cluster assembly in a protein must be linked to metal ion transport, availability, and homeostasis [82]. From the chemical viewpoint, however,
it is important to determine what reactions are feasible, regardless of biological
inuences.

Figure 14 Equilibria between [3Fe-4S] clusters and [M3Fe-4S] adducts showing the relationships between different oxidation
levels.

Voltammetric Investigations of Iron-Sulfur Clusters in Protein


171

Figure 15 Cartoon showing coordination of the [3Fe-4S] cluster in P.f. Fd and D.a. FdIII and the reaction with metal ions such as Fe
to form a cubane adduct. The labile metal ion may be coordinated by the carboxylate group of the aspartate that replaces cysteine in the
normal motif.

172
Armstrong

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

173

Among the proteins that accommodate interconvertible [3Fe-4S]/[M3Fe4S] clusters, the ferredoxin from Pyrococcus furiosus (P.f. Fd) has received most
attention in terms of spectroscopic and structural characterization [74,75,77,78].
The [3Fe-4S] cluster occupies a binding domain resembling that of typical
[4Fe-4S] clusters except that one of the cysteines (the central one in the triad) is
replaced by aspartate (D), i.e., the binding motif is Cys-Xaa-Xaa-Asp-Xaa-XaaCys . . . . . . Cys-Pro [83]. A cartoon of how this domain coordinates a [3Fe-4S]
cluster and how it might also accommodate a [M3Fe-4S] cubane is shown in Figure 15. A similar binding motif is found in the small 7Fe ferredoxin (FdIII) from
Desulfovibrio africanus (D.a.) where, although the crystal structure is not known,
there is overwhelming evidence that the [3Fe-4S] cluster is bound in the unusual
motif [80,84]. For M = Fe, direct coordination of Fe by the aspartate carboxylate
(supported by NMR studies on P.f. Fd [85]) or H2O (OH) renders it labile relative
to the other Fe atoms. So far, the only crystallographically dened example of carboxylato coordination at an Fe-S cluster is the C24D mutant of A.v. FdI, in which
one of the cysteine ligands of the [4Fe-4S] cluster has been changed to aspartic acid
[86]. This coordinates one of the Fe atoms in a monodentate manner.
For these small proteins, the one-electron reduced [3Fe-4S]0 cluster takes
up Fe(II) from solution to give a [4Fe-4S]2+ cluster, in a reaction that is analogous
to the activation of aconitase. This reaction, its reversal, and analogous reactions
with other metal ions have been examined by voltammetry, most extensively with
D.a. FdIII, but also with P.f. Fd [59,7173,76,79,80]. The following account
refers to D.a. FdIII: signicantly, it was the extremely reactive nature of the [3Fe4S] cluster in this protein that attracted attention at the time because it had proved
very difcult to study by conventional methods [84]. Not only does the [3Fe-4S]
cluster react very rapidly with Fe, but also the protein itself is rather unstable, so
it continually releases Fe into the medium, thereby fuelling the transformation.
As shown in Figure 16, an almost quantitative [3Fe-4S]-to-[4Fe-4S] cluster
transformation could be achieved for D.a. FdIII using electrochemical methods
[71]. After recording a cyclic voltammogram, the solution of 7Fe D.a. FdIII (containing 2mM neomycin to induce interaction with the PGE electrode, and 0.1 M
EGTA to sequester trace levels of Fe2+) was electrolyzed at 610 mV to reduce
both clusters (this was monitored by coulometry) then aliquots of Fe(II) were
added while holding the potential at 610 mV. Each aliquot produced a stepwise
increase in current until a total of one equivalent of Fe(II) had been added, as
[3Fe-4S]0 reacts to form [4Fe-4S]2+, which is then reduced to [4Fe-4S]1+. The
coulometry showed that, overall, there is consumption of a further electron equivalent; the cyclic voltammogram of the product showed further that the signal due
to [3Fe-4S]+/0 had vanished while the signal at the position of the original [4Fe4S]2+/+ couple had grown in intensity. The EPR spectrum of the product, as well
as the MCD spectrum of a chemically produced form, conrmed that the product
was an 8Fe ferredoxin containing a new [4Fe-4S]+ cluster, which could be distin-

174

Armstrong

guished spectroscopically because it has the unusual spin state S = 3/2. Referring
to Figure 14, the reaction steps follow the course 1 3 3 3 5.
Uptake of Fe(II) and other metals into a [3Fe-4S] cluster is detected and
quantied very conveniently in lms of 7Fe D.a. FdIII formed on a PGE electrode. Figure 17 (top) shows the changes observed in the voltammogram during
repeated cycling of a lm of FdIII contacting a solution of Fe(II) [72]. The other
voltammograms were obtained for reactions with Zn and Cd. Both rates and
extents of conversion depend on the metal ion, and titrations were carried out to
determine binding afnities under strict potential control; the point being that
because the equilibrium under study is Reaction 3 of Fig. 14, it is necessary to
apply a potential that is appropriate to lock the clusters into the correct oxidation levels (i.e., [3Fe-4S]0 and [M3Fe-4S]2+). Such potential control is easily
achieved if the sample is immobilized on the electrode. If metal uptake and
release occurs on a timescale that is relatively long compared to the voltammetric scan rate, the equilibrium populations of [3Fe-4S] and [M3Fe-4S] clusters can
be measured from their respective peak amplitudes. The dissociation constants Kd
determined for a variety of metal ions reacting with the [3Fe-4S] cluster of D.a.
FdIII are shown in Table 3. Equilibrium with Tl(I) is established so rapidly that
Kd values are determined instead from the shift in reduction potential as a function of log[Tl(I)] [73].
The observation that the [3Fe-4S] core should have such a high afnity for
Pb has possible biological relevance, although no Fe-S clusters corrupted with
this heavy metal have ever been found. Another surprise was the discovery that
for the [M3Fe-4S]2+ clusters that are formed in D.a. FdIII, Fe(II) (Kd / M = 30)
is bound considerably less tightly than other metal ions, e.g., Zn(II) (1.6), Cd(II)
(0.8), Tl(I) (1.5) or Cu(I) (<1) [72,73,76,79]. The respective products were prepared using bulk techniques and characterized by EPR and MCD spectroscopy,
and it was possible to make comparisons with the analogous [M3Fe-4S] species
reported for P.f. Fd, and Desulfovibrio gigas FdII [59]. Voltammetric studies with
P.f. Fd showed, by contrast, that Zn(II) is bound less tightly than Fe(II) (although
the kinetics are faster), thus ruling out any generalization on the order of afnities
for metals in clusters. Small changes in the way that the labile metal ion is coor-

Figure 16 Bulk electrolytic reduction of Desulfovibrio africanus ferredoxin III (7Fe


form) at a PGE electrode and titration with Fe2+. (a) Bulk reduction of 0.10 mM 7Fe FdIII at
a potential of 610 mv vs SHE in stirred cell, pH 7.4, temp. 3C, solution also contains 1.5
mM neomycin and 0.10 mM EGTA. (b) Additions of 0.25 equivalent aliquots of Fe2+ to the
reduced solution while holding at 610 mV. The current increases (after a lag during which
the excess EGTA is complexed) due to the formation of a [Fe3Fe-4S] cluster via reaction 3
of Figure 14, and spontaneous further reduction (reaction 5). (c) Continued bulk reduction at
610 mV. (From Ref. 71. Reprinted with permission from the Biochemical Journal 264,
275284 [1989]. Copyright, The Biochemical Society.)

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

175

176

Armstrong

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

177

dinated are probably sufcient to modulate both the metal-ion selectivity and the
reduction potential of the [M3Fe-4S] adduct.
The [3Fe-4S] cluster thus behaves as a tridentate ligand that is rather similar to a crown thioether, but with the capability of varying its coordinating ability
depending on oxidation level [13,79]. Clearly, [3Fe-4S]0 is a much better ligand
than [3Fe-4S]1+, on account of the increased electron density that is available to
the bridging suldo ligands; it is also relevant, as described in Section IIIB, that
[3Fe-4S]0 is also a Brnsted base in certain proteins, such as A.v. FdI. Only in the
so-called high-potential iron-sulfur proteins (HiPIPs) can a divalent ion (Fe(II))
be considered to bind to [3Fe-4S]1+, in this case to give [4Fe-4S]3+. The situation
is different for monovalent thiophilic metals like Tl(I) and Cu(I), for which binding is detected not only to [3Fe-4S]0 but also to [3Fe-4S]+, although afnities are
much lower in the latter case (KdOx for Tl(I) is 0.34 M) [73]. Paramagnetic clusters formulated as [Cu3Fe-4S]2+ have been spectroscopically characterized in

Figure 17 Formation of [M3Fe-4S] cubane adducts (M = Fe, Zn, and Cd) in D.a. FdIII
adsorbed as a lm on a PGE electrode with neomycin as co-adsorbate. Left-hand side
shows successive cycles (oxidative direction only) at 470 mV s1: [Fe2+] = 300 M; [Zn2+]
= 10 M; [Cd2+] = 10 M. Right-hand side shows the corresponding semi-log plots of the
appearance of peaks D due to [M3Fe-4S]2+/+ and disappearance of A and C due to [3Fe4S]+. (Reproduced with permission from J. Am. Chem. Soc. 1991, 113, 66636670. Copyright 1991 American Chemical Society.)

Table 3 Reduction Potentials and Dissociation


Constants (Equilibrium 3 of Figure 14) for [M3Fe-4S]2+
Clusters formed in Desulfovibrio africanus FdIII lms
at a PGE Electrode
Cluster

Eo/mV vs SHE

[3Fe4S]+/0
[Fe3Fe4S]2+/+
[Co3Fe4S]2+/+
[Cu3Fe4S]2+/+
[Zn3Fe4S]2+/+
[Cd3Fe4S]2+/+
[Tl3Fe4S]2+/+

150
393
247
+148
492
569
+80

[Pb3Fe4S]2+/+

440

Kd/M
30
130
<1a
1.6
0.8
34000b
1.5c
<<1

Conditions: pH 7.0, 0.1 M NaCl, 0C. a Cu is bound as Cu(I). b Value


for cluster in 2+ level (Equilibrium 2 of Figure 14). c Value for cluster in 1+ level.

178

Armstrong

both D.a. FdIII and P.f. Fd [76,77]. As with D.a. FdIII, voltammetry shows that
uptake of Tl(I) is also fast and reversible for P.f. Fd (rates approach diffusion control in both cases); this means that Tl(I) coordinates with minimal reorganization
of the cluster and its surrounding residues and suggests that it may be an effective
probe for detecting solvent-exposed 3Fe clusters [59].
More subtle still, the thermodynamic relationships between Kd values and
reduction potentials for different [M3Fe-4S] clusters predict a correspondence
between relative stabilities and ambient electrochemical potential. For D.a. FdIII,
the fact that the reduction potential of [Zn3Fe-4S]2+/+ is more negative than
[Fe3Fe-4S]2+/+ means that the reduced + level is stabilized more by Fe than Zn;
thus under mildly reducing conditions, Fe will displace Zn if the [Zn3Fe-4S] cluster remains oxidized (2+) while the [Fe3Fe-4S] cluster is reduced (+). This afnity reversal accounts for conicting observations in the EPR characterization of
[Zn3Fe-4S]1+ and [Cd3Fe-4S]1+ clusters, in which samples always showed contamination from [Fe3Fe-4S]+, despite Fe being less tightly bound based on the
titrations carried out under potential control [72]. The persistence of [Fe3Fe-4S]
is a result of the continual release of Fe by unstable FdIII (see above) and the fact
that during reductive titrations a considerable amount of time is spent at potential
values between 400 and 500 mV, sufciently negative to reduce [Fe3Fe-4S]
but not the clusters containing Zn or Cd. Under these conditions, the balance is
restored in favor of Fe.
E.

Oxidative Degradation

Degradation (disassembly) of Fe-S clusters in proteins is closely linked with


oxidative processes. The reactions are relevant in an organisms response to
oxidative stress, where examples include Fe-S cluster-containing transcription
factors such as FNR (fumarate/nitrate regulation protein) and SoxR (superoxide
regulation) as well as the citric acid cycle enzyme, aconitase [8,10,11]. So far, the
best established examples involve [4Fe-4S] clusters and their disassembly to
[3Fe-4S] or [2Fe-2S] forms [7,8]. As stated in the previous section, many proteins are able to interconvert freely between [3Fe-4S] and [4Fe-4S] forms, and the
4Fe 3 3Fe transformation represents the simplest example of oxidative disassembly. To help understand the chemistry involved in cluster damage, lm
voltammetry has been used to study how the two [4Fe-4S] clusters in Clostridium
pasteurianum ferredoxin respond to oxidation under controlled conditions [36].
Although this protein was among the rst ferredoxins to be characterized, and
established to contain two [4Fe-4S] clusters, it has long been known that chemical oxidants such as Fe(CN)63 cause conversion to [3Fe-4S] clusters, and a 7Fe
form of C.p. Fd has been identied using NMR [87]. The lm voltammetry experiment allows the sample to be subjected to brief pulses of oxidizing potential, following which cyclic voltammograms are recorded to detect and measure the damage. Such a pulse sequence is shown in Figure 18, and the results of an experiment
with 8Fe C.p. Fd are shown in Figure 19.

Figure 18 Sequence of stages in a pulsed-protein-lm voltammetry experiment. Cyclic voltammetry is carried out before and
after applying an oxidative pulse of precise potential and duration. Further pulses can be incorporated. (Reproduced with permission from Biochemistry 2000, 39, 1058710598. Copyright 2000 American Chemical Society.)

Voltammetric Investigations of Iron-Sulfur Clusters in Protein


179

180

Armstrong

Figure 19 Top: Cyclic voltammograms of C.p. 8Fe ferredoxin adsorbed on a PGE electrode, scanned before and after applying a 3-second oxidative pulse at 720 mV vs SHE
(result shown is third cycle (100 mVs1) after the pulse. Voltammetry was carried out at
0C, in presence of polymyxin as co-adsorbate. Bottom: Graph showing how the ratio of
[3Fe-4S] to [4Fe-4S] clusters depends on the pulse potential. Pulse duration 3 seconds;
cluster populations measured on the third cycle.

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

181

Before the pulse, the 8Fe protein shows a simple voltammogram; this consists of a single signal that is actually due to both of the [4Fe-4S] clusters, which
have similar reduction potentials. Application of a single 3-second pulse at +0.72
V followed by a period at normal potentials produces a form in which a [3Fe-4S]
cluster is clearly present (note the characteristic A + C signal pattern) in an
amount approximately equal to that of the remaining [4Fe-4S] population. Further examination revealed that immediately after the pulse, there is actually an
excess of [3Fe-4S] over [4Fe-4S]; this means that some proteins must contain two
[3Fe-4S] clusters, with the second one being rapidly repaired to regenerate
[4Fe-4S]. By varying the pulse potential and measuring the ratio of 3Fe/4Fe clusters present in the product, a bell-shaped curve was obtained that was analyzed in
terms of two competing reactionsformation of a species that is either [3Fe-4S]
itself or a precursor, and destruction of [3Fe-4S] (or its precursor). Double-pulse
experiments (applying a second pulse after allowing the lm to relax to the 7Fe
form) showed that the [3Fe-4S] cluster is more stable than [4Fe-4S] with regard
to oxidative degradation, consistent with the high-potential side of the bell curve
being due to a reaction occurring prior to (and not after) formation of [3Fe-4S].
Accordingly, a reaction sequence has been proposed (Figure 20), which invokes
formation of [4Fe-4S]3+ (reduction potential E1), which either expels Fe to give
[3Fe-4S]+ or is oxidized further (E2) to give a species that is possibly [4Fe-4S]4+,
which is otherwise unknown; this highly oxidized cluster breaks down rapidly to
give apoprotein. The [3Fe-4S] cluster thus appears to be a relatively stable product of [4Fe-4S] disassembly in C.p. Fd, and is capable of converting easily back
to [4Fe-4S] once reducing conditions are restored and Fe is available. Viewed
more generally, this may be important because it would enable Biology to limit
the damage due to release of Fe in the presence of oxidants, i.e. from the wellknown and dangerous Fenton reaction.
F.

Ligand Binding at Fe-S Clusters

Knowing how Fe-S clusters add or substitute ligands is important for understanding how these centers act as catalysts or sensors. However, the great majority of clusters have robust all-cysteine ligation and changes in coordination are
difcult to achieve. Apart from aconitase, a few studies have also been made with
ferredoxins in which there is a subsite differentiated [Fe3Fe-4S] cluster. Binding
of CN to the [Fe3Fe-4S] cluster of Pyrococcus furiosus Fd was established by
spectroscopic methods, but no details of kinetics or thermodynamics were
revealed [88]. Following this, lm voltammetry was used to study the kinetics and
thermodynamics of ligand binding at the transformed [Fe3Fe-4S] cluster of Da
FdIII, in which (see Section IIID) the labile Fe is coordinated not by cysteine but
either by carboxylate (aspartate) or H2O (OH). The reversible reaction of this
cluster with ethanol 2-thiolate (the mercaptoethanol anion) provides a wellbehaved model for oxidation-level dependent ligand interchange [89].

Figure 20 Oxidative disassembly of [4Fe-4S] clusters in C.p. Fd following an oxidative pulse applied to a lm of protein
adsorbed in a PGE electrode.

182
Armstrong

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

183

In typical experiments; after rst forming the [Fe3Fe-4S] cluster by cycling a


lm of 7Fe Da FdIII in 0.1 mM Fe2+, the coated electrode is transferred to a cell containing the ligand RS. The minuscule amount of protein being studied on the electrode is many orders of magnitude smaller than the amount of RS available in solution (which depends both on the concentration of thiol and on pH (pK = 9.2)).
Ethanol-2-thiolate is commonly used as a mild reducing agent to prevent disulde
bond formation during protein isolation and handling; but in this case, it is a particularly suitable subject for studying ligand binding at [Fe3Fe-4S] clusters because it
forms only a weak complex with Fe2+. Thus, even with very high levels of thiol in
the contacting solution, the [Fe3Fe-4S] cluster remains intact and the small excess
of Fe2+ (present to suppress dissociation of Fe from the cluster) is not quenched.
As shown in Figure 21, the lm voltammetry of the 8Fe form of D.a. FdIII is
very sensitive to scan rate. Normally, signal D from the transformed cluster
[Fe3Fe-4S]2+/1+ overlaps with signal B, which is due to the indigenous [4Fe-4S]
cluster. However, in the presence of thiolate, and as measured at 0.2 V s1, signal D
disappears and is replaced by new signals The results were analyzed using the
square scheme model shown in Figure 22. In this model, electron transfer is fast for
all species, while ligand binding occurs at both oxidized and reduced states of the
[Fe3Fe-4S]2+ cluster with differing afnities Kdox, Kdred, and rate constants konox,
koffox, konred, and koffred. At slow scan rates, only two signals are observed, and the
one at low potential (Eoobs) varies with thiolate concentration. At sufciently fast
scan rate (and commencing the cycle from the oxidizing limit), the species [Fe3Fe4S]2+-L is trapped (reduction potential EFo). From the variation of Eoobs with thiolate concentration, and knowing that Eoobs = EFo at the hypothetical high-thiolate
limit (the necessary concentration would be in the region of 1 M, under which conditions the clusters are destroyed), it was possible to determine Kdox and Kdred. The
kinetics of ligand exchange were analyzed by iterative simulation of the coupled
electron transfer (Butler-Volmer) and ligation (rst- and second-order) reactions.
In this way it was established that the increased afnity of thiolate for the oxidized
cluster [Fe3Fe-4S]2+ results almost totally from a much faster rate of ligand binding, and redox transformations of [Fe3Fe-4S] are thus gated by the binding and
release of the ligand.
The lm voltammetry experiment readily enables us to visualize this gating, since the cycle predominantly turns clockwise, with electrochemical energy
being used to drive ligand transfer. Hypothetically, if one could energize the system instead by enforcing unfavorable ligand binding (for example, by mechanically inducing a conformational change or hydrolyzing ATP), then this energy
would be conserved as a negative potential shift, thus providing an increased driving force for reduction processes. In this example, the exo-thiolate ligated cluster
[Fe3Fe-4S]1+ -L is a much stronger reductant than [Fe3Fe-4S]1+, and preparing a
sample for EPR characterization was not feasible due to the very high concentration of thiol required. Film voltammetry thus detects and records a cluster species
(the + oxidation level) that cannot exist under normal conditions.

184

Armstrong

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

185

Although it is certain that the reaction occurs at the noncysteine Fe subsite,


it is still not clear whether RS adds to or replaces an existing ligand. The observation that the reduction potential is lowered so signicantly (by nearly 0.2 volt)
suggests that RS adds to and expands the coordination sphere. That only the oxidized cluster is able to do this demonstrates the extent to which catalytic activity
at clusters might be controlled by oxidation state.
Figure 21 Voltammograms showing reversible binding of an exogenous ligand (RS =
ethanol 2-thiolate) to the transformed [Fe3Fe-4S] cluster in Desulfovibrio africanus FdIII,
as observed by increasing the scan rate. The protein is adsorbed to approximately monolayer coverage on an edge-oriented pyrolytic graphite electrode, using neomycin as coadsorbate, and the electrolyte contains 0.2 M NaCl at pH 8.0, with 0.1 mM Fe(II). Temperature 0C.

Figure 22 Square scheme depicting the reversible binding of an exogenous ligand to a


[4Fe-4S] cluster; represented as [Fe3Fe-4S]. Data are as determined by simulation for binding of 1-hydroxyethane-2-thiolate at the transformed cluster of Desulfovibrio africanus
FdIII, at 0C. Note that RS may add to the coordination sphere rather than replace an existing ligand.

186

Armstrong

G. Observing Fe-S Clusters in Enzymes: Fumarate Reductase


and NiFe Hydrogenase
In most Fe-Scontaining enzymes, the cluster provides a relay site or way station
for electrons. Using protein lm voltammetry, it has now proved possible to
observe Fe-S clusters in enzymes as discrete signals and to examine their role in
catalytic electron transport. Examples that have been studied are fumarate reductase (Frd) from E. coli [90,91] and the NiFe hydrogenase from Allochromatium
vinosum [92], the structures of which are outlined in Figure 23 (see color plate)
[93,94]. In each case, the buried active sitesa covalently bound FAD in the case
of Frd or a bis-cysteinato bridged NiFe center in the case of hydrogenaseare
wired to the protein surface by a series of three clusters each separated by 1015
. An intriguing feature is that in both cases, the medial cluster has a reduction
potential that is approximately 0.2 V lower (Frd) or 0.3 V higher (H2ase) than the
distal and proximal clusters, thereby requiring a thermodynamically uphill step in
the relay. Consideration of how this may affect electron-transport efciency in
enzymes has been presented in the article by Dutton and coworkers [18].
Fumarate reductase is a membrane-bound terminal respiratory enzyme that
catalyzes the oxidation of menaquinol by fumarate. It is related structurally to the
mitochondrial enzyme succinate ubiquinone oxidoreductase (Complex II) and is
induced when E. coli grows under anaerobic conditions [95]. It has four subunits:
subunits A and B contain, respectively, the FAD and three Fe-S clusters (center 1,
center 2, and center 3) and constitute what is termed the membrane-extrinsic catalytic subdomain. The other two subunits, C and D, are smaller; they constitute
the membrane-intrinsic subdomain that anchors the AB domain to the membrane
and provides a binding site for the quinol and quinone redox partners. A soluble
enzyme, FrdAB, can be obtained either by chemical resolution of the intact membrane form or by overexpression. Importantly, FrdAB adsorbs readily at a PGE
electrode, where its redox centers and its catalytic activity can be studied in considerable detail [90,91]. Figure 24 (top) shows voltammograms obtained for
FrdAB in the absence of substrate. Above 500 mV, two sets of signals are
observed (only the oxidation direction is shown), which can be deconvoluted
quantitatively in terms of the exact coulometric contribution expected for each
center. (A signal that may be assigned to the [3Fe-4S]0/2 couple is also observed if
the potential is taken lower.) The complex envelope at higher potential is dominated by the cooperative two-electron reaction of the FAD (the reduction potential
of which is pH dependent) but includes those of the [2Fe-2S]2+/+ (Center 1) and
[3Fe-4S]+/0 (Center 3) clusters. The signal due to [4Fe-4S]2+/+ (Center 2) is visible
just below 300 mV. Center 2 has otherwise proved difcult to observe by spectroscopic methods: this is because the EPR signal in the reduced protein is
extremely broad due to magnetic interactions with the other Fe-S clusters [95].
When fumarate is added to the solution and the electrode is rotated to control mass transport, these signals convert to sigmoidal catalytic waves that

Voltammetric Investigations of Iron-Sulfur Clusters in Protein

187

Figure 24 Voltammograms of a lm of E. coli fumarate reductase (approximately monolayer) adsorbed on a PGE electrode in the presence of polymyxin at pH 9.0. Top. No substrate, baseline corrected. FAD signal is clearly observed above the Fe-S clusters. Voltammogram also detects the low-potential (medial) [4Fe-4S] cluster. Bottom. In presence of
fumarate (0.2mM) with electrode rotating at 3000 rpm. Note the boost that is observed at a
potential corresponding to the [4Fe-4S] cluster. (Adapted with permission from J. Am.
Chem. Soc. 1997, 119, 1162811638. Copyright 1997 American Chemical Society.)

188

Armstrong

directly relate turnover rate (current) to potential. By analyzing the currents as a


function of electrode rotation rate and fumarate concentration, and knowing the
concentration of electroactive enzyme molecules adsorbed on the electrode, the
kinetic constants kcat and KM were determined. The values obtained conrmed
the high activity of fumarate reductase in this adsorbed state, thus supporting the
notion that physiologically relevant properties can be addressed in this experiment [90]. At low fumarate concentrations, the catalytic current is controlled by
mass transport of substrate, and the waveform is a simple sigmoid. However,
once mass transport limitations are relaxed (high fumarate concentration and high
rotation rate) the voltammogram shows a clearly dened boost in current as the
electrode potential is taken below a value corresponding to the [4Fe-4S] cluster.
The magnitude of this current boost does not vary among different samples, and
so it is unlikely to arise from damaged protein moleules. Before publication of the
crystal structure, this second wave was proposed to arise from the [4Fe-4S] cluster providing a dual pathway for electron transfer between FAD and electrode
[91]. However, because it is now known that the clusters are arranged in a linear
manner, the simplest explanation for the current boost is that some enzyme molecules are oriented on the electrode surface in such a way that the [4Fe-4S] cluster (rather than the more exposed [3Fe-4S]) provides the all-important lead-in
group. The crystal structure indicates that this is quite feasible [18,93]. An interesting alternative interpretation is that electron transport down the wire occurs
by a super-exchange mechanism, utilizing the large polarizable S-atoms as a pathway, and the reduced [4Fe-4S] cluster might afford superior coupling [96]. These
ideas are currently under further investigation.
The NiFe hydrogenase contains a buried NiFe center (which is the active site
for H2 oxidation) and a wire of Fe-S clusters leading to the protein surface [94].
In order of their distance from the active site, these are [4Fe-4S] (proximal), [3Fe4S] (medial), and [4Fe-4S] (distal). However, the reduction potential of the medial
cluster, [3Fe-4S], is more than 0.2 volts higher than its neighbors in the chain. The
enzyme adsorbs at a rotating disc PGE electrode to give a lm that is extremely
active in oxidizing H2 [92]. Under 0.1 atm H2, sigmoidal voltammograms are
obtained (Figure 25) for which the current shows an almost linear increase with
(rotation rate)1/2, as predicted by the Levich equation for a reaction that is controlled by mass transport of substrate to the electrode. Based on there being up to a
monolayer coverage of electrocatalytically active enzyme, the turnover number
lies in the range 15009000 s, which is much higher than values reported for studies using small-molecule electron donors. The enzyme is inhibited by CO, and
experiments carried out under a stream of CO reveal, quite clearly, the presence of
two reversible (non-turnover) signalsone at 30 mV and one (integrating to double amplitude) at 300 mVwhich can be assigned respectively to the [3Fe-4S]
and two [4Fe-4S] clusters. These reduction potentials are in quite close agreement
with values determined by EPR potentiometry, and their positions with respect to

Figure 25 Voltammetry of Allochromatium vinosum NiFe hydrogenase adsorbed on a PGE electrode, under 0.1 atm H2. Catalysis
is diffusion controlled (note rotation-rate dependence) and rapid even at a potential where the [3Fe-4S] cluster will be reduced.
(Adapted with permission from Biochemistry 1999, 38, 89928999. Copyright 1999 American Chemical Society.)

Voltammetric Investigations of Iron-Sulfur Clusters in Protein


189

190

Armstrong

the catalytic H2-oxidation wave are indicated in the gure. It is immediately evident that H2 oxidation is diffusion controlled even in the potential range where the
[3Fe-4S] cluster must remain almost entirely in the reduced form. Evidently, this
uphill electron transfer does not seriously compromise activity.

IV.

CONCLUDING REMARKS

This chapter has sought to provide a broad spectrum of results, illustrating how
voltammetric methods can provide an electrochemical eye for visualizing and
analyzing complicated, convoluted and even chaotic behavior of Fe-S clusters
and, moreover, metalloproteins in general. It may be appreciated that even for the
most complex of electron-transport enzymes, the methods can provide a direct
readout of characteristic reaction properties. The area is still in its infancy, and I
expect that this chapter will soon be out of date.

ACKNOWLEDGMENTS
I wish to thank numerous colleagues for their enthusiastic input throughout the
past years, in the area of Fe-S proteins. In particular Julea Butt, Artur Sucheta,
Barbara Burgess, Brian Ackrell, Andrew Thomson, Simon George, Jacques Breton, Lisa Martin, Alan Bond, Jill Duff, Judy Hirst, Sarah Fawcett, Harsh Pershad,
Gareth Tilley, Raul Camba, Anne Jones, and Joel Weiner have each contributed
to the studies that I have summarized above. I am grateful also to Allen Hill, who
rst interested me in this eld. Research has been supported over the past ten
years by the UK EPSRC and BBSRC, The Wellcome Trust, The Royal Society,
Exxon Education Foundation, The Petroleum Research Fund, National Science
Foundation, NATO, and the Human Sciences Frontier Program.

REFERENCES
1.
2.
3.
4.
5.
6.
7.

H. Beinert, R. H. Holm, and E. Mnck, Science 277, 653659 (1997).


M. K. Johnson, Current Opinion in Chemical Biology 2, 173181 (1998).
Iron-Sulfur Proteins, Advances in Inorganic Chemistry, Vol. 47. (A. G. Sykes and R.
Cammack, Eds.) Academic Press, San Diego (1999).
H. Beinert, J. Biol. Inorg. Chem. 5, 215 (2000).
W. R. Hagen, in Bioinorganic Catalysis (J. Reedijk and E. Bouwman, Eds.), Marcel
Dekker, New York, p. 209 (1999).
D. C. Rees and J. B. Howard, Current Opinion in Chemical Biology 4, 559566 (2000).
D. H. Flint and R. M. Allen, Chem. Rev. 96, 23152334 (1996).

Voltammetric Investigations of Iron-Sulfur Clusters in Protein


8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.

191

H. Beinert, M. C. Kennedy, and C. D. Stout, Chem. Rev. 96, 23352373 (1996).


N. B. Ugulava, B. R. Gibney, and J. T. Jarret, Biochemistry 39, 52065214
(2000).
C. V. Popescu, D. M. Bates, H. Beinert, E. Mnck, and P. J. Kiley, Proc. Natl. Acad.
Sci. USA 95, 1343113435 (1998).
H. Ding and B. Demple, Biochemistry 37, 1728017286 (1998).
J. N. Agar, P. Yuvaniyama, R. F. Jack, V. L. Cash, A. D. Smith, D. R. Dean, and
M. K. Johnson, J. Biol. Inorg. Chem. 5, 167177 (2000).
R. H. Holm in Advances in Inorganic Chemistry, Vol. 38. (R. Cammack and A. G.
Sykes, Eds), Academic Press, San Diego, pp. 171 (1993).
B. Guigliarelli and P. Bertrand, in Advances in Inorganic Chemistry, Vol. 47. (A. G.
Sykes and R. Cammack, Eds.) Academic Press, San Diego, pp. 421497 (1999).
P. L. Hagedoorn, M. C. P. F. Driesson, M. Van den Bosch, I. Landa, and W. R.
Hagen, FEBS Lett. 440, 311314 (1998).
L. D. Gills de Plichy and E. T. Smith, Biochemistry 38, 78747880 (1999).
A. M. Bond and K. B. Oldham, J. Phys. Chem. 87, 24922502 (1983).
C. G. Page, C. C. Moser, X. Chen, and P. L. Dutton, Nature 402, 4752 (1999).
G. Dryhurst, K. M. Kadish, F. Scheller, and R. Renneberg, Biological Electrochemistry, Academic Press, New York and London, 1982.
F. A. Armstrong, H. A. Heering, and J. Hirst. Chem. Soc. Rev. 26, 169179 (1997).
F. A. Armstrong and G. S. Wilson, Electrochim. Acta 45, 26232645 (2000).
J. F. Rusling, Acc. Chem. Res. 31, 363369 (1998).
A. E. Kasmi, J. M. Wallace, E. F. Bowden, S. M. Binet, R. J. Linderman J. Am.
Chem. Soc. 120, 225226 (1998).
F. A. Armstrong, P. A. Cox, H. A. O. Hill, V. J. Lowe, and B. N. Oliver, J. Electroanal. Chem. 217, 331366 (1987).
Z. G. Xiao, M. J. Lavery, A. M. Bond, and A. G. Wedd, Electrochemical Communications 1, 309314 (1999).
I. Taniguchi, K. Watanabe, M. Tominaga, and F. M. Hawkridge, J. Electroanal.
Chem. 333, 331338 (1992).
L. S. Wong, V. L. Vilker, W. T. Yap, and V. Reipa, Langmuir 11, 48184822 (1995)
C. M. A. Brett and A. M. Oliveira Brett, Electrochemistry: Principles, Methods and
Applications, Oxford Science Publications, Oxford (1993).
A. J. Bard and L. R. Faulkner, Electrochemical Methods: Fundamentals and Applications, Second Eidion, Wiley, New York (2001).
F. A. Armstrong, A. M. Bond, H. A. O. Hill, I. S. M. Psalti, and C. G. Zoski, J. Phys.
Chem. 93, 64856493 (1989).
A. M. Bond and H. A. O. Hill, in Metal Ions in Biological Systems, Vol. 27 (H. Sigel
and A. Sigel, Eds), Marcel Dekker, New York, pp. 431494 (1991).
C. Amatore, J. M. Savant, and D. Tessier, J. Electroanal. Chem. 147, 3951 (1983).
D. Hobara, K. Niki, C. Zhou, G. Chumanov, and T. M. Cotton, Colloids and Surfaces
93, 241250 (1994).
A.-E. F. Nassar, Z. Zhang, N. Fu, J. F. Rusling, and T. F. Kumosinski, J. Phys. Chem.
B. 101, 22242231 (1997).
M. Collinson and E. F. Bowden, Anal. Chem. 64, 14701476 (1992).

192
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.

Armstrong
R. Camba and F. A. Armstrong, Biochemistry 39, 1058710598 (2000).
E. Laviron, in Electroanalytical Chemistry, A. J. Bard (ed), Marcel Dekker, New
York, Vol. 12, pp 53157 (1982).
R. A. Clark and E. F. Bowden, Langmuir 13, 559565 (1997).
J. P. McEvoy and F. A. Armstrong, Chem. Commun 1635 (1999)
S. G. Mayhew, Eur. J. Biochem. 85, 535547 (1978).
H. A. Heering, Y. B. M. Bulsink, W. R. Hagen, and T. E. Meyer, Eur. J. Biochem.
232, 811817 (1995).
C. D. Stout, J. Biol. Chem. 268, 2592025927, (1993).
C. D. Stout, E. A. Stura, and D. E. McRee, J. Mol. Biol. 278, 629639 (1998).
C. G. Schipke, D. B. Goodin, D. E. McRee, and C. D. Stout, Biochemistry 38,
82288239 (1999).
P. J. Stephens, G. M. Jensen, F. J. Devlin, T. V. Morgan, C. D. Stout, A. E. Martin,
and B. K. Burgess, Biochemistry 30, 32003209 (1991).
Z. G. Hu, D. Jollie, B. K. Burgess, P. J. Stephens, and E. Mnck, Biochemistry 33,
1447514485 (1994).
F. A. Armstrong, S. J. George, A. J. Thomson, and M. G. Yates, FEBS Lett. 234,
107110 (1988).
S. E. Iismaa, A. E. Vzquez, G. M. Jensen, P. J. Stephens, J. N. Butt, F. A. Armstrong, and B. K. Burgess, J. Biol. Chem. 266, 2156321571 (1991).
H. S. Gao-Sheridan, H. R. Pershad, F. A. Armstrong, and B. K. Burgess, J. Biol.
Chem. 273, 55145519 (1998).
P. Kyritsis, O. M. Hatzfeld, T. A. Link, and J.-M. Moulis, J. Biol. Chem. 273,
1540415411 (1998).
M. Boll, G. Fuchs, G. Tilley, F. A. Armstrong, and D. J. Lowe, Biochemistry 39,
49294928 (2000).
P. J. Stephens, D. R. Jollie, and A. Warshel, Chem. Rev. 96, 24912513 (1996).
K. Chen, G. J. Tilley, V. Sridhar, G. S. Prasad, C. D. Stout, F. A. Armstrong, and
B. K. Burgess, J. Biol. Chem. 274, 3647936487 (1999).
J. L. Breton, J. L. C. Duff, J. N. Butt, F. A. Armstrong, S. J. George, Y. Ptillot,
E. Forest, G. Schfer, and A. J. Thomson, Eur. J. Biochem. 233, 937946 (1995).
J. L. C. Duff, J. L. J Breton, J. N. Butt, F. A. Armstrong, and A. J. Thomson, J. Am.
Chem. Soc. 118, 85938603 (1996).
F. A. Armstrong, in Advances in Inorganic Chemistry, Vol. 38 (R. Cammack and
A. G. Sykes, Eds), Academic Press, San Diego, pp. 117163 (1993).
F. A. Armstrong, in Bioelectrochemistry of Biomacromolecules: Bioelectrochemistry: Principles and Practice (G. Lenaz and G. Milazzo, Eds), Birkhauser Verlag,
Basel, pp. 205255 (1997).
L. I. Krishtalik, Biochim. Biophys. Acta 1458, 627 (2000).
S. E. J. Fawcett, D. Davis, J. L. Breton, A. J. Thomson, and F. A. Armstrong,
Biochem. J. 335, 357368 (1998).
J. N. Butt, A. Sucheta, L. L. Martin B. Shen, B. K. Burgess, and F. A. Armstrong, J.
Am. Chem. Soc. 115 1258712588 (1993).
J. Hirst, J. L. C. Duff, G. N. L. Jameson, M. A. Kemper, B. K. Burgess, F. A. Armstrong, J. Am. Chem. Soc. 120, 70857094 (1998).

Voltammetric Investigations of Iron-Sulfur Clusters in Protein


62.
63.

64.
65.
66.
67.
68.

69.
70.
71.
72.
73.
74.
75.
76.
77.
78.

79.
80.
81.
82.
83.

193

K. Chen, J. Hirst, R. Camba, C. A. Bonagura, C. D. Stout, and F. A. Armstrong


Nature 405, 814817 (2000).
B. Shen, L. L. Martin, J. N. Butt, F. A. Armstrong, C. D. Stout, G. M. Jensen, P. J.
Stephens, G. N. La Mar, C. M. Gorst, and B. K. Burgess, J. Biol. Chem. 268
2592825939 (1993).
H. Luecke, Biochim. Biophys. Acta 1460, 133156 (2000).
F. A. Armstrong, J. N. Butt, S. J. George, E. C. Hatchikian, and A. J. Thomson, FEBS
Lett. 259, 1518 (1989).
S. J. Yoo, H. C. Angove, B. K. Burgess, M. P. Hendrich, and E. Mnck, J. Am. Chem.
Soc. 121, 25342545 (1999).
J. Hirst, G. N. L. Jameson, J. W. A. Allen, and F. A. Armstrong, J. Am. Chem. Soc.
120, 1199411999 (1998).
H. F. Gilbert, in Bioelectrochemistry of Biomacromolecules: Bioelectrochemistry:
Principles and Practice (G. Lenaz and G. Milazzo, Eds), Birkhauser Verlag, Basel,
pp. 256324 (1997).
I. Moura, J. J. G. Moura, E. Mnck, V. Papaefthymiou, and J. LeGall, J. Am. Chem.
Soc. 108, 349351 (1986).
K. K. Surerus, E. Mnck, I. Moura, J. J. G. Moura, and J. Le Gall, J. Am. Chem. Soc.
109, 38053807 (1987).
S. J. George, F. A. Armstrong, E. C. Hatchikian, and A. J. Thomson, Biochem. J.
264, 275284 (1989).
J. N. Butt, F. A. Armstrong, J. Breton, S. J. George, A. J. Thomson, and E. C.
Hatchikian, J. Am. Chem. Soc. 113, 66636670 (1991).
J. N. Butt, A. Sucheta, F. A. Armstrong, J. Breton, A. J. Thomson, and E. C.
Hatchikian, J. Am. Chem. Soc. 113 89498950 (1991).
K. K. P. Srivastava, K. K. Surerus, R. C. Conover, M. K. Johnson, J.-B. Park, M. W.
W. Adams, and E. Mnck, Inorg. Chem. 32, 927936 (1993).
W. Fu, J. Telser, B. M. Hoffman, E. T. Smith, M. W. W. Adams, M. G. Finnegan,
B. C. Conover, and M. K. Johnson, J. Am. Chem. Soc. 116, 57225729 (1993).
J. N. Butt, J. Niles, F. A. Armstrong, J. Breton, and A. J. Thomson, Nature Structural
Biology 1, 9427433 (1994).
M. G. Finnegan, R. C. Conover, J.-B. Park, Z. H. Zhou, M. W. W. Adams, and
M. K. Johnson, M. K, Inorg. Chem. 34, 53585369 (1995).
C. R. Staples, I. K. Dhawan, M. G. Finnegan, D. A. Dwinell, Z. H. Zhou, H. Huang,
M. F. J. M. Verhagen, M. W. W. Adams, and M. K. Johnson, Inorg. Chem. 36,
57405749 (1997).
J. N. Butt, S. E. J. Fawcett, J. Breton, A. J. Thomson, and F. A. Armstrong, J. Am.
Chem. Soc. 119, 97299737 (1997).
J. L. H. Busch, J. L. Breton, B. M. Bartlett, F. A. Armstrong, R. James, and A. J.
Thomson, Biochem. J. 323, 95102 (1997).
M. A. Kemper, H. S. Gao-Sheridan, B. H. Shen, J. L. C. Duff, G. J. Tilley, F. A.
Armstrong, and B. K. Burgess, Biochemistry 37, 1282912837 (1998).
F. A. Armstrong and R. J. P. Williams, FEBS Lett., 451, 9194 (1999).
R. C. Conover, A. T. Kowal, W. Fu, J.-B. Park, A. Aono, M. W. W. Adams, and
M. K. Johnson, J. Biol. Chem. 265, 85338541 (1990).

194
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.

96.

Armstrong
F. A. Armstrong, S. J. George, R. Cammack, E. C. Hatchikian, and A. J. Thomson,
Biochem. J. 264, 265273 (1989).
L. Calzolai, C. M. Gorst, Z. H. Zhou, Q. Teng, M. W. W. Adams, and G. N. La Mar,
Biochemistry 34, 1137311384 (1995).
Y.-S. Jung, C. A. Bonagura, G. J. Tilley, H. S. Gao-Sheridan, F. A. Armstrong,
C. D. Stout, and B. K. Burgess, J. Biol. Chem. 275, 3697436983 (2000).
I. Bertini, I, F. Briganti, L. Calzolai, L. Messori, and A. Scozzafava, FEBS Lett. 332,
268272 (1993).
R. C. Conover, J.-B. Park, M. W. W. Adams, and M. K. Johnson, J. Am. Chem. Soc.
113, 27992800 (1991).
J. N. Butt, A. Sucheta, F. A. Armstrong, J. Breton, A. J. Thomson, and E. C.
Hatchikian, J. Am. Chem. Soc. 115, 14131421 (1993).
A. Sucheta, J. H. Weiner, and F. A. Armstrong, Biochemistry, 32, 54555465 (1993).
H. A. Heering, J. H. Weiner, and F. A. Armstrong, J. Am. Chem. Soc. 120,
1162811638 (1997).
H. R. Pershad, J. L. C. Duff, H. A. Heering, E. C. Duin, S. P. J. Albracht, and F. A.
Armstrong, Biochemistry 38, 89928999 (1999).
T. M. Iverson, C. Luna-Chavez, G. Cecchini, and D. C. Rees, Science 284,
19611966 (1999).
A. Volbeda, M.-H. Charon, C. Piras, E. C. Hatchikian, M. Frey, and J. C. FontecillaCamps, Nature 373, 580587 (1996).
B. A. C. Ackrell, M. K. Johnson, R. P. Gunsalus, and G. Cecchini, in Chemistry and
Biochemistry of Flavoenzymes, F. Mller (ed), CRC Press, Boca Raton, Florida, Vol.
3, pp. 229297, (1992).
A. Nitzan, J. Jortner, J. Wilkie, A. L. Burin, and M. A. Ratner, J. Phys. Chem. B. 104,
56615665 (2000).

6
Polyion and Surfactant Films on
Electrodes for Protein
Electrochemistry
James F. Rusling and Zhe Zhang
University of Connecticut, Storrs, Connecticut

I.

INTRODUCTION

As described in Chapters 4, 5, and 7, the past 1015 years have witnessed remarkable progress in technologies that facilitate direct electron exchange between
electrodes and proteins [1]. Direct exchange of electrons between native redox
enzymes and electrodes enables fundamental voltammetric studies of these biocatalysts and permits electrodes to replace the enzymes natural redox partners.
This approach avoids the necessity for electron mediators, simplifying the design
of biosensors and bioreactors.
By the end of the 1980s, it was clear that direct electrochemical observation
of the redox chemistry of proteins required control of electrode surface structure
and minimization of surface contamination [13]. In this chapter, we discuss several strategies that combine these critical features with stable immobilization of
native proteins in lms on electrodes. In Section II, we discuss ordered surfactant
lms that provide biomembranelike environments for electrochemistry. In Section III, studies on polyion-protein lms prepared by casting and grown layer by
layer are summarized. Section IV presents studies on prototype bioreactors that
make use of electrode-driven enzymelike catalysis. Section V speculates about
the future.

195

196

Rusling and Zhang

II. CAST LIPID AND SURFACTANT FILMS


INCORPORATING PROTEINS
A.

Background

Many enzymes function in nature while bound to membranes [4]. Biological membranes are about half protein and half phospholipid. Proteins are bound on the outer
surfaces of the membrane, partly inserted into the bilayer, or extended across the
biomembranes (Figure 1). Yokota and coworkers reported direct, reversible
voltammetry for cytochrome c (cyt c) on a phosphatidylcholine LangmuirBlodgett membrane transferred to an ITO electrode in 1987 [5]. Tien and Salamon [6] developed a simple method for adsorbing a lipid bilayer onto a metal electrode. An insulated metal wire is cleaved in a droplet of lipid in organic solvent.
Lipid molecules adsorb onto the freshly cleaved surface. The adsorbed lm thins
to a lipid bilayer in water and is stable for 36 hr. Salamon and Tollin employed
this method with mixed phosphatidylcholines lms to reduce cyt c [7,8]. Dipping a basal plane pyrolytic graphite electrode into a detergent solution also provided coatings on electrodes that facilitated direct electron transfer [9]. Reversible voltammetry of negatively charged cyt b5 was obtained on an electrode coated
with dodecylamine, but positively charged proteins gave no response. Reversible

Figure 1 Conceptual model of a lipid bilayer membrane showing three modes of protein
binding.

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

197

cyt c voltammetry was achieved with a coating of lauric acid (dodecylcarboxylic


acid).
These papers gave indications that bilayers of lipids and synthetic surfactants could be used to achieve direct electron exchange between proteins and electrodes. Surfactantsor their biological counterpart, lipidscontain charged or
polar head groups and one or more long hydrocarbon tails (Figure 2). To form
bilayers as in Figure 1, these amphiphilic molecules need to have a ratio v/aolc
between 0.5 and 1, where v is the volume of the hydrocarbon tail region, ao is the
optimal area per head group, and lc is the critical chain length [10]. Bilayer-forming molecules tend to have two or three hydrocarbon tails. Adding a second tail

Figure 2 Structures of surfactants and lipids that form bilayers.

198

Rusling and Zhang

increases the value of v while not changing aolc very much compared to a molecule with equivalent single chain length.

B.

Myoglobin

Myoglobin is a muscle protein (molecular weight about 17,000) that contains an


electroactive iron heme prosthetic group. It is a good model protein for voltammetric studies in lms because its crystal structure is known to high precision, pK
values of all of its amino acid side chains have been measured, and its solution
electrochemistry is slow under ordinary solution conditions. Although it is an
oxygen storage and transport protein, it can be made to act like an enzyme [11].
Proteins incorporated into lipid bilayers stacked on top of each other can
lead to larger concentrations of redox sites per unit electrode area. Self-assembled
lms of ordered stacks of bilayers can be made by spreading water-insoluble surfactants dissolved in organic solvents or in aqueous vesicle dispersions onto solid
surfaces [12]. In water, these lms feature signicant amounts of water between
the bilayers. Similar lms are made by replacing counterions of the surfactants
with ionic polymers [1315]. Ordered lms can be made from double- and triplechain surfactants, which have a strong tendency to self-assemble into bilayers. In
1991, Kunitaki at al. incorporated the protein myoglobin (Mb) between bilayers
of dialkylphosphate surfactants by self-assembly in a specic orientation [16].
Mb can be taken up from pH 5.57 buffers into cast liquid crystal lms of
didodecyldimethylammonium bromide (DDAB) on electrodes [17]. We observed
chemically reversible voltammetry for the heme FeIII/FeII redox couple of Mb
taken up from solution into 0.520 m DDAB lms (Figure 3) [1719]. No voltammetric peaks were observed for bare pyrolytic graphite (PG) electrodes in Mb
solutions, so the surfactant lm turns on the protein electrochemistry. MbDDAB lms retain more than 80% of the original steady state voltammetric peak
current after a months storage in buffer. DDAB and Mb are both positive at pH
5.5, so hydrophobic interactions are probably important for lm stability.
Similar lms can be made by mixing a protein solution with a vesicle dispersion of the surfactant, casting this mixture onto an electrode, and drying it.
This technique is preferable because it is easier and faster, provides a known
amount of protein in the lm, and gave Mb-surfactant lms that are slightly better ordered [20]. Both techniques give comparable voltammetry. Film structure
can be represented as in Figure 4.
The CV peaks for 20 m Mb-DDAB lms (Figure 3) have a characteristic
diffusional tail consistent with the inuence of diffusion. Peak separations
were about 100 mV at 0.1 V s1, and peak current Ip was proportional to the square
root of scan rate () consistent [21,22] with diffusion-controlled lm voltammetry. Analysis of the linear Ip vs 1/2 plot gave a charge transport diffusion coef-

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

199

Figure 3 Cyclic voltammograms at 100 mV s1 and 25C: (a) pH 5.5 buffer containing
no protein on a bare PG electrode; (b) 25 M Mb puried by ultraltration in buffer on bare
PG; (c) 20 m Mb-DDAB lm on PG electrode in buffer, no Mb in solution. (Adapted from
Ref. 17 with permission. Copyright 1995 American Chemical Society.)

cient (Dct) of 4 107 cm2 s1 [17]. This can indicate physical diffusion of the protein in the lm, or so-called electron hopping by self-exchange reactions between
Mb redox sites [23]. When 20 m DDAB lms containing no protein were placed
into a Mb solution, the protein arrived at the electrode in less than 10 s. Using the
(Dct) of 4 107 cm2 s1 from voltammetry, the relation between molecular displacement , D, and diffusion time t [24], 2 = 2Dt, predicts that t = 5 s. This
result suggests that charge transport involves physical movement of Mb through
the DDAB lms. Similar experiments showed that at room temperature, Mb diffuses rapidly through lms of phosphatidylcholines but not through lms of gelstate dihexadecylphosphate (DHP, Figure 2) [25].
DDAB lms are in a lamellar liquid crystal phase at 25C. The uidity of
this phase facilitates movement of the protein during voltammetry. In thick lms,
normal pulse voltammetry (NPV) limiting currents, a direct measure of Dct1/2,
[11,17] gave small NPV limiting currents for Mb-DDAB lms at temperatures
below the gel-to-liquid crystal phase transition temperature (Tc), where they are

200

Rusling and Zhang

Figure 4 Conceptual model of structures of lms cast from insoluble surfactants and
myoglobin.

in the solid-like gel state. As temperature is increased, the limiting current Ilim
increases sharply in a sigmoid-shaped Ilim versus T curve consistent with a phase
transition from the solid-like gel state into a uid liquid crystal lm. The increase
in limiting current begins near the Tc value of about 12C for Mb-DDAB lms
measured by differential scanning calorimetry. Sigmoid-shaped Ilim versus T
curves were not observed for phosphatidylcholine lms or for very thin DDAB
lms at scan rates where diffusion does not inuence the voltammogram.
Films of Mb and phosphatidylcholines had symmetric peak shapes and relatively small peak separations (Figure 5) [25,26]. Peak symmetry, equal heights
of oxidation and reduction peaks, and linear increases in peak height with scan
rate were characteristic of thin-layer voltammetry. Unlike Mb-DDAB, lms of
phosphatidylcholines became thinner after soaking in buffer for about 1 hr, but
then remained stable. Scanning electron microscopy cross-sections suggested
thicknesses after thinning less than or equal to 0.5 m. Midpoint potentials of
the CVs (Figure 5) shifted negative with increasing solution pH, as with all Mb-

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

201

Figure 5 Thin-layer cyclic voltammograms of approximately 0.5m Mb-DLPC lm on


PG electrode at 2 V s1 and 25C in buffers of different pH + 0.1 M NaBr. (Adapted from
Ref. 26 with permission. Copyright 1997 American Chemical Society.)

surfactant lms examined. Thus, the properties of Mb in these lms are controlled
by the external solution.
Mb gave direct, reversible heme FeIII/FeII electron transfer in all lms of
double-chain (e.g., DDAB, DMPC and DHP) [11,1820,26] or triple-chain surfactants [27] on PG, Au, ITO, or Pt surfaces. These electrodes do not give significant voltammetric peaks in a nominally pure Mb solution. However, quasireversible peaks developed after several hours of storage and repetitive scanning of
PG electrodes in Mb solutions [28]. Also, quasireversible voltammetry of Mb can
be obtained with freshly puried solutions on specially cleaned hydrophilic
ITO electrodes [29] and on PG electrodes [19]. Without ultrahigh Mb purity, considerably slower electron transfer was found on ordinary ITO [30].
How is electron transfer between electrodes and redox proteins facilitated
by cast surfactant lms? X-ray photoelectron spectroscopy (XPS), surface infrared spectroscopy, and voltammetric studies showed that competitive adsorption on an electrode surface between surfactant, protein, and macromolecular
impurities in the protein solution is invariably won by the surfactant when present in large enough amounts [19]. Adsorption of proteins and macromolecular

202

Rusling and Zhang

impurities, which can block electron exchange between the protein of interest and
the electrode, is inhibited by adsorption of the surfactant lm on the electrode
[14]. This clears and maintains a path for electron delivery between the electrode
and the protein redox site.
Square wave voltammograms (SWV) of Mb-surfactant lms conrmed
chemically reversible heme FeIII/FeII electrochemistry (Figure 6). CV or SWV
peaks for Mb-surfactant lms were much broader than predicted by models for a
single electroactive species conned to an electrode surface. SWV data were successfully t using a model featuring a dispersion of formal potentials [25,26].
SWV was used because of inherently better signal-to-noise and resolution than
CV [31]. The model combines a dispersion of p Eo -values with a SWV model
for a single electroactive moiety on the electrode surface [32]. Parameters used in
the nonlinear regression analyses were average standard heterogeneous rate constant ks (s1), average transfer coefcient , the p Eo values, and the p j* values.

Figure 6 Forward and reverse background subtracted square wave voltammograms for
Mb-DDAB lm on PG electrodes at 75 mV pulse height, 4 mV step height, and different
frequencies (f ). Points are experimental data and lines are best ts by nonlinear regression
onto the Eo-dispersion model using 5 Ejo and 5j*. Best ts for average ks = 48 2 s1,
= 0.48 0.01, and Eoavg = 0.003 0.001 V vs. NHE. (Adapted from Ref. 26 with permission. Copyright 1995 American Chemical Society.)

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

203

The model neglects rate limiting ion entry or ejection, electron selfexchange, and molecular interactions within the lms. Thus, the average apparent rate constant kS is best 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 rate constant.
The SWV Eo -dispersion model has been used to analyze the electrochemical parameters of myoglobin in various surfactant and polyion lms. Figure 6
illustrates typical good ts of the model onto SWV forward and reverse currents
for a range of frequencies. Parameters from the analysis are shown in the gure
caption. In general, values of were 0.50 0.05. Standard errors were within
about 5 mV for Eo , and within about 15% for kS. The ks-values indicate more
efcient electron transfer in lms than in solution (Table 1), and depended only
weakly on surfactant type. The ks for Mb in cast lms of Eastman AQ38 ionomer
(see Section III.A for structure) falls in mid-range.
Surfactant type inuences Eo of the protein signicantly (Table 1). Values
in DDAB lms are the most positive and closest to reported solution Eo s for Mb
[1,11], although still more negative than solution values. The most negative Eos
for Mb were found in lms of zwitterionic phosphatidylcholines (DMPC and
DLPC). The negatively charged DHP and AQ38 ionomer gave Eos slightly more
positive than in phosphatidylcholine lms. Values in DDAB lms are about 100
mV positive of those in DMPC lms. Similar trends were observed for hemoglobin (Hb) [1]. In DDAB-clay lms, where the negative clay platelets neutralize the
positive charges of the DDAB head groups, Eo for Hb was 80 mV negative of that
in DDAB lms. This suggests that head group charge may be important in determining Eo.

Table 1 Electrochemical Parameters for Mb at 25C


Film or solution

pH

Mb-DDAB
Mb-DMPC
Mb-DLPC
Mb-DHP
Mb in solution
High-purity Mb solution
Mb-AQ

7.0
7.0
7.0
7.4
7.0
6.5
7

Eo, mV vs. NHE


1
101
102
89
50
60
99

ks, s1

Method/electrodea

31
59
60
90
(0.07)b
(3)b
52

SWV/PG
SWV/PG
SWV/PG
SWV/PG
CV/ITO
CV/hydrophilic ITO
SWV/PG

aSWVsquare wave voltammetry; CVcyclic voltammetry; PGordinary basal plane pyrolytic


graphite; EPGedge plane pyrolytic graphite; ITOindium tin oxide.
bFrom bare electrode value in cm/s, converted to hypothetical value for 1 m lm.
Source: Data from Refs. 11, 25, 26 and 72.

204

Rusling and Zhang

Figure 7 Inuence of electrode material on formal potential (Eo) of Mb-DDAB lms in


pH 5.5 buffer.

The material used for the electrode also had a strong inuence on Eo (Figure 7). For Mb in DDAB lms, Eo values ranged from 50 to +120 mV versus
NHE at pH 5.5 in the order ITO<Au<PG<Pt [33]. The apparent rate constant ks
depended weakly on electrode material. Clearly, proteins in surfactant lms do
not give the same Eo-values as in solution. The inuence of surfactant type and
electrode material suggests a possible electrical double-layer effect at the electrode-lm interface on the potential felt by the protein. Surfactant-protein interactions may also be important.
A second chemically reversible set of peaks at potentials negative (approximately 1.05 V vs. SCE at pH 5.5) of the FeIII/FeII couple [34] was found for Mb
in DDAB lms. Spectroelectrochemical and metal substitution experiments [35]
suggest that this peak pair represents the FeII/FeI redox couple.
The mechanism of Mb FeIII/FeII reduction in surfactant lms involves protonation coupled to electron transfer. Details were investigated in thin liquid crystal lms of DDAB and phosphatidylcholines. Eo of the heme FeIII/FeII couple at

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

205

Figure 8 Inuence of pH on formal potential for lms of Mb-DDAB from CV () and


SWV (+), Mb-DMPC from CV () and SWV (), and Mb-DLPC from CV (+) and SWV
(). CV values are averages for scan rates from 0.1 to 2 V s1, SWV values are averages
from tting data at 60 and 75 mV pulse heights and frequencies between 100 and 240 Hz.
Solid lines are representative of a coupled protonation with pK a1s of 4.9 for Mb-DDAB
and 4.6 for Mb-PC lms. (Adapted from Ref. 26 with permission. Copyright 1997 American Chemical Society.)

25C shifted with pH from about pH 5 to 11 with slopes of 5156 mV/pH unit
(Figure 8), close to the predicted 59 mV/pH for a coupled one-proton, oneelectron reaction. Eo was roughly independent of pH from pH. 3.6 to 4.5, with an
intersection point at pH. 4.70.3 corresponding to a pKa for a weak acid [26].
Reectance-absorbance FT-IR (RAIR) and visible spectroscopy showed
that conformation of Mb in the lms was controlled by the pH of the external
solution. Changes in amide I bands (17001600 cm1) are caused by polypeptide
C A O stretching and reect conformational changes in secondary structure [36].
Second derivative RAIR spectra provide resolution of the amide I band into overlapped bands for different carbonyl environments, allowing semiquantitative
identication of structural features such as -helix, -sheet, and disordered
regions.
The second derivative amide I spectra of Mb in DDAB or DMPC lms
[26,37] at pH 6 and 7 were similar to that of pure Mb (Figure 9). The main fea-

206

Rusling and Zhang

Figure 9 Second-derivative reectance-absorbance infrared spectra of thin Mb-DMPC


lms at source incidence angle 60 prepared using buffers from pH 4 to 7 and cast onto
vapor deposited aluminum slides. Native Mb spectrum is similar to that at pH 7. (Adapted
from Ref. 11 with permission. Copyright 1998 American Chemical Society.)

ture (negative peak) at 1659 cm1 represents -helices, which comprise 76% of
native Mb. The nearly constant band at 1742 cm1 is from phosphatidylcholine
ester carbonyl stretching. As pH decreased from 6 to 4, a new band appeared at
1630 cm1 and the -helix band near 1659 cm1 decreased. The 1630 cm1 band
represents disordered regions formed from partial unfolding of Mb helices.
Unfolding in this pH range was conrmed by visible absorbance spectra. This
process appears similar to partial unfolding of Mb in solution at pH lower than 5
to a stable, molten globule state with roughly half the original helix content [38].
Linked acid-base equilibrium models explained the pH dependence of the
iron heme Soret band visible absorbance, formal potentials, ks, and electroactive
surface concentrations of Mb in the lms [26,37]. The pKa of 4.7 in the Mb-lipid

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

207

lms is associated with protonation of histidine residues in hydrophobic regions


of the protein structure and may involve the proximal histidine bound to iron
and/or the distal histidine in the heme pocket. At pH lower than 4.6, a protonated
form of Mb similar to the partly unfolded molten globule exists in the lms and
is reduced directly. A species with secondary structure very similar to native metmyoglobin [MbFeIII-H2O] appears to predominate in lms between pH 5.5 and 8.
Spectroscopic observation of the protonated form at lower pH suggests that protonation of MbFeIII-H2O between pH 5.5 and 8 occurs prior to electron transfer.
A protonated form of Mb, which may be a kinetic conformer but not the molten
globule, accepts the electron. MbFe(III)-OH is formed in lms at pH greater than
9, and its one-electron reduction is also coupled to protonation.
C.

Supramolecular Organization of Mb-Surfactant Films

As illustrated above, spectroscopic methods such as FT-IR and UV-Vis absorbance can be used to help characterize protein-surfactant lms. In addition,
spectroelectrochemistry, UV-Vis linear dichroism, ESR, and low-angle x-ray diffraction have also been used. Spectroelectrochemistry of Mb-DDAB lms on
transparent ITO electrodes conrmed that the CV peaks near 50 mV versus NHE
on ITO correspond to the MbFeIII/FeII redox couple [34]. The spectra were consistent with near native structures of MbFeIII and MbFeII in the lms.
Low-angle x-ray diffraction suggested that the surfactant lms are arranged
in bilayers. A Bragg diffraction was observed for Mb-DMPC lms corresponding
to a d-spacing of 5.4 nm [11], close to the 5.6 nm reported for DMPC lms alone
[39]. This suggests a tilt of hydrocarbon tails of about 10 with respect to the lm
normal. Mb-DDAB lms had a d-spacing of 2.8 nm, consistent with the shorter
chain length and smaller head group of DDAB compared to DMPC. The d-value
is consistent with a tilt angle of roughly 40 for the DDAB tails, similar to 2934
found by RAIR of DDAB lms without protein [40]. Phase transition studies also
suggest that surfactants in the lms are arranged in bilayers. Films containing Mb
and other proteins had gel-to-liquid crystal phase transition temperatures (Tc)
observed by DSC within several degrees of Tc for bilayer vesicle dispersions of
the same surfactant [11,17].
Anisotropic ESR spectroscopy of MbFeIII lms provided information about
spin state and orientation [20]. In lms made from solutions of pH between 6 and
8, myoglobin had an ESR spectrum at 10 K characteristic of native, high-spin
heme, suggesting water as an axial ligand of MbFeIII. Thus, UV-Vis spectroscopy, ESR, and reectance FT-IR spectroscopy all showed that at medium
solution pH values, secondary structures of Mb in these lms are similar to native
conformations.
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 [20]. Mb heme orientation angles were close to 60 irrespective

208

Rusling and Zhang

of head group charge or surfactant type [1,20]. The absolute direction of the tilt
cannot be obtained from these experiments. However, Mb orientation seems not
to depend on coulombic interactions. The protein may be partly imbedded in the
hydrophobic regions of the bilayers and depend on hydrophobic interactions for
its orientation and retention. NMR [37] showed that anions bind strongly to Mb
within the lms, neutralizing some of the surface charge and possibly allowing
the protein to reside in hydrophobic regions more easily.
A simplied model of Mb-surfactant lms based on all the characterization
results is shown in Figure 4. Mb is imbedded within tilted, stacked surfactant
bilayers. The Mb heme plane is tilted 60 on average to the lm normal. The
degree of Mb penetration into the bilayers is unknown, but strong electrostatic
binding is unlikely.
Boussard and Tao [41,42] used atomic force microscopy (AFM) to study
cast and self-assembled, adsorbed DDAB layers on highly ordered PG (HOPG)
electrodes. Consistent with Figure 4, their results suggested that Mb inserts into the
DDAB bilayers. A potential-driven phase transition of adsorbed DDAB was
observed by AFM, which may cause physical movement of the DDAB microstructures close to the electrode and possibly assist in mass transport of Mb during CV
scans.

D.

Cyt P450 Enzymes

Metabolism of lipophilic drugs and pollutants in the mammalian liver is catalyzed


by membrane-bound cytochromes P450 (cyt P450) in reactions involving the iron
heme of these enzymes. Cyt P450s often activate pollutants oxidatively for reactions with DNA. This activation is thought to be a major pathway for toxicity and
carcinogenesis [43,44]. Bacterial cyt P450cam catalyzes similar reactions to the
mamalian liver forms of the enzyme. In anaerobic environments, cyt P450s catalyze reductive dechlorination of organohalides.
Catalysis by cyt P450s involves the initial reduction of cyt P450FeIII to cyt
P450FeII by a reductase. This redox conversion has been achieved reversibly by
CV for cyt P450cam in thin lms of DMPC and DDAB (Figure 10) [45]. Bare PG
electrodes in cyt P450cam solutions gave no CV peaks. A Soret absorption band at
447 nm for the FeII-CO complex of cyt P450cam in these lms showed that the
enzyme was in its native state. CV midpoint potentials shifted by about +60 mV
as a consequence of CO complexation of cyt P450FeII.
Only routine enzyme purication was required for these experiments,
whereas quasireversible CVs of solution cyt P450cam were found only with highly
pure enzyme at low temperature [46]. Cast surfactant lms containing cyt
P450cam gave CV peaks that decreased only 10% during a months storage in
buffer. As with Mb-surfactant lms, solution pH controlled the Eo-values of cyt

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

209

Figure 10 Cyclic voltammograms on basal plane PG electrodes at 0.1 V s1 in pH 7


buffer: (a) cyt P450cam-DMPC lm in oxygen-free buffer containing no enzyme; (b) bare
electrode in oxygen-free buffer containing 40 M cyt P450cam; (c) DMPC lm in oxygenfree buffer containing no enzyme. (Adapted with permission from Ref. 45. Copyright 1997
Royal Society of Chemistry, UK)

P450cam in these lms [45]. At pH greater than 6, Eo shifted by about 59 mV/pH


unit, suggesting proton-associated electron transfer.
SWV data for cyt P450cam lms was t well with the Eo-dispersion model
for thin lms. Values of ks obtained from the regression analyses were similar for
cyt P450cam in DDAB and DMPC lms (Table 2), but a bit smaller than values
for Mb (Table 1). As with Mb, DDAB lms had Eo-values about 100 mV more
positive than DMPC. In this case, however Eos were considerably positive of the
solution values [1,47].
E.

Surfactant Films Containing Other Metalloproteins

Films of surfactants and hemoglobin (Hb) gave quasireversible voltammograms


similar to those of Mb [48,49]. However, uptake of Hb into surfactant lms was
much slower than for the smaller Mb. Hb took 10 min to traverse a 20 m DDAB

210

Rusling and Zhang

Table 2 Electrochemical Parameters for Cyt P450cam at pH 7


Film or solution

Eo, mV vs NHE

ks, s1

Method/electrodea

121a
22a
303a

25
26

SWV/PG
SWV/PG
Titration

272b

CV/EPG

cyt P450cam-DMPC
cyt P450cam-DDAB
cyt P450cam-solution
High-purity cyt
P450cam solution
a25

C; bO C.
Source: Data from Refs. 4547.

lm, compared to less than 10 s for Mb. Thus, the larger Hb probably does not
physically diffuse within the lms during a typical CV scan.
The mechanism of Hb electron transfer in DDAB lms was studied in the
presence and absence of oxygen [50]. Reaction pathways depended on concentration of oxygen and pH, and involve equilibrium between the so-called tense
and relaxed conformations.
Cytochromes c, c3, and c553 gave chemically reversible voltammograms in
phosphatidylcholinecholesterollauric acid lms [5153]. Spinach ferredoxin
(an iron-sulfur protein) in lms of DDAB, phosphatidylcholine-cholesterol doped
with dodecylamine, or dioctadecyldimethylammonium bromide also gave reversible voltammetry [54,55]. Chlorella ferredoxin gave reversible voltammograms in lms of synthetic surfactants [56,57], but these lms were not as stable
as the heme-protein surfactant lms. Iron sulfur protein putidaredoxin gave quasireversible voltammetry in DDAB lms [58].
F.

Composite Polymer-Surfactant Films

Composite polymeric lms can be made by replacing the counterion of a charged


surfactant with a polyion, or by polymerizing the surfactant head groups. Both
methods give multibilayer lms with the polymerized regions in between hydrocarbon tails [5961]. Myoglobin was incorporated into cast and LangmuirBlodgett (LB) lms of several polymer-surfactant composites on PG electrodes.
Reversible voltammetry in the liquid crystal phase, and characteristic gel-toliquid crystal phase transitions were observed in all cases [27]. Similar to the simpler cast surfactant lms, the formal potentials depended on surfactant structure
and charge, but the heterogeneous electron transfer rate constant did not.
Hu and coworkers investigated myoglobin in composite lms of DDA+ and
polyacrylate (PA) [62]. Similar to lms of DDAB alone, two voltammetric peak
pairs were observed for Mb, one for the FeIII/FeII redox couple and the second

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

211

attributed to FeII/FeI. Electrochemical results and a Tc of 18C compared to 11C


for DDAB-Mb lms suggest slightly less uidity in DDA-PA-Mb lms compared to DDAB-Mb.
Clay platelets can serve as inorganic polyanions. Hb retained its native conformation in lms made by casting aqueous dispersions of the protein and a
DDAB-clay composite [63]. Cyclic voltammetry of Hb-DDAB-clay lms on
pyrolytic graphite (PG) electrodes showed a pair of well-dened and nearly
reversible peaks at a midpoint potential of about 0.06 V vs. NHE at pH 7. Only
the inner layers of the lm closest to the electrode were electrochemically active.
These lms had excellent stability over several months storage. Hb in these lms
is not located completely between the clay platelets.

III.

POLYION-PROTEIN FILMS

A.

Films Cast from Protein-Polyion Mixtures

Polyelectrolyte lms can serve as suitable hosts for proteins, which are themselves
polyelectrolytes. Naon (Figure 11), a peruosulphonate ionomer with less than
15% ionizable sulfonate groups per monomer, has partly hydrophobic and partly
anionic character, resulting in high afnity for hydrophobic cations [6466].
Hill et al. reported cast lms of proteins sandwiched between layers of
Naon on PG electrodes [67] that gave reversible voltammetry for cyt c551, cyt b5,
and azurin. Cyt c gave no voltammetry, which was surprising because the charge
on cyt c was +8, compared to 2 for cyt c551, 8 for cyt b5, and +2 for azurin. Thus,
hydrophobic interactions between Naon and the latter proteins are implicated.
Similarly, electrodes coated with Naon lms did not incorporate positively
charged cyt c3 [68]. Voltammetry was observed if protein was rst cast onto the
electrode and coated with Naon, but the lms were unstable. Negatively charged
spinach ferredoxin deposited onto a PG electrode with polycationic poly-L-lysine,
dried and then coated with Naon, gave reasonably stable, reversible cyclic
voltammograms [68]. Carboxyalkylthiol layers on Au electrodes adsorbed positively charged poly(lysine), which then adsorbed the negative cyt b5 to obtain
reversible voltammetry [69].
Poly-L-lysine lms cast onto electrodes incorporated negatively charged

Figure 11 Structure of Naon.

212

Rusling and Zhang

ferredoxins and hydrogenase [70]. Electrochemical reduction of protons to


hydrogen was catalyzed using poly-L-lysine lms with incorporated hydrogenase.
Poly(ester sulfonic acids) known as Eastman AQ ionomers (Figure 12) also
proved useful for cast lms. Cyt c and cyt c3 were incorporated into AQ lms on
electrodes [71] to give reversible voltammetry. Films cast from mixtures of Eastman AQ38 and Mb on PG electrodes formed stable hydrated gel lms in water
[72]. Soret absorbance band positions suggested that Mb has a near-native conformation in these lms. Reversible cyclic voltammograms were reproducible for
several months when Mb-AQ38 lms were stored dry or in buffer. Mb in AQ
lms catalytically reduced oxygen and trichloroacetic acid. SWV of Mb-AQ
lms gave good ts onto the Eo-dispersion model (Table 1). A major advantage
of these lms is their ease of construction, but disadvantages include a relatively
low fraction of electroactive protein.
Protein-AQ lms were also used to study reaction of c-type cytochromes
with soluble FeIII ammonium citrate [73]. Polyheme c-type cytochromes from
sulfate or sulfur reducing bacteria acted as FeIII reductases, but mitochondrial ctype cytochromes did not. AQ-Mb lms on graphite electrodes catalyzed dechlorination of hexachloroethane in ethanolic solutions [74].
B.

Films Constructed Layer-by-Layer

Construction of lms in a layer-by-layer fashion enables the design of lm architecture according to desired specications. One elegant modern approach
involves utilization of the selective binding abilities of antigens with antibodies.
Antibody-enzyme conjugates have been bound to antigen layers adsorbed on
electrodes [75,76]. Additional layers of enzyme can be added with related strategies. The advantage of this technique is that high enzyme activity and orientation
are attainable. A disadvantage is that the enzyme conjugates must be synthesized
and puried, and the necessary antibodies must be available.
A simpler and more general technique involves alternate adsorption of
monolayers of biomolecules and polyelectrolytes. This method was used by Lvov
and coworkers to make ultrathin lms of a variety of proteins and oppositely
charged polyions [7779]. Figure 13 illustrates the general procedure. In this

Figure 12 Structure of Eastman AQ38 and AQ29 ionomers.

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

213

Figure 13 Illustration of the method of layer-by-layer adsorption for constructing


polyion-protein lms.

214

Rusling and Zhang

example, a negatively charged solid is rst immersed into a 13 mg mL1 solution of positively charged polyions, and the charge on the solid surface is effectively reversed. The solid is rinsed in water, then immersed in a 13 mg mL1
solution of negatively charged proteins. A protein layer is adsorbed, and the surface develops a negative charge. Both adsorption steps can be repeated many
times to obtain alternating multilayer assemblies with reproducible amounts in
each layer. Film growth can be monitored after each adsorption cycle with techniques such as quartz crystal microbalance (QCM) weighing, surface plasmon
resonance, or voltammetry.
Figure 14 shows an idealized structure of layer-by-layer lms. In reality,
nearest neighbor protein and polyion layers are probably more intimately mixed,
as documented for lms of linear polycations and polyanions [80]. Most enzymes
in these lms retain high activity. Polyions used to make such lms are shown in
Figure 15. Metal oxide nanoparticles can also be used as the polyanion component.
Our research group rst demonstrated that layer-by-layer adsorption can
provide ultrathin myoglobin and cyt P450cam lms featuring direct electron
exchange with electrodes [81]. An undercoating of mercaptopropanesulfonate
(MPS) on gold electrodes was necessary to facilitate direct voltammetry of the

Figure 14 Idealized drawing of a polyion-protein lm made by layer-by-layer adsorption.

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

215

POLYCATIONS

poly(diallydimethylamine)
(PDDA)

POLYANIONS

Figure 15 Polycations and polyanions used for layered lm formation.

proteins. MPS is chemisorbed to Au via Au-S bonds, placing a layer of negative


sulfonate groups at the interface on which to adsorb the rst layer of polycation
or protein. Chemically reversible cyclic voltammograms were found for the
FeIII/FeII redox couple of the proteins in these lms on smooth gold, as illustrated
for cyt P450cam (Figure 16). Peak currents of Mb and cyt P450cam in these lms
were proportional to scan rate at <1 V s1, but peak widths were considerably
larger than 90 mV, suggesting non-ideal thin-layer electrochemical behavior.
Reversible voltammetry was also found for lms of polycations and putidaredoxin, a Fe2S2 ferredoxin and natural redox partner of cyt P450cam [58]. No
voltammetric peaks were found for any of these proteins in solution on bare gold
electrodes.

216

Rusling and Zhang

Figure 16 Cyclic voltammogram at scan rate of 0.5 V s1 of Au-MPS/PEI/PSS/Cyt


P450cam lm in pH 7 TRIS buffer. (Adapted from Ref. 81 with permission. Copyright
1998 American Chemical Society.)

The frequency decrease of a piezoelectric QCM resonator is directly proportional to the mass on the metal coating of the resonator, provided the viscoelasticity of the interface does not change [82]. In situ QCM measurements on
gold-coated resonators showed that saturation of each adsorbed layer is reached
in 1520 min [81].
QCM of dry lms provides estimates of the weight of each layer and monitors the repeatability of the multiple adsorption steps. Drying the lms minimizes
bias from interfacial viscoelasticity changes. Film mass per unit area M/A (g
cm2) is related to QCM frequency shift F (Hz) for 9 MHz quartz resonators by
[77]:
M/A = F(1.83 108)

(1)
cm2 on

on a resonator of A = 0.16 0.01


one side. Cross-sectional images from
high-resolution scanning electron microscopy of lms made from a variety of
proteins and polyions provided direct scaling between F and nominal lm thickness (d) at both faces of the resonator:
d(nm) (0.016 0.002) F(Hz)

(2)

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

217

Figure 17 QCM frequency changes for dried lms at each step of assembly of a
Mb/DNA lm on a smooth Au-MPS underlayer. DNA was double-stranded calf thymus
(ds-CT). Measurements after adsorption cycles 911 were omitted. (Adapted from Ref. 81
with permission. Copyright 1998 American Chemical Society.)

Figure 17 shows the QCM frequency change after formation of each layer
of a DNA/Mb lm on a smooth MPS-Au electrode. The lm was dried before
each QCM reading. Frequency decreased linearly throughout lm formation,
indicating repeatable adsorption of the layers. From Eq. 2, the thickness of the
lm after deposition of 18 layers is 95 nm.
The amount of electroactive protein in each layer can be estimated, subsequent to QCM, by CV on the same electrode. Dividing the electroactive mass
obtained from integrating slow scan CVs by the total mass from QCM [81] gives
the fraction of electroactive protein. On smooth Au, only the protein layer closest
to the electrode was fully electroactive, while second layers were 2040% electroactive [81]. Additional protein layers on top of the electroactive ones did not
communicate with the electrode.
An increased number of electroactive layers could provide a larger active
enzyme loading per unit area for biosensors and bioreactors. Neighbor-layer mixing and increased protein loading may be benecial for electron transport, as it
might lead to a shorter average distance between redox sites and facilitate electron hopping transport of charge through the lm [83]. These ideas were borne
out in recent work, in which a greater number of electroactive protein layers was
achieved than on smooth gold. Films were grown on mechanically roughened PG

218

Rusling and Zhang

electrodes, which may provide a disorder-inducing template for the lm. In addition, 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 [84].
Films constructed on rough PG electrodes by adsorbing PSS from 0.5 M
NaCl and Mb from dilute pH 5.5 buffer gave CVs in which up to seven Mb layers were electroactive (Figure 18) [84]. The lms were built on an initial layer of
PSS, which binds strongly to rough PG. This approach provided many more electroactive layers than on smooth gold. Films were stable for several months upon
dry storage or in buffer solutions. Adsorption of PSS from 0.1 M NaCl on rough
PG gave lms with seven electroactive layers but about 27% less electroactive
Mb. Films constructed on smoothly polished PG electrodes showed a saturation
in electroactivity of Mb after three or four layers. Thus, roughness of the underlying electrode seems to be a very important factor in obtaining optimal electroactivity.
Analysis of SWV data on 75 nm {PSS(0.5 M NaCl)/Mb}6 lms on rough
PG with the Eo-dispersion model gave an average ks of 53 s1, in the same range
as Mb-surfactant and cast Mb-polyion lms (cf. Table 1). The average Eo for Mb

Figure 18 Background-subtracted cyclic voltammograms in pH 5.5 buffer at 0.3 V s1


of rough PG electrode coated with {PSS(0.5 M NaCl)/Mb}n lms with n = 1, 4, 6, and 7.
(Adapted from Ref. 84 with permission. Copyright 2000 American Chemical Society.)

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

219

in {PSS(0.5 M NaCl)/Mb}6 lms was 0.085 V versus NHE, similar to values for
Mb in cast phospholipid and AQ ionomer lms. In owing or stirred solutions,
the mechanical stability of the layer-by-layer lms is much better than that of the
cast surfactant and polyion lms. Both surfactant and layered polyion lms have
storage stability of several months as long as the protein is stable.
Negatively charged metal oxide nanoparticles were also explored for layerby-layer construction of lms on rough PG. Mb was adsorbed from an acetate
buffer at pH 5.2, where it is positively charged. MnO2 and SiO2 nanoparticles of
nominal diameters 2050 nm were adsorbed from weakly basic solutions, where
they are negatively charged. Protein-nanoparticle lms were constructed on a bed
of PSS/PDDA adsorbed separately from 0.5 M NaCl solutions onto rough PG.
CV peak currents increased with layer number for up to 10 bilayers of protein and
nanoparticles (Figure 19) [85]. Formal potentials were similar to those of other
Mb lms. Masses measured by QCM during formation of the SiO2-Mb lms
showed that there was enough protein in the lm to completely coat each
nanoparticle. Complete protein coating of each particle may provide connected
pathways of closely packed Mb molecules, which could facilitate propagation of
charge within the lm by a electron hopping mechanism.
Double-stranded, calf thymus DNA was also used as a polyanion in lms
of Mb and cyt P450cam. CVs similar to those in Figures 17 and 18 were obtained
[86].
The amount of electroactive protein in layered lms depends on the materials used. Figure 20 compares Mb loading per unit area on rough PG electrodes
using nanoparticles, PSS and DNA adsorbed from 0.5 M NaCl. PSS and DNA
gave larger loadings of Mb per unit electrode area for up to six bilayers, with PSS
giving slightly larger amounts than DNA. Electroactivity saturated above seven
bilayers in lms made with DNA or PSS. Increases in electroactivity extended to
10 bilayers for the MnO2 and SiO2 nanoparticle lms, but smaller total Mb loadings were obtained. Electroactivity in these systems represents charge transfer
through 350 nm for SiO2/Mb lms and 30 nm for MnO2/Mb lms. The MnO2/Mb
lms are thinner for nominally similar numbers of bilayers because of competitive adsorption processes during Mb layer growth [85].

IV.

ELECTROCHEMICAL BIOREACTORS

This section deals with prototype bioreactors made with the lms discussed
above. Electrochemical bioreactors can be based on thin lms of active enzyme
catalysts that convert reactants to useful products. The electrode can serve the
function of natural redox partners of the enzyme. Cytochromes P450, for example, employ NADH as an electron donor, and one or more reductases to deliver
electrons to the active iron heme enzyme [43,44]. If electrodes can take over the

220

Rusling and Zhang

Figure 19 Cyclic voltammograms at 0.3 V s1 in pH 5.5 buffer of PSS /PDDA/MnO2


(Mb/MnO2)n lms with n = 2, 5 and 10 on rough PG electrodes. (Adapted from Ref. 85
with permission. Copyright 2000 American Chemical Society.)

electron delivery functions, relatively simple electrochemically driven bioreactors could be made. Similar principles on a smaller scale can be used as the basis
for biosensors.
Electrochemical enzyme catalysis can often be identied by cyclic voltammetry. Mb, Hb and cyt P450 enzymes in various lms catalyze the reductive
dehalogenation of organohalides [11,17,33,34,48,62,63,72,84]. Figure 21 shows
CVs of lms of Mb and dimyristoylphosphatidylcholine (DMPC). In the absence
of reactant trichloroacetic acid (TCA), the voltammogram features reversible
reduction-oxidation peaks for the FeIII/FeII couple of Mb. In a solution 2 mM in

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

221

Figure 20 Inuence of number of bilayers deposited on the amount of electroactive


myoglobin in lms constructed on rough PG with various polyanions. PSS and ds-CT DNA
were adsorbed from 0.5 M NaCl solutions

TCA, the FeIII reduction peak current increases and the FeII oxidation current
decreases. This catalytic reduction peak current can be used to detect TCA. As
TCA concentration increases, the reduction peak grows and the oxidation peak
eventually disappears. This is because MbFeII is used up by the reducing TCA.
Two MbFeIII molecules are generated for each TCA reduced. The electrochemical catalytic reduction of TCA with Mb is as follows:
MbFeIII + e p MbFeII (in lm)
2MbFeII + Cl3CCOOH + H+ 3 2MbFeIII + Cl2HCCOOH + Cl

(3)

Electrochemical reduction of myoglobin in DDAB lms produces two


strong reductant species. At pH 5.5, MbFeII is formed from MbFeIII at about 0.2
V, and MbFeI seems to be generated at about 1.1 V (Figure 22) [34,35]. As
described above, catalytic reduction of trichloroacetic acid occurs by reaction
with MbFeII. CV of a Mb-DDAB electrode in buffer containing 30 mM
trichloroethylene (TCE) (Figure 23) shows that the MbFeIII/MbFeII redox peaks

222

Rusling and Zhang

Figure 21 Cyclic voltammograms showing effect of catalytic reduction of TCA by Mb


in DMPC lms at 0.1 V/s and pH 5.5. Numbers on curves show TCA concentration in mM;
dashed curve to the far right is for reduction of TCA on a DMPC-coated electrode without
Mb. (Adapted with permission from Ref. 25. Copyright 1997 Elsevier.)

at 0.2 V are unchanged, but a catalytic peak is found at 1.1 V, the potential of
the MbFeII/MbFeI redox couple. Thus, MbFeII does not react with TCE on the CV
time scale, but MbFEI does. Direct reduction of TCE occurs at potentials about
2.1 V on a DDAB electrode, so Mb provides a catalytic potential decrease of 1.0
V. Here Mb acts like an enzyme even though this is not its natural function.
Electrolysis of a series of organohalides was done using 20 m Mb-DDAB
lms on rough PG electrodes of about 3 cm2 area [34]. TCA was reduced at 0.4
V versus SCE to dichloroacetic, monochloroacetic, and acetic acids. Ethylene
dibromide (EDB) was reduced at 0.5 V to ethylene. Trichloroethylene and tetra-

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

223

Figure 22 Cyclic voltammograms at 100 mV s1 in pH 5.5 buffer + 50 mM NaBr.


Dashed line for bare PG electrode with 0.5 mM Mb in solution; solid line for fully loaded
Mb-DDAB lm in buffer containing no Mb. (Adapted from Ref. 34 with permission. Copyright 1995 American Chemical Society.)

chloroethylene were reduced at 1.3 V to dichloroethylene. These reactions persisted for thousands of catalyst turnovers. Rate enhancements were found in the
lms compared to the chemical reactions in Mb solutions and were attributed
mainly to preconcentration of the organohalides in the lms.
The reduction of nitrite ion catalyzed by Mb-DDAB lms was reported by
Farmer and coworkers. Two catalytic peaks were found by CV in nitrite solutions
at about 0.9 and 1.1 V vs SCE [87]. Electrolysis at 1.1 V and pH 5.5 gave
NH3OH+, N2O, N2, and NO as products. These ndings show that Mb can act like
a nitrite reductase in DDAB lms.
A Mb-DDAB lm in a pH 7.4 solution saturated with NO showed a catalytic peak at about 0.75 V vs SCE, and the peaks for MbFeIII/MbFeII disappeared [88]. Electrolysis at potentials negative of the catalytic peak gave N2O and
small amounts of nitrite. Results indicated chemical reduction of MbFeIII by NO,
ultimately yielding MbFeII-NO. Reductions of N2O and N3 with Mb-DDAB
lms were also studied [89].
Epoxidation of styrene and its derivatives was achieved by using Mb and
cyt P450cam in surfactant and layered polyion lms. This is similar to natural

224

Rusling and Zhang

Figure 23 Steady-state scan after repetitive cycling of Mb-DDAB electrodes in pH 5.5


buffer containing 30 mM trichloroethylene. (Adapted from Ref. 34 with permission. Copyright 1995 American Chemical Society.)

cyt P450-catalyzed oxidations, and involves an unusual combined oxidationreduction pathway. The detailed pathway (Figure 24) was rst investigated for
Mb in aqueous solutions and microemulsions [90].
The electrode converts MbFeIII into MbFeII, which reacts with oxygen to
give MbFeII-O2. Electrochemical reduction of MbFeII-O2 produces H2O2, which
converts MbFeIII to the radical MbFeIV = O, the active oxidant. This oxyferryl
radical epoxidizes styrene by oxygen transfer to the double bond. In this process,
a catalytic electrochemical reduction drives a catalytic enzymelike oxidation in a
doubly catalytic process.
Catalytic activity of lms containing Mb or cyt P450 enzymes on electrodes
was compared for styrene epoxidations [81,91]. Cyclic voltammetry for these
systems illustrated for a PDDA/cyt P450cam lm on Au-MPS with oxygen in
solution shows a large reduction peak at the FeIII reduction potential (Figure 25).
The catalytic reduction was 200 mV positive of the direct reduction of oxygen.
Saturating the solution with styrene caused only a slight increase in the catalytic
wave, probably because of increased utilization of oxygen for the epoxidation.
Electrolyses were done at 0.6 V versus SCE at 4C to protect the proteins
from damage by H2O2. Layer-by-layer polyion-protein lms on 1 5 cm gold
foils and surfactant lms on 1 6 cm carbon cloth were compared. The cathode

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

225

Figure 24 Electrochemical Mb-catalyzed oxidation of styrene.

compartment contained oxygenated pH 7.4 buffer saturated with styrene or a


styrene derivative. Ultrathin protein-polyion lms with two protein layers on AuMPS electrodes gave the best catalytic activities, with cyt P450cam giving the
largest turnover rates. The performance of rough carbon electrodes with multiple
protein layers has not yet been evaluated.
Improved catalysis of styrene epoxidation by protein-polyion lms was
related to their better mechanical stability compared to the surfactant lms. Product stereochemistry for the epoxidation of cis--methylstyrene depended on oxygen availability. There appear to be two pathways for olen oxidation. The stereoselective pathway utilizes the high valent iron-oxygen intermediate (Figure 24),
as does the natural enzyme system. The non-stereoselective pathway may involve
a peroxyl radical near the protein surface that forms in the presence of oxygen
[91]. The competing reactions are summarized in Figure 26.
Electrolysis for 1 hr produced 720 mM hydrogen peroxide in solution for
lms containing Mb or cyt P450. The presence of the peroxide-destroying
enzyme catalase shut down the catalytic epoxidation of styrene, conrming the
role of hydrogen peroxide as the iron heme activating agent.

226

Rusling and Zhang

Figure 25 Cyclic voltammograms at 0.2 V s1 in pH 7.4 buffer in a sealed cell for


(PDDA/P450cam)2 lms on Au-MPS electrodes: (a) solid line, under Ar; (b) solid line,
after passing 50 mL oxygen through the solutions; (c) dashed line, solution saturated in
styrene after passing 50 mL oxygen; and (d) dashed line, bare Au electrode after passing
50 mL oxygen. (Adapted from Ref. 91 with permission. Copyright 1999 American Chemical Society.)

Mb-catalyzed peroxidation of styrene was studied by spectroelectrochemistry in cast lms of DNA or Eastman AQ on optically transparent indium tin
oxide (ITO) electrodes [92]. Characteristic spectral features of ferrylmyoglobins
were observed during electrolysis of the Mb-coated ITO electrodes in oxygenated
solutions. Results were consistent with conversion of styrene to styrene oxide via
the MbFeIV = O radical.

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry

227

Figure 26 Oxygen-dependent epoxidation pathways catalyzed by Mb.

Styrene epoxidation with Mb lms clearly proceeds by the doubly catalytic


pathway in Figure 24. The active oxidant in cyt P450cam oxidations is too unstable [43,44] to be observed by spectroelectrochemistry, but the data are consistent
with a similar pathway.

V.

SUMMARY AND FUTURE PROSPECTS

The lms discussed in this chapter provide a complementary set of alternatives


for fundamental studies of redox enzymes, and for bioreactor and biosensor
development. The layer-by-layer polyion- and nanoparticle-protein lms have the
best mechanical stability. However, cast polyion and surfactant-protein lms are
stable for a month when hydrodynamic ow is avoided. The cast polyion and surfactant lms are the easiest to use for fundamental spectroscopic studies, which
can provide information complementary to voltammetry. The tiny amounts of
protein (nmol range) in all these lms lead to very signicant cost and time
economies.
We may be entering an era of signicant progress in applications of enzyme
lms to biosensors, bioreactors, biomedical devices, and fundamental biochemical research. We might expect future biosensors that can detect physiological
abnormalities from analysis of blood or urine or even from noninvasive analyses

228

Rusling and Zhang

that make measurements through the skin or in the saliva. Multiplexed systems
may be developed to provide many biomedical tests simultaneously. 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 by making the metabolite via enzyme catalysis in a
lm on an electrode containing DNA, and estimating the resulting DNA damage
with the same electrode [93]. Also, sensors based on direct protein electron transfer could be included in feedback loops to control implanted biomedical devices.
The use of enzymes in thin lms to catalytically oxidize olens and reduce
organohalides with an electrochemical driving force has been described in this
chapter. It should be possible to design electrochemical bioreactors based on
ultrathin enzyme lms to efciently catalyze regio- and stereospecic chemical
syntheses at low temperatures, with cyt P450 enzymes, for example. On a laboratory scale, syntheses require only sub-nanomolar amounts of enzyme and can
be done at room temperature or below. An electrode provides oxidizing or reducing power, so that electricity rather than chemical oxidants and reductants are
consumed. If the electrode can take over the electron delivery functions, natural
electron donor or acceptor molecules and reductase and oxidase enzymes that
ordinarily provide oxidizing and reducing equivalents to the enzymes are not
needed. If only the active enzyme is required, a less complicated and less expensive system can result, and product isolation and purication are simplied.

ACKNOWLEDGMENTS
The authors research described herein was supported by grant No. ES03154 from
National Institute of Environmental Health Sciences (NIEHS), NIH. Contents are
solely the responsibility of the authors and do not necessarily represent ofcial
views of NIEHS, NIH. We are very grateful to colleagues and students named
in joint publications for their essential contributions. We especially thank Prof.
John Schenkman of the University of Connecticut Health Center, whose continued collaboration makes possible the work with cytochrome P450s.

REFERENCES
1.
2.

J. F. Rusling and Z. Zhang, in H. S. Nalwa (Ed.) Handbook of Surfaces and Interfaces of Materials, Academic Press, San Diego, in press.
F. A. Armstrong (1990), in Bioinorganic Chemistry, Structure and Bonding 72,
Springer-Verlag: Berlin, pp. 137221.

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry


3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.

24.
25.
26.
27.
28.
29.
30.
31.
32.
33.

229

F. A. Armstrong, in G. Lenz and G. Milazzo (Eds.) Bioelectrochemistry of Biomacromolecules, Birkhauser Verlag: Basel, Switzerland, 1997, pp. 205255.
A. Kotyk, K. Janacek, and J. Koryta, Biophysical Chemistry of Membrane Function,
Wiley: Chichester, U. K., 1988.
T. Yokota, K. Itoh, and A. Fujishima, J. Electroanal. Chem., 216, 289 (1987).
H. Ti Tien and Z. Salamon, Bioelectrochem. Bioenerg., 22, 211 (1989).
Z. Salamon and G. Tollin, Bioelectrochem. Bioenerg., 25, 447 (1991).
Z. Salamon and G. Tollin, Bioelectrochem. Bioenerg., 26, 321 (1991).
A. Guerrieri. T. R. I. Cataldi, and H. A. O. Hill, J. Electroanal. Chem., 297, 541
(1991).
J. Israelachvili, Intermolecular and Surface Forces, 2nd Ed. Academic Press, San
Diego, CA., 1992
J. F. Rusling, Acc. Chem. Res., 31, 363 (1998).
N. Nakashima, R. Ando and T. Kunitake, Chem. Lett. 1577 (1983).
T. Kunitake, M. Shimomura, T. Kajiyama, A. Harada, K. Okuyama, M. Takayanagi,
Thin Solid Films, 121, L89 (1984)
T. Kunitake, A. Tsuge, and N. Nakashima, Chem. Lett. 1783 (1984)
M. Shimomura and T. Kunitake, Polymer J., 16, 187 (1984).
I. Hamachi, S. Noda and T. Kunitake, J. Am. Chem. Soc., 113, 9625 (1991).
J. F. Rusling and A.-E. F. Nassar, J. Am. Chem. Soc., 115, 11891 (1993).
J. F. Rusling and A.-E. Nassar, Langmuir, 10, 2800 (1994).
A.-E. Nassar, W. S. Willis, and J. F. Rusling, Anal. Chem., 67, 2386 (1995).
A.-E. F. Nassar, Z. Zhang, J. F. Rusling, V. Chynwat, H. A. Frank, and K. Suga, J.
Phys. Chem., 99, 11013 (1995).
R. W. Murray, in Electroanalytical Chemistry; A. J, Bard, Ed., Marcel Dekker: New
York, 1984; Vol. 13, pp. 191368.
R. W. Murray in R. W. Murray, Ed., Molecular Design of Electrode Surfaces; Techniques of Chemistry Series; Wiley-Interscience: New York, 1992; Vol. 22, pp. 148.
M. Madja in R. W. Murray, Ed., Molecular Design of Electrode Surfaces; Techniques of Chemistry Series; Wiley-Interscience, New York, 1992; Vol. 22, pp. 159
206.
A. J. Bard and L. R. Faulkner, Electrochemical Methods, Wiley, New York, 1980.
Z. Zhang and J. F. Rusling, Biophysical Chem., 63, 133 (1997).
A.-E. F. Nassar, Z. Zhang, N. Hu, J. F. Rusling, and T. F. Kumosinski, J. Phys.
Chem., 101, 2224 (1997).
A.-E. F. Nassar, Y. Narikiyo, T. Sagara, N. Nakashima, and J. F. Rusling, J. Chem.
Soc., Faraday Trans., 91, 1775 (1995).
A. Onuoha and J. F. Rusling, unpublished results.
I. Taniguchi, K. Watanabe, M. Tominaga, and F. M. Hawkridge, J. Electroanal.
Chem., 333, 331 (1992).
B. C. King, F. M. Hawkridge, and B. M. Hoffman, J. Am. Chem. Soc., 114, 10603
(1992).
J. Osteryoung and J. J. ODea, in Electroanalytical Chemistry; Bard, A. J., Ed.; Marcel Dekker, New York, 1986; Vol. 14, p 209.
ODea, J. J.; Osteryoung, J. Anal. Chem. 65, 3090 (1993).
J. F. Rusling, Prog. Coll. Poly. Sci. 103, 170 (1997).

230
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.

Rusling and Zhang


A.-E. Nassar, J. M. Bobbitt, J. D. Stuart, and J. F. Rusling, J. Am. Chem. Soc., 117,
10986 (1995).
P. J. Farmer, R. Lin, and M. Bayachou, Comm. Inorg. Chem. 20, 101 (1998).
J. F. Rusling and T. F. Kumosinski, Nonlinear Computer Modeling of Chemical and
Biochemical Data, Academic Press, New York, 1996, pp. 117134.
A.-E. F. Nassar, J. F. Rusling, and T. F. Kumosinski, Biophys. Chem., 67, 107
(1997).
A.-S. Yang and B. J. Honig, Mol. Biol., 237, 602 (1994).
E. B. Sirota, G. S. Smith, C. R. Sanya, R. J. Plano, and N. A. Clark, Science, 242,
1406 (1988).
K. Suga and J. F. Rusling, Langmuir, 9, 3649 (1993).
S. Boussaad, N. J. Tao, and R. Arechabaleta, Chem. Phys. Lett., 280, 397 (1997).
S. Boussaad and N. J. Tao, J. Am. Chem. Soc., 121, 4510 (1999).
J. B. Schenkman and H. Greim (Eds.), Cytochrome P450, Springer-Verlag, Berlin,
1993.
P. R. Ortiz de Montellano (Ed.), Cytochrome P450, 2nd Ed., Plenum, New York,
1995.
Z. Zhang, A-E. F. Nassar, Z. Lu, J. B. Schenkman, and J. F. Rusling, J. Chem. Soc.
Faraday Trans., 93, 1769 (1997).
J. Kazlauskaite, A. C. G. Westlake, L.-L. Wong, and H. A. O. Hill, J. Chem. Soc.
Chem. Commun. 2189 (1996).
S. G. Sligar, D. L. Cinti, G. G. Gibson, and J. B. Schenkman, Biochem. Biophys.
Res. Commun. 90, 925 (1979).
J. Yang and N. Hu, Bioelectrochem. Bioenerg., 48, 117 (1999).
Z. Lu, Q. Huang, and J. F. Rusling, J. Electroanal. Chem., 423, 59 (1997).
M. Cireanu, S. Goldstein, and M. A. Mateescu, J. Electrochem. Soc. 145, 533 (1998).
P. Bianco and J. Haladjian, J. Electroanal. Chem., 367, 79 (1994).
P. Bianco and J. Haladjian, Electrochim. Acta, 39, 911 (1994).
J. Hanzlik, P. Bianco and J. Haladjian, J. Electroanal. Chem., 380, 287 (1995).
P. Bianco and J. Haladjian, Electroanalysis, 7, 442 (1995).
P. Bianco and J. Haladjian, Electrochim. Acta, 42, 587 (1997).
M. Tominaga, J. Yanagimoto, A.-E. F., Nassar, J. F. Rusling and N. Nakashima,
Chem. Lett. 523 (1996).
A.-E. F., Nassar, J. F. Rusling, M. Tominaga, J. Yanagimoto, and N. Nakashima, J.
Electroanal. Chem., 416, 183 (1996).
Z. Lu, Y. M. Lvov, I. Jansson, J. B. Schenkman, and J. F. Rusling, J. Coll. Interface
Sci., 224, 162 (2000).
T. Kunitake, M. Shimomura, T. Kajiyama, A. Harada, K. Okuyama, and M. Takayanagi, Thin Solid Films, 121, L89 (1984)
T. Kunitake, A. Tsuge, and N. Nakashima, Chem. Lett. 1783 (1984).
M. Shimomura and T. Kunitake, Polymer J., 16, 187 (1984).
Y. Hu, N. Hu, and Y. Zeng, Talanta 50, 1183 (2000).
X. Chen, N. Hu, Y. Zeng, J. F. Rusling, and J. Yang, Langmuir, 15, 7022 (1999).
C. R. Martin and H. Freiser, Anal. Chem., 53, 902 (1981).
M. N. Szentirmay and C. R. Martin, Anal. Chem., 56, 1898 (1984).

Polyion, Surfactant Films on Electrodes for Protein Electrochemistry


66.
67.
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.

231

C. A. C. Sequeira, in Electrochemical Science and Technology of Polymers 2; R. G.


Linford (Ed.), Elsevier Applied Science, London, 1990; pp 319374.
C. E. W. Hahn, H. A. O. Hill, M. D. Ritchie, and J. W. Sear, J. Chem. Soc. Chem.
Commun. 125 (1990).
P. Bianco and J. Haladjian, Electroanalysis, 6, 415 (1994).
J. D. H. Glenn and E. F. Bowden, Chem. Lett., 399 (1996).
P. Bianco, J. Haladjian, and S. Giannandrea-Derocles, Electroanalysis, 6, 67 (1994).
P. Bianco, A. Taye, and J. Haladjian, J. Electroanal. Chem., 337, 299 (1994).
N. Hu and J. F. Rusling, Langmuir, 13, 4119 (1997).
E. Lojou, P. Bianco, and M. Bruschi, Electrochim. Acta, 43, 2005 (1998).
M. Wright, M. J. Honeychurch, and H. A. O. Hill, Electrochem. Commun. 1, 609
(1999).
C. Bourdillon, C. Demaille, J. Moiroux, J. M. Saveant, Acc. Chem. Res., 29, 529
(1996).
C. Demaille, J. Moiroux, J.-M. Saveant, and C. Bourdillon, in Y. Lvov and H. Mhwald (Eds.), Protein Architecture: Interfacing Molecular Assemblies and Immobilization Biotechnology, Marcel Dekker, New York, 2000, pp. 311335.
Y. Lvov, K. Ariga, I. Ichinose, and T. Kunitake, J. Am. Chem. Soc., 117, 6117
(1995).
M. Onda, Y. Lvov, K. Ariga, and T. Kunitake, J. Ferment. Bioeng., 82, 502 (1996).
Y. Lvov, in Y. Lvov and H. Mhwald (Eds.), Protein Architecture: Interfacing Molecular Assemblies and Immobilization Biotechnology, Marcel Dekker, New York,
2000, pp. 125167.
G. Decher, Science, 277, 1231 (1997)
Y. M. Lvov, Z. Lu, J. B. Schenkman, X. Zu, and J. F. Rusling, J. Am. Chem. Soc.,
120, 4073 (1998).
D. A. Buttry and M. D. Ward, Chem. Rev., 92, 1355 (1992).
M. Majda, in Molecular Design of Electrode Surfaces, R. W. Murray, Ed.; John
Wiley & Sons, New York, 1992, pp. 159206.
H. Ma, N. Hu, and J. F. Rusling. Langmuir, 16, 4969 (2000).
Y. Lvov, B. Munge, O. Giraldo, I. Ichinose, S. L. Suib, and J. F. Rusling, Langmuir,
16, 8850 (2000).
J. F. Rusling, L. Zhou, B. Munge, J. Yang, C. Estavillo, and J. B. Schenkman, Faraday Soc. Disc., 116, 77 (2000).
R. Lin, M. Bayachou, J. Greaves, and P. J. Farmer, J. Am. Chem. Soc., 119, 12689
(1997).
M. Bayachou, R. Lin, W. Cho, and P. J. Farmer, J. Am. Chem. Soc., 120, 9888
(1998).
M. Bayachou, L. Elkbir, and P. J. Farmer, Inorg. Chem., 39, 289 (2000).
A. C. Onuoha, X. Zu, and J. F. Rusling, J. Am. Chem. Soc., 119, 3979 (1997).
Zu, Z. Lu, Z. Zhang, J. B. Schenkman, and J. F. Rusling, Langmuir, 15, 7372 (1999).
J. Kong, J. N. Mbindyo, X. Wu, J. X. Zhou, and J. F. Rusling, Biophys. Chem., 79,
219 (1999).
J. Mbindyo, L. Zhou, Z. Zhang, J. D. Stuart, and J. F. Rusling, Anal. Chem., 72, 2059
(2000).

7
Electrochemistry of Peroxidases
Tautgirdas Ruzgas, Annika Lindgren, and Lo Gorton
Lund University, Lund, Sweden

Hans-Jrgen Hecht, Joachim Reichelt, and Ursula Bilitewski


German Research Center for Biotechnology (GBF),
Braunschweig, Germany

I.

INTRODUCTION

Heme peroxidases catalyze the oxidation of organic substrates by H2O2 or organic


hydroperoxides [1,2]. The catalytic cycle consists of several reactions including
multiple electron/proton transfer steps, which can be summarized by the following three reactions:
k1

E(FeIII) + H2O2 3 EI(FeIV = O, P+) + H2O


EI(FeIV = O, P+) + AH
EII(FeIV = O) + AH

k3

k2

3 EII(FeIV = O) + A

3 E(FeIII) + H2O + A

(1)
(2)
(3)

The resting state of the peroxidase enzyme is presented by E(FeIII), i.e., the
enzyme with Fe-heme redox center in the ferric oxidation state. Upon exposure
of peroxidase to a solution containing peroxide, the catalytic cycle starts (reaction
1) with a rapid oxygen transfer from the peroxide to the resting state of the
enzyme to yield compound I, EI(FeIV = O, P+). Formally, the process is a twoelectron oxidation of the resting enzyme. One oxidation equivalent is conserved

233

234

Ruzgas et al.

in the form of the oxy-ferryl state (FeIV = O) and the other is stored as an organic
radical, P+. The last is located at an oxidizable amino acid in case of cytochrome
c peroxidase or on the porphyrin as a porphyrin cation radical in case of plant
peroxidases [2]. Reduction of EI to E proceeds by two one-electron transfer steps
(reactions 2 and 3). The rst electron reduces the organic P+ radical, resulting in
an intermediate peroxidase compound II, EII(FeIV = O), and the second electron
converts EII to E, thus completing the peroxidase cycle. One-electron donors,
AH, e.g., phenolic compounds, are released as phenoxy radicals, A. Reactions (2)
and (3) also involve proton transfer, meaning that the way these reactions are presented above reects the usual simplication used to describe the peroxidase catalytic cycle. The reduction of EI to E can also proceed in a single two-electron
step, e.g., in the case of iodide acting as a two-electron donor [2].
Peroxidases and, especially, peroxidase from horseradish, HRP, are frequently exploited in different enzymatic analysis protocols for detection of H2O2
[3,4]. Due to the availability of a broad range of H2O2-producing oxidase
enzymes, peroxide detection is one of the most central subjects in the development of enzymatic methods for detection of compounds such as glucose, cholesterol, lactate, ethanol in food, biological (blood) and biotechnological (fermentation broth) uids [5]. Electrochemical detection of peroxide becomes relevant
when the analysis is composed to a small, cheap, and simple biosensor format.
The most popular is still detection of H2O2 through its direct oxidation at a platinum electrode [6]; however, other electrode formats, e.g., peroxidase-modied
or Prussian bluemodied electrodes, have become popular [7,8].
The basis for construction of peroxidase-modied electrodes for detection of
H2O2 relies on two different mechanisms of electronic communication between
the electrode and the active site of the peroxidase. The simpler of these is based on
direct electron transfer, DET, between the heme in the peroxidase and the electrode. DET-communication was rst discovered for HRP adsorbed on carbon
black electrodes [9] and was later conrmed for peroxidases from different origins,
including other plant peroxidases, as well as fungal and mammalian and on a variety of electrode materials [5,10,11]. DET is usually presented as a process substituting the EI-reducing reactions (2) and (3) by the following reaction (4):
EI(FeIV = O, P+) + 2e + 2H+

ks

3 E(FeIII) + H2O

(4)

Heme-containing peroxidases thus belong to the rather restricted group of the


redox enzymes for which DET has been shown [11,12]. If, however, the electronic communication between the cofactor of the peroxidase and the electrode is
slow, small redox molecules can be exploited as mediators to carry the electrons
between the enzyme and the electrode. This mediated electron transfer (MET)
proceeds in accordance with the peroxidase catalytic cycle (reactions 13), where
the role of the mediator can be carried by the reducing substrate AH. The differ-

Electrochemistry of Peroxidases

235

ence is that the oxidized A molecules are regenerated by an electrochemical reaction as follows:
A + e + H+

kS,m

3 AH

(5)

In both cases, i.e., in DET and MET, the peroxidase-modied electrode will generate a reduction current in the presence of H2O2 in the contacting solution. If
electron transfer (ET) mediators are present in the contacting solution, both
processes, DET as well as MET, will contribute to the reduction current depending on the relative efciency of each. In this context it should be noted that the
response signal of the peroxidase-modied electrodes based on either DET or
MET can be affected by interfering compounds in real applications. Compounds
such as acetaminophen in blood can enter the peroxidase cycle as a mediating/
reducing substance. Therefore, the exploitation of efcient mediators or the establishment of very facile DET between the peroxidase and the electrode is crucial
for the design of electrodes with a high selectivity toward H2O2. Thus, it is obvious that studies of both DET- and MET-based processes using different peroxidases and different electrode materials should help in designing selective, highly
sensitive, and stable peroxidase-modied electrodes for detection of H2O2.
In this chapter, we will basically discuss only a few practical developments
of peroxidase-modied electrodes, focusing more on recent results on the characterization of DET of peroxidases. A previous review on peroxidase-modied
electrodes was published by our group in 1996 [5]. Both fundamental and applied
aspects of heterogeneous ET of peroxidases were described. Since then, a few
new trends have appeared in the area of peroxidase electrochemistry. First, several new plant peroxidases have been studied and their electrochemical properties
have been compared with those observed for HRP-modied electrodes. The main
driving force for these studies was to understand what structural properties (e.g.,
glycosylation, pI) determine the rate of the heterogeneous ET of native peroxidases. Second, recombinant wild-type and mutant forms of peroxidases have been
brought into electrochemical investigations, strongly contributing to the understanding of the mechanism of ET between a peroxidase and an electrode. In analytical applications of peroxidase-modied electrodes, several new sensor
designs have appeared, especially in connection with enzyme-based transistors
and immunosensors. Only a few interesting examples of sensors based on peroxidase-modied electrodes are included in this chapter.

II.

THEORETICAL ESTIMATION OF ET PATHWAYS IN HRP

The electronic coupling of the heme (the cofactor of HRP) with the outer surface
of the protein was calculated using the program etunnel [13] with the coordinates

236

Ruzgas et al.

of the recombinant HRP structure ([14] pdb-id 2ATJ). The program approximates
the electron-tunneling rate by an empirical distance dependent expression. It
takes into account the faster and slower tunneling rates for covalent and nonbonded connections by a weighted mean of the barrier constant and by consideration of the atomic packing density, dened as the fraction of the distance
between two centers that is covered by the van der Waals radii of the intervening
atoms. In Figure 1A (see color plate) the calculated ET rates are shown as a colorcoded surface representation of HRP, drawn with GRASP [15], with red colors
indicating high ET rates and blue colors low ET rates. It can be seen that high
rates occur on a belt around the middle of the molecule, where the active site is
located. In the vicinity of the entrance to the active site (e.g., Phe68, Phe142,
Asn72) these calculated rates are in the order of 109 s1, whereas rates at the opposite side are between 105 s1 (Arg159) and 103 s1 (Gln271). The lowest rates are
calculated for the C-terminal end of the molecule, with values lower than 102 s1.
Most commonly, the heavily glycosylated native HRP is used instead of the
recombinant nonglycosylated HRP. Structure models of the major glycan species,
identied by Gray et al. [16], were therefore generated using the SWEET carbohydrate server [17] and attached to the known glycosylation sites [16] on the HRP
structure. The carbohydrate structures attached to the protein were subjected to 70
ps molecular dynamics calculations [18] carried out in a periodic boundary box
lled with water for simulation of glycosylated HRP in solution. For simulation
of glycosylated HRP in dry conditions, the same calculations were carried out in
the absence of water molecules surrounding the protein. In these simulations, the
carbohydrate moieties form large excursions from the protein surface under solution conditions (Figure 1B) but are folded back to the protein surface in calculations simulating dry conditions (Figure 1C). (See color plate.) Based on these
models, an inuence of the glycosylation on the measurable DET rates has to be
expected. In both cases these carbohydrate structures cover the protein surface
areas where the calculated ET rate is high. Even when the glycans are folded back
to the protein surface, they necessarily have lower calculated ET rates than the
corresponding protein residues due to the increase in distance. Especially, the
tips, of the extended glycans have very low calculated ET rates (< 1010 s1) due
to their large distance from the active center and thus would practically abolish
DET if they formed the sole contact to the electrode surface.
The possibility of directional variability in electronic coupling of HRP to an
electrode was examined by docking calculations using a modied version of the
autodock program [19]. The program attempts to determine preferential binding
modes for a ligand, in this case HRP, to a receptor, in this case a planar graphite
surface. Starting from random orientations of the ligand, the program calculates
the afnities (hydrophobic, hydrophilic, and electrostatic) between receptor and
ligand, and optimizes the ligand orientation for increased afnity. However, due

Electrochemistry of Peroxidases

237

to computing and program limitations, it was not possible to take into account
exible side chains on the protein or functional groups like hydroxyl or carboxyl
groups on graphite edges, and thus only hydrophobic interactions were optimized
in this case. For nonglycosylated HRP (Figure 2A; see color plate) the calculations suggest that the protein area in contact with the electrode surface is maximized during adsorption. Considering the shape of the nonglycosylated HRP
molecule as approximately a rotational ellipsoid, this means that the molecule is
preferentially oriented with the long axis parallel to the electrode surface. In this
case, only low directional variability in electronic coupling to an electrode is suggested, taking into account the distribution of surface areas with high and low calculated ET rates on a belt around the middle of the HRP molecule. For the models of glycosylated HRP, however, the calculations suggest that the glycans
inuence the orientation of the HRP molecule on the surface. In the case of the
extended carbohydrate moieties (solution conditions), HRP shows preferential
contact to the surface via the glycans (Figure 2B) and thus suggest very poor electronic coupling due to the low calculated ET rates at the tips of the glycans. For
the model of HRP with folded-back glycans (dry conditions), the calculations
show that in nearly all orientations a mixture of protein and carbohydrate contacts
to the electrode surface is formed (Figure 2C). This model is probably more realistic, taking the exibility of the glycans and a tendency to maximize hydrophobic contacts by adsorption to the surface into account. Anyway, as the carbohydrate moieties inherently have lower calculated ET rates than the protein residues,
the calculations suggest lower electronic coupling to the electrode for glycosylated HRP than for the recombinant enzyme.
Due to the close homology of plant peroxidases [20,21], one can expect that
in general other heme peroxidases will have calculated ET rates similar to HRP
(Figure 1) with good electronic coupling possible on a belt covering the middle
of the ellipsoidal molecule. However, the calculations suggest that this general
feature for the glycosylated wild-type enzymes is modulated by the extent of the
glycans and by the pattern of the glycosylation sites on the protein.

III. EXPERIMENTAL HETEROGENEOUS ET CHARACTERISTICS


OF PEROXIDASES
To characterize the electronic coupling in the peroxidase-electrode system, three
ET reactions can be distinguished. The rst is a one-electron transfer redox
process reecting the conversion between the ferro/ferri redox states, i.e.,
FeII/FeIII. This process is not directly related to the catalytic cycle of peroxidase;
however, this reaction could be of interest for comparison with similar heterogeneous ET reactions of heme, heme-peptides, cytochrome c, or other heme-con-

238

Ruzgas et al.

taining enzymes or proteins. The other two redox processes (1) native peroxidase/compound I (i.e., FeIII/FeIV = O, P+) and (2) compound I/compound II (i.e.,
FeIV = O, P+/FeIV = O), reect redox transformation of the active site in the course
of the catalytic cycle of peroxidase and are highly relevant for the performance of
peroxidase-modied electrodes.

A.

Direct Heterogeneous FeII/FeIII Conversion


of Heme-Containing Peroxidases

The value of the formal potential, Eo, of the FeII/FeIII redox process depends on
the origin of peroxidase and the solution pH [22]. For HRP this value is found to
be equal to 278 mV vs. NHE evaluated from homogeneous redox titration of the
enzyme at pH 7 [23,24]. For cytochrome c peroxidase, the similar process is characterized by E0 = 194 mV at pH 7 [25]. There are a number of reports stating
the observation of the direct heterogeneous FeII/FeIII redox process for HRP by
means of cyclic voltammetry. Accordingly, voltammograms with well-dened
redox peaks at negative potentials (usually, with a mid-point potential in the range
170 to 100 mV vs. NHE) have been recorded with HRP entrapped into anionic
exchange resin [26], DNA lm [27,28], didodecyldimethylammonium bromide
lm [29], or deposited from an organic solvent solution [30] on glassy carbon.
Unfortunately, the conclusion of direct electrochemistry of HRP deduced from
the well-pronounced redox signals were never supported by proper/correct complementary results on the bioelectrocatalytic reduction of H2O2 at these electrodes. This process should start at approximately 860 mV vs. NHE, i.e., a potential close to that determined by the redox reactions (2) and (3) of the catalytic
cycle of HRP. The electrocatalysis at the mentioned electrodes, however, invokes
at potentials close to the observed FeII/FeIII waves and seems to coincide well
with the electrocatalytic reduction of H2O2 at heme-modied electrodes [31].
Similar effects on the electrocatalysis of H2O2 have been observed for hemepeptide-(microperoxidase 11)-modied electrodes in organic solvents [32]. All these
investigations indicate that the mechanism of H2O2 electroreduction at the mentioned HRP-modied electrodes is different from that described by reactions (1)
and (4), i.e., it proceeds through or involves reaction intermediates different from
compounds I and II.
The values of Eo estimated from cyclic voltammograms of the mentioned
HRP-modied electrodes, about 60 [29], 90 [28], 136 [26], and 155 mV vs.
NHE at pH 7 [30], are more than 100 mV more positive than the value of Eo =
278 mV of FeII/FeIII found for soluble HRP. The electrochemically measured Eo
values seem to be closer to the Eo of the redox process of surface-conned heme,
specically, 91, 110, 127 mV [33], 118 mV [11], and 178 mV [32] or surface-conned hemepeptide (microperoxidase-11), for which the following values

Electrochemistry of Peroxidases

239

of Eo have been reported: 118 mV [32], 155 mV [34,35]. Keeping in mind the
discussion on bioelectroreduction of H2O2 (above) and this comparison between
Eo values, it should be mentioned that these facts indicate a substantial exposure
of the active site of HRP to the solution at the HRP-modied electrodes, perhaps
as a result of the electrode-modication procedures.
An exception to the above could be the observed cyclic voltammograms
recorded for HRP at a tin-doped indium oxide electrode [36]. This contribution
reports on a midpoint potential of the voltammograms equal to 270 mV vs.
NHE, which agrees well with the expected Eo of FeII/FeIII redox conversion for
HRP, however, any data on the bioelectrocatalytic reduction of H2O2 at these
electrodes remain absent.

B.

Direct Heterogeneous Compound I/FeIII Conversion


of Heme-Containing Peroxidases

Efcient electronic coupling between superoxidized peroxidase [compounds I


and II, i.e., EI and EII in reactions (13)] and an electrode is very important in
designing highly sensitive peroxidase-modied electrodes for detection of H2O2
possessing also a broad linear response range. Even more, it is well known that
for most peroxidases compound I is unstable. It is inferred that EI is spontaneously reduced by the amino acid residues in the protein, thus damaging the protein [3739] and consequently causing reduced stability of the electrodes. Thus
rapid reduction of compound I back to the FeIII redox state through DET (discussed in this work) or MET (elsewhere [40,41]) is of crucial practical importance. This is the practical motivation for thorough investigations of the kinetics,
thermodynamics, and mechanism of the heterogeneous ET process for the
EI/E(FeIII) couple.
As mentioned in the introduction, in homogeneous solution compound I
can be reduced to FeIII by two one-electron transfer steps (through the intermediate, compound II) or directly in one two-electron transfer step, depending on the
nature of the electron donor. What is known about this process at peroxidasemodied electrodes? The thermodynamic Eo potentials of FeIII/compound II and
compound II/compound I depend on the origin of peroxidase and solution pH
[42]. For HRP the values of Eo (compound I/compound II) = 879 and Eo (compound II/FeIII) = 903 mV vs. NHE have been determined from homogeneous
redox titrations at pH 7 [43]. There is no doubt that direct heterogeneous reduction of compound I to FeIII is possible on a broad variety of electrode materials as
reviewed previously [5]. The process is usually described by reaction (4), above.
The question is whether this is a one-step two-electron cooperative process, or if
it proceeds by two single-electron transfer steps. A two-electron transfer cooperative process is inferred for cytochrome c peroxidase adsorbed on highly ordered

240

Ruzgas et al.

edge plane pyrolytic graphite [44]. This conclusion is based on the analysis of
cyclic voltammograms, which show only one narrow peak characterized with a
midpoint potential value of 736 mV vs. NHE at pH 5.8 [45]. Even more, the data
demonstrate that reversible electrochemical cycling between compound I and
FeIII redox states of cytochrome c peroxidase is possible. Striking here is that one
would probably expect the process to be extremely slow because formation of
compound I as well as compound II requires covalent attachment of an oxygen
atom to give the oxoferryl iron. However, this might not be the case. Extremely
low energies of reorganization, , were found for the process of reduction of compound I to FeIII by phenols, = 0.6 eV [46], by ferrocene derivatives = 0.45 eV
[47], and by indoleacetic acid, = 0.01 eV [48]. These data suggest that the redox
reactions, compound I/FeIII or compound II/FeIII, are electron-transfer limited
processes and thus support the electrochemical data revealed by cyclic voltammetry, indicating good reversibility of the corresponding processes driven electrochemically.
Similar experimental data with HRP, however, do not support a cooperative
two-electron process. Cyclic voltammograms of HRP immobilized on gold
[49,50] or soluble at tin oxide [36] electrodes have been reported. The voltammograms indicate either one or two one-electron transfer processes and probably
cannot be ascribed to electrochemical cycling of HRP between compound I, compound II, and FeIII states. The reason for this critical conclusion is that abovementioned cyclic voltammograms possess redox peaks with midpoint potentials
far more negative than should be expected for interconversions involving compound I, compound II, and FeIII. As suggested by the authors [36], the electrochemical processes could be related to the oxidation of the porphyrin ring to a cation radical in HRP. Such a redox conversion was found to be the initial
reaction on the way from FeIII to compound II in HRP upon its photochemical
oxidation by Ru-bipyridine [51]. Keeping in mind all these facts, it can be stated
that more experimental data are necessary to judge on the cooperativity of the ET
process, which returns compound I to the FeIII redox state.
C.

Kinetics of Electrochemical Reduction of Compound I to FeIII

Currently it is hard to distinguish with purely electrochemical means whether the


mediator-free electrochemical reduction of compound I to FeIII proceeds through
compound II. Thus, the rate of the reduction of compound I to FeIII is considered
as the rate reecting the overall process converting superoxidized peroxidase to
the resting FeIII form. The most extensive experimental data on the heterogeneous
reduction of compound I come from measurements of ks (see reaction 4) obtained
from rotating disc electrode (RDE) technique with HRP-modied electrodes. As
it was reviewed recently [5], the most efcient ET has been found for HRP-

Electrochemistry of Peroxidases

241

adsorbed on spectrographic (ks 1 s1 [52]) and for cytochrome c peroxidase on


highly ordered pyrolytic (ks 110 s1 [53]) graphite electrodes. It has to be mentioned that in the mechanism proposed for HRP-modied electrodes, it was
assumed that not the entire population of adsorbed enzyme molecules is directly
electronically coupled with the electrode [11,54]. It was found that about 50% of
the adsorbed peroxidase molecules do not possess this property at graphite electrodes (electronically connected only via MET) [52,54]. Understanding the reasons for this was addressed by studying the electrochemical properties of plant
peroxidases from different origins as well as recombinant HRP forms (see
below).
All native plant peroxidases are glycosylated enzymes and thus the rst
question that has been tackled was whether the glycosylation hindered DET
between peroxidases and graphite. HRP, peanut (PNP), sweet potato (SPP) and
tobacco (TOP) peroxidases possessing 21, 20, 16, and 11% glycosylation by
weight, respectively, have been studied using the RDE techniques [55]. In general, it appeared that lower glycosylation of the enzyme enables a higher percentage of adsorbed peroxidase molecules to be directly electronically connected with
the graphite surface (see Table 1); e.g., 68% of TOP adsorbed on the electrode
was electroactive via DET, which is higher than 48% found for HRP. The percentage of the TOP molecules in direct ET with graphite increased to 83% at the
presence of Ca2+ [56], which probably affected the conformation of the surfacebound enzyme. However, SPP gave the best electronic coupling with the electrode. Similarly, all other kinetic characteristics were found to be better for
adsorbed SPP when compared with other plant peroxidases used in this study; see
Table 1 [55]. The conclusion that could be drawn here is that not only the extent
of the glycosylation but also the pattern of the glycosylation sites on the peroxidase strongly determines the electronic coupling between the enzyme and the
electrode, as suggested by theoretical ET calculations (above).
The nal proof that glycosylation of plant peroxidases hinders DET connection of the protein with the electrode has been achieved by comparing electrochemically derived kinetic characteristics of electrodes modied with native
HRP or its glycosylation-free recombinant forms [57]. By removing the carbohydrate shell, the percentage of HRP molecules possessing DET contact
increased from 48% to 63% (Table 1). The absence of the glycosylation also had
strong effect on the rate of DET (ks in reaction 4), which increased from 1.9 to 7.6
s1 for native and recombinant HRP on graphite, respectively. A more thorough
interpretation of the results is, however, difcult because peroxidases from different origins possess not only different degrees of glycosylation but also different pIs, as well as small differences in structure, which might strongly affect the
direct electronic coupling of the peroxidase with the electrode. Additionally, the
graphite surface is highly heterogeneous [58] providing a broad variety of inter-

21%
20%
1617%?
11%
0%
0%
0%
0%

Glycosylation
48 4
44 7
91 6
68 10
63 7
64 5
100a
100a

% in direct ET
1.3 0.2
1.3 0.3
2.8 0.8
0.53 0.03
2.8 0.6
1.5 0.3
0.52 0.07
0.53 0.13

k1/105 M1s1
1.9 0.3
1.3 0.3
4.8 2.3
2.6 1.0
7.6 2.5
2.6 1.0
2.4 0.6
2.2 0.5

Rate constants ks/s1

3.8 1.6
5.0 3.3
35 18
1.9 0.7
14 7
4.3 2.0
a
a

k3/104 M1s1

2.1 0.2
2.3 0.5
1.9 0.1
2.0 0.2
1.8 0.2
1.8 0.1
1.8 0.4
1.8 0.2

Number of electrons

PNP, peanut peroxidase; TOP, tobacco peroxidase; SPP, sweet potato peroxidase; recHRP, recombinant wild type (WT) HRP. HRP Phe221Trp, HRP Asn70Val, HRP
Asn70Asp, are HRP mutants.
aNo mediated ET was observed with p-cresol, catechol, p-aminophenol, or guaiacol.

8.8
7.8
3.5
3.5
9.05

pI

Electrochemically Determined Characteristics of Direct and Mediated ET for Plant Peroxidases Adsorbed on Graphite Electrodes

HRP
PNP
SPP
TOP
recHRP
HRPPhe221Trp
HRP Asn70Val
HRP Asn70Asp

Table 1

Electrochemistry of Peroxidases

243

actions between the electrode and the peroxidase and thus masking the effects that
might be related to relatively small structural differences in the structures of plant
peroxidases. Exploitation of metal electrodes with a more dened surface might
be preferable in this kind of study.
Recent experiments with gold electrodes for the investigation of the heterogeneous properties of HRPs have revealed that the electrode-peroxidase interaction depends on the structure of the HRP mutant as well as on the quality of the gold
surface, which is probably dened by the method of its pretreatment [5961]. The
most striking nding at the moment is that glycosylated native HRP does not result
in stable HRP-modied electrodes. The instability is especially pronounced in the
presence of H2O2, and is probably caused by gradual desorption of HRP from gold.
A similar problem with native HRP is mentioned in the preparation of HRP-modied diamond electrodes [62]. Recombinant, glycosylation-free HRP enables 58%
of the adsorbed peroxidase molecules being in DET contact with gold [61]. The
rate of DET, ks, reaches 18 s1 (Table 2), indicating substantial improvement when
compared with HRP on graphite (Table 1).
A more rational way of studying the peroxidase-electrode contact and optimizing the electronic connection could be achieved by exploiting genetically
engineered enzymes. In this context, the surface of HRP has been designed to
incorporate six-histidine tags at either the C- or the N-terminus of the enzyme
[63]. The electrochemical characteristics obtained with gold electrodes modied
when using C- and N-terminus mutants [61] are summarized in Table 2. It can be
seen that the histidine tags bring further improvement on direct electronic connection between HRP and gold. The highest average fraction of HRP in DET with
gold reaches 75% for the C-terminus histidine mutant and the rate constant for

Table 2
Characteristics of Bioelectrochemical Reduction of H2O2 at Gold Electrodes Modied with Native or Different Recombinant Forms of HRP

native HRP
rec-HRP
CHisrec-HRP
NHisrec-HRP

% in DET

ks/s1

k1/M1s1

k3/M1s1

*
58 11
75 17
63 31

1
18.0 3.4
33.2 6.2
32.7 5.6

0.02 106
(1.00 0.20) 106
(0.84 0.21) 106
(1.57 0.32) 106

*
(14.1 6.3) 104
(16.7 8.2) 104
(7.2 3.1) 104

The values are calculated from the data of RDE experiments performed in 0.01 M phosphate buffer
(pH 7.4) containing 0.1 M NaCl. DET indicates the fraction (in %) of the peroxidase molecules on
the electrode, which possesses a direct (mediatorless) electronic coupling with the surface of the
gold. Rate constant k1 represents the rate of H2O2 reaction with HRP(FeIII), i.e., the rate of the production of EI. Constants ks and k3 represent the rate of the reduction of EI to HRP(FeIII) by direct
heterogeneous ET from the electrode or by soluble catechol (mediated ET), respectively.
*not determined due to low stability of gold electrodes modied with native HRP.

244

Ruzgas et al.

direct ET further increases to 33 s1. At the moment, however, it is too early to


speculate on the exact reasons resulting in the positive effect brought by the histidine tag. At neutral pH, histidine molecules interact strongly with gold. It was
found that histidine is preferentially adsorbed to gold through the deprotonated
nitrogen atom of the imidazole ring [64]. This might impose some orientation of
the HRP mutants on gold. As presented above, the calculations of ET tunneling
in HRP exclude the possibility of the histidine tags being a part of the ET pathway between the active site of the peroxidase and the electrode.
A practical consequence of the efcient electronic coupling between the
histidine mutants of HRP and gold is that for the rst time, HRP-modied electrodes were found to enable a diffusion-limited sensitivity for H2O2 (1.5 A M1
cm2) in the range of 0.140 M [60]. This sensitivity is similar to the most sensitive electrodes based on MET of HRP [65]. From this comparison it should be
concluded that the statement mediated ET between HRP and an electrode is
more efcient than direct, being exploited from the discovery of direct heterogeneous ET of HRP [9], is no longer valid. Current results demonstrate that at low
concentrations of H2O2 (i.e., below 40 M), ET mediators do not have any
enhancing effect on the electronic coupling of HRP to gold [60].
Are further improvements on DET between the electrode surface and the
peroxidase practically needed and possible? As mentioned above, the better the
electronic coupling between the peroxidase and the electrode is, the better the stability of the peroxidase-modied electrodes should be. So, even if the sensitivity
of the HRP-modied gold electrodes cannot be increased above its diffusion limitations, from the stability point of view a further enhancement of the electronic
coupling is of obvious interest. The possibility of increasing the rate of direct heterogeneous ET between HRP and the electrode surface denitely exists. This is
indicated by the electron coupling calculations described above and by the homogeneous measurements of the ET rates between compound I of cytochrome c peroxidase and cytochrome c [66]. The rate of the inter-protein ET in this system
could reach values such as 2 106 s1.

IV. EXAMPLES OF NEW APPLICATIONS OF PEROXIDASEMODIFIED ELECTRODES


In general, peroxidase-modied electrodes can be used for detection of H2O2 and
organic hydroperoxides [5,6769], phenolic substances [7072], peroxidase
inhibitors [73,74], and other metabolites through combined use of peroxidases and
H2O2-producing oxidases [75,76]. During the past ve years, more than 200 published articles could be found through thorough literature search describing further
improvements relevant to the development of HRP-modied electrodes for detec-

Electrochemistry of Peroxidases

245

tion of different analytes. This chapter does not cover these developments and
applications in an extensive manner. Instead, a few interesting developments in the
area of peroxidase-modied electrodes, e.g., detection based on the determination
of a precipitation product derived from the peroxidase catalyzed reaction [77] or
monitoring of nitric oxide [78], are exemplied below, some of them being distinctly different from the classical use of peroxidase-modied electrodes.
One of the most interesting applications is the development of an enzyme
transistor [79]. The idea is based on the DET between HRP and a poly(aniline)
lm, which loses electrical conductivity due to its oxidation by H2O2 catalyzed by
the enzyme. As mentioned by the authors, the enzyme transistor can form the
basis for a new generation of devices distinctly different from usual amperometric electrodes; however, a number of limitations still need to be overcome, such
as fast reproducible switching of the transistor and inactivation of peroxidase. The
last limitation might not be a problem in the format of a disposable sensor.
Another interesting application of DET between peroxidase and an electrode is the development of separation-free electrochemical immuno sensors.
Such a format of immunoassays is especially attractive, because no washing procedures are required. The application of an immunosensor incorporating disposable screen-printed HRP-modied electrodes as the detector in conjunction with
an atrazine immune-membrane was demonstrated [80]. The idea of such an
immunosensor is based on the fact that the enzyme-labeled antigen is complexed
by the surface-bound antibody. As a label, an H2O2-producing oxidase is used. In
the presence of the oxidase substrate, H2O2 is generated, which is detected electrochemically by the HRP-modied screen-printed electrode. To conne the reaction only to the electrode surface, the H2O2 generated in the bulk of the solution
is destroyed by catalase. Other peroxidase-based electrochemical immunosensor
formats also are possible; they have been recently reviewed elsewhere [81]. A
similar idea, however, based on mediated HRP electrodes was proposed for
immunoanalysis of analytes in blood [82] and for verication of DNA amplication using polymerase chain reaction [83].
A substantial improvement of the selectivity of amperometric biosensors
based on peroxidase/H2O2-producing oxidase enzyme can be achieved by choosing recombinant HRP instead of the native glycosylated enzyme. Recombinant
HRP has a much better electronic coupling to the electrode and thus possible interferences, which act as mediators, e.g., acetaminophen and dopamine, will affect
the biosensor signal to a much lesser extent. As an example, calibration plots for
glucose detection in a ow injection system are presented in Figure 3. As can be
seen, the presence of acetaminophen in the solution together with glucose strongly
increases the response signal at the electrode modied with glucose oxidase/native
HRP reecting the action of acetaminophen as an efcient mediator. A similar calibration plot for an electrode based on coadsorbed glucose oxidase/recombinant
HRP exhibits less difference between the slopes of current vs. glucose concentra-

246

Ruzgas et al.

Figure 3 Calibration plots for glucose in the presence () and in the absence () of 0.1
mM acetaminophen. The calibrations were recorded with graphite electrodes modied with
bienzyme layer of (A) native HRP and glucose oxidase or (B) recombinant HRP and glucose oxidase. The measurements were done in a ow-injection mode, with a ow carrier of
20 mM phosphate buffer, pH 7 containing 0.1 M KCl. Applied potential was 50 mV vs.
Ag/AgCl/0.1 M KCl.

tion in the presence and the absence of acetaminophen. Additional selectivity


improvement of peroxidase-modied electrodes can be achieved by genetic engineering of HRP by single-point mutations of the amino acid residues within the
active site of the enzyme [8486]. The mutations suppress the reactivity of the
enzyme with mediators such as phenols, acetaminophen, or dopamine. The
increase in the selectivity was studied with graphite electrodes modied with such
HRP mutants. The selectivity was evaluated by calculating the ratio between the
response of the electrode to H2O2 in the presence and in the absence of acetaminophen. This ratio was the following: 1.6 (for native HRP), 1.2 (recombinant HRP),
1.14 (Asn70Val HRP mutant), and 1.09 (Asn70Asp HRP mutant). Comparing the
obtained results, it can be seen that the use of the Asn70Asp mutant resulted in the
most selective electrodes for H2O2 detection.

Electrochemistry of Peroxidases

247

Figure 3B

The majority of the examples mentioned here were based on DET of peroxidases and especially DET of HRP, the most commonly used peroxidase. However, because native HRP exhibits limited direct electronic coupling with electrodes, the principle of MET has been most frequently exploited for design of
HRP-modied electrodes. A good example is the phenothiazine-based HRPmodied electrode with detection limit for H2O2 below 1 nM [87].

V.

CONCLUSIONS

The electrochemistry of peroxidases continues to be an intensive area of research.


Critical evaluation of the results on the bioelectrochemistry of peroxidases leads
to the conclusion that efcient direct electronic coupling between peroxidases and
electrodes is possible. Rapid methods of genetic engineering of peroxidases
[8890] offer a new tool of tuning peroxidase-electrode interactions. This will
facilitate the understanding of the electrochemistry of peroxidases as well as lead
to the development of more selective and stable peroxidase-modied electrodes.

248

Ruzgas et al.

New and interesting developments might appear by exploiting peroxidases containing several heme centres [91,92] as well as nonheme peroxidases [9395].
ACKNOWLEDGMENTS
The authors thank the following organizations for nancial support: the Swedish
Board for Industrial and Technical Development (NUTEK), the Swedish Natural
Science Research Council (NFR), and the European Commission (contract number BIO4-CT97-2199).
REFERENCES
1.
2.
3.

4.
5.
6.

7.

8.

9.
10.
11.

12.

H. B. Dunford, Heme Peroxidase Nomenclature. Plant Peroxidase Newsletters, 13


(1999) 6571.
H. B. Dunford, Heme Peroxidases. Wiley-VCH, NY, 1999, 480.
I. Karube, Analytical Application of Enzymes: Enzyme Sensors for Clinical,
Process, and Environmental Analyses, in: J. F. Kennedy (Ed.), Biotechnology. VCH,
Weinheim, 7a (1987) 685708.
Y. Kasahara, New Advancement of Enzymatic Methodologies in Clinical Laboratory Analysis. J. Clin. Lab. Anal., 8 (1994) 3543.
T. Ruzgas, E. Csregi, J. Emnus, L. Gorton, G. Marko-Varga, Peroxidase-Modied
Electrodes: Fundamentals and Application. Anal. Chim. Acta, 330 (1996) 123138.
S. B. Hall, E. A. Khudaish, A. L. Hart, Electrochemical Oxidation of Hydrogen Peroxide at Platinum Electrodes. Part V: Inhibition by Chloride. Electrochim. Acta, 45
(2000) 35733579.
I. L. De Mattos, L. Gorton, T. Ruzgas, A. A. Karyakin, Sensor for Hydrogen Peroxide Based on Prussian Blue Modied Electrode: Improvement of the Operational
Stability. Anal. Sci., 16 (2000) 795798.
A. A. Karyakin, G. V. Presnova, M. Y. Rubtsova, A. M. Egorov, Oriented Immobilization of Antibodies onto the Gold Surfaces via Their Native Thiol Groups. Anal.
Chem., 72 (2000) 38053811.
A. I. Yaropolov, M. R. Tarasevich, S. D. Vorfolomeev, Electrochemical Properties
of Peroxidase. Bioelectrochem. Bioenerg., 5 (1978) 1824.
A. L. Ghindilis, P. Atanasov, E. Wilkins, Enzyme-Catalyzed Direct Electron Transfer: Fundamentals and Analytical Applications. Electroanalysis, 9 (1997) 661674.
L. Gorton, A. Lindgren, T. Larsson, F. D. Munteanu, T. Ruzgas, I. Gazaryan, Direct
Electron Transfer between Heme-Containing Enzymes and Electrodes as Basis for
Third Generation Biosensors. Anal. Chim. Acta, 400 (1999) 91108.
H. A. O. Hill, N. I. Hunt, Direct and Indirect Electrochemical Investigations of Metalloenzymes, in: J. F. Riordan, B. L. Vallee (Eds.), Methods in Enzymology, Metallobiochemistry. Part D, 227 (1993) 501522.

Electrochemistry of Peroxidases
13.
14.

15.

16.
17.
18.

19.

20.
21.

22.

23.
24.

25.

26.

27.

28.

249

C. C. Page, C. C. Moser, X. Chen, P. L. Dutton, Natural Engineering Principles of


Electron Tunneling in Biological Oxidation-reduction. Nature, 402 (199) 4752.
A. Henriksen, D. J. Schuller, K. Meno, A. T. Smith, K. G. Welinder, M. Gajhede,
Structural Interactions between Horseradish Peroxidase C and the Substrate Acid
Determined by X-Ray Crystallography. Biochemistry, 37 (1998) 80548060.
A. Nicholls, K. A. Sharp, B. Honig, Protein Folding and Association: Insights from
the Interfacial and Thermodynamic Properties of Hydrocarbons. Proteins Struct.
Funct. Gen., 11 (1991) 281296.
J. S. Gray, B. Y. Yang, R. Montgomery, Heterogeneity of Glycans at Each N-glycosylation Site of Horseradish Peroxidase. Carbohydr. Res., 311 (1998) 6169.
A. Bohne, E. Lang, C. W. von der Lieth, W3-SWEET: Carbohydrate Modelling by
Internet. J. Mol. Model., 4 (1998) 3343.
D. A. Case, D. A. Pearlman, J. W. Caldwell, T. E. Cheatham III, W. S. Ross,
C. L. Simmerling, T. A. Darden, K. M. Merz, R. V. Stanton, A. L. Cheng, J. J. Vincent, M. Crowley, D. M. Ferguson, R. J. Radmer, G. L. Seibel, U. C. Singh,
P. K. Weiner, P. A. Kollman, AMBER 5, University of California, San Francisco.
http://www.amber.ucsf.edu/amber/index.html, (1997)
G. M. Morris, D. S. Goodsell, R. S. Halliday, R. Huey, W. E. Hart, R. K. Belew,
A. J. Olson, Automated Docking Using a Lamarckian Genetic Algorithm and Empirical Binding Free Energy Function. J. Comput. Chem., 19 (1998) 16391662.
K. G. Welinder, Superfamily of Plant, Fungal and Bacterial Peroxidases. Curr. Opin.
Struc. Biol., 2 (1992) 388393.
G.-H. Huh, S.-J. Lee, Y.-S. Bae, J. R. Lui, S.-S. Kwak, Molecular Cloning and Characterization of cDNAs for Anionic and Neutral Peroxidases from Suspension Cultured-Cells of Sweet Potato and Their Differential Expression in Response to Stress.
Mol. Gen. Genet., 255 (1997) 382391.
C. D. Millis, D. Cai, M. T. Stankovich, M. Tien, Oxidation-Reduction Potentials and
Ionization States of Extracellular Peroxidases from Lignin-Degrading Fungus
Phanerochaete chrysosporium. Biochemistry, 28 (1989) 84848489.
H. A. Harbury, Oxidation-Reduction Potentials of Horseradish Peroxidase. J. Biol.
Chem., 225 (1957) 10091024.
H. Yamada, R. Makino, I. Yamazaki, Effect of 2,4-Substituents of Deuteroheme
upon Redox Potentials of Horseradish Peroxidases. Arch. Biochem. Biophys., 169
(1975) 344353.
C. W. Conroy, P. Tyma, P. H. Daum, J. E. Erman, Oxidation-Reduction Potential
Measurements of Cytochrome c Peroxidase and pH Dependent Spectral Transitions
in the Ferrous Enzyme. Biochim. Biophys. Acta, 537 (1978) 6269.
T. Ferri, A. Poscia, R. Santucci, Direct Electrochemistry of Membrane-Entrapped
Horseradish Peroxidase. Part I. A Voltammetric and Spectroscopic Study. Bioelectrochem. Bioenerg., 44 (1998) 177181.
X. Chen, C. Ruan, J. Kong, J. Deng, Spectroelectrochemical Investigation of Direct
Electron Transfer between Resting Horseradish Peroxidase and its Oxidation States
Promoted by DNA. Fresenius J. Anal. Chem., 367 (2000) 172177.
X. Chen, C. Ruan, J. Kong, J. Deng, Characterization of the Direct Electron Trans-

250

29.

30.

31.

32.

33.

34.

35.
36.

37.

38.
39.

40.

41.

42.

43.

Ruzgas et al.
fer and Bioelectrocatalysis of Horseradish Peroxidase in DNA Film at Pyrolytic
Graphite Electrode. Anal. Chim. Acta, 412 (2000) 8998.
X. Chen, X. Peng, J. Kong, J. Deng, Facilitated Electron Transfer from an Electrode
to Horseradish Peroxidase in a Biomembrane-like Surfactant Film. J. Electroanal.
Chem., 480 (2000) 2633.
Y. Guo, A. R. Guadalupe, Direct Electrochemistry of Horseradish Peroxidase
Adsorbed on Glassy Carbon Electrode from Organic Solution. Chem. Commun.,
(1997) 14371438.
T. Ferri, A. Pascia, R. Santucci, Direct Electrochemistry of Membrane-Entrapped
Horseradish Peroxidase. Part II: Amperometric Detection of Hydrogen Peroxide.
Bioelectrochem. Bioenerg., 45 (1998) 221226.
A. N. J. Moore, E. Katz, I. Willner, Electrocatalytic Reduction of Organic Peroxides
in Organic Solvents by Microperoxidase-11 Immobilized as a Monolayer on a Gold
Electrode. J. Electroanal. Chem., 417 (1996) 189192.
H. Zimmermann, A. Lindgren, W. Schuhmann, L. Gorton, Anisotropic Orientation
of Horseradish Peroxidase by Reconstitution on a Thiol-Modied Gold Electrode.
Chem. Eur. J., 6 (2000) 592599.
T. Ltzbeyer, W. Schuhmann, E. Katz, J. Falter, H.-L. Schmidt, Direct Electron
Transfer between Covalently Immobilized Enzyme Microperoxidase MP-11 and a
Cystamine-Modied Gold Electrode. J. Electroanal. Chem., 377 (1994) 291294.
T. Ruzgas, A. Gaigalas, L. Gorton, Diffusionless Electron Transfer of Microperoxidase-11 on Gold Electrodes. J. Electroanal. Chem., 469 (1999) 123131.
K. Chattopadhyay, S. Mazumdar, Direct Electrochemical Oxidation of Horseradish
Peroxidase: Cyclic Voltammetric and Spectroelectrochemical Studies. New J.
Chem., (1999) 137139.
G. Wu, C. Wei, R. J. Kulmacz, Y. Osawa, A. Tsai, A Mechanistic Study of Self-inactivation of the Peroxidase Activity in Prostaglandin H Synthase-1. J. Biol. Chem.,
274 (1999) 92319237.
H. Wright, J. A. Nicell, Characterization of Soybean Peroxidase for the Treatment of
Aqueous Phenols. Bioresour. Technol., 70 (1999) 6979.
W. A. Doyle, W. Blodig, N. C. Veitch, K. Piontek, A. T. Smith, Two Substrate Interaction Sites in Lignin Peroxidase Revealed by Site-directed Mutagenesis. Biochemistry, 37 (1998) 1509715105.
S. J. Sagedhi, G. Gilardi, A. E. G. Cass, Mediated Electrochemistry of Peroxidase
Effects of Variations in Protein and Mediator Structures. Biosens. Bioelectron., 12
(1997) 11911198.
A. D. Ryabov, V. S. Kurova, V. N. Goral, M. D. Reshetova, J. Razumiene, R.
Simkus, V. Laurinavicius, p-Ferrocenylaniline and p-Ferrocenylphenol: Promising
Materials for Analytical Biochemistry and Bioelectrochemistry. Chem. Mater., 11
(1999) 600604.
Y. Hayashi, I. Yamazaki, The Oxidation-Reduction Potentials of Compound I/Compound II and Compound II/Ferric Couples of Horseradish Peroxidase A2 and C. J.
Biol. Chem., 254 (1979) 91019106.
B. He, R. Sinclair, R. B. Copeland, R. Makino, L. S. Powers, I. Yamazaki, The Structure-Function Relationship and Reduction Potentials of High Oxidation States of
Myoglobin and Peroxidase. Biochemistry, 35 (1996) 24132420.

Electrochemistry of Peroxidases
44.

45.

46.

47.

48.

49.
50.
51.
52.

53.
54.
55.

56.

57.

58.

59.

251

M. S. Mondal, D. B. Goodin, F. A. Armstrong, Simultaneus Voltammetric Comparisons of Reduction Potentials, Reactivities, and Stabilities of the High-Potential Catalytic States of Wild-Type and Distal-Pocket (W51F) Yeast Cytochrome c Peroxidase. J. Am. Chem. Soc., 120 (1998) 62706276.
H. A. Heering, J. Hirst, F. Armstrong, Interpreting the Catalytic Voltammetry of
Electroactive Enzymes Adsorbed on Electrodes. J. Phys. Chem., 102 (1998)
68896902.
L. K. Folkes, L. P. Candeias, Interpretation of the Reactivity of Peroxidase Compounds I and II with Phenols by the Marcus Equation. FEBS Letters, 412 (1997)
305308.
A. D. Ryabov, V. N. Goral, E. V. Ivanova, M. D. Reshetova, A. Hradsky, B. Bildstein, Linear Free-energy Relationships and Inverted Marcus Region in the Horseradish Peroxidase-catalyzed Oxidation of Ferrocenes by Hydrogen Peroxide. J.
Organomet. Chem., 589 (1999) 8591.
L. P. Candeias, L. K. Folkes, P. Wardman, Factors Controlling the Substrate Specicity of Peroxidases: Kinetics and Thermodynamics of the Reaction of Horseradish
Peroxidase Compound I with Phenols and Indole-3-acetic Acids. Biochemistry, 36
(1997) 70817085.
S. Dong, J. Li, Self-assembled Monolayers of Thiols on Gold Electrodes for Bioelectrochemistry and Biosensors. Bioelectrochem. Bioenerg., 42 (1997) 713.
J. Li, S. Dong, The Electrochemical Study of Oxidation-Reduction Properties of
Horseradish Peroxidase. J. Electroanal. Chem., 431 (1997) 1922.
J. Berglund, T. Pascher, J. R. Winkler, H. B. Gray, Photoinduced Oxidation of
Horseradish Peroxidase. J. Am. Chem. Soc., 119 (1997) 24642469.
A. Lindgren, F.-D. Munteanu, I. Gazaryan, T. Ruzgas, L. Gorton, Comparison of
Rotating Disk and Wall-Jet Electrode Systems for Studying the Kinetics of Direct
and Mediated Electron Transfer for Horseradish Peroxidase on a Graphite Electrode.
J. Electroanal. Chem., 458 (1998) 113120.
D. L. Scott, E. F. Bowden, Enzyme-Substrate Kinetics of Adsorbed Cytochrome c
Peroxidase on Pyrolytic Graphite Electrodes. Anal. Chem., 66 (1994) 12171223.
T. Ruzgas, L. Gorton, J. Emnus, G. Marko-Varga, Kinetic Models of Horseradish
Peroxidase Action on a Graphite Electrode. J. Electroanal. Chem., 391 (1995) 4149.
A. Lindgren, T. Ruzgas, L. Gorton, E. Csregi, G. B. Ardila, I. Y. Sakharov,
I. G. Gazaryan, Biosensors Based on Novel Peroxidases with Improved Properties in
Direct and Mediated Electron Transfer. Biosens. Bioelectron., 15 (2000) 491497.
F.-D. Munteanu, L. Gorton, A. Lindgren, T. Ruzgas, J. Emnus, E. Csregi, I. G.
Gazaryan, I. V. Ouporov, E. A. Mareeva, L. M. Lagrimini, Direct and Mediated
Electron Transfer Catalyzed by Anionic Tobacco Peroxidase. Effect of Calcium
Ions. Appl. Biochem. Biotech., 84 (2000) 113.
A. Lindgren, M. Tanaka, T. Ruzgas, L. Gorton, I. Gazaryan, K. Ishimori, I. Morishima, Direct Electron Transfer Catalysed by Recombinant Forms of Horseradish
Peroxidase: Insight into the Mechanism. Electrochem. Comm., 1 (1999) 171175.
R. L. McCreery, Carbon Electrodes: Structural Effects on Electron Transfer Kinetics, in: A. J. Bard (Ed.), Electroanalytical Chemistry. Dekker, New York, 1991, vol.
17, 221374.
G. Presnova, V. Grigorenko, A. Egorov, T. Ruzgas, A. Lindgren, L. Gorton, T.

252

60.

61.

62.
63.

64.
65.

66.

67.

68.

69.

70.

71.

72.

Ruzgas et al.
Brchers, Direct Heterogeneous Electron Transfer of Recombinant Horseradish Peroxidase on Gold. Faraday Discuss, 116 (2000) 281284.
E. E. Ferapontova, V. G. Grigorenko, A. M. Egorov, T. Brchers, T. Ruzgas, L. Gorton, Mediatorless Biosensor for H2O2 Based on Recombinant Forms of Horseradish
Peroxidase Directly Adsorbed on Polycrystalline Gold. Biosens. Bioelectron. 16
(2001) 147157.
E. E. Ferapontova, V. G. Grigorenko, A. M. Egorov, T. Brchers, T. Ruzgas, L. Gorton, Direct Electron Transfer in the System Gold ElectrodeRecombinant Horseradish Peroxidase. J. Electroanal. Chem. 509 (2001) 1926.
T. Tasuma, H. Mori, A. Fujishima, Electron Transfer from Diamond Electrodes to
Heme Peptide and Peroxidase. Anal. Chem., 72 (2000) 29192924.
V. Grigorenko, T. Chubar, Y. Kapeliuch, T. Brchers, F. Spener, A. Egorov, New
Approaches for Functional Expression of Recombinant Horseradish Peroxidase C in
Escherichia coli. Biocatalysis Biotransform. 17 (1999) 359379.
R. Slojkowska, M. Jurkiewicz-Herbich, Adsorption Study of Amino Acids on a
Polycrystalline Gold Electrode. Colloids Surf. A., 178 (2001) 325336.
M. Vreeke, R. Maidan, A. Heller, Hydrogen Peroxide and -Nicotinamide Adenine
Dinucleotide Sensing Amperometric Electrodes Based on Electrical Connection of
Horseradish Peroxidase Redox Centers to Electrodes Through a Three-Dimensional
Electron Relaying Polymer Network. Anal. Chem., 64 (1992) 30843090.
H. Mei, K. Wang, N. Peffer, G. Weatherly, D. S. Cohen, M. Miller, G. Pielak, B.
Durham, F. Millett, Role of Conformational Gating in Intracomplex Electron Transfer from Cytochrome c to the Radical Cation in Cytochrome c Peroxidase. Biochemistry, 38 (1999) 68466854.
E. I. Iwuoha, I. Leister, E. Miland, M. R. Smyth, C. O. Fagain, Reactivities of
Organic-Phase Biosensors. 1. Enhancement of the Sensitivity and Stability of
Amperometric Peroxidase Biosensors Using Chemically Modied Enzymes. Anal.
Chem., 69 (1997) 16741681.
K. Yamamoto, T. Ohgaru, M. Torimura, H. Kinoshita, K. Kano, T. Ikeda, Highlysensitive Flow Injection Determination of Hydrogen Peroxide with a PeroxidaseImmobilized Electrode and its Application to Clinical Chemistry. Anal. Chim. Acta,
406 (2000) 201207.
W. Li, Z. Wang, C. Sun, M. Xian, M. Zhao, Fabrication of Multilayer Films Containing Horseradish Peroxidase and Polycation-bearing Os Complex by Means of
Electrostatic Layer-by-layer Adsorption and its Application as a Hydrogen Peroxide
Sensor. Anal. Chim. Acta, 418 (2000) 225232.
S. A. Kane, E. I. Iwuoha, M. R. Smyth, Development of a Sol-gel Based Amperometric Biosensor for the Determination of Phenolics. Analyst (Cambridge, U. K.),
123 (1998) 20012006.
A. Lindgren, J. Emnus, T. Ruzgas, L. Gorton, G. Marko-Varga, Amperometric
Detection of Phenols Using Peroxidase-Modied Graphite Electrodes. Anal. Chim.
Acta, 347 (1997) 5162.
F.-D. Munteanu, A. Lindgren, J. Emneus, L. Gorton, T. Ruzgas, E. Cseregi, A.
Ciucu, R. B. van Huystee, I. G. Gazaryan, L. M. Lagrimini, Bioelectrochemical
Monitoring of Phenols and Aromatic Amines in Flow Injection Using Novel Plant
Peroxidases. Anal. Chem., 70 (1998) 25962600.

Electrochemistry of Peroxidases
73.
74.
75.

76.

77.

78.
79.

80.
81.
82.

83.

84.

85.

86.

87.

88.

253

T. M. Park, E. I. Iwuoha, M. R. Smyth, Development of a Sol-gel Enzyme Inhibitionbased Amperometric Biosensor for Cyanide. Electroanalysis, 9 (1997) 1120 1123.
M. Espinosa, P. Atanasov, E. Wilkins, Development of a Disposable Organophosphate Biosensor. Electroanalysis, 11 (1999) 10551062.
A. Narvez, G. Surez, I. C. Popescu, I. Katakis, E. Domnguez, Reagentless Biosensors based on Self-deposited Redox Polyelectrolyte-oxidoreductases Architectures.
Biosens. Bioelectron., 15 (2000) 4352.
L. Campanella, G. Favero, L. Persi, M. Tomassetti, New Biosensor for Superoxide
Radical Used to Evidence Molecules of Biomedical and Pharmaceutical Interest
Having Radical Scavenging Properties. J. Pharm. Biomed. Anal., 23 (2000) 6976.
F. Patolsky, M. Zayats, E. Katz, I. Willner, Precipitation of an Insoluble Product on
Enzyme Monolayer Electrodes for Biosensor Applications: Characterization by
Faradaic Impedance Spectroscopy, Cyclic Voltammetry, and Microgravimetric
Quartz Crystal Microbalance Analysis. Anal. Chem., 71 (1999) 31713180.
E. Casero, M. Darder, F. Pariente, E. Lorenzo, Peroxidase Enzyme Electrodes as
Nitric Oxide Biosensors. Anal. Chim. Acta, 403 (2000) 19.
P. N. Bartlett, P. R. Birkin, J. H. Wand, F. Palmisano, G. Benedetto, An Enzyme
Switch Employing Direct Electrochemical Communication between Horseradish
Peroxidase and a Poly(aniline) Film. Anal. Chem., 70 (1998) 36853694.
R. W. Keay, C. J. McNeil, Separation-free Electrochemical Immunosensor for Rapid
Determination of Atrazine. Biosens. Bioelectron., 13 (1998) 963970.
A. Warsinke, A. Benkert, F. W. Scheller, Electrochemical Immunoassays. Fresenius J. Anal. Chem., 366 (2000) 622634.
C. N. Campbell, T. Lumley-Woodyear, A. Heller, Towards Immunoassay in Whole
Blood: Separationless Sandwich-type Electrochemical Immunoassay Based on insitu Generation of the Substrate of the Labeling Enzyme. Fresenius J. Anal. Chem.,
364 (1999) 165169.
T. Lumjey-Woodyear, C. N. Campbell, E. Freeman, A. Freeman, G. Georgiou, A.
Heller, Rapid Amperometric Verication of PCR Amplication of DNA. Anal.
Chem., 71 (1999) 535538.
A. Morimoto, M. Tanaka, S. Takahashi, K. Ishimori, H. Hori, I. Morishima, Detection of a Tryptophan Radical as an Intermediate Species in the Reaction of Horseradish Peroxidase Mutant (Phe-221 3 Trp) and Hydrogen Peroxide. J. Biol. Chem.,
273 (1998) 1475314760.
S. Nagano, M. Tanaka, K. Ishimori, Y. Watanabe, I. Morishima, Catalytic Roles of
the Distal Site Asparagine-Histidine Couple in Peroxidases. Biochemistry, 35 (1996)
1425114258.
S. Nagano, M. Tanaka, Y. Watanabe, I. Morishima, Putative Hydrogen Bond Network in the Heme Distal Site of Horseradish Peroxidase. Biochem. Biophys. Res.
Commun., 207 (1995) 417423.
S. S. Razola, E. Aktas, J.-C. Vire, J.-M. Kauffmann, Reagentless Enzyme Electrode
Based on Phenothiazine Mediation of Horseradish Peroxidase for Subnanomolar
Hydrogen Peroxide Determination. Analyst, 125 (2000) 7985.
J. R. Cherry, M. H. Lamsa, P. Schneider, J. Vind, A. Svendsen, A. Jones, A. H. Pedersen, Directed Evolution of a Fungal Peroxidase. Nature Biotechnol., 17 (1999)
379384.

254
89.
90.

91.

92.

93.
94.
95.
96.

Ruzgas et al.
A. Ifand, P. Tafelmeyer, C. Saudan, K. Johansson, Directed Molecular Evolution
of Cytochrome c Peroxidase. Biochemistry, 39 (2000) 1079010798.
B. Morawski, Z. Lin, P. Cirino, H. Joo, G. Bandara, F. H. Arnold, Functional Expression of Horseradish Peroxidase in Saccharomyces cerevisiae and Pichia pastoris.
Protein Eng., 13 (2000) 377384.
J. A. Zahn, D. M. Arciero, A. B. Hooper, J. R. Coats, A. A. DiSpirito, Cytochrome
c Peroxidase from Methylococcus capsulatus Bath. Arch. Microbiol, 168 (1997)
362372.
H. Lopes, G. W. Pettigrew, I. Moura, J. J. G. Moura, Electrochemical Study on
Cytochrome c Peroxidase from Paracoccus denitricans: a Shifting Pattern of Structural and Thermodynamic Properties as the Enzyme Is Activated. JBIC, 3 (1998)
632642.
J. Kulys, R. Vidziunaite, The Determination of Halides and Pseudohalides by the
Vanadium Haloperoxidase based Biosensor. Anal. Lett., 31 (1998) 26072623.
J. Kulys, R. Vidziunaite, Electrocatalysis by Vanadium Haloperoxidase. J. Electroanal. Chem., 455 (1998) 161167.
C. Lehmann, U. Wollenberger, R. Brigelius-Flohe, F. W. Scheller, Bioelectrocatalysis by a Selenoenzyme. J. Electroanal. Chem., 455 (1998) 259263.
P. Kraulis, MOLSCRIPT: A Program to Produce Both Detailed and Schematic Plots
of Protein Structures. J. Appl. Cryst., 24 (1991) 946950.

8
Mechanisms and Kinetics
of Neurotransmission Measured
in Brain Slices with
Cyclic Voltammetry
Joshua D. Joseph and R. Mark Wightman
University of North Carolina at Chapel Hill, Chapel Hill, North Carolina

I.

INTRODUCTION

Dopamine (DA) is an important chemical messenger in the central nervous system. Understanding the mechanisms and kinetics of DA neurotransmission is
necessary to discern the role of DA in many biological functions, including locomotion, motivation, pathologies such as Parkinson disease, and the use of drugs
of abuse. The investigation of DA neurotransmission poses challenges that are
common in analytical chemistry: selectivity and sensitivity. In addition, biocompatible sensors that are both rapid and small are needed. Background-subtracted
cyclic voltammetry at carbon-ber microelectrodes represents an excellent tool to
address these challenges in brain slices.
To understand the challenges associated with these measurements, one
must consider the environment in which DA functions [1]. Neurons are cells in
the brain that gather, process, and relay information (Figure 1). A typical neuron
consists of a cell body with two types of projections: dendrites, which serve as
inputs for information, and the axon, which serves as an output device. Information is relayed within a neuron in the form of electrical potentials across the cell
membrane. Normally, the potential of the inside of the neuron is negative with
respect to the outside due to an ionic concentration gradient across the cell mem255

Joseph and Wightman

AT

256

Figure 1 Anatomy of a neuron.

brane. When the local potential of a portion of the cell membrane is altered, various ion channels can open because their conductivity is voltage dependent. This
causes the membrane to depolarize. A rapid voltage change, called an action
potential, propagates down the axon to the presynaptic terminals, the end structure of a neuron. At the terminal, the voltage change causes an inux of Ca2+ that
leads to the release of a neurotransmitter, such as DA. A neurotransmitter is a
chemical that neurons use to relay information.
Within the DA neuron, DA is synthesized from its precursor, tyrosine, and
packaged into vesicles, small spherical cellular structures utilized for neurotransmitter storage. The arrival of an action potential leads to a process termed exocytosis [2]. For this process, Ca2+ inux causes the fusion of the DA-containing
vesicles with the cell membrane within the synapse, the space between consecutive neurons. A fusion pore forms initially and then opens further to release the
contents of the vesicle into the extracellular uid. DA can then diffuse across the
synapse (1100 nm) and interact with receptors on the membrane of either

Mechanisms and Kinetics of Neurotransmission

257

the presynaptic or postsynaptic neurons. At this point, the released DA can meet
three fates: uptake back into the releasing presynaptic cell, metabolism, or diffusion to a remote location. Uptake by the dopamine transporter (DAT) on the
presynaptic neuron is the predominant fate and occurs on a millisecond time scale
so that the cell can be reset for the next message. After uptake into the releasing cell, DA may be repackaged into vesicles for re-release, or metabolized.
In our laboratory, we use intact brain slices (thin slabs of brain tissue) to
study DA neurotransmission. Approximately 400 m thick slices of brain tissue
containing the desired dopaminergic region are prepared. Some of the major
dopaminergic regions in the brain include the caudate-putamen (CP), prefrontal
cortex, anterior cingulate cortex, nucleus accumbens (NAc), and the olfactory
tubercle [1]. The slices are bathed with an articial cerebral-spinal uid that is
preheated to 32C and saturated with 95% O2, 5% CO2. The slices retain structural and metabolic integrity and are viable for up to 10 hours after the animal is
sacriced. The use of brain slices to investigate neurotransmission has many
advantages. The brain slice preparation allows physiological and pharmacological manipulation of the neuron terminals while maintaining system integrity
i.e., all the physiological mechanisms are still functioning.
Common pharmacological manipulations involve the application of antagonists and agonists. Antagonists act on receptors by binding directly to the receptor and preventing binding by other molecules. They in no way directly alter the
function of the receptor. In contrast, agonists bind to and activate the receptor in
some fashion by changing the function of the receptor. Agonists may directly or
indirectly bring about this effect and often mimic the action of the naturally
occurring substance. Because drugs are applied directly into the perfusion buffer,
the use of drugs that are not able to cross the blood-brain barrier is possible. Also,
the drug concentration applied to the neurons is easily quantiable, facilitating the
construction of dose-response curves, something that is not true in vivo. Visual
placement of the stimulating and working electrodes ensures that the experiment
is performed in the desired region.
The slice is electrically stimulated with an enamel-coated bipolar wire electrode causing action potentials that evoke DA release. In the slice preparation, the
stimulation is applied directly at the neuron terminals. This is in contrast to an in
vivo experiment, where the stimulation is performed at the cell body of the neuron and the DA release is monitored remotely at the presynaptic terminals. DA
concentration in the extracellular uid rises and quickly returns to baseline at the
cessation of the stimulation [3]. Fast scan cyclic voltammetry (FSCV) at carbonber microelectrodes is used to detect the resulting concentration changes in the
extracellular uid. This analytical technique provides a method for the determination of uptake kinetics in intact brain tissue. Thus, the secretion and subsequent
clearance of DA in the tissue is observed in real time.

258

Joseph and Wightman

II.

ANALYTICAL CHALLENGES

A.

Size

Ideally, to monitor the chemical dynamics of this process, one would like to
directly measure the concentration prole of DA within the synapse. Although this
is not currently possible, it is important to minimize the size of the probe in order
to get as close to the synapse as possible and to minimize the disruption of the tissue being investigated. To minimize the size of the probe, microelectrodes have
been fabricated from a variety of carbon bers (630 m). In addition to their small
size, the use of carbon bers is advantageous because carbon tends to be more compatible with biological samples than other more commonly used electrode materials. These electrodes, although not in the synapse, are able to monitor the concentration of DA in the extracellular uid surrounding the neurons.
B.

Selectivity

Complicating the measurement of electroactive neurotransmitters are numerous


endogenous compounds in the brain. An inventory of easily oxidized compounds
in the extracellular uid is important to the electrochemical study of neurotransmission because their electrochemistry may interfere with the measurement of
DA or other target molecules. Using a push-pull cannula coupled to liquid chromatography with electrochemical detection (LC-EC), several easily oxidized
compounds in the extracellular uid of the CP, a well-studied dopamine containing region, were identied, including ascorbate, uric acid, homovanillic acid, and
DOPAC [4]. Further, using microdialysis it is possible to measure the low basal
level of DA (~30 nM) [5] although the stimulated release typically studied in
brain slices is in the micromolar range [6,7]. These experiments have demonstrated that the electrode must provide selectivity against ascorbate and DOPAC
that are present at much greater concentrations than DA. Other previously mentioned electroactive compounds are oxidized at more positive potentials and thus
can be discriminated electrochemically.
The high concentrations of ascorbate and DOPAC are avoided by the application of a small lm of Naon, a peruorinated ion-exchange membrane, to the
electroactive surface of the microelectrode [8,9]. Both ascorbate and DOPAC are
anions at physiological pH (7.4) and are mostly excluded by this cation-exchange
material. DA, however, is positively charged and can accumulate in this membrane, as rst demonstrated by Ralph Adams group [8,9]. Naon imparts a 200fold greater sensitivity for DA than for DOPAC or ascorbate. Although this lm
decreases the response time of the electrode, it is sufciently thin that DA can
penetrate on a sub-second time scale [10].
The primary electrochemical technique employed in these investigations is
background-subtracted fast-scan cyclic voltammetry. For DA, the waveform is

Mechanisms and Kinetics of Neurotransmission

259

scanned from 400 mV to 1000 mV and back at 300 V/s and the current at the
peak of oxidation is monitored. The background charging current at carbon ber
microelectrodes is remarkably stable, and may be digitally subtracted. Voltammograms recorded at 300 V/s take less than 10 ms to acquire and typically are
repeated every 100 ms. At fast scan rates, the voltammogram provides a unique
ngerprint for identication of DA because the shapes of the voltammogram
are kinetically controlled, and different from those of DOPAC and ascorbate, the
interferents previously discussed. This discrimination increases the selectivity for
DA over ascorbate to approximately 1000-fold [11]. Integrating the current over
the oxidation peak and plotting the integrated value as a discrete point versus time
allows the temporal prole of DA to be monitored. The electrochemical discrimination afforded by FSCV allows for the discrimination of target molecules from
interferents (such as DOPAC and ascorbate), simultaneous monitoring of additional molecules of interest, and the detection and identication of additional neurotransmitters such as serotonin (5-hydroxytryptamine) and norepinephrine when
this technique is applied to different regions in the brain.
It has been shown that carbon ber microelectrodes are responsive to pH as
well as redox-active molecules [12]. Changes in solution pH lead to shifts in the
background voltammetric features that arise from oxygen-containing functional
groups on the carbon ber electroactive surface [13]. These voltammetric
responses occur over the same potentials as DA oxidation and reduction. Changes
in pH are seen in slices, usually accompanying large stimulations [12]. Multiple
resolutions to this problem have been used to remove the pH interference from the
temporal prole of DA concentration changes. For example, Ro4-1284, a drug
that depletes catecholamines (such as DA and norepinephrine) from vesicles, was
used to eliminate DA release. The electrical response from pH changes were still
present, so a data le of only the pH interferences could be subtracted from the
original data le providing a temporal prole of DA concentration free from interferences [12]. Recently, the utility of three-dimensional false-color imaging was
demonstrated for examining all of the current-potential-time data simultaneously
[14]. By examining the full-color images, it was clear that some potentials reect
both pH and DA, and others reect only pH, because pH voltammetric shifts
occur over a much broader potential range than DA. A potential does not exist that
is sensitive to DA but not to pH. Using these images, it is easy to extract the temporal pH data and subtract it from the mixed signal. An alternative to mathematically subtracting the pH interference is to create a sensor that is pH-insensitive.
Removing the surface oxides on the surface of the carbon ber microelectrode
[15] signicantly reduces the pH sensitivity of the carbon ber microelectrode
[16] (Figure 2). The removal of surface oxides was accomplished by polishing the
electroactive surface in a reduced oxygen environment (degassed cyclohexane).
The simultaneous detection of multiple analytes is possible providing that both
molecules react electrochemically at sufciently separated potentials. This

260

Joseph and Wightman

Figure 2 Cyclic voltammograms showing the effects of cyclohexane at pH 7.4. Before


cyclohexane polishing (left). After cyclohexane polishing (right). (a,b) Background; (c,d)
[dotted line shown in (c) represents simulated voltammogram] 10 M DA; (e,f) acidic pH
change (pH = 0.4 units). Scan rate for all was 300 V/s. (Reproduced from Analytical
Chemistry with permission [16]) Electrode carbon ber radius was 6 M.

method was used to measure changes in DA concentration and extracellular


decreases in O2 concentration evoked by brief electrochemical stimulations. The
electrode potential was rst scanned sufciently positive to oxidize DA, and then
subsequently altered to a value where O2 reduction occurs [17,18]
DA is not the only electroactive neurotransmitter that may be monitored
with microelectrodes and FSCV. It is also possible to detect norepinephrine
[19,20], following similar electrochemical procedures in a norepinephrine-containing anatomical region of the brain. By altering the waveform to maximize
oxidative current and minimize fouling, it is also possible to study serotonin
[21,22].
Less successful in our hands were attempts to couple proteins to the ber to
make sensors for other nonelectroactive neurotransmitters. Enzyme-modied
amperometric sensors for choline and acetylcholine were constructed and performed with high sensitivity and selectivity in the ow analysis system. However,
in brain slices, both electrodes were found to measure acetylcholine [23,24]. This
was perhaps due to the adsorption of acetylcholinesterase onto the electrode from
the brain tissue, where it is highly abundant.

Mechanisms and Kinetics of Neurotransmission

261

The investigation of glutamate via enzyme-modied carbon-ber microelectrodes [2528] has been performed in vivo [25,28]. Glutamate is a nonelectroactive neurotransmitter that is important in the mammalian central nervous
system [29]. The detection of glutamate is usually performed using either glutamate dehydrogenase, where the sensor monitors the electrochemistry of the
NAD/NADH redox couple [26,27], or glutamate oxidase, where the production
of hydrogen peroxide is monitored [28]. The enzymes are attached to the surface
either via avidin-biotin technology [26,27] or via a cross-linkable redox polymer
[25]. These electrodes are sensitive to glutamate on the micromolar range and are
able to monitor changes on a subsecond time scale.
C.

Sensitivity and Speed

In the CP, the terminals of dopaminergic neurons are known to be an average of


4 m apart [30]. If the terminals release their neurotransmitters simultaneously,
as is the case with local electrical stimulation, the concentration in the extracellular uid between neurons would become homogeneous in 3 ms as a result of diffusion [31]. The ideal probe to monitor this situation would have a total sampling
time longer than 3 ms and a sampling frequency high enough to accurately
describe the rapid rate of uptake. In this way, the measured DA disappearance rate
is similar to the rate of uptake by the DAT. The fast-scan cyclic voltammetry that
we use approaches such an ideal by lasting 10 ms and scanning once every 100
ms.
A simple way to improve electrode sensitivity is to increase the electroactive area by increasing the size of the carbon ber. While attractive in terms of
sensitivity, for reasons previously stated, this causes a loss of anatomical specicity. While even large electrodes are not currently sufciently sensitive to monitor basal DA in vivo, microelectrodes are able to monitor the dynamics of stimulated DA release and disappearance both in vivo and in slices, both with content
typically in the M range.
An alternative to increasing the microdisk surface area by using a larger
carbon ber is to use the cylinder electrode geometry. The cylinder electrode has
a carbon ber extending 10150 M beyond the tip of glass capillary and epoxy
[32]. This geometry is advantageous because it increases the sensitivity without
increasing damage to the tissue because the electrode tip is not larger. These electrodes were found to increase the signal-to-noise ratio by a factor of three after in
vivo use. This improvement is attractive, but these electrodes do not currently
allow the kinetic modeling of the release and uptake of data because of slow DA
adsorption/ desorption kinetics at these surfaces. The model used to describe
uptake assumes concentration changes monitored by the probe are uniform
throughout the sensing volume. DA release has been shown to be heterogeneous

262

Joseph and Wightman

over distances similar to the length of the cylinders electroactive area [33], providing an additional reason cylinder electrodes are currently considered unsuitable for kinetic modeling.
Recently, changes in our data acquisition system have improved signal-tonoise ratios, thereby increasing sensitivity [34]. Minimization of noise is critical
when measuring the low concentrations of neurotransmitters. The new system is
based on a high-speed, 16-bit analog to digital/digital to analog acquisition board
where the computer not only collects the current data but also generates the waveform, thereby minimizing timing errors. Noise from frequency drift of line frequency is eliminated by the use of a phase-locked loop that allows data to be collected exactly in phase with line frequency. Although direct comparison of
current versus time data indicated a mere two-fold increase in signal-to-noise
ratio, the three-dimensional false color plots clearly demonstrated improved quality of data [34].
It has been reported that by electrochemically pretreating a carbon ber
microelectrode, the sensitivity to DA could be greatly increased [3538]. This was
shown to be the result of an oxide lm forming on the surface of the electrode as
well as cracking of the surface [37]. This treatment generates more surface area and
increases the adsorption of DA to the electrode. This increase in adsorption
improves the sensitivity but greatly decreases the time response of the electrode.
This is extremely unattractive for modeling the DA kinetics, which occur on a millisecond time scale, thus making these electrodes less suitable for kinetic analysis.

III.

NEUROTRANSMISSION

A.

Release

Because of its small size, selectivity, time response, and sensitivity, the carbon
ber microelectrode coupled with fast-scan cyclic voltammetry currently represents the most ideal sensor for the measure of kinetics and mechanisms of DA
neurotransmission. Release and uptake sites (neuron terminals) are depicted in
Figure 3 along with a typical example of DA release and uptake data monitored
with a carbon-ber microelectrode using FSCV. These sites are generically
described as a cartoon merely to give the reader an idea of the relative size of the
electrode versus the neuron terminals (release and uptake sites).
Before DA was monitored in brain tissue using voltammetry, it was not
possible to assign rises and decreases in DA tissue content or dialysis efuent as
due to metabolism, uptake, or release [3941]. However, the high time resolution
allows the instantaneous rise in DA accompanying electrical stimulation to be
attributed to release. Using various pharmacological agents, it is possible to
determine that the observed DA release is physiologically relevanti.e., release
is sensitive to its ionic environment, inhibition of vesicular storage, and to the

Mechanisms and Kinetics of Neurotransmission

263

Figure 3 Schematic of the scale of release and uptake sites versus the microelectrode.
Below, representative data showing electrically stimulated release and subsequent uptake
of DA. Boxes represent data points between which stimulation occurred. Data points are
100 ms apart; amplitude of the DA response is approximately 2 M.

electrical stimulus parameters [17]. Because DA release is exocytotic, it is


directly related to the external Ca2+ concentration, and release is abolished by the
removal of the external Ca2+. Depletion of vesicular DA also results in the loss of
all DA release. Lastly, the application of tetrodotoxin (TTX), a sodium channel
blocker, completely abolishes electrically evoked DA release, indicating the electrical stimulus evokes release by generating action potentials. The major presynaptic mechanisms that effect DA release are depicted in Figure 4.
B.

Uptake

The classic method for measuring neurotransmitter uptake kinetics was to incubate
synaptosomes (dissociated, pinched off neuron terminals) or chopped brain tissue
in different concentrations of radioactive substrate for varying periods of time. The

264

Joseph and Wightman

Figure 4 Schematic of presynaptic mechanisms affecting the release of dopamine (DA).

accumulation of the substrate in the tissue was measured and a classic MichaelisMenten plot was constructed. The apparent values of the kinetic parameters Vmax
and Km for various experimental conditions [42] were then determined. Unfortunately, these experiments present several problems. Although the synaptosome
preparation allows measurements to be made without the effects of diffusion barriers, synaptosomes lack the cellular compartments found in true physiological
systems, and the DA concentrations employed may differ from those found physiologically in brain tissue. In chopped brain tissue, diffusion, both into and through
the tissue, coupled with the uptake process, greatly distorts the results. Diffusionuptake coupling causes the DA concentration within the tissue to be nonuniform
and lower than in the surrounding uid as a result of the many uptake sites DA
encounters as it diffuses through the tissue. Indeed, the concentration is lowest at
the center of the tissue [43]. The nonuniform concentration throughout the tissue
and the fact that the concentration of DA at the uptake site is not known causes an
inaccurate estimation of the kinetic values [44].
We have used electrical stimulation in the intact brain of an anesthetized rat
and in rat and mouse brain slices to release endogenous, i.e., physiologically syn-

Mechanisms and Kinetics of Neurotransmission

265

thesized, DA [6,45]. Electrical stimulation of endogenous DA decreases many of


the previously discussed concerns. Concerns over nonphysiological concentrations are greatly reduced and there is no diffusion because DA levels are elevated
more uniformly. Thus, as previously noted, the ability to make spatially and temporally resolved measurements is a necessity for the meaningful evaluation of
extracellular neurotransmitter dynamics.
There are many pieces of evidence that indicate the rapid disappearance of
DA following electrical stimulation observed in brain slices is due to uptake. The
rate of disappearance of DA is far too quick in comparison to diffusion. Also,
when the effect of diffusion through the thin Naon lm is removed, no time lag
between stimulation and DA uptake is observed in the disappearance as would be
expected for diffusion [43]. Additionally, well-studied DA uptake inhibitors
including bupropion, cocaine, and nomifensine [46,47], both in vivo and in brain
slices, affect the rate of clearance. Lastly, a strain of mice was created that had the
gene for DA transporter deleted [48]. In brain slices prepared from these uptakedecient animals, stimulated DA remained in the extracellular uid on a time
scale consistent with that predicted for diffusion.
Another mechanism that could contribute to the rapid disappearance of DA
is metabolism by catechol O-methyltransferase (COMT) or monoamine oxidase
(MAO). If this were the case, then the addition of a metabolic inhibitor would
cause a decrease in the disappearance rate. Neither the addition of an MAO
inhibitor, pargyline, nor a COMT inhibitor had a signicant effect on the disappearance rate of DA in vivo [49]. Taken together these data indicate that the disappearance of DA observed with FSCV and microelectrodes is due solely to
uptake of DA by the DA transporter. The presynaptic mechanisms involved in
DA uptake are depicted in Figure 5.
DA uptake is a saturable process that is described by Michaelis-Menten
kinetics:
V = {d[DA]/dt}uptake = Vmax/({Km/[DA]+1})

(1)

Rather than converting our data to traditional Michaelis-Menten plots, we have


locally written software that numerically integrates the above equation based on
estimates of Vmax, Km, and [DA]p (the concentration of DA released into the
extracellular uid by each electrical stimulation). This software produces a simulated curve that is compared with the raw data. Typically the value for Km is
taken from a literature value calculated from synaptosome studies. The Vmax is
determined from the maximum slope of the descending side of the DA concentration prole. The [DA]p is optimized based on the best t obtained when compared to the actual data. Naon slows the diffusion of DA to the electrode, which
is accounted for by the convolution of the simulation with the effects of diffusion
through the Naon lm [50]. Alternatively, the actual data could be deconvoluted

266

Joseph and Wightman

Figure 5 Schematic of presynaptic mechanisms involved with the uptake of dopamine


(DA). MAO, monoamine oxidase (located on the surface of the mitochondrion); VMAT,
vesicular monoamine transporter DAT dopamine transporter. Cellular machinery has been
enlarged to show detail.

to remove the slow diffusion caused by Naon; however, this approach introduces noise artifacts.

IV.

INVESTIGATIONS USING MICROELECTRODES WITH FSCV

A.

O2 Levels Following Stimulations

As previously discussed, the simultaneous measurement of O2 levels and DA


concentrations is possible by scanning the electrode rst sufciently positive to
oxidize DA, and subsequently negative enough to reduce O2 [17]. The investigation of O2 consumption is important because oxygen is necessary for normal brain
function and is an index of energy use. The process of neurotransmission consumes energy and thus there should be a link between it and O2 levels. With an
electrical stimulation typical of studies in brain slices (2 pulses at 10 Hz, 480 A)
an immediate, short-lived rise in DA concentration is observed that quickly

Mechanisms and Kinetics of Neurotransmission

267

Figure 6 [Left] Typical responses to two-pulse stimulation for both dopamine (DA) and
O2. The short bars indicate the stimulus. Two stimulations are shown; 180 s was allowed
between each stimulus. (A) Current for DA (average current between 500 and 700 mV);
(B) average current from 100 to 100 mV, which is a region of only background; (C) current for O2 (average current from 1100 to 1295 mV). [Right] Background-corrected
cyclic voltammograms obtained from data for rst stimulation in data on left. All potentials
were measured versus a saturated calomel electrode at a scan rate of 420 V/s. The scan originated at 0 V and was initially to positive potentials. The upper cyclic voltammogram was
obtained by subtracting the average of ve scans immediately following the rst stimulus
pulse from the average of ve scans during the baseline preceding the stimulus pulse. The
lower cyclic voltammogram was obtained by subtracting the average of ve scans 13 s after
the stimulus was applied from the average of ve scans obtained during the prestimulus
baseline. (Reproduced from Neuroscience with permission [17].)

returns to baseline. A decrease in O2 levels occurs that is delayed 300700 ms


after stimulation and approaches its original value much more slowly (Figure 6).
The release of DA and consumption of O2 are Ca2+ sensitive, but the only substance that affects the DA system and also causes a change in O2 consumption is
sulpiride, a pharmacological antagonist of specic DA receptors. This may be the
result of receptor modulation of Ca2+ channels, and is the only example of a
receptor-mediated alteration of O2 consumption. The most striking result in this
investigation was the discovery that immediate oxygen consumption was not
required for neurotransmission. This was demonstrated by both the delay between

268

Joseph and Wightman

stimulation and O2 consumption, and the existence of neurotransmission without


oxygen consumption when cyanide is applied [17]. This implies that although
Ca2+ entry triggers oxygen consumption, and therefore oxidative energy production, the energy is not used to support ongoing neurotransmission.
B.

Investigation of Autoreceptors

Autoreceptors are found at the terminals of presynaptic neurons and provide feedback to these neurons. An autoreceptor responds to the neurotransmitter released
by that neuron and can greatly inuence neurotransmission. They are capable of
controlling the concentration prole and lifetime of neurotransmitters in the
synapse by modulating release [51,52] or uptake [5356]. Regulation of release
can be broken into two subcategories: synthesis regulation and secretion regulation. This down-regulation of DA concentration within the extracellular uid
(uid that lls the space between neurons) caused by autoreceptor activation is
thought to prevent an over-stimulation of the postsynaptic receptors. Autoreceptors typically have a much higher afnity for the neurotransmitter than the postsynaptic receptors. The multiple functionalities of receptors are thought to be a
result of localization and intracellular coupling.
DA receptors have been divided into 2 classes: D1-like and D2-like. These
are separated into ve subclasses with D1-like including D1, D4, and D5 receptors
and D2-like encompassing D2 and D3 receptors. DA autoreceptors are classied
as D2-like receptors, and both D2 and D3 receptors have been found presynaptically and are also proposed to function as autoreceptors [51,57,58]. D1-like receptors do not seem to play a role as presynaptic receptors. All ve subclasses are
reported to function as postsynaptic receptors.
Brain slices provide an excellent environment in which to study presynaptic effects. The slicing procedure conserves most of the structural integrity within
the slice but removes all extrinsic interaction, i.e., postsynaptic effects. This
allows all alteration of neurotransmission to be attributed solely to presynaptic
effects. This is especially important when studying autoreceptors, as D2 and D3
receptors are found both as presynaptic autoreceptors and post-synaptic receptors.
Therefore any agonist that stimulates these receptors may have both presynaptic
and postsynaptic effects in the intact animal.
Autoreceptors have been studied in slices in several groups. Palij et al. [59]
reported on the autoreceptor-mediated regulation of DA release in slices. The
function of autoreceptors was probed by showing DA release was inhibited by
quinpirole and N,N-dipropyl-5,6-ADTN, two selective D2 receptor agonists. A
D1 agonist, SKF 38393, had no effect on release, as was expected because D1
receptors are found exclusively on postsynaptic cells. The activation of autoreceptors was completely antagonized with potencies close to those published for
post-synaptic D2 receptors by haloperidol and metoclopramide, well-studied

Mechanisms and Kinetics of Neurotransmission

269

autoreceptor antagonists [69]. This study demonstrated that autoreceptors had


direct control of DA release and were identical to postsynaptic receptors previously investigated in binding studies.
In the Kruk laboratory, the effect of autoreceptors on stimulation trains was
investigated to probe the effect of endogenous DA on DA neurotransmission [60].
Trains of four or ten electrical pulses were applied at varying frequencies (0.2 Hz,
1 Hz, or 5 Hz). The train of electrical pulses at 0.2 Hz caused completely separated peaks of DA; however, the second through fourth pulses were all smaller
than the rst, but similar in amplitude to each other. The application of an autoreceptor antagonist, metoclopramide, caused no change in the single pulse, or train
stimulated at 0.2 Hz. This indicated there is no autoregulation of single-pulse
stimulations and the lower amplitude of the second through fourth pulses were the
result of a mechanism other than autoregulation. If there was an autoreceptor suppression of the initial pulse or subsequent pulses in the 0.2 Hz train, the application of an autoreceptor antagonist would have increased the amplitude of the DA
release. Trains delivered at higher frequencies resulted in either an initial peak
with an extended shoulder on the disappearance section of the curve, or monophasic pulse that was larger than that elicited by a single pulse. Metoclopramide
caused an increase in release seen with a 1 Hz or 5 Hz stimulation. In the presence of nomifensine, an uptake inhibitor, metoclopramide caused an increase in
release for all stimulations. This study demonstrated that presynaptic autoregulation of DA neurotransmission occurs immediately after the rst stimulation; however, it is not the only mechanism responsible for the decrease in release seen in
subsequent stimulations.
In our laboratory, similar results were reported for agonist and antagonist
effects of autoreceptors [6]. Additionally, the amount of time required for a second single-pulse stimulation to recover completely to the initial pulses amplitude
was investigated. The data were t to a biexponential curve, and control data was
compared to that obtained in the presence of an autoreceptor antagonist, sulpiride
(Figure 7). The control data had a half-time of the slower exponential of approximately 17 seconds. This half-time was decreased to 5 seconds by sulpiride. This
again demonstrated that the decrease seen after the rst stimulation for trains is
only partially restored by autoreceptor antagonists. Multiple explanations for the
inability of autoreceptor antagonists to completely restore the release phenomenon have been suggested. It may be that the applied drugs are unable to compete
with the endogenously released DA, or another neurotransmitter released by the
stimulation inhibits the release of DA, although neither a GABA-antagonist nor
glutamate affected DA release. GABA and glutamate are two neurotransmitters
released by neurons near the DA neuron terminals that were investigated. It also
has been proposed that recruitment of vesicles for release may be too slow to
maintain release at higher stimulation frequencies [60].
These studies indicate a rapid, signicant, and long-lasting inhibition of DA

270

Joseph and Wightman

Figure 7 Recovery of amplitude of dopamine (DA) release in response to one-pulse


stimulations. A second pulse was given at 1, 5, 10, 30, or 60 s after the rst pulse. Inset:
Rate of recovery measured as the height of the second pulse relative to the rst pulse for
the control response (; n = 5) and the response in the presence of 2 M sulpiride (; n =
4). Solid lines in the inset are double exponentials. For data with sulpiride present, the relative percentage was 100 [1/2 exp (1.1t) + 1/2 exp (0.15t)]. For control data the relative percentage was 100 [1/2 exp (1.1t) + 1/2 exp (0.4t)]. (Reproduced from Journal of
Neurochemistry with permission [6].)

Mechanisms and Kinetics of Neurotransmission

271

release mediated by autoreceptors. When this information is combined with previously studied uptake data, the high concentration of DA in the synapse lasts for
only a brief period of time before uptake and autoreceptors reset the neuron in
preparation for the next stimulation.
C.

Effects of Pharmacological Agents on DA Neurotransmission

As previously stated, uptake inhibitors have been used to demonstrate that the disappearance of DA observed in brain slice is in fact due to uptake. Drugs can be
investigated to determine their mechanism of action and effect on uptake. They
also can be used as tools to investigate uptake itself.
Uptake inhibitors, such as cocaine and bupropion, have their action at the
DAT and were used to alter the rate of uptake in slices. Their actions as competitive uptake inhibitors were conrmed by their alteration of apparent Km values
for the rate of uptake of DA by the DAT, and lack of effect on [DA]p, and Vmax.
The Ki for each drug was calculated based on the apparent Km, and the values correspond well to those measured in synaptosomes [47]. In this way, these drugs
were investigated and their previously reported action on uptake was conrmed
with FSCV in slices.
D.

Uptake Rates in Multiple Regions

DA uptake has been investigated in several regions of the brain, including the CP
and the NAc [7]. It was found that there is a great degree of variability in uptake
from region to region; uptake in the CP was approximately twice that found in the
core of the NAc. In addition there was variability within each region. Uptake
within shell of the NAc was one third of that in the NAc core, in spite of the fact
that DA tissue concentrations were signicantly higher in the core than the shell
of the NAc [61]. Additionally, the rate of uptake in the CP was found to decrease
as the electrode was lowered further into the CP [7]. The difference in uptake rates
was attributed solely to the density of DATs found within each area since it has
been shown to be the same protein in each area [43].
E.

Uptake Investigated Through Physiological Manipulation

Pharmacological agents also have been used to probe the mechanism by which
DA transport occurs. It has been shown, in a simplied preparation, that the kinetics of the catecholamine transporters can be described Figure 8A [62,63]. This
scheme describes transport as a series of steps: the competition between binding
and dissociation of the substrate to the transporter on the extracellular side, a same
competition on the cytoplasmic side, and an equilibrium between the unoccupied
transporter facing either side.

272

Joseph and Wightman

Figure 8 (A) Kinetic scheme for a simple carrier adapted from Ref. 68 for the dopamine
transporter. The subscripts o and i represent the outside and inside, respectively, of a
dopamine (DA) neuron. (B) A possible mechanism for the cotransport of another species
(such as amphetamine). All symbols are as in (A) except that A corresponds to amphetamine. (Reproduced from Journal of Neurochemistry with permission [65]).

To investigate the dependence of transport on the ionic gradient, ouabain,


an inhibitor of the Na+, K+-ATPase, was used [65]. The Na+, K+-ATPase is the
protein responsible for maintaining the Na+/K+ gradient across the cell membrane
that establishes the membrane potential of the neuron. Experimentally, the application of ouabain altered both uptake and release. At low concentrations, Vmax
values were decreased with no changes in Km, at high concentrations, the change
in Vmax was accompanied by a large change in Km. Release was increased in the
presence ouabain, consistent with each stimulation causing a greater depolarization of the membrane as a result of the altered membrane potential of the neuron.
These observations were consistent with previous experiments preformed in
synaptosomes [66] and indicate that the rate-limiting step in DA uptake precedes
the binding of DA to the DAT.
Although the mathematics describing the chemical kinetics illustrated in
Figure 8A are complicated, they predict Michaelis-Menten behavior under zerotrans conditionsi.e., when [DA]i << [DA]o, where the subscript i indicates cytoplasmic concentration and the subscript o indicates the extracellular concentration of DA. As a result of this DA gradient, outward transport would be low
compared to inward transport. An inhibitor of the vesicular transporter, Ro41284, was used to disrupt DA storage, thus increasing cytoplasmic DA, in order
to test the assumption of zero-trans conditions [65]. While DA release was
decreased as the vesicle storage was disrupted, uptake was unchanged and no outward transport was observed. This was interpreted as a failure to increase [DA]i
sufciently to cause outward transport, and indicates the zero-trans condition is a
fair assumption under physiological concentration of DA.
It has been demonstrated that amphetamine causes reverse transport of
endogenous DA into the extracellular uid [67]. This is consistent with Figure

Mechanisms and Kinetics of Neurotransmission

273

8B, indicating the transporter takes up a second substrate, in this case amphetamine (A). The multiple substrates for the DAT result in a greater number of
inward-facing transporters due to the increased transport of molecules from the
extracellular uid into the cytoplasm. After amphetamine is applied, DA uptake
by the DAT is altered as Km is greatly increased. Additionally, DA levels no
longer return to baseline, and this elevation was conrmed to be DA by the release
and uptake plot of DA in the presence of amphetamine (Figure 9). This shows that
under normal conditions, DA levels return quickly to baseline and show no indication of reverse transport into the extracellular uid. However, when amphetamine is present, the DA level hangs up; indicating possible reverse transport of
DA by the DAT. Figure 8B is further conrmed by similar results when high concentrations of norepinephrine, and even DA are applied to act as an additional
substrate for the DAT [65].
These pharmacological manipulations of transport indicate a simple transporter model such as the one described in Figure 8A,B can describe the process
of DA uptake and that drug interaction with the DAT can be explained by a few
general principles. The process is predominantly unidirectional, as the inward

Figure 9 Effects of amphetamine (10 M) on release and uptake of DA. Measurements


were made 30 min after drug application. DA release was evoked in the mouse caudateputamen by a single electrical impulse. Measured concentrations were recorded at 100 ms
intervals. Concentrations were measured before drug application () and 30 min after 10
M amphetamine (). Solid lines represent simulations obtained with the best-t parameters: before drug, [DA]p = 9.0 M, Vmax = 3.5 M/s, Km = 0.16 M, and 30 min after 10
M amphetamine, [DA]p = 2.7 M, Vmax = 3.5 M/s, Km = 8.0 M. A Naon thickness of
225 nm was used in all measurements. (Reproduced from Journal of Neurochemistry with
permission [65].)

274

Joseph and Wightman

transport of DA is driven by the co-transport of Na+ along its concentration gradient. Michaelis-Menten kinetics are a fair assumption, as the zero-trans condition appears to hold. The mechanistic information about the DAT enables us to
clarify the role the DAT plays in the regulation of DA.

V.

CONCLUSIONS

The detection of electrically evoked DA overow by FSCV in brain slices provides an excellent paradigm in which to study the mechanisms and kinetics of DA
neurotransmission. The microelectrode with FSCV provides the size, selectivity,
time response, and sensitivity and represents the most ideal sensor for these measurements available at this time. This tool has enabled dynamic measurements of
complex phenomena that regulate both release and uptake of DA directly at its
site of action.

ACKNOWLEDGMENTS
We would like to acknowledge support from NIH (NS 15841) for funding this
research.

REFERENCES
1.
2.
3.
4.

5.

6.
7.

Cooper JR, Bloom FE, Roth RH (1996) The biochemical basis of neuropharmacology. New York: Oxford University Press.
Feldman RS, Meyer JS, Quenzer LF (1997) Principles of neuropsychopharmacology. Sunderland, Mass.: Sinauer Associates.
Kelly RS, Wightman RM (1987) Detection of dopamine overow and diffusion with
voltammetry in slices of rat brain. Brain Res 423: 7987.
Salamone JD, Hamby LS, Neill DB, Justice JB, Jr. (1984) Extracellular ascorbic acid
increases in striatum following systemic amphetamine. Pharmacol Biochem Behav
20: 609612.
Church WH, Justice JB, Jr., Byrd LD (1987) Extracellular dopamine in rat striatum
following uptake inhibition by cocaine, nomifensine and benztropine. Eur J Pharmacol 139: 345348.
Kennedy RT, Jones SR, Wightman RM (1992) Dynamic observation of dopamine
autoreceptor effects in rat striatal slices. J Neurochem 59: 449455.
Jones SR, Garris PA, Kilts CD, Wightman RM (1995) Comparison of dopamine
uptake in the basolateral amygdaloid nucleus, caudate-putamen, and nucleus accumbens of the rat. J Neurochem 64: 25812589.

Mechanisms and Kinetics of Neurotransmission


8.
9.

10.

11.
12.

13.
14.
15.
16.

17.

18.

19.

20.

21.

22.

23.

24.

275

Gerhardt GA, Oke AF, Nagy G, Moghaddam B, Adams RN (1984) Naon-coated


electrodes with high selectivity for CNS electrochemistry. Brain Res 290: 390395.
Nagy G, Gerhardt GA, Oke AF, Rice ME, Adams RN, Moore RB, Szentirmay MN,
Martin CP (1985) Ion-exchange and transport of neurotransmitters in Naon lms
on conventional and microelectrode surfaces. J Electroanal Chem 188: 8594.
Kristensen EW, Kuhr WG, Wightman RM (1987) Temporal characterization of peruorinated ion exchange coated microvoltammetric electrodes for in vivo use. Anal
Chem 59: 17521757.
Baur JE, Kristensen EW, May LJ, Wiedemann DJ, Wightman RM (1988) Fast-scan
voltammetry of biogenic amines. Anal Chem 60: 12681272.
Jones SR, Mickelson GE, Collins LB, Kawagoe KT, Wightman RM (1994) Interference by pH and Ca2+ ions during measurements of catecholamine release in slices
of rat amygdala with fast-scan cyclic voltammetry. J Neurosci Methods 52: 110.
Kawagoe KT, Garris PA, Wightman RM (1993) pH-dependent process at Naon-coated carbon-ber microelectrodes. J Electroanal Chem 359: 193207.
Michael D, Travis ER, Wightman RM (1998) Color images for fast-scan CV measurements in biological systems. Anal Chem 70: 586A592A.
Chen P, McCreery RL (1996) Control of electron transfer kinetics at glassy carbon
electrodes by specic surface modication. Anal Chem 68: 39583965.
Runnels P.L., Joseph J.D., Logman M.J., Wightman R.M. (1999) Effect of pH and
surface functionalities on the cyclic voltammetric responses of carbon-ber microelectrodes. Anal Chem 71: 27822789.
Kennedy RT, Jones SR, Wightman RM (1992) Simultaneous measurement of oxygen and dopamine: coupling of oxygen consumption and neurotransmission. Neuroscience 47: 603612.
Zimmerman JB, Kennedy RT, Wightman RM (1992) Evoked neuronal activity
accompanied by transmitter release increases oxygen concentration in rat striatum in
vivo but not in vitro. J Cereb Blood Flow Metab 12: 629637.
Capella P, Ghasemzadeh MB, Adams RN, Wiedemann DJ, Wightman RM (1993)
Real-time monitoring of electrically stimulated norepinephrine release in rat thalamus: II. Modeling of release and reuptake characteristics of stimulated norepinephrine overow. J Neurochem 60: 449453.
Ghasemzadeh MB, Capella P, Mitchell K, Adams RN (1993) Real-time monitoring
of electrically stimulated norepinephrine release in rat thalamus: I. Resolution of
transmitter and metabolite signal components. J Neurochem 60: 442448.
Bunin MA, Wightman RM (1998) Quantitative evaluation of 5-hydroxytryptamine
(serotonin) neuronal release and uptake: an investigation of extrasynaptic transmission. J Neurosci 18: 48544860.
Bunin MA, Prioleau C, Mailman RB, Wightman RM (1998) Release and uptake
rates of 5-hydroxytryptamine in the dorsal raphe and substantia nigra reticulata of the
rat brain. J Neurochem 70: 10771087.
Xin Q, Wightman RM (1997) Enzyme modied amperometric sensors for choline
and acetylcholine with tetrathiafulvalene tetracyanoquinodimethane as the electrontransfer mediator. Anal Chem Acta 341: 4351.
Xin Q, Wightman RM (1997) Transport of choline in rat brain slices. Brain Res 776:
126132.

276
25.

26.

27.

28.

29.
30.
31.

32.

33.

34.
35.
36.

37.
38.

39.
40.
41.

Joseph and Wightman


Kulagina NV, Shankar L, Michael AC (1999) Monitoring glutamate and ascorbate
in the extracellular space of brain tissue with electrochemical microsensors. Anal
Chem 71: 50935100.
Pantano P, Kuhr WG (1993) Dehydrogenase-modied carbon-ber microelectrodes
for the measurement of neurotransmitter dynamics. 2. Covalent modication utilizing avidinbiotin technology. Anal Chem 65: 623630.
Kuhr WG, Barrett VL, Gagnon MR, Hopper P, Pantano P (1993) Dehydrogenasemodied carbon-ber microelectrodes for the measurement of neurotransmitter
dynamics. 1. NADH voltammetry. Anal Chem 65: 617622.
Hu Y, Mitchell KM, Albahadily FN, Michaelis EK, Wilson GS (1994) Direct measurement of glutamate release in the brain using a dual enzyme-based electrochemical sensor. Brain Res 659: 117125.
Fonnum F (1984) Glutamate: a neurotransmitter in mammalian brain. J Neurochem
42: 111.
Pickel VM, Beckley SC, Joh TH, Reis DJ (1981) Ultrastructural immunocytochemical localization of tyrosine hydroxylase in the neostriatum. Brain Res 225: 373385.
Garris PA, Ciolkowski EL, Pastore P, Wightman RM (1994) Efux of dopamine
from the synaptic cleft in the nucleus accumbens of the rat brain. J Neurosci 14:
60846093.
Cahill PS, Walker QD, Finnegan JM, Mickelson GE, Travis ER, Wightman RM
(1996) Microelectrodes for the measurement of catecholamines in biological systems. Anal Chem 68: 31803186.
Wightman RM, Amatore C, Engstrom RC, Hale PD, Kristensen EW, Kuhr WG,
May LJ (1988) Real-time characterization of dopamine overow and uptake in the
rat striatum. Neuroscience 25: 513523.
Michael DJ, Joseph JD, Kilpatrick MR, Travis ER, Wightman RM (1999) Improving data acquisition for fast-scan cyclic voltammetry. Anal Chem 71: 39413947.
Michael AC, Justice JB, Jr. (1987) Oxidation of dopamine and 4-methylcatechol at
carbon ber disk electrodes. Anal Chem 59: 405410.
Feng JX, Brazell M, Renner K, Kasser R, Adams RN (1987) Electrochemical pretreatment of carbon bers for in vivo electrochemistry: effects on sensitivity and
response time. Anal Chem 59: 18631867.
Kovach PM, Deakin MR, Wightman RM (1986) Electrochemistry at partially
blocked carbon-ber microcylinder electrodes. J Phys Chem 90: 46124617.
Sujatitvanichpong S, Aokik, Tokuda AK, Matsuda, H (1986) Electrochemicalbehavior of dopamine at carbon-ber electrodes. Electroanal Chem Interfacial Electrochem 198: 195203.
Baumann PA, Maitre L (1976) Is drug inhibition of dopamine uptake a misinterpretation of in vitro experiments? Nature 264:789790.
Horn AS (1978) Characteristics of neuronal dopamine uptake. Adv Biochem Psychopharmacol 19: 2534.
Raiteri M, Cerrito F, Cervoni AM, del Carmine R, Ribera MT, Levi G (1978) Studies on dopamine uptake and release in synaptosomes. Adv Biochem Psychopharmacol 19: 3556.

Mechanisms and Kinetics of Neurotransmission


42.

43.
44.
45.
46.

47.

48.

49.

50.

51.

52.

53.
54.

55.
56.

57.

277

Annunziato L, Leblanc P, Kordon C, Weiner RI (1980) Differences in the kinetics of


dopamine uptake in synaptosome preparations of the median eminence relative to
other dopaminergically inervated brain regions. Neuroendocrinology 31: 316320.
Bunin MA, Wightman RM (1998) Measuring uptake rates in intact tissue. Methods
Enzymol 296: 689707.
Near JA, Bigelow JC, Wightman RM (1988) Comparison of uptake of dopamine in
rat striatal chopped tissue and synaptosomes. J Pharmacol Exp Ther 245: 921927.
Kuhr WG, Wightman RM (1986) Real-time measurement of dopamine release in rat
brain. Brain Res 381: 168171.
May LJ, Kuhr WG, Wightman RM (1988) Differentiation of dopamine overow and
uptake processes in the extracellular uid of the rat caudate nucleus with fast-scan in
vivo voltammetry. J Neurochem 51: 10601069.
Jones SR, Garris PA, Wightman RM (1995) Different effects of cocaine and
nomifensine on dopamine uptake in the caudate-putamen and nucleus accumbens. J
Pharmacol Exp Ther 274: 396403.
Giros B, Jaber M, Jones SR, Wightman RM, Caron MG (1996) Hyperlocomotion
and indifference to cocaine and amphetamine in mice lacking the dopamine transporter. Nature 379: 606612.
Garris PA, Wightman RM (1995) Distinct pharmacological regulation of evoked
dopamine efux in the amygdala and striatum of the rat in vivo. Synapse 20:
269279.
Kawagoe KT, Garris PA, Wiedemann DJ, Wightman RM (1992) Regulation of transient dopamine concentration gradients in the microenvironment surrounding nerve
terminals in the rat striatum. Neuroscience 51: 5564.
Gainetdinov RR, Grekhova TV, Sotnikova TD, Rayevsky KS (1994) Dopamine D2
and D3 receptor preferring antagonists differentially affect striatal dopamine release
and metabolism in conscious rats. Eur J Pharmacol 261: 327331.
Gainetdinov RR, Sotnikova TD, Grekhova TV, Rayevsky KS (1996) In vivo evidence for preferential role of dopamine D3 receptor in the presynaptic regulation of
dopamine release but not synthesis. Eur J Pharmacol 308: 261269.
Cass WA, Gerhardt GA (1994) Direct in vivo evidence that D2 dopamine receptors
can modulate dopamine uptake. Neurosci Lett 176: 259263.
Meiergerd SM, Patterson TA, Schenk JO (1993) D2 receptors may modulate the
function of the striatal transporter for dopamine: kinetic evidence from studies in
vitro and in vivo. J Neurochem 61: 764767.
Rothblat DS, Schneider JS (1997) Regionally specic effects of haloperidol and
clozapine on dopamine reuptake in the striatum. Neurosci Lett 228: 119122.
Dickinson SD, Sabeti J, Larson GA, Giardina K, Rubinstein M, Kelly MA, Grandy
DK, Low MJ, Gerhardt GA, Zahniser NR (1999) Dopamine D2 receptor-decient
mice exhibit decreased dopamine transporter function but no changes in dopamine
release in dorsal striatum. J Neurochem 72: 148156.
Sokoloff P, Giros B, Martres MP, Bouthenet ML, Schwartz JC (1990) Molecular
cloning and characterization of a novel dopamine receptor (D3) as a target for neuroleptics. Nature 347: 146151.

278
58.

59.

60.

61.

62.
63.

64.

65.
66.
67.

68.
69.

Joseph and Wightman


Tepper JM, Sun BC, Martin LP, Creese I (1997) Functional roles of dopamine D2
and D3 autoreceptors on nigrostriatal neurons analyzed by antisense knockdown in
vivo. J Neurosci 17: 25192530.
Palij P, Bull DR, Sheehan MJ, Millar J, Stamford J, Kruk ZL, Humphrey PP (1990)
Presynaptic regulation of dopamine release in corpus striatum monitored in vitro in
real time by fast cyclic voltammetry. Brain Res 509: 172174.
Limberger N, Trout SJ, Kruk ZL, Starke K (1991) Real time measurement of
endogenous dopamine release during short trains of pulses in slices of rat neostriatum and nucleus accumbens: role of autoinhibition. Naunyn Schmiedebergs Arch
Pharmacol 344: 623629.
Jones SR, ODell SJ, Marshall JF, Wightman RM (1996) Functional and anatomical
evidence for different dopamine dynamics in the core and shell of the nucleus
accumbens in slices of rat brain. Synapse 23: 224231.
McElvain JS, Schenk JO (1992) A multisubstrate mechanism of striatal dopamine
uptake and its inhibition by cocaine. Biochem Pharmacol 43: 21892199.
Burnette WB, Bailey MD, Kukoyi S, Blakely RD, Trowbridge CG, Justice JB, Jr.
(1996) Human norepinephrine transporter kinetics using rotating disk electrode
voltammetry. Anal Chem 68: 29322938.
Sonders MS, Zhu SJ, Zahniser NR, Kavanaugh MP, Amara SG (1997) Multiple
ionic conductances of the human dopamine transporter: the actions of dopamine and
psychostimulants. J Neurosci 17: 960974.
Jones SR, Joseph JD, Barak LS, Caron MG, Wightman RM (1999) Dopamine neuronal transport kinetics and effects of amphetamine. J Neurochem 73: 24062414.
Krueger BK (1990) Kinetics and block of dopamine uptake in synaptosomes from
rat caudate nucleus. J Neurochem 55: 260267.
Jones SR, Gainetdinov RR, Wightman RM, Caron MG (1998) Mechanisms of
amphetamine action revealed in mice lacking the dopamine transporter. J Neurosci
18: 19791986.
Stein WD, Lieb WR (1986) Transport and diffusion across cell membranes. Orlando:
Academic Press.
Richeld EK, Penney JB, Young AB (1989) Anatomical and afnity state comparisons between dopamine D1 and D2 receptors in the rat central nervous system. Neuroscience 30: 767777.

9
Electrochemical Monitoring
of Exocytosis from Individual
PC12 Cells in Culture
Leslie A. Sombers and Andrew G. Ewing
Pennsylvania State University, University Park, Pennsylvania

I.

INTRODUCTION

The study of the brain and its function has been an area of research for many
years, with work focusing on aspects that vary from whole brain systems to single cells. Chemical analysis of single cells is an area of great interest in the biological and medical sciences, as knowledge of the dynamics of single cells should
advance our understanding of such diverse processes as neuronal communication,
neurotransmitter transport and secretion, receptor-mediated signal transduction,
and voltage-gated ion channels in regulating cellular functions. Several analytical
techniques have provided information regarding the neurophysiology and neuropharmacology of the brain. However, one technique, electroanalysis, is especially useful for these types of studies as it has the ability to provide highly sensitive qualitative and quantitative information for electroactive neurochemicals
that are easily oxidized or reduced. Electroanalytical techniques possess unique
characteristics that make them suitable for studying dynamic chemical processes
in real time as they occur in complex biological matrices.
The fundamental building block of the brain is the single nerve cell or neuron [1,2]. The classical model of neuronal communication involves the transduction of the electrical signal of an action potential in a neuron to a chemical signal,
which then traverses across a small gap from one neuron to another [3]. This small

279

280

Sombers and Ewing

gap, known as the synapse, is classically dened as the small space between the
membranes of two different cells through which chemical messages are transferred [4]. Exocytosis forms the key component of this neuronal communication.
The process involves the precise targeting of a presynaptic intracellular secretory
vesicle (storage compartment) to the plasma membrane, and its subsequent fusion
with that membrane, which is triggered by a rise in the cytosolic Ca2+ concentration [5]. Thus, the vesicular contents are released into the extracellular space. The
release event takes place in a few milliseconds, therefore requiring a rapid detection scheme that is both selective and sensitive so as to allow the precise quantitation of the chemical messenger released [3].
Previous studies have examined the release of chemical messengers on the
whole-cell level through capillary electrophoresis and microcolumn liquid chromatography with electrochemical detection [6]. Although these techniques provide both qualitative and quantitative information about the average vesicular
chemical content, they lack the temporal and spatial resolution needed for the precise detection of individual release events [6]. Methods to observe and quantitate
individual events have traditionally revolved around electron microscopy and
patch-clamp capacitance measurements [7].
In the past decade, however, signicant advances in microelectrochemical
techniques have made it possible to directly monitor individual exocytotic events
involving easily oxidized messengers using electrochemical measurements at
microelectrodes [8]. Voltammetric microelectrodes [911] are ideally suited for
monitoring dynamic chemical changes from discrete neurochemical events in real
time, as they possess rapid response times (s). Furthermore, their small size
(m) allows measurement of the local concentration of transmitter in very small
spaces, such as at the surface of a single cell. Although there are many compounds
of neurological interest, only a few demonstrate redox chemistry at potentials
attainable for electrochemical research. This gives electrochemical methods a
unique chemical selectivity so that a single, or perhaps a few, compounds can be
measured in a mixture of many other compounds, such as in the complex mixture
of biological matrices [6]. The spatial selectivity provided by the small size of the
probes combined with the chemical selectivity provided by electrochemistry
make this an ideal technique for measuring target compounds that act as messengers and modulators present at low quantities and sequestered in small volumes
in specic regions of the brain. Several neurotransmitters such as the catecholamines, serotonin (5-hydroxytryptamine [5-HT]), and insulin are easily oxidized and are thus readily detectable, thereby permitting the precise identication
and quantitation of messenger released during each exocytotic event [12].
Electrochemical detection is generally carried out in one of two modes:
fast-scan voltammetry or constant potential amperometry [3]. The characteristic
voltammograms of different neurotransmitters obtained by fast-scan voltamme-

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

281

try provide qualitative information that enables the identication of the released
substance, whereas amperometry provides highly sensitive quantitative information with a rapid response time. In a typical experiment, a microelectrode is
manipulated up to a cell, and subsequently the cell is chemically or electrically
stimulated to elicit exocytosis. Upon stimulation, vesicles inside the cell fuse with
the cell membrane and release their contents into the extracellular uid. Any electroactive material released from vesicles that fuse directly adjacent to the electrode is completely oxidized, allowing quantication of the contents of individual
vesicles. In effect, the electrode can be thought of as a postsynaptic cell that
receives the chemical signal(s) from the presynaptic cell.
The earliest experiments of this type employed voltammetry as a detection
scheme. Fast-scan cyclic voltammetry can be used to discriminate between
released substances with different oxidation/reduction properties. Voltammetric
measurements of absolute concentrations released from single vesicles undergoing exocytosis have proven to be difcult; however, because the amount of transmitter released is only at zeptomole levels and because the events occur on the
millisecond time scale [13]. Furthermore, although elicited by chemical stimulation, these events do not occur at precise times. Although several neurotransmitters have been identied on the basis of characteristic voltammograms, it is difcult to distinguish catecholamines with this technique. However, fast-scan cyclic
voltammetry has been used to identify the secreted catecholamines norepinephrine and epinephrine [14].
In contrast, constant potential amperometry has allowed the quantitative
aspects of single exocytotic release events to be studied in detail. This technique
provides specic information on the amplitude, kinetics, and location of individual release events from single cells. Secretion is resolved as a series of current
spikes that represent the electrooxidation of released substances. Wightman et al.
have shown that each amperometric current spike detected represents the oxidation of neurotransmitter from a single exocytotic event [8]. In addition, the technique holds the potential to provide clues about the fusion pore complex, which
manifests itself as a pre-spike foot that is observed directly prior to some release
events. A drawback of this technique, however, is that chemical identication
must be sacriced for temporal resolution. This is a concern when one considers
the complex biological matrix present in synaptic vesicles.
The recent development of these electrochemical techniques for the study
of vesicular release from single cells provides a powerful tool for the study of
chemical signaling in the nervous system. However, the application of these
methods to mammalian neurons is difcult, mainly due to size constraints, the
extremely small quantities of chemical transmitter released at a single synaptic
vesicle, and the complex working environment of nerve cell cultures [15]. Neuronal synaptic vesicles are only about 50 nm in diameter and have been estimated

282

Sombers and Ewing

to contain only about 1,000 to 3,000 catecholamine molecules, thus approaching


the current limits of detection [16]. In addition, mammalian neurons are very difcult to work with in the laboratory because they stop proliferation once they
have matured. However, much can be learned about exocytosis by studying this
phenomenon at model cell systems with larger vesicles and higher concentrations
of neurotransmitter.
The rst model system used for the electrochemical detection of exocytosis
was the bovine adrenal medullary chromafn cell line [17]. Since then, many different cultured cell lines, including beige mouse mast cells, human pancreatic cells, and rat pheochromocytoma (PC12) cells, have been used in electrochemical studies to probe the process of exocytosis. The PC12 clonal cell line,
established from a transplantable rat adrenal pheochromocytoma, provides an
excellent neuronal model system for the development of methods to monitor
synaptic quantal release. PC12 cells share many physiological properties with primary cultures of sympathetic ganglion neurons [18]. For instance, PC12 cells
synthesize acetylcholine as well as the catecholamines dopamine and norepinephrine. They also store, release, and take up these neurotransmitters in a manner similar to sympathetic neurons. They contain voltage-sensitive Na+, K+, and
Ca2+ channels, receptor-mediated Na+ and Ca2+ channels, and Ca2+ -sensitive K+
channels. PC12 cells also express muscarinic and nicotinic acetylcholine receptors. On chemical stimulation, these cells exhibit Ca2+ -dependent exocytosis. In
addition, PC12 cells have the ability to differentiate into neuronal-like cells upon
treatment with nerve growth factor, thus providing a working model of the developing nerve cell [3]. Furthermore, it is also relatively easy to maintain the PC12
cell line in culture. We will focus on electrochemical measurements from rat
PC12 cells for the purpose of this chapter; however, many of the electrochemical
principles and techniques described have been developed for, and are applicable
to, many types of cultured cells.

II.

TECHNIQUES

A.

Amperometry at PC12 Cells

The initial electrochemical studies on PC12 cells have focused on quantitation of


catecholamine release. Zeptomole levels of catecholamine have been observed
for a single-release event using amperometric detection with carbon ber microelectrodes [15]. To measure the time course of these release events the electrode,
held at a potential where dopamine is readily oxidized, is gently placed against
the cell surface. Catecholamines are released from individual vesicles following
perfusion of the cell with a solution containing nicotine, elevated potassium chloride, or other chemical stimulants. Such agents serve to directly depolarize the

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

283

cell membrane to initiate exocytosis. Figure 1A is a schematic of the typical setup


for this experiment. The stimulus results in multiple current transients due to the
oxidation of catecholamine released during individual exocytotic events (Figure
1B,C). Because the catecholamines (dopamine and norepinephrine) are the only
known electrochemically active neurotransmitters present at signicant levels in
PC12 cell cultures, the easily oxidized substances responsible for the current transients observed have been tentatively identied as catecholamines. Figure 2 displays a characteristic amperogram detecting exocytotic release. The current transients are on the time scale expected for exocytotic release of catecholamines (9.3
ms half-width), and each vesicle has been reported to contain approximately 190
zmol (114,300 molecules) of catecholamine [15].
The quantitation of amperometric data is straightforward. When the electrode is held at a potential where analyte oxidation is diffusion limited, analyte in
the conned space between the cell and the electrode is oxidized [12,19]. Therefore, the amount of charge passed at the electrode surface is directly proportional
to the number of moles of analyte by Faradays law:
Q = nNF

(1)

where Q is the charge passed at the electrode surface, n is the number of moles of
electrons transferred per mol of analyte oxidized (two for the catecholamines), N
is the number of moles of detected analyte, and F is the Faraday constant (96,485
C/mol). It is important to note that events that occur at sites that are not directly
underneath the electrode are essentially undetected [19]. Diffusion from the
release site is equal in all directions; hence, events occurring at points removed
from the electrode contribute only to a broad, dilute background and not to the
millisecond current transients.
The reported quantity of neurotransmitter released from individual vesicles
in a cell type is an average value, but the individual amounts are broadly distributed about this range. The distribution can be visualized by plotting a histogram
of the total cellular exocytosis data [6]. The total number of moles of catecholamine detected for each exocytosis event are collected into bins and plotted
as the percent of the total number of vesicles undergoing exocytosis. Figure 3
shows a typical histogram generated from exocytotic events measured from PC12
cells [15]. The average vesicular content for this set of cells is 190 zmol; however,
the distribution ranges from 66 to 1290 zmol. This distribution is non-gaussian
(skewed) and relatively narrow, with 85% of current transients corresponding to
less than 330 zmol (200,000 molecules).
Studies have shown that plotting either a cubed root or log transformation
of the amperometrically recorded quantal sizes results in a normal distribution
[12]. Two suggested explanations have been proposed for the skewed distribution. The rst is based on the idea that the vesicles, which have a gaussian distribution of radii, possess relatively uniform concentrations of neurotransmitter [8].

284

Sombers and Ewing

Figure 1 Schematic depicting the experimental arrangement used for measuring neurotransmitter secretion from a single cell (A) and representative amperometric data obtained
from a single PC12 cell (B). A portion of the data shown in B is also displayed at a smaller
time and current scale (C). Data was obtained using a 5 m carbon ber electrode held at
0.65 V vs. SSCE.

Figure 2 Amperograms of individual vesicular exocytosis events from PC12 cells. (A)
Amperogram of a control experiment with the tip of the working electrode 200 m from
the cell. When the electrode tip is bathed with 30 nL of stimulating solution (1 mM nicotine in 105 mM K+ balanced salt solution), a slow change in the background amperometric
response is observed. (B) Amperogram of vesicular exocytosis induced by bathing a cell
(legend continued on next page)

286

Sombers and Ewing

Figure 3 Frequency distribution of the amount of catecholamine contained in individual


current transients from single cells. The bins are divided into 66 zmol intervals. Numbers
on the x-axis represent the mean of each interval. Data shown have been obtained from the
rst stimulation of 13 individual cells; 1912 spikes have been included. (Reproduced from
Anal. Chem. with permission [15].)

Figure 2 (Continued) with stimulating solution (described above), with the electrode
placed on top of the cell. Narrow current transients are clearly observed at times corresponding to the rst two chemical stimulations. Only background current is observed following a third cell stimulation. It appears that the vesicles responsible for the release of catecholamines are depleted with multiple stimulations on this time scale. (C) A one-second
portion of data from the rst stimulation in B is displayed here with an expanded time
scale. All data were obtained using a 5 m carbon ber electrode held at 0.65 V vs. SSCE.
(Reproduced from Anal. Chem. with permission [15].)

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

287

As the vesicular concentration is equivalent to moles/volume and V = 4/3r3, so


r is proportional to the cubed root of the vesicle content [8,20]. Thus, the data for
each release event can be replotted as the frequency of release events versus
the cubed root of vesicular amount (in moles). A second explanation suggests that
a lognormal distribution of vesicular amount will result if multiplicative deviations from the mean occur; such as if integral differences in the number of uptake
transporters per vesicle result in multiplicative transmitter accumulation rates
[21]. Thus, release event data can also be plotted as the frequency of release
events vs. the log transform of quantal size. When the amperometrically recorded
data are plotted after either of the aforementioned transformations, a dramatically
different histogram with a nearly gaussian distribution results that is said to reect
the vesicular size distribution. Small deviations in the gaussian shape are possibly due to variations in the vesicular concentration, which is assumed to be
constant.

B.

Comparison of Data at PC12s to Data at Adrenal, Mast,


and b-Cells

The majority of work concerning exocytosis at the single-cell level has been carried out on bovine adrenal chromafn cells [8,17,19,21a]. However, amperometric studies of exocytosis have also been performed at mast cells [12], pancreatic
-cells [12,22], and of course at PC12 cells. For each of these cell types, the
vesicular radii have been shown previously to have a gaussian distribution [12].
As previously discussed, spherical volumes are proportional to the cube of the
vesicular radius. Thus, histograms depicting the cubed root of spike areas of the
amperometrically detected events for each cell type have been found to yield a
more gaussian distribution of events, with a relative standard deviation similar to
that calculated for the distribution of vesicular radii [12].
Finnegan et al. report the average quantity of neurotransmitter per adrenal
medullary chromafn vesicle to be 3.3 attomole, which corresponds to 2.0 106
molecules; however, the individual vesicular amounts are broadly distributed
about this range [12]. This broad distribution had been attributed to a wide range
of individual vesicular volumes. As expected, the distribution of volumes exhibits
a skewed, non-gaussian distribution. The distribution of Q1/3, however, is gaussian and distributed similarly to that for the vesicular radii, with a similar relative
standard deviation. The integrated Q value has been determined for repetitive
stimulations at a single adrenal chromafn cell, and a histogram has been plotted
(Figure 4A). The gaussian histogram of the cubed root of charges for this data set
is also given in Figure 4B. The relative standard deviation for this data set (26%)
is similar to that of the distribution of radii from epinephrine-containing vesicles

288

Sombers and Ewing

Figure 4 Q and Q1/3 histograms from a single bovine chromafn cell. (A) Area histogram for spikes obtained during 15 repetitive 60 mM K+ stimulations at a single chromafn cell. The mean SD value for this data set is 0.89 1.03 pC. (B) Histogram of the
cubed root of the areas for the same data set. The gaussian t has a mean SD value of 0.86
0.22 pC1/3. 334 spikes are included. (Reproduced from J. Neurochem. with permission
[12].)

in adrenal tissue slices, which has been determined to be 35% [23]. Using a mean
radius for epinephrine vesicles of 170 nm [23], the mean vesicular catecholamine
concentration for this cell can be estimated to be 0.15 M.
Beige mouse mast cells, which contain 5-hydroxytryptamine (serotonin)
and histamine, have also been utilized for amperometric investigations of exocytotic release. Both histamine and serotonin are easily oxidizable molecules that
perform a variety of functions in the nervous and immune systems of mammals.
One advantage of using this cell system is the large size of the vesicles present.
The larger vesicles make mast cells especially useful for studies involving the
fusion process of exocytosis, as the size allows release to be correlated with
simultaneous changes in membrane capacitance [24,24a]. Capillary chromatography studies utilizing electrochemical detection have quantied both agents,
yielding 150 18 and 3.8 1.3 fmol of histamine and serotonin/cell, respectively
[25]. Both agents have been shown to co-release from a single vesicle.
Finnegan et al. have also performed the amperometric detection of 5-HT
release from rat mast cells, in a manner similar to the study done on bovine chromafn cells [12]. One main difference observed is a delay of 3060 s between the

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

289

Figure 5 Q1/3 histograms for amperometric spikes from rat mast, human pancreatic -,
and rat PC12 cells. Histograms are from 234 A23187-induced spikes from seven rat peritoneal mast cells (open columns), 228 tolbutamide-induced spikes from six human pancreatic -cells (gray columns), and 475 K+-induced spikes from 17 rat PC12 cells (solid
columns). The tted gaussian curves have mean SD values of 1.06 0.19, 0.58 0.12,
and 0.34 0.050 pC1/3, respectively. (Reproduced from J. Neurochem. with permission
[12].)

application of the calcium ionophore and detection of the rst release event, as
compared to an average 1 s delay observed for the chromafn cells [12]. Amperometric data from these cells has been pooled to create a Q1/3 histogram (Figure
5) with a relative SD (18%) similar to that reported for the distribution of vesicular radii (12%) [26]. Using the cubed root of detected spike areas and a mean
vesicular radius of 770 nm [26], the vesicular 5-HT concentration can be calculated to be 3.8 mM.
Exocytosis from pancreatic -cells also exhibits a Ca2+ ion dependence,
similar to the other secretory cells described. -cells secrete insulin based on the
level of glucose in the body, and thus amperometric monitoring of insulin secretion from individual cells is useful in efforts to gain insight into the function of
these cells in the regulation of blood sugar levels in the body [20]. Huang et al.
have used chromatographic methods to show that the material secreted during

290

Sombers and Ewing

exocytosis at these cells is indeed insulin [27]. The fact that insulin is stored
within vesicles provides a common link between exocytosis from -cells and
from cells of the nervous system. For these reasons, Finnegan et al. have also plotted 228 amperometric spike areas from six human pancreatic -cells as a Q1/3 histogram (Figure 5) [12]. The distribution of Q1/3 values displays a relative SD
(20%) that is almost identical to the previously reported distribution of vesicular
radii (21%) [28]. A mean vesicular radius of 150 nm [28] has been used to calculate a mean vesicular insulin concentration of 0.13 M.
In comparison, Finnegan et al. have also utilized amperometric detection to
examine exocytotic release from PC12 cells [12]. A Q1/3 histogram, also presented in Figure 5, exhibits a relative SD of 15%. In agreement with the results
from the other cell types, this SD is similar to the previously reported relative SD
of PC12 vesicular radii (25%) [29]. A mean vesicular radius of 79 nm [29] has
been used to determine a mean PC12 vesicular catecholamine concentration of 99
mM.
C.

Voltammetry at PC12 Cells

Although storage vesicles in PC12 cells are known to contain a variety of neurotransmitters, the catecholamines dopamine and norepinephrine are the only
known electrochemically active neurotransmitters present at signicant levels in
this cell line [13]. Thus, the easily oxidized substances responsible for the amperometric current transients observed have been tentatively identied as catecholamines; however, there has been little direct evidence that the molecular
species detected during amperometric monitoring of exocytosis is indeed a catecholamine. For this reason, Kozminski et al. have used a combination of voltammetry and pharmacological manipulations to identify the approximately 1019
moles of neurotransmitter released from individual exocytosis events at PC12
cells as a catecholamine, most likely dopamine [13]. Interpreting these voltammograms, which have been obtained in the attoliter volume affected between the
electrode and the cell, and which are dened by the size of the exocytosis pore
during exocytosis, has proven to be difcult.
Fast-scan cyclic voltammetry at microelectrodes is dominated by a doublelayer charging current that masks the faradaic current arising from the oxidationreduction of electroactive neurotransmitters. This background current is digitally
subtracted from the faradaic current to give a background subtracted voltammogram. Because individual fast cyclic voltammograms are obtained on a time
scale similar to release events following cellular stimulation, the catecholamines
are the most likely compounds to change in concentration during the time frame
of each scan.
Generally, the shape of background-corrected voltammograms varies with
different chemical species. At the scan rates used in fast-scan cyclic voltammetry, voltammogram shapes are kinetically controlled. As a result, most electroac-

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

291

tive analytes present in the extracellular uid have a unique voltammetric ngerprint that can be used to tentatively identify the compound(s) that have
changed in concentration during the time course of the scan. Unfortunately, however, qualitative chemical analyses cannot be based solely on the shape of the
voltammograms. For instance, the neurotransmitters serotonin, epinephrine, and
dopamine all oxidize at potentials between +0.2 and 0.5 V versus Ag-AgCl with
untreated electrodes, and the shape of the signal from each of these compounds is
not readily distinguishable. Furthermore, interpreting the shape of the voltammograms can become complicated if more than one neurotransmitter changes in
concentration. Thus, when using fast-scan cyclic voltammetry to detect electroactive neurotransmitters in the brain, it is essential that independent pharmacological manipulations be used to verify the identity of the detected neurotransmitter [30].
In an attempt to qualitatively identify the released substance from PC12
cells, cyclic voltammetry has been performed at 300 V/s as PC12 cells were stimulated with an elevated K+ solution [13]. At this rate, a complete voltammogram
has been collected in 10 ms and thus it should capture single exocytosis events as
they occur in real time. A characteristic of cyclic voltammetry is the presence of
a high background current at fast scan rates. This stable non-faradaic background
current is digitally subtracted from the faradaic current to yield analytically useful data. To determine which scans (0.5 to 1.0 V) exhibit an increased faradaic
signal due to the release of neurotransmitter, an average of 10 data points collected within a 60-mV window of the peak oxidation potential of dopamine is
continuously monitored (Figure 6A). When one focuses in further (Figure 6B), it
becomes apparent that individual current transients rise sharply and decay rapidly. Removal of background signal (average of scans 687690) from the faradaic
response (scan 693) reveals a voltammogram (Figure 6C) that exhibits the characteristic features of dopamine in solution. For comparison, Figure 6D depicts a
background-subtracted voltammogram of a standard 25 M dopamine solution
collected from a ow injection apparatus. For this measurement, the electrode tip
is positioned a short distance inside the exit tube of a loop injector apparatus.
Physiological saline (identical to that used for single cell sampling) is pumped
through the ow system with a syringe at a rate of 1.0 mL/min, and the dopamine
standard is injected manually into the owing stream. In response to the data
depicted in Figure 6, the released compound at PC12 cells has been tentatively
identied as dopamine.
Calibration plots of peak current versus standard DA concentration are prepared for each electrode used to evaluate the concentration of measured exocytotic events, and an estimate of the concentration during exocytosis can be
obtained by comparison of the peak current for voltammograms obtained for individual exocytosis events to the calibration plot [13]. However these concentration
estimates are lower than expected for vesicular concentrations. This is because in
measuring exocytotic events, the released catecholamine only accesses a very

292

Sombers and Ewing

Figure 6 Fast-scan cyclic voltammetry data collected during stimulated exocytosis of a


single PC12 cell. Cyclic voltammetry was performed in the two- electrode mode using an
SSCE reference electrode for all measurements. A scan rate of 300 V/s was employed using
a triangular wave form that was ramped between 0.5 and +1.0 V. (A) Average current
within a 60-mV window of the peak oxidation potential of DA. Individual current spikes
represent faradaic current. (B) Enlargement of the region between scans 600 and 800 showing two current transients used to generate voltammograms. (C) Background-subtracted
voltammograms generated by subtracting the average signal of scans 687690 from the signal of scan 693. (D) Background-subtracted voltammogram of a standard 25 M DA solution collected using the ow system described in the text. The same buffer medium is used
for both cellular experiments and for standards. (Reproduced from Anal. Chem. with permission [13].)

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

293

small area of the electrode. However, the calibration is carried out in a solution
accessing the entire electrode surface. Due to this limitation, the absolute concentration of vesicular species is not accurately measured; however, an estimated
fraction of the true vesicular concentration is accomplished. A simple model, outlined in pictorial form in Figure 7, has been developed to explain this concept.
Consider the general cyclic voltammetry equation for peak current:
ip = (2.69 105)n3/2 AD01/2v1/2C0*

(2)

where ip is peak current, n is the number of electrons transferred, A is the electrode area of oxidation/reduction, D is the diffusion coefcient of the analyte, v is
the scan rate, and C is the concentration of analyte [13]. As illustrated in Figure
7, the area of the electrode used for the oxidation/reduction of analyte is signicantly smaller for substances released from a single vesicle compared to a standard solution of the same substance. Area differences between the total electrode
area (calculated from the average of the major and minor radii) and the vesicular
area suggest that vesicular concentrations should be ~0.061% of their true value
when compared to measurements in standard solution [13]. Although this model
indicates that absolute intravesicular concentrations cannot be measured with
voltammetry, relative concentrations can be evaluated and qualitative analysis
can be performed.
The model allows determination of relative concentrations from individual
release events and has been used to examine events at control cells and cells incubated with the dopamine precursor, L-3,4-dihydroxyphenylalanine (L-DOPA)
[13]. Exposure to L-DOPA (100 M for 1 hr) results in 145 detectable exocytosis
events for 11 cells compared to 77 events for 29 control cells, again suggesting
that the released substance is indeed dopamine and that vesicles can be loaded
with the catecholamine. Additionally, each event has a larger half-width (t1/2) in
the release of vesicular contents when compared to cells that are not incubated
with L-DOPA. This data is signicant kinetically, as it suggests a longer release
time for the increased quantity of dopamine. The data are summarized in Figure
8. As described previously, the non-faradaic background current in Figure 8A and
Figure 8B is stable, whereas the individual current transients rise very sharply and
decay rapidly.
Histograms of frequency versus apparent concentration that have been prepared for voltammograms collected from cells with and without L-DOPA exposure provide similar distributions for both types of cells [13]. These histograms
are presented in Figure 9. The apparent average vesicle concentration measured
in this experiment is 3.47 0.17 M for events from L-DOPAexposed cells, as
compared to 3.35 0.24 M for control cells. These data have led to the conclusion that loading vesicles by increased transmitter synthesis does not lead to elevated concentrations at individual release sites. It is proposed instead that vesicles
increase in size as catecholamine content increases [13].

294

Sombers and Ewing

Figure 7 Simple model of the concentration of vesicular events as determined by fastscan rate cyclic voltammetry. (A) Pictorial demonstration that the area of the electrode used
for oxidation/reduction of the DA species is very different for the cellular case compared
with that in standard solution. (B) A head-on view depicting the difference in electrode area
used in the above two cases. Beveling a carbon ber on a 45 angle creates an elliptical surface with major and minor radii of about 3.5 and 2.5 m, respectively. It is apparent that a
large difference exists between the vesicular area and that of the total electrode. (Reproduced from Anal. Chem. with permission [13].)

D.

A Mathematical Model to Estimate Vesicle Size


and Concentration

Amperometric measurements of exocytosis events at single cells provide information about the total amount of neurotransmitter released, the frequency of that
release, and the time course of each event. If an accurate distribution of the vesi-

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

295

Figure 8 Fast-scan rate cyclic voltammetry data collected during stimulated exocytosis
of a single PC12 cell after exposure to 100 M L-DOPA for 1 hr. Cyclic voltammetry was
performed in the two-electrode mode using an SSCE reference electrode for all measurements. A scan rate of 300 V/s was employed using a triangular wave form that was ramped
between 0.5 and +1.0 V. (A) Average current within a 60-mV window of the peak oxidation potential of DA. Individual current spikes represent faradaic current. (B) Enlargement
of the region between scans 1010 and 1225 showing multiple current transients used to generate voltammograms. (C) Background-subtracted voltammogram generated by subtracting the average signal of scans 10201023 from the signal of scans 10251026. (D) Background-subtracted voltammogram of a standard 25 M DA solution. (Reproduced from
Anal. Chem. with permission [13].)

296

Sombers and Ewing

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

297

cle size is available (via electron microscopy measurements), then the actual concentration and vesicle size can also be determined for each event. This adds a
great deal of information to that gained from typical amperometric measurements, especially when one considers that the neurotransmitter-receptor interaction is an equilibrium-based reaction that will be highly dependent on transmitter
concentration in the synapse. For these reasons, Anderson et al. have developed
a mathematical model that allows vesicle size and neurotransmitter concentration
distributions to be quantied for individual vesicles following amperometric electrochemical detection of exocytotic release [31].
Figure 10 is a pictorial representation of the space between the cell and
electrode surface (the drawing is not to scale) [31]. Following complete fusion of
a spherical intracellular vesicle to the plasma membrane, the surface area of the
vesicle creates a disk-shaped pore with a radius of 2rv. The contents of the vesicle diffuse from this disk and spread at an angle, , until they reach the surface of
the carbon ber electrode positioned at a distance, h, away from the cell surface.
The electrochemical reaction occurs at an area dened by re, which is representative of the size of each detected vesicle. This being the case, the quantity re is
related to the vesicle radius and the cell/electrode distance (h) by Eq. (3) [31].
re = 2rv + h tan()

(3)

Skipping a great deal of algebra, it can be shown that the ratio of rv3/re2 has a quadratic dependence on h, tan , and rv [31]. Since this quantity is a nontrivial function
of these variables, it is difcult to solve. However, by comparing the distribution of
vesicle size for a population of vesicles from electron microscopy to that for a
reduced radius (radius in units of the above variables) obtained from amperometry,
one can isolate values for each vesicle in the amperometric data set. Using parameters contained within amperometric exocytosis data and the distribution of vesicle size from electron microscopy data, a histogram of vesicle size can be calculated directly from the electrochemical data (Figure 11). Anderson et al. have
calculated the relative standard deviation of the size distribution calculated from
electrochemical data to be 25%, which closely matches the relative standard deviation of the vesicle size distribution measured by electron microscopy [29].

Figure 9 Apparent distributions of vesicular concentrations detected by fast-scan rate


cyclic voltammetry following K+-stimulated exocytosis. (A) Distribution generated from
control PC12 cells (n = 29 cells, 77 total events). (B) Distribution generated from
L-DOPAexposed (100 M for 1 hr) PC12 cells (n = 11 cells, 145 total events). For each
data set, the peak current of the anodic wave of all events is matched to its respective calibration plot to determine the detected concentration. All data are collected into bins having increments of 0.25 M and plotted as the percent of the total number of release events
detected. (Reproduced from Anal. Chem. with permission [13].)

298

Sombers and Ewing

Figure 10 Contents of a single vesicle after complete exocytosis. is the angle at which
diffusion of analyte occurs following release, h is the distance between the electrode and
the cell surface, re is the apparent vesicle radius that denes the electroactive area of the
electrode for each release event, and 2rv is the radius of the membrane for a spherical vesicle after it has completely fused with the plasma membrane, where rv is the original radius
of the intact vesicle. (Reproduced from J. Neurosci. Meth. with permission [31].)

Again, simplifying the algebra, a straightforward equation for the concentration of transmitter in each vesicle is available [31].
CDA = 3Q/(4nFrv3)

(4)

Here Q represents the charge passed at the surface of the electrode, n is the number of electrons transferred per analyte molecule (two for catecholamines), and F
is the Faraday constant (96,485 C/mol). The charge, Q, is obtained by integration
of the current with respect to time of a current vs. time trace, such as the one
shown in Figure 2C. Equation (4) has been used to obtain a quantitative distribution of transmitter concentration in a population of vesicles, as shown in Figure
12. The mean vesicular dopamine concentration calculated using this model is
148 7 mM. Previous estimates of concentration have been made using a known
average vesicle radius (i.e., by electron microscopy) and an average vesicle
amount (by amperometry). Using this method, a value of 99 mM has been
reported by Finnegan et al. [12], which is in close agreement to the average value
reported here. However, Finnegan et al. assumed a constant concentration of
transmitter in the vesicles for their calculations. The model presented here sug-

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

299

Figure 11 Normalized histogram of calculated vesicle radius (open bars) overlaid with
the known radius histogram (shaded bars) from Schubert et al. (1980) (14 bins, 10 nm/bin
each). (Reproduced from J. Neurosci. Meth. with permission [31].)

300

Sombers and Ewing

Figure 12 Vesicular DA concentration histogram (20 bins, 49 mM/bin) calculated from


Eq. (4) where each rv is taken from the normalized histogram presented in Figure 11.
(Reproduced from J. Neurosci. Meth. with permission [31].)

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

301

gests a distribution of vesicular neurotransmitter concentration exists and, therefore, previous estimates may have been limited in scope.
This vesicular catecholamine concentration distribution has important ramications to neurotransmission. Although the mean concentration reported in this
study is 148 mM, it is clear from the histogram in Figure 12 that the most likely
release event (mode) contains catecholamine at a concentration three times
smaller than the mean. This fact becomes important if a small number of vesicles
are released per stimulus. If hundreds of vesicles are released, the total amount of
messenger released will follow the distribution shown. However, if only a few
vesicles undergo release, the most likely release event will contain neurotransmitter at a much smaller concentration and, therefore, may not have as great an
effect on the postsynaptic cell or, perhaps more important, on the outlying cells
surrounding the synapse. Again, this is particularly important when one considers that the neurotransmitter-receptor interaction is an equilibrium-based reaction
that could be highly dependent on transmitter concentration if the receptors are
not saturated.
E.

Other Techniques

Electrochemical detection methods such as amperometry and fast cyclic voltammetry are highly effective for making rapid and sensitive measurements of exocytotic release events when the released transmitter can be readily oxidized or
reduced. Although attempts have been made to enzymatically convert other neurotransmitters into readily oxidizable intermediates [32], this additional step has
proven to be too slow to permit the direct observation of individual exocytotic
release events in real time. Thus, non-electroactive transmitters must be detected
via an alternate means. One alternative is to use the sniffer-patch detection technique [33].
The sniffer-patch method, which was rst used in 1983 to detect the release
of acetylcholine from developing growth cones [34,35], provides a highly sensitive means of detection by exploiting the natural high afnity of ligand-gated ion
channels for their native neurotransmitter [33]. The rst step in the detection
process is to excise a small patch of receptor-rich membrane from a donor cell. If
the receptor density is suitably high, this patch can be used to measure release.
The membrane patch is voltage-clamped at a potential that provides a good driving force for ion-movement through the detector channels but also produces little
or no activation of voltage-gated ion channels. Using a second patch electrode
and amplier, a whole-cell patch-clamp recording is taken from the soma of the
cell to be studied. This cell is repeatedly stimulated to elicit an action potential
while the detector patch is moved close to the cell in order to probe for sites of
neurotransmitter release. Once a release site is located, the position of the detector patch is nely tuned to give an optimal response (Figure 13). The small size

302

Sombers and Ewing

Figure 13 Schematic of the basic sniffer-patch recording conguration used to detect


elicited release. (A) The initial step in the process is to excise a membrane patch from a
donor cell (not shown) expressing a high density of the desired receptors. This is then voltage-clamped at the desired potential using a suitable amplier (1). A second electrode and
amplier (2) are used to record from and to stimulate the cell of interest while the detector
patch is simultaneously used to probe the cell until a release site is detected (B and C). Typical records from an experiment showing the somal action potential (B) and subsequent
release of ACh detected from a rat magnocellular basal forebrain neuron (C). In this example, the detector consists of a patch of rat skeletal muscle membrane expressing embryonictype nicotinic ACh receptors voltage-clamped to 80 mV (C). Unitary channel currents are
visible on the decaying phase of the response. The expanded portion of the data shows a
small artifact preceding the rst channel opening. This is due to extracellular pickup of the
somal action potential by the patch electrode and conveniently can be used to estimate the
delay between stimulation and transmitter release. (Reproduced from Trends Neurosci.
with permission [33].)

of membrane patches (typically 23 m) makes them particularly well suited for


detecting release from discrete loci.
To date, the sniffer-patch technique has mainly been used as a highly sensitive but largely qualitative means of detecting transmitter release from neurons.
The list of transmitters detected in this way includes acetylcholine [33,34,3638]
adenosine 5-triphosphate [39,40], glutamate [4143], and GABA [44]. Mem-

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

303

brane patches can theoretically be used to detect the release of any neurotransmitter from any cell type, including PC12 cells, provided that an appropriate
source of receptors can be found to act as a detector. However, one must carefully
consider the receptor subtypes expressed by the donor cell, as subtype variation
could complicate the detection process.
If calibrated, the sniffer-patch technique can be used to make quantitative
measurements of neurotransmitter release. Unfortunately, however, calibration is
not a straightforward procedure as ligand-gated receptors exhibit nonlinear doseresponse relationships that are susceptible to concentration-dependent desensitization [33]. Dose-response curves can be constructed by measuring the frequency
of channel openings at different concentrations. Agonists can be applied using
rapid (<1 ms) solution exchange systems to minimize complications due to desensitization [33].
The detection of neurotransmitter release using the sniffer-patch technique
is a difcult and time-consuming process. Nevertheless, it is one of the few methods available for the detection of quantal non-electroactive neurotransmitter
release from discrete sites. In the future, the range of suitable donor cells available could be expanded by using molecular biological techniques to express specic receptors at high densities in selected cell types [33]. In addition, by coexpressing different receptor combinations in the same donor cell a single patch
could be used, in combination with selective antagonists, to study the transmission of more than one neurotransmitter [33]. This type of approach has been used
to detect the co-release of nucleotide and neurotransmitter in frog cutaneous pectoris nerve-muscle preparations, using dissociated guinea-pig sympathetic neurons as the receptor source [39].
Recently, another type of sensor has been developed to directly detect individual exocytotic events. Xin et al. have simultaneously measured Ca2+ and catecholamine following their secretion from individual cells using a multidimensional microsensor based on both electrochemistry and uorescence [45]. The
surface of a carbon ber microelectrode is modied with the uorescent dye calcium green-1 dextran, as this dye is a selective chelator for Ca2+. The uorescence
response linearly increases with bound Ca2+, and the large size of this molecule
prevents overconcentration of the reagent at the sensor tip. The dye is attached to
the tip of a carbon ber electrode by cross-linking with 5% glutaraldehyde.
The utility of this dual microsensor has been tested at bovine adrenal
medullary cells [45]; however, the technique has the potential to be used at other
cultured cell types, including PC12 cells. The sensor has a subsecond response
time for both Ca2+ and catecholamine concentration changes, and Ca2+ concentrations as low as 100 nM can be detected while the detection limit for catecholamine is in the micromolar range.
Exocytosis of catecholamines has been evoked by the transient application
of histamine, and detected by amperometry [45]. The simultaneous Ca2+ release
is measured by uorescence from the same sensor (Figure 14). Catecholamine

304

Sombers and Ewing

Figure 14 Experiments with a carbon ber electrode at a single adrenal chromafn cell
exposed to 50 M histamine for 5 s (indicated by the long bar). A 3 s exposure to a solution containing 100 nM Ca2+ and 50 M epinephrine was used to postcalibrate the
microsensor (shown by the short bar). In panel A, release from a single cell was evoked by
histamine exposure and resulted in the amperometric spikes due to catecholamine oxidation (lower panel). At the same time, Ca2+ release was detected. In panel B, neither catecholamine nor Ca2+ release was detected from a single cell after histamine exposure. I is
the uorescence intensity. Data were obtained using a 5m carbon ber electrode held at
0.65 V vs. SSCE. (Reproduced from Anal. Chem. with permission [45].)

secretion is observed as a series of sharp amperometric spikes arising from the


electrooxidation of the catecholamine content of individual vesicles. The uorescence signal, which rises slowly to a plateau and then slowly returns to background, reveals that Ca2+ levels increase at the cell surface at the same time as catecholamine secretion occurs. A faster-responding Ca2+ sensor is required to
determine whether spike-shaped Ca2+ concentrations can be observed. This type
of signal would suggest a vesicular origin for Ca2+.
The main disadvantage of this approach is the instability of the uorescent
signal, which renders the technique unsuitable for long-term use [45]. The stabil-

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

305

ity is only sufcient for experiments that can be completed in 100 s or less [45].
For this reason, the measurements are necessarily qualitative, as the short lifetime
of the uorescent sensor precludes calibration. The sensor is simply an indicator
that reports on an elevation in Ca2+ levels in the vicinity of the probe tip. If this
type of sensor is to make a signicant contribution in this type of application,
quantitative rather than simply qualitative measurements of transmitter release
are required.
III.

APPLICATIONS

A.

Latency of Release Following Chemical Stimulation

The quantitation of exocytosis events at PC12 cells has been used to investigate
the effect of different mechanisms of stimulation on the latency of exocytosis following stimulation [18]. It is well established that intracellular Ca2+ is generally
required for exocytosis. At sufcient levels in the cytoplasm, Ca2+ causes the
mobilization of vesicles to the plasma membrane where rapid exocytosis follows.
For the PC12 cell line, the entry of Ca2+ into the cytoplasm and subsequent exocytosis can be accomplished in three basic ways [18].
Elevated extracellular K+ levels are known to directly depolarize the cell
membrane and so trigger exocytosis by opening voltage-sensitive sodium channels that cause the subsequent opening of voltage-gated calcium channels
(VGCC) [18]. The opening of Ca2+ channels allows a rapid increase in the intracellular Ca2+ concentration to a level sufcient to trigger the mobilization of catecholamine-containing vesicles to the plasma membrane for exocytosis. The
application of nicotine, however, induces exocytosis in a different way [18]. This
mechanism involves the binding of the stimulant to a membrane nicotinic acetylcholine receptor. This ligand binding subsequently induces conformational
changes in the receptor, which opens channels permeable primarily to Na+. The
Na+ inux then causes sufcient depolarization of the cell membrane to open
voltage-sensitive Ca2+ channels, thus allowing the rapid inux of Ca2+ to induce
exocytosis. In contrast to both of these mechanisms, the application of muscarine
activates muscarinic acetylcholine receptors [18]. The signal is subsequently
transduced through intracellular second messengers to release Ca2+ from intracellular stores, thus triggering exocytosis.
These mechanisms are similar in that they each require Ca2+ for exocytosis;
however, differences in the latency period between stimulation and exocytotic
secretion are expected because those mechanisms with increasing degrees of
complexity involve various steps prior to the point where Ca2+ levels have
reached the threshold required for catecholamine exocytosis. Thus, the more
complex mechanisms should exhibit longer latencies between stimulation and
secretion. Accordingly, Zerby et al. report that application of elevated K+ results
in immediate and multiple current transients, which cease within 30 3 s (Figure

306

Sombers and Ewing

15A) [18]. The mean lag time between the stimulant administration and transmitter release is reported to be about 6 1 s for 17 cells (475 total events). In contrast, when 1 mM nicotine is administered to a cell, the release of catecholamine
is not immediate (Figure 15B) and continues for 55 10 s following a delayed
onset. The average latency between stimulation and secretion for nicotine has
been reported at 37 5 s for 16 cells (232 total events). Upon stimulation with
1 mM muscarine, the response is delayed to an even greater extent (Figure 15C)
and lasts a relatively long time (133 20 s). The mean delay time for release following muscarine stimulation has been calculated to be 103 11 s for 19 cells
(439 total events). The average vesicle catecholamine content is apparently unaltered by the mechanism of release, but the latencies obviously vary signicantly.
Figure 16 depicts the histograms created for data from the three mechanisms of stimulated release [18]. The 6 s latency before release following K+ stimulation apparently represents the diffusion time from the stimulation pipet. However, the relatively long latencies before the onset of exocytosis after nicotine and
muscarine stimulation are surprising, as exocytotic events generally occur on the
millisecond time scale. The longer times might reect a reduced number of Na+
channels on these cells, or a slow step in the G-protein coupling by these receptors in PC12 cells, or another potential rate-limiting mechanism [18]. Regardless,
the observed differences in latency could play an important role in intercellular
communication by providing another level of modulation in the processing of
neuronal signals.
B.

Release from Varicosities

PC12 cells have also been used as a model for developing neurons to investigate
changes in the location of exocytotic release following cell differentiation. The
PC12 cell line is an ideal model for studying neuronal differentiation and the
changes that occur during development and maturation, as these cells can be
induced to become more neuronal-like both morphologically and physiologically
by exposure to nerve growth factor (NGF). Thus, NGF-treated PC12 cells have
been used to probe the site of exocytosis during the differentiation process [3].
Upon treatment with NGF, PC12 cells extend processes, or neurites, from the cell
body. As the cells grow, these processes become more numerous and increase in
length and degree of branching, often intersecting with one another. Along the
neurites, bulbous varicosities appear that have previously been shown to contain
aggregates of vesicles that are each 2070 nm in diameter [46]. These spherical
varicosities, which are 14 m in diameter, develop all along the extensions.
Carbon-ber microelectrodes have been used to monitor the response of the
differentiated PC12 cells to stimulation with elevated levels of K+ (Figure 17).
The small size of the electrodes permits placement specically on the cell body,

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

307

Figure 15 Current-time traces for exocytosis at single PC12 cells. A 6 s ejection of stimulant [105 mM K+ (A), 1 mM nicotine (B), or 1 mM muscarine (C)] from a microinjector
was administered at each arrow. The resulting current transients correspond to the oxidation of DA at the electrode tip as it is released from the cell. Detection was performed using
a 5 m carbon ber electrode in the amperometric mode at 0.65 V versus SSCE. (Reproduced from J. Neurochem. with permission [18].)

308

Sombers and Ewing

Figure 16 Distributions of latencies observed between stimulation and secretion. The


time between the application of the stimulus and the detection of released dopamine for
each exocytosis event in the rst 40 s of release is plotted versus the percentage of the total
number of events observed for each type of stimulus. The differences in the time courses
n=
of release stimulated by 105 mM K+ (; n = 17 cells; 475 events), 1 mM nicotine (;
16 cells; 232 events), and 1 mM muscarine (; n = 19 cells; 439 events) are apparent. All
data were obtained using a 5 m carbon ber electrode held at 0.65 V vs. SSCE. (Reproduced from J. Neurochem with permission [18].)

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

309

Figure 17 Differentiated PC12 cells with varicosities. Bright-eld microphotograph of


4-day-old PC12 cells that have been exposed to 100 ng/mL levels of -NGF continually
from initial subculturing observed through the 40X objective of an IX70 Olympus microscope. The arrow points to a varicosity from which dopamine release is monitored. This
structure is completely covered by the electrode when it is placed on the varicosity.

smooth process or varicosity of a single cell, thus providing sufcient spatial resolution to allow the sensitive, site-specic, and time-resolved amperometric
detection of neurotransmitter release.
In one report, no release was detected from the cell body, smooth processes
or varicosities until after the sixth day in culture [3]. However, release is sometimes detected earlier than this. Experiments carried out on days 1014 show no
release from the cell body (n = 3), only very infrequent responses from the smooth

310

Sombers and Ewing

regions of the neurites (n = 5), and frequent release (Figure 18) when the electrode
is placed at a varicosity (n = 16) [3]. Release is either immediate upon stimulation
(n = 9) (Figure 18A), as observed at undifferentiated cells, or delayed by 54 10
s (n = 7) (Figure 18B). The duration of the delayed response is generally longer
than that of the immediate response.
Exocytosis from differentiated cells occurs preferentially at the varicosities
with very little release, if any, from the smooth processes or cell body [3]. The
absence of exocytotic events at the cell body of the differentiated cells could be
the result of changes in surface morphology and in the expression and distribution of ion channels in PC12 cells upon differentiation [3]. The absence of catecholamine release from the smooth processes is most likely due to the fact that
vesicles do not appear in these locations. The average vesicular catecholamine
content observed for exocytosis at varicosities is 178 9 zmol (107,000 molecules), which is not signicantly different from that observed at undifferentiated
cells; however, a more narrow distribution is observed for the NGF-treated cells
(Figure 19).
This study demonstrates that functional changes do indeed occur during
differentiation. Although the average vesicular catecholamine content appears to
be unaltered, the number of exocytosis events, the distribution of vesicle catecholamine content, and the site of release are all affected to some degree. It is possible that the narrow distribution in the amount released from varicosities may
indicate that larger vesicles are excluded from the relatively small bulbous
regions, thus resulting in a tighter distribution of vesicle radii [3].
C.

PC12 Cells as a Genetic Model: A Transfection Study

Amperometry at single PC12 cells has also been used in conjunction with a
genetic cell transfection protocol to examine the effects of toxin expression on
basal and evoked exocytosis. PC12 cells have been transfected with the specic
endoprotease Botulinum neurotoxin C1 light chain (BoNT/C1), which cleaves the
proteins syntaxin and SNAP-25 [5]. The molecular dissection of the mechanisms
underlying exocytosis has been motivated by the SNARE hypothesis, which postulates that exocytosis requires the assembly of the plasma membrane proteins
syntaxin 1, SNAP-25, and the vesicle associated membrane protein (VAMP) into
a complex [5]. This SNARE complex then acts as a receptor for cytosolic components of the proposed fusion machinery. Direct evidence for the role of the
SNARE proteins in neurotransmission comes from molecular genetic studies in
which syntaxin and VAMP have been shown to be required for neurotransmission
in Drosophila [4749] and Caenorhabditis elegans [50,51]. To assess the effects
of the disruption of SNARE proteins on exocytosis in PC12 cells, amperometry
has been used in conjunction with a genetic cell transfection assay to establish a

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

311

Figure 18 Representative current-time traces for exocytosis from varicosities of two differentiated PC12 cells. Potassium chloride (105 mM) was ejected (at arrows) from the
microinjector to perfuse the area surrounding the electrode and varicosity of interest. (A)
Immediate release of dopamine (DA) following potassium stimulation, representative of 9
of the 16 cells. The second stimulation and release indicates that the varicosity is still
responsive after initial stimulation. (B) Representative data from one of the seven cells that
exhibited a delay between the time of stimulation with 105 mM potassium chloride and
secretion of DA. The mean vesicle content in both immediate and delayed release is 178
zmol. The 5 m carbon ber electrode is held at a potential of 0.65 V vs. SSCE for all measurements. The inset is an example of a typical current transient observed at a varicosity
displayed on an expanded time scale. The release time (3 ms rise-time of the current spike)
is similar to that observed at an undifferentiated cell (Reproduced from Brain Res. with permission [3].)

312

Sombers and Ewing

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

313

new approach for evaluating the effects of changes in the function of specic proteins on the characteristics of individual exocytosis release events [5].
PC12 cells have been cotransfected with either a BoNT/C1-encoding plasmid or a control pcDNA3 plasmid, together with a plasmid encoding enhanced
green uorescent protein, EGFP [5]. This allows transfected cells to be identied
for subsequent quantitative amperometric measurements of individual catecholamine release events. Four to six days after transfection, the cells are loaded
with dopamine prior to the experiments to enhance the amperometric signal, and
the ATP-elicited amperometric responses from cells under three different conditions are compared. Nontransfected cells that have undergone the transfection
protocol but do not express EGFP show qualitatively similar responses to those
of normal control cells. EGFP-expressing cells that are cotransfected with the
control pcDNA3 plasmid in the absence of the toxin pBoNT/C1 also exhibit
responses similar to those of control cells, thus indicating that the expression of
EGFP alone has little or no effect on the response to ATP in terms of spike amplitude or frequency. In contrast, however, the response to ATP stimulation is dramatically altered for uorescent cells from dishes cotransfected with pEGFP and
pBoNT/C1 (Figure 20).
The transfection protocol does not seem to have any deleterious effects on
PC12 cells over the post-transfection periods, as BoNT/C1-expressing cells
detected by EGFP uorescence survive and are morphologically indistinguishable from nontransfected cells on examination by phase-contrast microscopy [5].
In contrast, the results of this study indicate that the expression of BoNT/C1 does
markedly inhibit exocytosis. The number of current spikes upon stimulation, as
compared to nontransfected or cells expressing only EGFP and the control plasmid cDNA3, is signicantly reduced. Furthermore, it has been determined that
this inhibition is not caused by a lack of dopamine uptake, as dopamine release
can be readily detected from BoNT/C1-transfected cells following mechanical
disruption of the plasma membrane.
The results of this study indicate that cell transfection, with subsequent
amperometric detection of exocytosis, allows valid and efcient analysis of the
roles of specic proteins on the quantitative and kinetic release aspects of individual exocytotic events. Studies utilizing this combined methodology, therefore,
hold the potential to provide a powerful tool for the future analysis of other components of the exocytotic machinery.

Figure 19 Distribution of vesicle content for potassium-stimulated release at varicosities plotted as the cubed root of catecholamine released. Plots of the percent of total events
observed in the rst 40 s following initiation of release vs. the cubed root of the amount of
catecholamine released upon elevated potassium stimulation for (A) 17 undifferentiated
PC12 cells (475 total release events) and (B) 16 differentiated PC12 cells (156 total release
events). (Reproduced from Brain Res. with permission [3].)

314

Sombers and Ewing

Figure 20 Comparison of single PC12 cells cotransfected with reporter pEGFP and
either control pcDNA3 plasmid (A), pBoNT/C1 (C) or a nontransfected cell from this condition (B). Cells were transfected with 20 g pEGRP and either 20 g control pcDNA3 or
15 g pBoNT/C1 and 5 g pcDNA3. Four to six days post transfection, cells were examined by uorescence microscopy. A uorescent control transfectant (A), a nonuorescent
untransfected cell from the toxin transfection dish (B), and a uorescent toxin-transfected
cell (C) were identied. Cells were loaded with dopamine and stimulated with a 3 s pulse
of 100 M ATP at 100 s following the start of the amperometric recording. All amperometric data were obtained using a 5m carbon ber electrode held at 0.70 V vs. Ag/AgCl
(Reproduced from Pugers Arch. with permission [5].)

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

D.

315

Chemical and Pharmacological Manipulation


of Messenger Vesicles

The PC12 cell line has also been used to probe the mechanisms of action of various neurologically active drugs. For instance, Sulzer et al. have investigated the
effects of amphetamine on exocytosis from PC12 cells [52]. Two distinct mechanisms of action have been proposed: (1) exchange diffusion [53] and (2) vesicle
depletion [54,55]. The exchange-diffusion model involves the attachment of
extracellular amphetamine to the dopamine transporter, and the subsequent transport of amphetamine into the cell while dopamine is simultaneously transported
out of the cell. This model appears to predominate at low concentrations of
applied amphetamine [52]. The second mechanism of action, apparently operational at high doses, involves the depletion of catecholamine from intracellular
storage vesicles followed by reverse transport out of the cell [52]. The amperometric detection of individual exocytosis events at PC12 cells provides a unique
method to probe this second mechanism.
To evaluate the effects of amphetamine on quantal size, PC12 cells have
been incubated in 10 M amphetamine for 10 min followed by K+ stimulation and
the amperometric detection of the amount of catecholamine released per event
(Figure 21) [52]. After exposure to amphetamine, the average amount of catecholamine released per exocytotic event is reduced to 48 5.4% of controls (n =
7). These results demonstrate that amphetamine can attenuate stimulus-dependent release by reducing quantal size, presumably resulting in a redistribution of
catecholamine to the cytosol. Cytosolic catecholamine is, in turn, rapidly released
by reverse transport. These ndings have important implications for neurotransmission, as they suggest that amphetamine affects synaptic transmission both by
increasing activity-independent reverse transport and by reducing the amount of
transmitter released during activity-dependent vesicular exocytosis [52]. The end
result is a decrease in the signal-to-noise ratio of stimulation-induced exocytotic
events. Thus, physiological and pharmacological alterations of quantal size may
well play an important role in the neuromodulation of catecholamine release.
Quantal size has also been investigated amperometrically using the tyrosine
hydroxylase product, L-3,4-dihydroxyphenyalanine (L-DOPA). The addition of
L-DOPA to cultured PC12 cells can markedly increase the levels of dopamine
within the vesicles [16]. Experiments conducted with nomifensine, a dopamine
transporter inhibitor, suggest that reverse transport does not make a signicant
contribution to release in these studies [16]. An amperometric trace for exocytosis after incubation with L-DOPA is shown in Figure 22A. Figure 22B shows the
cubed root histogram for release from both control and L-DOPA incubated cells.
Incubation with L-DOPA does not appear to increase the number or frequency of
exocytotic events; however, it does increase the average size of quantal release to
a value that is approximately 250% larger than that seen from control cells.
Pothos et al. (1996) report the average values for exocytosis from a single vesicle

316

Sombers and Ewing

Figure 21 AMPH exposure decreases the quantal amplitude of individual release events
from PC12 cells. Randomly chosen amperometric electrochemical records of quantal
release for (A) a control cell and (B) a cell that has been incubated in 10 M d-AMPH for
10 min. Cells were stimulated by local perfusion (30 nL of 1 mM nicotine in 105 mM KCl
saline) to induce vesicular exocytosis. (C) A histogram displaying the percentage of peak
sizes in paired control and AMPH-treated cells. Peaks are combined in intervals of 30,000
molecules. The lower limit of each bin size is shown on the abscissa. All data were obtained
using a 5 m carbon ber electrode held at 0.65 V vs. SSCE. (Reproduced from J. Neurosci. with permission [52].)

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

317

Figure 22 (A) Amperometric trace from a PC12 cell incubated for 40 min in 50 M
L-DOPA. Individual quanta are represented by the peaks after stimulation and the quantal
size is determined by the charge of each peak. The sharp inection in the baseline is due to
a 6 s perfusion application of the stimulation saline (arrow). The width at half height (t1/2)
is determined by measuring the width of individual peaks at a point halfway between the
baseline and maximal peak height. (B) Histogram of the cubed root of quantal sizes by percentage. L-DOPA exposure shifts the distribution of quantal sizes to the right. The lower
limit of each bin size is displayed on the abscissa. Data sets are not signicantly different
from normal distribution (Kolmogorov-Smirnov normality test; p > 0.05). All data were
obtained using a 5 m carbon ber electrode held at 0.65 V vs. SSCE. (Reproduced from
J. Neurochem. with permission [16].)

318

Sombers and Ewing

obtained from a PC12 control cell and one treated with L-DOPA are 125 zmol
(75,000 11,000 molecules) and 310 zmol (187,000 31,000 molecules) of catecholamine, respectively [16]. PC12 secretory vesicles can therefore be loaded
with catecholamine, via treatment with L-DOPA, to attain quantal sizes far
beyond their usual capacity.
Thus, it is clear that the quantal size of a vesicle depends on the extracellular conditions. Pharmacological manipulations of PC12 cells with L-DOPA and
amphetamine have both been shown to increase dopamine release; however,
these two chemicals work in different ways. L-DOPA acts to increase catecholamine content in the vesicles by increasing cytosolic dopamine synthesis,
whereas amphetamine acts to increase dopamine content in the cytosol by depleting it from the vesicles. These results are highly signicant, as they dene a mechanism for neuroplasticity whereby the amount of messenger released is modulated under different cellular conditions.
Additional studies have investigated the effects of other chemical agents on
the characteristics of exocytotic release. In an effort to directly observe a receptor-mediated mechanism that alters quantal size, the effect of autoreceptor activation on quantal exocytotic neurotransmitter release events has been measured
at PC12 cells via amperometry [56]. Although it is known that autoreceptors generally inhibit stimulation-dependent neurotransmitter release, it is unclear
whether the effects of autoreceptor stimulation are attributable to a decreased frequency of exocytotic release or a decrease in quantal size. In order to examine
this, PC12 cells have been exposed to the D2 agonist quinpirole, followed by the
amperometric detection of neurotransmitter release.
By comparing quantal sizes from multiple cells, quinpirole has been found
to reduce the average quantal size to about 51% of control values [56]. Quinpirole
also decreases the maximum amperometric spike amplitude (imax), the spike width
at half-height(t1/2) and the spike frequency. Each of these D2-mediated effects on
quantal size can be blocked by co-incubation in the D2 antagonist sulpiride; however, exposure to sulpiride alone does not affect quantal size and the individual
exocytotic event characteristics are very similar to those of controls.
It is known that the activity of the rate-limiting enzyme in the biological
synthesis of dopamine, tyrosine hydroxylase, can be modulated through a D2receptor mediated pathway. To determine whether quinpirole reduces exocytotic
quantal size by inhibiting this enzyme or by blocking the vesicular uptake of
cytosolic dopamine via inhibition of the vesicular monoamine transporter,
VMAT1, the tyrosine hydroxylase product L-DOPA has been measured in the
presence of an aromatic amino acid decarboxylase inhibitor that blocks the conversion of L-DOPA to dopamine [56]. Interestingly, it has been determined that
quinpirole decreases tyrosine hydroxylase activity to approximately 59% of control values, a value close to the percentage of the reduced quantal size reported

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

319

above. Thus, it has been postulated that quinpirole reduces quantal size by inhibiting the activity of tyrosine hydroxylase. Accordingly, the reduction in quantal
size is reversed by the product of the enzyme, L-DOPA, which elevates the average quantal size of release events.
Because dopamine itself is also a D2-agonist, one might assume that it
would exhibit quinpirole-like effects on quantal release; however, incubation in
50 M dopamine serves to increase, rather than decrease, the quantal size of individual release events [56]. Because dopamine is a universal agonist at all
dopamine receptor subtypes, it also activates the D1 receptor subtype present in
PC12 cells and serves as a substrate for the plasma membrane dopamine transporter and for the vesicular monoamine transporter, VMAT1. Thus, incubation in
dopamine does not provide a straightforward approach to the analysis of receptor-mediated modulations of quantal size.
These ndings provide direct evidence that the activation of neurotransmitter autoreceptors reduces both the quantal size and frequency of individual release
events, apparently by inhibition of neurotransmitter synthesis. This is an important and relatively unexplored mechanism of neuromodulation.
As previously discussed, amperometric results at PC12 cells have consistently shown that pharmacological manipulations utilizing the dopamine precursor, L-DOPA, increase the quantal amount of electroactive transmitter(s) released
[16]. Dramatic increases in the magnitude of individual current transients are
observed when PC12 cells are incubated in L-DOPA prior to amperometric experimentation. In contrast to these ndings, the described study utilizing fast-scan
cyclic voltammetry at PC12 cells has determined that the relative concentration
of electroactive transmitter(s) released from single PC12 cell vesicles is
unchanged following exposure to L-DOPA [13]. Assuming that the measurements
obtained with amperometry and FCV reect changes in the amount and concentration, respectively, of electroactive transmitter stored within single vesicles
prior to release, it can be hypothesized that when secretory vesicles increase their
neurotransmitter content they also increase their volume [57].
Initial pharmacological investigations combining amperometry with the
application of reserpine, an inhibitor of the vesicular monoamine transporter
(VMAT1) that displaces catecholamine from neurotransmitter vesicles, provides
further biological evidence that the volume of secretory vesicles may be modulated. When PC12 cells are incubated with 1 M reserpine, total release per stimulation diminishes to half its initial value after about ten minutes, and steadily
declines throughout the remainder of the experiment [13]. In addition, the frequency and amplitude of individual events diminish rapidly. A series of amperometric current transients are shown in Figure 23 for two PC12 cells. Initial conditions are identical for both cells; however, 2 min after the rst K+ stimulation the
medium surrounding the cell shown in the second column is adjusted to one con-

320

Sombers and Ewing

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

321

taining 1 M reserpine. Thus, it has been proposed that the VMAT-mediated modulation of quantal size ultimately affects the duration of quantal release and signaling, whereas neurotransmitter concentration remains relatively constant [13,57].
In order to further examine this possibility, amperometry and transmission
electron microscopy (TEM) have been used simultaneously to directly determine
if the L-DOPA and reserpine-mediated effects on quantal release are accompanied
by changes in vesicular volume [57]. Pharmacological manipulations utilizing
L-DOPA and reserpine directly affect the VMAT-mediated transport of catecholamines into PC12 vesicles [57]. Electron microscopy has been used to measure changes in the size of vesicles not actively involved in exocytosis, and these
measurements can be correlated with the amount of neurotransmitter released
using amperometry.
A same-cell paradigm for amperometry experiments has been used in
which the same electrode is used to measure release from a cell before and after
drug exposure. Ratios for spike area, half-width (t1/2), and maximum amplitude
(imax) have been created for cells by dividing the mean of log values after the
incubation period by the mean of log values before treatment [59]. In agreement
with previous work, Colliver et al. report that exposure of PC12 cells to 100 M
L-DOPA for 90 min signicantly increases spike area and t1/2 values [59].
Although the imax ratio is also increased by treatment with L-DOPA, this change
is not signicant when compared to that measured at control cells. When PC12
cells are treated for 90 min with 100 nM reserpine, a signicant decrease in each
of the aforementioned ratios results. In contrast to the effects independently
elicited by the drug treatments, when PC12 cells are simultaneously exposed to
both L-DOPA and reserpine, spike characteristic ratios are not signicantly different from those measured at control cells, which consistently measure close to
one. These results are summarized in Figure 24.
TEM images from single PC12 cells treated with each of these drugs are
shown in Figure 25. For each of the dense core vesicles depicted, a space between
the dense core and the vesicular membrane can be identied. This clear spatial
component is referred to as the vesicle halo [57]. Overall, the volume of the PC12
dense core vesicles tracks the changes in quantal size observed in response to
treatment with L-DOPA and/or reserpine. When PC12 cells are treated with

Figure 23 Effect of reserpine on quantal release from PC12 cells. In the rst column, a
single PC12 cell was repeatedly stimulated with 80 mM KCl for 3 s at the arrows; amperometric spikes were elicited in each case. In the second column, a different cell was treated
identically except that at 2 min following the rst stimulation, 1 M reserpine was added
to the medium. Subsequent stimulations indicate that quantal release is abolished by reserpine. All data were obtained using a 5 m carbon ber electrode held at 0.65 V vs. SSCE.
(Reproduced from Anal. Chem. with permission [13].)

322

Sombers and Ewing

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

323

Figure 25 Representative TEM images from single PC12 cells following treatment with
(A) physiological saline, (B) 100 M L-DOPA, (C) 100 nM reserpine, and (D) 100 M
L-DOPA and 100 nM reserpine for 90 min. A portion of the nucleus can be seen for the cells
shown in A and B (at the top and bottom, respectively). Scale bars in each gure represent
500 nm. (Reproduced from J. Neurosci. with permission [57].)

Figure 24 Summary of spike characteristic ratio values created at individual PC12 cells.
Under control conditions, there was an average of approximately 194,000 13 molecules
released per vesicle. For the ratios presented, an average of 169 27 and 102 12 amperometric values were used to determine pre and post means, respectively. Bars represent the
mean SEM of spike characteristic ratio values for the different experimental conditions
(control, n = 6; L-DOPA, n = 7, reserpine, n = 5; reserpine and L-DOPA, n = 5). Values
marked with *** are statistically different, with p < 0.001; ** indicates a signicant difference, with p < 0.01; and * indicates a signicant difference, with p < 0.05 vs. control (ttest). (Reproduced from J. Neurosci. with permission [57].)

324

Sombers and Ewing

L-DOPA, a dramatic increase in the volume of the dense core vesicle halo results
(Figure 25B). When the cells are treated with reserpine, a signicant decrease in
the size of the vesicle halo is apparent (Figure 25C). Similar to the amperometric
results, no signicant change in vesicle volume results when PC12 cells are
simultaneously treated with both drugs (Figure 25D). In addition, neither drug
signicantly affects the volume of the dense core itself.
These results suggest that PC12 vesicles can adjust their volume when
accommodating different amounts of neurotransmitter to regulate a relatively
constant neurotransmitter concentration. When combined with the results
reported from the fast cyclic voltammetry experiments, these ndings also suggest that the modulation of quantal release ultimately affects the time course of
receptor stimulation following exocytosis.

IV.

CONCLUDING REMARKS

Electrochemistry in ultrasmall environments is emerging as an increasingly


important technique for fundamental studies of single cell neurochemistry. Many
questions in neuroscience are currently being addressed using electrochemical
techniques for the investigation of exocytotic release. Several model systems and
experimental protocols have been developed utilizing amperometric or voltammetric measurements of exocytosis to elucidate various cellular functions related
to neurocommunication, as it is difcult to look at exocytotic events at single
mammalian neurons. This work is important to the scientic community in that it
is leading to new hypotheses concerning neurocommunication and providing new
methods to determine pharmacological mechanisms.
Understanding the chemistry of single cells is an area of great interest
because an understanding of the dynamics of single nerve cells should lead to better models of the cellular neurotransmission process. This type of information
promises to advance our knowledge of neurotransmitter storage, exocytosis, and
the physiological effects of external stimuli, such as drugs and toxins.
Although the methodologies and applications reviewed in this work have
provided valuable information, the future of exocytosis studies depends on the
development of smaller electrodes and a minimization of the limits of detection
with more sensitive instrumentation. The current voltammetric and amperometric techniques using microelectrodes are not able to detect very low concentration
species that may be involved in exocytosis. Furthermore, with the development
of smaller electrodes it will become possible to insert an electrode directly into a
functioning neuronal synapse to detect individual exocytosis events during
synaptic transmission. This type of experiment will allow investigators to test the
synaptic importance of concentration estimates for exocytosis developed from the
PC12 cell model. The goal of making measurements in single synapses of atto-

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells

325

liter volume represents the cutting edge of modern analytical science and a means
to discover a great deal more about how individual neurons in the brain communicate.
In addition to smaller probes and more sensitive measurements, future
experiments will require fast probes to measure the exocytotic release of nonelectroactive neurotransmitters. Most neurotransmitters and neuromodulators
(for example, glutamate, acetylcholine, glycine, vasopressin, and substance P) are
not easily oxidized, and at this time the methodology does not exist to allow quantiable real time investigation of exocytosis of these neurochemicals. The development of new technology to both enhance the experimental data currently
obtained and to achieve the detection of nonelectroactive neurotransmitters as
they are released in real time is necessary to further our understanding of the fundamental aspects of exocytosis in neurotransmission, including the role that pharmacologically induced neuroplasticity plays in exocytosis.
ACKNOWLEDGMENTS
This work was funded in part by grants from the National Institutes of Health and
the National Science Foundation. In addition, we gratefully acknowledge our
coworkers, past and present, for the studies cited in this review.
REFERENCES
1.

2.
3.
4.
5.
6.
7.
8.

9.
10.
11.
12.

Siegel, G. J. A., B. W.; Albers, R. W.; Fisher S. K.; Uhler, M. D., Eds. Basic Neurochemistry: Molecular, Cellular and Medical Aspects; 6th ed.; Lippincott Williams
and Wilkins, 1998.
Cooper, J. R. B., F. E.; Roth, R. H., Eds. The Biochemical Basis of Neuropharmacology; Oxford University Press, 1986.
Zerby, S. E.; Ewing, A. G. Brain Res 1996, 712, 110.
Shepherd, G. M. Neurobiology; 3rd ed.; Oxford University Press: New York, 1994.
Fisher, R. J.; Burgoyne, R. D. Pugers Arch 1999, 437, 75462.
Clark, R. A.; Ewing, A. G. Mol Neurobiol 1997, 15, 116.
Neher, E.; Marty, A. Proc Natl Acad Sci U S A 1982, 79, 67126.
Wightman, R. M.; Jankowski, J. A.; Kennedy, R. T.; Kawagoe, K. T.; Schroeder,
T. J.; Leszczyszyn, D. J.; Near, J. A.; Diliberto, E. J., Jr.; Viveros, O. H. Proc Natl
Acad Sci U S A 1991, 88, 107548.
Wightman, R. M. Analytical Chemistry 1981, 53, 1125A1134A.
Wightman, R. M. Science 1988, 240, 415420.
Wightman, R. M. W., D. O. In Electroanalytical Chemistry; Bard, A. J., Ed.; Marcel Dekker, Inc.: New York, 1988; Vol. 15.
Finnegan, J. M.; Pihel, K.; Cahill, P. S.; Huang, L.; Zerby, S. E.; Ewing, A. G.;
Kennedy, R. T.; Wightman, R. M. J Neurochem 1996, 66, 191423.

326
13.

Sombers and Ewing

Kozminski, K. D.; Gutman, D. A.; Davila, V.; Sulzer, D.; Ewing, A. G. Anal Chem
1998, 70, 31233130.
14. Ciolkowski, E. L. C., B. R.; Jankowski, J. A.; Jorgenson, J. W.; Wightman, R. M. J
Am Chem Soc 1992, 114, 28152821.
15. Chen, T. K.; Luo, G.; Ewing, A. G. Anal Chem 1994, 66, 30315.
16. Pothos, E.; Desmond, M.; Sulzer, D. J Neurochem 1996, 66, 62936.
17. Leszczyszyn, D. J.; Jankowski, J. A.; Viveros, O. H.; Diliberto, E. J., Jr.; Near, J. A.;
Wightman, R. M. J Biol Chem 1990, 265, 147367.
18. Zerby, S. E.; Ewing, A. G. J Neurochem 1996, 66, 6517.
19. Wightman, R. M.; Schroeder, T. J.; Finnegan, J. M.; Ciolkowski, E. L.; Pihel, K. Biophys J 1995, 68, 38390.
20. Bekkers, J. M.; Richerson, G. B.; Stevens, C. F. Proc Natl Acad Sci USA 1990, 87,
535962.
21. Van der Kloot, W. Prog Neurobiol 1991, 36, 93130.
21a. Jankowski, J. A.; Schroeder, T. J.; Ciolkowski, E. L.; Wightman, R. M. J Biol Chem
1993, 268, 1469414700.
22. Kennedy, R. T.; Huang, L.; Atkinson, M. A.; Dush, P. Anal Chem 1993, 65, 18827.
23. Coupland, R. E. Nature 1968, 217, 3848.
24. Alvarez de Toledo, G.; Fernandez-Chacon, R.; Fernandez, J. M. Nature 1993, 363,
5548.
24a. Brodwick, M. S.; Curran, M.; Edwards, C. Membrane Biol 1992, 126, 159169.
25. Pihel, K.; Hsieh, S.; Jorgenson, J. W.; Wightman, R. M. Anal Chem 1995, 67,
451421.
26. Helander, H. F.; Bloom, G. D. J Microsc 1974, 100, 31521.
27. Huang, L.; Shen, H.; Atkinson, M. A.; Kennedy, R. T. Proc Natl Acad Sci USA 1995,
92, 960812.
28. Larsson, L. I.; Sundler, F.; Hakanson, R. Diabetologia 1976, 12, 21126.
29. Schubert, D.; LaCorbiere, M.; Klier, F. G.; Steinbach, J. H. Brain Res 1980, 190,
6779.
30. Justice, J. J. Voltammetry in the Neurosciences; Humana Press: Clifton, N. J., 1987.
31. Anderson, B. B.; Zerby, S. E.; Ewing, A. G. J Neurosci Meth 1999, 88, 16370.
32. Kawagoe, K. T.; Jankowski, J. A.; Wightman, R. M. Anal Chem 1991, 63, 158994.
33. Allen, T. G. Trends Neurosci 1997, 20, 1927.
34. Hume, R. I.; Role, L. W.; Fischbach, G. D. Nature 1983, 305, 6324.
35. Young, S. H.; Poo, M. M. Nature 1983, 305, 6347.
36. Poo, M. M.; Sun, Y. A.; Young, S. H. J Physiol 1985, 80, 2839.
37. Young, S. H.; Grinnell, A. D. J Physiol (Lond) 1994, 475, 20716.
38. Tatsumi, H.; Katayama, Y. Neuroscience 1999, 92, 85565.
39. Silinsky, E. M.; Redman, R. S. J Physiol (Lond) 1996, 492, 81522.
40. Hollins, B.; Ikeda, S. R. J Neurophysiol 1997, 78, 306976.
41. Copenhagen, D. R.; Jahr, C. E. Nature 1989, 341, 5369.
42. Billups, B.; Szatkowski, M.; Rossi, D.; Attwell, D. Meth Enzymol 1998, 296,
61732.
43. Maeda, T.; Kaneko, S.; Akaike, A.; Satoh, M. Neurosci Lett 1997, 224, 1036.
44. Isaacson, J. S.; Solis, J. M.; Nicoll, R. A. Neuron 1993, 10, 16575.
45. Xin, Q.; Wightman, R. M. Anal Chem 1998, 70, 167781.

Electrochemical Monitoring of Exocytosis from Individual PC12 Cells


46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.

327

Greene, L. A.; Rein, G. Brain Res 1977, 129, 24763.


Wu, M. N.; Fergestad, T.; Lloyd, T. E.; He, Y.; Broadie, K.; Bellen, H. J. Neuron
1999, 23, 593605.
Yoshihara, M.; Ueda, A.; Zhang, D.; Deitcher, D. L.; Schwarz, T. L.; Kidokoro, Y.
J Neurosci 1999, 19, 243241.
Schulze, K. L.; Broadie, K.; Perin, M. S.; Bellen, H. J. Cell 1995, 80, 31120.
Saifee, O.; Wei, L.; Nonet, M. L. Mol Biol Cell 1998, 9, 123552.
Sassa, T.; Harada, S.; Ogawa, H.; Rand, J. B.; Maruyama, I. N.; Hosono, R. J Neurosci 1999, 19, 47727.
Sulzer, D.; Chen, T. K.; Lau, Y. Y.; Kristensen, H.; Rayport, S.; Ewing, A. J Neurosci 1995, 15, 41028.
Fischer, J. F.; Cho, A. K. J Pharmacol Exp Ther 1979, 208, 2039.
Sulzer, D.; Rayport, S. Neuron 1990, 5, 797808.
Sulzer, D.; Maidment, N. T.; Rayport, S. J Neurochem 1993, 60, 52735.
Pothos, E. N.; Przedborski, S.; Davila, V.; Schmitz, Y.; Sulzer, D. J Neurosci 1998,
18, 557585.
Colliver, T. L.; Pyott, S. J.; Achalabun, M.; Ewing, A. G.J Neurosci 2000, 20,
527682.

10
Milestones of Electrochemical
Immunoassay at Cincinnati
C. Ajith Wijayawardhana, H. Brian Halsall,*
and William R. Heineman*
University of Cincinnati, Cincinnati, Ohio

I.

INTRODUCTION

For over 20 years, our group has been developing voltammetric detection methods
for immunoassays. These assays have come to be generally referred to as electrochemical immunoassay (ECIA). The main impetus for this work has been the
exciting possibility of developing very sensitive assays by combining the excellent
sensitivity of voltammetry with the exceptional selectivity of antibody-antigen
binding. We have also been encouraged by the promise of ECIA for overcoming
problems associated with other major types of immunoassays. Radioimmunoassays, which use radioactive labels, pose health hazards and disposal problems; and
immunoassays with optical detection are often unsuitable for the analysis of colored or turbid samples. Electrochemical immunoassay is free of these problems;
and, moreover, has the advantage of being able to handle extremely small samples
because electrochemistry is an interfacial process in which the important reactions
occur at the electrode-sample interface, not in bulk solution.
Our early ECIA work focused on using electroactive functionalities and
metal ions as labels for electrochemical detection. Since then, the focus has primarily been to use enzyme labels that catalyze the formation of electroactive
products. The main advantage of using enzyme labels is the remarkable signal

*Corresponding authors.
329

330

Wijayawardhana et al.

amplication that may be gained by the high turnover of enzymatic product molecules per each enzyme label. Having demonstrated success with enzyme labels
in conventional immunoassay formats, we rened the immunoassay methodology
for carrying out ultrasensitive assays inside capillaries. This also marked the
beginning of our investigation on various miniaturized immunoassay platforms
aimed at extending ECIA to analysis in extremely small volumes, possibly with
multi-analyte detection capabilities.
It is our aim in this chapter to describe the various types of ECIA we have
developed and to highlight some of the major accomplishments. It is beyond the
scope of this chapter to discuss in detail all the different immunoassays that have
been developed in our labs, so we have tabulated their key features in Table 1. The
reader is referred to an encyclopedia chapter [1], book chapters [2,3], and journal
review articles [4,5] for work done by other groups in the eld.

II.

EARLY DAYS OF ELECTROCHEMICAL IMMUNOASSAY

The earliest immunoassays developed in our labs were homogeneous assays (no
separation steps) that used electroactive functionalities for labels. These immunoassays were limited to low-molecular-weight analytes (antigens) called haptens that showed little or no electrochemical activity over the potential range that
the label did. Although homogeneous ECIA assays based on nonmetal labels have
long since been surpassed by other more sensitive ECIA methods, they still serve
as good examples for understanding homogeneous immunoassays in general.
The analyte we chose for the rst immunoassay was estriol, a hapten in the
class of compounds called estrogens [6,7]. Estriol was particularly attractive
because of the large number of synthetic procedures available for its derivatization
with labels. In our work, estriol was derivatized with either mercuric acetate [6] or
nitro groups to form 2,4-dinitroestriol (DNE) [7]. The immunoassay with the nitro
group labels was based on the competitive binding of the analyte estriol and DNE
to a limited number of estriol-specic antibodies in solution. Key to the success of
the assay was the fact that unlike DNE free in solution, DNE bound to antibody
gives a diminished electrochemical signal. Therefore, in a sample containing a
high concentration of estriol where more DNE remained as free DNE, there was a
correspondingly high current. Similarly, low estriol concentrations led to smaller
currents. This fundamental difference between the electroactivity of free and
bound DNE is worthy of discussion as it serves to illustrate the principles of homogeneous immunoassay including the enzyme-based assays discussed later.
The loss of electroactivity of a labeled hapten like DNE upon binding to
antibody has been attributed to two sources. One is the possibility that upon binding, the electroactive groups become sequestered within the antibody, thus hindering the facile transfer of electrons between the electroactive groups and the

He,E,S
Ho,E,C

He,M,C
He,E
He,E,C
He,E,C
He,E,C

Rabbit IgG
Theophylline

Human serum albumin (HSA) and human-IgG


Spatially separated mouse IgG

Environmental testing
Atrazine
2,4 Dichlorophenoxycetic acid (2,4-D)
Indole-3-acetic acid
ALP/PAPP
ALP/PP
ALP/PAPP

ALP/PP
ALP/PP
ALP/PAPP
ElectroactiveNO2 groups
ALP/PAPP
Ni2+
ALP/PAPP
ALP/PAPP
ALP/PAPP
ALP/PP
ALP/PAPP
Glucose-6-phosphate
dehydrogenase/NAD+/2,6dichloroindophenol
ALP/PP
Glucose-6-phosphate
dehydrogenase/NAD+/2,6dichloroindophenol
Bi3+ and In3+
ALP/PAPP

Label/substrate

13
69

01.8, 0.6 g/mL : HAS, IgG


NA
Capillary-FIA-EC
0.1010.0 g/L
Bi-enzymatic recycling biosensor 0.0011000 g/L
Capillary-FIA-EC
3 pg/L

Anodic stripping voltammetry


Chronoamperometry

45
70
44

68
36

5600 molecules
2.540 g/mL

FIA-EC

Capillary-FIA-EC
FIA-EC

66
24
27
7
28
11
20
50
61
23
67
17

Reference

Differential pulse polarography


IDA electrode
RDE
SECM
LC-EC (carbon paste electrode)
Capillary-FIA-EC
FIA-EC

Limit of detection
and/or range
40500 U/L
50 pg/mL
101000 pg/L
0.022.2 mg/mL
0.163 g/L, 0.316100 g/L
333 nM
101000 ng/mL
505000 ng/mL
52500 ng/mL
1.010.0 ng/mL
303000 g/L
2.530 g/mL

FIA-EC
FIA-EC
Capillary-FIA-EC
Differential pulse polarography

Detection scheme

Source: Adapted from C. A. Wijayawardhana, H. B. Halsall, W. R. Heineman, in Encyclopedia of Electrochemistry (Vol. 8) (Eds. A. J. Bard, P. Stratmann), Wiley-VCH, New York, submitted.
ALP: alkaline phosphatase; C: competitive assay; E: enzime; FIA-EC: ow-injection analysis with electrochemical detection; GC: glassy carbon; He: heterogeneous; Ho: homogeneous; LCEC: liquid chromatography with electrochemical detection; N: non-enzyme; PAPP: 4-aminophenyl phosphate; PP: phenyl phosphate; RDE: rotating disk electrode; S: sandwich assay; SECM:
scanning electrochemical microscopy.

He,E
He,E,C
He,E,S
Ho,N,C
He,E,S
He,M,C
He,E,S
He,E,S
He,E,S
He,E,C
He,E,C
Ho,E,C

Assay type

Biomedical & pharmaceutical analysis


Alkaline phosphatase
Digoxin
Digoxin
Estriol
Human alpha-fetoprotein
Human serum albumin (HSA)
Mouse IgG
Mouse IgG
Mouse IgG
Orosomucoid (OMD, 1-acid glycoprotein)
Phenobarbital
Phenytoin

Analyte

Table 1 Selected Applications of Electrochemical Immunoassay

332

Wijayawardhana et al.

electrode. The other possibility is that the decrease in diffusivity of the labeled
hapten as it binds to the larger antibody is large enough to cause a decrease in the
diffusion-dependent voltammetric current. The comparison of some typical diffusion coefcients indeed suggests that this decrease can be signicant. For
example, the diffusion coefcient of DNE is about 100 Fick (1 Fick = 107 cm2/s)
[8]. In contrast, the diffusion coefcient of the larger DNE-antibody complex is
only about 1 Fick [9]. Therefore, it is evident that the above decrease in current is
at least in part due to a decrease in the diffusion coefcient. This is also corroborated by the fact that larger antigens such as IgG (MW ~ 150,000), which undergo
only a small fractional change in diffusivity upon binding to antibody, are difcult to detect with homogeneous ECIA assays.

III.

METAL LABELS

Our major motivation for using metal labels was that it opened the possibility of
applying the remarkably sensitive stripping voltammetric techniques [10] to
enhance the sensitivity of ECIA. The detection in stripping voltammetry is a twostep procedure. In the rst step, the metal ion is reduced to its metallic state at a
mercury electrode set at a constant reduction potential. Because the metal thus
formed is soluble in mercury, the rst step also results in a pre-concentration of
the metal in the electrode. In the second step, the potential is swept in a positive
direction whereupon the dissolved metal is oxidized and stripped from the electrode. The anodic current produced during the stripping is directly proportional to
the concentration of the metal ion in solution. Of the several different stripping
voltammetric techniques available, we chose differential-pulse-anodic-strippingvoltammetry (DPASV) because of its exceptional sensitivity. Detection limits as
low as 109 M to 1010 M are routinely obtained with DPASV [10].
Our search for a universal labeling protocol that could use several different
metal ions led to metal chelating groups that could act as bridges between metal
labels and antigens [11]. The successful use of chelating groups depended on several criteria. Of particular importance was that the chelating agent should bind the
metal ions tightly, but yet have a mechanism to release the ions for detection that
could be activated by some simple change in the sample conditions, such as the
pH. If the assay were to be homogeneous, the choice of the metal label, and hence
its matching chelating group, had to be limited to metals not present at detectable
levels in the sample itself.
The feasibility of using metal labels with chelating groups was rst tested
in a heterogeneous assay for human serum albumin (HSA) [11]. Diethylenetriaminepentaacetic acid (DTPA) and In3+ were chosen for the chelating ligand and
the metal ion label, respectively, based on their high formation constant (Kf =
1029) and the relative ease with which In3+ could be released by acidication. The

Milestones of Electrochemical Immunoassay at Cincinnati

333

chelation was done in two steps. First, DTPA was attached to HSA via the amino
groups of HSA using a four-step anhydride procedure [12]. In3+ was then added
to the HSA-DTPA mixture to form the HSA-DTPA-In3+ complex. The dialyzed
and puried HSA-DTPA-In3+ complex was found to be of high purity (95%) with
the remainder consisting mainly of aggregates that did not interfere in the assay.
The antigen thus labeled was ready for use in the immunoassay.
The HSA immunoassay, as shown in Figure 1, was based on a competitive
format where the labeled HSA and any HSA present in the sample competed for
a limited number of antibody binding sites. The bound complexes were then separated by centrifugation and resuspended for the release of In3+ for detection. The
DPASV voltammograms obtained from an immunoassay for HSA are shown in
Figure 2. The anodic wave corresponding to the released In3+ appears at 588
mV. The wave at 633 mV due to Cd2+ present in the reagents is sufciently
resolved from the In3+ peak and did not interfere with the detection of In3+. The
plot of the anodic current versus the concentration of HSA shown in Figure 3 has
the inverse relationship expected from a competitive assay. These results were
compared to the more common microbiuret-based immunoassay for HSA, and
the correlation was found to be linear except at low HSA concentrations. This was
attributed to the poorer detection limits of the microbiuret method.
An exciting extension of metal labelbased immunoassays is to use several
metal ions to carry out multi-analyte immunoassays. Because stripping voltammetric methods can detect up to six metals simultaneously, it should be possible to
detect just as many analytes simultaneously as well. This concept was successfully
demonstrated in a dual-analyte immunoassay for HSA and human IgG [13].
Rising concerns over the toxicity of mercury have made the use of mercury
electrodes unfavorable, and hence, the application of metal ionbased ECIA less
desirable. Yet this can change in the future as advances in both the design of mercury electrodes and the automation and miniaturization of systems using mercury
electrodes are expected to make the use of mercury electrodes safe [14,15]. However, as discussed next, enzyme labels that followed metal labels in ECIA are
often superior to metal labels. The niche for metal labels probably lies in
immunoassays where using enzymes is difcult because of unfavorable sample
conditions such as extreme temperatures to which metals are invulnerable.

IV.

ENZYME LABELS

Enzymes have become the most common type of label in ECIA owing to the signal amplication that can be derived from enzyme catalysis. Virtually all of our
current ECIA assays use enzyme labels. In enzyme-based ECIA, the enzyme
product is usually the species detected, and its concentration is directly or indirectly related to the analyte (antigen) concentration. The enzyme immunoassays

334

Wijayawardhana et al.

Figure 1 The protocol for metal-label based immunoassay for human serum albumin
(HSA) using releasable In3+.

Milestones of Electrochemical Immunoassay at Cincinnati

335

Figure 2 Differential pulse anodic stripping voltammograms from an HSA immunoassay as illustrated in Figure 1 for a series of HSA standards of increasing concentration
(AF) in 0.1 M citrate pH 1.5. Voltammetry conditions: hanging mercury drop electrode
(volume 0.37 L), deposition time 5 min, deposition potential 800 mV vs. Ag/AgCl scan
rate 2 mV s1, modulation amplitude 25 mV, clock time 0.5 s. (From Ref. 11).

developed in our lab have all relied on amperometric detection, although potentiometric and conductimetric methods have also been applied in ECIA to a lesser
degree [2,4].
A.

Electrochemical Detection

Five types of amperometric detection have been applied for enzyme-based ECIA.
They are ow-injection analysis with electrochemical detection (FIAEC), liquid
chromatography with electrochemical detection (LCEC), amperometric detection
with interdigitated array electrodes (IDA), rotating disk electrode (RDE) amperometry, and scanning electrochemical microscopy (SECM). Of these, the two
conventional types, FIAEC and LCEC, shall be discussed in this section, leaving
the discussion of the other types to Section V on miniaturized immunoassays.
Both FIAEC and LCEC have become powerful micro-analytical methods
in batch analysis, capable of handling samples as small as a few microliters. This

336

Wijayawardhana et al.

Figure 3 A plot of the anodic peak current (A) for the 588 mV indium wave vs. the
concentration of HSA (g/mL) in a series of standard solutions. (From Ref. 11.)

makes them ideal for electrochemical detection in conventional microwell-based


enzyme immunoassays where there are typically large numbers of samples to be
assayed routinely.
The basic apparatuses for FIAEC and LCEC are shown in Figure 4. The two
are identical except that for LCEC a chromatographic column is included
between its injection port and detector. In both FIAEC and LECE, the sample is
introduced to the system via the injection valve, whereupon the mobile buffer carries the sample through the electrochemical detector downstream. The most popular type of electrochemical detector is the commercially available thin-layer ow
cell of 120 L volume whose design provides optimum ow characteristics and
a high electrode surface-area-to-cell-volume for sensitive detection.
The chromatographic column allows LCEC to separate species in homogeneous assays that might otherwise interfere with detection. This is particularly
important in homogeneous assays where other separations steps are disallowed.
Furthermore, the chromatographic column can differentiate the analyte peak from
the ubiquitous capacitance peak caused by the difference in composition between
the sample buffer and the mobile phase. Owing to this, LCEC has a better detec-

Milestones of Electrochemical Immunoassay at Cincinnati

337

Figure 4 A schematic of LCEC (top) and FIAEC (bottom) instrumentation.

tion limit (109 M or 1015 moles for a 1 L sample) than FIAEC (107 M). These
advantages, however, are often offset by the signicantly longer detection times
required in LCEC. Therefore, FIAEC is the better choice for detection when neither high sensitivity nor chromatographic separation is crucial.
An important aspect of any kind of amperometric detection in enzymebased ECIA is the choice of electrode potential for detecting the enzyme product.
The optimum potential is obtained by plotting the oxidation or reduction current
response of the enzyme product as a function of electrode potential. In methods
such as FIAEC and LCEC, where convection is the main source of mass transport, such a plot is referred to as a hydrodynamic voltammogram. As shown in the
hypothetical plot in Figure 5 for an oxidizable product, a hydrodynamic voltammogram for an electroactive analyte has three distinctive regions: one, a current
response near zero that corresponds to the potential window in which the analyte
is electroinactive; two, a region of sharply rising current region dened by the
Nernst equation; and last, a limiting current plateau independent of the potential.
The limiting plateau current is directly proportional to the analyte concentration
and is described by the equation iL = nFADCo/d where iL is the limiting current, n
the number of electrons involved in the redox reaction, F the Faraday constant, A
the surface area of the electrode, D the diffusion coefcient, Co the bulk analyte
concentration, and d the thickness of the diffusion layer. The ideal potential for
detection is at the low end of the plateau region where small changes in the

338

Wijayawardhana et al.

Figure 5 A hypothetical hydrodynamic voltammogram for electroactive enzyme product and substrate.

applied potential will not signicantly affect the current response, and the possibility of co-oxidizing other species that leads to higher background signals is
reduced.
B.

Enzyme-Substrate Pairs for Electrochemical Immunoassay

Several enzyme-substrate (E-S) pairs have been used in ECIA, some of which are
listed in Table 2 with the respective enzyme products. The two enzymes that have
been used in our labs are glucose-6-phosphate dehydrogenase (G6PDH) for
homogenous assays and alkaline phosphatase for heterogeneous assays.
The enzyme G6PDH converts glucose-6-phosphate to 6-phosphoglucono-lactone (6-PG--L) in the presence of the cofactor NAD+ [16]. The progress of
the enzymatic reaction can be followed electrochemically by detecting the NADH
as it is formed. The relevant reactions are given below.
Enzyme reaction:
glucose-6-phosphate + NAD+ G6PDH 6-PG--L + NADH
3
Electrochemical reaction:

(1)

NADH 3 NAD+ + 2e + H+

(2)

The direct detection of NADH has several drawbacks. One, in the homogeneous assays for which this E-S pair is intended, the high oxidation potential for
detecting NADH often causes the oxidization of other species in the sample (e.g.,

Glucose-6-phosphate dehydrogenase
-Galactosidase

Horseradish peroxidase (HRP)

Alkaline phosphatase (ALP)

Enzyme label
4-aminophenyl phosphate (PAPP)
1-naphthyl phosphate
glucose-6-phosphate
4-hydroxynaphthyl-1-phosphate (HNP)
3-indoxyl phosphate
phenyl phosphate
5-bromo-4-chloro-3-indolyl phosphate ester
3,3,5,5-tetramethylbenzidine (TMB)
hydroquinone
NAD+ + glucose-6-phosphate
4-aminophenyl-beta-d-galactopyranoside (PAPG)

Substrate

Product
4-aminophenol (PAP)
1-naphthol
glucose
dihydroxy naphthalene
indigo blue
phenol
H2O2
TMB (ox)
Benzoquinone
NADH
4-aminophenol (PAP)

Table 2 Common Enzyme-Substrate-Product (E-S-P) Systems in Electrochemical Immunoassay

19
71
72
73
74
18
75
76
76
33
77

Reference

Milestones of Electrochemical Immunoassay at Cincinnati


339

340

Wijayawardhana et al.

antibodies, hemoglobin, glutathione, ascorbic acid, uric acid, and acetaminophen,


etc.), which contributes to a signicant background current. Another, when the
NADH generated in the enzymatic reaction reaches concentrations greater than
several micromoles per liter, it tends to foul carbon (graphite and glassy carbon)
electrode surfaces [16]. Thus, a large dilution of product or a very short incubation is required to keep the NADH concentrations low enough to avoid electrode
fouling. One solution to these problems is to add 2,6-dichloroindophenol (DCIP)
to the sample and allow NADH to rst react with DCIP, and then detect the product DCIPH2 at the much lower electrode potential of 0.20 V versus Ag/AgCl [17].
The enzyme that we have used the most is alkaline phosphatase (ALP), an
enzyme that hydrolyzes a number of phosphate esters (RHPO4) under alkaline
conditions:
Enzyme reaction:
RHPO4 + H2O ALP ROH + H2PO4
3

(3)

Electrochemical reaction:
ROH 3 R = O + ne + nH+

(4)

Phenyl phosphate was the rst enzyme substrate used for ALP in ECIA [18],
but the high oxidation potential of the enzyme product phenol and the susceptibility of phenol to foul electrodes by electropolymerization posed difculties during
detection and motivated us to nd a better substrate. Of the many phosphate esters
investigated, 4-aminophenyl phosphate was shown to have the best performance
characteristics [19]. Its enzyme product, 4-aminophenol, has an oxidation potential of 275 mV versus Ag/AgCl at glassy carbon, nearly 600 mV less than for phenol, and shows no signs of electrode fouling. Furthermore, 4-aminophenol, unlike
phenol, is electrochemically reversible, which has allowed us and several other
groups to apply new sensitive detectors that depend on reversibility for signal
amplication like the interdigitated array electrode (IDA) [20,21].
C.

Heterogeneous Enzyme Immunoassays

Heterogeneous enzyme immunoassays constitute the largest body of our groups


work. Many of our early heterogeneous assays were based on the standard
enzyme-linked immunosorbent immunoassay (ELISA) format, where the antibody is passively immobilized on the walls of microtiter wells or cuvettes [22].
The protocols of the two types of heterogeneous assays we performed, competitive and sandwich, are illustrated in Figure 6. In both, the rst step is attaching
the relevant primary antibody to the reaction surface covalently or by passive
adsorption. Blockers such as Tween 20, a nonionic surfactant, and bovine serum
albumin (BSA) are applied to the surface during and/or after antibody attachment
to block nonspecic adsorption (see Section IV.C.2).

Milestones of Electrochemical Immunoassay at Cincinnati

341

Figure 6 General protocol for competitive and sandwich enzyme immunoassays.

The terms competitive and sandwich of the two immunoassay types are
descriptive of each assay procedure. In competitive immunoassay, an aliquot of
enzyme-labeled antigen (Ag*) of known concentration is added to the sample for
competitive binding to the primary antibody. Following this, the reaction surface
is rinsed and the enzyme substrate (S) added. After allowing sufcient time for
enzymatic conversion, the sample is analyzed for the electroactive enzyme product (P). The concentration of P bears an inverse relationship to the analyte (Ag)
concentration because of the competitive binding of Ag*.
Sandwich immunoassays follow a very different protocol from that of competitive assays. Here, the Ag in the sample is allowed to bind freely to the primary
antibody on the surface. Following this, the reaction surface is rinsed and a second enzyme-labeled antibody (Ab*) also specic for Ag is added. After its incubation, the unbound Ab* is rinsed and the enzyme substrate added for electrochemical detection of P. In contrast to competitive assays, successful sandwich
assays will have a direct linear relationship between the measured P concentration and the Ag concentration because high numbers of Ag will result in correspondingly high numbers of Ab* retained in the vessel to generate more P, and
vice versa. Although sandwich assays require more reagents, incubations, and
rinses than competitive assays, sandwich assays are more selective owing to the
use of two antibodies, and more sensitive owing to the use of excess reagents.
1.

Competitive Enzyme Immunoassays

The rst two competitive enzyme immunoassays we developed were for the
macromolecule 1-acid glycoprotein (orosomucoid) [23] and the low-molecularweight hapten digoxin [24]. Both assays were carried out in polystyrene cuvettes

342

Wijayawardhana et al.

using ALP and phenyl phosphate for the E-S pair. The assay for digoxin used
both LCEC and FIAEC for detection and provided a good comparison of the two
methods.
Two key experimental parameters that require optimizing in competitive
assays are the incubation times for the sample and the enzyme substrate. For sample incubation, it is clear that the assay sensitivity is highest when the binding
between the antigen and antibody has reached equilibrium. We determined the
equilibration time for digoxin by incubating aliquots of ALP-labeled digoxin in
antibody-coated cuvettes for varying lengths of time. The amount of labeled antigen bound in each well was determined indirectly by adding enzyme substrate
phenyl phosphate to the rinsed cuvettes and electrochemically measuring the
amount of phenol formed after a xed incubation time. The results, as shown in
Figure 7, show that although it takes about 3 hours to reach equilibrium, shorter
incubations could be used to decrease the assay time in situations where very high
assay sensitivities are not mandatory.
The second parameter, substrate incubation time, was also optimized using
aliquots of labeled antigen. The results showed that as long as the substrate incu-

Figure 7 Plot of LCEC chromatogram peak current vs. hapten/antibody incubation


interval obtained by using a 0.0 ng/mL digoxin standard. (From Ref. 24.)

Milestones of Electrochemical Immunoassay at Cincinnati

343

bation is done under substrate saturation conditions, the amount of enzyme product formed increased linearly with the incubation time. Therefore, the investigator once again has the option of longer incubations for assays requiring high sensitivity, or shorter incubations for assays that do not.
The LCEC chromatograms from the digoxin immunoassay using optimum
conditions thus obtained are shown in Figure 8. The results indeed show a very sensitive assay throughout the therapeutic range with a detection limit of 50 pg/mL.

Figure 8 Heterogeneous enzyme immunoassay with LCEC detection for a series of


digoxin standards in plasma solutions. Concentration of digoxin in plasma samples was (A)
5.0, (B) 2.0, (C) 1.0, (D) 0.5, and (E) 0.0 ng/mL. Assay conditions were as follows: 10
g/mL primary antibody concentration, 1/125 digoxin-ALP conjugate dilution, 4-h antigen/antibody incubation interval, and 40-min substrate reaction time. (From Ref. 24.)

344

Wijayawardhana et al.

The relative standard deviation for any given assay concentration ranged from 4%
to 12% on various days. The immunoassay with FIAEC detection showed inferior
sensitivity to that of LCEC (about a 100-fold less limit of detection) owing to the
same detrimental effects of the capacitance current noted earlier. However,
because of the extra chromatographic separation step, the detection time for LCEC
at 2.5 min was signicantly longer than that for FIAEC at 30 s.

2.

Sandwich Enzyme Immunoassays

Sandwich immunoassay, as noted earlier, can be used with greater sensitivity on


larger antigens than can be achieved with competitive assays. The limitation of
sandwich assays to large antigens is due to the necessity for sufcient surface area
on the antigen to accommodate two antibodies simultaneously. The analyte that
we used to demonstrate ECIA sandwich assays was IgG, either rabbit or mouse
[25,26]. These assays, as well as some of the later ones including assays for
digoxin [27] and human -fetoprotein [28], were done in cuvettes or microtiter
wells. Heterogeneous sandwich assays developed using capillaries and
microbeads are discussed in Section V.
The calibration curve for a rabbit IgG sandwich immunoassay using optimum antigen/antibody incubation times and dilutions and a 20 min enzyme substrate incubation is shown in Figure 9. The concentration of IgG ranged from 0 to
250 ng/mL with a detection limit (the analyte concentration that yields a signal
equal to the blank plus two standard deviations) of 100 pg/mL. The coefcient of
variation for 10 repetitions with the 10 ng/mL standard was 3.8%. In a separate
experiment, the detection limit was lowered to 50 pg/mL when the enzyme substrate incubation was extended to 60 min. These results were proof of the remarkable sensitivity of sandwich enzyme ECIA and motivated us to obtain even lower
detection limits.
The detection limit in the above assay was dened not by the ability to
detect phenol, but by a relatively large blank signal. After verifying that this signal did not arise from either the contamination of any of the biochemical preparations with AP or the presence of phenol generated by nonenzymatic hydrolysis
of phenyl phosphate, we investigated the possibility of Ab* binding nonspecically to the surface.
Nonspecic adsorption (NSA) of proteins is a well-known phenomenon in
heterogeneous assays [29], and NSA due to Ab* can occur by its direct binding
to the cuvette surface and/or to some surface-bound species other than the Ag.
One possibility of the latter is the binding of Ab* to the primary antibody. If it
were so, then increasing the density of the primary antibody on the surface ought
to increase the blank signal. However, our data showed that the opposite was true:
NSA decreased with an increase of the population of primary antibodies on the

Milestones of Electrochemical Immunoassay at Cincinnati

345

Figure 9 Standard calibration curve for sandwich enzyme immunoassay for rabbit IgG
using the ALP-phenyl phosphate E-S pair. The bracketed point is the zero dose response.
(From Ref. 25.)

surface. This led to the conclusion that the blank signal was due to the Ab*
adsorbing nonspecically to the cuvette surface, not to the primary antibody.
Direct binding of proteins to surfaces is surface dependent but is generally
a combination of entropically driven hydrophobic interactions with nonpolar
areas of surfaces, and electrostatic interactions, even with plastics such as polystyrene [30]. Tween 20 and BSA are two reagents commonly used to block
hydrophobic interactions [31,32]. Our results showed that the presence of Tween
20 in all assay solutions including the primary antibody solution decreased NSA
nearly 20-fold compared to assays using no Tween 20 at all. Furthermore, the
absence of Tween 20 in the primary antibody solution caused nonspecic adsorption to double, suggesting that Tween 20 served in some way as a preconditioning agent during the long primary antibody incubation step.
The benets of using BSA as an additional site blocker are illustrated in Figure 10. When BSA is present in all solutions after the primary antibody coating
(Figure 10A, B), conjugate adsorption decreased an additional 96% from that
observed with using only Tween 20 (Figure 10D). The preconditioning effect of
Tween 20 is evident here, too, in the increase of NSA when Tween 20 is absent

346

Wijayawardhana et al.

Figure 10 Nonspecic adsorption of the conjugate (Ab*) with and without the use of
BSA. The primary antibody was present in either phosphate buffer (PB), or phosphate
buffer with 0.05% Tween 20 (PBT), or phosphate buffer with 0.05% Tween 20 and 1%
BSA (PBTBSA). Rinse + refers to the buffer components of all solutions beginning with
the rst rinse solution. (From Ref. 3.)

from the primary antibody solution even when BSA is present in all other solutions
(Figure 10C). One difference between BSA and Tween 20 we observed was that
BSA, unlike Tween 20, slightly hindered the binding of primary antibody to the
cuvette surface by competing for the same binding sites. Hence, in all future
assays, Tween 20 (0.5% v/v) was used in all reagent solutions, and BSA (1% w/v)
in all except the primary antibody solution. These improved solution conditions for
minimizing NSA resulted in a 13-fold improvement in the detection limit for IgG.
It became clear with these results that ECIA could be as good, and in some
cases better, than other clinical immunoassays. However, it also became clear that
if we were to improve the assay sensitivity even further, there would need to be
some fundamental changes made in the assay format as the levels of enzyme
product generated at the detection limit in the above assays had reached the detection limit of electrochemical detection of the product. This led to the capillary
immunoassays described separately in Section V.A.

Milestones of Electrochemical Immunoassay at Cincinnati

D.

347

Homogeneous Enzyme Immunoassays

Homogeneous enzyme immunoassays are more attractive than heterogeneous


immunoassays because of the simpler and shorter assay protocol. Yet they are
more challenging than heterogeneous assays because removing the effects of
potential interferents is nontrivial in the absence of separation steps.
The rst homogeneous enzyme immunoassay developed in our group was
for phenytoin (a drug for the treatment of epileptic seizures) using the G6PDH
and NAD+ enzyme-cofactor system [33]. The product NADH was determined by
oxidation using LCEC consisting of a thin-layer cell with a glassy carbon electrode held at 0.75 V versus Ag/AgCl. A reverse-phase C-18 chromatographic column was required to remove proteins in the samples that would otherwise have
passivated the electrode. Furthermore, the serum samples had to be diluted 121fold before the assay because the amount of NADH generated with undiluted
samples exceeded the upper limit of the NADH calibration curve within a few
seconds. Although the necessity of a dilution step complicated the immunoassay
for phenytoin, it suggested that in situations where the therapeutic level of the
analyte was much smaller than for phenytoin, it should not be necessary to dilute
the samples.
This was tested in an assay for digoxin [34], a cardiovascular drug for the
treatment of chronic heart disease, which has about a 1,000-fold lower therapeutic level than phenytoin. An improved chromatography system that involved a
column-switching valve placed between the sample injection port and the
reverse-phase C-18 column was used in the assay. The valve contained a small
size exclusion column whose eluate could be switched to either (1) the reversephase column and the electrochemical detector or (2) the waste container. In the
assay for digoxin, the undiluted serum sample, which contained high levels of
proteins, was rst passed through the size exclusion column. The early eluting
proteins were vented to waste. Then the size exclusion column was switched on
line to the reverse-phase column in which the NADH could be separated from
electroactive interferences for detection. This electrochemical assay demonstrated a linear dynamic range from 0.55 to 7.5 ng/mL, which encompasses the
therapeutic range (0.9 to 2.0 ng/ml), and had a high level of precision (2.3% relative standard deviation) and a shorter enzyme incubation time (5 min) than for
spectrophotometric detection (30 min). The analysis time for one sample was less
than 12 min. The analysis of 46 human serum samples from patients receiving
digoxin therapy gave a good correlation (0.942) with radioimmunoassay using
sample values of 0.3 to 4.1 ng/mL.
The next major improvement in homogeneous enzyme ECIA came from coupling NADH with 2,6-dichloroindophenol to lower the detection potential as
described in Section IV.B. This method has been applied in immunoassays for

348

Wijayawardhana et al.

phenytoin [17] and theophylline in whole blood [35], and theophylline in hemolyzed, lipemic, and icteric sera [36]. Being able to do immunoassays directly on
blood or serum samples is seen as a signicant advantage over homogeneous immunoassays with spectrophotometric detection where detection is severely hampered by spectral interferences caused by the turbidity and/or the color of samples.

V. MINIATURIZED ENZYME ELECTROCHEMICAL


IMMUNOASSAYS
The effects of reducing the volume of the reaction vessel on the detection limit
and the speed of enzyme immunoassays can be profound. This arises mainly from
the fact that in smaller volumes where enzyme product dilution is minimized, the
product concentration can be brought to detectable levels more rapidly than in
larger volumes. Also, the smaller dimensions that accompany smaller volumes
make the distances that species have to diffuse in solution before reacting at the
heterogeneous surface correspondingly smaller, thus allowing shorter incubations, and hence faster immunoassays. The importance of miniaturized assays,
however, extends far beyond just increasing assay sensitivity or speed. The ability to carry out immunoassays in extremely small volumes can be very important
in situations where minimizing the sample volume is absolutely necessary such
as is in the testing of neonates or critically ill patients, or in the analysis of physiologically restricted sites such as the brain, eye, or spine [37].
Our rst successful effort at miniaturizing assays came with capillary-based
immunoassays. More recently, we have developed several miniaturized immunoassays using paramagnetic microbeads to serve as the heterogeneous surface. Currently we are investigating different automated electrochemical immunosensor systems with the aim of extending ECIA to lab-on-a-chip systems.
In the remainder of this chapter, we will present the accomplishments in each of
these areas.
A.

Capillary Enzyme Immunoassays

Microcapillaries are an excellent choice for reaction vessels in miniaturized heterogeneous enzyme immunoassays because of their high surface-area-to-volume
ratios that ensure small volumes while providing a high surface area for the heterogeneous reactions. Although capillary immunoassays with optical detection
preceded our use of capillaries in ECIA [38], our goal of pushing the detection
limit as low as possible required a number of special renements with respect to
the immobilization of primary antibodies and the minimization of NSA.

Milestones of Electrochemical Immunoassay at Cincinnati

349

The sensitivity of a heterogeneous enzyme assay can be greatly affected by


the way in which the primary antibody is immobilized on the surface. In particular, it is important that antibodies be immobilized such that their binding sites are
sterically unhindered to capture the antigen. In classical adsorption methods
where the antibody adsorbs in a rather random fashion, the fraction of the total
primary antibody adsorbing with an orientation suitable for efcient antigen binding can be small. One way to overcome this problem is to bind the antibody covalently to the surface. Unfortunately, most convenient covalent attachment methods popular at the time, including the glutaraldehyde immobilization method,
were unsuitable for this because they required the coupling via the amino groups
of the protein, which includes the N termini located in the antigen-binding
domain of the antibody [39]. Therefore, an early focus in our capillary assays was
to develop an immobilization technique that made the attachments via the carboxyl termini of the antibody located at the opposite end of the molecule. This
was achieved by a multistep procedure in which the inner wall of the capillary
was rst treated with poly(vinylbenzyl chloride) to provide a robust polymer
base, which was subsequently treated with tris(2-aminoethyl) amine to substitute
the chloride with amine groups to enable the application of one of the convenient
amine-to-carboxyl coupling procedures for binding the antibody [40]. The actual
coupling, as shown in Figure 11, was done using carbodiimide (EDC) chemistry
[41] where the antibody (protein) was activated with EDC prior to coupling. A
simplex optimization method [42] was used to determine the optimum concentration of EDC and the reaction time.
Capillaries thus prepared had one serious problem in that the electrostatic
interactions between the Ab* and the unreacted positively charged amine groups
of the capillary gave rise to signicant NSA. Therefore, several of the ion-pairing
agents used in ion chromatography for the complexation of amino groups were
investigated for their effectiveness in removing the electrostatic interactions.
Phosphates and phosphonates, though very effective pairing agents, could not be
used because of their inhibition of ALP activity. Therefore, sulfonates and sulfates that have intermediate formation constants with amino groups, but do not
inhibit enzyme activity, were chosen for the ion-pairing agents. These agents did
indeed reduce NSA arising from electrostatic interactions, but the alkyl groups of
the ion-pairing agents gave rise to NSA due to hydrophobic interactions. Pentane
sulfonate, with its moderately long alkane chain, yielded the best balance between
these opposing effects and was chosen for the ion-pairing agent in our capillary
immunoassays.
Capillaries modied as above have been used in several heterogeneous
enzyme immunoassays, both competitive and sandwich. The calibration curve for
a rabbit IgG assay in human serum controls is shown in Figure 12. The zero doseresponse was 7.5 nA using the ion-pairing agent pentane sulfonate. The nal

350

Wijayawardhana et al.

Figure 11 Reaction of the activated carboxyl group of the protein with the amino groups
on the inner surface of the capillary (From Ref. 3.)

assay required 30 min working with 70 L samples and had a detection limit of
121 fg/mL or 5.6 1020 moles.
Another successful way of attaching antibodies to capillaries has been to
use capillaries modied with polyethylene glycol (PEG) linkers containing amine
termini that can be coupled to the antibody glycan chains [43]. These capillaries
have several advantages. Coupling the PEG to the glycan reduces the possibility
of deactivating much of the antibody as occurs in passive adsorption, and the exibility of the long PEG chain provides movement to the antibodies that increases
the capture efciency of the antibody. The PEG linker decreases nonspecic
adsorption (NSA) of the Ab* to the capillary wall, and PEG added to the solution
lowers NSA further, presumably by acting as a competitive inhibitor of interactions between the Ab* and the linker PEG [43]. Capillaries modied with PEG
have been used in immunoassays for digoxin [27], indole-3-acetic acid [44], and

351

Log Current, nA

Milestones of Electrochemical Immunoassay at Cincinnati

Log Concentration, pg/mL

Figure 12 Final calibration curve for the rabbit IgG capillary assay performed in human
serum controls using pentane sulphonate as blocking agent. The bracketed point is the zero
dose response.

atrazine [45]. In general, capillary ECIA has yielded some of the most sensitive
immunoassays.
B.

Paramagnetic Microbead-Based Electrochemical


Immunoassays

Paramagnetic micro-immunobeads of 13 m diameter have gained popularity in


the past decade as a mobile solid phase for heterogeneous assays [46,47]. Our
main reason for using microbeads was a proposal to use them as the heterogeneous surface in the extension of ECIA to automated lab-on-chip systems, (Section V.D). Apart from this long-term goal, we were also interested in exploring
other ways of using microbeads in ECIA.
The main advantage of using beads is that they can be discarded from a
reaction vessel after an assay so that the vessel can be reused using fresh aliquots
of beads. Also in contrast to capillary assays, bead-based assays raise the interesting possibility of using large sample volumes to enhance the sampling capacity without compromising assay detection limit from a large enzyme product dilution. This is because the beads can be dispersed in large volumes during sampling,
but collected and concentrated in small volumes during enzyme substrate incubation for detection. Furthermore, when beads are dispersed in the sample by gentle shaking, they become active participants in capturing target molecules, thus
eliminating the dependence on the slow processes of diffusion for the major

352

Wijayawardhana et al.

means of mass transport. Also, the high surface area-to-volume ratio of


microbeads provides a far greater number of total receptor sites than is obtainable
using a reaction vessel identical in size to that used to hold beads but where the
receptors are bound to the walls.
The bead-based ECIAs developed thus far in our labs have used polystyrene culture tubes to hold the beads. Sandwich enzyme immunoassay was chosen for the assay format, and Figure 13 gives a schematic representation of the
experimental setup. The details of the assay procedure including the various
buffers, the sample volumes, and incubation times can be found elsewhere [48].
The fact that beads are a mobile heterogeneous surface holds several advantages during the immunoassay. This suggested to us the possibility of using a new
detection scheme for bead-based assays that could have a signicant advantage
over the two popular methods FIAEC and LCEC. In cuvette- and microwellbased assays, it was necessary to add enzyme substrate to the vessels, allow for a
suitable incubation time, and sample each container with FIAEC or LCEC. With
microbeads, on the other hand, there was the possibility of combining these three
steps of electrochemical detection in one by adding the beads directly to an electrochemical cell that already contained the enzyme substrate. The search for an
electroanalytical technique that could best harness this advantage of microbeads
led us to amperometric detection with rotating disk electrodes (RDE).
Amperometric RDE detection is one of the most sensitive electrochemical
detection methods available [49]. Moreover, with respect to its application for
detection in bead-based assays, it has the additional advantage of having a rotating
electrode that could efciently mix the beads in solution. However, the conventional RDE experiment, which requires a relatively large volume of sample, posed
a problem to bead-based immunoassays because of the adverse affects of enzyme
product dilution already cited. We circumvented this problem by applying the
RDE electrode directly to a microdrop of enzyme substrate held between the base
of the electrode and a hydrophobic surface as shown in Figure 14. The hydrophobicity of the surfaces and the high surface tension of the drop were able to contain
drops of 20150 L in a symmetric bulge at electrode spin rates up to 3000 rpm
[50]. When necessary, reagents were added to the drop by simply pipetting them
directly to the drop as shown for the addition of microbeads in Figure 14. At the end
of the experiment, a magnet was brought in contact with the drop to remove all the
beads and thus prevent them from contaminating the electrodes. The evaporation
effects that can be considerable in such small volumes were minimal here because
of both the short times involved and the controlled humidity of the lab. The RDE
setup was renewed for a fresh test by blowing the drop along the hydrophobic surface to waste with a stream of air.
We have carried out several successful sandwich immunoassays in this
way, including assays for mouse IgG [50], ovalbumin [51], and an articial
microorganism called the Bugbead [52]. To illustrate the features of bead-based

Figure 13 Schematic of the overall sandwich immunoassay based on paramagnetic beads. (From Ref. 3.)

Milestones of Electrochemical Immunoassay at Cincinnati


353

354

Wijayawardhana et al.

Figure 14 Schematic of the experimental setup for the rotating disk electrode (RDE)
detection in a bead-based immunoassay using the ALP-PAPP enzyme-substrate pair.

immunoassays with RDE, we will focus on the Bugbead assay. The Bugbead was
developed as a simulant for potentially lethal pathogenic microorganisms in the
early testing of immunosensor systems (Section V.D.) where direct handling of
pathogens had to be avoided for safety concerns. Briey, the version of the Bugbead used here is a Neutravidin-dendrimer coated bead of 0.39 m diameter modied with biotinylated mouse IgG and guinea pig IgG. In the sandwich immunoassay, mouse IgG on the Bugbead was recognized by 2.8 m paramagnetic
beads coated with the primary antibody, and guinea pig IgG by anti-guinea pig
IgG conjugated to alkaline phosphatase. Further details of the Bugbead and its
preparation have been presented elsewhere [52], and we will focus here only on
describing the general features of RDE in bead-based assays.
The RDE current-time plots from the Bugbead immunoassay using the
detection scheme described above are shown in Figure 15. The curves correspond
to assays of Bugbead standards ranging from 18 to 1.2 105 Bugbeads/L. The
curves for the low concentrations are shown in larger scale in the inset. All the

Milestones of Electrochemical Immunoassay at Cincinnati

355

Figure 15 Current-time plots for a bead sandwich immunoassay with RDE detection for
the model analyte Bugbead. (From Ref. 52.)

curves have near zero currents in the beginning. The slight negative current step
at the point of the enzyme substrate PAPP addition is due to the presence of trace
amounts of PAP formed by non-enzymatic hydrolysis of PAPP. However, this
current decays rapidly upon the exhaustive oxidation of the trace amounts of
PAP. The sharp change in the current prole, which occurs at the point of the
addition of beads, marks the beginning of enzymatic conversion of PAPP to PAP.
The expected trend from a successful sandwich immunoassay, therefore, would
be a PAP oxidation current that grows proportionally to the antigen concentration.
This indeed is observed in Figure 15. The minimum detected Bugbead concentration is 180 Bugbeads/L, and because only 2 L of this sample were tested, the
absolute number of Bugbeads detected was 360.
The ability to combine small volumes with excellent sensitivity has made
amperometric detection with RDE ideal for bead-based immunoassays. Direct
comparison of the results obtained in the above Bugbead assay with those
obtained with FIAEC detection showed that the detection limit of RDE was 2
orders of magnitude better than for FIAEC [53]. The other important feature of
RDE is the extremely short detection time that could be reached by combining
enzyme turnover with electrochemical detection. The RDE detection time of 2
min in the above Bugbead assay is over 10 times faster than the typical detection
time for FIAEC that requires a 20 min enzyme substrate incubation. It is also clear

356

Wijayawardhana et al.

from Figure 15 that the detection time could easily be further reduced. Indeed,
this has been accomplished in other assays where the detection time was reduced
30-fold over that of FIAEC detection [50].
C.

Immunoassays with Scanning Electrochemical


Microscopy (SECM)

One of the most promising areas of miniaturized immunoassay has risen with the
recent remarkable advancements made in methodologies for patterning surfaces
with biologically active microdomains (e.g., ink-jet technology [54], photolithography [55], screen-printing [56]), and scanning probe microscopies (SPM) to
probe such microdomains. The formation of several different biologically active
regions in closely packed arrays is also at the forefront of the burgeoning area of
multi-analyte detection in biosensors based on both conventional immunoassay
formats and the new technologies of genomics and proteomics. With such promise, we began the investigation of applying scanning electrochemical microscopy
(SECM), the SPM of choice in electrochemistry, for miniaturized heterogeneous
immunoassays [57].
Scanning electrochemical microscopy has become a powerful tool for
imaging the electrochemical activity of surfaces [58,59]. Two major modes of
operation can be distinguished: feedback mode and generation-collection (GC)
mode. In the GC mode used in our work, a probe microelectrode placed very close
to the surface is scanned two-dimensionally over the surface with the aid of inchworm actuators or stepper-motors. The computer records the current of the probe
microelectrode to give the SECM image. The SECM image, therefore, is essentially a topographical map of the electroactivity of the surface. Our main focus in
applying SECM to ECIA has been to use the probe microelectrode as a detector
to measure enzymatic product generated in the vicinity of heterogeneous
immunoassay microspots as shown in Figure 16. In the assays, ALP was used for
the label and PAPP for the substrate.
Our most recent application of SECM in immunoassay, which we will focus
on here, has been to image microscopic immunomagnetic bead domains [60,61].
Figure 17 (A) illustrates, and its caption describes, the essential features of the
technique that was developed to microspot bead domains. The size of the bead
domain was dened by the concentration of the beads in the droplet as shown in
Figure 17 (B). Using bead domains similar in size to that corresponding to the
microspot for 4.1 107 beads/mL in Figure 17 (B), we recently carried out a sandwich enzyme immunoassay with SECM detection for the model analyte mouse
IgG using the E-S pair ALP-PAPP [61]. The SECM probe microelectrode was
placed approximately 70 m from the surface for the two-dimensional scan
to detect the enzymatic product PAP. As the SECM images in Figure 18 illustrate,

Milestones of Electrochemical Immunoassay at Cincinnati

357

Figure 16 Schematic representation of scanning electrochemical microscopy (SECM)


for imaging heterogeneous enzyme immunoassay microsurfaces.

the quasi-stationary state concentration of PAP established over the beadmicrodomains increases with increasing mouse IgG concentration. These preliminary results along with those of other researchers in the area [62,63] have shown
that SECM can be applied successfully in miniaturized ECIA with micropatterned
surfaces. Furthermore, it has been shown that a surface micropatterned with an
array of antibodies, each targeting a different analyte, can be coupled easily to
SECM to yield multi-analyte ECIA [63].

D.

Automated Electrochemical Immunosensor Systems

Over the past three years, our group has been involved in a multi-disciplinary
effort aimed at extending enzyme ECIA to micrototal analysis systems (TAS).
The goal of TAS is to develop lab-on-a-chip type systems that integrate

358

Wijayawardhana et al.

(A)

(1)

(2)

(3)

(4)

pipette tip with


1 l bead suspension

0.3 L

rare earth
magnet

hydrophobic
surface

microspotted
beads

(B)

(1)

(2)

(3)

Figure 17 (A) Schematic representation of bead deposition. (1) Drop of bead suspension
is extruded out of a positive-displacement micropipet; (2) after concentrating the beads at
the bottom of the drop, the drop is reduced to 0.3 L; (3) the drop is set on the hydrophobic surface; (4) the drop will stay at the pipet tip if the pipet is retracted slowly from the surface, leaving the beads behind. (B) SEM images of bead domains produced using 1 L of
(1) 4.1 107, (2) 4.1 106, (3) 4.1 105 beads/mL.

microfabricated components and control electronics in a single miniaturized platform [64]. Some of the advantages of such systems for carrying out ECIA are
obvious. The extremely small size of lab-on-a-chip sensors permits samples in the
picoliter range. This reduces the cost of reagents and allows very small samples
to be analyzed, a critical issue in clinical immunoassay, as discussed above. The
small size of the lab-on-a-chip systems also holds much promise for a wide variety of novel ECIA applications ranging from light, hand-held immunosensors to
immunosensors implanted in humans for continuous monitoring.
In reaching our goal of a lab-on-a-chip electrochemical immunosensor system, we have adopted a scale-down approach in which the initial focus was to
develop a meso-scale immunosensor system that may be partially tted with

Milestones of Electrochemical Immunoassay at Cincinnati

359

Figure 18 SECM images and proles of bead domains prepared from 1 L of 4 107
beads/mL after a sandwich immunoassay of mouse IgG. Concentration of mouse IgG in
ng/mL: (a) 1150, (b) 115, (c) 11.5. (From Ref. 61.)

360

Wijayawardhana et al.

microfabricated components. Figure 19 shows the fully automated meso-scale


system that is currently in use. The reagents necessary for enzyme ECIA are contained in reservoirs that open through a combination of valves to a central
microuidic path where the immunoassay takes place. Paramagnetic microbeads
coated with antibody serve as the heterogeneous surface for the immunoassay.
Two electromagnets are used to capture the micro beads for the various assay
incubations. All incubations up to the enzyme substrate incubation are done at the
position of the rst electromagnet with the rinses being taken out through a Tjunction placed upstream of the second electromagnet. This ensures that no conjugate (Ab*) adsorbs to the wall of the microuidic channel along the length of
the second electromagnet where the beads are held for the enzyme substrate incubation. This reduces the nonspecic adsorption effects of Ab* greatly.
Recently, a sandwich enzyme immunoassay for the model analyte mouse
IgG was carried out in the meso-system using the ALP-PAPP E-S pair [65]. The
incubation volumes were 9 L as dened by the volume of the channel along the
length of the electromagnet. The microbeads were coated with the primary antibody before use. The total assay time was 34 min and it included 7 min incubations for both the antigen (mouse IgG) and Ab*, and a 4 min incubation for the
enzyme substrate. The current-time plots obtained using a BAS LC-EC owthrough electrochemical cell for 0, 50, 500, and 1000 ng/mL mouse IgG
immunoassays are shown in Figure 20. The plot of the peak-current versus the
mouse IgG concentration showed excellent linearity. The signal for 0 ng/mL is
due to the presence of PAPP from the nonenzymatic hydrolysis of PAPP.

VI.

CONCLUSION

Electrochemical immunoassays have evolved greatly over the past two decades.
It has been shown that ECIA works well in several heterogeneous and homogeneous immunoassay formats and that a variety of electroanalytical techniques
may be used for detection. Signicant improvements in the assay sensitivity have
resulted in zeptomole (1021) detection limits for a number of analytes. Although
most of the ECIA concepts were demonstrated on relatively pure samples, there
have also been several thorough evaluations on blood and serum samples as well
as environmental samples. The ability to thus perform homogeneous ECIA
directly on colored and/or turbid samples is a denite advantage over spectrophotometric immunoassays where colored and turbid samples pose numerous
difculties in detection. Another important asset of ECIA compared to spectrophotometric immunoassay that we have been exploring is its amenability to
miniaturization. There has been a steady trend in the area of biosensors to miniaturize sensors so that they may be used as hand-held diagnostic tools in a wide

Figure 19 An automated meso-scale electrochemical immunosensor system with the components (a) microbead reservoirs, (b) rinse buffer reservoir, (c) conjugate (Ab*) reservoir, (d) enzyme substrate reservoir, (e) valves, (f ) electromagnets,
(g) central micro-uidic silicone channel, (h) rotary pump.

Milestones of Electrochemical Immunoassay at Cincinnati


361

362

Wijayawardhana et al.

Figure 20 Current-time plots of a mouse IgG immunoassay done in the meso-scale


immunosensor shown in Figure 19.

Milestones of Electrochemical Immunoassay at Cincinnati

363

range of situations including home-based diagnosis [37]. All these advantages of


ECIA are embodied in our current efforts aimed at developing lab-on-a-chip electrochemical immunosensor systems.

REFERENCES
1.

2.

3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.

C. A. Wijayawardhana, H. B. Halsall, W. R. Heineman, in Encyclopedia of Electrochemistry. Volume 9: Bioelectrochemistry (G. S. Wilson, ed.) Wiley-VCH
accepted for publication.
P. Treloar, J. Kane, P. Vadgama, in Principles and Practices of Immunoassay 2nd
ed. (C. P. Price, D. J. Newman, eds.), Macmillan Reference Ltd, London, 1997, pp.
483509.
S. H. Jenkins, H. B. Halsall, W. R. Heineman, in Advances in Biosensors (A. P. F.
Turner, ed.), JAI Press Ltd., 1991, 171228.
P. Skldal, Electroanalysis 9:737745 (1997).
M. A. Cousino, T. B. Jarbawi, H. B. Halsall, W. R. Heineman, Anal. Chem.
69:544A549A (1997).
K. R. Wehmeyer, H. B. Halsall, W. R. Heineman, Clin. Chem. 28:19681972
(1982).
M. A. Brooks, J. A. F. DeSilva, M. R. Hackman, Anal. Chim. Acta 64:165175
(1973).
T. A. Miller, B. Lamb, K. Prater, J. K. Lee, R. N. Adams, Anal. Chem. 36:418420
(1964).
J. Wang, Stripping Analysis, VCH, Deereld Beach, FL, 1985.
J. P. Zodda, The Synthesis and Electrochemical Evaluation of Electroactive Labels.
Ph.D. Dissertation, University of Cincinnati, 1980.
M. J. Doyle, H. B. Halsall, W. R. Heineman, Anal. Chem. 54:23182322 (1982).
G. E. Krejcarek, K. L. Tucker, Biochem. Biophys. Res. Commun. 77:581585
(1977).
F. J. Hayes, H. B. Halsall, W. R. Heineman, Anal. Chem. 66:18601865 (1994).
H. Huiliang, C. Hua, D. Jagner, L. Renman, Anal. Chim. Acta 193:6169 (1987).
J. Wang, B. Tian, K. R. Rogers, Anal. Chem. 70:16821685 (1998).
K. R. Wehmeyer, M. J. Doyle, D. S. Wright, H. M. Eggers, H. B. Halsall, W. R.
Heineman, J. Liq. Chromatogr. 6:21422156 (1983).
H. T. Tang, H. B. Halsall, W. R. Heineman, Clin. Chem. 37:245248 (1991).
M. J. Doyle, H. B. Halsall, W. R. Heineman, Anal. Chem. 56:23552360 (1984).
H. T. Tang, C. E. Lunte, H. B. Halsall, W. R. Heineman, Anal. Chim. Acta
214:187195 (1988).
O. Niwa, Y. Xu, H. B. Halsall, W. R. Heineman, Anal. Chem. 65:15591563 (1993).
U. Wollenberger, M. Paeschke, R. Hintsche, Analyst 119:12451249 (1994).
B. R. Clark, E. Engvall in Enzyme Immunoassay (E. T. Maggio, ed.), CRC Press,
Boca Raton, 1981, pp. 167179.
M. J. Doyle, H. B. Halsall, W. R. Heineman, Anal. Chem. 56:23552360 (1984).
K. R. Wehmeyer, H. B. Halsall, W. R. Heineman, Anal. Chem. 58:135139 (1986).

364
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.
51.
52.
53.

Wijayawardhana et al.
K. R. Wehmeyer, H. B. Halsall, W. R. Heineman, Clin. Chem. 31:15461548
(1985).
S. H. Jenkins, H. B. Halsall, W. R. Heineman, Anal. Biochem. 168:291294 (1988).
N. Kaneki, Y. Xu, A. Kumari, H. B. Halsall, W. R. Heineman, P. T. Kissinger, Anal.
Chim. Acta 287:253258 (1994).
Y. Xu, H. B. Halsall, W. R. Heineman, Clin. Chem. 36:19411944 (1990).
E. T. Maggio in Enzyme Immunoassay (E. T. Maggio, ed.), CRC Press, Boca Raton,
1981. pp. 145.
R. I. Leininger, C. W. Copper, R. D. Falb, G. A. Grode, Science 17:16251626
(1966).
B. R. Clark, E. Engvall in Enzyme Immunoassay (E. T. Maggio, ed.) CRC Press,
Boca Raton, FL, 1981, pp. 167180.
G. H. Parson, Meth. Enz., 73:224233 (1981).
M. E. Eggers, H. B. Halsall, W. R. Heineman, Clin. Chem. 28:18481851 (1982).
D. S. Wright, H. B. Halsall, W. R. Heineman, Anal. Chem. 58:29952998 (1986).
H. Yao, H. B. Halsall, W. R. Heineman, S. H. Jenkins, Clin. Chem. 41:591598
(1995).
H. Yao, S. H. Jenkins, A. J. Pesce, H. B. Halsall, W. R. Heineman, Clin. Chem.
39:14321434 (1993).
P. R. Ekins, Clin. Chem. 44:20142021 (1998).
P. A. Nagainis, C. H. Nakagawa, S. L. Baron, S. A. Fuller, H. M. Chandler, J. G. Hurrell, Clin. Chim. Acta 163:273277 (1986).
P. N. Alzari, M.-B. Lascombe, R. J. Poljack, Annu. Rev. Immunol. 6:555580 (1988).
H. B. Halsall, W. R. Heineman, S. H. Jenkins, Clin. Chem. 34:17011702 (1988).
L. S. Croft, Introduction to Protein Sequence Analysis, John Wiley & Sons, New
York, 1980, pp. 7085.
R. D. Krause, J. A. Lott, Clin. Chem. 20:775 (1974).
A. Kumari, M. S. Thesis, University of Cincinnati, 1991.
H. Gao, T. Jiang, W. R. Heineman, H. B. Halsall, J. L. Caruso, Fresenius J. Anal.
Chem. 364:170174 (1999).
T. Jiang, H. B. Halsall, W. R. Heineman, J. Agric. Food. Chem. 43:10981104
(1995).
C. H. Pollema, J. Ruzicka, Anal. Chem. 66:18251830 (1994).
S. Santandreu, S. Sole, E. Fabregas, S. Alegret, Biosens. Bioelectron. 13:712
(1998).
C. A. Wijayawardhana, S. Purushothama, M. A. Cousino, H. B. Halsall, W. R.
Heineman, J. Electroanal. Chem. 468:28 (1999).
A. J. Bard, L. R. Faulkner, Electrochemical Methods, John Wiley & Sons, New
York, 1980.
C. A. Wijayawardhana, H. B. Halsall, W. R. Heineman, Anal. Chim. Acta 399:311
(1999).
S. Purushothama, S. Kradtap, C. A. Wijayawardhana, H. B. Halsall, W. R. Heineman, Analyst 126: 337341 (2001).
S. Kradtap, C. A. Wijayawardhana, K. T. Schlueter, H. B. Halsall, W. R. Heineman,
Anal. Chim. Acta, 444: 1326 (2001).
S. Kradtap, Ph.D. Dissertation, University of Cincinnati, 2000.

Milestones of Electrochemical Immunoassay at Cincinnati


54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.

365

J. D. Newman, A. F. P. Turner, G. Marazza, Anal. Chim. Acta 262:1317 (1992).


P. Connolly, Trends in Biotechnology 12:123127 (1994).
J. P. Hart, S. A. Wring, Electroanalysis 6:617624 (1994).
G. Wittstock, K. Yu, H. B. Halsall, T. H. Ridgway, W. R. Heineman, Anal. Chem.
67:35783584 (1995).
A. J. Bard, F.-R. F. Fan, D. T. Pierce, P. R. Unwin, D. O. Wipf, F. Zhou, Science
254:6873 (1991).
M. V. Mirkin, Mikrochim. Acta 130:127153 (1999).
C. A. Wijayawardhana, G. Wittstock, H. B. Halsall, W. R. Heineman, Anal. Chem.
72:333338 (2000).
C. A. Wijayawardhana, G. Wittstock, H. B. Halsall, W. R. Heineman, Electroanalysis 12/9:640644 (2000).
H. Shiku, Y. Hara, T. Matsue, I. Uchida, T. Yamauchi, J. Electroanal. Chem.
438:187190 (1997).
H. Shiku, T. Matsue, I. Uchida, Anal. Chem. 68:12761278 (1996).
Micro-total Analysis Systems (Ed.: A. van der Berg, P. Bergveld), Kluwer Academic
Press, Boston, 1995.
C. A. Wijayawardhana, Ph.D. Dissertation, University of Cincinnati, 2000.
S. D. Jackson, H. B. Halsall, A. J. Pesce, W. R. Heineman, Fresenius J. Anal. Chem.
346:869862 (1993).
J. Zhang, W. R. Heineman, H. B. Halsall, J. Pharm. Biomed. Anal. 19:145152
(1999).
S. H. Jenkins, H. B. Halsall, W. R. Heineman, J. Clin. Immunoassay 13:99104
(1990).
Y. Ding, L. Zhou, H. B. Halsall, W. R. Heineman, J. Pharm. Biomed. Anal.
19:153161 (1999).
C. C. Bauer, A. V. Eremenko, E. Ehrentreich-Frster, F. F. Bier, A. Markower,
H. B. Halsall, W. R. Heineman, F. W. Scheller, Anal. Chem. 68:24532458 (1996).
M. Cardosi, S. Birch, J. Talbot, A. Phillips, Electroanalysis 3:169176 (1991).
R. Renneberg, W. Schssler, F. Scheller, Anal. Lett. 16:12791289 (1983).
M. Masson, T. Haruyama, E. Kobatake, M. Aizawa, Anal. Chim. Acta 402:2935
(1999).
C. Fernndez-Snches, Anal. Chim. Acta 402:119127 (1999).
W. O. Ho, D. Athey, C. J. McNeil, Biosens. Bioelectron. 10:683691 (1995).
G. Volpe, D. Compagnone, R. Draisci, G. Palleschi, Analyst 123:113031307
(1998).
M. Masson, Z. Liu, T. Haruyama, E. Kobatake, Y. Ikariyama, M. Aizawa, Anal.
Chim. Acta 304:353359 (1995).

11
Electrochemical Detection
of Peptides
Eskil Sahlin, Amy T. Beisler, and Stephen G. Weber
University of Pittsburgh, Pittsburgh, Pennsylvania

Mats Sandberg
Gteborg University, Gteborg, Sweden

I.
A.

INTRODUCTION
Importance of Peptide Determination

Signaling between nerve cells and between nerve cells and different organs is
effected by neuroactive substances such as amino acids, monoamines, acetylcholine, purines, and neuropeptides. This chemical language forms the basis of
life for multicellular organisms and is a major player in diverse functions such as
mood, memory, circadian rhythm, sexual behavior, brain development, and
degeneration. The neuropeptides are evolutionarily very old and constitute a
plethora of compounds in contrast to the relatively few so-called classical transmitters [1]. Neuropeptides are typically localized together with a nonpeptide
transmitter, which suggests that the role of the peptides is modulation of the basic
chemical transmission exerted by classical transmitters. The diversity of neuropeptides is exemplied in a simple organism such as the nematode C. elegans, which has 41 genes that encode neuropeptide precursors despite there being
only about 1000 somatic cells in the whole animal [2]. In humans, neuropeptides
are involved in regulation of the release of hormones from the pituitary, probably
in pain and analgesia, and in regulation of food intake [1]. A neuropeptide antagonist has also recently been used in treatment of depression. The nding that
many neuropeptides show a plastic expression and are upregulated following dif367

368

Sahlin et al.

ferent forms of injury implies long-term trophic effects that make this signaling
system interesting from a future therapeutic point of view [1].
The complete sequencing of different genomes will generate new information concerning neuropeptide and neuropeptide precursor genes [2]. The transgene
technology will be (and has been) used in order to try to dissect specic roles of neuropeptides [1]. A prerequisite for the success of such research is that the expression
of the gene products can be identied in a simple and sensitive manner. The antibody-based techniques are normally very sensitive and easy to use for analysis of
established neuropeptides. However, when it comes to identication of new
endogenous peptide ligands expressed after injury or after a transgene experiment,
which could trigger compensatory neuropeptide expression, a general and highly
sensitive technique for separation and detection of neuropeptides is essential.
Besides the discovery of new peptides, the routine investigation of peptide
inuence and metabolism requires a separations-based technique. Biologically
active peptides are numerous. They are all synthesized in the same general way.
Transcription and translation leads to prepropeptides [3]. These proteins may be
the source of several different active peptides. Specic enzymes are used to create
rst the propeptide and then active peptides. Active peptides are then inactivated,
usually by a peptidase that hydrolyses an amide bond in the peptide [4]. It is therefore certain that a large number of different molecules share the same primary
sequence, and it is likely that a large number of molecules may cross-react with a
peptide-reactive antibody. Thus, any immunoassay is bedeviled by selectivity
issues. If a perfectly selective antibody were available, it would be useful. However, it is undeniable that ultimately a complete understanding of, for example, the
action of peptides in the brain [35] will only come about when the dynamics of all
of the relevant processes are understood. This can only be done with chemical analytical techniques that can track the concentrations of all of the peptides relevant to
a particular process.
B.

Electrochemical Detection

Electrochemistry provides a powerful method for the determination of peptides


following their separation by high-performance liquid chromatography (HPLC) or
capillary electrophoresis (CE). The general principles of amperometric detector
function have been worked out and reviewed [610]. While there is still some optimization that can be done to maximize the signal-to-electronic noise level [11], in
practice it is not electronic noise but environmental noise, such as temperature
uctuations and composition uctuations in the mobile phase, that limits detection. To achieve the lowest detection limits under these circumstances, selectivity
is key. If systems can be found that turn on or increase the electrochemical reaction
rate for the analyte, but do nothing or suppress the reaction rate of impurities, then
an increase in signal-to-background ratio will result. In many circumstances, the
signal current is much smaller than the background current, so the limiting noise is

Electrochemical Detection of Peptides

369

related to the magnitude of the background current. As a result, lowering the background lowers the noise.
In principle, temperature uctuations can be minimized by thermostating or
insulating. The effects of impurities are harder to remove. It is generally true in electrochemistry that the rate of oxidation of a species increases as the potential applied
to the working electrode increases. Thus, signals from impurities are minimized by
using the lowest possible potential at an anode. Other means of generating selectivity rely on kinetics. Some electrochemical reactions are slow at electrode surfaces,
so the signal from them is poor. Catalysts or surface treatments that increase the
reaction rate increase the signal. This increase in reaction rate can be selective, so
that the rate of background processes does not increase as much as the rate of reaction of the analyte. Some electrochemical reactions have stable products that can
undergo a reverse redox reaction, for example hydroquinone oxidation leads to
quinone, which is reducible. Dual electrodes can be used in an amperometric detector to achieve an improvement in selectivity by detecting the reverse redox reaction
at the downstream electrode. Some electrochemical detectors have high efciency,
so they can be used to scrub the mobile phase, removing impurities.
As a general rule, one-electron, outer sphere electron transfer reactions, such
as the oxidation of ferrocene to ferricenium ion, have the property that the result of
an electrochemical experiment changes very little, or changes in a simple and predictable manner as conditions, such as solvent, pH, electrolyte composition and
concentration, electrode material, electrode surface treatment, are changed. This
becomes less true the more chemical and electrochemical reactions there are
between the reagent or analyte and the product that is stable at a given potential and
pH. Most, but certainly not all, of the reactions discussed in this chapter are somewhat complex. In these complex reactions, besides there being a signicant sensitivity of the electrochemistry to the chemical surroundings of the analyte, the result
of an electrochemical experiment can vary dramatically with the time scale of the
experiment. Although sometimes this sensitivity can be an advantagee.g., selective conditions for a particular reaction may be foundit is usually a disadvantage.
When a decision is made to use electrochemical detection, it is generally useful to
understand the electrochemistry of the analyte in some depth so that the practitioner is prepared for troubleshooting. An understanding of the literature and the
basic voltammetry at a variety of timescales and electrode materials can be useful
in interpreting difculties in detection or in developing better detection conditions.
II.

DETERMINATION BASED ON OXIDATION OR REDUCTION


OF UNDERIVATIZED PEPTIDES

A.

Introduction

The most common electrochemical approach to detecting peptides in separation


systems is by amperometry using oxidation or reduction of the amino acids com-

370

Sahlin et al.

prising the peptide. The electrochemical behavior of amino acids is well studied
but very complicated. The electrode material has a major impact, and an amino
acid can undergo unique redox reactions at different electrode materials. At carbon
and bare metal electrodes (e.g., platinum and gold), most work has focused on peptides containing tryptophan, tyrosine, cysteine, or cystine. Other amino acids are
oxidized at much higher potentials, resulting in high background currents that prevent determination of trace levels of peptides without tryptophan, tyrosine, cysteine, or cystine. The oxidation potentials for free tyrosine, tryptophan, cysteine,
and cystine at a parafn waximpregnated graphite electrode at pH 4 and 8.4 are
given in Table 1 [12]. These oxidation potentials are, however, strongly dependent
on electrochemical conditions such as pH and electrode material. In addition, the
oxidation potential of amino acids in peptides can be shifted in positive or negative
directions compared to free amino acids [1315]. Peptides containing tryptophan
and tyrosine are detected under almost the same electrochemical conditions; peptides containing cysteine and cystine can be detected with several different detection schemes. In addition to being susceptible to oxidation, cystine can also be
reduced. Metal oxide electrodes can also be used for peptide detection. The redox
reactions taking place at these electrodes are very different from the carbon and
bare electrodes and are therefore described separately below.
B.
1.

Detection of Peptides Containing Tryptophan or Tyrosine at


Carbon and Bare Metal Electrodes
Electrochemistry of Tryptophan

Free tryptophan (see Figure 1) can be oxidized at several electrode materials


including gold, platinum, and carbon [12,1620]. Studies of pH dependence show

Table 1 Oxidation Peak Potentials for Different Tyrosine,


Tryptophan, Cysteine, and Cystine at a parafn waximpregnated
graphite electrode
Oxidation peak potential (V vs. SCE)
Tyrosine

Tryptophan

Cysteine

Cystine

pH 4

0.90

0.92

1.37

pH 8.4

0.65

0.70

1.23a
1.41b
1.00a
1.22b

1.15

Data from linear sweep voltammetry (scan rate of 8.3 mV/s for tryptophan and
tyrosine, and 16.7 mV/s for cysteine and cystine) [12].
aMore negative peak.
bMore positive peak.

Electrochemical Detection of Peptides

371

Figure 1 Structure of tryptophan and tyrosine.

that the oxidation potential increases with decreasing pH [12]. The electrochemical behavior is complicated and different reaction pathways, each pathway
including several steps, have been suggested [12,16,1820]. The number of electrons transferred to the electrode surface can range from one to over four depending on the electrochemical conditions [12,16,18,19]. Oxidation of tryptophan
may, particularly at high concentrations, also result in formation of polymeric
products forming a lm on the electrode surface [16,1820].
2.

Electrochemistry of Tyrosine

Free tyrosine (see Figure 1) can be oxidized at gold, platinum, and carbon electrodes [12,15,1719]. Similar to that of tryptophan, the oxidation potential
increases with decreasing pH [12,15]. At platinum electrodes (pH 7), a three-step
oxidation reaction has been suggested with one electron transferred in both the
rst and third steps (i.e., ECE) [17]. At glassy carbon and gold electrodes, tyrosine oxidation produces hexacyclodienone and polyphenylene oxide alanyl in
another reaction pathway [18,20]. At a glassy carbon electrode, the anodic reaction is irreversible involving one electron and one proton [15]. At a graphite electrode, 3,4-dihydroxyphenylalanine is proposed as a product from an irreversible
two-electron process [12]. Under some conditions a polymeric product, suggested
to be polyphenylene oxide alanyl, has been formed [18,19].
3. Amperometric Detection of Peptides Containing Tryptophan or
Tyrosine at Carbon Electrodes
Because tryptophan and tyrosine are susceptible to oxidation at relatively low
potentials, peptides containing these amino acids can easily be detected amperometrically. Among the citations listed below, carbon materials such as glassy carbon and graphite are the most widely used working electrode materials. For
microscale separations, a carbon ber can be employed. An optimal signal-tonoise ratio is often found at potentials lower than the potential needed for maximum sensitivity because the signal is accompanied by a background that rapidly

372

Sahlin et al.

rises at high potentials. Both the signal and the background are highly pH dependent and, as mentioned, an optimal detection potential is shifted toward more positive potentials with decreasing pH. In an effort to improve the performance (e.g.,
sensitivity, selectivity, or stability) of the measurements, one type of electrode
surface modication has been used. Coating carbon bers with poly(3-methylthiophene) resulted in improved sensitivity and stability together with decreased
overpotentials for oxidation of tryptophan and tyrosine and has been applied to
the detection of dipeptides containing these amino acids [21].
Oxidation of tyrosine- and tryptophan-containing peptides has been mainly
used for detection in HPLC [13,2145] and occasionally in CE [46]. Examples of
tyrosine- or tryptophan-containing peptides that have been determined include
enkephalin peptides and their metabolites [2126,2931,33,34,36,40,41,4346],
luteinizing hormone-releasing hormone [25,43], [CH2S] pseudopeptides [25],
cholecystokinin tetrapeptide and octapeptide sulfate [26,40], microcystins [27],
neurotensin and its fragments [13,28,43], bombesin and its fragments [28], endorphins [30,32,34,35,43], oxytocin [33,42,43], Lys-vasopressin [33,42,43],
angiotensin I and II [33,43], Tyr-Ala [21], [D-Arg1, D-Phe5, D-Trp7,9, Leu11]-substance P [37], [Arg6, D-Trp7,9, N-MePhe8]-substance P(611) [37], adipokinetic
hormone [38], hypertrehalosemic hormone [38], allatostatin I [38], Arg-vasopressin [42,43], vasopressinoic acid [42], pressinoic acid [42], Arg-vasotocin
[43], ACTH 124 [43], and -MSH [43]. Detection limits are typically in the
110 nM range (without on-column preconcentration) [22,23,31,34,37,40,42].
A system has also been described for in vivo monitoring of Met-enkephalin
using microdialysis sampling and capillary liquid chromatography [22].
B. Detection of Peptides Containing Cysteine or Cystine at Carbon
or Bare Metal Electrodes
1.

Electrochemistry of Cysteine

Free cysteine (see Figure 2) can be oxidized at several electrode materials including carbon, platinum, and gold [12,19,4752]. As exhibited with tyrosine and
tryptophan, the oxidation potential of cysteine increases with decreasing pH [12].
Cysteine (RSH) is oxidized to cystine (RSSR) [19,4749,51,52] according to:
2RSH 3 RSSR + 2H+ + 2e

(1)

Cystine can then be further oxidized as described in the section below


[19,47,48,52]. It has also been suggested that cysteine is oxidized to RS [4850],
which is subsequently oxidized to RSO3 [49,50]. In addition, RSH adsorbed on
platinum is thought to be oxidized directly to cysteic acid (cysteine sulfonic acid),
RSO3H, [51]:
RSHads + 3H2O 3 RSO3H + 6H+ + 6e

(2)

Electrochemical Detection of Peptides

373

Figure 2 Structure of cysteine and cystine.

Here, the subscript ads indicates that the species is adsorbed on the electrode surface. Two approaches are used to overcome the disadvantages associated
with the high overpotentials required for cysteine oxidation. In the rst approach,
carbon electrodes are modied with compounds that decrease the overpotential.
Such compounds include different macrocyclic transition metal complexes (often
containing cobalt and phthalocyanines or porphyrins) [50,5359], vitamin B12 (a
cobalt containing compound) [56], a mixed-valence ruthenium(IV,III) oxide lm
stabilized by cyano cross-links [6062], indium ferricyanide [63], or a Prussian
blue lm [64]. In addition, a bismuth(V)-doped lead dioxide PbO2 lm on gold
has been used [65]. The most commonly used compound is cobalt phthalocyanine, and it has been reported that with its use the overpotential for cysteine oxidation is decreased up to 0.75 V [53]. It has also been suggested that the
cobalt(III/II) redox couple acts as an electron mediator according to [53,54,59]:
Co(II) 3 Co(III) + e

(3)

2RSH + 2Co(III) 3 RSSR + 2Co(II) + 2H+

(4)

At a glassy carbon electrode modied with mixed-valence ruthenium oxide lms


stabilized by cyano cross-links, it has been shown that cysteine is oxidized
directly under coulometric conditions in the absence of dioxygen to cysteine
sulnic acid, RSO2H, according to [60]:
RSH + 2H2O 3 RSO2H + 4H+ + 4e

(5)

In the second and more common approach, cysteine is detected at a mercury


or amalgamated gold electrode. The reaction has been studied frequently and different reactions have been suggested. These are summarized in the literature [50].
Today cysteine is considered to undergo the following reaction at a mercury electrode [50,6668]:

374

Sahlin et al.

2RSH + Hg 3 (RS)2Hg + 2H+ + 2e

(6)

Under some conditions, (RS)2Hg2 or RSHg have been suggested to be formed


instead of (RS)2Hg [67,68]. Note that in Eq. (6), it is mercury that undergoes oxidation and not cysteine. A detailed study of Eq. (6) under different conditions is
available in the literature [67,68].
2.

Electrochemistry of Cystine

Cystine (see Figure 2) can be reduced at mercury and gold electrodes to cysteine
according to [50,66,69,70]:
RSSR + 2H+ + 2e 3 2RSH

(7)

Equation (7) has been studied in several articles and a summary is available in the
literature [50,70]. The mercury-cystine system is, however, more complex than
what is implied in Eq. (7) and can, depending on the electrode potential, include
adsorbed species such as (RS)2Hg and (RS)2Hg2 [70]. The system has been studied and explored in detail [70]. Cystine can also be reduced at carbon electrodes
modied with conducting polymers containing xed metal thiolate sites [50,71],
different macrocyclic transition metal complexes (often containing cobalt and
phtalocyanines or porphyrins) [50,55,57,72,73], or vitamin B12 [56], which lower
the overpotential necessary for reduction.
Cystine can also be further oxidized at carbon, platinum, and gold electrodes [12,19,47,5052,69,74] with an increase in the oxidation potential seen
when the pH is decreased [12]. In the oxidation reaction, cystine is oxidized to
cysteic acid according to [19,47,50,51,69,72]:
RSSR + 6H2O 3 2RSO3H + 10H+ + 10e

(8)

In addition, oxidation of cystine at a gold electrode is suggested to occur through


several electron transfer steps to RSO3 [19,47,50,69] and RSSO3 [19,47,50,52].
At a platinum electrode, oxidation has been suggested to result in RSO3
[50,51,74] and RSO2ads (an adsorbed radical) [50,74]. Furthermore, a bismuth(V)-doped lead dioxide PbO2 lm on gold [65] and a mixed-valence ruthenium (IV, III) oxide lm stabilized by cyano cross-links on glassy carbon [60,62]
have been used.
3. Amperometric Detection of Peptides Containing Cysteine or Cystine
at Carbon or Bare Metal Electrodes
Carbon electrodes are also often used for oxidation of cysteine and cystine-containing peptides [7586]. With a dual carbon working electrode system (in
series), cysteine containing peptides can be oxidized at the upstream electrode,

Electrochemical Detection of Peptides

375

and cystine-containing peptides can be oxidized at a higher potential at the downstream electrode [75,76,86]. Pretreatment of a glassy carbon electrode at +1.9 V
versus Ag/AgCl has been shown to result in increased sensitivity and decreased
overpotential for glutathione (-Glu-Cys-Gly) oxidation [85].
In order to decrease the overpotential for cysteine oxidation in cysteinecontaining peptides, graphite-epoxy resin or carbon paste electrodes with immobilized cobalt phthalocyanine have been employed in HPLC systems [53,8789].
Glassy carbon electrodes modied with a mixed-valence ruthenium(IV,III) oxide
lm stabilized by cyano cross-links [61,62], indium ferricyanide [63], or a Prussian blue lm [64], and a bismuth(V)-doped lead dioxide PbO2 lm on gold [65]
have also been used for detection following HPLC separation. In addition, a carbon ber modied with a mixed-valence ruthenium(IV,III) oxide lm stabilized
by cyano cross-links has been used in CE [46].
Amperometric detection of cysteine- and cysteine-containing peptides frequently occurs at mercury or amalgamated gold electrodes [39,90107] through
the oxidation of mercury in the presence of thiols that occurs at much lower
potentials than the oxidation of thiols. By employing a dual working electrode
system (in series), cystine-containing peptides can be reduced at the upstream
electrode and reoxidized at the downstream electrode together with cysteine containing peptides [90,91,93,95,96,98104] as shown in Figure 3. A comparison of
the oxidation of mercury in the presence of thiols to oxidation of thiols at gold or
glassy carbon electrodes showed that the former was superior with respect to
detection limits and baseline drift [105].
In another approach, electrogenerated bromine oxidizes cysteine and cystine in peptides. The concentration of bromine is then measured amperometrically
[108,109].
Single, dual, and even triple working electrode congurations for determination of cysteine- and cystine-containing peptides (e.g., glutathione) have been
used with HPLC [39,53,6165,7589,9193,95105,109] or CE [46,90,92,106,

Figure 3 Detection scheme for peptides containing cystine (RSSR) and cysteine (RSH)
employing a dual gold amalgam working electrode system.

376

Sahlin et al.

107]. Cysteine- and cystine-containing peptides that have been determined include glutathione [39,46,53,6265,7579,81102,104109], tryptic digest of
ribonuclease [90], Cys-Gly [93,96,98], -Glu-Cys [98], glutathione cysteine
disulde (CSSG) [98], ergothioneine [98], Lys-Cys-Thr-Cys-Cys-Ala [80,101,
104], Cys-Gln-Asp-Ser-Glu-Thr-Arg-Thr-Phe-Tyr [101], [Arg8]-vasopressin
[61,101,102,104], chymotryptic digest of bovine A-crystallin [101], peptic
digest of partially reduced insulin [101], oxytocin [61,102104], tryptic digest of
synthetic rat atrial natriuretic factor (ANF 833) [102], vasopressin analogues
[103], SK&F 101926 and SK&F 101498 [103], SST-22a, SST-22b, SST-22c,
SST-14 [104], tocinoic acid [61], and isotocin [61]. Detection limits are typically
in the 10500 nM range (without on-column preconcentration) [46,53,61,62,65,
75,76,78,83,84,8790,92,93,98,103,105,106,108]. In addition, HPLC with amalgamated gold electrodes has also been used to localize disulde bridges in peptides and proteins [102,104].
C.

Amperometric Detection of Peptides Containing Amino Acids


with Low or No Electrochemical Activity at Carbon Electrodes

Most amino acids are electroactive by being susceptible to oxidation. However,


their oxidation potentials at carbon or bare metal electrodes are too high (>+1.0
V vs. Ag/AgCl [3 M NaCl]) to be useful for trace determinations, and their electrochemistry will not be described here. Despite this, several di- and tripeptides
without tryptophan, tyrosine, cysteine, and cystine can still be oxidized at potentials suitable for trace determinations (<+1.0 V vs. Ag/AgCl [3 M NaCl]). For
instance, several dipeptides and tripeptides containing alanine, arginine, aspartic
acid, glutamic acid, glycine, leucine, lysine, and valine have been oxidized at a
glassy carbon electrode used as detector in an HPLC system [110,111]. The electrochemistry is, however, unclear. Improved sensitivity and selectivity for
-dipeptides (over - and -dipeptides and amino acids) can be obtained by
anodic pretreatment of the glassy carbon electrode at +1.2 to +1.4 V versus
Ag/AgCl (3 M NaCl) in a solution containing Cu(II) and tartrate [110,111]. The
changed electrochemical behavior of the glassy carbon electrode seems to be due
to an increase in the oxygen to carbon ratio at the surface rather than formation of
copper-containing species on the surface.
D.

Detection of Peptides at Metal OxideCovered Electrodes

Many compounds, including almost all amino acids and several peptides, can be
oxidized at relatively low potentials at some metal oxidecovered electrodes in
alkaline media. For peptides, oxides of copper, gold, and platinum have been utilized. However, the oxidation potentials are primarily determined by the state of

Electrochemical Detection of Peptides

377

the metal oxide rather than by the redox potentials (E0) of the peptidei.e., almost
no voltammetric resolution can be obtained. Hence, the main application of metal
oxidecovered electrodes is for amperometric detection in separation systems.

1.

Copper OxideCovered Electrodes

Oxidation of peptides and amino acids at copper oxide electrodes is achieved at


very high pH (typically in 0.1 M hydroxide) and potentials in the range of +0.40
to +0.60 V versus Ag/AgCl [112114]. In this potential range, a copper electrode
is covered with a Cu(II) oxide lm (CuO) containing hydroxyl radicals that can
described as CuO(OH) [113115]. Amino acids that have been oxidized include
Ala, Arg, Asn, Asp, Cys, Gln, Gly, His, Leu, Lys, Met, Phe, Pro, Ser, Thr, Trp,
Tyr and Val [112,107], and it has been suggested that oxidation occurs according
to [113]:
RCH(NH2)CO2 3 RCH = NH + CO2 + H+ + 2e
RCH = NH 3 RCN + 2H+ + 2e

(9)
(10)

Copper oxidecovered electrodes in alkaline media have been employed for


detection of peptides in both HPLC [112,114] and CE [116]. Peptides that have
been determined include small peptides such as Gly-Gly-amide [112,116], GlyGly [112,116], tri-, tetra-, penta-, and hexaglycine [114,116], aspartame [116]
cyclic Leu-Trp-Asp-Pro-Val [116], cyclic Leu-Tyr-Asp-Pro-Val [116], and desTyr-leucine enkephalin [114]. Detection limits (without on-column preconcentration) are poor, typically in the 5100 M range [114,116].
At copper oxidecovered electrodes, another detection mechanism is also
possible. At lower potentials and pH (typically at +0.15 V vs. Ag/AgCl (satd
KCl) at pH 611), a passivating Cu(II) oxide or hydroxide lm is formed on the
copper electrode [117119]. The dissolution of this lm, i.e., the oxidation of
metallic copper to Cu(II), is enhanced in the presence of amino acids and short
peptides due to complexation between Cu(II) and the amino acids or peptides
[117]. Thus, the current generated originates from oxidation of metallic copper to
Cu(II)i.e., the amino acids are not oxidized. The exact composition of the lm
is not known but it is believed that a copper(I) oxide or hydroxide layer is located
between the copper electrode surface and the Cu(II) oxide or hydroxide layer
[118,119]. The sensitivity is dependent on complexation kinetics [117119] and
decreases with increasing peptide length [117,120]. This detection scheme has
been applied in both HPLC [117,120] and CE [121] determinations. Peptides that
have been detected include different di-, tri-, and tetrapeptides containing glycine
[117,120]. Detection limits are high, typically in the 70 to 200 M range for
dipeptides [117,120,121].

378

2.

Sahlin et al.

Gold and Platinum OxideCovered Electrodes

Amino acids and peptides can be oxidized during initial oxidation of gold or platinum electrodes typically at +0.50 V versus. SCE at pH>11 [122125]. Oxidation
occurs in the presence of adsorbed hydroxyl radicals, i.e., AuOH or PtOH, that
are formed as intermediates in the oxidation of gold to AuO or platinum to PtO
[123,125]. Oxidation of amino acids occurs at both the amine group and, if present, the sulfur-containing groups [125]. For sulfur-containing compounds, the
following reactions have been suggested [126]:
RSH 3 RSads + H+ + e

(11)

RSads + 3H2O 3 RSO3 + 6H+ + 5e

(12)

RSSR 3 2RSads

(13)

2RSads + 6H2O 3 2RSO3 + 12H+ + 10e

(14)

RSR 3 RSRads

(15)

RSRads + 2H2O 3 RSO2R + 4H+ + 4e

(16)

Here, the subscript ads indicates that the species is adsorbed on the electrode surface. In addition, cysteine [125129], cystine [125129], methionine [125129],
tryptophan [127], and tyrosine [127] can be oxidized under mildly acidic conditions. Due to rapid inactivation of the electrode surface by oxide growth and
adsorption of different species, a pulsed potential sequence, shown in Figure
4(A), must be employed. The current measurement is performed a short time (typically 500 ms) after the detection potential is applied. This is followed by a positive and a negative potential step in order to clean and reactivate the electrode surface. During the positive potential step, most species are desorbed and the metal
surface is further oxidized to AuO or PtO. In the negative potential step, the metal
oxide is reduced and the original metal surface is restored. This particular pulse
sequence is known as pulsed amperometric detection (PAD). A common variant
is integrated pulsed amperometric detection (IPAD), in which a rapid cyclic scan
is applied within the detection step with simultaneous integration of the current
as shown in Figure 4(B) [123125,129]. The anodic charge for oxide formation
during the positive sweep tends to be compensated by the cathodic charge from
dissolution of the oxide during the subsequent negative sweep. Hence, the background current is greatly reduced. A more general name is pulsed electrochemical detection (PED) which encompasses many waveforms. PED has been
described in detail in the literature [123125]. Amino acids that have been oxidized using this technique include Ala, Arg, Asn, Asp, Cys (and cystine), Gln,

Electrochemical Detection of Peptides

379

Figure 4 Typical potential step sequence in (A) pulsed amperometric detection (PAD)
and (B) integrated pulsed amperometric detection (IPAD) of peptides.

380

Sahlin et al.

Glu, Gly, His, Ile, Leu, Lys, Met, Phe, Pro, Ser, Thr, Tyr, and Val [122]. PED for
peptides has only been applied for detection in HPLC [123128,130]. Leu-Tyr
[127], Met-Leu [127], Leu-Trp [127], tryptic digest of -Casein A2 [127] glutathione [126,128,129], and -L--aminoadipyl-L-cysteinyl-D-valine [130] have
all been determined using PED with detection limits (without on-column preconcentration) typically in the 40 nM1.0 M range [126,128,130].

III.

DERIVATIZATION TECHNIQUES

A.

General Considerations

Peptides may also be detected using derivatizing agents to impart electrochemical, UV, or uorescence properties on the analyte of interest. Though a vast number of possible derivatizing reagents exist, discussion here will be limited to those
that have been used with electrochemical detection.
Derivatization may be achieved in a variety of ways. The choice will
depend on a number of factors including the stability of the derivatization agent
and tagged analyte, required reaction time, and the required temperature. There
are three basic approaches to performing a derivatization reaction: pre-column
(off line), on-column, or post-column. Pre-column derivatization is the most commonly employed technique because the reaction time is not limited and excess
derivatization agent can be removed before analysis. For on-column and post-column derivatization a few limitations exist. First, the reaction should be complete
in a short amount of time. Second, the derivatization reagent should not be electroactive or it should elute well separated from the analytes of interest. In addition, using post-column derivatization may require the use of a pump to deliver
the derivatization agent.
Many derivatization reagents target primary amines and thus are not specic for peptides. In this case, identication of peptides in complicated matrices
is extremely difcult. A number of approaches for minimizing interference from
nonpeptide analytes have been undertaken and will be discussed in context.
B.

Derivatizing Agents

1.

Naphthalene-2,3-Dicarboxaldehyde (NDA)

NDA reacts with primary amines in the presence of cyanide to produce 1cyanobenz[f]isoindole (CBI) derivatives (Figure 5) [131]. Cyclic voltammetry
reveals that the oxidation of the CBI involves the loss of an electron from the heterocyclic nitrogen forming a radical cation that further reacts and is therefore not
available for reduction. In addition, the electrochemistry of NDA derivatives is
virtually pH independent, indicating no gain or loss of a proton.

Electrochemical Detection of Peptides

381

Figure 5 Reaction of NDA with a primary amine. (From Ref. 131.)

Analysis of a number of derivatized enkephalins and other peptides (ArgGly-Gly, Tyr-Gly, Gly-Gly-Phe, Tyr-Gly-Gly to name a few) revealed an average half-wave potential of +0.65 V versus Ag/AgCl. Only a slight increase in
half-wave potential (+35 mV) is seen with increasing chain length from glycine
to tetra-glycine and the half-wave potential is largely independent of scan rate.
When the derivatized peptide contains an electroactive amino acid such as tyrosine or tryptophan, a signal may be seen from both the derivative and the amino
acid. In addition, the presence of arginine in the derivatized peptide can signicantly affect the oxidation potential, depending on the position of the amino acid
in the sequence. The limited data suggest that the presence of arginine at the carboxy terminus may lower the half-wave potential. In addition, doubly derivatized
peptides (i.e. those containing lysine) are oxidized at lower applied potentials
than those that are single derivatives.
2. o-Phthalaldehyde (OPA)
OPA reacts with primary amines in the presence of a thiol to yield isoindole derivatives that are electroactive (Figure 6) [132,133]. OPA with -mercaptoethanol as
the thiol (OPA-ME) has been used to derivatize -glutamyl di- and tripeptides
(GGPs) in microdialysate samples [134]. OPA-ME reacts quickly (3060 s for
maximum response) through the primary amine with all of the GGPs studied. In
addition, the derivative is reasonably stable with a product half-life of 6 hours.
Half-wave potentials for GGPs are in the range of +0.50 V to +0.55 V versus.
Ag/AgCl. Although the half-wave potential for ME is 0.35 V more positive than
its derivatives, a large interference peak is produced in the chromatogram when
+0.70 V versus. Ag/AgCl is applied to the working electrode. A number of
reagents have been used to eliminate or minimize excess ME, including iodoacetamide; however, these attempts were unsuccessful owing to the appearance of
interfering peaks [135]. A Hg/Au amalgam electrode positioned upstream of the
detector may be used to eliminate a large amount of the ME [134]. By this
approach, thiols are eliminated through mercury oxidation at Hg/Au electrode

382

Sahlin et al.

Figure 6 Reaction of OPA with a primary amine and a thiol. (From Ref. 132.)

(described previously) thereby reducing the ME peak by 75%. Isocratic separation and electrochemical detection of peptide derivatives showed a factor of 15
improvement over the system without the amalgam scrubber with detection limits of 13 fmol (1.3 nM) achieved. OPA-ME has also been used for derivatization
of dipeptides of neurochemical interest, including carnosine (-Ala-His). Detection was performed at +0.70 V vs. Ag/AgCl with detection limits of 0.66 pmol
(50 nM) achieved [136].
OPA has shown potential as a derivatization agent for glutathione (-GluCys-Gly) determination using electrochemical detection at glassy carbon electrodes [137]. The derivatization reaction is fast (complete in 10 s) as the thiol
required to form the electroactive derivative is part of the analyte molecule. The
peak potential of the resulting product is pH dependent in the pH 811 range.
Using this methodology, glutathione derivatives were analyzed and detection
limits around 40 pmol (0.67 M) were achieved at +0.80 V versus. Ag/AgCl.
3.

6-Aminoquinolyl-N-Hydroxysuccinimidyl Carbamate (6-AQC)

As a result of high derivative yields (~100%) and derivative stability, 6-AQC has
shown promise as a peptide derivatization reagent [138]. 6-AQC reacts easily
with primary and secondary amines to form derivatized peptides (Figure 7) containing an aromatic amino group (aminoquinoline) that is used for electrochemical detection. Cyclic voltammetry reveals a chemically irreversible oxidation
unaffected by changes in pH in the range 2.37. Scan rate studies suggest that a
chemical reaction follows the oxidation, yielding a product that adsorbs on the
electrode surface. Hydrodynamic voltammograms (HDVs) indicate that the halfwave potential for a number of dipeptide derivatives is +1.0 V versus Ag/AgCl.
Chromatographic analysis demonstrated separation of six peptides and detection
using a glassy carbon electrode with detection limits (determined for amino acids)
near the 2.5 pmol (0.25 M) level. The high potential required for oxidation of 6AQC derivatives is a disadvantage compared to other electrochemically detected
derivatives (OPA and NDA). In addition, the presence of unreacted 6-AQC or its

Electrochemical Detection of Peptides

383

Figure 7 Reaction of 6-AQ with a primary amine. (From Ref. 132.)

hydrolysis product 6-AQ (both electroactive) may interfere with detection of


trace analytes. However, a potential advantage of this technique is that 6-AQC
derivatives have proven to be more stable than OPA and NDA analogues.
4.

Ferrocenecarboxylic Acid Chloride (FAC)

FAC has been successfully used to derivatize primary and secondary amines
[139]. Derivatives are optimally detected at +0.60 V vs. Ag/AgCl, and the oneelectron oxidation is reversible. Ferrocenecarboxylic acid (FCA), an electroactive
hydrolysis product of FAC, is a potential chromatographic interference. Fortunately, being somewhat hydrophilic in nature, FCA elutes before the majority of
analytes in reverse-phase HPLC. The FAC derivatives studied are extremely stable, with identical chromatograms obtained for samples analyzed one week apart
when stored at 5C. This derivatization technique may be used to monitor the
enzymatic hydrolysis of 200 nmol of Gly-Phe-Leu. Using FAC, a substance P
fragment that has proline at the N-terminus and therefore cannot be tagged with
either OPA or NDA was successfully detected with detection limits in the 500
fmol (25 nM) range [139].
5.

3,6-Dinitrophthalic Anhydride (DNPT)

DNPT has been implemented as a derivatization reagent for the electrochemical


detection of peptides. DNPT reacts quickly (second timescale) through the following scheme (Figure 8) with a variety of simple peptides, including Val-Val.
One advantage of this derivatization approach is that only single derivatives are

384

Sahlin et al.

Figure 8 Reaction of DNPT with a primary amine. (From Ref. 140.)

formed even when the peptide is multiply substituted, as in the case of Lys-LysLys. Additionally, the derivatives are stable for several days. Following HPLC
separation and reduction at 0.24 V versus Ag/AgCl, detection limits in the 1
pmol range are achieved for DNPT-Val-Val [140].

6. p-Nitrophenyl 2,5-Dihydroxyphenylacetate bis-Tetrahydropyranyl


Ether (NDTE)
Hydroquinone-based derivatization reagents have also been suggested for reaction
with primary and/or secondary amines to form electroactive products. Recently
two new reagents for this purpose were developed [141]. The most promising is
NDTE. The reaction scheme for the NDTE derivatization of peptides is shown in
Figure 9 [141]. Maximum yield for step 1 is achieved in approximately 30 min at
25C; 3 hours is required for maximum yield to be achieved for step 2. Once
formed, the derivative is stable for one day when stored in an acidic environment.
One drawback associated with NDTE is that it is converted to electroactive
homogentisic acid during the derivatization process, which may cause interference
in trace chromatographic analyses. Nonetheless, oxidation at +0.30 V versus
Ag/AgCl allows for determination of Ile-Leu-OMe with detection limits in the 250
nM range [141]. NDTE has also been used as a derivatization reagent for analysis
of angiotensin II [142]. Using a radial-ow thin-layer electrochemical detection
cell at a potential of +0.25 V versus. Ag/AgCl, detection of angiotensin II is accomplished with a detection limit of 62.5 nM. A wide separation window between the
derivatization agent and the analyte is achieved using gradient elution, thereby
eliminating interference from NDTE. Determination of angiotensin II transport
across an in vitro cell culture model of the blood-brain barrier is possible, with
mean concentrations over 2 hours in the range of 80 to 495 nM.

Electrochemical Detection of Peptides

385

Figure 9 Reaction of NDTE with a primary amine. (Adapted from Ref. 141.)

7.

Photolysis

The use of light has also shown promise as a derivatization agent. This technique
simplies the derivatization procedure and is not characterized by the drawbacks
associated with chemical methods, including interference from unreacted derivatizing agent. The method may be implemented on-line (e.g., in a post-column
reactor following a separation) with limited band broadening. However, photolysis derivatization is limited to detection of peptides composed of aromatic or sulfur-containing amino acids. Photolysis increases selectivity at a number of
applied potentials. For example, with peptides containing tryptophan and tyrosine, no electrochemical signal is seen at +0.60 V without photolysis, but a
response is identied at +0.80 V versus Ag/AgCl [143]. In comparison, with photolysis a signal is observed at +0.60 V and the signal obtained at +0.80 V vs.
Ag/AgCl is two or three times higher than without photolysis [143]. This is attributed to the fact that the stable photoproducts of some electroactive amino acids,
including 3-hydroxytyramine (DOPA, the major product of tyrosine photolysis),
are readily oxidized [144]. A similar increase in selectivity after photolysis is

386

Sahlin et al.

exhibited with peptides containing methionine and cystine [143]. In comparison,


cysteine displayed decreased sensitivity following photolysis [143,144].
Using this approach, separation of a number of peptides including, but not
limited to, Phe-Gly-Gly-Phe, Tyr-Leu, and Lys-Cys-Thr-Cys-Cys-Ala has been
achieved using ion-pair chromatography and gradient elution with detection limits around 21 pmol (1 M) obtained for ion-pair chromatography at +0.80 V versus Ag/AgCl [143]. This technique may be applied as an alternative to UV
absorbance detection for peptide mapping of tryptic peptides from bovine
cytochrome C. Using electrochemical detection, detection limits of 20 g (1
mg/mL, 20 L injected) of cytochrome C are achieved [145]. This is an order of
magnitude improvement over UV absorbance. In addition, determination of peptide 520 in serum is achieved using post-column photolysis as the derivatization
method. However, in this case, electrochemical detection is inferior to UV
absorbance. This is attributed to the peptides lack of an aromatic amino acid,
disulde, or thiol (peptide 520 contained only a thio-ether functionality) [146].

8.

Biuret Reaction

Theory and Mechanism. The classic biuret reaction is another approach to


peptide derivatization and is characterized by reaction of a peptide with a Cu(II)tartrate solution at alkaline pH. A generic complexation reaction of Cu(II) with a
tetrapeptide is given in Figure 10 [147]. Cu(II) will rst coordinate with the amine
terminus and then with the carbonyl of the amide bond. Coordination with the carbonyl draws electron density away from the amide nitrogen, thereby lowering the
amide pKa [148]. This process enables the Cu(II) to coordinate to successive
amido nitrogens. The redox potentials for the Cu(II)-NNNO (atoms bonding to
the Cu(II) are three nitrogens and an oxygen), and Cu(II)-NNNN forms are +0.72
and +0.42 V versus Ag/AgCl, respectively [149].
A number of complexes, including Cu(II)-Gly-Gly-Gly, undergo quasireversible oxidation to Cu(III)-Gly-Gly-Gly at potentials ranging from +0.50 V
to +0.90 V versus NHE [148,150]. Because the reaction is chemically reversible,
a dual electrode detection system can be easily employed for added selectivity.
Specically, Cu(II)-peptide is oxidized at an upstream anode and the resulting
Cu(III)-peptide complex is reduced at a downstream cathode back to Cu(II)-peptide (Figure 11). The biuret detection scheme holds advantages over others in that
the reaction chemistry occurs through the amide backbone and therefore allows
for determination of peptides lacking an electroactive group or primary amine.
Furthermore, the degree of selectivity for peptide detection using the biuret
scheme is high because the applied anodic potential is relatively low and the
cathodic signal is useable. Figure 12 shows the half-wave potential ranges for a
number of peptide classes.

Electrochemical Detection of Peptides

387

Figure 10 Reaction of Cu(II) with a peptide. (Adapted from Ref. 147.)

Figure 11 Application of a dual-electrode electrochemical detector to products of the


biuret reaction.

388

Sahlin et al.

Figure 12 Half-wave potential ranges for tyrosine, tryptophan, and a number of peptide
classes versus Ag/AgCl. Here XXD represents a peptide composed of any amino acid in
the rst and second position with Asp in the third position.

Weber et al. [111,147,149156] have extensively studied the biuret reaction as a means of detecting peptides separated by HPLC. They identied a number of operating parameters that inuence sensitivity, including post-column temperature, reaction time, buffer composition, Cu(II) concentration, peptide
structure, and pH.
Post-Column Temperature. In general, most of the peptides studied possess increased anodic sensitivities when the post-column temperature is elevated
(50C compared to 30C) demonstrating a kinetic barrier to complex formation
[154]. At a higher temperature, more of the Cu(II)-peptide complexes form in the
reaction window. At the cathode, results for nonelectroactive (in the absence of
Cu(II)) and electroactive peptides vary. The nonelectroactive peptides have
increased sensitivities at elevated temperatures because there is more current at
the upstream electrode, therefore more Cu(III)-peptide is formed. However, the
electroactive peptides generally are characterized by a decrease in cathode sensi-

Electrochemical Detection of Peptides

389

tivity at 50C resulting from degradation of the Cu(III)-peptide complexes (Figure 11).
Reaction Time. Intuitively, a longer post-column reactor will increase the
electrochemical signal. This is true for larger peptides, N-acylated peptides, and
peptides with pyroglutamate in the rst position, but for small peptides (up to
eight amino acids) without those attributes listed, the longer post-column reaction
time does not increase peak height and only leads to peak broadening [154]. Tsai
et al. [151] examined pre-column formation of the biuret complex as an alternative to post-column derivatization. This allows for the increased reaction time
needed for longer peptides (> eight amino acids) and those containing proline.
Buffer Composition. Investigation of a number of buffer systems (pH 9.8)
including bicarbonate/carbonate (pH 9.8), boric acid/sodium hydroxide (pH 9.8
and 8.0), boric acid/glycerol/sodium hydroxide (pH 9.8), and phosphate (pH 8)
revealed the bicarbonate buffer system to be optimal [154]. Phosphate buffers
appear to inhibit, in comparison to borate, the formation or the electrochemistry
of the complex and should be avoided.
Cu(II) Concentration. In addition to the above-mentioned factors affecting sensitivity, Chen and coworkers [155] determined the biuret reaction to be
zero order with respect to Cu(II), implying that an excess concentration of Cu(II)
will not improve reaction rate or sensitivity.
Amino Acid Composition and Peptide Structure. Similar to some of the
previously mentioned derivatization agents, peptides containing electroactive
amino acids exhibit increased anodic sensitivities over nonelectroactive peptides
in the presence of Cu(II), resulting from the additive effect of the amino acid and
Cu(II) oxidation. At the cathode, sensitivities of electroactive peptides are equal
to those for nonelectroactive peptides except for shorter tripeptides [157] where,
in some cases, cathode signals are repressed by the presence of tyrosine.
The inuence of tyrosine on the detection of Cu(II) peptide complexes has
been examined by Tsai et al. [157]. Generally, the presence of tyrosine decreases
cathodic sensitivity with the highest sensitivity observed at both anode and cathode when tyrosine is in the amine terminal position. In addition, there are a number of electron transfer reactions that can occur when the Cu(II) peptide complex
is oxidized or reduced that may further affect the electrochemical measurement.
These are discussed elsewhere in detail [157].
The biuret reaction is best suited for detection of peptides three amino acid
residues and longer but, nonetheless, a number of dipeptides are detectable at reasonable potentials [156]. Although N-terminal dipeptide amides require a high
pH (pH 12) for Cu(II) complexation, they are detected at relatively low potentials
(~+0.50 V vs. Ag/AgCl). In comparison, C-terminal dipeptide amides exhibit a

390

Sahlin et al.

half-wave potential of +0.65 V versus Ag/AgCl, and dipeptides including GlyLeu have half-wave potentials of +0.90 V versus Ag/AgCl [156].
Peptides containing proline, including bradykinin and substance P, have
low sensitivities resulting from structural limitations caused by the amido nitrogen being unavailable for Cu(II) complexation [155]. This type of structural limitation is not encountered with pyroglutamyl peptides or those containing disulde bonds.
In Table 2, starting conditions necessary for method development with the
ultimate goal of detecting Cu(II)-peptide complexes are summarized. The term
ordinary is used to describe peptides in which there is nothing particularly noteworthy about their electrochemistry or Cu(II) coordination chemistry.
Applications: HPLC. The biuret detection scheme is compatible with the
most common mobile phase used in peptide determinations, namely 0.1% TFA
(triuoroacetic acid) with an acetonitrile gradient [153]. Detection limits for nonelectroactive peptides at the cathode range from 20200 fmol (100 L of 0.21
nM) and 640 fmol (100 L of 60400 pM) at the anode for electroactive peptides.
Additionally, the biuret detection scheme has been used to determine levels
of TP9201, an Arg-Gly-Asp integrin-binding peptide [152]. TP9201 is an ideal
candidate for the biuret chemistry because it is a cyclic cystine-bridged nonapeptide with an acetylated amine terminus, thus eliminating the possibility of detection using amine-reactive uorescent labels. Employing reverse-phase HPLC followed by electrochemical detection, TP9201 is separated and detected in serum

Table 2 Starting Conditions for Post-Column Peptide Determination Using Electrochemical Detection of Cu(II)-Peptide Complexes
Peptide
class
Small
(<8 AAs)
Ordinary
Large
(>18 AAs)
Ordinary
N-Blocked
XXDX . . .
and XXX

Post-column buffer
composition and pH

Applied potential vs.


Ag/AgCl (3M KCl)

Reaction
temperature

Reaction
time (min)

pH 9.8
Bicarbonate/carbonate

+0.50 V Anode
+0.10 V Cathode

50C

pH 9.8
Bicarbonate/carbonate

+0.50 V Anode
+0.10 V Cathode

50C

1.5

pH 9.8
Bicarbonate/carbonate
pH 6
Maleate

+0.50 V Anode
+0.10 V Cathode
+0.80 V Anode

50C

1.5

50C

Electrochemical Detection of Peptides

391

samples. The detection limit was found to be 20 nM (limited by interfering sample peaks), an improvement over conventional analysis.
Typically when the biuret reaction is used for post-column derivatization of
peptides (as above), a second pump must be employed to deliver the biuret
reagent. To eliminate this problem, a method using a solid-phase or in-line reactor containing copper metal has been employed [158,159]. This technique is
based on the formation of Cu(II)-peptide complexes by reaction of the peptides,
under basic conditions, with Cu(II) ions formed by passing through a reactor containing ne particle metallic copper. Using a two-electrode detection system with
the upstream electrode set at a potential to screen out interfering peaks and the
second electrode to detect the Cu(II)-peptide complexes, detection of Ala-AlaAla with detection limits of 5 ng (5 L injected) was achieved.
Applications: Capillary HPLC. Kennedy and coworkers [160] examined
the utility of pre-column Cu(II) peptide complex formation for detection of neuropeptides in dialysate using capillary HPLC. They discovered a noteworthy difference between glassy carbon and carbon ber electrodes. Data show that the
additive effect of tyrosine and Cu(II) oxidation is not convincingly seen at a carbon ber electrode. In the presence of Cu(II), the signal for Met-enkephalin is
increased 105 8% compared to des-Tyr-enkephalin at a glassy carbon electrode
but only 5 4% at a carbon ber electrode. It is hypothesized that Cu(II) modies the ber electrode surface, affecting the kinetics of either Cu(II) or tyrosine
oxidation. Employing preconcentration techniques, capillary HPLC, and amperometry using a carbon ber electrode, detection of vasopressin, bradykinin, oxytocin, and neurotensin was achieved with 7, 5, 20, and 59 pM (559 amol) detection limits respectively. This permits identication of vasopressin and bradykinin
in dialysate samples.
Applications: CE. In addition to its prevalent use in detection following
HPLC separation, the biuret detection strategy has also been used for electrochemical detection of peptides following CE separation. A Cu(II)-coated capillary has been used to eliminate the need for pre-column or on-column derivatization using a Cu(II)-containing buffer [161]. Detection limits on the order of 0.7
M were obtained for nonelectroactive peptides (di-, tri-, tetra-, and pentaglycines). Other peptides including Pro-Leu-Gly-amide were also successfully
detected with a 2 M detection limit.
Lunte and coworkers [162] recently showed the utility of on-column complexation using a Cu(II)-containing buffer for monitoring Leu-enkephalin metabolism in plasma. Using CE separation, biuret complexation, and electrochemical
detection, a complete separation of Leu-enkephalin and its ve metabolites was
achieved. Applicability of this technique to plasma samples allowed two major

392

Sahlin et al.

metabolites, Tyr and des-Tyr-Leu-enkephalin, to be identied after 30 min of


sampling.
9. Determination Based on photoluminescence Following
Electron Transfer
In photoluminescence following electron transfer (PFET), oxidation of the analyte
occurs in solution by a dissolved oxidant. The oxidant is a non-photoluminescent
species that becomes photoluminescent when reduced in the redox reaction with
the analyte. The only published work uses tris(2,2-bipyridyl) ruthenium(III/II),
Ru(bpy)33+/2+, which is non-photoluminescent as Ru(bpy)33+ but photoluminescent as Ru(bpy)32+ [163]. The structure of Ru(bpy)33+/2+ is shown in Figure 13. The
PFET detection scheme can be represented as:
Ru(bpy)32+ 3 Ru(bpy)33+ + e

(17)

Ru(bpy)33+ + peptide 3 Ru(bpy)32+ + peptideox

(18)

Ru(bpy)32+ + hvex 3 Ru(bpy)32+* 3 Ru(bpy)32+ + hvem

(19)

3+,

The reagent, Ru(bpy)3 is not stable in aqueous solution for extended time, so it
is generated externally by oxidation of Ru(bpy)32+ [Eq. (17)]. A high-efciency
(>99%) ow-through electrode containing carbon powder as working electrode
material is employed for this purpose. The Ru(bpy)33+ solution is then mixed with
the mobile phase downstream of the column where Eq. (18) occurs. Ru(bpy)33+ is
a strong oxidant (E0 = +1.24 V vs. NHE) and is able to oxidize both tyrosine and

Figure 13 Structure of Ru(bpy)33+/2+.

Electrochemical Detection of Peptides

393

tryptophan in peptides containing these amino acids [163]. From the use of
Ru(bpy)33+ as a reagent in electrochemiluminescence (or electrogenerated chemiluminescence) (ECL), it is known that Ru(bpy)33+ also oxidizes amine groups.
The reaction mechanisms for the reaction between Ru(bpy)33+ and amines is complicated, and several different mechanisms have been suggested in the literature
[164166]. However, the importance of amine oxidation on the PFET signal is
not known. Detection of the formed Ru(bpy)32+ (reaction (19)) is then achieved
employing a normal uorescence detector. The emission from Ru(bpy)32+* is formally phosphorescence but with a decay-time shorter than normal phosphorescence [165,167,168]. A detailed description of the photoluminescence properties
of Ru(bpy)32+ is available in the literature [167,168].
So far the PFET detection scheme has only been applied in HPLC systems
for determination of peptides containing tyrosine or tryptophan, such as dynorphin A and its fragments [163]. Detection limits were found to be around 0.57 nM
(with on-column preconcentration of 140 L injected sample) [163].

IV.

PRACTICAL CONSIDERATIONS

In order to achieve ultralow detection limits, the background must be reduced.


Background in the detection techniques discussed above mainly originates from
contaminants susceptible to oxidation present in the mobile phase, e.g., Fe(II),
chloride and different amines. Fe(II) often comes from the stainless steel used in
the pumps and can be avoided by using metal-free pumps built with PEEK (poly
ether ether ketone). Chloride and different amines most often are contained in the
solvents, electrolyte, or buffer components and can be eliminated by distillation
of the solvents and recrystallization of the salts. Contamination can also, in some
cases, be removed in-line by a scrubber or guard ow-through electrode (i.e., a
high-efciency ow-through electrode) [42,163]. For instance, a scrubber electrode can be constructed from porous electrode materials such as reticulated vitreous carbon or packed carbon particles (e.g., glassy carbon powder).
The amperometric signal is very sensitive to pulsation of the ow rate
because a change in the ow rate will result in a change in the diffusion layer
thickness and thereby a change in the mass transport of analytes toward the electrode surface. Pulsation of the ow rates is a consequence of the moving parts in
the pump and can be decreased by employing a pulse damper.
It has also been shown that baseline stability, when detection is performed
at high potentials (> +0.90 V vs. Ag/AgCl), is very sensitive to temperature
changes. Even minor variations in temperature (<1C) can result in drift in the
baseline [169].

394

Sahlin et al.

ACKNOWLEDGMENTS
We acknowledge NIH for nancial support through Grant GM 44842. E.S.
acknowledges The Swedish Foundation for International Cooperation in
Research and Higher Education (STINT) for a post doctoral scholarship. M.S. is
supported by the Swedish Natural Science Research Council.

REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22
23
24.
25.
26.

T. Hkfelt; C. Broberger; Z. Q. Xu; V. Sergeyev; R. Ubink; M. Diez, Neuropharmacol., 39: 1337 (2000).
D. R. Nssel, Naturwissenschaften, 87: 439 (2000).
F. L. Strand, Neuropeptides: Regulators of physiological processes, The MIT
Press, Cambridge, 1999.
G. OCuinn, Ed., Metabolism of brain peptides, CRC Press, New York, 1995.
K. Fuxe; L. F. Agnati, Eds., Volume transmission in the brain. Novel mechanisms
for neural transmission, Raven Press, New York, 1991.
S. G. Weber; J. T. Long, Anal. Chem., 60: 903A (1988).
S. G. Weber, in Detectors for Liquid Chromatography, E. S. Yeung, Ed., John
Wiley & Sons, Inc., New York, 1986, p. 229.
D. C. Johnson; S. G. Weber; A. M. Bond; R. M. Wightman; R. E. Shoup; I. S. Krull,
Anal. Chim. Acta, 180: 187 (1986).
D. M. Morgan; S. G. Weber, Anal. Chem., 56: 2560 (1984).
J. M. Elbicki; D. M. Morgan; S. G. Weber, Anal. Chem., 56: 978 (1984).
S. G. Weber, Anal. Chem., 61: 295 (1989).
V. Brabec; V. Mornstein, Biophys. Chem., 12: 159 (1980).
C. D. Kilts; D. L. Knight; C. B. Nemeroff, Life Sci., 59: 911 (1996).
G. W. Bennett; M. P. Brazell; C. A. Marsden, Life Sci, 29: 1001 (1981).
N. C. Reynolds Jr.; B. M. Kissela; L. H. Fleming, Electroanalysis, 7: 1177 (1995).
N. T. Nguyen; M. Z. Wrona; G. Dryhurst, J. Electroanal. Chem., 199: 101 (1986).
S. M. MacDonald; S. G. Roscoe, Electrochim. Acta, 42: 1189 (1997).
B. Malfoy; J. A. Reynaud, J. Electroanal. Chem., 114: 213 (1980).
J. A. Reynaud; B. Malfoy; A. Bere, J. Electroanal. Chem., 116: 595 (1980).
N. T. Nguyen; M. Z. Wrona; G. Dryhurst, J. Electroanal. Chem., 211: 257 (1986).
L. Ag; A. Gonzlez-Corts; P. A. Yez-Sedeo; J. M. Pingarrn, Anal. Chim.
Acta, 401: 145 (1999).
H. Shen; M. W. Lada; R. T. Kennedy, J. Chromatogr. B, 704: 43 (1997).
V.-P. Ranta; A. Urtti; S. Auriola, J. Chromatogr. A, 766: 85 (1997).
V. P. Ranta; K. M. Hmlinen; S. Auriola; A. Urtti, J. Chromatogr. B, 709: 1
(1998).
A. F. Spatola; D. E. Benovitz, J. Chromatogr., 327: 165 (1985).
A. Sauter; W. Frick, J. Chromatogr., 297: 215 (1984).

Electrochemical Detection of Peptides


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.
60.
61.

395

J. Meriluoto; B. Kincaid; M. R. Smyth; M. Wasberg, J. Chromatogr. A, 810: 226


(1998).
A. L. Drumheller; H. Bachelard; S. St-Pierre; F. B. Jolicoeur, J. Liq. Chromatogr.,
8: 1829 (1985).
C. Kim; R. Cheng; S. R. George, J. Chromatogr., 494: 67 (1989).
L. H. Fleming; N. C. Reynolds, J. Chromatogr., 375: 65 (1986).
S. Shibanoki; S. B. Weinberger; K. Ishikawa; J. L. Martinez, J. Chromatogr., 532:
249 (1990).
L. S. Monger; C. J. Olliff, J. Chromatogr., 577: 239 (1992).
M. W. White, J. Chromatogr., 262: 420 (1983).
S. Mousa; D. Couri, J. Chromatogr., 267: 191 (1983).
L. S. Monger; C. J. Olliff, J. Chromatogr., 595: 125 (1992).
S. A. Mousa; G. R. Van Loon, Life Sci., 37: 1795 (1985).
J. Cummings; A. MacLellan; S. P. Langdon; E. Rozengurt; J. F. Smyth, J. Chromatogr. B, 653: 195 (1994).
C. E. Linn Jr.; W. L. Roelofs, Insect Biochem. Molec. Biol., 23: 367 (1993).
D. L. Rabenstein; R. Saetre, Clin. Chem., 24: 1140 (1978).
A. Sauter; W. Frick, Anal. Biochem., 133: 307 (1983).
T. Nabeshima; M. Hiramatsu; S. Noma; M. Ukai; M. Amano; T. Kameyama, Res.
Commun. Chem. Pathol. Pharmacol., 35: 421 (1982).
G. W. Bennett; J. V. Johnson; C. A. Marsden In Neuropeptides: A Methodology; G.
Fink and A. J. Hamar, Eds., John Wiley & Sons Inc., New York, 1989, p 125.
R. Dawson Jr.; J. P. Steves; J. F. Lorden; S. Oparil, Peptides, 6: 1173 (1985).
G. Forsberg; I. Bednar; G. A. Qureshi; P. Eneroth; P. Sdersten, J. Neuroendocr.,
3: 79 (1991).
L. H. Fleming; N. C. Reynolds Jr., J. Chromatogr., 431: 65 (1988).
J. Zhou; S. M. Lunte, Anal. Chem., 67: 13 (1995).
J. A. Reynaud; B. Malfoy; P. Canesson, J. Electroanal. Chem., 114: 195 (1980).
J. Koryta; J. Prad, J. Electroanal. Chem., 17: 185 (1968).
D. G. Davis; E. Bianco, J. Electroanal. Chem., 12: 254 (1966).
J. P. Millington; F. C. Walsh, J. Electroanal. Chem., 375: 1 (1994).
Z. Samec; Z. Malysheva; J. Koryta, J. Electroanal. Chem., 65: 573 (1975).
V. A. Bogdanovskaya; A. Y. Safronov; M. R. Tarasevich; A. S. Chernyak, J. Electroanal. Chem., 202: 147 (1986).
S. A. Wring; J. P. Hart; B. J. Birch, Analyst, 114: 1563 (1989).
M. K. Halbert; R. P. Baldwin, Anal. Chem., 57: 591 (1985).
J. H. Zagal; P. Herrera, Electrochim. Acta, 30: 449 (1985).
J. H. Zagal; M. J. Aguirre; C. G. Parodi; J. Sturm, J. Electroanal. Chem., 374: 215
(1994).
Z. Wang; D. Pang, J. Electroanal. Chem., 283: 349 (1990).
Z. Deng; Z. Wang; C. Li; Q. Zha; G. Xuexiao; H. Xuebao, Chem. Abstr., 108:
12940t (1987).
X. Qi; R. P. Baldwin, J. Electroanal. Chem., 143: 1283 (1996).
J. A. Cox; T. J. Gray, Anal. Chem., 62: 2742 (1990).
A. Przyjazny; J. A. J. A. Cox, Electroanalysis, 5: 657 (1993).

396

Sahlin et al.

62.
63.
64.
65.
66.
67.

J. A. Cox; E. Dabek-Zlotorzynska, J. Chromatogr., 543: 226 (1991).


H. Li; E. Wang, Microchem. J., 49: 91 (1994).
W. Hou; E. Wang, J. Electroanal. Chem., 316: 155 (1991).
N. D. Popovic; D. C. Johnson, Electroanalysis, 11: 934 (1999).
M. T. Stankovich; A. J. Bard, J. Electroanal. Chem., 75: 487 (1977).
M. Heyrovsky; P. Mader; S. Vavricka; V. Vesel; M. Fedurco, J. Electroanal.
Chem., 430: 103 (1997).
M. Heyrovsky; S. Vavricka, J. Electroanal. Chem., 423: 125 (1997).
J. Koryta; J. Prad, J. Electroanal. Chem., 17: 177 (1968).
M. Heyrovsky; P. Mader; V. Vesel; M. Fedurco, J. Electroanal. Chem., 369: 53
(1994).
G. Arai; T. Sugaya; M. Sakamoto; I. Yasumor, Bull. Chem. Soc. Jpn., 65: 594
(1992).
Z. Wang; C. Li; Z. Deng; Q. Zha; H. Xuebao, Chem. Abstr., 105: 199030z (1986).
Z. Deng; Z. Wang; C. Li; Q. Zha; H. Xuebao, Chem. Abstr., 108: 64491s (1987).
J. Prad; J. Koryta, J. Electroanal. Chem., 17: 167 (1968).
P. M. Krien; V. Margou; M. Kermici, J. Chromatogr., 576: 255 (1992).
L. Manna; L. Valvo; P. Betto, J. Chromatogr. A, 846: 59 (1999).
A. Rodriguez-Ariza; F. Toribio; J. Lpez-Barea, J. Chromatogr. B, 656: 311
(1994).
N. C. Smith; M. Dunnett; P. C. Mills, J. Chromatogr. B, 673: 35 (1995).
J. Lakritz; C. G. Plopper; A. R. Buckpitt, Anal. Biochem., 247: 63 (1997).
G. Bordin; F. Cordeiro Raposo; A. R. Rodriguez, Chromatographia, 34: 146
(1994).
R. C. Rose; A. M. Bode, Biochem. J., 306: 101 (1995).
C. Y. OGara; K. R. Maddipati; L. J. Marnett, Chem. Res. Toxicol., 2: 295 (1989).
I. Mefford; R. N. Adams, Life Sci., 23: 1167 (1978).
E. Bousquet; N. A. Santagati; T. Lancetta, J. Pharm. Biomed. Anal., 7: 643 (1989).
K. Iriyama; T. Iwamoto; M. Yoshiura, J. Liq. Chromatogr., 9: 955 (1986).
P. R. C. Harvey; R. G. Ilson; S. M. Strasberg, Clin. Chim. Acta, 180: 203 (1989).
S. A. Wring; J. P. Hart; B. J. Birch, Talanta, 38: 1257 (1991).
M. K. Halbert; R. P. Baldwin, J. Chromatogr., 345: 43 (1985).
S. A. Wring; J. P. Hart; B. J. Birch, Analyst, 114: 1571 (1989).
M. Zhong; S. M. Lunte, Anal. Chem., 71: 251 (1999).
L. A. Allison; R. E. Shoup, Anal. Chem., 55: 8 (1983).
T. J. OShea; S. M. Lunte, Anal. Chem., 65: 247 (1993).
A. F. Stein; R. L. Dillis; C. D. Klaassen, J. Chromatogr., 381: 259 (1986).
E. G. Demaster; F. N. Shirota; B. Redfern; D. J. W. Goon; H. T. Nagasawa, J. Chromatogr., 308: 83 (1984).
S. M. Lunte; P. T. Kissinger, J. Chromatogr., 317: 579 (1984).
S. M. Lunte; P. T. Kissinger, J. Liq. Chromatogr., 8: 691 (1985).
D. L. Rabenstein; R. Saetre, Anal. Chem., 49: 1036 (1977).
W. A. Kleinman; J. P. Richie Jr., J. Chromatogr. B, 672: 73 (1995).
J. P. Richie Jr.; C. A. Lang, Anal. Biochem., 163: 9 (1987).

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.

Electrochemical Detection of Peptides


100.
101.
102.
103.
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.
129.
130.
131.
132.
133.
134.
135.
136.
137.

397

D. Dupuy; S. Szabo, J. Liq. Chromatogr., 10: 107 (1987).


Y. Sun; D. L. Smith; R. E. Shoup, Anal. Biochem., 197: 69 (1991).
C. Lazure; J. Rochemont; N. G. Seidah; M. Chrtien, J. Chromatogr., 326: 339
(1985).
C. T. Garvie; K. M. Straub; R. K. Lynn, J. Chromatogr., 413: 43 (1987).
Y. Sun; P. C. Andrews; D. L. Smith, J. Protein Chem., 9: 151 (1990).
D. Shea; W. A. MacCrehan, Anal. Chem., 60: 1449 (1988).
W. Jin; W. Li; Q. Xy, Electrophoresis, 21: 774 (2000).
W. Jin; Y. Wang, Anal. Chim. Acta, 343: 231 (1997).
M. zcimder; A. J. H. Louter; H. Lingeman; W. H. Voogt R. W. Frei; M. Bloemendal, J. Chromatogr., 570: 19 (1991).
K. Isaksson; J. Lindquist; K. Lundstrm, J. Chromatogr., 324: 333 (1985).
J.-G. Chen; E. Vinski; K. Colizza; S. G. Weber, J. Chromatogr. A, 705: 171 (1995).
S. G. Weber; H. Tsai; M. Sandberg, J. Chromatogr., 638: 1 (1993).
P. Luo; F. Zhang; R. P. Baldwin, Anal. Chem., 66: 1702 (1991).
N. A. Hampson; J. B. Lee; K. I. MacDonald, J. Electroanal. Chem., 34: 91 (1972).
P. Luo; R. P. Baldwin, Electroanalysis, 4: 393 (1992).
Y. Xie; C. O. Huber, Anal. Chem., 63: 1714 (1991).
J. Ye;
 R. P. Baldwin, Anal. Chem., 66: 2669 (1994).
K. Stulk; V. Packov; G. Jokuszies, J. Chromatogr., 436: 334 (1988).
W. T. Kok; H. B. Hanekamp; P. Bos; R. W. Frei, Anal. Chim. Acta, 142: 31 (1982).
K. Stulk; V. Packov; K. Le;
 B. Hennissen, Talanta, 35: 455 (1988).
H. Wang; V. Packov; K. Stulk, J. Chromatogr., 509: 245 (1990).
C. E. Engstrom-Silverman; A. G. Ewing, J. Microcolumn Sep., 3: 141 (1991).
L. E. Welch; W. R. LaCourse; D. A. Mead Jr.; D. C. Johnson; T. Hu, Anal. Chem.,
61: 555 (1989).
W. R. LaCourse, Analysis, 21: 181 (1993).
D. C. Johnson; W. R. LaCourse, Anal. Chem., 62: 589A (1990).
W. R. LaCourse, Pulsed electrochemical detection in high-performance liquid
chromatography, John Wiley & Sons, New York, 1997.
W. R. LaCourse; G. S. Owens, Anal. Chim. Acta, 307: 301 (1995).
J. A. M. van Riel; C. Oliema, Anal. Chem., 67: 3911 (1995).
P. J. Vandeberg; D. C. Johnson, Anal. Chem., 65: 2713 (1993).
P. J. Vandeberg; D. C. Johnson, Anal. Chim. Acta, 290: 317 (1994).
M. J. Donaldson; M. W. Adlard, J. Chromatogr., 509: 347 (1990).
M. A. Nussbaum; J. E. Przedwiecki; D. U. Staerk; S. M. Lunte; C. M. Riley, Anal.
Chem., 64: 1259 (1992).
M. Koller; H. Eckert, Anal. Chim. Acta, 352: 31 (1997).
L. A. Allison; G. S. Mayer; R. E. Shoup, Anal. Chem., 56: 1089 (1984).
O. Orwar; S. Folestad; S. Einarsson; P. Andin; M. Sandberg, J. Chromatogr., 566:
39 (1991).
S. Einarsson, J. Chromatogr., 348: 213 (1985).
L. Canevari; R. Vieira; M. Aldegunde; F. Dagani, Anal. Biochem., 205: 137 (1992).
W. Buchberger; K. Winsauer, Anal. Chim. Acta, 196: 251 (1987).

398

Sahlin et al.

138.
139.
140.
141.

G. D. Li; I. S. Krull; S. A. Cohen, J. Chromatogr. A, 724: 147 (1996).


R. L. Cox; T. W. Schneider; M. D. Koppang, Anal. Chim. Acta, 262: 145 (1992).
J. L. Meek, J. Chromatogr., 266: 401 (1983).
M. J. Rose; S. M. Lunte; R. G. Carlson; J. F. Stobaugh, Anal. Chem., 71: 2221
(1999).
M. J. Rose; J. M. Rose; S. M. Lunte; K. L. Audus; R. G. Carlson; J. F. Stobaugh,
Anal. Chim. Acta, 394: 299 (1999).
L. Dou; I. S. Krul, Anal. Chem., 62: 2599 (1990).
L. Dou; I. S. Krull, Electroanalysis, 4: 381 (1992).
L. Chen; I. S. Krull, Electroanalysis, 6: 1 (1994).
L. Chen; J. Mazzeo; I. S. Krull; S. L. Wu, J. Pharm. Biomed Anal., 11: 999 (1993).
J. G. Chen; S. J. Woltman; S. G. Weber, Adv. Chromatogr., 36: 273 (1996).
D. W. Margerum, Pure Appl. Chem., 55: 23 (1983).
S. J. Woltman; M. R. Alward; S. G. Weber, Anal. Chem., 34: 541 (1995).
A. M. Warner; S. G. Weber, Anal. Chem., 61: 2664 (1989).
H. Tsai; S. G. Weber, J. Chromatogr., 542: 345 (1991).
S. J. Woltman; J. G. Chen; S. G. Weber; J. O. Tolley, J. Pharm. Biomed. Anal., 14:
155 (1995).
J. G. Chen; S. J. Woltman; S. G. Weber, J Chromatogr. A, 691: 301 (1995).
J. G. Chen; S. G. Weber, Anal. Chem., 67: 3596 (1995).
J. G. Chen; M. Logman; S. G. Weber, Electroanalysis, 11: 331 (1999).
H. Tsai; S. G. Weber, J. Chromatogr., 515: 451 (1990).
H. Tsai; S. G. Weber, Anal. Chem., 64: 2897 (1992).
H. Kubo; A. Tsujimura; T. Kinoshita, Anal. Sci., 10: 579 (1994).
H. Kubo, Anal. Chim. Acta, 287: 211 (1994).
H. Shen; S. R. Witowski; B. W. Boyd; R. T. Kennedy, Anal. Chem., 71: 987 (1999).
M. Deacon; T. J. OShea; S. M. Lunte, J. Chromatogr. A, 652: 377 (1993).
A. J. Gawron; S. M. Lunte, Electrophoresis, 21: 3205 (2000).
S. J. Woltman; W. R. Even; S. G. Weber, Anal. Chem., 71: 1504 (1999).
A. W. Knight, Trends Anal. Chem., 18: 47 (1999).
R. D. Gerardi; N. W. Barnett; S. W. Lewis, Anal. Chim. Acta, 378: 1 (1999).
W.-Y. Lee, Microchim. Acta, 127: 19 (1997).
J. R. Lakowicz, Principles of Fluorescence Spectroscopy, Volume 4, Kluwer Academic/Plenum Publishers, New York, 1999, p 573.
J. N. Demas; B. A. DeGraff, in Topics in Fluorescence Spectroscopy, Volume 4:
Probe Design and Chemical Sensing; J. R. Lakowicz, Ed., Plenum Press, New York,
1994, p 71.
J. T. Bretz; P. R. Brown, J. Chromatogr. Sci., 26: 310 (1988).

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.
167.
168.

169.

12
Microfabrication of Electrode
Surfaces for Biosensors
Steven E. Rosenwald and Werner G. Kuhr
University of California, Riverside, Riverside, California

I.

INTRODUCTION

Sensitive and reliable discrimination of the analytical signal ascribed to one specic chemical species within a functioning organismfor example, in vivo
detection of neurotransmittersis invariably complicated by the complex matrix
of the biological system being monitored. Similarly, in situ detection of a single
analyte of interest without prior treatment (e.g., separation) to remove interfering
species from the complex mixture of a real-life sample continues to be one of the
most formidable challenges facing analytical chemists today. In a recent review
of chemical sensors in Analytical Chemistry: Fundamental Reviews, it was noted
that the quest for better selectivity remains the cornerstone of the chemical sensing research; and biosensing has been the most reviewed topic in the past 5 years
[1]. A biosensor has been described as an analytical detector (such as an electrode
or beroptic transducer) whose selectivity is enhanced by immobilizing a sensitive and selective biological element (typically an enzyme) within close proximity of the sensor [2]. The development of new biosensors is seen as essential to
solving the inherent difculties of many challenging in situ and in vivo analyses.
It is beyond the scope of this chapter to attempt to describe the myriad combinations of biological and sensor elements possible, so only improvements in the
design of amperometric biosensors, commonly referred to as enzyme-modied
electrodes, will be discussed. The favored conguration for biosensors utilizes
amperometry (measurement of electric current) for transduction of the chemical
signal into an electronic signal [3]. Electrochemical instrumentation is simple and
399

400

Rosenwald and Kuhr

inexpensive; it is sensitive over a wide concentration range; and colored or turbid


samples (a potential problem in spectroscopic analyses) do not cause interference
in electrochemical analyses [3].
In the simplest format, amperometric biosensors are constructed in such a
way that in the absence of other electroactive sample components, the faradaic
current produced at the electrode surface is produced solely due to presence of the
product of an enzyme-catalyzed reaction as illustrated in Figure 1. Three possible
congurations available for an amperometric biosensing electrode system [4] are
illustrated in Figure 2:
1. The enzyme is added to the analytical solution. As depicted in Figure
2A, an electrode is placed in a solution of enzyme and cofactor. On
addition of substrate, the change in concentration of reduced cofactor
may be monitored with respect to time. For example, an electrode is
immersed in a phosphate buffer solution containing glutamate dehydrogenase (enzyme) and NAD+ (cofactor) at a pH of 8.5. Upon addition
of a small amount of the glutamate (substrate), an increase in NADH
concentration (reduced cofactor) may be measured using a carbon ber
microelectrode.
2. The enzyme is immobilized upstream from the sensing electrode. This
system requires the ow of a solution containing enzyme cofactor, from
left to right as drawn in Figure 2B. In a manner analogous to addition of
substrate to a solution containing the enzyme, the immobilized enzyme
(e.g., glutamate dehydrogenase) catalyzes production of a reduced
cofactor (e.g., NAD+ to NADH) only in the presence of the substrate
(e.g., glutamate.) A downstream electrode then monitors this enzymatically generated reduced cofactor. Note that in this conguration the
ow rate will inuence the time resolution of the measurement.
3. The enzyme is attached to the sensing electrode itself. Solution ow is
shown in Figure 2C only to suggest a method of analyte introduction to
the biosensor. This system does not require ow past the enzyme and
detector but instead relies on diffusion of substrate to the enzyme, and
diffusion of the enzymatically generated reduced cofactor from the
immobilized enzyme to the working electrode. As will be shown, immobilization of enzymes on the electrode surface often reduces the detector
efciency because both enzyme conversion efciency and diffusion of
analytes can limit the time response of such a system. This type of system is amenable to the immobilization of the cofactor as well (e.g.,
wired enzyme electrodes [57], which use intrinsic FAD/FADH2 as
cofactors).
Immobilized enzyme biosensors will allow both accurate spatial resolution of the
measurement area and reuse of enzymes, which are sometimes costly. In order to

Microfabrication of Electrode Surfaces for Biosensors

401

Figure 1 Enzymes as signal transducers. As drawn, an enzyme specic for only one substrate is immobilized in close proximity to the electrode surface. The substrate (e.g., glutamate) is oxidized by the enzyme to the corresponding enzyme product (e.g., -ketoglutarate) with concurrent reduction of a cofactor in a 1:1 ratio. (NAD+ is reduced to NADH
for dehydrogenase enzymes, and molecular oxygen is reduced to peroxide in the case of
oxidase enzymes.) Of the four species present in solution, only the product of the cofactor
reduction (NADH or H2O2) is electrochemically active and produces an analytical signal,
so the enzyme substrate (glutamate in this example) is transduced to an electroactive
species by the enzyme. Barring introduction of NADH (or peroxide) to the solution, any
increase in faradaic current may then be attributed to the presence of the enzyme substrate
alone.

402

Rosenwald and Kuhr

Figure 2 Amperometric biosensor congurations. Detailed descriptions are in the text.


(A) Addition of substrate to a solution containing enzyme and cofactor. (B) Immobilized
enzyme upstream from sensing electrode. (C) Enzyme immobilized directly on detecting
surface.

Microfabrication of Electrode Surfaces for Biosensors

403

satisfy the further requirement that the biosensor must be implanted with minimal
tissue damage in order to make in vivo neurological measurements, with subsecond time response possible, only systems involving enzymes attached directly to
the working electrode surface will be given further consideration.
II.

ENZYME SELECTION

The utility of the immobilized enzyme electrode depends on factors such as the
method of immobilization, the chemical and physical conditions of use (pH, temperature, ionic strength of the sample, long-term stability of the biocomponent,
interfering species, enzyme inhibitors, availability of cofactors, etc.), the activity
and stability of the enzyme once immobilized, the stability of the electrochemical
sensor, the response time, and the storage conditions required. The signal transducer (electrode surface) must be able to oxidize (or reduce) the enzyme product
(or the enzyme itself) while producing a measurable faradaic current; it must possess some quality that will allow immobilization of the enzyme (such as the availability of functional groups on carbon [8] or the ability of thiols to form selfassembled monolayers [SAMs] on gold [Au]); and it must be capable of working
in the environment in which the analysis will be carried out. However, protein
adsorption is well known at graphite [9], carbon [10] gold [11,12] and platinum
[13,14] and may cause inhibition of electron transfer processes [15] and/or diffusion of analytes or enzymatic products [16]. Interfering compounds, both those
that are electroactive at useful potentials and those that are inhibitory to the
biosensing enzymes, must be identied and determined to be absent from the
sampling environment or eliminated, if possible. Immobilization of the enzyme
on the electrode surface must be reproducible; and the microenvironment of the
immobilized enzyme must be amenable to diffusion of both analyte and product,
allow normal functioning of the protein, and be stable enough to permit a reasonable working lifetime for the sensor.
Fabrication of an enzyme-modied electrode, and indeed any biosensor,
requires stepwise selection and integration of a number of complex components.
First, the redox enzyme chosen must have the required selectivity for the analyte
as well as a product that is detectable at the signal transducer (in this case the electrode surface). A search of the literature reveals that a relatively small number of
enzymes have been reported as functional on biosensors, and two classes of
enzymes (oxido-reductase and dehydrogenase) predominate. Notably, the oxidoreductase family, and glucose oxidase in particular, account for the majority of all
reports in the literature. Other factors that complicate the choice of enzyme
include the necessity for a derivatization procedure that does not destroy the
active site of the enzyme. In some cases, proper orientation of the enzymes active
site must also be considered. Glutamate dehydrogenase, a protein whose conformation (and therefore activity) is dependent on electrostatic forces, has been

404

Rosenwald and Kuhr

shown to strongly physisorb at a carbon ber surface with subsequent loss of


activity [15], and changes in the pH within the microenvironment at the electrode
surface have been shown to inuence enzyme activity [17].
Another consideration in the fabrication of enzyme-modied electrodes is
the time resolution required for the measurement. Typically, the process of
enzyme immobilization slows the response of the electrode. In some systems of
this type, the enzyme itself, rather than the product of the enzyme catalysis, is oxidized (or reduced) [6,7]. If electron transfer to (or from) the enzyme itself produces the faradaic current being monitored, true monolayer coverage and maximum contact between the surface and the enzyme is needed, along with proper
orientation of the redox center of the enzyme. However, if the faradaic current is
due to electron transfer to or from the product of enzyme catalysis, coverage of
the electrode surface with a monolayer or more of protein can hinder analyte diffusion and/or electron transfer (commonly known as electrode fouling) and
adversely affect measurement sensitivity. There are relatively few enzymes capable of direct electron transfers, so the majority of these systems measure the electron transfer of an enzyme product instead of the enzyme itself. This chapter will
discuss methods used to enhance surface coverage with active enzyme while
maintaining a rapid time response and overcoming the problems associated with
electrode fouling.
III.

ELECTRODE CONFIGURATION

The majority of reports of enzyme-modied electrodes describe macroelectrodes


(diameter greater than 100 m). These macroelectrodes are then used in constant
potential experiments and commonly show time resolution of 3 to 12 s [18,19].
The Ewing group has shown that there is a linear relationship between sensor tip
diameter and response time [20]. Sensors with tip diameters of less than 100 nm
might provide response times of a few milliseconds if enzyme kinetics are not rate
limiting. The only limitation might be whether there is sufcient signal to provide
adequate sensitivity for the measurement.
This review will apply general information regarding modication of electrodes with enzymes to the modication of microelectrodes (<100 m diameter)
to make biosensors in order to take advantage of the numerous advantages associated with microelectrodes, such as enhanced diffusion and reduced IR drop
[2022]. Microelectrodes also possess the ability to perform cyclic voltammetry
at fast scan rates (200 V/s), thereby increasing time resolution [23]. For some analytes, fast-scan cyclic voltammetry has been shown to reduce fouling due to analyte adsorption [24] and has proven useful for in vivo measurement [23,25].
Although microelectrodes are constructed in a number of ways, two congurations are most common in biosensor production: barrel electrodes and disk
electrodes. The Michael group has immobilized glutamate oxidase on barrel elec-

Microfabrication of Electrode Surfaces for Biosensors

405

trodes and glutamate release was measured in vivo [26]. Unfortunately, due to the
necessity of using HNO3 to remove a coating applied during fabrication of the
carbon ber, the time resolution for this electrode is limited to a scale of minutes.
In addition the spatial resolution of an electrode with a barrel conguration is
large in comparison to a disk. (See Figure 3.) Disk electrodes, fabricated with 10
or 32 m carbon bers, present the possibility of much improved spatial resolution, although it will be seen that derivatization process affects time resolution in
a similar manner to that reported for barrel electrodes. An additional problem

Figure 3 Comparison of barrel and disk microelectrode congurations. The barrel


conguration allows for a much larger surface area for immobilization of enzymes, but this
larger size can be a disadvantage when spatial resolution on a smaller scale is important.
Both types of electrodes are constructed by inserting a thin carbon ber, usually from 4 to
32 m in diameter, into a glass capillary, then pulling the assembly to a sharp point using
a pipet puller, thus making two electrodes at once. The volume between the glass capillary
wall and the ber is then lled with epoxy to make a tight seal, and electrical contact is
made via the back side of the capillary [25]. For a barrel electrode (A), the glass and
epoxy are then removed for some distance from the tip, and the outer surface of the ber is
usually chemically treated to remove any epoxy, etc. This cleaning has been found to hinder electron transfer as compared to an electrochemically pretreated disk electrode. To
make a disk electrode (B), the end of a pulled electrode is rst cleaved with a scalpel, then
polished/beveled on polishing wheel. Following electrochemical pretreatment, for example
pre-anodization/pre-cathodization [8], this surface exhibits facile electron transfer and
much better time resolution than the barrel conguration.

406

Rosenwald and Kuhr

associated with disk microelectrodes is the small area available for the dual role
of enzyme immobilization and electron transfer.
IV.

ENZYME IMMOBILIZATION, STABILITY, AND ACTIVITY

The overall performance of the biosensor is improved, up to a point, if more


enzyme can be immobilized [27]. However, when a fast time response is necessary, it is required that a monolayer (or less) of enzyme be covalently immobilized to the electrode surface. Any effective strategy for construction of an
enzyme-modied electrode surface must balance the need for an adequate amount
of biomolecule with recognition of the effects of surface derivatization. Immobilization strategies have employed various methods for attachment of the biosensing molecules, including entrapment in membranes [28], thin lms [29,30], Langmuir-Blodgett lms [31,32], sol-gels [33,34], and microcavities [35]; by
chemisorption [36], electrodeposition/polymerization [37,38], electrodeposited
thin lms [39,40], photochemical attachment [41], photopolymerization [4244],
and SAM deposition [4549]. Enzymes have been covalently attached directly to
the surface of platinum [50] or carbon [51], and enzymes have been immobilized
within the electrode itself (e.g., carbon paste) [5254]. Linking reagents such as
biotin/avidin [2,55], or antibody/antigen interactions [5658] have been employed, and immobilization has been accomplished through cross-linking to the
amino group of gluteraldehyde [59] or to a conducting polypyrrole surface [60,
61]. Linking reagents have also been employed in conjunction with other surface
modication techniques, such as SAM deposition followed by biotin/avidin linking [62] and the linkage of biotinylated glucose oxidase to avidin adsorbed on a
Langmuir-Blodgett membrane [63].
The simplest method of enzyme immobilization is physisorption. Glucose
oxidase has been incorporated into the micropores (approximately 200 m) of a
micro-platinized platinum electrode by immersion in a solution containing glucose oxidase. These electrodes demonstrated glucose sensitivity of 500 nM and
3 s response times for repeated detection of glucose [18]. One factor common to
all of these methods is that the surface is derivatized in a global fashion; therefore,
the biosensor surface is randomly, and often completely, covered with the biomolecule and/or modied by the attachment chemistry, so diffusion and/or electron transfer is hindered. This severely degrades both the time resolution and the
sensitivity of the amperometric response for biosensors [8,15,64]. For example,
cross-linking with glutaraldehyde is well known to cause loss of as much as 90%
of the enzymes activity, and entrapment in a gel or behind a semipermeable
membrane will affect both the time response and the sensitivity of the measurement signicantly [3].
Biotin/avidin chemistry has been used extensively for the immobilization
of proteins at electrode surfaces [2]. The interaction between one molecule of

Microfabrication of Electrode Surfaces for Biosensors

407

avidin and four molecules of biotin is one of the strongest known noncovalent
interactions between a protein and ligand (Kd = 1015 M) [65]. In association with
the covalently linked tether, the biotin-avidin molecular sandwich attaches the
enzyme close (within 100 ) to the carbon surface [55]. Avidin can be coupled to
a biotinylated enzyme that has the appropriate characteristics (i.e., substrate consumption and product formation) needed for chemical selectivity for sensing the
analyte. This worked extremely well for oxidase enzymes and allowed the production of sensors for peroxide and glucose with response times of less than 200
ms [55,66].
Carbon ber microelectrodes can also be covalently modied with dehydrogenase enzymes (including glutamate, alcohol and glyceraldehyde 6-phosphate dehydrogenase), and these sensors also show subsecond response times
[8,67]. Biotin-conjugated enzymes have been immobilized in an on-line reactor
containing avidin-conjugated agarose, and equilibrium concentrations were
reached within seconds (unpublished results), and experiments with biosensors
fabricated with biotin-conjugated enzymes show good sensitivity and fast
response times [55]. Existing methods for conjugation of enzymes with biotin,
and subsequent tethering of the enzyme utilizing avidin-biotin technology, which
has been used extensively for the immobilization of proteins at electrode surfaces
[2], would therefore be expected to produce fast, sensitive biosensors.

V.

WIRED ENZYME ELECTRODES

Conducting polymer-based immobilization or wired enzymes is a global


enzyme immobilization method that differs in many respects from those just
described. In one example, a redox polymer is formed on the surface by the oxidation of pyrrole molecules to pyrrole radical cations, which then polymerize on
the surface to form conductive polypyrrole [60,68]. Other conducting polymers
include polyvinylpyridine, polythiophene, polyaniline, and polyindole. If
enzymes are present in the solution as polymerization takes place, they are
entrapped within the polymer. When these polymers are cross-linked with redox
mediators such as [Os(bpy)2Cl]+/2+ the resulting amperometric (or potentiometric) biosensors are referred to as wired enzyme electrodes [57]. The distance
between the redox centers of the polymer and the FADH2 centers of the reduced
enzyme is reduced sufciently for electrons to be transferred and, therefore, for
the mediated electro-oxidation of glucose on conventional electrodes. These electrodes do not require diffusing redox mediators or membranes to contain the
enzyme and the redox polymer.
Glucose electrodes made with the redox polymer modied enzyme are stable and sensitive and the glucose response of the electrodes can be less than 1 s
[6,69,70]. In yet another variation of this technique, horseradish peroxidase

408

Rosenwald and Kuhr

(HRP) is co-immobilized with the redox polymer, and as dissolved O2 is reduced


to H2O2 by the HRP, the electrode oxidizes the wired HRP, producing a current
and an analytical signal. Similarly, if an oxido-reductase enzyme (e.g., glucose
oxidase) and HRP are co-immobilized, H2O2 is produced only by the oxidoreductase, HRP reduces the H2O2 and the wired HRP is oxidized by the electrode,
producing a current and an analytical signal that exhibits specicity due to the
combined action of the two enzymes.

VI. ISOLATION OF ATTACHMENT


AND ELECTRON TRANSFER SITES
With the exception of wired enzyme electrodes discussed above, the enzymatically produced cofactor needs an available electron transfer site for the electrochemical reaction to take place to be monitored. As has been discussed, the procedures used for the derivatization of the electrode surface can reduce the rate of
electron transfer for the redox mediator (NADH) [8,15], decreasing the sensitivity of the measurement dramatically. Thus, at least two principal factors inuence
the response of an enzyme-modied electrode: the amount of enzyme present
near the electrode surface and the availability of electron transfer sites. Therefore,
to get a more general protocol for the immobilization of any enzyme to an electrode surface, it would be advantageous to covalently modify only specic parts
of the microelectrode surface, leaving other (directly adjacent) regions available
for facile electron transfer.
One early approach takes advantage of the heterogeneous surface of carbon
electrodes. Selective linking reagents make it possible to derivatize specic functional groups with a linking molecule such as biotin. For example, the formation
of amide bonds is commonly employed to link amine-containing tethers to surface carboxylate groups after activation with 1-ethyl-3-[(dimethylamino)propyl]
carbodi-imide (EDAC). An appropriately modied enzyme is then attached to the
linking reagent [55]. Notably, the separation between enzyme attachment site and
electron transfer site is on the order of hundreds of angstroms, which would allow
for extremely fast time responses on the order of microseconds. Unfortunately,
surface activation with EDAC has since been shown to greatly reduce the rate of
electron transfer [64] and because the nal enzyme immersion step is known to
result in protein adsorption and inhibited electron transfer [15,71], the signal-tonoise ratio (and limit of detection) is adversely affected. This situation illustrates
one of the major obstacles in the fabrication of any analytical measurement system that relies on facile electron transfer for sensitivity. If the nal step in derivatization of the electrode with biomolecules results in poor electron transfer, the
ultimate sensitivity of the measurement will be reduced.
Another approach to separate electron transfer sites and enzyme immobi-

Microfabrication of Electrode Surfaces for Biosensors

409

lization sites is to use microfabrication techniques (as opposed to chemical techniques) to manufacture specic isolated features on the electrode surface. The
Einstein diffusion equation, d = (Dt)1/2, may be used to estimate the optimal size
range necessary for these features. Using a range of typical values for the diffusion coefcient (D) of 1 106 to 2 105 cm2/second and a desired time response
range (t) of 0.01 to 1 s, it can be estimated that the optimal feature size should be
in the range of 1 to 45 m. A number of methods have been employed to form
surface features of micron dimensions.
Conventional lithography techniques have been applied to spatially pattern
a variety of biomolecules, notably antibodies. A recent patent describes a method
to produce a patterned multiple antibody substrate by coating it with a material
that resists antibody adsorption [72]. Ion beam sputtering, laser ablation, or
mechanical scribing is then used to remove the coating at specic sites on the substrate, and then specic antibodies are adsorbed at the sites. Ligler et al. have
demonstrated a method by which surfaces containing thiol groups, epoxy groups,
or vicinal diol groups may be irradiated with UV light, resulting in surfaces that
resist adsorption of biomolecules. This irradiation may be carried out in such a
way as to produce patterned surfaces [73]. Although the effect of this treatment
on electron transfer is not addressed in the patent, chemical modication of the
surface would be expected to reduce electron transfer facility.
Self-assembled monolayers (SAMs), formed by adsorption of -substituted alkanethiols on the surface of gold, can be patterned on the micron scale by
microcontact printing (CP) [7476]. A stamp produced by conventional
lithography techniques is inked with the appropriate alkanethiol, then applied
to the gold surface. This method is convenient for producing and patterning surfaces relevant to biosensors, and costs are low, but uses are limited to surfaces that
allow SAM formation (i.e., gold).
More recently, scanning probe techiques involving the microreagent
mode of scanning electrochemical microscopy (SECM) have been employed. In
one variation, called the positive or direct-write mode, the SECM probe tip is
used as an electrochemical pen depositing linker molecules such as biotin in
micron-sized lines on the carbon substrate as it is scanned across its surface [77
79]. In the negative mode, the SECM probe tip is used as an electrochemical
eraser, cleaning attached molecules off the surface and leaving clean spots on the
surface of a globally derivatized carbon surface [78,80]. These types of micromodications of the surface of an electrode will allow the fabrication of extremely small features in the range of 10 to 40 m that can potentially be tailormade for a variety of applications.
A number of different techniques have focused on the use of photo-activated compounds and lasers for the patterning of biomolecules on electrode surfaces. A procedure has been described for the selective deposition of antibodies
onto a sensor surface using biological self-assembly by photoactivation of a lig-

410

Rosenwald and Kuhr

and previously bound to the sensor surface [81]. So-called confocal patterning,
where a laser of the appropriate wavelength (e.g., ~350 nm for photobiotin) is
focused to a small spot size that is then scanned in the desired pattern to attach
photoactivatable tethering molecules (e.g., photobiotin), has been shown to be
useful for the attachment of tethering sites for biomolecules [82]. Another technique, positive maskless photolithography, uses the interference pattern of a laser
to deposit tethering molecules such as photobiotin [83]. Note that this method is
most effective on microelectrodes with a diameter up to that of the laser beam
spot, so that only one laser shot is required.
Unfortunately, all of the techniques described above require attachment of
the biological molecule to the patterned surface as the nal step in biosensor production, a situation that is analogous to positive maskless lithography. Immersion
of the electrode surface in any protein, including enzymes, is known to result in
adsorption of the protein to the surface, so these methods are ineffective in producing a biosensing electrode with facile electron transfer for any analyte whose
electron transfer characteristics are affected by surface fouling. Another approach
to spatially pattern the biosensor is to rst cover the microelectrode surface completely with the enzyme of interest, then selectively remove it from specic regions
of the surface, which is analogous to negative maskless lithography. Because activation of the electrode surface can be performed with the interference pattern of a
Nd:YAG laser, this approach has the additional benet of eliminating protein fouling in a spatially ordered fashion as the nal step in fabrication of the biosensor
[15]. Negative maskless photolithography by laser interference pattern ablation
(LIPA) has been shown to effectively modify the electrode surface after attachment of biomolecules without loss of biological activity/specicity in the unaffected (negative interference) zones of the laser pattern when appropriate power
levels are used [71]. The enzyme-binding sites are then spatially segregated from
and directly adjacent to active electron transfer sites on the same electrode surface.

VII.

ENZYME LOADING AND STABILIZATION

Increasing the amount of enzyme attached to the surface will increase the amount
of electrochemically detectable cofactor (e.g., NADH, H2O2) produced in the
enzymatic reaction, up to a point. It has been shown for at least one system that
the optimum maximum concentration of enzyme on the surface can clearly be
reached [27]. In this work two enzymes, NADP+-dependent L-malate dehydrogenase and p-hydroxybenzoate hydroxylase, were immobilized on a Clark electrode. The resulting bienzyme sensor behaved as if a single enzyme with a selectivity for L-malate was immobilized, and the optimal surface concentration for
both enzymes was found to be approximately 5 U/cm2. Furthermore, in order to
fabricate a microelectrode with the fast time response necessary to measure neu-

Microfabrication of Electrode Surfaces for Biosensors

411

rotransmitter release on a relevant time scale, monolayer coverage of the electrode surface with the transducing enzyme would clearly be the most desirable situation to achieve fast diffusion and to minimize the impact of enzyme immobilization on electron transfer [2]. Because LIPA removes attached enzyme (while
restoring electron transfer) as the nal step in biosensor fabrication, it is now
desirable to develop methodology to increase the amount of stable enzyme present on the electrode surface prior to application of the Nd:YAG laser interference
pattern. It would then be expected that the amount of active enzyme remaining
after application of LIPA will also be increased.
Building up thicker membranes or lms to contain enzyme should easily
increase the amount of enzyme present on a surface, but it would also be expected
that diffusion and electron transfer will be impeded. Even with restoration of electron transfer by LIPA, the amount of enzyme available for transduction of analyte
is not signicantly increased if millisecond time resolution is required, because
transport of the analyte across the membrane has been previously identied as the
rate-limiting process [84].
A more practical approach involves the buildup of dendrimeric structures
on the surface of the electrode, thus allowing relatively unhindered diffusion to
and from the immobilized enzymes. Yoon and Kim have reported the construction of a surface based on prefabricated dendrimers [85]. Alternating layers of
G4 poly(amidoamine) dendrimers and glucose oxidase were deposited on a gold
electrode (area = 0.033 cm2) by oxidation of glutamate oxidase with IO4 followed by amide bond linkage to the preexisting dendrimeric structure. In this
way, up to ve bilayers of enzyme were deposited on the surface, with a resulting
enhancement in glucose sensitivity as measured by cyclic voltammetry (5 mV/s),
corresponding to the number of bilayers present. Although there is signicant
void space in these dendrimeric structures to allow diffusion of analytes, the ratelimiting process in multilayer amperometric biosensors will still be transport to
and from the enzyme through the membrane [84]. Because no time resolution
data were presented in this study, it is estimated that at 5 mV/s scan rate, and using
a potential window of 0 to 800 mV, the time resolution as measured is greater than
2 min per cyclic voltammogram.
Another approach involves the synthesis of a new compound, biotinylated
Jeffamine. Two schemes for enzyme attachment are presented using either a
dually biotinylated form to stack avidin-conjugated enzymes on the electrode surface, or a mono-biotinylated form conjugated to the enzyme to stack alternating
layers of enzyme and avidin. Although signal enhancement was not seen with
increasing number of enzyme layers, it is possible that this is due to the rate-limiting transport effect reported by Albery [86] on the fast measurements made
(cyclic voltammetry at 100 V/s, 200 ms time resolution). An overall increase in
the successful production of biosensors was viewed as the primary gain by use of
this technique. It is expected that some percentage of the enzyme attached to any

412

Rosenwald and Kuhr

biosensor surface will be inactive, due to effects such as improper orientation and
attachment of linking agent within the active site, and so there is a large increase
in the availability of enzyme attachment sites when stacking techniques are used.
Additionally, with the stability afforded by multi-point covalent linkage, the
amount of enzyme available for transduction is increased [87,88].

VIII.

SUMMARY

Laser patterning can be used in either constructive (e.g., when the laser is used to
create structures on the surface) or destructive (e.g., LIPA, when the laser is used
to remove material from the surface and restore electron transfer) modes to create biosensors with enhanced sensitivity. In either case, only part of the electrode
surface is derivatized with the biological element, leaving a substantial part of the
electrode available for facile electron transfer. The effective concentration of the
biological element in these patterned structures can be increased by using dendrimer-like chemistries. This more than compensates for the lost surface area
available for biomolecules immobilization in the patterned surface. A combination of these approaches may ultimately provide the ideal amperometric biosensor surface.

REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.

Janata, J.; Josowicz, M.; Vanysek, P.; DeVaney, D. M. Anal. Chem. 1998, 70,
R179R208.
Pantano, P.; Kuhr, W. G. Electroanal. 1995, 7, 405416.
Scouten, W. H.; Luong, J. H. T.; Brown, R. S. Trends Biotech. 1995, 13, 178185.
Hall, E. A. H.; Gooding, J. J.; Hall, C. E. Mikrochim. Acta 1995, 121, 119145.
Gregg, B. A.; Heller, A. Anal. Chem. 1990, 62, 258263.
Pishko, M. V.; Katakis, I.; Lindquist, S. E.; Heller, A.; Degani, Y. Mol. Cryst. Liq.
Cryst. 1990, 190, 221249.
Kenausis, G.; Chen, Q.; Heller, A. Anal. Chem. 1997, 69, 10541060.
Pantano, P.; Kuhr, W. G. Anal. Chem. 1993, 65, 24522458.
Brabec, V.; Schindlerova, I. Bioelectrochem. Bioenerg. 1981, 8, 451458.
Downard, A. J.; Roddick, A. D. Electroanal. 1995, 7, 376378.
Davis, J. J.; Halliwell, C. M.; Hill, H. A. O.; Canters, G. W.; vanAmsterdam, M. C.;
Verbeet, M. P. New Journal of Chemistry 1998, 22, 11191123.
Safronov, A. Y.; Kashevskii, A. V.; Serysheva, N. V.; Chernyak, A. S. Elektrokhimiya 1993, 29, 858863.
Guo, B.; Anzai, J.; Osa, T. Chem. Pharm. Bull., 1996, 44, 800803.
Bernabeu, P.; Tamisier, L.; De Cesare, A.; Caprani. A. Electrochim. Acta 1988, 33,
11291136.

Microfabrication of Electrode Surfaces for Biosensors


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.

413

Rosenwald, S. E.; Dontha, N.; Kuhr, W. G. Anal. Chem. 1998, 70, 11331140.
Buchi, F. N.; Bond, A. M. J. Electroanal. Chem. 1991, 314, 191206.
OBrien, J. C.; Shumaker-Parry, J.; Engstrom, R. C. Anal. Chem. 1998, 70, 1307
1311.
Ikariyama, Y.; Yamauchi, S.; Yukiashi, T.; Ushioda, H. Anal. Lett. 1987, 20,
14071416.
Alva, K. S.; Marx, K. A.; Samuelson, L. A.; Kumar, J.; Tripathy, S. K.; Kaplan,
D. L. Proc. SPIE-Int. Soc. Opt. Eng. 1996, 2716, 152163.
Meyerhoff, J. B.; Ewing, M. A.; Ewing, A. G. Electroanal. 1999, 11, 308312.
Pons, S.; Fleischmann, M. Anal. Chem., 1987, 59, A1391A1399.
Howell, J. O. Curr. Sep., 1987; 8, 216.
Michael, D. J.; Joseph, J. D.; Kilpatrick, M. R.; Travis, E. R.; Wightman, R. M. Anal.
Chem. 1999, 71, 39413947.
Baur, J. E.; Kristensen, E. W.; May, L. J.; Wiedemann, D. J.; Wightman, R. M. Anal.
Chem. 1988, 60, 126872.
Hochstetler, S. E.; Puopolo, M.; Gustincich, S.; Raviola, E.; Wightman, R. M. Anal.
Chem. 2000, 72, 489496.
Kulagina, N. V.; Shankar, L.; Michael, A. C. Anal. Chem. 1999, 71, 50935100.
Gajovic, N.; Warsinke, A.; Huang, T.; Schulmeister, T.; Scheller, F. W. Anal. Chem.
1999, 71, 46574662.
Karasavova, M.; Yotova, L.; Krysteva, M.; Shopova, B. Biotechnol. Biotechnol.
Equip. 1998, 12, 110115.
Moser, I.; Schalkhammer, T.; Mann-Buxbaum, E.; Hawa, G.; Rakohl, M.; Urban, G.;
Pittner, F. Sens. Actuators, B 1992, B7, 35662.
Zhang, X.; Shen, J. C. Adv. Mater. (Weinheim, Ger.) 1999, 11, 11391143.
Berzina, T. S.; Alliata, D.; Troitsky, V. I.; Tronin, A. Y. Artif. Nat. Percept., Proc.
Ital. Conf. Sens. Microsyst., 2nd 1997, 8488.
Berzina, T. S.; Piras, L.; Troitsky, V. I. Thin Solid Films 1998, 327, 621626.
Wang, J. Anal. Chim. Acta 1999, 399, 2127.
Wu, J.; Suls, J.; Sansen, W. Anal. Sci. 1999, 15, 10291032.
Walt, D. R.; Taylor, L. In PCT Int. Appl.; 9945357: WO, 1999.
Ryan, M. R.; Lowry, J. P.; Oneill, R. D. Analyst 1997, 122, 14191424.
Curulli, A.; Carelli, I.; Trischitta, O.; Palleschi, G. Biosens. Bioelectron. 1997, 12,
10431055.
Curulli, A.; Carelli, I.; Trischitta, O.; Palleschi, G. Talanta 1997, 44, 16591669.
Bartlett, P. N.; Cooper, J. M. J. Electroanal. Chem. 1993, 362, 112.
Warriner, K.; Higson, S.; Christie, I.; Ashworth, D.; Vadgama, P. Biosens. Bioelectron. 1996, 11, 615623.
Dontha, N.; Nowall, W. B.; Kuhr, W. G. Anal. Chem. 1997, 69, 26192625.
Nakayama, Y.; Zheng, Q.; Nishimura, J.; Matsuda, T. Asaio J. 1995, 41, M418
M421.
Liu, J.-H.; Chung, Y.-C.; Lin, M.-T. Angew. Makromol. Chem. 1994, 219, 4354.
Miyasaka, T.; Koyama, K.; Watanabe, T. Chem. Lett. 1990, 627630.
Mizutani, F.; Sato, Y.; Yabuki, S.; Sawaguchi, T.; Iijima, S. Electrochim. Acta 1999,
44, 38333838.

414
46.
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.
72.
73.
74.
75.

Rosenwald and Kuhr


Hou, S.-F.; Yang, K.-S.; Fang, H.-Q.; Chen, H.-Y. Talanta 1998, 47, 561567.
Gooding, J. J.; Praig, V.; Hall, E. A. H. Anal. Chem. 1998, 70, 23962402.
Dong, S.; Li, J. Bioelectrochem. Bioenerg. 1997, 42, 713.
Creager, S. E.; Olsen, K. G. Anal. Chim. Acta 1995, 307, 277289.
Moody, G. J.; Sanghera, G. S.; Thomas, J. D. R. Analyst (London) 1986, 111, 1235
1239.
Ianniello, R. M.; Yacynych, A. M. Anal. Chim. Acta 1981, 131, 123132.
Bergmann, W.; Rudolph, R.; Spohn, U. Anal. Chim. Acta 1999, 394, 233241.
Wang, J.; Liu, J.; Cepra, G. Anal. Chem. 1997, 69, 31243127.
Gorton, L.; Marko-Varga, G.; Persson, B.; Huan, Z.; Linden, H.; Burestedt, E.;
Ghobadi, S.; Smolander, M.; Sahni, S.; Skotheim, T. Adv. Mol. Cell Biol. 1996, 15B,
421450.
Pantano, P.; Morton, T. H.; Kuhr, W. G. J. Am. Chem. Soc. 1991, 113, 18321833.
Anicet, N.; Bourdillon, C.; Moiroux, J.; Saveant, J.-M. Langmuir 1999, 15, 6527
6533.
Anicet, N.; Bourdillon, C.; Moiroux, J.; Saveant, J.-M. J. Phys. Chem. B 1998, 102,
98449849.
Anicet, N.; Anne, A.; Moiroux, J.; Saveant, J.-M. J. Am. Chem. Soc. 1998, 120,
71157116.
Walt, D. R.; Agayn, V. I. Trac-Trends in Anal. Chem. 1994, 13, 425430.
Schuhmann, W. Mikrochim. Acta 1995, 121, 129.
Schuhmann, W.; Lammert, R.; Uhe, B.; Schmidt, H. L. Sens. Actuators. B 1990, B1,
537541.
He, P.; Ye, J.; Fang, Y.; Anzai, J.; Osa, T. Talanta 1997, 44, 885890.
Lee, S.; Anzai, J.; Osa, T. Sens. Actuators. B 1993, 12, 153158.
Hayes, M. A.; Kuhr, W. G. Analytical Chemistry 1999, 71, 17201727.
Savage, M. D.; Mattson, G.; Desai, S.; Nielander, G. W.; Morgensen, S.; Conklin,
E. J. Avidin-Biotin Chemistry: A Handbook; Pierce Chemical Company: Rockford,
IL, 1992.
Pantano, P.; Kuhr, W. G. In Monitoring Molecules in Neuroscience; Rollema, H.,
Westerink, B., Drijfhout, W. J., Eds.; Krips Repro: Mepples, The Netherlands, 1991.
Pantano, P.; Kuhr, W. G. Proc. Electrochem. Soc. 1992, 921, 829830.
Hammerle, M.; Schuhmann, W.; Schmidt, H. L. Sensors and Actuators B-Chemical
1992, 6, 106112.
Chen, T.; Friedman, K. A.; Lei, I.; Heller, A. Anal. Chem. 2000, 72, 37573763.
Schmidtke, D. W.; Heller, A. Anal. Chem. 1998, 70, 21492155.
Rosenwald, S. E., Nowall, W. B., Dontha, N., and Kuhr, W. G. Submitted April 2000
2000.
Brizzolara, R. A. U. S. Pat. 816337, 1998.
Conrad, D. W.; Davis, A. V.; Golightley, S. K.; Bart, J. C.; Ligler, F. S. Proc. SPIEInt. Soc. Opt. Eng. 1997, 2978, 1221.
Kane, R. S.; Takayama, S.; Ostuni, E.; Ingber, D. E.; Whitesides, G. M. Biomaterials 1999, 20, 23632376.
Marzolin, C.; Terfort, A.; Tien J.; Whitesides, G. M. Thin Solid Films, 1998, 315,
912.

Microfabrication of Electrode Surfaces for Biosensors


76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.

415

Mrksich, M.; Whitesides, G. M. Trends Biotechnol., 1995, 13, 228235.


Wijayawardhana, C. A.; Wittstock, G.; Halsall, H. B.; Heineman, W. R. Anal. Chem.
2000, 72, 333338.
Nowall, W. B.; Wipf, D. O.; Kuhr, W. G. Anal. Chem. 1998, 70, 26012606.
Shiku, H.; Hara, Y.; Takeda, T.; Matsue, T.; Uchida, I. ACS Symp. Ser. 1997, 656,
202209.
Ratcliff, B. B.; Klancke, J. W.; Koppang, M. D.; Engstrom, R. C. Anal. Chem. 1996,
68, 20102014.
Morgan, H.; Pritchard, D. J.; Cooper, J. M. Biosens. Bioelectron. 1995, 10, 841846.
Brooks, S. A.; Ambrose, W. P.; Kuhr, W. G. Anal. Chem. 1999, 71, 25582563.
Dontha, N.; Nowall, W. B.; Kuhr, W. G. J. Pharm. Biomed. Anal. 1999, 19, 8391.
Albery, W. J.; Bartlett, P. N.; McMahon, A. J. J. Electroanal. Chem. Interfacial
Electrochem. 1985, 182, 723.
Yoon, H. C.; Kim, H. S. Analytical Chemistry 2000, 72, 922926.
Albery, W. J.; McMahon, A. J. J. Electroanal. Chem. Interfacial Electrochem. 1985,
182, 16.
Martinek, K.; Klibanov, A. M.; Goldmacher, V. S.; Berezin, I. V. Biochimica et Biophysica Acta 1977, 485, 112.
Martinek, K.; Klibanov, A. M.; Goldmacher, V. S.; Tchernysheva, A. V.; Mozhaev,
V. V.; Berezin, I. V.; Glotov, B. O. Biochimica et Biophysica Acta 1977, 485, 1328.

13
Electrocatalytic Determination
of Biochemical Compounds
James A. Cox and Long Cheng
Miami University, Oxford, Ohio

I.

INTRODUCTION

From a thermodynamic point of view, electrochemical oxidation is predicted to


be a universal method for the detection of biochemical compounds; however, in
practice, only a small fraction of these species are oxidized directly at conventional electrodes in the potential range of common electrolytes. Unfavorable electron-transfer kinetics in combination with passivation of electrode surfaces by
either the test compounds or their oxidation products (and intermediates) places
this limitation on the use of electrochemical oxidation in analytical biochemistry.
An obvious strategy is to employ a catalyst to accelerate the electron-transfer
kinetics. One objective of this chapter is to demonstrate that some surface-modied and composite electrodes not only lower the potential of oxidation of various
classes of biochemical compounds into a range accessible in aqueous solution but
also are sufciently stable for practical analytical methodology.
To achieve the needed stability requires the electrocatalysis step to inuence the pathway of the oxidation, thereby yielding products that do not passivate
the electrode. In addition, the catalyst needs to be either strongly immobilized on
the surface or readily renewed. A well-known regeneration method is pulsed electrochemical detection (PED) [1]. The common example involves a gold electrode.
A series of potential steps is applied to form, sequentially, pristine metal, the
lower oxide, and the higher oxide of gold. In many cases, the lower oxide is catalytically active, and the formation of the higher oxide is accompanied by conversion of surface-active products to forms that desorb. By returning to the initial,
417

418

Cox and Cheng

oxide-free surface in every measurement sequence, reproducible data are


obtained. Applications of PED primarily are to quantify analytes that are separated by high-performance liquid chromatography (HPLC). Carbohydrates and
sulfur-containing biochemical compounds are among the analytes that are
detected by this strategy [1,2]. The formation of conducting composite electrodes
(CCEs) with catalytic centers as dopants also achieves long-term stability by
renewing the surface [3]. Here, the regeneration is accomplished by polishing the
composite. Such electrodes often use the same type of catalyst as developed for
surface-modied electrodes.
Because of their stability against chemical dissolution and their effective
promotion of oxidations of biochemical compounds, catalysts comprising inorganic polymers and other macromolecules that contain platinum-group metal
centers will be the focus of the present report. The applications primarily will be
as amperometric detectors in HPLC and in ow-injection analysis, in which case
both surface-modied electrodes and doped CCEs are suitable. Specically
addressed will be the use of sol-gel processing as a means of immobilizing catalysts in composites.

II.

CATALYSIS BY MIXED-VALENCE RUTHENIUM


OXIDE POLYMERS

The electrochemical reversibility and the facile redox reactions of various species
with high oxidation states of ruthenium make this metal well suited to mediate
electrode processes. Immobilizing Ru(CN)63 in a lm of protonated poly(4vinylpyridine), PVP, on glassy carbon yielded a modied electrode in which the
RuIV formed by electrochemical oxidation of RuIII reacted rapidly with various
solutes, including AsIII [4]. The enhanced current for the electrochemical oxidation of RuIII to RuIV that results from regeneration of RuIII by the chemical step
provided a basis for quantifying AsIII; however, this modied electrode was
unstable because of gradual loss of the ruthenium complex from the PVP lm. For
practical exploitation of the merits of the catalytic activity of the ruthenium centers, it was necessary to immobilize them in a stable manner. An initial attempt
was made to adsorb the ruthenium analogue of Prussian blue onto a glassy carbon
electrode. A mixture of RuCl3 and Ru(CN)62 was used to form ruthenium purple, and glassy carbon was dip-coated with the product. A thick lm with a resistance that is sufcient to distort the shape of the voltammograms was formed.
However, when the potential of a glassy carbon electrode was scanned continuously at 0.1 Vs1 over the range 0.2 to 0.9 V versus Ag/AgCl for several minutes
in this mixture, a highly conductive lm was formed [4]. In acid solution, the stepwise oxidation of RuII to RuIII, RuIV, and (presumably) RuVI was observed. Evi-

Electrocatalytic Determination of Biochemical Compounds

419

dence that a mixed-valence ruthenium oxide structure with cyano cross-links,


mvRuOx, composed the lm was subsequently reported [5].
The oxidation of biochemical compounds is the main application of
mvRuOx. Initial studies dealt with the oxidation of cysteine, glutathione, cystine,
and methionine. Voltammetry was performed at mvRuOx-coated glassy carbon
in aqueous solution. The mediated oxidation of cysteine is demonstrated by Figure 1. That the process is mediated is suggested by the combination of the augmentation of the current for the oxidation of RuIII at around 0.8 V and the diminution of the current for the reduction of RuIV during the reverse scan. An important
observation was that the mvRuOx-modied electrode was not passivated during
these oxidations, which is in contrast to the electrochemical oxidation of thiols in
aqueous solution at bare carbon electrodes. For example, to obtain reproducible
results for amperometric detection of benzenethiol and cyclohexanethiol at bare
carbon after HPLC separation, it is necessary to include an organic solvent such
as methanol in the mobile phase and to renew the surface at a regular interval [6].
However, the mvRuOx-coated electrodes were stable for at least 10 days when
used to determine cysteine and related compounds in the 1100 M range after
separation by HPLC with an aqueous mobile phase [7].
A more detailed study of the long-term stability of a mvRuOx-coated electrode was performed by ow-injection analysis with methionine as the test compound [8]. Unlike thiols, this sulde is not oxidized at a useful rate at bare electrodes in aqueous solution at potentials that do not overlap the oxidation of water.

Figure 1 Cyclic voltammetry of (A) 5.6 mM cysteine and (B) blank electrolyte at a
glassy carbon electrode coated with a lm of mvRuOx. Electrolyte, 0.2 M K2SO4 adjusted
to pH 2.0 with H2SO4; scan rate, 10 mV s1. A hydrodynamic voltammogram of 8.0 M
cysteine had the same features as those in Figure 2 for cystine.

420

Cox and Cheng

Table 1 Analytical Figures-of-Merit for the Determination of Biochemical Compounds by Electrochemical Oxidation at Modied Electrodes Using Flow-Injection
Methodology
Analyte
Methionine
Cystine
Insulin
Heparin
Heparin
Novobiocin
Methionine
Methionine
Cystine
Insulin

Modier
a

mvRuOx
mvRuOx
mvRuOx
mvRuOx
Cu2O
RuO2
Rh2POMb
RuDenc
RuDen
IrOxd

Sensitivity
(nA nM1)

pH

LDR (M)

DL (nM)

Ref.

20
11
31

0.3
1.3
0.85
24
38
35

2.0
2.0
2.0
2.0
13
13.7
7.4
7.0
2.3
7.4

2180
0.23.4
0.22
Nonlinear
1.00.01
6400
5160
1.010
1.010
0.050.5

30

20
140
9
600e
200e
600e
500e
20

8
9
10
16
16
24
33
40
40
14

amvRuOx, mixed-valent ruthenium oxide; bRh POM, phosphotungstic acid with a tungstate
2
replaced by dirhodium acetate; cRuDen, pentaerythritol-based metallodendrimer with RuIIterpyrid
dine units; iridium oxide electroplated from a 0.20 mM Na3IrCl6, 0.1 M KNO3 solution; econducting composite electrodes for which background currents are higher than those at surface-modied
electrodes; LDR, linear dynamic range; DL, detection limit (concentration that gives a signal three
times the standard deviation).

Analytical gures-of-merit are shown in Table 1. Calibration curves obtained


over a 3-week period had only a 5% relative standard deviation of slope. This system served as an amperometric detector for HPLC with various citrate, phosphate
buffers in the pH 2.13.0 range and with 0.1 M monochloroacetic acid in 1% acetonitrile at pH 3.0 as mobile phases.
The electrocatalytic oxidation of disuldes is an important application of
mvRuOx-coated electrodes. As in the case of methionine, the oxidation of cystine
at a bare carbon electrode in aqueous solution generally is not practical for analytical purposes. The hydrodynamic voltammogram, HDV, in Figure 2 illustrates
that cystine is oxidized at this modied electrode. The use of constant potential
makes the background current from charging of the electrical double layer and the
oxidation of the RuIII centers in the lm to the RuIV state negligible. The presence
of a plateau in Figure 2 is unusual for a process that is mediated by an immobilized lm. The expected result is a drop in current at applied potentials well positive of that of the onset of the oxidation. Anodic protection of the oxidized state
of the mediator is the factor that diminishes the mediated current. The reason for
the observed plateau is that the mvRuOx catalyzes the oxidation of water, thereby

Electrocatalytic Determination of Biochemical Compounds

421

Figure 2 Hydrodynamic voltammogram of 8.1 M cystine at a mvRuOx-coated glassy


carbon electrode with 0.2 M K2SO4 at pH 2.0 as the carrier solution. Flow rate, 1.0 mL
min1; injection volume, 7.5 L; electrode area, 0.071 cm2. Each plotted entry is the average and standard deviation of the peak current from ve trials at the given potential. The
current is not corrected for the onset of the catalyzed oxidation of water.

adding to the measured current after about 0.98 V. The calibration curve that was
obtained by ow-injection analysis with an applied potential of 1 V was linear
over the concentration range of 0.410 M cystine [9].
Because the disulde of cystine was oxidized, the study of cystine was
extended to the determination of insulin, which is a protein that contains two peptide chains that are linked by disuldes. The bovine insulin that was used had a
molecular weight of 5733 Da and three disulde bonds. Evidence for the oxidation of insulin at pH 2.0 at a mvRuOx-coated electrode was obtained by owinjection analysis with amperometric detection [10]. Initially, a HDV (Figure 3)
was developed. As described above, HDV provides current-potential information
at concentrations lower than those suited to conventional cyclic voltammetry
because of a higher signal-to-background ratio. Under the conditions in Figure 3
with 0.96 V versus Ag/AgCl as the applied potential, the gures-of-merit in Table
1 were obtained. As shown in Figure 4, the oxidation of 95 nM bovine insulin (7.5
L injection) yielded a signal well above the baseline. The tabulated detection
limited was calculated from results represented in Figure 4.

422

Cox and Cheng

Figure 3 Hydrodynamic voltammogram for the oxidation of 0.58 M insulin with a


mvRuOx-coated glassy carbon electrode. (From J.A. Cox and T.J. Gray, Anal. Chem. 61,
2462, 1989, with permission of the publisher). The conditions are those in Figure 2 except
that the currents are corrected for the component from the oxidation of water.

The electrocatalytic oxidation of insulin at mvRuOx-coated electrodes was


employed in a study by Kennedy et al. [11] to elucidate the mechanism of release
of this protein from pancreatic cells. The lms were coated on electrodes with
dimensions in the 10 m diameter range. Amperometric measurement of insulin
released from single cells by glucose stimulation was successful. The study was
limited by the instability of the mvRuOx lm at pHs above about 5. Apparently,
the oxidized form of mvRuOx slowly dissolves in aqueous solution except at low
pH values [11,12].
A systematic study by Kennedy and coworkers was performed to nd a catalyst that promoted the oxidation of insulin at physiological pH but was more stable than mvRuOx in these solutions [13]. Polynuclear hexacyanometallates,
binary metal oxides, and metallophthalocyanines with various metal centers were
tested as electrode modiers using Ru, Cr, Fe, Co, and Os alone or in combination with Pb, Pd, and Ir. Of the modiers tested, only with lms originating by
oxidation of RuCl3 or (NH4)RuCl6 was oxidation of insulin promoted (an extension of this work, detailed later, described an iridium-based catalyst for the oxidation of insulin [14]). Voltammetric deposition of a ruthenium oxide lm from
a 0.2 mM RuCl3, 10 mM HClO4 mixture onto glassy carbon yielded a modied

Electrocatalytic Determination of Biochemical Compounds

423

Figure 4 Flow-injection analysis of 95 nM bovine insulin (MW 5733) by oxidation at a


mvRuOx-coated electrode. (From J.A. Cox and T.J. Gray, Anal. Chem, 61, 2462, 1989,
with permission of the publisher). The conditions are those in Figure 2 with an applied
potential of 0.96 V vs. Ag/AgCl.

electrode that permitted the amperometric determination of insulin at pH 7.4 at


the sub-micromolar level over a 4-day period (with the electrode stored in 0.1 M
KNO3 when not in use). Films that were formed with 0.1 M KNO3 rather than 10
mM HClO4 as the supporting electrolyte yielded a slower response to insulin. In
ow-injection analysis, the detection limit was 23 nM insulin in a pH 7.4 phosphate buffer.
Ethanol is an important analyte in clinical chemistry and in biotechnology.
Cataldi et al. [15] used mvRuOx-coated glassy carbon as the working electrode
in the amperometric determination of ethanol by ow-injection analysis. A gradual loss of amperometric response to ethanol was observed even in acidic media.
Inclusion of RuCl3 and K4Ru(CN)6 in the carrier solution electrolytically replaced
the lost catalytic sites, thereby providing stability. The instability that was
observed was attributed to the high concentration of analyte in this study (5250
mM).
The oxidation of heparin was investigated with a mvRuOx-coated electrode
in acidic solution and with copper oxide as the catalyst in basic media [16].
Heparin is a member of the glucosaminoglycan family that is widely used as an
anticoagulant during heart surgery. The need for rapid methods of analysis is

424

Cox and Cheng

related to the monitoring of its removal in postsurgical treatment. Three types of


heparin are commonly used, namely full-size heparin (1315 kDa); low-molecular-weight heparin, made by depolymerization by nitrous oxide followed by
reduction with NaBH4; and low-molecular-weight heparin (5 kDa) that is the
product of depolymerization by hydrogen peroxide. The structural characteristic
of heparin is repeating polysaccharide units that contain sulfate sites.
With mvRuOx-coated electrodes and 1 M H3PO4 as the supporting electrolyte, full-size heparin, which is chemically stable under these conditions, was
oxidized at 1.0 V versus Ag/AgCl. The other forms of heparin were not electroactive under these conditions. Voltammetry, which yielded currents consistent
with a two-electron transfer, suggested that the pathway was that found for chemical oxidation of heparin by periodate, which cleaves the C2-C3 bond of the glucuronic acid residue. For studies in basic media, a Cu2O-coated glassy carbon,
which was prepared as reported by Prabhu and Baldwin [17], was used. In basic
media, the heparin undergoes a slow decomposition, but it is not important on the
time scale over which these measurements were performed (hours). All three
types of heparin were electrochemically oxidized at about 0.5 V versus Ag/AgCl,
3 M NaCl in 1 M KOH. Oxidation of saccharide moieties with concomitant
release of sulfate functionalities was demonstrated. Analytical data for the
mvRuOx and Cu2O electrodes are compared in Table 1.
Investigations in various laboratories with biochemical and inorganic analytes are consistent with the efcacy of mvRuOx as a catalyst being related to
mediation of electron transfer. We hypothesized that, at least in the case of oxidation of sulfur-containing compounds, passivation of mvRuOx did not occur
because of a transfer of oxygen in conjunction with the transfer of electrons. This
pathway yields sulnates and/or sulfonates. If these polar compounds are the
products, they will distribute into the aqueous electrolyte. In contrast, with bare
electrodes, the oxidation yields passivating lms derived from thiyl radicals
and/or dimers [18]. To test the hypothesis of oxygen transfer in conjunction with
electron transfer, controlled potential electrolysis of cysteine was performed at a
mvRuOx-coated electrode at pH 2.0 in the presence and absence of dissolved
oxygen [19]. In the absence of dissolved oxygen, the number of electrons, n,
transferred per mole was 4.1 0.3, whereas with oxygen present, n was 4.9 0.1.
The former was consistent with the formation of the sulnic acid (n = 4), and the
latter suggested a mixture of the sulfonic acid (n = 6) and sulnic acid [19].
Additional support for the above pathway was obtained by thin-layer chromatography. Controlled potential electrolysis of methionine was performed at a
mvRuOx-coated electrode at pH 2. The Rf-values (the ratio of the distance moved
by the solute to the distance that the solvent front moves on the solid phase) of the
electrolysis product and of standards were compared. The value for the electrolysis product agreed with that for methionine sulfone to two signicant gures.

Electrocatalytic Determination of Biochemical Compounds

425

The oxygen transfer in conjunction with electron transfer when mvRuOx is


the mediator was conrmed by mass spectrometric identication of products of a
solid-state electrolysis [20]. Here, the test compound, glutathione, and a mediator
were doped into a sol-gel layer in contact with the working electrode. The counter
electrode was isolated from the working electrode by contacting it to second solgel layer. The thickness of each layer was on the order of 10 m. Although the
electrolysis was performed in the absence of a contacting liquid phase, the
approximately 20% (by mass) water in the silica sol-gel provided an aqueous
electrolyte for the process. The electrolyte in the pore water was approximately
0.5 M KCl at pH 2. The pH value was estimated by performing independent solgel experiments with pH indicator dyes included in the gelling mixture. Product
identication was by matrix-assisted laser ionization/desorption time-of-ight
mass spectrometry.
Controlled potential electrolyses in the solid-state, bilayer cell were performed at 1.2 V versus Ag/AgCl for 22 hr. The mass spectrum that was developed
after electrolysis (Figure 5) demonstrated that the glutathione was consumed and
that a major product (m/z 355) was its cysteic acid analogue. No evidence of
dimerization of glutathione was seen. With ferrocyanide as the mediator, electron

Figure 5 Mass spectrum of a glutathione-doped silica sol-gel after a 22 h electrolysis at


1.2 V vs. Ag/AgCl in the presence of mvRuOx. Before electrolysis, a peak related to glutathione appears at 308.1 m/z. The peak at 355.4 m/z is related to the cysteic acid analogue
of glutathione. Other peaks are from the -cyano-4-hydroxy cinnamic acid matrix. (From
J.B. Laughlin, K. Miecznikowski, P.J. Kulesza, and J.A. Cox, Electrochem. Solid-State
Lett. 2, 574, 1999, with permission of the publisher.)

426

Cox and Cheng

transfer occurred without oxygen transfer. Here, the ferrocyanide was incorporated into the gel in contact with the working electrode, and the reaction was initiated by the electrochemical oxidation of FeII to FeIII. The primary product was
dimerized glutathione (Figure 6).
The pathway of oxidations with mvRuOx as the catalyst differs from those
at other common inorganic modiers, metallophthalocyanines and doped lead
oxide. For example, Qi and Baldwin [21] investigated the oxidation of cysteine
and glutathione at a reticulated vitreous carbon electrode that was coated with
cobalt phthalocyanine. The products were identied as disuldes. The formation
of the corresponding sulnic and sulfonic acids was not seen. Johnson and
coworkers have extensively investigated doped lead oxide as electrocatalysts.
Because they function in basic media, they are only indirectly comparable to
mvRuOx. A similarity in function is that characterization of the products demonstrates that oxygen transfer occurs in conjunction with electron transfer. However, the role of the dopant (e.g., bismuth oxide) is that it apparently provides an
adsorption site for reactant species such as cysteine [22]. Because the bismuth
oxide was present in its highest oxidation state, promotion of the oxidation of cysteine by electron-transfer mediation by the dopant was precluded.
Related to mvRuOx as a catalyst is RuO2. Its merit relative to mvRuOx is
stability in basic media. The thermal oxide is commercially available as an industrial-scale catalytic electrode. A convenient means of employing RuO2 in analytical-scale electrodes is to use it as a dopant in a conducting composite. An early

Figure 6 Mass spectrum of glutathione after oxidation in a silica sol-gel with hexacyanoferrate as the mediator. Conditions are reported in Figure 5.

Electrocatalytic Determination of Biochemical Compounds

427

application had carbon paste as the electrode matrix [23]. Selected amino acids
such as methionine, glycine, and phenylalanine, which were not derivatized, were
electroactive at this electrode in 1 M KOH; however, applications to concentrations below the mM level were not reported.
More robust composites are made with conducting solid matrices such as
carbon-doped epoxies or sol-gels. Leech et al. [24] doped a polishable graphiteepoxy composite with RuO2 to yield an electrode at which saccharide-based
antibiotics (e.g., streptomycin and novobiocin) were oxidized. Typical owinjection analysis data (0.5 M NaOH carrier solution) with amperometric detection at 0.4 V vs. Ag/AgCl are summarized in Table 1.
Incorporation of mvRuOx into a polishable composite electrode was
achieved [25]. The mvRuOx was electropolymerized on graphite particles using
a ow-through, packed-bed as the working electrode. The modied graphite was
incorporated into epoxy. The catalytic behavior of this electrode was the same as
that of mvRuOx-coated glassy carbon. Flow-injection analysis of compounds
such as glutathione and isotocin, a small peptide (molecular weight, 966 Da) with
cysteine and disulde sites, was performed at pH 2.0. Detailed evaluation of the
analytical gures-of-merit was not performed; however, the respective sensitivities were determined as 0.28 and 0.15 nAnM1cm2. The lower sensitivity with
isotocin than with glutathione was perhaps related to partial passivation of the
electrode by adsorption, which is a greater concern with the higher mass peptide.
To minimize this problem, 0.01 M hexadecylcetyl trimethylammonium chloride
(HTAC) was included in the carrier solution. The amount of the HTAC was sufcient to form micelles, which can be predicted to increase the solubility of these
compounds and their oxidation products relative to that in simple aqueous solution. Moreover, unlike analogous surface-modied systems, polishing its surface
can restore the catalytic activity of a passivated composite electrode. In this
regard, 19 replicate trials of the determination of glutathione in the absence of
HTAC, with the electrode polished after each, yielded a repeatability of 4%. A
second merit of a composite electrode is the long-term stability. The mvRuOxmodied graphite composites were stable in dry storage for at least one year.
In summary, mvRuOx as a lm on an electrode or in a conducting composite mediates the oxidation of a wide range of biochemical compounds. In acid
solution media, the catalyst is stable and active in ow systems, so it is well suited
to detection in conjunction with HPLC and, presumably, related methods. Among
the compounds that have been determined are sulfur-containing amino acids,
insulin, myoglobin, ribonuclease A, oxytocin, isotocin, vasopressin, and heparin
[79,26]. The variation of the method of formation of the mixed-valence ruthenium oxide and/or the use of RuO2 has promise to extend the range of applicability to neutral and basic solutions. In these solutions, the promotion of the oxidation of carbohydrates and perhaps amines is promising. Further details and

428

Cox and Cheng

descriptions of applications to inorganic analytes can be found in recent reviews


[27,28].

III. TRANSITION METAL-SUBSTITUTED POLYOXOMETALLATES


AS ELECTROCHEMICAL OXIDATION CATALYSTS
The structure of a typical Keggin-type polyoxometallate, POM, is shown in Figure 7. Important features of the chemistry and electrochemistry are detailed in a
recent issue of Chemical Reviews [29]. The voltammetry of POMs is characterized by stepwise reductions of the multi-cation centers. For example, Figure 8
contains a cyclic voltammogram of PMo12O403. The three reduction steps are
described by the following equations [30]:

Figure 7 Structure of phosphomolybdate ion, a Keggin-type polyoxometalate.

Electrocatalytic Determination of Biochemical Compounds

429

Figure 8 Cyclic voltammogram of 5 mM PMo12O403 in a mixed organic-aqueous solvent (1:1 acetonitrile-water by volume) containing 0.5 M H2SO4. Working electrode,
glassy carbon (0.3 cm2); reference electrode, Ag/AgCl; scan rate, 0.1 V s1.

PMo12O403 + 2e + 2H+ p H2PMo12O403

(1)

H2PMo12O403 + 2e + 2H+ p H4PMo12O403 (2)


H4PMo12O403 + 2e + 2H+ p H6PMo12O403 (3)
Each step is an overall two-electron reduction of MoVI centers, during which two
of them are converted to MoV.
In terms of utilization as electrocatalysts, most studies have been related to
the promotion of reductions of inorganic species. A means of employing POMs
as oxidation catalysts is to modify them into transition metal-substituted analogues, TMSPs [31]. Using a reported procedure to modify phosphotungstic acid

430

Cox and Cheng

[32], we removed a tungsten center and its terminally bonded oxygen from the
Keggin structure and replaced them with a center hypothesized to serve as an
electrochemical catalyst, dirhodium acetate [33]. The resulting compound,
Rh2POM, was of interest for two reasons. The molecular size of the TMSP suggested that as a dopant in a sol-gel material, it would be stable against leaching.
The basis of this prediction is that the molecular size of the TMSP is greater than
the width of the interstitial spaces in a microporous silica sol-gel (described
below). As a result, it is trapped in pockets within the matrix and cannot diffuse
through the pores. This model has been established for the immobilization of
enzymes in sol-gel materials [34]. Second, the reversible electron transfer in
POMs suggests that that these compounds will be effective electron-transfer
mediators. In this regard, there is a virtual absence of major bond-length changes
during redox of these compounds, so the electron transfer rates are rapid.
Prior to using dirhodium acetate, Rh2(OAc)4, in a TMSP, it was shown that
this compound is an effective electron-transfer mediator in homogeneous solution
(Figure 9). The increase in the current for oxidation of the RhII-RhII centers to
RhIII-RhII and the attenuation of the corresponding cathodic current when methionine is added to the solution are evidence of mediation. When the Rh2(OAc)4 was
incorporated into a TMSP, the catalytic strength was increased. For example, the
simple dimer only catalyzed the oxidation of AsIII over a limited pH range,
8.110.6, whereas the TMSP was effective over the pH 2.212.1 [33].
In our work, immobilization of POMs and TMSPs on electrodes has been
by incorporating them in microporous sol-gel materials. The most common solgel processing method is initiated by the hydrolysis of metal alkoxides such as
tetramethyl orthosilicate, TMOS. The product condenses into chains with repeating Si-O-Si linkages. Depending on the relative rates of hydrolysis and condensation, which are pH dependent, the gelation (with concomitant water and alcohol loss) leads to solid structures that are dominated by either cross-linked strands
or clusters. The interstitial spaces constitute pores in the ngstrom domain. Other
sol-gel processes lead to different structures; for example, hydrolysis of vanadium oxide leads to a ribbonlike material. An important modication of the solgel processing of TMOS and related compounds is to include surfactants at levels well above their critical micelle concentration in the sol. The gelation occurs
around organized structures (liquid crystals, for example), yielding mesoporous
silica (pore widths in the nm domain). Sol-gels with mixed-valence character
(e.g., VIV,V) are intrinsic conductors, whereas materials such as silica are conductive as long as they contain sufcient pore liquid electrolyte. Details are available elsewhere on general sol-gel chemistry [34], liquid-crystal templating [35],
and silica-based electrochemical platforms [36].
As described above, the size of the TMSPs (15 range) relative to pore
widths of sol-gels (down to a few angstroms for microporous silica) prevents

Electrocatalytic Determination of Biochemical Compounds

431

Figure 9 Cyclic voltammetry at a glassy carbon electrode in (a) 1.8 mM Rh2(OAc)4 and
(b) 1.5 mM Rh2POM solutions in the presence () and absence ( ) of 3.7 mM
methionine. Supporting electrolyte, 0.5 M phosphate buffer at pH 7.0; scan rate, 0.25 V s1.

432

Cox and Cheng

leaching of the catalytic centers into the contacting medium. Yet, access of small
substrates to the catalytic centers is maintained. Regarding the latter, effective
diffusion coefcients in silica are in the range 106108 cm2 s1, depending on
processing conditions [37].
The potential utility of immobilized Rh2POM as an oxidation catalyst was
demonstrated by using it in a sol-gel based composite [33]. The study had two
goals, namely to characterize this CCE and to compare the immobilized Rh2POM
to mvRuOx as an electrocatalyst for the oxidation of biochemical compounds.
The CCE was prepared from a sol of methyltrimethoxysilane that was doped with
Rh2POM and graphite powder. The resulting polishable solid comprised the
working electrode in cyclic voltammetry and in an amperometric detector coupled to a reverse-phase HPLC system.
The cyclic voltammetry of methionine at pH 7.9 with the above-described
CCE was analogous to the results in Figure 9. The data were compared to those
obtained at a CCE that did not include Rh2POM. With a blank CCE (Rh2POM
absent), oxidation of methionine was not observed out to 1.3 V versus Ag/AgCl,
whereas with the doped electrode, a peak was developed at 1.0 V. After the initial scan, the current dropped by 30%, suggesting partial passivation of the electrode. However, the variation was less than 10% for the next several scans as long
as the solution was stirred for a few minutes between trials.
The partial passivation demonstrated the importance of reproducibility of
polishing steps; in this regard, the relative standard deviation of the peak current
for a comparable system, the mediated oxidation of cystine, was 6% when the
electrode was polished after each of the six trials. Flow-injection amperometry at
the micromolar level was not inuenced by this partial passivation. The results
obtained for cystine and for methionine are included in Table 1.
As in the case of mvRuOx, to determine biochemical compounds, electrodes modied with TMSPs need to be used in conjunction with a separation
method such as HPLC to overcome the inherent lack of selectivity. Sol-gel composite electrodes with Rh2POM as a dopant are suited to such application. An
attractive feature of the Rh2POM - doped sol-gel composite relative to various
mvRuOx systems as the working electrode in an amperometric detector for HPLC
is the applicability over a wide pH range. Not only is the former electrode more
stable in neutral and basic media but also the response is quite constant over a
wide pH range. For example, in the above study on the ow-injection amperometry of methionine at the Rh2POM-modied electrode, the peak current varied by
only 4% over the pH range of 2.1 to 7.9 [33]. Hence, it can be predicted that variation of pH of the mobile phase can be used to optimize an HPLC separation without compromising the sensitivity of detection. In combination with data on the
effect of pH on sensitivity (stated below), the compatibility of an amperometric
detector using this composite with HPLC separation at various pHs is shown by
the results in Table 2, and the quantitative results are in Table 3.

Electrocatalytic Determination of Biochemical Compounds

433

Table 2 Separation of Selected Peptides Followed by Amperometric


Detection at a Composite Electrode Doped with Rh2POM
k for:
Eluent pH

Met

Gly-Met-Gly

Met-Met

6.7
4.6
2.3

0.39
0.39
0.85

0.64
0.58
1.8

3.6
1.9
9.2

Met-Phe
27
15
>57

Mobile phase, 0.05 M phosphate in 10% methanol; ow rate, 1.0 mL min1; column, C18; and applied potential, 0.8 V vs. Ag/AgCl. Met, Gly, and Phe
are methionine, glycine, and phenylalanine, respectively, and k is the retention
factor.

Table 3 Quantitative Results for Selected Small Peptides Obtained


by HPLC with Amperometric Detection at a Rh2POM-Doped
Composite Electrode
Analyte
Met
Gly-Met-Gly
Met-Met
Phe-Met
Gly-Met-Gly
Gly-Met-Gly
Met
Gly-Met-Gly

Conditions

Sensitivity

Detection limit

pH 6.7
pH 6.7
pH 6.7
pH 6.7
pH 2.3
pH 4.6
pH 9.5a
pH 9.5

5.0 nA M1
3.4 nA M1
9.2 nA M1
1.4 nA M1
3.5 nA M1
4.2 nA M1
1.9 nA M1
3.5 nA M1

0.2 M
0.1 M
0.2 M
0.8 M
Not determined
Not determined
Not determined
Not determined

aAt pH 9.5 an anion-exchange column was used because C18 silica is unstable, and
acetonitrile replaced methanol; otherwise, the conditions are reported in Table 2.
Met, methionine; Gly; glycine; Phe, phenylalanine.

In summary, Rh2POM and presumably other TMSPs have promise as electrochemical oxidation catalysts. Unlike mvRuOx, they are stable at physiological
pH. Further study is needed to determine if the range of application to biochemical compounds is as extensive as with mvRuOx. The work cited above also
demonstrates the utility of sol-gel processing as a means of immobilizing catalysts; particularly promising is the immobilization of catalysts that have molecular sizes greater than the pore widths of the sol-gel materials.

434

IV.

Cox and Cheng

METALLODENDRIMERS AS ELECTROCHEMICAL
OXIDATION CATALYSTS

Although the immobilization of Rh2POM in microporous silica yielded a stable


electrode, the disparity in molecular size relative to pore width is insufcient to
extend this approach to use of mesoporous sol-gel materials as supports for these
catalysts. Mesoporous structures have an intrinsic advantage over microporous
materials in that effective diffusion coefcients are greater in the former materials [36]. Macromolecular dendrimers have several attractive features for use in
sol-gel materials other than projected stable encapsulation. They may provide a
template around which mesoporous silica can form [37]. Metal centers that are
potential catalysts can be incorporated readily into dendrimeric compounds. Further, the structure of dendrimers assures spatial distribution of metal centers,
which is optimum geometry for most types of catalysis.
Polyamidoamine (PAMAM) systems are perhaps the most studied in the
area of electrochemistry in the presence of dendrimers. The uptake of CuII by
coordination to amine sites followed by reduction to Cu0 yielded clusters that
have potential use for catalysis [38]. The stability of the small clusters suggested
that they reside in the cavities of these starburst molecules. Because PAMAM is
a nonspecic ligand, the use of this method of preparing clusters is applicable to
a wide range of metals. Moreover, the geometry of dendrimers and the variation
of size that is available by controlling the generation number allow the prediction
that they can be immobilized by a variety of methods at electrode surfaces.
Our studies have focused on a pentaerythritol-based metallodendrimer with
RuIIterpyridine units (RuDen), which was synthesized and puried based on a
procedure by Constable et al. [39]. The structure is shown in Figure 10. Mass
spectrometry, NMR, and UV-visible spectrophotometry established the veracity
of the synthesis, and the oxidation state (+2) of the ruthenium centers in the product was veried by EPR [40]. Cyclic voltammetry of RuDen in homogeneous
solution demonstrated the reversible, one-electron oxidation of RuII. The voltammogram of a 1.0 mM RuDen solution in acetonitrile with 0.1 M tetrabutylammonium perchlorate as the supporting electrolyte showed an oxidation peak at 1.1 V
versus Ag/AgCl. With a scan rate of 0.2 V s1, the difference between anodic peak
potential and that for the corresponding reduction peak was 60 mV, and the ratio
of the anodic to cathodic peak current was 1.0.
The RuDen was included as a component of a sol-gel based CCE to test its
catalysis. Graphite powder and RuDen were added to methyltrimethoxysilane in
the formation of the sol [40]. The graphite powder served two purposes. It limits
shrinkage of the material during the gelation step, and the electrical conductivity
is increased. The catalysis of AsIII oxidation was the initial test of the utility of the
RuDen-doped CCE. The sample was 10 mM AsIII in 0.1 M phosphate buffer at
pH 7.0. In the absence of RuDen, AsIII was not oxidized at the composite. In the

Electrocatalytic Determination of Biochemical Compounds

435

Figure 10 Structure of a pentaerythritol-based metallodendrimer with RuIIterpyridine


units.

doped-CCE, the current for the oxidation of RuII to RuIII at 1.1 V versus Ag/AgCl
was increased by 20 A at 0.1 V s1 at a 0.3 cm2 electrode.
The application of a RuDen-doped CCE to the determination of biochemical compounds was investigated by ow-injection amperometry with methionine
and with cystine as analytes. Hydrodynamic voltammogram showed that at 1.10
V versus Ag/AgCl, current responses to methionine at pH 7.0 and to cystine at pH
2.3 (0.10 M phosphate buffer) carrier were achieved (Table 1). The catalysis of
the oxidation of the disulde, cystine, suggested that the oxidation of insulin will
be promoted by RuDen. Cyclic voltammetry at a glassy carbon electrode of a
solution containing 40 M RuDen and 150 M insulin (in 88% [vol.] pH 11
buffer and 12% CH3CN) showed the mediated oxidation of insulin [40]. Extending the study to CCEs and to measurement at physiological pH is the next step in
the test of the utility of RuDen.

V.

RELATED SYSTEMS

The focus of the above studies was the development of stable modied electrodes
for the electrochemical determination of peptides by ow-injection analysis or by
HPLC. Weber and coworkers have described the electrochemical detection of

436

Cox and Cheng

peptides by converting them to copper complexes, which are oxidized at a low


potential [41,42]. The biuret reagent, which forms the complexes, is added in a
post-column reactor. Determinations at the nanometer level were reported.
Somewhat related to mvRuOx as a catalyst is Prussian blue and its analogues. A typical example is CuII-hexacyanoferrate, CuHCF. Zhou and E. Wang
formed deposits of CuHCF on glassy carbon and coated this surface with Naon
to obtain a modied electrode that promoted the oxidation of sulfhydryl compounds over the pH 2.07.0 range [43]. When used in conjunction with HPLC,
detection limits (k = 3 criterion) for cysteine, N-acetylcysteine, and glutathione
were 0.25, 0.66, and 0.6 M, respectively.
An important new catalyst for the electrochemical oxidation of biochemical compounds is an iridium oxide lm that is formed on glassy carbon during
cyclic voltammetry of 0.20 mM IrCl63 in 0.10 M KNO3 [14]. The resulting lm,
an iridium oxide, mediated the oxidation of insulin at pH 7.4 at 0.70 V vs.
Ag/AgCl. Quantitative results obtained by ow-injection amperometry are
included in Table 1.
Other related work includes the use of metallophthalocyanines, doped lead
oxide, and pulsed electrochemical detection. For details, see References 1,2,21,22
and citations therein.

VI.

SUMMARY

Electrochemical oxidation catalysts with ruthenium centers are the basis of several amperometric detectors that permit determination of a variety of biochemical
compounds at the sub-micromolar level in ow systems. Modication of glassy
carbon electrodes with lms of mvRuOx is particularly useful for peptides that
contain sulfur. Oxidation to polar products that are not adsorbed on the electrode
occurs by concurrent oxygen and electron transfer. These electrodes have longterm stability as long as the analytes are at concentrations below about 10 M.
Two problems related to the ruthenium oxide lms are being addressed.
First, the most-studied system, mvRuOx, is not stable during use at physiological pH because of slow dissolution of the oxidized form above pH 4-5. Modifying the deposition chemistry, using alternative means of immobilizing ruthenium on surfaces (e.g., incorporation in dendrimers) and substituting iridium for
ruthenium centers are being explored. Second, when passivation does occur, an
alternative to polishing surface-modied electrodes as a restoration method is
needed. Immobilizing catalysts in polishable conducting composites is being
investigated.

Electrocatalytic Determination of Biochemical Compounds

437

REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.

10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.

D.C. Johnson, W.R. LaCourse, Anal. Chem. 62, 589A597A, 1990.


P.J. Vandeberg, D.C. Johnson, Anal. Chem. 65, 27132718, 1993.
M. Tsionsky, G. Gun, V. Glezer, O. Lev, Anal. Chem. 66, 17471753, 1994.
J.A. Cox, P.J. Kulesza, Anal. Chem. 56, 10211025, 1984.
P.J. Kulesza, J. Electroanal. Chem. 220, 295309, 1987.
J.A. Cox, A. Przyjazny, Anal. Lett. 10, 869885, 1977.
J.A. Cox, E. Dabek-Zlotorzynska, J. Chromatogr. 543, 226232, 1991.
J.A. Cox, E. Dabek-Zlotorzynska, Electroanalysis 3, 239242, 1991.
James A. Cox and T.J. Gray, Stable Modied Electrodes for Use in Amperometric
Detectors in Flow Systems, in Contemporary Electroanalytical Chemistry, A.
Ivaska, A. Lewenstam, and R. Sara, Eds., Plenum Press, New York, 1990.
J.A. Cox, T.J. Gray, Anal. Chem. 61, 24622464, 1989.
R.T. Kennedy, L. Huang, M.A. Atkinson, P. Dush, Anal. Chem. 65, 18821887,
1993.
W. Gorski, J.A. Cox, J. Electroanal. Chem. 389, 123128, 1995.
W. Gorski, C.A. Aspinwall. J.R.T. Lakey, R.T. Kennedy, J. Electroanal. Chem. 425,
191199, 1997.
M. Pikulski, W. Gorski, Anal. Chem. 72, 26962702, 2000.
T.R.I. Cataldi, D. Centonze, E. Desimoni, V. Forastiero, Anal. Chim. Acta 310,
257262, 1995.
K. Lewinski, Y. Hu, C.C. Grifn, J.A. Cox, Electroanalysis 9, 675679, 1997.
S.B. Prabhu, R.P. Baldwin, Anal. Chem. 61, 852856, 1989.
J. Pradac, J. Koryta, J. Electroanal. Chem. 17, 167171, 1968.
J.A. Cox, T.J. Gray, Anal. Chem. 62, 27422744, 1990.
J.B. Laughlin, K. Miecznikowski, P.J. Kulesza, J.A. Cox, Electrochem. Solid-State
Lett. 2, 574576, 1999.
X.H. Qi, R.P. Baldwin, J. Electrochem. Soc. 143, 12831287, 1996.
N. Popovic, J.A. Cox, D.C. Johnson, J. Electroanal. Chem. 455, 153160, 1998.
J. Wang, Y. Lin, Electroanalysis 6, 125129, 1994.
D. Leech, J. Wang, M.R. Smyth, Analyst 115, 14471450, 1990.
J.A. Cox, K. Lewinski, Electroanalysis 6, 976981, 1994.
A. Przyjazny and J.A. Cox, Electroanalysis, 5, 657661, 1993.
J.A. Cox, M.E. Tess, T.E. Cummings, Reviews in Analytical Chemistry, vol. XV, M.
Zangen, ed., Fruend Publishing House, London, England, 1996, p. 173223.
J.A. Cox, R.K. Jaworski, P.J. Kulesza, Electroanalysis 3, 869877, 1991.
The February 5, 1998 issue (volume 98, no. 1) of Chemical Reviews is devoted to
polyoxometallates.
M. Sadakane and E. Steckhan, Chem. Rev., 1998, 98, 219237.
L.C.W. Baker, V.S. Baker, K. Eriks, M.T. Pope, M. Shibata, O.W. Rollins,
J.H. Fang, I.L. Koh, J. Am. Chem. Soc. 88, 23292331, 1966.
X. Wei, M.H. Dickman, M.T. Pope, Inorg. Chem. 36, 130131, 1997.

438
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.

Cox and Cheng


M.E. Tess, J.A. Cox, Electroanalysis 10, 12371240, 1998.
D. Avnir, S. Braun, O. Lev, M. Ottolenghi, Chem. Mater. 6, 16051614, 1994.
C.J. Brinker, G.W. Scherer, Sol-Gel Science: The Physics and Chemistry of Sol-Gel
Processing, Academic Press, NY, 1990.
S.D. Holmstrom, B. Karwowska, J.A. Cox, P.J. Kulesza, J. Electroanal. Chem. 456,
239243, 1998.
S.A. Bagshaw, E. Prouzet, T.J. Pinnavaia, Science 269, 1244421244, 1995.
M. Zhao, L. Sun, R.M. Crooks, J. Am. Chem. Soc. 120, 48774878, 1998.
E.C. Constable, C.E. Housecroft, M. Cattalini, D. Phillips, New J. Chem. 22,
193200, 1998.
S.D. Holmstrom, J.A. Cox, Anal. Chem. 72, 31913195, 2000.
J.-G. Chen, S.G. Weber, Anal. Chem. 67, 35963604, 1995.
J.-G. Chen, M. Logman, S.G. Weber, Electroanalysis 11, 331336, 1999.
J. Zhou, E. Wang, Electroanalysis 6, 2934, 1994.

14
Electrodes Based on the
Electrical Wiring of Enzymes
Charles N. Campbell and Adam Heller
University of Texas, Austin, Texas

Daren J. Caruana
University College London, London, England

David W. Schmidtke
University of Oklahoma, Norman, Oklahoma

I.

INTRODUCTION

The electrical connection of reaction centers of redox enzymes to electrodes,


through their electrical wiring with cross-linked redox polymers, provides
enzyme electrodes having no leachable components. It forms the basis for electrodes that can be used in vivo and in ow systems and that can be miniaturized
to micrometer dimensions. The three parts of this chapter discuss a subcutaneous
glucose anode, an immunosensing cathode, and a DNA-sensing microelectrode.
The glucose electrooxidizing anode is an example of an electrode designed for
use in vivo; the rotating immunosensing electrode exemplies a wired enzyme
electrode operating in a fast-owing solution, and the DNA-sensor is an example
of a wired enzyme microelectrode.

439

440

Campbell et al.

II. MINIATURE SUBCUTANEOUSLY IMPLANTED


GLUCOSE ANODES
Continuous monitoring of glycemia will soon improve the dosage and timing of
insulin administration; warn of impending or actual hypoglycemia; and, through
better glycemic control, reduce the likelihood of diabetes complications. The
wired glucose oxidase electrode of this section is designed to be self-implanted
by the user and replaced twice a week [1]. It is calibrated in vivo by one-point calibration, meaning withdrawal and analysis of a single sample of blood. Such calibration requires independence of the intercept of the concentrations of all potentially electroreducible and electrooxidizable components of serum [2,3].
The counter/reference electrode of the system is an external ECG Ag/AgCl
electrode. The subcutaneously implanted glucose electrooxidizing anode is a 0.29
mm diameter wire. Its structure is shown in Figure 1A [4]. Glucose is electrooxidized in the electrode-contacting layer, composed of wired glucose oxidase
(GOX) and a redox polymer (Equation 1). The Os3+ center of the redox polymer
is reduced by glucose to Os2+, then re-electrooxidized (Equations 2 and 3).
glucose + GOX(FAD) + 2H+ 3 gluconolactone + GOX(FADH2)

(1)

GOX(FADH2) + 2 Os3+ 3 GOX(FAD) + 2Os2+ + 2H+

(2)

2 Os2+ 3 2 Os3+ + 2e

(3)

Miniaturization requires a high current density, which in turn requires rapid


electrooxidation of glucose and efcient collection of the current. The electrode
kinetics is fast when the redox polymer conducts electrons and is highly permeable to both glucose and gluconolactone [58]. Cross-linked poly[(N-vinyl-imidazole) Os(bipyridine)2Cl]+/2+ [9] and poly[(4-vinyl-pyridine)Os (bipyridine)2
Cl]+/2+ [10] are examples of GOX-wiring redox polymers.
Serum contains electrooxidizable ascorbate, urate, and acetaminophen.
These may contribute to the measured current, and thereby prevent one-point calibration of the implanted electrodes. To reduce the rate of the electrooxidation of
these compounds, their ux to the wired enzyme layer is reduced by a special
micromembrane [11]. The same membrane also reduces, to a lesser extent, the
ux of glucose, thereby extending, at least to 30 mM, the linear part of the curve
representing the dependence of the electrooxidation current on glucose concentration [11]. The required micromembranes of 5 104 cm2 area are synthesized
in situ on the wired GOX layers by sequentially adsorbing polyanions and polycations. The micromembrane allows the simultaneous adjustment of the linear

Electrodes Based on the Electrical Wiring of Enzymes

441

Figure 1 (A) Schematic drawing of the three-layer glucose sensor. (B) Comparison of
the estimated glucose concentration of a subcutaneously implanted three-layer sensor with
independently measured blood glucose concentrations before, during, and after an
intraperitoneal infusion of glucose in a rat. (C) Comparison of the estimated glucose concentration of a three-layer sensor subcutaneously implanted in a diabetic chimpanzee with
the capillary blood glucose concentrations measured with an Accu-Chek glucose meter.

part of the dynamic range to overlap that encountered in diabetes (230 mM),
reduction of the drift at 15 mM glucose concentration to less than 5% in 24 hr at
37C, and limitation of the current produced by the ux of all interferants to less
than 5% of the current (for the combination of 0.1 mM ascorbate, 0.2 mM acetaminophen, and 0.5 mM urate). It also makes the electrodes insensitive to variation in the partial pressure of O2 and reduces the transition metal ioncaused loss

442

Campbell et al.

of sensing range, while the sensor maintains a greater than 2 nA/mM sensitivity.
The third, solution-side, layer of the anode is a biocompatible coating formed by
photocross-linking poly(ethylene oxide) (PEO) [12]. The quality of the one-point
in vivo calibration is evaluated by implanting subcutaneously an anode and periodically withdrawing blood samples for independent glucose analysis, e.g., with
a YSI analyzer (Figure 1B & 1C).
In the management of diabetes today, clinical decisions are based on the
blood glucose concentration. The concentration measured by the implanted sensor is, however, that in the subcutaneous interstitial uid. At steady state and in
periods of slow change of the glycemia, the two are well correlated and clinically
identical. This is not the case in periods of rapid change, particularly after insulin
injection. To determine the relationship between the subcutaneous and intravascular electrode-measured glucose concentrations, pairs of electrodes are
implanted in the rat. Simultaneous tracking of the glycemia in the jugular vein and
in the subcutaneous uid of the rat shows that one-point in vivo calibration is feasible even when the blood glucose level rises or falls rapidly after intravenous
glucose or insulin injection as long as a simple correction algorithm is applied
[4,11]. The duration of the time lag between blood and subcutaneous glucose levels following intravenous injection of a glucose bolus is affected by the blood glucose level itself [13].
The transient difference between the blood and the subcutaneous glucose
concentrations after insulin injection is most substantial (Figure 2). Nevertheless,
when this difference is modeled and corrected through an appropriate algorithm,
the blood and subcutaneous uid concentrations are well correlated [14]. In the
simplest correction algorithm, it is assumed that after glucose injection the lag is
of 3 min and after insulin injections it is of 9 min. Even with these simple assumptions, 93% of the subcutaneous and intravenous sensor-measured glucose concentrations are clinically accurate and the remaining 7% are clinically acceptable
in experiments where the glycemia is forced to change very rapidly in the normal
rat [11]. In the brittle diabetic chimpanzee, there is also a signicant difference
between the blood and subcutaneous glucose concentrations when the glucose
concentration declines rapidly following insulin injection (at a rate > 1.8 mg dL1
min1). This difference is, however, not clinically signicant when the concentration changes at a rate smaller than 1 mg dL1 min1 [15].
The stability of the current of glucose electrooxidizing anodes comprising
electrocatalysts based on the electrical wiring of glucose oxidase is usually poorer
in serum than it is in common buffers. In electrocatalysts based on wiring of glucose oxidase, the rate of electrooxidation is controlled by the mobility of the tethered redox polymer segments, which affects the transfer of electrons from the
enzyme to the redox polymer and through the redox polymer to the electrode.
Analysis of the causes of the difference between the current stability in serum and

Electrodes Based on the Electrical Wiring of Enzymes

443

Figure 2 (A) Comparison of the glucose concentration measured by an intravascularly


implanted sensor with the conventionally measured blood glucose concentrations following the intravenous injection of glucose and insulin boli in a rat. (B) Comparison of the glucose concentrations of two sensors, one implanted intravascularly in the jugular vein and
one implanted subcutaneously in the rats back, with the measured glucose concentration
in withdrawn blood samples, following an intravenous insulin injection. The prediction of
a model translating the subcutaneously measured glucose concentration to the blood glucose concentration is also shown.

in buffer shows that among the constituents of serum, transition metal ions and an
electrooxidation product of urate are particularly damaging to the electrocatalytic
process [11]. Reduction of their concentration in the electrocatalytic coatings
through proper design of the mass transport controlling membrane reduces the
decay of the glucose electrooxidation current in serum and provides the basis for
electrodes drifting by less than 5% in their 3-day period of intended use [11].

III.

SEPARATIONLESS AMPEROMETRIC IMMUNOASSAY

Separationless immunoassays in whole blood present a number of challenges.


Hydrogen peroxide, the substrate of the commonly used immunolabeling peroxidases, is eliminated by catalase and catalase-like blood constituents, and proteins
may quickly foul the sensors. Additionally, mass transport both to and within the
sensors denes the time required for separationless immunoassays. The problems
are alleviated in a redox hydrogel-based separationless immunoassay. The redox
polymer used is a copolymer of acrylamide and N-vinylimidazole complexed
with osmium-4,4-dimethyl-2,2-bipyridine and treated with hydrazine to provide

444

Campbell et al.

Figure 3 Structure of the hydrazine-treated copolymer of poly-acrylamide and poly-N(vinylimidazole) complexed with osmium-4,4-dimethyl-2,2,-bipyridine(Cl)

hydrazide functions for cross-linking (PAA-PVI-Os-Hz, Figure 3). The redox


polymer is then mixed with avidin, choline oxidase, and an amine-reactive crosslinker, 400 MW poly (ethylene glycol) diglycidyl ether, and deposited on a 3 mm
diameter glassy carbon electrode. As the mixture dries, the reactive groups on the
cross-linker react with terminal amines on the polymer and proteins to form a
rugged adhering lm on the electrode. The avidin of the lm provides a means for
attaching a biotin-labeled antibody. Although avidin has amine-containing lysine
residues in its biotin-binding regions, these are not accessible to most cross-linking agents [16]. Thus, the cross-linking reaction has little effect on the biotinbinding activity of avidin. Choline oxidase is an 80 kDa enzyme with covalently
bound FAD [17]. The redox-active centers of choline oxidase, unlike those of
most other redox enzymes, are buried deeply in the interior of the protein shell
and thus are not affected by cross-linking with the hydrogel [18].
The electrode is rotated in all experiments at 1000 rpm to enhance the transport of reagents to the electrode. To make the electrode specic to a particular
antigen it is incubated with a biotin-labeled antibody such as the biotin-labeled
F(ab)2 fragment of goat anti-rabbit IgG. This fragment binds to the Fc portion of
the rabbit IgG. Following the attachment of the antibody, the electrode is placed

Electrodes Based on the Electrical Wiring of Enzymes

445

in a solution containing the antigen. After a period of 530 min, a second antibody specic to the rabbit IgG and labeled with horseradish peroxidase (HRP) is
added and allowed to bind with the rabbit IgG (Figure 4).
The reaction cycles producing the electrical signal are shown in Figure 5.
When choline is added to the test cell, the choline oxidase in the hydrogel converts the choline to betaine and reduces dissolved oxygen present to hydrogen
peroxide. As the H2O2 diffuses through the hydrogel it comes in contact with
HRP bound by the immunoreaction. HRP reduces the H2O2 to H2O and is in turn
oxidized. Next, an Os2+ site near the HRP reduces the oxidized HRP returning the
HRP to its active state. The redox polymers Os3+ is then electroreduced [19,20].
When tested in buffer solutions, the hydrogen peroxide reduction current
resulting from the immunoreaction increases about linearly with the logarithm of
the antigen (rabbit IgG) concentration over the 11000 ng/mL range. To assess
the reproducibility of making these electrodes, 21 electrodes made at different
times over a 3-month period were tested at a rabbit IgG concentration of 86
ng/mL. The mean current obtained was 1.10 A/cm2 with a standard deviation of
0.35 A/cm2. The residual current density, following deletion of any of the essential system components, equaled, or was smaller than, that when the antigen was
absent. In the presence of choline, but without any immunoreagent, the current
density was only 0.03 A/cm2, conrming that choline oxidase was not wired
and that the electrooxidation of choline did not interfere with the assay. When the
avidin-containing redox hydrogel was not activated with biotin-labeled antibody,
the current density was about one-fth that obtained in a typical assay, showing
that some antigen did bind nonspecically to the gel. The major cause of noise or
background current was, however, the nonspecic binding of HRP-labeled antibody to the redox polymer coated electrode. This nonspecic binding made the
coating catalytic for H2O2 electroreduction.
The dependence of the signal and background current densities on the concentration of the HRP-labeled antibody is shown in Figure 6. In the absence of
antigen, the current is small. At a 1:1 ratio of HRP-labeled antibody to antigen,
the signal-to-noise ratio is 20:1, and at a 4:1 ratio, the signal-to-noise ratio is 18:1.
Nonspecic interaction of the HRP-labeled antibody fragment with the redox
hydrogel or other electrode components can lead to high background signals,
result in lower signal-to-noise ratios, and a less sensitive assay. These nonspecic
interactions may involve electrostatic attractions or hydrophobic interactions. In
almost all immunoassays, blocking agents such as salts, proteins, or surfactants
are used to partially block nonspecic adsorption. In this system, low concentrations (0.1 wt%) of BSA, combined with 0.16 M NaCl, pH 7.4 phosphate buffer
eliminate nearly all nonspecic adsorption.
Because the electrode is rotated at 1000 rpm and the avidin-biotin complex
has a very high binding constant, the attachment of the biotin-labeled antibody is

446

Campbell et al.

2H2O + O2

Figure 4 Composition of the lm on the immunosensing cathode.

Electrodes Based on the Electrical Wiring of Enzymes

447

Figure 5 The reaction cycles leading to the observed current.

complete in about 5 min. The binding of the antigen rabbit IgG and the labeling
with the HRP-labeled antibody takes much longer. Binding of the HRP-labeled
antibody shows a two-phase behavior. Initially the current response is slow;
however, after 1015 min the current rise accelerates and approaches a rst-order
Langmuir-type binding progress curve.
= 1 e (kf C

bt)

(4)

where is the surface coverage and kf is the rate constant. This type of binding
has been observed in other studies of antibody binding to surfaces [21]. For the
binding of the HRP-labeled antibody, a rate of constant of 1 105 L/(molesec)
was calculated. Binding of the rabbit IgG is somewhat faster with a calculated rate
constant of 2.4 106 L/(molesec).
The cause of the slow initial response is hindered mass transport of the
HRP-labeled antibody into the relatively thick hydrated copolymer lm. A frac-

448

Campbell et al.

Figure 6 Dependence of the current density on the specic binding and on the nonspecic adsorption of the HRP-labeled antibody on its concentration.

tal model that accounts for the geometry of the electrode surface gives a good t
to the data for the entire binding curve. Improvements in the binding rate are realized by decreasing the mass transport effects both to and within the lm by using
microelectrodes and by using thinner, less cross-linked layers on the electrodes.
However, with the thinner layers, more nonspecic attachment is also observed.
The overall time required to produce a readable signal is about 20 min.
Catalase, which decomposes H2O2, is found in blood, apparently because
of the rupture of erythrocytes, and its level increases in some diseases [22]. The
effect of catalase on the assay is seen in Figure 7. The current loss is only 5
10% even when the catalase concentration is 10 units/mL, typical of human blood
[22]. Higher catalase levels (4070 units/mL) typical in blood of people with respiratory distress [22] causes a current decrease of slightly more than 20%. Adding serum causes a signicant decline in sensor response. Additives, such as
Pluronic F-68, that inuence the hydrophilic/hydrophobic interactions between
the sensor and serum constituents are effective in reducing the effects of low
serum concentrations.

Electrodes Based on the Electrical Wiring of Enzymes

449

Figure 7 Dependence of the decrease in the immunoreaction-caused hydrogen peroxide


electroreduction reduction current on the catalase concentration.

IV. AMPEROMETRIC DETECTION


OF OLIGONUCLEOTIDE HYBRIDIZATION
The principle behind DNA chips was born from the basic Southern blot technique, and it is now a fast-moving eld at the cutting edge of genome research.
The two main scientic elements of DNA chips are the manufacture of microarrays with single-stranded oligonucleotides attached to pixels on the surface and
hybridization analysis with spatial resolution.
Measures of the relative advantages of different oligonucleotide sensing
systems include size, sensitivity (the number of copies detected), selectivity (the
ability to sense mutations) and system cost [23]. There is no accepted gure of
merit for the combination of these four measures. In some applications, particularly those relevant to sensing in combinatorial arrays where the samples are
small, the combination of the size of the sensing element, the number of copies
detected and the ability to differentiate between long oligonucleotide sequences
differing in a single base are particularly relevant. The size of the detecting ele-

450

Campbell et al.

ments denes their surface density in the array; the number of copies detected
denes the number of PCR cycles needed to detect a particular sequence, the error
rate increasing with the number of cycles; and the ability to differentiate between
oligonucleotides of increasing length, differing only in a single base, denes the
balance between the ability to locate mutants and the specicity [24,25].
The ability to identify single base changes in a nucleic acid sequence in an
inexpensive, sensitive, rapid, accurate, and high-throughput manner is becoming
an increasingly important goal of both academic and diagnostic laboratories.
Oligonucleotide chip and microsequencing technologies, both using uorescence to detect hybridization, are commonly being applied to high-volume mutation screening. Both detect the hybridization of target nucleic acids (PCR amplied from the sample or patient) to oligonucleotide probes (containing the
sequence of interest) that are immobilized, or synthesized in situ, on a glass slide or
a nylon membrane. The target is generally labeled with a uorescent dye and the
hybridization is measured relative to a wild-type sequence (control) labeled with a
different uorescent dye. Whether the chips are made in house or are purchased, a
uidic hybridization station, a uorescence scanning device, a liquid handling
robot, and resources for high-throughput PCR are also required. Running costs of
the uorescence-based hybridization technologies are quite high, as the uorescent dyes are expensive and large numbers of PCR cycles need to be performed.
The sensitivity of the uorescence-based technology is also a concern. PCR
amplication of the target sequence is required prior to hybridization. Amplication bias of some sequences over others and the many problems inherent in PCR
(sensitivity of reaction to contaminants, nonspecic amplication, misincorporation of nucleotides, etc.) may affect the results of the mutation screening. There
are also problems with the use of uorescent dyes. Background uorescence from
the slides, bleaching, quenching of the uorescent signal, and high signal-tonoise ratio are technical problems that are far from trivial. These problems are
currently being addressed by measuring the uorescent signal from the sample
relative to the uorescent signal from a control, but this is not an ideal situation,
because the appropriate control is not always available in a diagnostic setting.
A technology that addresses the above concerns, especially by increasing
the sensitivity and thus obviating the need for PCR, would be of enormous benet in terms of cutting running costs, eliminating potential false positives (misincorporation of nucleotides and nonspecic amplication, etc.) and reducing the
time required to do the screening. Electrochemical techniques are well suited for
detection and measurement of surface processes such as hybridization processes
in an array device as described here. The strength of this approach is twofold:
rst, the method does not require individually addressable microdots; second, it
incorporates an amplifying enzyme that improves the sensitivity/cost ratio relative to that of uorescence techniques.

Electrodes Based on the Electrical Wiring of Enzymes

451

The scheme for detecting the hybridization of the enzyme-labeled target to


the probe attached to an electron-conducting hydrogel is shown in Figure 8. A
redox polymer lm is electrophoretically deposited on the electrode, then reacted
in a second electrophoretic deposition step with the 5-activated-poly-T spacer of
the probe. Upon hybridization with the SBP-labeled target, electrons ow from
the electrode through the redox polymer to heme-centers of the label, reducing
these. Soybean peroxidase (SBP) is preferred over the more active horseradish
peroxidase for labeling the target because of its superior thermal stability [26].
Because electrophoretic deposition is restricted to the conductive area of a
glass-embedded carbon ber, redox polymer lms of reproducible dimensions
and characteristics are deposited when the electrode and solution were also reproducible. This is important because if the microelectrodes are to be used in genedetecting arrays, then differences between their currents will be measured and the
signicance of small differences in current will depend on the reproducibility of
the coatings. The dimensions of the electrodes are likely to be reproduced when
made by the processes used in the manufacture of microelectronic circuits, within
0.05 m. In 25 successively deposited lms, the voltammetric peaks showed a
normal distribution, with the standard deviation, , being 5%. Typically the
faradaic charge was 5.6 1011 C, corresponding to 5.8 1016 moles of osmium
complex on the electrode. Because of the fast electron exchange and because the
redox polymer is well adsorbed on the electrode, the peaks of the anodic and
cathodic waves of the voltammogram of the redox-polymer coated microelectrodes are separated only by about 20 mV. The peak separation increases to
180mV after the probe oligonucleotide is bound to the redox polymer, indicative
of much slower electron transport [27,28]. The apparent cause of the sluggish
transport is the formation of ion bridges between the polyanionic probe and the
polycationic redox polymer, which restricts the segmental mobility of the redox
polymer and thereby the frequency of electron transferring collisions.
The purpose of precoating the microelectrodes with a thin layer of redox
polymer is to make the electrical contact between the reaction centers of SBP and
the electrode independent of the orientation of the target-labeling SBP. In absence
of a redox polymer lm, electrical contact is established only with those SBP
heme centers that are near the electrode surface. In the case of horseradish peroxidase on vitreous carbon, a small fraction of the HRP molecules is always electrically connected to the carbon surface, because some of the molecules are so oriented that their heme redox centers are close enough to the surface of the
electrode for electrons to be transferred. This fraction is usually approximately
1% [29]. In contrast, when the peroxidase is embedded in an electron-conducting
redox polymer, electrons are transferred from the redox polymer to all the heme
centers of the enzyme irrespective of their orientation through collisions of hemecenters with randomly moving segments of the cross-linked redox polymer lm.

452

Campbell et al.

B
Figure 8 (A) Schematic diagram of the DNA detection system. The 18-base probe
oligonucleotide is covalently bound to the electron-conducting redox polymer on the
microelectrode through a 12-base poly-T spacer arm. The SBP-label is covalently bound to
the target oligonucleotide through a 12-carbon spacer arm. Upon hybridization, electrical
contact is established between the SBP-heme centers and the electrode via the redox polymer. This contact enables the electrocatalytic reduction of H2O2 to water through the steps
shown in B. (B) The electron-transferring steps underlying the enzyme-amplied signaling
of hybridization.

Electrodes Based on the Electrical Wiring of Enzymes

453

As a result, the current density is approximately 100-fold higher. The mobility of


these tethered segments increases when the redox polymer is hydrated. Figure 8
shows the steps of electron-transport between the electrodes and the SBP labels
of the targets, resulting in the catalysis of the electroreduction of H2O2 to water
(Equation 5) at 0.06V (Ag/AgCl). At this potential, H2O2 is not electroreduced
on carbon in the absence of the catalytic steps.
H2O2 + 2 e + 2 H+ 3 2 H2O

(5)

Figure 9 shows that overloading of the redox polymer lm with the probe
oligonucleotide reduces the current after the SBP-labeled hybrid is formed. The
current increases initially (Figure 9, curves a, b, and c) as the amount of probe is
built up and more of the SBP-labeled target is captured. However, when the
amount of probe is excessive, the current decreases and the rate of hybridization,
evidenced by the rate of change in current, also slows (Figure 9, curve d). The
apparent cause of the reduction in current is again the restriction of the movement

Figure 9 The increase in the catalytic hydrogen peroxide electroreduction current upon
adding the soybean peroxidaselabeled target in microelectrodes loaded with different
amounts of the probe oligonucleotide. The probe was deposited for (a) 1 minute, (b) 2.5
minutes, (c) 5 minutes, and (d) 10 minutes.

454

Campbell et al.

of the segments of the redox polymer upon excessive ion-bridging, compounded


by dilution of the density of electron-exchanging redox centers.
The optimal duration of the deposition process is about 5 min. The time
dependence of the current, which represents the rate of hybridization, is well
described (R2 = 0.99) by the diffusion-limited Langmuir equation (Equation 6)
[30]
1/2

e = emax[1 exp(kDt )]

(6)

where kD is the surface binding rate (here the rate of hybridization of the SBPlabeled target to the electrode-bound probe), related to the rate of diffusional mass
transport, and emax is the maximal surface concentration of enzyme, reached after
hybridization of all possible probes that can hybridize. For a thin redox polymer
lm and neglecting diffusion of the substrate (H2O2 in the present case), the saturation current is given by Equation 7.
Icat = 2k[Os2+][S]FAe

(7)

In Equation 7, k is the rate of the reaction between the mediator and enzyme,
[Os2+] is the concentration of the reduced redox centers in the lm, and [S] is the
substrate (H2O2) concentration. Because the substrate concentration is high, the
current is limited either by electron diffusion in the lm or by the enzymes
turnover rate. Combining Equation 6 with the simplied Equation 7 gives
1/2

Icat = b[1 exp(kDt )]

(8)

where b = 2k[Os2+][S]FAemax. This equation ts the curves of Figures 9 and 10.


The best-t parameters for the hybridization transients of Figure 9 are summarized in Table 1. The kD values for lms a, b, and c, which are not overloaded

Table 1 Best-Fit Parameters to Equation 8 for Microelectrodes


with Different Probe Loadingsa
Oligonucleotide
deposition time (min)
1
2.5
5
10

b/pA

kD/sec1

R2

2.26
3.86
5.79
4.6

0.089
0.093
0.095
0.077

0.99
0.99
0.99
0.99

a
The calculated and measured currents (Figure 9) t Equation 8 with a
mean difference of <2% in all the experiments.

Electrodes Based on the Electrical Wiring of Enzymes

455

with the probe, are similar, but kD of the overloaded lm is signicantly lower,
suggesting that the hybridization-causing diffusive step is restricted in a matrix
with an excessive amount of ion-bridgeforming oligonucleotide. The time
dependence of the current also ts the diffusion-limited Langmuir equation when
the hybridization is carried out at different temperatures and with mismatched
bases in the oligonucleotides. However, the noise increases with the temperature
and R2 is reduced.
Figure 10 shows the increase in the current with temperature for a perfectly
matched hybrid (25C, 6pA; 45C, 28pA; and 57C, 45pA). This data yields an
activation energy of 60 kJ mol1, similar to activation energy reported for a
related redox polymer [19]. The similarity in activation energies of electron transfer through the redox polymer and the measured current suggests that the current
is limited by the transport of electrons through the redox polymer.
Because the SBP-label contacts electrically the redox polymer only below
the melting temperature of the hybrid, the activation energies for the currents at
25C, 45C, and 57C differ signicantly when the hybrids are perfectly matched,
have a single base-pair mismatch, or contain four mismatched base pairs. The theoretically estimated melting temperature of the 18-base hybrid with a single mismatched base pair is approximately 57C below that of the perfect hybrid when
the mismatch is in the middle of the oligonucleotide and the mismatched base pair
is GC [31]. For the four base-pair mismatch, the theoretically estimated melting
temperature is 2023C below that of the perfectly matched hybrid. The actual
melting points of the 18-base pair hybrids when perfectly matched, mismatched
in a single base pair, and mismatched in four base pairs are 59.5C, 54C, and
3740C, respectively. Consequently, a current should ow in the case of the perfectly matched hybrid at any of the three temperatures, 25C, 45C, or 57C. In
the case of the hybrid with a single mismatched base pair, a current should ow
at 25C and at 45C, but not at 57C; and in the case of the hybrid with four mismatched base pairs, a current should ow only at 25C, not at 45C or at 57C.
That this is indeed the case is seen in Figure 10. For example, at 57C the current
for the perfectly matched hybrid was 45 pA, whereas the current for the hybrid
with a single mismatch was 11 pA. Figure 11 shows the result from an experiment
carried out at 45C, where rst the SBP-labeled target hybridizing below
3740C with four mismatched bases was added, followed by the complementary
SBP-labeled target, hybridizing at temperatures up to 59.5C. This experiment
shows that the presence of an extraneous oligonucleotide with a partially matching sequence does not interfere with the hybridization of the matched target, nor
does it affect the magnitude of the current (about 30 pA) reached upon hybridization (Figures 9b and 11).
The number of copies producing the current was estimated from the
turnover rate of the SBP label to be about 34,000. The rate of turnover of the SBP
label is 460 s1 at 25C. With two electrons being transferred per turnover, this

456

Campbell et al.

Figure 10 Increase in the catalytic current of microelectrodes at 25C, 45C, and 57C
after adding the SBP-labeled fully complementary or partially mismatched targets. (A) perfectly matched target; (B) target with a single mismatched base; (C) target with four mismatched bases. The dashed lines represent the best t of the data to Equation 8.

Electrodes Based on the Electrical Wiring of Enzymes

457

Figure 11 Current-time plot of the catalytic current of a microelectrode coated with the
probe-bearing redox polymer. An aliquot of the SBP-labeled target with four mismatched
bases was introduced at A, then an aliquot of the SBP-labeled perfectly matching target at B.

turnover rate corresponds to a current of 1.5 1016 A per label. At 25C the saturating current measured upon complete hybridization is 5 pA, which is the output of 34,000 wired and active labels. For the 7 m diameter electrode, the corresponding surface coverage is 1.4 1013 moles cm2, well in the theoretically
calculated surface density range of 0.03 to 3.8 1013 moles cm2 for a probe with
18 base pairs on a solid surface [32].
In summary, a single-base mismatch in an 18-base oligonucleotide can be
amperometrically sensed with, and amplied by, a redox polymercoated microelectrode. The detected current is generated by about 34,000 active and wired
copies of the thermostable SBP-labeled hybrid. In the future, a complete gene
might be hybridized to the 18-base peptide or deoxyribose oligonucleotide
[33,34] bound to the redox polymer, then the presence of the hybridized and
thereby redox polymer bound gene would be queried with an SBP-labeled 18base sequence hybridized to a different region of the gene.

458

Campbell et al.

REFERENCES
1.
2.
3.
4.
5.

6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.

17.
18.
19.
20.
21.
22.
23.

24.

Heller, A. Annual Reviews in Biomedical Engineering 1999, 1, 153175.


Csoregi, E.; Quinn, C. P.; Schmidtke, D. W.; Lindquist, S.; Pishko, M. V.; Ye, L.;
Katakis, I.; Hubbell, J. A.; Heller, A. Analytical Chemistry 1994, 66, 31313138.
Csoregi, E.; Schmidtke, D. W.; Heller, A. Analytical Chemistry 1995, 67,
12401244.
Schmidtke, D. W.; Heller, A. Analytical Chemistry 1998, 70, 21492155.
Rajagopalan, R.; Heller, A. Electrical Wiring of Glucose Oxidase in Electron Conducting Hydrogels; Jortner, J., and Ratner, M., Ed.; Blackwell: Oxford, 1997, pp
241255.
Rajagopalan, R.; Aoki, A.; Heller, A. Journal of Physical Chemistry 1996, 100,
37193727.
Aoki, A.; Rajagopalan, R.; Heller, A. Journal of Physical Chemistry 1995, 99,
51025110.
Aoki, A.; Heller, A. Journal of Physical Chemistry 1993, 97, 1101411019.
Ohara, T. J.; Rajagopalan, R.; Heller, A. Analytical Chemistry 1993, 65, 35123517.
Gregg, B.; Heller, A. Analytical Chemistry 1990, 62, 258263.
Chen, T.; Friedman, K. A.; Lei, I.; Heller, A. Analytical Chemistry 2000, 72,
37573763.
Quinn, C. P.; Pathak, C. P.; Heller, A. Biomaterials 1995, 16, 389396.
Quinn, C. P.; Pishko, M. V.; Schmidtke, D. W.; Ishikawa, M.; Wagner, J. G.; Raskin,
P.; Hubbell, J. A.; Heller, A. American, Journal of Physiology 1995, 269, E155161.
Schmidtke, D. W.; Freeland, A. C.; Heller, A.; Bonnecaze, R. T. Proceedings of the
National Academy of Sciences 1998, 95, 294299.
Wagner, J. G.; Schmidtke, D. W.; Quinn, C. P.; Fleming, T. F.; Bernacky, B.; Heller,
A. Proceedings of the National Academy of Sciences 1998, 95, 63796382.
Savage, M. D.; Mattson, G.; Desai, S.; Nielander, G. W.; Morgensen, S.; Conklin, E.
J. Avidin-Biotin Chemistry: A Handbook; 2nd ed.; Pierce Chemical Company: Rockford, Illinois, 1994.
Ohta-Fukuyama, M.; Miyake, Y.; Emi, S.; Yamano, T. Journal of Biochemistry
1980, 88, 197203.
Ohara, T. J.; Vreeke, M. S.; Battaglini, F.; Heller, A. Electroanalysis 1993, 5,
825831.
Gregg, B.; Heller, A. Journal of Physical Chemistry 1991, 95, 59705975.
Gregg, B.; Heller, A. Journal of Physical Chemistry 1991, 95, 59765980.
Bourdillon, C.; Demaille, C.; Moiroux, J.; Saveant, J.-M. Journal of the American
Chemical Society 1999, 121, 24012408.
Leff, J. A.; Parsons, P. E.; Day, C. E.; Moore, E. E.; Moore, F. A.; Oppegard, M. A.;
Repine, J. E. American Review of Respiratory Disease 1992, 146, 985989.
Winzler, E. A.; Richards, D. R.; Conway, A. R.; Goldstein, A. L.; Kalman, S.;
McCullough, M. J.; McCusker, J. H.; Stevens, D. A.; Wodicka, L.; Lockhart, D. J.;
Davis, R. W. Science 1998, 281, 1194.
Pease, A. C.; Solas, D.; Sullivan, E. J.; Crownin, M. T.; Holmes, C. P.; Fodor, S. P.
A. Proceedings of the National Academy of Science, USA 1994, 91, 5022.

Electrodes Based on the Electrical Wiring of Enzymes


25.

26.
27.
28.
29.
30.
31.
32.
33.
34.

459

Yershov, G.; Barsky, V.; Belgovsky, A.; Kirillov, E.; Kreindlin, E.; Ivanov, I.; Parinov, S.; Guschin, D.; Drobishev, A.; Dubiley, S.; Mirzabekov, A. Proceedings of the
National Academy of Science, USA 1996, 93, 4913.
Vreeke, M. S.; Yong, K. T.; Heller, A. Analytical Chemistry 1995, 67, 42474249.
Chidsey, C. E. D. Science 1991, 251, 919.
Laviron, E. Journal of Electroanalytical Chemistry 1979, 101, 1928.
Gorton, L.; Jonsson-Patterson, G.; Csoregi, E.; Johansson, K.; Dominguez, E.;
Marko-Varga, G. Analyst 1992, 117, 12351241.
Peterlinz, K. A.; Georgiadis, R. Langmuir 1996, 12, 47314740.
Anderson, M. L. M.; Hames, B. D. and Higgins, S. J., Ed.; Oxford University Press,
New York, 1995, pp 129.
Chan, V.; Graves, D.; McKenzie, S. Biophysical Journal 1995, 69, 22432255.
Wang, J.; Palecek, E.; Nielson, P. E.; Rivas, G.; Cai, X.; Shiraishi, H.; Dontha, N.;
Luo, D.; Farias, P. A. M. Journal of the American Chemical Society 1996, 118, 7667.
Wang, J.; Cai, X.; Fernandes, J. R.; Grant, D. H.; Ozsoz, M. J. Journal of Electroanalytical Chemistry 1998, 441, 167.

15
Capillary Electrophoresis/
Electrochemistry
Instrument Design and Bioanalytical Applications
Susan M. Lunte, R. Scott Martin, and Craig E. Lunte
University of Kansas, Lawrence, Kansas

I.

INTRODUCTION

Over the past two decades, capillary electrophoresis (CE) has grown from a laboratory curiosity to an important method for the analysis of small volume samples
[1]. It has shown its greatest promise for DNA analysis, chiral separations, and
the analysis of single cells and microdialysis samples [2,3]. More recently, capillary electrophoresis in the microchip format has become popular due to its ease
of construction and ability to perform parallel processing [4].
CE as a separation method has several advantages for the analysis of biological samples. Injection volumes are very small (nanoliter to picoliter), allowing the analysis of limited volume samples. Due to the plug ow associated with
CE, it is possible to obtain rapid, highly efcient separations by applying a high
eld across the fused silica capillary. Separation selectivity can be adjusted by
changes in pH or through the use of modiers. The pH range is larger than that
which can be used with most silica-based chromatography columns. Due to the
small volume of run buffer needed for the separation, exotic additives that would
be too expensive for LC-based separations can be employed.
Several types of detectors that are suitable for use with capillary electrophoresis have been reported [5]. Absorbance detection is most popular because
conventional spectrometric cells can be modied for the fused silica capillary.
Detection can be accomplished simply by removing some of the polyimide coat461

462

Lunte et al.

ing from the outside of the capillary to produce the detection cell. For high-sensitivity analyses, laser-induced uorescence detection is the most commonly used
method. However, although very low limits of detection can be obtained, this
method almost always requires derivatization of the analyte. Capillary electrophoresis has also been interfaced with mass spectrometric detection. In general, the concentration limits of detection are much higher than those obtained by
LC-MS, due to the lower mass loading in CE [2].
Electrochemical (EC) detection is particularly well suited to capillary electrophoresis. In contrast to many optical methods, EC detection can be miniaturized without a loss of sensitivity. It is very selective because there are few compounds in biological samples that are electroactive. The selectivity is tunable and
dependent on the electrode material and the applied potential. Modied electrodes and multiple electrode systems can be used to increase selectivity for certain classes of analytes. Electrodes and potentiostats can also be integrated into
chip-based CE systems using procedures similar to those employed for the manufacture of the analytical chips.
In this chapter, we will review the fundamental parameters involved in the
development of electrochemical detection for CE and describe the application of
CEEC to the analysis of microdialysates and biological uids. The focus will be
primarily on the results obtained in our laboratories. Several excellent reviews are
suggested for individuals desiring further information about applications of
amperometric, conductometric, and potentiometric detection in CE [611].
II.

THE BASIC CEEC SYSTEM

Two typical congurations for CEEC are shown in Figure 1. In both cases, the
separation system consists of a buffer-lled fused silica capillary (typically 5100
m inner diameter [i.d.]) placed between two buffer reservoirs containing platinum wires. A high-voltage power supply is used to apply a potential eld across
the capillary. Optical detection can be performed directly on-capillary by removing some of the polyimide coating from the fused silica capillary to create a window. Electrochemical detection can be accomplished either off-column or endcolumn; these detection modes are discussed in detail in the next section. To
inject a sample, the anodic buffer reservoir is replaced with a sample vial, and
either pressure or a voltage is applied for a predetermined amount of time. Injection volumes are generally in the low nanoliter range. The anodic buffer reservoir
is then replaced, and a high voltage is applied across the capillary. Analytes
migrate in the capillary based on the combination of electrophoretic and electroosmotic forces. At pH values greater than 4, a substantial amount of electroosmotic ow results. In most cases, this ow is greater than the electrophoretic
mobility of the analyte and causes nearly all analytes, regardless of charge, to

Capillary Electrophoresis/Electrochemistry

463

Figure 1 CEEC systems with (A) off-column detection and (B) end-column detection.

migrate toward the cathode. Compounds are separated based on their mass-tocharge ratios. The order of migration is positive ions rst, then neutrals and,
nally, negative ions. Neutral analytes cannot be separated by conventional capillary electrophoresis; however, they can be separated by micellar electrokinetic
chromatography (MEKC). In that case, a charged surfactant is added to the run
buffer above its critical micelle concentration. This causes separation of the neutral analytes due to differing degrees of partition into the charged micelle.
One advantage of CE is that both the separation speed and efciency are a
function of the applied potential. This makes it possible to obtain extremely fast,
high-efciency separations using short capillaries and high eld strengths. To
obtain the best results for electrochemical detection in CE, the separation voltage
should be isolated from the electrochemical detector. This can be accomplished
using either off-column or end-column detection.
A.

Off-Column Detection

The very rst application of electrochemical detection in CE employed off-column detection [12] (Figure 1A). In this method, the electrochemical detector is
isolated from the separation voltage by a decoupler, which typically consists of a

464

Lunte et al.

small fracture in the capillary between the separation capillary and the detection
electrode. The fracture provides a current path to ground through the cathodic
buffer reservoir. The electroosmotic ow produced in the separation capillary
forces the analyte over the joint to the detection cell. If the length of the detection
capillary is short (12 mm) and the pH of the run buffer is greater than 4, the efciency of the separation is maintained.
The simplest decoupler for off-column detection is the bare fracture joint,
which is produced by scoring the capillary and threading it into a piece of exible tubing such as polyetheretherketone (PEEK). Prior to insertion of the capillary, a hole is made in the tubing to provide a window for the bare fracture. After
the scored portion of the capillary is aligned with the hole in the tubing, it is xed
in place with epoxy, and gentle pressure is applied to create a crack. The decoupler is then placed in the cathodic buffer reservoir, which holds the ground electrode. This method works best with CE separations that generate enough electroosmotic ow so that very little analyte is lost through the fracture into the
cathodic buffer reservoir as it travels to the electrochemical cell.
If the electroosmotic ow is low or leakage from the fracture is a problem,
Naon or another type of ion-permeable tubing can be placed over the scored part
of the capillary prior to fracturing [13]. For best alignment, the capillary is rst
mounted on a support. The resulting crack, which is encased in ion-permeable
tubing, produces a path to ground without directly exposing the fracture to the
free solution. These joints can also be used as postcolumn mixers to change the
pH or ionic strength of analytes exiting the fused silica capillary [14].
If sub-nanomolar limits of detection are required, the use of a longer Naon
tube (12 mm) that is the same diameter as the fused silica capillary can be produced by building the tube around a wire [15]. Figure 2 shows the detection of a
mixture of catecholamines at 10 nM using this approach. This long ion-permeable
tube leads to more effective grounding, less noise at the detector, and lower limits of detection than the bare fracture. Detection limits of 500 pM for hydroquinone were reported using a phosphate buffer. The limits of detection for catecholamines were approximately 5 nM. Alternatively, the decoupler can be
constructed from the fused silica capillary itself. This is accomplished by using
hydrouoric acid to etch the wall so that it becomes very thin. If the wall is made
thin enough, it becomes ion-permeable, and the capillary can be grounded at that
point [16,17].
Several other types of decouplers have been described in the literature,
including a palladium joint [18], carbon tubes [19], and a cellulose acetate joint
[14,20]. In general, off-column detection yields the best limits of detection for
CEEC, but this conguration is the most difcult to implement due to the instability of the fracture.

Capillary Electrophoresis/Electrochemistry

465

Figure 2 Electropherogram of 10 nM each (1) dopamine, (2) norepinephrine, and (3)


isoproterenol. (Reprinted with permission from Ref. 15.)

B.

End-Column Detection

An alternative (and very widespread) approach to off-column detection is endcolumn detection (Figure 1B). In this case, small-diameter fused silica capillaries
are employed for the separation. It has been shown by Huang et al. that when capillaries with internal diameters of less than 25 m are employed, most of the voltage is dropped across the capillary [21]. Therefore, if the electrode is placed just
outside the end of the capillary, the separation voltage has minimal inuence on
the applied detection potential. This obviates the need for a decoupler, making the
overall system easier to construct and more rugged because the capillary is made
in one piece.
End-column detection is more popular than off-column detection due to the
ease of alignment and elimination of the often fragile fracture. One potential

466

Lunte et al.

drawback of end-column detection is the need to align the electrode with the outlet of the fused silica capillary without inserting the electrode. Initially, etched
conical outlets were used to align the electrode with the end of the capillary [22].
Because the i.d. of the cone was much larger than that of the capillary, most of the
voltage drop was across the capillary. Another technique involves constructing a
Naon tube at the end of a fused silica capillary and placing the electrode inside
it [23]. Detection limits using this approach were similar to those obtained with
off-column detection using a Naon tube, but adsorption of positive analytes was
not as great a problem. Detection limits for catecholamines were 3 nM.
Ever since Baldwin and Ye [24] rst demonstrated that macroelectrodes
could be used for end-column detection, this has been a common approach for
CEEC. Alignment of macroelectrodes is much simpler because they can just be
butted up to the end of the fused silica capillary. However, the sensitivity is not
as good as that obtained with off-capillary or optimized end-column detection. If
selectivity rather than sensitivity is the primary goal of the analytical system, this
method is the easiest to implement. Limits of detection for end-column detection
are usually in the micromolar range.
An important consideration in the use of end-column detection is the effect
of the separation voltage on the detector voltage. Both Matysik and Wallenborg
et al. [25,26] have shown that, even with 25 m i.d. capillaries, the separation
eld has a considerable effect on the applied voltage at the detector. Therefore,
the detection voltage must be adjusted for the separation voltage or the maximum
current response will not be obtained. This problem does not occur with off-column detection because in that case the detector is more efciently isolated from
the separation voltage.
Figure 3 shows a comparison of hydrodynamic voltammograms (HDVs)
obtained using end-column and off-column detection. A positive shift in potential of 130 mV was observed for catechol with end-column detection. The shift
that occurs is a function of both the separation voltage and the distance of the electrode from the end of the fused silica capillary. If the separation current is high,
the voltage drop across the capillary is low, leading to a higher voltage at the end
of the capillary and a greater inuence on the detection potential.
C.

Electrode Types

Selectivity in CEEC can be achieved through judicious selection of both the


detection potential and the type of working electrode employed. Obviously, the
lowest potential necessary to detect all analytes of interest is optimal for CEEC.
Most applications employ carbon ber or disk electrodes. However, other electrode materials have been used for the detection of specic classes of compounds.
For example, copper and nickel electrodes have been successfully employed for
the detection of carbohydrates, amino acids, and peptides following CE separa-

Capillary Electrophoresis/Electrochemistry

467

Figure 3 HDVs recorded for dopamine and catechol using end-column detection at capillary-to-electrode distances of 20 and 60 m without (A) and with (B) a fracture decoupler. Symbols: dopamine () and catechol () at 20 m; dopamine () and catechol ()
at 60 m. Open and closed symbols represent different trials. (Reprinted with permission
from Ref. 26.)

468

Lunte et al.

tion [2730]. Platinum and gold electrodes have been used for pulsed amperometric detection of carbohydrates and thiols [31].
Gold/mercury amalgam electrodes have been successfully employed for
the detection of thiols by CEEC. In this case, it is the catalytic oxidation of Hg in
the presence of thiols that provides the response.
Hg + 2RSH 3 RS2(Hg) + 2H+ + 2e

(1)

This oxidation occurs at a potential at which very few other compounds are oxidized (+150 mV vs. Ag/AgCl). In the rst application of this approach for CEEC,
a 50-m amalgamated gold wire was used in the off-column mode for the detection of glutathione in rat brain tissue homogenate [32]. The run buffer and the
samples were deoxygenated to avoid on-capillary oxidation of the thiols. Detection limits for glutathione were 21 nM at S/N = 3.
Chemically modied electrodes have also been employed as detectors for
CEEC [33,34]. Specically, carbon paste modied with cobalt phthalocyanine
was explored in our laboratories for selective detection of thiols [33]. Electrodes
were produced by packing a 150 m i.d. fused silica capillary with modied carbon paste to a depth of approximately 5 mm. This electrode was used for the
selective detection of cysteine in urine. Later, ruthenium cyanidemodied electrodes were used for the simultaneous detection of thiols and disuldes at +850
mV (vs. Ag/AgCl) [34]. Modied carbon ber microelectrodes were prepared by
cycling the potential between 500 and 1000 mV at a scan rate of 50 cycles in an
acidic deoxygenated plating solution containing RuCl3, K4Ru(CN)6, and KCl.
Detection limits for cystine were 3 M. The selective detection of cystine in urine
of a patient with kidney stones was demonstrated.
Biosensors have also been explored as detectors for CE. Carbon paste modied with glucose oxidase was used for the selective detection of glucose in serum
[33]. Platinum electrodes modied with acetylcholine esterase (AChE) and
choline oxidase have been used for the screening of acetylcholine esterase
inhibitors [35]. AChE converts acetylcholine to choline, which is then oxidized
by choline oxidase, leading to the production of hydrogen peroxide. The hydrogen peroxide is detected at the platinum electrode. CEEC with an AChE sensor
was used to identify peptide components that inhibit AChE activity in snake
venom. In these experiments, acetylcholine was added to the run buffer, leading
to a constant amperometric current for the oxidation of hydrogen peroxide. Negative peaks indicated inhibition of the enzyme. Components inhibiting AChE
were then identied by off-line MALDI-TOF analysis.
One of the advantages of electrochemical detection is that voltammetric
characterization can be used along with migration time for conclusive identication of compounds. This can be accomplished either by injecting the sample several times at different detection potentials or by using voltammetric detection.
The small sample requirements make the rst approach possible because only a

Capillary Electrophoresis/Electrochemistry

469

few nanoliters are injected into the capillary each time. This can be done with
great sensitivity. Both Ewings and Luntes groups [36,37] have described
voltammetric detectors for CE. In this case, voltammetry can be used to identify
particular analytes.
D.

Cell Holders

The rst reports of CEEC described the use of micromanipulators to position the
electrode into the end of the fused silica capillary. Since that time, numerous cell
holders have been designed [3841]. In 1998, Kok and coworkers reported the
direct coupling of a BAS Unijet cell with a fused silica capillary [39]. More
recently, Everett et al. reported the use of a commercial CEEC interface for offcolumn detection in CEEC [41]. A diagram of this cell conguration is shown in
Figure 4. This cell holder is used with electrodes with an i.d. larger than that of
the separation capillary. Either off-column or end-column detection can be
employed.
Instead of using a cell holder, the detector can be directly integrated into the
fused silica capillary. Metal electrodes can be permanently afxed to the end of
the capillary perpendicular to the ow [42,43]. Alternatively, metal can be
deposited directly onto the end of the capillary by vapor deposition [44]. One of
the rst of these employed a gold wire placed perpendicular to the opening at the

Figure 4 Commercial interface for CEEC with off-column detection. (Courtesy of Bioanalytical Systems, Inc.)

470

Lunte et al.

end of the fused silica capillary [42]. The electrode was afxed to the side of the
capillary and connections made with a copper wire. Off-column detection was
employed and fast analysis times were possible through the use of short separation capillaries. Separation of hydroquinone and dopamine in less than 30 s was
achieved using this capillary format (Figure 5) [42].
Another possibility is to afx a gold tube of the same inner diameter as the
external diameter of the fused silica capillary to the end of the fused silica capillary. Although these detectors exhibit a loss of efciency caused by band broad-

Figure 5 Fast separation of 100 M dopamine (DA) and hydroquinone (HQ). (Reprinted
with permission from Ref. 42.)

Capillary Electrophoresis/Electrochemistry

471

ening at the end of the capillary, they have been found to be especially useful for
pulsed amperometric detection (PAD) due to the high surface area of the electrode [43].

E.

Dual-Electrode Detection

Dual-electrode detection permits the selective detection of compounds undergoing chemically reversible redox reactions. Lin et al. rst reported a dual-electrode
detector for CE in 1994 [45]. An amalgamated gold wire inserted into a hole
drilled in the capillary served as the generator electrode with a similar wire in the
end of the capillary serving as the detector. Using this approach, the authors were
able to detect both thiols and disuldes. However, separation efciencies were
compromised due to the electrode positioning. This led to very high limits of
detection for disuldes (100 M).
Several dual electrode designs have been described by our group [4649]
(Figure 6). The rst detector consisted of a micro ring disk electrode produced by
chemical vapor deposition [46]. The disk consisted of a carbon ber that was surrounded by silicon and then a layer of carbon. The end of the electrode could be
cut with a razor to produce a planar ring-disk electrode that was easily aligned at
the end of the capillary. Figure 6A shows a schematic of this design. Selective
detection of several chemically reversible phenolic compounds was accomplished using this conguration.
Later, Zhong and Lunte described a dual electrode detector consisting of a
gold tube and a wire inserted into the fused silica capillary [47] (Figure 6B). This
conguration was used for the selective detection of thiols and disuldes. Disudes were reduced to their corresponding thiols at the tubular electrode. Thiols
were then detected at the amalgamated wire electrode by the catalytic oxidation
of mercury described previously. With a careful choice of the size of the tube and
the wire, it was possible to get high conversion efciencies at the rst electrode.
Detection limits for disuldes were over two orders of magnitude lower than in
the previously reported conguration employed by Lin et al. [45]. Figure 7 shows
the detection of cysteine, glutathione, and oxidized glutathione by using the tubular-wire electrode.
A third design developed by Holland et al. [48,49] consisted of two wires
at the end of the capillary, one inside the tube and a second across the end of the
capillary (Figure 6C). The wire provided high coulometric transformation at the
rst electrode. The second wire was used to detect the product of the rst reaction. Postcolumn amperometric reaction was evaluated for the indirect detection
of thiols and thioethers. Bromide present in the run buffer was oxidized to
bromine at the rst electrode. The resulting bromine was detected at the second
electrode downstream. Thiols and thioethers are oxidized by the bromine generated at the rst electrode, leading to a reduced response at the second electrode

472

Lunte et al.

Figure 6 Dual-electrode congurations for CEEC: (A) ring-disk, (B) tubular-wire, and
(C) wire-wire.

Capillary Electrophoresis/Electrochemistry

473

Figure 7 Simultaneous detection of thiols and disuldes with Hg/Au amalgamated tubular-wire dual electrode. Detection of (1) 50 M cysteine, (2) 100 M glutathione, and (3)
500 M glutathione disulde. (Reprinted with permission from Ref. 47.)

474

Lunte et al.

that is proportional to the analyte concentration. Detection limits for cysteine


were in the high nanomolar range. Dual-electrode detection of phenolic acids in
beer was also accomplished using this conguration [49].

III.

MICROCHIP SEPARATIONS

Over the past ten years, there has been an explosion of interest in the development
of analytical systems in the microchip format. CE is used both to manipulate uids and to achieve separations [4] in these devices. Very fast, highly efcient separations have been reported for such microchip CE systems. The use of CE in the
microchip format allows the use of high separation eld strengths (which
increases the efciency of the separation) because the materials typically used to
construct the microchips are very efcient in dissipating heat. Most microchip CE
devices have been constructed using glass [4], but devices have also been fabricated from materials such as plastics [50], low temperature co-red ceramics
(LTCC) [51], and poly(dimethylsiloxane) (PDMS) [52].
Most applications of microchip CE have used laser-induced uorescence
detection because of the ease of focusing directly onto the small channel on the
chip. Electrochemical detection is ideally suited for miniaturization to the
microchip format. The photolithographic techniques that are already used to
make the microchips can also be used to fabricate micron-sized detection electrodes reproducibly and inexpensively without a loss in sensitivity. If the power
supply and potentiostat are also miniaturized, it is possible to envision a complete
miniaturized total analysis system [4,53].
A schematic of a typical microchip CEEC device is shown in Figure 8. The
i.d. of the channel can vary from 10 to 100 m with typical straight separation
channels between 3 and 10 cm in length. A serpentine or semicircular design can
be implemented to increase the separation channel length up to 30 cm. In the basic
setup shown in Figure 8, buffer solution is introduced into the buffer and sample
waste reservoirs while the analytes of interest are placed in the sample reservoir.
The reservoir volume is often dened by afxing pipet tips to the chip, which
results in capacities of 200 L or less. The overall size of the microchip varies,
but is on the order of 10 cm2 or less. High voltages are applied to the reservoirs
via high voltage power supplies (230 kV) and platinum electrodes placed in the
reservoirs.
In contrast to the one-dimensional manner in which samples are injected
and then separated in conventional CE, microchip CE devices inject and separate
in two different dimensions; channels are rst lled with sample by electrophoresis across an injection channel that is perpendicular to the separation
channel (Figure 8). Once this channel is lled, a high voltage is applied to the separation channel and the separation is initiated. The issue of injection is one that

Capillary Electrophoresis/Electrochemistry

475

Figure 8 Schematic of a microchip CEEC device with dual working electrodes.

has been the point of numerous studies with microchip CE devices. The methods
that can be used become especially limited in microchip CEEC devices due to the
requirement that the detection reservoir, which contains the necessary electrodes
connected to a potentiostat, must always be held at ground.
Several groups have described end-column EC detection with microchip
CE devices fabricated from glass. Ewing and coworkers rst demonstrated EC
detection for separations in open-channel electrophoresis chips by using Pt microelectrode arrays to continuously monitor various analytes that had been sampled
by a capillary [54]. Woolley and coworkers later demonstrated a single-channel
glass microchip CEEC system that employed fabricated Pt electrodes [53]. Wang
and coworkers have reported two different glass microchip CEEC systems that
utilize single working electrodes in an end-column conguration; one uses a sputtered Au electrode [55] and a second uses an externally mounted screen-printed
electrode [56]. EC detection in plastic microchip CE devices using electrodes
made from carbon ink has been reported by Rossier et al. [57].
Our group has explored the use of LTCC substrates to construct a microchip
CE system utilizing a 25-m wire for EC detection [51]. The LTCC substrates are
relatively inexpensive to fabricate and are similar to glass in their electroosmotic

476

Lunte et al.

properties. The ceramics are opaque, but they do not need to be transparent for
electrochemical detection. The electroosmotic ow of the ceramic chips was
found to be similar to that of silica. Separation of catechol and dopamine was
accomplished in about 45 s.
More recently, PDMS has been investigated by a number of groups as the
substrate for microchip CE systems. It has several advantages for microchip CE.
These include ease of construction, the ability to make many devices from a single master, and the fact that most of the fabrication can be done outside the clean
room. Bonding of the separation layer can be done in either a reversible or an irreversible manner. Reversible bonding is especially useful when electrodes that are
expensive or difcult to construct are employed in the detection layer. The
reversible bonding makes it possible to clean or replace the separation channels
without having to make a new detection layer.
We recently described the rst example of EC detection in PDMS-based
microchip CE devices using a device similar to that depicted in Figure 8 [58].
Multiple gold electrodes were deposited on the glass substrate that was then covered with the PDMS separation channels. For dual electrode detection, two of
these electrodes were used. The microchip format is ideal for multiple electrode
detection in CEEC because it is much simpler to construct and align electrodes at
the end of a planar channel than to reproducibly place two electrodes at the end
of a fused silica capillary.
Figure 9 shows an example of the selectivity that can be gained by using
dual electrode detection with microchip CEEC. Tyrosine, 5-hydroxyindoleacetic
acid (5-HIAA), and catechol are all separated by CE. All of the compounds are
detected at E1, as all three undergo oxidation at the applied potential (+750 mV).
However, only catechol undergoes a chemically reversible redox reaction, and the
resulting quinone is detected at E2 with a potential of 100 mV. As shown in this
example, these dual electrode devices hold great promise for the analysis of complex biological samples containing analytes such as neurotransmitters and Cu(II)complexed peptides that possess a reversible redox couple. In general, the limits
of detection achieved with microchip CEEC devices have been inferior to those
possible with conventional CEEC, although this will probably change as more
studies with microchip CEEC devices are completed.
IV.

BIOANALYTICAL APPLICATIONS OF CEEC

A.

Biological Fluids

CEEC has been used extensively for the detection of small molecules in blood
and urine. Pulsed amperometric detection at a gold electrode was used to detect
glucose in blood following capillary electrophoresis separation [59]. It was also
detected using a glucose oxidasemodied carbon paste electrode [33]. As previ-

Capillary Electrophoresis/Electrochemistry

477

Figure 9 Electropherogram showing dual-electrode detection in microchip CEEC using


gold microelectrodes. Detection of 100 M tyrosine, HIAA, and catechol. (Reprinted with
permission from Ref. 58.)

ously described, cysteine and cystine have been detected in urine by CEEC using
chemically modied electrodes [33,34].
More recently, the activity of serum peptidases was investigated by capillary electrophoresis with electrochemical detection [60]. Increased peptidase
activity in blood is characteristic of a number of disease states. In this application,
leu-enkephalin was used as a model substrate. Leu-enkephalin and its metabolites
were separated and detected with CEEC following on-capillary copper complexation. By incorporating copper in the run buffer, peptides were complexed
directly on-capillary [61]. The copper(II) complexes could then be detected at
+700 mV by oxidation to Cu(III). The method shows good selectivity for peptides
over amino acids. This method was used to monitor the metabolism of leuenkephalin by enzymes present in a serum sample (Figure 10).
The use of CEEC for the detection of biomarkers for cancer has also been
investigated. 8-Hydroxydeoxyguanosine (8-OHdG) is present in urine as a result
of the oxidative DNA damage associated with cancer and aging. Detection of 8OHdG is typically accomplished using multiple column switching and/or solid

478

Lunte et al.

Figure 10 Metabolism of leu-enkephalin in plasma monitored by copper complexation


and CEEC: (A) plasma blank, (B) t = 30 min, (C) t = 60 min, (D) t = 90 min. Peak identication: (1) tyrosine, (2) leu-enkephalin, (3) des-tyr leu-enkephalin. (Modied with permission from Ref. 60.)

phase extraction steps prior to analysis. This procedure was signicantly simplied by Weiss and Lunte using solid-phase extraction and CEEC. Detection limits for 8-OHdG were 50 nM (S/N = 3). The concentration of 8-OHdG varied from
6 to 86 M with an average value of 42 26.9 M for four healthy female and
four healthy male volunteers [62].
B.

Microdialysis Sampling

Microdialysis is a technique capable of monitoring drugs and neurotransmitters


in vivo [63]. Figure 11 shows a typical microdialysis sampling system. A probe
consisting of a dialysis membrane is implanted in the tissue of interest. A solution
of ionic strength similar to that of the extracellular uid of interest is slowly perfused through the dialysis probe. Compounds that are in the extracellular uid diffuse through the dialysis membrane and are pumped to the fraction collector. To
obtain maximum recovery of analyte through the probe, very slow ow rates are
generally employed (0.11 L/min).
A number of different probes can be employed, depending on the tissue
being sampled (Figure 12). Brain probes consist of two concentric cannulas and

Capillary Electrophoresis/Electrochemistry

479

Figure 11 Diagram of a typical off-line microdialysis sampling system.

have diameters of approximately 250 m. The linear probe consists of a longer


piece of dialysis membrane and is most often used for tissue sampling and transdermal drug delivery. The exible probe is used for blood sampling, and the
shunt probe is most useful for bile sampling.
Microdialysis has several advantages as a sampling method. Due to the
low-molecular-weight cutoff of the dialysis membrane, proteins are excluded;
this makes possible direct injection of the sample into the LC or CE. Enzymes are
also excluded from the sample, obviating the need for enzyme inhibitors to ensure
sample integrity. Because there is no net uid loss, sampling can be done for long
periods of time. Last, compounds (drugs, peptides) can be delivered directly to
the site of interest to look at local metabolism or pharmacological action.
The temporal resolution obtained in a microdialysis experiment is both a
function of the sample volume requirements and the sensitivity of the analytical
method. Low ow rates provide larger analyte recoveries, which puts less constraint on the sensitivity of the analytical method. However, if one desires a temporal resolution of one minute or less, a technique capable of analyzing very small
sample volumes is necessary.
Capillary electrophoresis is well suited for the analysis of microdialysis
samples [64]. The injection volumes in CE are generally in the nanoliter to picoliter range. CE separations take place under predominantly aqueous conditions,
making them very compatible with the microdialysis samples. For off-line analysis, the sample volumes required are limited not by the CE separation step but by

480

Lunte et al.

Figure 12 Typical microdialysis probes for various tissues of rats or similar-size animals. (A) intracerebral guide cannula and rigid concentric cannula probe for brain tissue,
(B) linear probe used for sampling peripheral tissues such as liver, muscle, dermis, tumor,
kidney, (C) vascular probe for sampling from the jugular vein, and (D) shunt probe for sampling from the bile duct. Unlabeled arrows indicate direction of dialysate ow into and out
of probes. (Courtesy of Bioanalytical Systems Inc.)

the ability of the analyst to physically manipulate submicroliter samples. In particular, irreproducibility of sample transfer due to surface tension and evaporation
becomes an issue. Therefore, off-line analysis is usually performed with sample
volumes greater than one microliter.
Off-line analysis of microdialysis samples was rst used to monitor the
pharmacokinetics of L-dopa [65]. In this example, a exible probe was employed
for I.V. microdialysis sampling. The probe was perfused at 1 l/min. Five
microliter samples were collected for analysis, leading to a temporal resolution
of 5 min. The concentration of L-dopa in the blood was monitored for over

Capillary Electrophoresis/Electrochemistry

481

2 hr with microdialysis sampling, and the half-life of the drug was determined to
be 10 min.
One advantage of collecting samples off-line is that because of the small
sample requirements of CE (15 nL), it is possible to analyze the same small volume sample (15 L) several times without an appreciable loss in volume. Therefore, if the potential of the working electrode is changed between CE runs,
voltammetric characterization of attomole quantities of analytes is possible. The
purity of the peaks corresponding to L-dopa and its metabolites was determined
using this approach. Comparison of the current ratios obtained for the methyldopa
peak with those of a standard showed that the methyldopa peak was not pure.
More recently, dual electrode detection and scanning voltammetry methods that
make it possible to obtain voltammetric information in a single run have been
developed for CEEC [37].
Amino acid neurotransmitters have also been detected following microdialysis sampling. OShea et al. used microdialysis sampling and derivatization
with naphthalene-2,3-dicarboxaldehyde and cyanide (NDA/CN) to monitor the
release of excitatory amino acids in the brain [66]. Several amino acids, including Asp, Glu, and GABA, were detected in the electropherogram. Using the
migration time in conjunction with voltammetric characterization, positive identication of several peaks in the electropherogram was possible. Changes in the
concentration of aspartate and glutamate were monitored following an infusion of
potassium.
In a separate study, amino acids were detected in a brain microdialysis sample by CEEC without derivatization. A copper electrode was employed as the
working electrode and the CE separation was accomplished using a zwitterionic
buffer at a high pH. Under these conditions, Zhou and Lunte were able to detect
several amino acids in microdialysis samples [30].
Another example of the use of off-line microdialysis and CEEC is monitoring the transport of compounds across the blood-brain barrier [67]. Tryptophan
is metabolized to kynurenine, a precursor to kynurenic acid (KA), which has been
shown to be a naturally occurring N-methyl-D-aspartate (NMDA) receptor antagonist. However, KA cannot cross the blood-brain barrier and, thus, must be synthesized in the brain from either tryptophan or kynurenine. Microdialysis sampling was used to investigate the transport and metabolism of tryptophan across
this barrier by Malone et al. [67]. Rats were given intraperitoneal injections of
tryptophan, and the appearance of kynurenine and tryptophan in the brain was
monitored. Samples were analyzed by CEEC using a carbon ber electrode. Figure 13 shows an electropherogram of the microdialysis sample after 6 hr. In this
example, a perfusate ow rate of 200 nL/min (total sample volume, 3 L) was
employed to maximize recovery, leading to a temporal resolution of approximately 15 min. The identity of kynurenine in the dialysate sample was conrmed
by voltammetric characterization.

482

Lunte et al.

Figure 13 Electropherogram of dialysate sample taken after intraperitoneal tryptophan


administration. Peak identication: kynurenine (KN) and tryptophan (Trp). (Reprinted
with permission from Ref. 67.)

Microdialysis with off-line analysis has also been successfully employed


for the investigation of the pharmacokinetics of an enantiomeric drug, isoproterenol [68,69]. CE has been shown to be an extremely powerful method for the
separation of enantiomeric compounds. It is possible to place the chiral selector
in the run buffer, precluding the need for very expensive modied LC columns.
Due to the small volume requirements of CE, very little additive is required.

Capillary Electrophoresis/Electrochemistry

483

CEEC with methyl-O--cyclodextrin as the modier was used to monitor the


pharmacokinetics of () and (+) isoproterenol by I.V. microdialysis sampling.
Using a combination of peak stacking and electrochemical detection, limits of
detection of 0.6 ng/mL were achieved. Isoproterenol has a very short half-life. A
ow rate of 1 L/min was used for the microdialysis sampling. It was possible to
analyze the 1 L samples and monitor the pharmacokinetics of the isoproterenol
enantiomers with a temporal resolution of one minute (Figure 14).
The advantage of the small sample volume requirements of CE can be best
realized using on-line systems. In this case, the CEEC system is directly coupled
to the microdialysis sampling system (Figure 15). This approach was rst
described using laser-induced uorescence as the detection method [70,71]. With
CEEC, both the animal and the detection system must be grounded. This is
accomplished using two joints, one isolating the rat from the high voltage applied
at the CE-MD interface and the second used to decouple the CE separation voltage from the electrochemical detector [72]. The 60 nL valve and injection interface converts the continuous microdialysis samples to discrete volumes for CE
analysis. Nicotine was separated from other electroactive components by CEEC
(Figure 16). A linear probe was implanted in the skin and the concentration of
nicotine was continuously monitored by CEEC. Figure 17 shows the results

Figure 14 Pharmacokinetic prole for protein-free isoproterenol: (A) (+) isoproterenol


and (B) () isoproterenol. (Reprinted with permission from Ref. 69.)

Figure 15 Schematic of an on-line microdialysis-CEEC system.

Capillary Electrophoresis/Electrochemistry

485

Figure 16 Electropherograms of transdermally delivered nicotine in the microdialysate


at various time points. Arrows indicate peak corresponding to nicotine. (Reprinted with
permission from Ref. 72.)

Figure 17 Time course of nicotine in the microdialysate following patch administration.


(Reprinted with permission from Ref. 72.)

486

Lunte et al.

obtained for the transdermal delivery of nicotine obtained using the on-line system. An important advantage of an on-line system is that the entire experiment is
automated so that the analyst can devote his or her attention to the well-being of
the animal.

V.

SUMMARY

In the future, capillary electrophoresis with electrochemical detection will continue to be important for the analysis of small volume samples. The direct coupling of microdialysis with CEEC yields a separation-based sensor that is capable of near real-time monitoring of drugs and neurotransmitters. New on-line
systems for the detection of catecholamines and peptides are currently under
development. Electrochemical detection is very amenable to miniaturization and,
therefore, is uniquely compatible with the microchip format. In the future, the
integration of microchip CEEC with microdialysis sampling will lead to truly
portable separation-based sensors.

ACKNOWLEDGMENTS
The authors would like to thank the National Science Foundation (Grant CHE9702631) and Bioanalytical Systems Inc. for nancial support. Support for
R.S.M. through a National Institutes of Health postdoctoral fellowship (F32
NS11053-01) is gratefully acknowledged. The authors would also like to thank
Nancy Harmony and Malonne Davies for their help in the preparation of this
manuscript.

REFERENCES
1.
2.
3.
4.
5.
6.

Jorgenson, J.W.; Lukacs, K.D. Zone Electrophoresis in Open Tubular Glass Capillaries. Anal. Chem. 1981 53, 12981302.
Holland, L.A.; Chetwyn, N.P.; Perkins, M.; Lunte, S.M. Capillary Electrophoresis in
Pharmaceutical Analysis. Pharm. Res. 1997 14, 372387.
Xu, Y. Capillary Electrophoresis. Anal. Chem. 1999 71, 309R313R.
Dolnik, V.; Liu, S.; Jovanovich, S. Capillary Electrophoresis on Microchip. Electrophoresis 2000 21, 4154.
Swinney, K.; Bornhop, D.J. Detection in Capillary Electrophoresis. Electrophoresis
2000 21, 12391250.
Kappes, T.; Hauser, P.C. Recent Developments in Electrochemical Detection Methods for Capillary Electrophoresis. Electroanalysis 2000 12, 165170.

Capillary Electrophoresis/Electrochemistry
7.
8.

9.
10.
11.

12.
13.

14.
15.

16.
17.

18.
19.

20.
21.
22.
23.

24.

487

Holland, L.A.; Lunte, S.M. Capillary Electrophoresis Coupled to Electrochemical


Detection: A Review of Recent Advances. Anal. Commun. 1998 35, 1H4H.
Wang, A.; Fang, Y. Applications of Capillary Electrophoresis with Electrochemical
Detection in Pharmaceutical and Biomedical Analysis. Electrophoresis 2000 21,
12811290.
Voegel, P.D.; Baldwin, R.P. Electrochemical Detection in Capillary Electrophoresis.
Electrophoresis 1997 18, 22672278.
You, T.; Yang, X.; Wang, E. Applications of Microelectrodes in Capillary Electrophoresis/Electrochemical Detection. Electroanalysis 1999 11, 459464.
OShea, T.J. Electrochemical Detection for Capillary Electrooresis. In Pharmaceutical and Biomedical Applications of Capillary Electrophoresis; Lunte, S.M.,
Radzik, D.M., Eds., Vol. 2 in Progress in Pharmaceutical and Biomedical Analysis,
Series Eds., C.M. Riley and A.F. Fell, Oxford: Oxford, U.K.; Pergamon, Tarrytown,
N.Y., 1996; 277306.
Wallingford, R.A.; Ewing, A.G. Capillary Zone Electrophoresis with Electrochemical Detection. Anal. Chem. 1987 59, 17621766.
OShea, T.J.; Greenhagen, R.D.; Lunte, S.M.; Lunte, C.E.; Smyth, M.R.; Radzik,
D.M.; Watanabe, N. Capillary Electrophoresis with Electrochemical Detection
Employing an On-Column Naon Joint. J. Chromatogr. 1992 593, 305312.
Zhou, J.; Lunte, S.M. Membrane-Based On-Column Mixer for Capillary Electrophoresis/Electrochemistry, Anal. Chem. 1995 67, 1318.
Park, S.; Lunte, S.M.; Lunte, C.E. A Peruoronated Ionomer Joint for Capillary
Electrophoresis with On-Column Electrochemical Detection. Anal. Chem. 1995 67,
911918.
Hu, S.; Wang, Z.; Li, P.; J. Cheng, Amperometric Detection in Capillary Electrophoresis with an Etched Joint. Anal. Chem. 1997 69, 264267.
Quian, J.; Wu, Y.; Yang, H.; Michael, A.C. An Integrated Decoupler for Capillary
Electrophoresis with Electrochemical Detection: Application to Analysis of Brain
Microdialysate. Anal. Chem. 1999 71, 44864492.
Kok, W.T.; Sahin, Y. Solid State Decoupler for Off-Column Detection in Capillary
Electrophoresis. Anal. Chem. 1993 65, 24972501.
Yik, Y.F.; Lee, H.K.; Li, S.F.Y.; Khoo, S.B. Micellar Electrokinetic Chromatography of Vitamin B6 with Electrochemical Detection. J. Chromatogr. 1991 585,
139144.
Wang, C.W.; Chen, I.C. Cellulose acetate porous polymer joint for capillary zone
electrophoresis. Anal. Chem. 1992 64, 24612464.
Huang, X.; Zare, R.N.; Sloss, S.; Ewing, A.E. End-Column Detection for Capillary
Electrophoresis. Anal. Chem. 1991 63, 189192.
Sloss S.; Ewing A.G. Improved Method for End-Column Detection for Capillary
Electrophoresis. Anal. Chem. 1993 65, 577581.
Park, S.; Lunte, C.E. A Peruorosulfonated Ionomer End-Column Electrical Decoupler for Capillary Electrophoresis/Electrochemical Detection. Anal. Chem. 1995 67,
43664370.
Ye, J.; Baldwin, R.P. Amperometric Detection in Capillary Electrophoresis with
Normal Size Electrodes. Anal. Chem. 1993 65, 35253527.

488
25.
26.

27.

28.

29.

30.

31.
32.

33.
34.

35.

36.
37.
38.
39.
40.
41.

42.

Lunte et al.
Matysik, F.-M. Improved End-column Amperometric Detection for Capillary Electrophoresis. J. Chromatogr. A. 1996 742, 229234.
Wallenborg, S.R.; Nyholm, L.; Lunte, C.E. End-Column Amperometric Detection in
Capillary Electrophoresis: Inuence of Separation-Related Parameters on the
Observed Half-Wave Potential for Dopamine and Catechol. Anal. Chem. 1999 71,
544549.
Fermier, A.M.; Coln, L.A. Capillary Electrophoresis with Constant Potential
Amperometric Detection Using a Nickel Microelectrode for the Detection of Carbohydrates. J. High Res. Chromatogr. 1996 19, 613616.
Coln, L.A.; Dadoo, R.; Zare, R.N. Determination of Carbohydrates by Capillary
Electrophoresis with Amperometric Detection at a Copper Microelectrode. Anal.
Chem. 1993 65, 476481.
Ye, J.; Baldwin, R.P. Determination of Amino Acids and Peptides by Capillary Electrophoresis and Electrochemical Detection at a Copper Electrode. Anal. Chem. 1994
66, 26692674.
Zhou, J.; Lunte, S.M. Direct Determination of Amino Acids by Capillary Electrophoresis/Electrochemistry Using a Copper Microelectrode and Zwitterionic
Buffers. Electrophoresis 1995 16, 498503.
Weber, P.L.; Lunte, S.M. Capillary Electrophoresis with Pulsed Amperometric
Detection of Carbohydrates and Glycopeptides. Electrophoresis 1996 17, 302309.
OShea, T.J.; Lunte, S.M. Selective Detection of Free Thiols by Capillary Electrophoresis-Electrochemistry Using a Gold/Mercury Amalgam Microelectrode.
Anal. Chem. 1993 65, 247250.
OShea, T.J.; Lunte, S.M. Chemically Modied Electrodes for Capillary Electrophoresis/Electrochemistry. Anal. Chem. 1994 66, 307311.
Zhou, J.; OShea, T.J.; Lunte, S.M. Simultaneous Detection of Thiols and Disuldes
by Capillary Electrophoresis-Electrochemical Detection Using a Mixed Valence
Ruthenium Cyanide-Modied Microelectrode. J. Chromatogr. A 1994 680,
271277.
Chetwyn, N.P. Rapid Screening of Enzyme Inhibitors Using Capillary Electrophoresis with Enzyme-Based Biosensors. Ph.D. Dissertation, University of
Kansas, Lawrence, KS, 1997.
Ferris, S. S.; Lou, G.; Ewing, A.G. Scanning Electrochemical Detection in Capillary
Electrophoresis. J. Microcol. Sep. 1994 6, 263268.
Park, S.; McGrath, M.J.; Smyth, M.R.; Diamond D.; Lunte, C.E. Voltammetric
Detection for Capillary Electrophoresis, Anal. Chem. 1997 69, 29943001.
Fermier, A.M.; Gostkowski, Coln. L.A. Rudimentary Capillary-Electrode Alignment for Capillary Electrophoresis. Anal. Chem. 1996 68, 16611664.
Durgbanshi, A.; Kok, W.T. Capillary Electrophoresis and Electrochemical Detection with a Conventional Detector Cell. J. Chromatogr. A 1998 798, 289296.
Chen, M.C.; Huang, H.J. An Electrochemical Cell for End-Column Amperometric
Detection in Capillary Electrophoresis, Anal. Chem. 1995 67, 40104014.
Everett, W.R.; Bohs, C.; Davies, M.I. Use of a New Interface Cell for Off-Column
CEEC Determination of Catecholamine Neurotransmitters. Current Separations
2000 19, 2528.
Zhong, M.; Lunte, S.M. Integrated On-Capillary Electrochemical Detector for Capillary Electrophoresis. Anal. Chem. 1996 69, 24882493.

Capillary Electrophoresis/Electrochemistry
43.

44.

45.
46.

47.

48.

49.

50.
51.

52.

53.
54.

55.

56.
57.

58.

59.

489

Zhong, M.; Lunte, S.M. Development and Characterization of an Integrated OnCapillary Tubular Electrode for Capillary Electrophoresis/Electrochemistry. Anal.
Commun. 1998 35, 209212.
Voegel, P.D.; Zhou, W.; Baldwin, R.P. Integrated Capillary Electrophoresis/Electrochemical Detection with Metal Film Electrodes Directly Deposited onto the Capillary Tip. Anal. Chem. 1997 69, 951957.
Lin, B.L.; Coln, L.; Zare, R.N. Dual Electrode Detection of Cysteine and Cystine
in Capillary Electrophoresis. J. Chromatogr. A 1994 680, 263270.
Zhong, M.; Zhou, J.; Lunte, S.M.; Zhao, G.; Giolando, D.M.; Kirchhoff, J.R. Dual
Electrode Detection for Capillary Electrophoresis/Electrochemistry. Anal. Chem.
1996 68, 203207.
Zhong, M.; Lunte, S.M. Tubular-Wire Dual Electrode for Detection of Thiols and
Disuldes by Capillary Electrophoresis/Electrochemistry. Anal. Chem. 1999 71,
251255.
Holland, L.A.; Lunte, S.M. Postcolumn Reaction Detection with Dual Electrode
Capillary Electrophoresis-Electrochemistry and Electrogenerated Bromine. Anal.
Chem. 1999 71, 407412.
Holland, L.A.; Harmony, N.M.; Lunte, S.M. Characterization of an Integrated OnCapillary Dual Electrode for Capillary Electrophoresis-Electrochemistry. Electroanalysis 1999 11, 327330.
Becker, H.; Gartner, C. Polymer Microfabrication Methods for Microuidic Analytical Applications. Electrophoresis 2000 21, 1226.
Henry, C.S.; Zhong, M.; Lunte, S.M.; Moon, K.; Bau, H.; Santiago, J.J. Ceramic
Microchips for Capillary Electrophoresis-Electrochemistry. Anal. Commun. 1999
36, 305307.
McDonald, J.C.; Duffy, D.C.; Anderson, J.R.; Chiu, D.T.; Wu, H.; Schueller,
O.J. A.; Whitesides, G.M. Fabrication of Microuidic Systems in Poly(dimethylsiloxane). Electrophoresis 2000 21, 2740.
Woolley, A.T.; Lao, K.; Glazer, A.N.; Mathies, R.A. Capillary Electrophoresis
Chips with Integrated Electrochemical Detection. Anal. Chem. 1998 70, 684688.
Gavin, P.F.; Ewing, A.G. Characterization of Electrochemical Array Detection for
Continuous Channel Electrophoretic Separations in Micrometer and Submicrometer
Channels. Anal. Chem. 1997 69, 38383845.
Wang, J.; Tian, B.; Sahlin, E. Integrated Electrophoresis Chips/Amperometric
Detection with Sputtered Gold Working Electrodes. Anal. Chem. 1999 71,
39013904.
Wang, J.; Tian, B.; Sahlin, E. Micromachined Electrophoresis Chips with ThickFilm Electrochemical Detectors. Anal. Chem. 1999 71, 54365440.
Rossier, J.S.; Schwarz, A.; Reymond, F.; Ferrigno, R.; Bianchi, F.; Girault, H.H.
Electrochemical Detection in Polymer Microchannels. Electrophoresis 1999 20,
727731.
Martin, R.S.; Gawron, A.J.; Lunte, S.M.; Henry, C.S. Dual Electrode Electrochemical Detection for Poly(dimethylsiloxane)-Fabricated Capillary Electrophoresis
Microchips. Anal. Chem. 2000 72, 31963202.
OShea, T.J.; Lunte, S.M.; LaCourse, W.R. Detection of Carbohydrates by Capillary
Electrophoresis with Pulsed Amperometric Detection. Anal. Chem. 1993 65,
948951.

490
60.

61.

62.

63.
64.
65.

66.

67.

68.

69.

70.
71.

72.

Lunte et al.
Gawron, A.; Lunte, S.M. Detection of Neuropeptides Using On-Capillary Copper
Complexation and Capillary Electrophoresis with Electrochemical Detection. Electrophoresis 2000 21, 32053211.
Deacon, M.; OShea, T.J.; Lunte, S.M.; Smyth, M.R. Determination of Peptides by
Capillary Electrophoresis-Electrochemical Detection Using On-Column Cu(II)
Complexation. J. Chromatogr. 1993 652, 377383.
Weiss, D.J.; Lunte, C.E. Detection of a Urinary Biomarker for Oxidative DNA Damage: 8-Hydroxyguanosine by Capillary Electrophoresis. Electrophoresis 2000 21,
20802085.
Hansen, D.K.; Davies, M.I.; Lunte, S.M.; Lunte, C.E. Microdialysis Sampling for
Pharmacological Studies. J. Pharm. Sci. 1999 88, 1427.
Lunte, S.M.; Malone, M.A.; Zou, H. Capillary Electrophoresis/Electrochemistry for
the Analysis of Microdialysis Samples. Current Separations 1994 13, 7579.
OShea, T.J.; Telting-Diaz, M.W.; Lunte, S.M.; Lunte, C.E.; Smyth, M.R. Capillary
Electrophoresis-Electrochemistry of Microdialysis Samples for Pharmacokinetic
Studies. Electroanalysis 1992 4, 463468.
OShea, T.J.; Weber, P.L.; Bammel, B.P.; Lunte, C.E.; Smyth, M.R.; Lunte, S.M.
Monitoring Excitatory Amino Acid Release In Vivo by Microdialysis with Capillary
Electrophoresis/Electrochemistry. J. Chromatogr. 1992 608, 189195.
Malone, M.; Zou, H.; Lunte, S.M.; Smyth, M.R. Determination of Tryptophan and
Kynurenine in Brain Microdialysis Samples by Capillary Electrophoresis with Electrochemical Detection. J. Chromatogr. A 1995 700, 7380.
Hadwiger, M.E.; Torchia, S.R.; Park, S.; Biggin, M.E.; Lunte, C.E. Optimization of
the Separation and Detection of the Enantiomers of Isoproterenol in Microdialysis
Samples by Cyclodextrin-Modied Capillary Electrophoresis with Electrochemical
Detection. J. Chromatogr. B. 1996 681, 241249.
Hadwiger, M.E.; Park, S.; Torchia, S.R.; Lunte, C.E. Simultaneous Determination of
the Elimination Proles of the individual Enantiomers of Racemic Isoproterenol
Using Capillary Electrophoresis and Microdialysis Sampling. J. Pharm. Biomed.
Anal. 1997 15, 621629.
Hogan, B.L.; Lunte, S.M.; Stobaugh, J.F.; Lunte, C.E. On-line Coupling of Microdialysis Sampling and Capillary Electrophoresis. Anal. Chem. 1994 66, 596602.
Zhou, S.Y.; Zuo, H.; Stobaugh, J.F.; Lunte, C.E.; Lunte, S.M. Continuous In Vivo
Monitoring of Amino Acid Neurotransmitters by Microdialysis Sampling, On-line
Derivatization and CE Separation. Anal. Chem. 1995 67, 594599.
Zhou, J.; Heckert D.M.; Zuo, H.; Lunte, C.E.; Lunte, S.M. On-Line Coupling of In
Vivo Microdialysis with Capillary Electrophoresis/Electrochemistry. Anal. Chim.
Acta 1999 379, 307317.

salient comp dept. peach typesetting dekker 150644 toth

16
Ultrahigh Sensitivity Analysis of
Amino Acids and Peptides by
Capillary Liquid Chromatography
with Electrochemical Detection
Brendan W. Boyd and Robert T. Kennedy
University of Florida, Gainesville, Florida

I.
A.

INTRODUCTION
Overview

One of the most important analytical applications of electrochemistry is as a


detector in liquid phase separations such as high-performance liquid chromatography (HPLC). Since the seminal work of Kissenger and Adams [1], it has
become apparent that the resolving power of HPLC nicely complements the
selectivity and sensitivity of amperometric electrodes, allowing routine determination of electroactive compounds in complex mixtures. Catechols, indoles,
disuldes, phenolic acids, aromatic amines, carbohydrates, organometallics, and
metal ions are some of the compounds that are sensitively measured by this
approach. Applications including analysis of neurotransmitters, pharmaceuticals,
sugars, amino acids, peptides, and nucleic acids have all been described. Electrochemical techniques used for HPLC detectors include amperometry (reductive
and oxidative), voltammetry, pulse methods, and potentiometry. As a result of
this versatility, electrochemical detection has become a valuable tool in the arsenal of the analytical chemist.

491

salient comp dept. peach typesetting dekker 150644 toth

492

Boyd and Kennedy

When coupled to HPLC, electrochemical detection provides detection limits of 1 fmol for favorable analytes. Although this is an impressively low detection limit, many problems, especially in the biological sciences, require even better sensitivity. Examples include analysis of single cells or small volume uids
such as dialysates collected in vivo. Over the past 15 years, considerable effort
has been devoted to improving the sensitivity of liquid chromatography (LC) with
electrochemical detection by miniaturizing the column and electrode. At this
point, low attomole amounts [2] can be detected and concentration detection limits down to picomolar have been achieved [3,4]. In this chapter, we give background on the development of capillary liquid chromatography (CLC) and electrochemical detection. Detailed discussion of sensitivity improvements made by
(1) on-column preconcentration, (2) improved derivatization chemistry, and (3)
recycling, microfabricated electrochemical detectors will be presented. Examples
of applications to determination of amino acids and peptides will be used to illustrate the potential of high sensitivity methods.

B.

Capillary Liquid Chromatography

The development of CLC, where the inner diameter (i.d.) of the separation column is less than 50 m, has been driven by both the general trend in miniaturization of analytical devices for improved performance and by the need for methods
capable of nanoscale analysis. Three styles of CLC columns are currently under
development: open tubular (OTLC), monoliths, and packed capillaries. In OTLC
the column is a small bore (125 m i.d.) glass or fused silica capillary with a stationary phase bonded on the surface of the inner wall. In theory, OTLC offers the
highest performance of any LC method; however, low sample capacity and difculty of fabrication at the diameter needed for best performance (theoretical calculations suggest 1 m i.d. to be the best) have resulted in relatively slow development [5]. In a monolithic column, a porous support network with stationary
phase on the surface is formed inside a capillary tube (or on a microfabricated
channel) with i.d. of 10 to 150 m. Monolithic columns are receiving attention
mainly for use with capillary electrochromatography [6]. In packed capillary
columns, conventional HPLC particles (1 to 5 m diameter spheres with stationary phase bonded to the surface) are packed into fused silica capillaries with i.d.
from 10 to 350 m. While theoretical performance is not as high as OTLC
columns, packed capillaries have become the most popular approach for the following reasons: they (1) yield higher efciencies than HPLC columns [7,8], (2)
are commercially available or can be easily prepared in-house with simple equipment, (3) have excellent column-to-column reproducibility [3,4], and (4) have
good sample capacity. Extensive reviews of CLC and CEC can be found elsewhere [911].

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

C.

493

Electrochemical Detectors for Capillary


Liquid Chromatography

A major challenge in use of miniaturized chromatographic systems is detection.


This challenge arises from the high sensitivity required and the need for low volume detectors [12]. Columns with i.d.s of 2550 m use less than 10 nL injection volumes requiring attomole sensitivity to detect analytes at nanomolar concentrations. Low volume of the detector is also important to prevent loss of
chromatographic efciency due to extra-column band broadening. It is generally
attempted to keep the volume of the detector cell less than half of the standard
deviation of the narrowest peak. This requires detector cells with nanoliter volumes for CLC. Two congurations of electrochemical detectors for capillary liquid chromatography, end-column and on-column, that t the needed criteria have
been described.
The end-column detector, also called a wall-jet, involves positioning the
working
 electrode near the column outlet (Figure 1A) as originally described by
Krejc [13]. In this initial report, a 0.1 cm diameter Pt disk working electrode was
positioned approximately 0.1 mm from the outlet of a 14 m i.d. OTLC column.
Characterization of the detector revealed a thin-layer cell performance with 27%
conversion efciency and a detection limit of 680 amol for hydroquinone (see
chromatogram in Figure 2B). A disadvantage of the end-column conguration
observed in this work was the loss of efciency that occurs when the solute peak
elutes from the column outlet into bulk solution prior to being detected (Figure 2).
The contribution of this extracolumn band-broadening was determined to be
ow-rate dependent and resulted in four-fold higher plate heights than predicted
by the Golay equation (Figure 2) [13,14]. This loss in chromatographic separation
efciency (i.e., theoretical plates) and electrochemical conversion efciency is
characteristic of all end-column-based electrochemical detectors. Thus, end-column detectors are characterized by ease of use, moderate sensitivity, and signicant extra-column band broadening.
In on-column detectors, the working electrode is actually inserted into the
tube used to make the column as illustrated in Figure 1B. (Technically, these are
not truly on-column detectors unless the stationary phase actually extends to
the end of the tube; however, for this discussion, on-column will refer to the case
of the electrode inserted into the same tube used for the column.) The rst on-column electrochemical detector was a 1 m tip potentiometric, ion-selective electrode inserted into the outlet of a 25 m i.d. OTLC column [15]. The detection
cell volume of 20 fL was proportional to the surface area of the electrode and the
length of the stagnant diffusion layer from the electrode surface. Unlike the endcolumn detector, experiments demonstrated that the on-column detector did not
cause a measurable loss in theorical plates. Little follow-up on use of micropotentiometric electrodes as detectors has been reported.

salient comp dept. peach typesetting dekker 150644 toth

494

Boyd and Kennedy

Fused silica capillary

Wire for electrical


connection

Figure 1 Illustrations of electrochemical detector congurations coupled to capillary


liquid chromatography columns. (A) End-column or wall jet electrode conguration with
the working electrode positioned proximal to the fused silica capillary outlet. (B) On-column electrode showing a cylindrical carbon ber positioned in the fused silica capillary
outlet forming a thin annular layer between the ber surface and the inner capillary wall.

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

495

Figure 2 Chromatographic efciency (A) and a chromatogram (B) obtained using an


end-column electrochemical detector coupled to an OTLC column. (A) Experimentally
measured data of the OTLC system (plotted points) is compared to the theoretical efciency as predicted by the Golay equation (solid curve) for a column diameter of 14 m and
diffusion coefcient of 105 cm2/s. (B) Chromatogram obtained for an injection of 0.76 pg
hydroquinone onto a 14 m 1.2 m column operated at 6 mm/s. (Adapted from Ref. 13.)

salient comp dept. peach typesetting dekker 150644 toth

496

Boyd and Kennedy

The more common approach to on-column detection utilizes an amperometric microelectrode inserted into the column outlet as originally described in
1984 [16]. Insertion of the microelectrode, typically a cylindrical ber or wire,
creates a thin-layer ow path in the annulus between the working electrode and
capillary wall. This thin-layer effect results in conversion efciencies that can
approach 100%. For example, in the original work, a 9 m diameter by 0.7 mm
long C-ber microelectrode was inserted into a 15 m i.d. reverse-phase column
producing a thin-layer ow path with a thickness of 3 m. This detector was operated in the anodic mode (oxidative current detected) with a simple two-electrode
potentiostat. The detection limit for hydroquinone in this initial study was 1 fmol
for a 0.7 nL injection volume; however, subsequent improvements including electrical shielding, lower noise ampliers, and better electrode construction have
reduced the noise to as low as 60 fA and detection limits to 110 amol [2,3]. In
addition to excellent sensitivity, this approach results in negligible band broadening [8,16]. The main disadvantage of this approach is that a micromanipulator and
microscope are required for proper positioning of the electrode. The simplicity
and effectiveness of this design have caused considerable spread of its use. In
addition to CLC, this approach has been extensively used with capillary electrophoresis (CE) as well [17]. In this application, the electrical current from the
electrophoresis column must be separated from the detector current for sensitive
detection. Similar isolation from the separation current would also be required for
electrochemical detection in CEC. Commercial versions of these detectors are not
yet available; however, technology is available to make this a reality.
Besides xed-potential anodic measurements, other electrochemical methods have also been coupled to CLC. These modes typically offer detection of
other analytes or improved selectivity over the anodic amperometric method.
Reductive mode detection for OTLC using an on-column, carbon ber microelectrode has been described; however, detection limits are approximately 100fold worse than anodic detection because of the high background currents [18].
Differential pulse detection has been applied for improved selectivity [18]. Integrated, pulsed amperometric detection (IPAD) at Au microwires has also been
applied to capillary columns for determination of thiols [19]. Scanning voltammetric detection using carbon ber microelectrodes has been used to increase the
amount of electrochemical information garnered by the on-column detector and
to improve resolution by allowing simultaneous chromatographic and voltammetric analysis [20,21]. A recent innovation, sinusoidal voltammetry, has yet to
be used with CLC but has been used with CE and ow injection analysis (FIA)
[22,23]. In this method, the electrode potential is varied in a sinusoidal wave with
the resulting current traces analyzed by Fourier transform. (In this method the
potential waveform is a sinusoid, with no other DC potential or linear ramps
applied.) The background and signal can have different frequency spectra, so

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

497

analysis in the frequency domains allows signicant discrimination against the


background, resulting in improved S/N. This method is best used with electrodes
that exhibit mainly capacitive background currents. This is because electrodes
that generate faradaic currents in their background signal (i.e., surface waves) will
have frequency spectra similar to faradaic current from analytes, thus preventing
the discrimination offered by frequency analysis. Using this approach with Cu
electrodes allowed concentration limits of detection as low as 70 nM for ATP
[22]. Because the detector response increases with the number of bases in a
nucleotide chain, this technique permits high sensitivity detection for oligonucleotides. For a 9.5 kilobase chain of double-stranded DNA, the concentration
limit of detection was 3.2 pM, corresponding to 14 zmol or 8400 molecules (4 nL
injection volume) using the sixth harmonic where the signal-to-noise ratio is maximal (Figure 3) [23].
D.

Derivatization

The high sensitivity and selectivity of amperometric electrochemical detection


make it attractive to extend its capabilities to non-electroactive compounds by
chemical derivatization of selected functional groups to generate an electroactive
product. An important example of this approach is use of the amine-reactive
reagents o-phthalaldehyde (OPA) and naphthalene-2,3-dicarboxaldehyde (NDA)
for analysis of amino acids. The OPA reaction, in which OPA and a nucleophile
(typically a thiol) react with amines to produce an isoindole that can be easily oxidized at carbon electrodes (see Figure 4 for reaction). The capabilities of this
approach in a CLC format were demonstrated by separation by gradient elution
and electrochemical detection of 18 proteinacious amino acids derivatized by
NDA [24]. The method obtained a 36 amol limit of detection for NDA-labeled
asparagine corresponding to a concentration limit of detection of 5 nM in a separation time of 50 min.
If analytes must be derivatized, it may seem preferable to use laser-induced
uorescence (LIF) detection because LIF detection is often more sensitive than
electrochemical detection; however, several advantages of electrochemical detection make it worth considering. One advantage is low cost of equipment. A
microelectrode system requires an inexpensive manipulator and current amplier.
In contrast, the LIF system will require a laser (lasers with multiple wavelengths
can be particularly expensive), optics, multiple manipulators, a photomultiplier
tube with associated power supply, and a current amplier. Another advantage is
in quantication. With a 100% conversion efciency and knowledge of the number of electrons transferred per oxidation (or reduction), it is possible to quantify
without a calibration curve using Faradays law. Finally, in some cases, derivatization chemistry may favor electrochemical detection. One example is detection

salient comp dept. peach typesetting dekker 150644 toth

498

Boyd and Kennedy

Figure 3 Flow injection analysis of 1 nM of a 9.5 kilobase oligonucleotide at a Cu electrode with sinusoidal voltammetry. Upper graph shows the signal at different frequencies
after Fourier transform of the voltammograms for single-stranded (ss) and double-stranded
(ds) DNA. Lower trace shows a ow injection trace for the ds DNA at the sixth harmonic.
The excitation sine was from 0.05 to 0.55 V at 2 Hz. The FIA ow rate was 0.5 mL/min
and the electrolyte was 0.1 M NaOH. (Adapted from Ref. 23.)

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

499

Figure 4 Derivatization and scavenging reactions. Top: Derivatization of amino acid


(GABA) by OPA with t-buSH as a nucleophile. Middle: Elimination of excess OPA by
addition of glycine (GLY). Bottom: Elimination of excess t-buSH by addition of iodoacetamide (IAA).

of -amino butyric acid (GABA), an important amino acid neurotransmitter. OPA


and NDA derivatives of this amino acid are not stable unless a bulky nucleophile,
such as tert-butyl thiol (t-buSH), is used in the reaction. However, use of t-buSH
instead of the other nucleophiles greatly decreases the uorescence yield. Thus,
GABA and several related amino acids are best detected by OPA/ t-buSH derivatization with electrochemical detection [25]. Another example of derivatization
chemistry that favors electrochemical detection is the biuret reaction for peptides
[26]. This method will be discussed below.
E.

Applications

CLC with electrochemical detection (EC) has been used in a variety of applications including microscale amino acid analysis of proteins [24] and analysis of
microdialysate fractions [4]. Perhaps the application that best exemplies the

salient comp dept. peach typesetting dekker 150644 toth

500

Boyd and Kennedy

capabilities of CLC-EC, however, is single cell analysis. CLC-EC was the rst
modern instrumental technique that was used for quantitative determination of
multiple species in single cells [27]. The ability to quantitate the contents of single cells in biology is relevant due to the heterogeneity of tissue, especially in neurobiology. When large populations of cells are analyzed, there is an averaging
effect for the cellular components that is dependent on the number of different
types of cells and their relative abundance. In microorgans where a number of different cell types are present, it is benecial to know the chemical composition of
the individual cells in order to differentiate their physiological function. Several
examples of single cell analysis by LC-EC have been reported, including determination of catecholamines, indoleamines, and amino acids in single neurons and
adrenal chromafn cells [2729]. Both voltammetric and amperometric detection
have been used.

II. SENSITIVITY ENHANCEMENTS:


ON-COLUMN PRECONCENTRATION
As discussed above, the inherent high mass sensitivity of microelectrodes coupled with CLC columns has allowed interesting developments in nanoscale analysis. Although the mass sensitivity is extraordinary, the concentration detection
limit is no better than large-scale methods. That is, nanomolar concentrations are
typical detection limits for both CLC-EC and conventional HPLC-EC. Improving
concentration detection limits would be of benet in at least two ways. First,
analysis of microscale samples requires sophisticated sample handling in order to
prevent dilution. For example, most single cell work has involved use of manually operated micromanipulators and micropipettes to transfer and derivatize
samples of nanoliter volume. If higher concentration sensitivity was available, it
would be possible to dilute samples to convenient volumes, thus avoiding the
need for specialized sample handling, allowing analysis to be performed with
ordinary dispensers and injectors. Second, determination of trace level constituents naturally requires better concentration detection limits. One example is
determination of peptide hormones or neurotransmitters, which are often present
at picomolar concentrations.
A simple approach to sensitivity enhancement is to utilize the preconcentrating ability of column chromatography. In on-column preconcentration, the
sample is injected in a weak mobile phase so that it is retained at the head of the
column and concentrated [30,31]. Analytes are eluted with a stronger mobile
phase after the less retained solutes are washed out. The greater the capacity factor (k) for analyte in the sample solution, the greater the preconcentration that can

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

501

be achieved. When retention is poor (i.e., the k of the analyte in the injected solution is low), the peaks can begin to migrate down the column during a large volume injection, resulting in broadening or breakthrough (where analytes actually elute during the injection), giving rise to variable retention times and peak
areas or excessive peak widths. Thus, a successful preconcentration requires identication of columns that will yield high k for the analyte of interest. For biological samples, which are frequently aqueous, a reversed-phase column is an excellent match for preconcentration because aqueous solutions are weak mobile
phases. The ease of preconcentration is a key advantage of CLC over CE, making CLC quite valuable for trace level work on small samples [31]. Importantly,
the miniaturization of the column means that it is possible to use a small sample
(e.g., 5 L), and gain 100- to 1000-fold preconcentration down to nanoliter volumes. To gain a similar level of preconcentration on a conventional-sized column
would require 250 mL of sample.
As a demonstration of this concept, we can consider the example of detecting met-enkephalin in a microdialysate sample collected from the brain of a living rat [32]. In preliminary studies we determined a mass detection limit of 40
amol for met-enkephalin by CLC-EC. This detection limit was obtained with an
on-column C-ber electrode in a 25 m i.d. column packed with 5 m particles.
The mass detection limit of this method compares favorably to the 100400 amol
detection limit that is typical for the most common method of bioactive peptide
detection, radioimmunoassay (RIA); however, the concentration detection limit
is worse. RIA concentration detection limits are approximately 1 pM (100 L
sample volume) whereas the LC-EC method had a concentration detection limit
of 2 nM (20 nL injection volume).
We found that an Alltima C-18 stationary phase allowed excellent retention, and therefore extensive preconcentration of met-enkephalin when injected
in aqueous solutions was readily achieved as illustrated by the data in Figure 5
and Table 1. In this experiment, the total amount of analyte was kept constant
while the injection volume was increased. As shown by the gure and quantied
in the table, a constant peak area was obtained indicating no loss of the analyte
during preconcentration. The retention time was also unaffected by the injection
volume, indicating that met-enkephalin did not elute during the sample loading.
Finally, the peak widths (measured as the variance of the chromatographic zone)
were also unchanged, indicating no broadening of the analyte zone during preconcentration. The results in Table 1 show that even with a sample volume of 2
L, which is equivalent to approximately 59 column volumes (capillary volume
is estimated to be 34 nL), the met-enkephalin did not signicantly migrate
through the column. As shown in Table 1, preconcentration of 2 L onto the column results in a detection limit that is 100-fold lower than that achieved with a
typical injection volume of 20 nL.

salient comp dept. peach typesetting dekker 150644 toth

502

Boyd and Kennedy

Figure 5 Comparison of chromatograms for different volume injections but constant


mass (800 amol) of met-enkephalin (peak indicated by arrow). Injections were 10 nL of 80
nM (A), 100 nL of 8.0 nM (B), and 1000 nL of 0.80 nM (C). Capillary column was 8 cm
long, 25 m i.d. fused silica capillary packed with 5 m Alltima C-18 reversed-phase particles. After loading sample, the column was rinsed with 5% solvent B95% solvent A for
5 min before gradient elution started. Gradient elution started at 5% solvent B, linearly
changed to 50% solvent B in 5 min, kept at this composition for 2 min, and then stepped
back to 5% solvent B. Solvent A was 1 mM phosphate buffer in 10 mM sodium sulfate
adjusted to pH 7.0. Solvent B was 40% (v/v) of solvent A with 60% acetonitrile. Detection
utilized a 9 m diameter by 2 mm long C-ber electrode inserted into the outlet of the column. Detector electrode potential was set at +1.0 V versus Ag/AgCl reference.

Table 1 Effect of Large Injection Volume on Characteristics


of Met-Enkephalin Peak
Injection
volume (nL)a
10
30
100
500
1000
2000
aFor

Concentration
of sample (nM)

Retention
time (s)

Height of
peak (pA)

Limit of
detection
b
(pM)

Variance of
peak (s2)

80.0
26.7
8.00
1.60
0.80
0.40

382
383
387
386
388
384

23.1
24.2
20.4
21.5
23.3
22.1

4200
1300
470
89
41
22

0.13
0.14
0.14
0.13
0.14
0.13

each injection, the mass injected was the same.


as the concentration that would give a signal 3 times the RMS noise.

bCalculated

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

503

Larger injection volumes give rise to better concentration detection limits,


but the large time required to inject L volumes onto columns with nL volumes
can preclude larger preconcentration. Increasing injection volume from 0.02 L
to 1 L improves the detection limit 50-fold to 40 pM, for an increase in injection
time of 8.1 min. (At 2 nL/s, the injection time was 0.16 min for 0.02 L and 8.3
min for 1 L.) A further two-fold improvement in detection limit to 20 pM by
injecting 2 L would require another 8.3 min for injection, for a total sample loading time of 16.6 min. This added time is signicant relative to the total analysis
time and will often be considered an inconsequential improvement at the expense
of total analysis time.
The data in Figure 5 and Table 1 show that large preconcentration factors
improve the detection limit with no negative effect on the mass detection limit or
the chromatographic efciency; however, some practical problems emerged. The
rst was that the preconcentration step concentrated impurities in the water and
buffers, resulting in baseline drift. For example, in Figure 5 a large broad peak,
presumably due to detection of preconcentrated impurities, increased with injection volume. (Subsequent studies have shown that the magnitude of this drift
depends strongly on the quality of water used to dissolve the sample.) Another
problem encountered with large volume injections was that the electrode sensitivity decreased by up to 50% after loading a 1 L sample. This effect was
observed only with samples dissolved in solutions containing high levels of salts.
The reason for this deactivation is unclear; however, it can be avoided by setting
the potential of the detector electrode at 0.0 V during the loading of large samples
and then resetting to the detection potential during the separation.

III.

PRECONCENTRATION WITH DERIVATIZATION: PEPTIDES

Although CLC-EC with on-column preconcentration yields high sensitivity, it is


only applicable to naturally electroactive peptidesi.e., oligopeptides that contain tyrosine or tryptophan. Derivatization methods may allow one to extend this
approach to other neuropeptides. One possible route to peptide derivatization for
electrochemical detection is to use the biuret reagent (cupric sulfate and sodium
potassium tartrate in borate buffer at pH 10.5) as described in a series of articles
by Webers group [26,3340]. In this elegant method, peptides react with Cu2+
under basic conditions to form a complex (see Figure 6) that can be detected electrochemically based on oxidation of Cu2+-peptide to Cu3+-peptide. This derivatization method is unique in that the reaction does not require primary amines but
rather involves the peptide backbone. As a result, the method is highly selective
for peptides and allows detection of peptides, such as those with post-translational
modications to the N-terminus, which are undetectable by reagents that act on
amines. Webers group has carefully explored the chemistry of this method and

salient comp dept. peach typesetting dekker 150644 toth

504

Boyd and Kennedy

Figure 6 Possible coordination congurations of Cu2+ by peptides. Cu2+ can be coordinated by amide nitrogens (II) or carboxylic acid oxygens (I) from acidic amino acids. The
Cu2+ center can be oxidized to Cu3+ for amperometric detection.

found that a wide range of bioactive peptides with 3 to 18 amino acids can be
detected with detection limits of 6 to 100 fmol on a microbore HPLC column
[37,38].
Based on this promising background, we explored the biuret reagent as a
pre-column reagent combined with CLC-EC. Previous work with HPLC-EC
detection of biuret complexes has demonstrated that both pre-column derivatization and post-column derivatization are possible [26,34,37]; however, most work
has emphasized post-column derivatization primarily because of difculties associated with performing separations at the high pH needed for complex formation
and detection. The difculty of working with post-column reactors in capillary
separations prompted us to pursue pre-column derivatization. The advent of polymer-based HPLC supports tolerant of high pH mobile phases made this strategy
feasible.
Our goal was to determine if the biuret reagent would be effective at low
concentrations and coupled with preconcentration. Preconcentration with derivatization requires (1) derivatization chemistry that does not produce extraneous
compounds that are also preconcentrated and give rise to background peaks, (2)
derivatization chemistry that is compatible with weak mobile phases (aqueous
solutions for reversed-phase LC), and (3) derivatization chemistry that has adequate kinetics and specicity to work quantitatively at trace concentrations.
Figure 7A illustrates successful detection of four neuropeptides at 100 pM
to 1 nM using pre-column biuret derivatization, on-column preconcentration, and

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

505

Figure 7 High sensitivity detection of biuret peptide complexes by CLC-EC. Chromatograms are from injection of 1 L of 100 pM bradykinin (B), 100 pM vasopressin (V),
400 pM oxytocin (O), and 1 nM neurotensin (N). (A) Raw chromatogram. (B) Chromatogram after median ltering. (C) Blank after median ltering. Chromatographic conditions same as Figure 5. Capillary column was 15 cm long by 25 m i.d. fused silica capillary packed with 5 m Astec C-18. The aqueous portion of the mobile phase (solvent A)
was 40 mM sodium carbonate buffer in 0.75 mM sodium potassium tartrate and 0.25 mM
cupric sulfate adjusted to pH 10.5 with sodium hydroxide. The organic phase (solvent B)
consisted of 40% (v/v) of solvent A with 60% acetonitrile. Gradient elution started at 5%
solvent B, linearly changed to 35% solvent B in 5 min, kept at this composition for 2 min,
and then stepped back to 100% solvent A. Other conditions similar to Figure 5.

salient comp dept. peach typesetting dekker 150644 toth

506

Boyd and Kennedy

EC detection. As with electroactive peptides, the main negative effect of preconcentrating on-column with derivatization is the baseline drift associated with
impurities in the solvents (see Figure 7A for example). The large difference in the
shape of drift compared to the analyte peaks allows it to be removed by appropriate data-processing. Figure 7 illustrates the effect of a median lter [41], a type
of high-pass lter, on a chromatogram for trace level peptides. The chromatograms are compared to an injection of a blank containing the same solution
but no peptides. The featureless background allows condent assignment of
peaks even at low concentrations. Detection limits for these oligopeptides varied
between 7 and 60 pM for 1 L injection volumes [3]. The chromatograms in Figure 7 illustrate that the biuret derivatization is well suited for use with preconcentration because (1) it allows narrow peaks to be obtained even with large volume injections on reversed-phase columns because it does not require organic
solvents, allowing the sample to be injected in a weak mobile phase, (2) it does
not produce substantial background peaks that interfere with analyte detection,
resulting in surprisingly clean blank chromatograms even for large volume injections, and (3) it yields derivatization even at trace level concentrations.
The LC-EC method is attractive for peptides because of the high sensitivity possible. The concentration detection limit is comparable to the low pM detection limits achieved by RIA [42,43] but with the added advantage of simultaneous detection of multiple analytes. The main limitation of the biuret approach is
the uncertainty of detection limits for a given peptide. Exploration of other peptides has revealed that detection limits can vary by over an order of magnitude [3].
The detection limit variability may be due to several factors, including variable
loss of samples due to adsorption, differences in electron transfer rates for the different peptides at C-ber surfaces and the +0.8 V potentials that were used, and
differences in the yields of the derivatization reaction.

IV. PRECONCENTRATION WITH DERIVATIZATION:


AMINO ACIDS
The biuret reagent is unusual in that when the reagent is added to a blank solution,
minimal additional background peaks are observed. It is much more common
when using chemical derivatization reagents to observe a plethora of background
peaks associated with reagent. These background peaks can be considered chemical noise and are frequently the noise that sets the detection limit rather than the
noise of the detector itself. The background peaks result from either contamination or undesirable, nonselective products of the derivatization reaction. Unfortunately, when a sample is preconcentrated, chemical noise is also preconcentrated,
resulting in no improvement in detection limit if the analyte elutes near a back-

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

507

ground peak. In addition, chemical noise can impair assay selectivity if the chemical noise is not identied, because the signal being measured can result from both
analyte and background peak. Successful implementation of preconcentration
when chemical noise is present requires a strategy that reduces the production of
background chemicals.
A classic example of a derivatization reagent that produces chemical background is the amine-selective reagent OPA/t-BuSH. The chemical background
observed by using this reagent to derivatize a blank solution composed of de-ionized water without any intentionally added amines is illustrated in Figure 8A. In
this case, the blank solution was preconcentrated with 500 nL injected onto a 50
m i.d. by 22 cm long column (~400 nL total volume). As shown, the resulting
chromatogram is littered with electroactive peaks of different magnitudes
throughout the elution window. The presence of these peaks can cause ambiguity
in identifying peaks, makes resolution more difcult, and increases detection limits by 100-fold [44]. This phenomena is not unique to the OPA/t-BuSH reagent;
similar effects of chemical noise have been observed using other uorogenic or
electrogenic reagents in pre-column derivatization [4547]. In one example,
derivatization of arginine with a sheath-ow cuvette and LIF detection, the detection limits were more than 10,000-fold larger when direct derivatization was performed compared to derivatizing arginine at a high concentration and then diluting the sample extensively to measure the instrumental detection limits [48].

V.

METHODS TO REDUCE CHEMICAL NOISE


FOR AMINE DERIVATIZATION

The chemical noise observed from derivatization reagents can come from several
sources, including (1) contamination of sample with amines, which then react
with the reagent, (2) nonselective reactions of the reagent other than those
intended during derivatization, and/or (3) naturally electroactive compounds that
contaminate the reagent solutions themselves. Strategies to reduce chemical noise
require that the individual components be identied and then selectively targeted
for minimization.
Contamination of samples can arise from reagents, their solvents, and surfaces that contact the sample during preparation. Reagent-based contamination is
minimized by purication of the derivatization reagent using purication methods such as recrystallization or column chromatography [46,49]. Another
approach to reducing derivatization reagent contamination is to age the derivatization solution prior to use. This approach is limited to derivatization reagents
that produce unstable amine derivatives such as OPA/-mercaptoethanol and uorescamine. Background peaks attributed to solvents can often be cleaned by
passing them through a 0.2 m hydrophobic Teon lter [3,4].

salient comp dept. peach typesetting dekker 150644 toth

508

Boyd and Kennedy

Any surface that contacts the sample or reagents, including stir bars, pH
meters, sample vials, and pipette tips, can be a source of contamination [4]. This
contamination has been attributed to airborne particles and chemicals associated
with the laboratory environment including food particles and dust particles contaminated with amino acids and proteins from pest urine and feces [50]. Noise arising from surface-contamination (aside from the pH meter) can be eliminated with
cleaning protocols to remove adsorbed contaminants. A cleaning protocol for
polypropylene surfaces (derivatization vessels and pipette tips) using 1 M HCl, deionized water, and ethanol has been demonstrated to reduce the level of surface
contamination signicantly [4,46]. While successful for a number of amino acids,
background peaks that coeluted with serine and glycine are not completely eliminated by this approach [4,51]. To avoid contamination from pH meters, buffers for
mobile phase and derivatization solutions can be prepared by adding the needed
amount of acid and conjugate base to be self-adjusting in pH and then ltered using
a 0.2 m Teon lter. Usually surface contamination becomes a greater source of
noise as the dimensions of the derivatization are miniaturized.
Although contamination can be problematic, the dominant source of noise
encountered using amine-reactive reagents such as OPA, NDA, and 3-(4-carboxybenzoyl)-quinoline-2-carboxaldehyde (CBQCA) arise from nonselective reactions of the excess derivatization reagent [4,44,45,48,52]. A large excess (typically
200-fold) of derivatization reagent is required to provide pseudorst order reaction kinetics, short reaction times, and a quantitative derivatization reaction.
(Some groups have attempted to circumvent the background problem by using a
minimal excess of reagent; however, this results in a method that does not allow
quantitative reactions.) The combination of high reagent concentration and high
concentration sensitivity of preconcentration techniques mean that even minor
reaction products of 0.0001% yield can result in large peaks in the chromatogram.
One method to minimize this source of chemical noise is to use scavenging
reactions to drive the excess reagent to a single, stable product that does not interfere with the assay [4,44,45,52]. In one example of an assay for GABA using
OPA/t-BuSH [44], the excess OPA/t-BuSH was scavenged by adding glycine to
excess after the analytical reaction (Figure 4). The glycine derivative is electroactive but is not retained with the mobile phase conditions used; thus, it did not
interfere [44]. The remaining t-BuSH (present at a 2:1 molar ratio to OPA) was
removed using iodoacetamide to form a non-electroactive sulde (see reactions
in Figure 4).
The use of glycine as a scavenger proved useful for GABA detection, but it
was necessary to use other scavenging conditions to allow analysis of a broader
range of amino acids. One solution was to follow a protocol of scavenging excess
t-buSH using iodoacetamide and then scavenging excess OPA using cysteic acid
and sodium sulte. The latter reaction formed a product that incorporated a highly
polar amine moiety from cysteic acid and a polar, charged nucleophile (sulte) to

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

509

form N-alkyl-1-isoindole sulphonate, which is unretained on the capillary column


during sample preconcentration [4]. When these reactions are performed consecutively, the chemical noise is reduced by 90% (measured as the area of all background peaks in the elution window) as shown in Figure 8A and 8B. This low
background permits high sensitivity analysis of multiple amino acids as illustrated by the chromatogram at 25 nM in Figure 8C). The detection limits for this
method using 0.25 L injection volumes range from 90340 pM for the neurotransmitter and neuromodulator amino acids. The upper limit of injection volume
for this reagent system was 0.25 L; after this point, more chemical noise was
observed that could not be removed using the aforementioned cleaning or chemical scavenging strategies.
It seems likely that scavenging reactions could be used with other derivatization strategies to improve detection limits. Other approaches to reduce chemical noise have also been explored to reduce the chemical noise associated with
background peaks, including acid hydrolysis [53], photodestruction [53], and
electrochemical scavenging (or coulombic preoxidation) [5456]. However, all
these approaches require that the derivatization reaction by-products possess
lower chemical stability than the target species.
VI. NANOSCALE ANALYSIS USING
SAMPLE PRECONCENTRATION
As discussed above, an advantage of using preconcentration with capillary
columns is the ability to gain substantial improvements in detection limits on
microliter samples. In working with samples where the total volume is 1 L, however, it was observed that the major source of chemical noise in the system was
not derivatization reaction by-products but surface contamination. As the surface
area to volume ratio of reaction vessels increased during the initial scale-down
work, the chemical noise increased signicantly with different aspect ratio vials.
With cleaning protocols for vials and micropipette tips (see Section V), the chemical noise for the OPA/t-buSH reaction and chemical scavenging in a 1 L
dialysate sample was reduced to match that of the larger scale derivatization reactions (see Figure 9 for sample chromatograms).
VII. AUTOMATED MICROSCALE ANALYSIS: IMPROVING
PRECISION, SENSITIVITY, AND THROUGHPUT
The sensitivity of clean derivatization/preconcentration/CLC-EC method is
impressive, but its lack of precision becomes apparent when working with small
samples. With manual reagent dispensing for derivatization, the relative standard
deviation (RSD) varies from 5 to 8% for amino acids. Furthermore, the entire 2

salient comp dept. peach typesetting dekker 150644 toth

510

Boyd and Kennedy

Figure 8 Demonstration of chemical noise produced by amine derivatization coupled


with gradient elution. (A) Blank of deionized, HPLC-grade water without intentionally
added amines after OPA-buSH derivatization. (B) Blank after derivatization and scavenging using IAA followed by cysteic acid/sulte. (C) 25 nM amino acid standard derivatized
as in (B). All sample injection volumes were 100 nL from a 500 L sample. Chromatography was performed using a 48 m 40 cm packed capillary column (5 m Alltima C8) and
a linear, binary gradient of mobile phase A (50 mM phosphate pH 6.8 w/ 1 mM EDTA) and
B (35% mobile phase A and 65% acetonitrile (v/v) with 1 mM EDTA) from 35% to 70%
B using a slope of 2% B/min using a ow rate of 0.3 L/min. Detection performed using a
9 m 1.5 mm cylindrical carbon, ber microelectrode held at + 0.75 V (vs. Ag/AgCl reference). Amino acids: L-aspartate (ASP), L-glutamate (GLU), L-asparagine (ASN), L-histidine (HIS), L-glutamine (GLN), L-citrulline (CIT), L-arginine (ARG), L-serine (SER), Lthreonine (THR), glycine (GLY), L-tyrosine (TYR), L-alanine (L-ALA or ALA), taurine
(TAU), and GABA.

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

511

Figure 9 Microscale analysis of amino acid standards and 10 s microdialysate fractions


(120 nL sample collected from live rat brain). Blank of saline (bottom trace) and an amino
acid standard (top trace) were diluted to 2 L and derivatized in a polypropylene microvial.
Concentrations of the amino acids (nM): ASP (10), GLU (20), ASN (20), HIS (40), GLN
(500), CIT (20), ARG (50), SER (100), O-phosphoelthanolamine (PEA, 20), THR (50),
GLY (100), TYR (20), L-ALA (100), -alanine (-ALA, 25), TAU (50), and GABA (10).
Inset is expanded portion of the chromatogram where GABA elutes. For all experiments,
an injection volume of 250 nL was used. Separation and detection conditions are as
described in Figure 8 except the mobile phase pH was 6.51.

L sample must be consumed to perform the sample injection because the dead
volume of the sample injection valve must be overlled. Automation of the system using a microscale autosampler/reagent dispenser can enhance precision and
minimize the amount of sample required to perform an injection.
Coupling a packed capillary column to a conventional scale autosampler is
not feasible due to the large internal volume of conventionally sized injection
valves and the extra-valve system volume (or mobile phase dwell volume) that
can be several orders of magnitude greater than the capillary column volume.
Large injection valve volumes result in chromatographic dispersion and undesirable mobile phase mixing that can disturb the integrity of the gradient. Recently
however, microscale autosamplers/reagent dispensers with low dead volume
injection valves have become commercially available (FAMOS, LC Packings,
San Francisco, CA). Using such a system and a mobile phase delivery system
designed for capillaries, the entire analysis including sample derivatization, injection, and separation can be automated with a 50 m i.d. packed capillary column.

salient comp dept. peach typesetting dekker 150644 toth

512

Boyd and Kennedy

Packed Capillary Column


Faraday Cage

Current Amplier
Fused Silica
Needle

Syringe Pump Controler

Figure 10 System conguration for the automated, capillary-scale amino acid analyzer.
Liquid chromatography is performed using a high pressure syringe pump (Isco 100-DM,
Lincoln NE). The mobile phase exits the syringe pump through a set of check valves (C)
and 0.2 m stainless steel lters (F) prior to mixing. The volumetric ow from the pump is
reduced from 6 to 0.3 L/min using a fused silica capillary splitter and carried from the
splitter to the autosampler using a short length of 50 m i.d. capillary. The 50 m i.d.
packed capillary column is mounted directly to the low-dead volume injection valve. The
electrochemical detector is similar to that described in previous gures.

A block diagram of the system is shown in Figure 10. In this conguration, the
mobile phase dwell volume can be kept low by placing a mobile phase splitter as
close to the injection valve as possible. The system is not ideal, because lightly
retained peaks (k < 10 in aqueous mobile phase) will exhibit band broadening
under isocratic conditions using this conguration because of the large volume of
mobile phase that passes through the column during injection; however, the main
goal here is to separate species that can be efciently preconcentrated at the column head. Use of this system improved RSDs to 13% while maintaining comparable sensitivity, as shown in Figure 11.
The sensitivity and analysis time were further improved by optimizing the
gradient elution program using the linear solvent strength gradient procedure
[57]. As shown in Figure 12AC, increasing the steepness of the gradient resulted

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

513

Figure 11 Ultra-trace level analysis of an amino acid standard using the automated, capillary-scale amino acid analyzer. A 2 L, low nanomolar amino acid standard (top trace)
and blank (bottom trace) is derivatized and preconcentrated using the automated amino
acid analyzer. The samples were contained in a polypropylene, 384-well microtiter plate.
The reagents were added, mixed, and siphoned for injection using the fused silica needle of
the autosampler (Figure 14). The time axis is truncated to highlight the elution window of
the amino acid derivatives. All separation and detection parameters are as described in Figure 9.

in faster separations, narrower peak widths, and increased peak heights. The
increase in peak heights improved detection limits to approximately 150 pM for
these analytes.

VIII.
A.

THE FUTURE: MICROFABRICATED DEVICES


Background

Microfabricated instrumentation has become an active area of research over the


past 10 years and it promises to play a dominant role in the future of chemical
analysis. Although microfabricated devices such as sensors, electrode arrays, and
even gas chromatography have been available for some time, the current revolution has been brought about by the implementation of microuidics and integration of sample preparation and separation on chips. Both CE and LC systems have

salient comp dept. peach typesetting dekker 150644 toth

514

Boyd and Kennedy

Figure 12 Chromatographic resolution and analysis time for mobile phase gradients
designed to produce different average capacity factor. Amino acid separations using an
average capacity factor (k of 23 (A), 11.6 (B), and 5.8 (C). The gradient amplitude
(3563%B), ow rate (0.3 L/min), and injection volume (250 nL) for all the separations
were held constant while the gradient slope was varied. Analyte and retention order the
same as in Figure 9.

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

515

been integrated with a variety of sample preparation steps in micro total analysis
systems [58,59]. Electrochemical detection is especially attractive for microuidic systems because the entire detection device can easily be integrated into the
system (as opposed to optical detection where technology for incorporation on a
chip is still in the development stage).
Several reports have already appeared describing the use of microfabricated
electrochemical detectors integrated with separations, mainly CE, on chips
[6065]. In general the detection limits in microfabricated EC detectors are considerably higher than what has been achieved with capillary designs because the
micofabricated designs used so far have lower conversion efciency and higher
noise than the on-column capillary EC detectors. Newer designs and implementation of other materials such as carbon electrodes, which are lower noise than Pt
or Au electrodes, are required to achieve detection limits comparable to capillary
systems.
B.

Recycling Electrochemical Detection

With appropriate design, it may be possible to achieve considerably higher sensitivity with a microfabricated system by taking advantage of the ability to control
the electrode geometry with great precision. We have explored the approach of
using two working electrodes with one electrode xed at a potential sufcient to
oxidize the analyte and the other xed at a potential sufcient to reduce the oxidized product back to its original state. If the electrodes are spaced closely, individual analyte molecules can be electrochemically recycled by diffusing back and
forth between the electrodes as they ow through the cell, resulting in detection
of individual molecules multiple times and enhanced signals. The amount of recycling observed is dependent on the time required for molecules to diffuse from
one electrode to the other and the residence time of analyte in the detector [66,67].
This approach can also enhance selectivity because only chemically reversible
compounds are recycled [66,67].
Recycling electrochemical detection, also called regenerative electrochemical detection, has been explored in owing streams before [6670]. Most of these
studies have utilized a parallel-opposed geometry (identical electrodes on opposite sides of the ow channel) with macroelectrodes (areas ~0.2 cm2) separated by
Teon spacers 20100 m thick and ow rates compatible with HPLC, i.e., 10 to
1000 L/min [6671]. Under these conditions, residence times of analyte in the
channel representing the detector cell are short and signal enhancements over single electrode cells are 5- to 10-fold [67,68]. At low ow rates found with CLC
systems, the residence time of analytes for a given electrode separation should be
relatively long, allowing more efcient recycling [67]. One difculty with implementing the idea of recycling electrochemical detection compatible with microcolumn separations and low ow rates is the fabrication of electrochemical cells.
The total volume of the cell must be in the nanoliter range and channel depths

salient comp dept. peach typesetting dekker 150644 toth

516

Boyd and Kennedy

must be on the order of micrometers for the electrodes to be analytically useful.


The precision possible with microfabrication offers a route to achieve this high
sensitivity.
With the parallel-opposed geometry, it is apparent that decreasing the channel depth (i.e., the distance between electrodes) will decrease the time required to
diffuse between the electrodes and therefore increase signal. However, if the
channel width and length are kept the same as the channel depth is decreased, then
the volume of the channel will decrease, which will decrease residence time (for
a given ow rate) and the amount of recycling possible. An alternative is to maintain the same channel volume as the channel depth is decreased by increasing the
width or length of the channel and electrodes. In this case, the noise, which is
nominally proportional to electrode area, will increase as the signal from recycling is increased. The effects of these different strategies are illustrated by the
results from a random walk simulation shown in Table 2. As seen in the table,
decreasing the channel depth while maintaining the maximum electrode area
results in the greatest ratio of current to electrode area (assumed to be proportional
to S/N)
We have prepared electrodes with geometry suitable for high recycling efciency and volume compatible with CLC and CE. A diagram of the ow cells is

Table 2 Simulation of Effect of Electrochemical Cell Geometry on Performancea


Channel Electrode
depth
width
(m)
(m)b
1.0
2.0
5.0
10
20
1.0
2.0
5.0
10
20

50
50
50
50
50
50
25
10
5.0
2.5

Electrode
Current/
area
Detector
Conversion
electrode
(cm2
volume Residence efciency Current area (nA/cm2
103)
(nL)
time (s)
(%)
(nA)
103)c
2.50
2.50
2.50
2.50
2.50
2.50
1.30
0.50
0.25
0.13

0.25
0.50
1.30
2.50
5.00
0.25
0.25
0.25
0.25
0.25

1.5
3.0
7.5
15
30
1.5
1.5
1.5
1.5
1.5

68,000
33,900
16,400
8,100
4,000
68,000
16,900
3,200
700
100

1092
546
274
131
64
1091
272
52
12
2

437
218
109
52
25
436
217
103
48
17

aSimulations were performed using a random walk. For all simulations, channel length was 5 mm, analyte concentration was 0.1 mM, ow rate was 10 nL/min, diffusion coefcient was 6 106 cm2/s, and
one-electron was transferred per oxidation.
bChannel width was the same as the electrode width.
cUsed as a relative S/N.

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

517

Figure 13 Design of recycling electrochemical detector. (A) Top view of nished electrodes on Pyrex and SiO2. (B) Cross-section of assembled detector. Arrows indicate ow
path. The following dimensions were used: channel depth = 510 m, channel width = 85
m, electrode width = 50 m (see top view, section A), electrode and channel length = 26
mm. A SCE reference electrode is placed in the ow outlet.

shown in Figure 13. Gold electrodes were placed into a channel on an insulated
Si wafer and on the surface of a glass slide. The two pieces were fused together
by anodic bonding and connected by epoxy to a fused silica capillary for ow
injection. The effect of signal recycling is shown in Figure 14, which illustrates a
60-fold enhancement in signal for injection of a plug of 100 M ferricyanide. In
our initial work, we found that the noise tended to increase when using the recycling detector as compared to using a single microfabricated electrode for conventional amperometric detection; however, modication of the potentiostat
according the model of Weber [67] greatly reduced this effect (unpublished
results). Estimated detection limits are 100 pM with the present design for ferricyanide. Much remains to be done to determine the feasibility of this approach for
actual analysis, however. Future work will require coupling the regenerative system with a separation and developing fabrication procedures suitable for carbon
electrodes that are compatible with preparation of the overall system.

salient comp dept. peach typesetting dekker 150644 toth

518

Boyd and Kennedy

Figure 14 Flow injection current-time traces for injections of a pulse of 100 M


Fe(CN)64 detected by a dual electrode cell with and without the recycling electrode connected. The detector electrode was at +0.3 V vs. SCE and the recycling electrode was at
0.1 V vs. SCE. The channel depth was 5 m, the channel length was 3 mm, and the ow
rate was 50 nL/min.

IX.

SUMMARY

CLC-EC allows determination of peptides and amino acids in real samples at low
picomolar concentrations on 1 L volumes. This extraordinary sensitivity is
equivalent to the most sensitive methods, including uorescence, mass spectrometry, and immunoassay. The approach of low noise derivatization should be
equally applicable to CE and CEC with EC detection; however, these methods
will require better preconcentration methods than those currently available.

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection

519

Applications of this approach include measurement of neurotransmitters in


dialysates and single cell analysis. Microfabricated instrumentation with recycling electrochemical detector may further improve the detection limit for species
that can be oxidized and reduced multiple times. Microfabrication may also make
this approach more widely available. The main limitation to greater use of CLC
with microelectrodes for EC detection is lack of commercial instrumentation that
would allow simple implementation of the principles outlined above.
ACKNOWLEDGMENTS
Our work in this area was supported by NIH GM46960.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.

Refshauge, C.; Kissinger, P.T.; Dreiling, R.; Blank, L.; Freeman, R.; Adams, R.N.
Life Sci., 1974, 14(2), 311322.
Kennedy, R.T.; Oates, M.D., Cooper, B.R.; Nickerson, B.; Jorgenson, J.W. Science,
1989, 246, 57.
Shen, H.; Witowski, S.; Boyd, B.W.; Kennedy, R.T. Anal. Chem. 1999, 71, 987994.
Boyd, B.W.; Witowski, S.R.; Kennedy, R.T. Anal. Chem, 2000; 72, 865871.
Jorgenson, J.W.; Guthrie, E.J. J. Chromatogr. 1983, 255, 33548.
Svec, F.; Peters, E.C.; Sykora, D.; Frechet, J.M.J. J. Chromatogr. 2000, 887, 329.
Karlsson, K.E.; Novotny, M. Anal. Chem. 1988, 60, 16621665.
Kennedy, R.T.; Jorgenson, J.W. Anal. Chem. 1989, 61, 11281135.
Kennedy, R.T.; German, I.; Thompson, J.E.; Witwoski, S. Chem. Rev. 1999, 99,
3081.
LaCourse, W.R. Anal. Chem. 2000, 72, 37R51R.
Cikalo, M.G.; Bartle, K.D.; Robson, M.M.; Myers, P.; Euerby, M.R. Analyst 1998,
123, 87R102R.
Knox, J.H.; Gilbert, M.T J. Chromatogr. 1979, 186, 405.
Slais, K.; Krejci, M. J. Chromatogr. 1982, 235(1), 2129.
dos, A.J.; Van Dyck, M.M.C.; Poppe, H.; Kok, W.T. Chromatographia 1993, 37,
7985.
Manz, A.; Simon, W. J. Chromatogr. Sci., 1983, 21, 326330.
Knecht, L.A.; Guthrie, E.J.; Jorgenson, J.W. Anal. Chem. 1984, 56, 479482.
Ewing, A.G.; Mesaros, J.M.; Gavin, P.F. Anal. Chem. 1994, 66, 527A537A.
St. Claire, R.L.; Jorgenson, J.W. J. Chromatogr. Sci. 1985, 23, 186191.
LaCourse, W.R.; Owens, G.S. Anal. Chim. Acta. 1995, 307(23), 301319.
White, J.G.; St. Claire, R.L.; Jorgenson, J.W. Anal. Chem., 1986, 58, 293298.
Howell, J.O.; Wightman, R.M. Anal. Chem., 1984, 56, 542529.
Singhal, P.; Kawagoe, K.T.; Christain, C.N.; Kuhr, W.G. Anal. Chem., 1997, 69,
16621668.
Singhal, P.; Kuhr, W.G. Anal. Chem., 1997, 69, 48284832.

salient comp dept. peach typesetting dekker 150644 toth

520
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.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.

Boyd and Kennedy


Oates, M.D.; Jorgenson, J.W. Anal. Chem. 1989, 61, 19771980.
Allison, L.A., Mayer, G.S.; Shoup, R.E. Anal. Chem. 1984, 56, 10891096.
Warner, A.M.; Weber, S.G. Anal. Chem. 1989, 61, 26642668.
Kennedy, R.T.; Jorgenson, J.W. Anal. Chem. 1989, 61, 441446.
Pihel, K.; Hsieh, S.C.; Jorgenson, J.W.; Wightman, R.M. Biochemistry 1998, 37,
10461052.
Oates, M.D.; Cooper, B.R.; Jorgenson, J.W. Anal. Chem. 1990, 62, 15731577.
Karger, B.L.; Martin, M.; Guiochon, G. Anal. Chem. 1974, 46, 16401647.
Vissers, J.P.C.; de Ru, A.H.; Ursem, M.; Chervet, J.P. J. Chromatogr. A, 1996, 746,
17.
Shen, H.; Lada, M.W.; Kennedy, R.T. J. Chromatogr. B, 1997, 704, 4352.
Tsai, H.; Weber, S.G. Anal. Chem. 1992, 64, 28972903.
Tsai, H.; Weber, S.G. J. Chromatogr. 1991, 542, 345350.
Tsai, H.; Weber, S.G. J. Chromatogr. 1990, 515: 451457.
Tsai, H.; Chen, J.G.; Woltman, S.J.; Weber, S.G. Anal. Chem. 1995, 67, 541551.
Chen, J.G.; Weber, S.G. Anal. Chem. 1995, 67, 35963604.
Chen, J.G.; Woltman, S.J.; Weber, S.G. J. Chromatogr. A 1995, 691, 301315.
Woltman S.J.; Alward, M.R.; Weber, S.G. Anal. Chem. 1995, 67, 541551.
Woltman, S.J.; Chen, J.G.; Weber, S.G.; Tolley, J.O. J. Pharm. Biomed. Anal. 1995,
14, 155164.
Moore, A.W.; Jorgenson, J.W. Anal. Chem. 1993, 65, 188196.
Maidment, N.T.; Siddall, B.J.; Rudolph, V.R.; Erdelyi, E.; Evans, C.J. Neuroscience
1991, 45, 8193.
Maidment, N.T.; Evans, C.J. In Microdialysis in the Neurosciences, Robinson, T.E.;
Justice, J.B., Eds., Elsevier: New York, 1991, pp. 275304.
Boyd, B.W.; Kennedy, R.T. Analyst 1998, 123, 21192124.
Robert, F.; Bert, L.; Denoroy, L.; Renaud, B. Anal. Chem. 1995, 67, 18381844.
Orwar, O.; Sandberg, M.; Jacobson, I.; Sundahl, M.; Folestad, S. Anal. Chem. 1994,
66, 44714482.
Rowley, H.L.; Martin, K.F.; Marsden, C.A. J. Neurosci. Meth. 1995, 57, 9399.
Arriaga, E.A.; Zhang, Y.; Dovichi, N.J. Anal. Chim. Acta 1995, 299, 319326.
Oates, M.D., Cooper, B.R., Joregenson, J.W. Anal. Chem. 1990, 62, 15731577.
http://web.mit.edu/biopolymers/www/amino acid analysis.html
http://www.abrf.org/ABRF/ResearchCommittees/aaaarticles/aaales/aaa.html
Smolders, I.; Sarre, S.; Michotte, Y.; Ebinger, G. J. Neurosci. Meth. 1995, 57, 4753.
Orwar, O.; Sandberg, M.; Jacobson, I.; Sundahl, M.; Folestad, S. Anal. Chem. 1994,
66, 44714482.
Bourdelais, A.; Kalivas, P.W. J. Neurosci. Meth. 1991, 39, 115121.
Phillips, Jr., J.B.; Cox, B.M. J. Neurosci. Meth. 1997, 77, 211220.
Orwar, O.; Folestad, S.; Einarsson, S.; Andin, P.; Sandberg J. Chromatogr. A 1991,
566, 3955.
Snyder, L.R.; Dolan, J.W.; Gant, J.R. J. Chromatogr. 1979, 165, 330.
Bousse, L.; Cohen, C.; Nikiforov, T.; Chow, A.; Kopf-Sill, A.R.; Dubrow, R.; Parce,
J.W. Annu. Rev. Bioph. Biom. 2000, 29, 155181.
Dolnik, V.; Liu, S.R.; Jovanovich, S. Electrophoresis 2000, 21, 4154.
Gavin, P.F.; Ewing, A.G. Anal. Chem. 1997, 69, 38383845.

salient comp dept. peach typesetting dekker 150644 toth

High Sensitivity Detection


61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.

521

Woolley, A.T.; Lao, K.Q.; Glazer, A.N.; Mathies, R.A. Anal. Chem. 1998, 70,
684688.
Wang, J.; Tian, B.M.; Sahlin, E. Anal. Chem. 1999, 71, 39013904.
Martin, R.S.; Gawron, A.J.; Lunte, S.M.; Henry, C.S. Anal. Chem. 2000, 72,
31963202.
Hilmi A..; Luong, J.H.T. Environ. Sci. Tech. 2000, 34, 30463050.
Wang, J.; Chatrathi, M.P.; Tian, B.M. Anal. Chim. Acta 2000, 416, 914.
Fenn, R.J.; Siggia, S.; Curran, D.J. Anal. Chem. 1978, 50, 1067.
Purdy, W.C.; Weber, S.G. Anal. Chem. 1982, 54, 1757.
Matsue, T.; Aoki, A.; Abe, T.; Uchida, I. Chem. Lett. 1989, 134.
McClintock, S.A.; Purdy, W.C. Anal. Chim. Acta. 1984, 166, 171.
Goto, M.; Zou, G.; Ishii, D. J. Chromatogr. 1984, 268, 157.
Goto, M.; Zou, G.; Ishii, D. J. Chromatogr. 1983, 275, 271.

salient comp dept. peach typesetting dekker 150644 toth

Index

Acetaminophen, interference in glucose


assay, 245246
Acetylcholine:
Ca release with, 305, 306
detection, 260
exocytosis with, 305
nicotine receptors, 282
sniffer-patch detection, 301, 302
Acetylcholine esterase, in CEEC, 468
Adenosine-5-triphosphate (ATP):
detection in PC12 cells, 313
sniffer-patch detection, 302
Adrenal tissue, epinephrine vesicles in,
287
Adsorption, of labeled oligonucleotides,
94
AFM:
morphology of DNA lms, 15
of surfactant layers, 208

Agonists, 257
DA receptors of, 268
Alkaline phosphatase:
4-aminophenol product of, 340
4-aminophenyl phosphate substrate
of, 340
ECIA label, 331, 338
phosphate hydrolysis with, 340
substrate of phenyl phosphate, 340
use in heterogeneous ISA, 338
Amalgamated gold (Hg/Au) electrode,
373, 375, 376
thiol removal, 381
Amphetamine, uptake by DAT, with
DA, 273
Amino acids:
CEEC analysis, 477
detection at gold oxide electrodes,
378

523

524
[Amino acids]
microdialysis sampling, 481
nonelectroactive, detection, 376
electrochemical detection in CLC,
499
photolysis, 385
4-Aminophenol:
electrochemistry, 340
product of alkaline phosphatase,
340
6-Aminoquinoyl-N-hydroxysuccinimidyl carbamate (6-AQC):
derivatives, 382383
peptide derivativization with, 382
Amperometric detection (see also Electrochemical detection):
amino acids nonelectroactive, 376
background in, 369, 393
with biuret reaction, 390
in catalysis, 369
with CE, 368, 372, 375, 391
at copper oxide, 377
at dual electrodes, 369, 374
with ow through electrodes, 392,
393
in integrated pulse amperometry
(IPAD), 378
at gold oxide, 378
with HPLC, 368, 372, 375, 377, 382,
383, 384, 386, 387, 388, 390,
392
oxidation with, 369, 370, 372, 374,
376, 378, 392, 393
with photolysis, 385
at platinum oxide, 378
in pulsed amperometric detection
(PAD), 378
selectivity, 368, 369
sensitivity, 369
signal-to-noise ratio, 368
thin-layer, 384
Amperometry:
constant potential, 280, 281
quantitative, 283

Index
Amphetamine:
effect on DA transport, 272
exocytosis with, 318
release with catecholamine, 315
Angiotensin II, 384
Anhydride, label attachment with, 332
Antagonists, 257
DA receptors of, 267
of GABA, 269
O2 consumption in, 267
Antibody immobilization:
in capillaries, 349, 350
in glycan chains, 350
with polyethylene glycol, 350
Anthraquinone, use as photooxidant, 9
Arsenic(III), detection with conducting
composite electrodes, 434
Ascorbate, in the brain, 258
ATP, detection in PC12 cells, 313
Atrazine, ISA analysis, 351
Avidin:
alkaline phosphatase conjugate, 88
binding of biotin-labeled antibody,
444
HRP conjugate, 57
Automation, in CLC, 511513
Autoreceptors, 268
antagonists of, 269
regulation of release, 268
Bead-based ISA (see also Paramagnetic
microbeads), 355
in RDE detection, 355
in SECM, 357
sensitivity of, 355
volumes in, 355
-cells, pancreatic:
exocytosis from, 282, 287
insulin from, 289
vesicular radii in, 287
Biocatalysis, of DNA modied electrodes, 8092
Bioreactors, electrochemical, 219
227

Index
Biosensors, amperometric congurations, 400402
Biotin/avidin, protein immobilization,
406
Bilayer lipid membranes, for monitoring
DNA hybridization, 36
Biuret reaction, 386
in amino acid composition, 389
in bradykinin detection, 390, 391
with capillary HPLC, 391
in CE, 391
conditions, 390
Cu(II) detection, 389
in dialysis, 391
with HPLC, 386, 390
on-column, 391
in plasma samples, 391
of proglutamyl peptides, 390
in solid-phase extraction, 391
in substance P, 390
preconcentration derivatization, 504,
506
tyrosine in, 389
with vasopressin, 391
Bovine serum albumin (BSA):
in enzyme ISA, 340, 345
NSA with, 340, 345
as site blocker, 345, 346
Brain:
electrical stimulation of, 257
DA release in, 257, 264265
DA uptake in, 264, 265
microdialysis sampling, 478, 481
O2 consumption after, 266267
presynaptic effects in, 268
5-Bromo-4-chloro-3-indoylphosphate,
oxidative hydrolysis of, 56
Bugbead assays, 354
detection in, 354, 355
dimensions, 354
guinea pig IFF in, 354
minimum detected concentration, 355
mouse IgG in, 354
neutavidine-dendrimer in, 354

525
[Bugbead assays]
paramagnetic bead detection, 354
RDE detection in, 352, 355
Caenorhabditis elegans, neurotransmission in, 310
Calcium:
channels, 305
detection in chromafn cells, 303
role in exocytosis, 305
release with acetylcholine, 305, 306
Capillary electrophoresis (CE):
in biological analysis, 461, 476
capillaries for, 462
of cysteine, with peptides, 375
of cystine, with peptides, 375
detectors (see also CEEC), 461, 462
electroosmotic ow in, 464
electrochemical detection (CEEC),
462
end-column, 462, 464
off-column, 462, 463
micellar electrokinetic chromatography (MEKC), 463
microchip, 461, 462, 474
neutral analytes, 463
pH in, 461
speed of, 463
voltage in, 462, 463
Capillary enzyme ISA, 348, 349
antibody immobilization in, 349
of atrazine, 351
competitive, 349
of digoxin, 350
of indole-3-acetic acid, 350
poly(vinylbenzyl chloride) treatment
with, 349
sandwich, 349
sensitivity, 351
Capillary liquid chromatography (CLC):
automation, 511
columns, 492, 511
derivatization with, 492, 497, 508,
509

526
[Capillary liquid chromatography]
dialysis with, 519
electrochemical detection, 492
of amino acids, 499
of -amino butyric acid (GABA),
499
background, 496, 507, 508
detection limits, 496, 497, 500,
516
of DNA, 497
end-column, 493
microfabrication in, 492, 513,
515
with naphthalene-2,3-dicarboxaldehyde (NDA), 497, 499,
508
of neurotransmitters, 519
of nonelectroactive compounds,
497
on-column, 493, 496
with o-phthalaldehyde (OPA),
497, 499, 507, 508
of proteins, 499
pulsed amperometric (IPAD),
496
recycling, 492
selectivity, 496
sensitivity, 492, 496, 500
of single cells, 500
of thiols, 496, 497
voltammetry, 496, 500
ow rate, 493
microdialysis, 499, 519
preconcentration with, 492, 500, 509
sensitivity, 492, 493, 496, 500
volumes in, 511, 518
Carbon ber electrodes:
amino acids, detection, 372, 391
CEEC detector, 466
cylinder, 261
in vivo, 259
peptide detection, 372, 391
pH response, 259
selectivity, 259
Carbon monoxide, inhibition of NiFe
hydrogenase, 188

Index
3-(4-Carboxybenzoyl)-quinoline-2carboxaldehyde (CBQCA),
508
Carnosine, OPA derivative, 382
Catecholamine,
concentration in PC12 vesicles, 290
release from PC12 cells, 382
Catalase, effect on immunoassay sensor,
448449
CEEC:
of biological uids, 476477
capillaries, 469
cell holders, 469
dual-electrodes for, 471
electrodes in, 466470
efciency, 470
end-column, 465
macrocyclic complexes for, 468
off-column, 463464
pulsed amperometric detection
(PAD), 468, 471, 476
of thiols, 468
voltammetry in, 468
volumes, 469
Chemical noise, in LC/EC, 507, 508
4-Chloro-1-naphthol,
HRP stimulated oxidation of, 76
mediated oxidation of, 56
Choline, detection of, 260
Choline oxidase, component of
immunoassay sensor, 444
445
Chromafn cells:
adrenal bovine medullary, 282
Ca detection in, 303
exocytosis in, 282, 287
serotonin in, 288
vesicles, 287
Chronoabsorptometry, of myoglobin,
115
Chronocoulometry:
detection of mismatched duplexes,
21
of DNA redox markers, 5051
of intercalated methylene blue, 16
of redox-labeled DNA, 85

Index
Chronopotentiometry, of DNA assemblies, 7881
Circular dichroism:
of cytochrome c, 121
of iron sulfur clusters, 159
Co(bpy)33+, as redox indicator for dsDNA, 33
Cocaine, DA uptake inhibition, 271
Competitive ISA, 341
for (1-acid glycoprotein (orosomucoid), 341
in capillaries, 349
for digoxin, 341
incubation time in, 342
Conducting composite electrodes (CCE),
418, 432, 434
with graphite powder, 434
with Rh2POM, 432
with RuDen , 434, 435
with RuO2, 427
Conductivity, electrical, measurement in
DNA, 10
Copper oxide electrodes, 374, 377, 423,
424
Cottrell equation, 50
Cryosolvent, use in protein lm voltammetry, 170
Cudate-putamen(CP), 257
DA uptake in, 271
DAT in, 271
Cyanodenz[f]isoindole (CBI), peptide
derivatization, 380
Cyanometmyoglobin, 123
Cysteic acid, elimination of chemical
noise in LC/EC, 508
Cysteine, 372
CEEC analysis, 477
from cysteic acid (cysteine sulfonic
acid), 372
from cystine, 372
electrode reactions, 372
oxidation, 372374
with peptides, 374376
Cystine, 374
CEEC analysis, 477
with peptides, 374376

527
[Cystine]
reduction, 374
with mvRuOx, 420
oxidation, 374
Cytochrome c:
comparison of tuna and horse heart
voltammetry, 137
controlled potential amperometry,
133
cyclic voltabsorptometry, 133
derivative cyclic voltabsorptometry,
116
electron transfer kinetics, 115116
ow injection analysis, 134
importance of protein purity, 117
mediation of cytochrome c oxidase
titration, 124127
promoted electrode response, 54
SERS, 121
Cytochrome c oxidase:
biphasic kinetics, 135
coulometric titration, 124127
cyclic voltammetry in bilayer membrane, 132
electron microscopy, 127
topology in surface lms, 137
Cytochrome c peroxidase, voltammetry,
239240
Cytochrome P450, thin lm voltammetry, 208209
Daunomycin, use as redox-active intercalator, 16, 33
7-Deazaguanine, photooxidation, 7
Demetallation, of iron-sulfur clusters,
178181
Dendrimers, in enzyme electrodes, 411
Diabetes, and glucose detection in
blood, 442
Didodecyldimethylammonium
bromide, as electrode modier,
198
Differential pulse anodic stripping
voltammetry (DPASV):
detection in ECIA, 332
detection limits, 332

528
Digoxin:
capillary enzyme ISA, 350
detection, 341, 342, 347
detection time, 344
FIAEC ECIA LOD, 344
LCEC ECIA LOD, 343, 347
radioimmunoassay, 347
in serum samples, 347
2,4-Dinitroestriol (DNE), 330, 332
3,6-Dinitrophthalic anhydride (DNPT),
peptide derivatives, 383, 384
Dissociation constants, iron-sulfur proteins, 177
DNA:
biosensors, mediated electron transfer, 12
biosensors, regeneration strategies,
32
bulk conductivity of dry, 10
differentiation of mutants, 92, 9698
electrochemical detection in CLC,
497
electrochemical detection of point
mutations in, 18
electrode surface coverage, 1315
enzyme amplied detection of, 55
hybridization:
amperometric detection, 449457
temperature effects, 455456
immobilization schemes, 30
polyanionic lms of calf thymus,
219
potential dependence of monolayer
thickness, 15
quartz crystal microgravimetry,
7376
self-assembled monolayers, 13
surface hybridization, 3132
surface morphology, 15
DNA polymerase, surface reaction,
8492
Dopamine(DA):
agonist as, 319
antagonists, 267
brain slices in, 257
calibration plots of, 291

Index
[Dopamine(DA)]
concentration, mathematical model,
298
detection selectivity of, 258
exocytosis of, 256
FSCV in PC12 cells, 291
neurotransmitter, 256
nomifensine, DA transporter inhibition, 315
PC12 cells vesicles in, 290
receptors, classes of, 268
release, 257, 262263
recovery, 269270
with ouabain, 272
transport, 271273
reverse, 272, 273
unidirectional, 273, 274
uptake, 257, 262, 263
in brain tissue, 264
inhibitors, 271
inhibition kinetics, 271
kinetics, 264265, 271
ouabain with, 272
in synaptosomes, 263
vesicular concentration, 293
L-DOPA with, 293
Dopamine transporter (DAT), 257
CP in, 271
DA uptake by, 272
DA transport regulation in, 273274
DOPAC, 258
Drosophila elegans, neurotransmission
in, 310
Eastman AQ, use for protein lm
voltammetry, 203, 212
Electrocatalysis, at DNA-modied electrodes, 21
Electrochemical detectors (see also LCEC):
amperometric, 496, 500
carbon ber, 496, 501
for CLC, 493
copper electrodes, 497
detection limits, 500
end-column, 493

Index
[Electrochemical detectors]
at xed potential, 496
gold electrodes, 496
mass sensitivity, 496, 500
microelectrodes for, 496
on-column, 493, 496
potentiometric, 493
sensitivity, 500
volume of, 492, 493
wall-jet, 493
Electrochemical immunoassay (ECIA),
32
amperometric detection, 335
applications, 331
conductometric detection, 335
detection limits, 360
differential pulse anodic stripping
voltammetry (DPASV) detection, 332
in ow-injection analysis (FIA), 335
heterogeneous, 332
hydrodynamic voltammetric detection, 337
with LC/EC, 344
potentiometric detection, 325
rotating disk electrode (RDE) detection, 335
scanning electrochemical
microscopy (SECM) detection,
335
stripping voltammetry, 332
labels for, 329333
in lab-on-a-chip, 358, 363
limit of detection (LOD), 331
in liquid chromatography (LC), 335
in microcapillaries, 348
miniaturized enzyme, 348
Electrochemiluminescence (ECL), 393
Electrodes, surface modied, 418
Electron microscopy, 297
Electron transfer:
between DNA bases, 7
distance dependence through the
DNA helix, 11
history of long-range in DNA, 2
microscopic model for protein, 150

529
[Electron transfer]
photoinduced, 7
photophysics in DNA, 3
proton coupled, 165167
Enzyme electrodes:
dendrimetric, 411
immobilization strategies, 406407
requirements, 403404
wired, 407
Enzyme labels, as immunosensors, 34
Enzyme-substrate (E-S) pairs, 331, 339
AP-PAP, 360
with ECIA, 338
in lab-on-a-chip, 360
in mesoscale immunosensor, 360
mouse IgG for, 356
PAP-PAPP, 355
Enzyme transistor, 245
Epinephrine, in vesicles, 287, 288
ESR spectroscopy, myoglobin lms,
207
Estriol, 330
Estrogens, 330
Ethanol-2-thiolate, binding to [Fe3Fe4S] clusters, 181183
Ethidium:
biphasic kinetics of Et*, 7
photochemistry, of intercalated, 3
Ethylene dibromide, electrocatalytic
reduction, 222
Exocytosis:
with acetylcholine, 305
amphetamine with, 315
BoNT/C1 transfected cells with, 313
Ca in, 305
of chromafn cells, 282, 287
cell differentiation after, 306, 309
FSCV of, 291
by histamine, 303
latency of, 305,306
mathematical model, 297
mechanism, 305, 306
SNARE hypothesis, 310
Faradaic impedance
of DNA surface layers, 78, 8789

530
[Faradaic impedance]
detection of functionalized liposomes, 7680
detection of insoluble product formation, 5960
detection of vesicular stomatitis
virus, 90
of dendritic interfaces, 73
Fast scan cyclic voltammetry (FSCV),
280
background subtracted, 258259
catecholamines of, 290, 291
concentrations with, 281
DA release with, 262
DA uptake with, 271
O2 consumption with, 266267
selectivity of, 281, 290291
Ferricyanide:
catalysis of chemical reduction, 21
use as redox probe, 60
Ferrocenecarboxylic acid chloride
(FAC) derivatives, 383
detection limits, 383
electrochemistry, 383
ferrocene carboxylic acid from, 383
Ferrocenyl naphthalene diimide, as
DNA redox indicator, 33
Ferrocyanide, mediator with mvRuOx,
425
Ferredoxin:
A.v. FdI voltammetry, 154
comparison of voltammetries,
156157
[3Fe-4S]/[M3Fe-4S] interconversion,
172
voltammetry in polymyxin lms,
157
Ferredoxin-NADP reductase, 110
Flow injection analysis (FIA), electrochemical detection, 335, 336
apparatus, 336
detection time, 344, 356
of digoxin, 344
ECIA detection in, 344
limit of detection (LOD), 337
PAD detection, 418

Index
[Flow injection analysis]
at Rh2POM, 432
thin-layer ow cell for, 336
Fractal model, of hydrogel
immunoassay sensor, 447
448
Fumarate reductase, voltammetry,
186188
Fumarate, electrocatalytic reduction,
186187
GABA, 269
electrochemical detection in CLC,
499
sniffer-patch detection, 302
Glucose oxidase electrode, 440
Glucose-6-phosphate dehydrogenase
(G6PDH), 338
with homogeneous ISA, 347
Glutamate, 269
detection, 261
microdialysis sampling, 481
sniffer-patch detection, 302
Glutaraldehyde, antibody immobilization in ISA, 349
Glutathione:
CEEC analysis, 468
detection, 375, 376
derivatives, 382
with mvRuOx, 424, 427
OPA derivative, 382
with Prussian blue, 436
Glycosylation, and electron transfer of
heme proteins, 236
Gold nanoparticles, use for detection of
DNA, 7376
Gold oxide electrodes, 378, 380
Guanine, DNA-medated photooxidation,
8
Hemoglobin, surfactant lm voltammetry, 209219
Heparin, 423, 424
Heterogeneous ISA:
antibody immobilization in, 349
in capillaries, 349

Index
[Heterogeneous ISA]
enzyme, 347
with paramagnetic beads, 351
sensitivity, 349
Histamine:
Ca release with, 303
capillary chromatography of, 288
exocytosis by, 303
LCEC, 288
in mast cells, 288
Hoecht 33258, as DNA redox indicator,
33
Homogeneous ISA, 347
with LCEC, 336
Homovanillic acid, 258
Horseradish peroxidase:
antibody label, 445
calculation of electron tunneling rate,
245246
discussion of Eo values, 238
electrocatalysis by histidine mutants,
243244
as oligonucleotide label, 62
use in ds-DNA indicator, 35
voltammetry, 238
HPLC:
amino acid detection, 368, 372, 375,
377, 382384, 386388, 390,
392
copper oxide electrode detector, 377
derivatization in, 504
electrochemical detection, 491, 492,
504
microbore, 504
peptide detection, 386, 387, 388, 390,
504
photoluminescence detection, 392,
393
Rh2POM, 432
Human serum albumin (HAS):
diethylenetriaminepentaacetic
(DTPA) label, 332
dual ISA, 333
for heterogeneous assay, 332
metal labels for, 332
microbiuret ISA of, 333

531
Hydrogel layers, in wired enzyme electrodes, 54
Hydrogen peroxide:
CEEC electrode, 468
electrocatalyzed reduction, 9396,
453
8-Hydroxydeoxyguanosine, CEEC
detection, 477
IgG:
detection limits, 346, 350
ECIA, 344, 352
E-S pair, 356
human, dual ISA, 333
mouse, 352, 356, 360
rabbit, 349, 350
SECM detection, 356
Immunoassay (ISA), 330, 331, 340,
347349
enzyme, 340, 341, 348, 349
for hapten, 330
with LCEC, 336
microbiuret based, 333
nonspecic adsorption (NSA) in, 344,
345
Indole-3-acetic acid, ISA analysis, 350
Insulin:
detection with conducting composite
electrodes, 435
with mvRuOx, 420, 422, 423
release from pancreatic -cells, 289
in vesicles, 290
Integrated pulse amperometry (IPAD):
amino acid detection, 378
CLC detector, 496
at platinum oxide electrodes, 378
Iridium oxide, 436
Isoindole, OPA derivative, 381
Jeffamine, 411
Kynurenine, microdialysis sampling,
481
Lab-on-a-chip (see also Paramagnetic
microbeads), 357

532
[Lab-on-a-chip]
ECIA in, 348, 357
in meso-scale immunosensor, 358,
360
microfabricated components, 358
micro-total analysis system, 357
sample size, 358
volume in, 360
Langmuir binding:
of DNA duplexes, 454
of enzyme-labeled antibodies, 447
Laser interference ablation, 410
L-DOPA (L-3,4-dihydroxyphenylalanine):
DA concentration with, 293
DA precursor of, 293
determination, 318
exocytosis with, 318
microdialysis sampling, 490
product of tyrosine hydroxylase, 315
vesicle size with, 293
Ligand interchange, of iron-sulfur proteins, 183
Lipid complexes, of aligned DNA lms,
10
Lipid membranes, for enzyme support,
127129
Liposomes, HRP-functionalized amplication probes, 7680
Liquid chromatography electrochemical
detection (LC-EC), 258, 336
capillary, 280
column switching, 347
detection time, 344
of digoxin, 341
ECIA with, 344
glassy carbon electrode in, 347
hydrodynamic voltammetry in, 337
homogeneous ISA with, 336
limit of detection (LOD), 336337
microcolumn, 280
thin-layer ow cell for, 336
Liquid crystal, detection of phase transitions, 199200, 210
Luminescence (see also Photoluminescence), 392

Index
Mast cells, 287, 288, 289
Methionine:
detection with conducting composite
electrodes, 435
with mvRuOx, 419, 424
-Mercaptoethanol (ME), with OPA,
381
Mercuric acetate, estriol derivative, 330
Metallation:
of [3Fe-4S] by Fe(II), 172
of iron-sulfur clusters, 170178
Metallodendrimers, 434
catalysis with RuDen, 434
Cu(II) in, 434
pentaerythriol in, 434
polyamidoamine (PAMAM) in, 434
RuIIterpyridine in, 434
sol-gel formation, 434
Methylene blue, reduction of intercalated, 15
Methyl viologen, 112
Michaelis-Menten kinetics, 272274
DA release, with ouabain, 272
DA uptake, 264265
DA uptake inhibition, 271
unidirectional DA transport, 274
Microchip, 474476
CE in, 474
ceramics (LTCC) for, 474, 475
dimensions, 474
DNA detection, 39
photolithography for, 474
plastics for, 474
poly(dimethylsiloxane) (PDMS), 474,
476
separations, 461, 462, 474, 476
voltage in, 474
volumes in, 474
Microdialysis sampling, 478, 479
amino acids, 481
aspartate, 481
of bile, 479
of blood, 479
of brain, 478, 481
with CE, 481, 482, 483
detection, 481, 483

Index
[Microdialysis sampling]
detection limits, 483
l-DOPA, 490
enantiomeric drugs, 482
ow rates, 478, 480, 481, 483
glutamate, 481
isoproterenol, 482
kynurenine, 481
in membrane, 478
with naphthalene-2,3-dicarboxylaldehyde (NDA), 481
of neurotransmitters, 481
with preconcentration in CLC, 501
in rat, 483
recovery in, 478
in tissue, 478, 479
of tryptophan, 481
Microelectrodes:
congurations for enzyme electrodes,
404406
single cell measurements, 280, 281
Microfabrication:
in CLC, 492, 514, 515
of electrochemical detectors,
515517
with glass, 517
in HPLC/EC, 515
with silicon, 517
Microuidics, with EC detection, 515
Muscarine acetylcholine receptors,
305
Ca release in, 305
exocytosis in, 305
mvRuOx, 418, 419, 420
applications, 420, 427
composite stability, 427
detection limits with, 421
disuldes with, 421
electropolymerization of, 427
ethanol with, 423
FIA with, 419, 423
glutathione with, 424, 427
heparin with, 423, 424
hexadecacylcetyl trimethylammonium chloride (HTAC) with,
427

533
[mvRuOx]
insulin with, 420, 422, 423
isotocin sensitivity, 427
methionine with, 419, 424
oxygen with, 424
passivation, 424
Myoglobin:
chronoabsorptometry, 115
dispersion of Eo values, 203
electrocatalysis, 220227
electron transfer kinetics, 115, 123
multilayer lms, role of surface
roughness, 218
spectroelectrochemistry, 207
voltammetry:
in composite lms, 210
in presence of cyanide, 124
role of protonation, 204
in surfactant lms, 198207
NADH detection, 338, 340, 347
with 2,6-dichloroindophenol (DCIP),
340, 347
interference in, 340
phenytoin determination of, 347
Naon:
for casting protein lms, 211
coated microelectrode, 258
Nanoelectrodes, electrical transport
through DNA bridge, 11
Nanoparticles, preparation of multilayer
lms, 219
Naphthalene-2,3-dicarboxaldehyde
(NDA), 380, 381
doubly derivatized peptides with,
381
in CLC, 497, 499, 508
enkephalins with, 381
with microdialysis sampling, 481
peptide derivatives, 380, 381
Neomycin, role in ferredoxin voltammetry, 154
Nerve growth factor, 306
Neurites, from PC12 cells, 306
Neuron, 255
DA of, 256

534
Neuronal communication, 279280
exocytosis in, 280
presynaptic intracellular vesicle in,
280
Neuropeptides, preconcentration in
CLC, 504
Neurotransmitters, 256, 260, 280 (see
also Acetylcholine, Adenosine5-triphosphate, choline, DA,
GABA, Glutamate, 5-Hydroxytryptamine, Insulin, Norepinephrine, Serotonin)
concentration mathematical model of,
297
nonelectroactive sniffer-patch detection of, 301
release amount of, 283
sniffer-patch calibration, 303
Nicotine:
acetylcholine receptors, 282, 305
exocytosis with, 305
microdialysis sampling, 483
perfusion in PC12 cells, 282
NiFe hydrogenase, voltammetry under
H2, 188
Nitrite, electrocatalytic reduction,
223
Nitrogen oxide, electrocatalytic reduction, 223
p-Nitrophenyl-2,5-dihydroxyphenylacetate bis-tetrahydropyranyl
ether (NDTE), 384
angiotensin II derivatization, 384
derivatization with, 384
homogenistic acid from, 384
peptide derivatives, 384
Nonspecic surface adsorption (NSA),
344, 345
amine group complexation in, 349
with BSA, 345, 346
ion pairing in, 349
with Tween 20, 345
Norepinephrine:
detection, 260
FSCV of, 281
PC12 cells vesicles in, 290

Index
Nucleus accumbens (NAc), 257
DA concentrations in, 271
DA uptake in, 271
Octadecyl mercaptan, cytochrome oxidase containing SAMs,
129132
Oligothymine thiophosphate, 51
Oligreen, 66
Organohalides, reductive dehalogenation, 222
Ouabain, Na, K-ATPase inhibitor of,
272
Oxygen, O2:
consumption, 266
determination, 266267
FSCV of, 266
with mvRuOx, 424
Paramagnetic microbeads:
Bugbead detection, 352
diameter, 351
with ECIA, 348, 351
with heterogeneous ISA, 351
for IgG detection, 352
in lab-on-a-chip system, 351
in ovalbumin detection, 352
in polystyrene cuvettes, 352
with RDE, 352
with sandwich ISA, 352
PCR amplication, amperometric monitoring, 35
Peptide nucleic acid, as DNA probe,
32
Peptides, derivatives, 380
with DNPT, 383
with FAC, 383
with NDA, 381
with NDTE, 384
with OPA, 381
of primary amines, 380, 381, 382,
383
separation, 380
stability, 383
Peptides, derivatization, 380
on-column, 380
photolytic, 385

Index
[Peptides]
post-column, 380
pre-column, 380
reagents for, 380
specicity, 380
Peptides, detection:
amino acids of, 370
arginine in, 381
6-AQC of, 382
biuret reaction, 386
with carbon ber electrodes, 372,
391
in CE, 372, 391
coatings for, 372
at copper oxide, 376
Cu(II)-tartrate complexes in, 386
derivatives, 380, 381
detection limits, 372, 376, 377, 380,
382, 383, 384, 390, 391
disulde bridges, 376
DNPT derivatives, 383
at dual electrodes, 374
electrode materials, 370, 371
FAC derivatives, 383
at gold oxide, 376
in HPLC, 372
in vivo, 372
of metabolites, 372
at metal oxide electrodes, 376
NDA derivatives, 381
with NDTE derivatives, 384
OPA derivatives, 381
oxidation in, 369, 370, 372, 374, 376,
378, 392, 393
photolytic, 385
at platinum oxide, 376
potentials in, 371, 375, 381
of reduction, 369
with Ru(bpy)2+
3 , 392, 393
with tryptophan, 370, 371, 381, 392
of underderivatized, 369
Peroxidases:
electron transfer rate constants,
241242
glycosylated, voltammetry of, 241
use in sensors, 244245

535
Phenytoin:
detection in serum, 347
LCEC ECIA of, 347
Pheochromocytoma cells (PC12),
313319
ATP detection in, 313
autoreceptor activation in, 318
BoNT/C1 in, 313
catecholamine release from, 282
with L-DOPA, 318
DA in, 290
differentiation, 306
enhanced green uorescent protein
(EGFP) in, 313
exocytosis in, 282
amphetamine with, 315
agonist with, 318
autoreceptor activation with
amphetamine, 315
catecholamine release with
amphetamine, 315
after differentiation, 306, 309
FSCV of, 291
mechanism with amphetamine, 315
nomifensine with, 315
quinpirole with, 318319
SNARE hypothesis of, 310
sulpiride with, 318
transfection after, 313
K perfusion of, 282
L-DOPA, DA release with, 315
as neuronal model, 282
NGF (nerve growth factor) treatment
of, 306
nicotine acetylcholine receptors in,
282
nicotine perfusion of, 282
norepinephrine in, 290
nomifensine with, 315
protein syntaxin1 in, 310
rat adrenal, 282
receptors in, 282
SNAP-25 in, 310
transfected, with botulinium neurotoxin C1 light chain
(BoNT/C1), 310

536
[Pheochromocytoma cells]
varicosities in, 306
vesicular radii in, 287, 290
Phosphatidylcholine, as electrode modier, 196, 210
Photoluminescence, with (2,2bipyridyl)ruthenium(III/II),
392, 393
Photolysis, 385
o-Phthalaldehyde (OPA), 381
and chemical noise in LCEC, 508
-mercaptoethanol ( (ME)) with,
381
dipeptide derivatives, 381, 382
glutathione derivative, 382
isoindole derivative, 381
peptide derivatives, 382
preconcentration in CLC, 507
primary amines with, 381
thiols with, 381
tripeptides (GGPs) derivatives, 381
Platinum oxide electrodes, 378380
Polyaniline, use in enzyme transistor,
245
Poly(dimethylsiloxane) (PDMS),
microchip fabrication, 476
Polyethylene glycol (PEG), 350
antibody immobilization with, 350
glycan coupling with, 350
Polyethylene oxide, enzyme electrode
component, 442
Polymerization, electrochemically initiated, 113
Polymyxin, role in ferredoxin voltammetry, 149
Polynucleotide kinase, phosphorylation
of DNA surface layers, 82
Polyoxometalate (POM), 428
as electrocatalyst, 429
electrodes, 430
immobilization, 430
Keggin structure, 428, 430
oxidation with, 429
PMo12O403, 428, 429
with sol-gel, 430

Index
[Polyoxometalate]
transition metal substituted (TMSP),
428
Polypyrrole lms, as ds-DNA detectors,
36
Polystyrene, with ISA, 345
Postsynaptic effects, 268
Preconcentration, 500509
in CLC, 500, 503
derivatization, 503507
detection limits, 501, 506, 509
detectors, 501
stationary phase, 501
volume in, 501, 509
Presynaptic effects, 268
Proglutamyl peptides, biuret reaction of,
390
Protein mutants, voltammetry, 161
Proton transfer:
gating of protein electron transfer,
159167
role in ferredoxin voltammetry, 161
Prussian blue, 436
cysteine with, 436
glutathione with, 436
oxidation with, 435
use in enzyme electrodes, 234
Pulsed amperometric detection (PAD),
417
amino acid detection, 378
in CEEC, 468, 471, 476
with ow injection analysis (FIA), 418
at gold electrodes, 417
HPLC detector, 418
at platinum oxide, 378
Putidaredoxin, voltammetry, 210, 215
Pyridyl disulde, as electrode modier,
119
Quartz crystal microgravimetry:
of cytochrome c oxidase, 135
detection of insoluble products, 63
DNA sensing using gold nanoparticles, 7376
of functionalized liposome interfaces,
64

Index
[Quartz crystal microgravimetry:]
of multilayer lms, 216217
of viral DNA, 85
Rabbit IgG, redox hydrogel sensor,
445447
Randles circuit, 46
Recombinant HRP:
electrocatalytic reduction of H2O2,
244
use in biosensors, 245
Redox hydrogel, immunoassay sensor
component, 443444
Redox polymer, in wired enzyme
electrodes, 440, 451
Reserpine, vesicular monoamine
transporter inhibition, 319
Rh(phi)2bpy3+, role in photoinduced
electron transfer, 3
Rh2POM, 430
Rotating disk electrode (RDE)
detection:
in Bugbead assays, 352, 355
with paramagnetic beads, 352, 355
in sandwich immunoassay, 352,
355
Ru(bpy)32+, use as redox mediator 36
Ru(CN)3
6 , 418
As(III) catalysis, 418
immobilization, 418
in poly(4-vinylpyridine), 418
RuO2, 426, 427
Sandwich ISA, 341344
antigens in, 344
in capillaries, 349
cuvettes for, 344
detection limit, 344
ECIA, 344
heterogeneous, 344
in lab-on-a-chip, 360
microtiter wells for, 344
paramagnetic beads for, 352
selectivity, 341
sensitivity, 344
Sauerbrey equation, 49

537
Scanning electrochemical microscopy
(SECM):
of bead microdomains, 357
ECIA detection, 335
feedback mode, 356
generation-collection mode, 356
heterogeneous assay with, 356
microelectrode probe for, 356
in miniaturized ISA, 356
mouse IgG detection, 356
multi-analyte detection, 356
with sandwich ISA, 356
surface patterning, 409
two-dimensional scanning, 356
Scanning force microscopy, cytochrome c electrode surfaces,
137138
Screen-printed electrodes, 245
Self-assembled monolayers, of DNA
monolayers, 13
Serotonin(5-hydroxytryptamine):
capillary chromatography (LCEC)
of, 288
in chromafn cells, 288
detection, 260
in mast cells, 288289
SERS, cytochrome c, 121
Sniffer-patch detection, 301
of acetylcholine, 301, 302
of ATP, 302
calibration, 303
in discreet loci, 302
of GAGA, 302
in neurons, 302
of neurotransmitters, 301303
of nucleotides, 303
voltage-clamp in, 301
in whole-cell patch-clamp recording,
301
Sol-gel, 430
diffusion coefcients, 432, 434
as mesoporous silica, 430
microporous, 430
POMs in, 430
pore widths, 430, 433
processing, 418, 430

538
[Sol-gel]
structure, 430
TMSPs in, 430
Soy bean peroxidase:
as enzyme label, 451453
as surface catalyst, 93
Spinach ferredoxin:
coulometric titration, 112
cyclic voltammetry, 114, 210211
reduction potential, 112
Stern-Volmer analysis, of DNA mediated electron transfer, 8
Stripping voltammetry:
ECIA detection with, 332
simultaneous detection, 333
Styrene, epoxidation, 225
Sulfur groups, detection at metal oxide
electrodes, 378
Surface patterning of electrodes,
409410
Synapse, 280
Tay-Sachs disease, 100
Tetramethylbenzidine, mediated oxidation, 56
Tetramethyl orthosilicate (TMOS), in
sol-gels, 430
Theophylline, homogeneous ECIA, 348
in serum, 348
in whole blood, 347
Thymine dimers, long-range repair of, in
DNA, 10
Transition metal substituted polyoxometalate (TMSP), 428
with As(III), 430
catalytic strength, 430
as dopant, 430
immobilization, 430
in microporous sol-gel, 430
pH with, 430
Rh2(OAc)4 with, 430
size, 430
Transport kinetics, 271273
Trichloroacetic acid, electrocatalytic
reduction, 222
Triplet energy transfer, between DNA
intercalators, 5

Index
Trumpet plots, 164
Tryptophan, 371
electrode reactions, 371
detection, 370
electrode coatings for, 372
oxidative, 370, 393
microdialysis sampling, 481
oxidation, 370, 393
peptides with, 370, 372
photoluminescence, 393
photolysis, 385
preconcentration in CLC, 503
Tween 20:
in ISA, 340, 345
NSA with, 345, 346
Tyrosine, 371
biuret reaction in, 389, 392
detection, 371372
electrode reaction, 371
oxidation, 371, 392
peptides with, 369, 371, 372
photoluminescence, 393
photolysis, 385
preconcentration in CLC, 503
Ru(bpy)2+
3 with, 392
Urate, interference of glucose sensor,
443
Uric acid, 258
Vasopressin, biuret reaction with,
391
Vesicles, 256
associated protein (VAMP), 310
catecholamines in, 282, 290, 298,
301
core halo, 324
electron microscopy, 297
epinephrine in, 287288
insulin, 290
neuronal dimensions, 281
in PC12 cells, 290
radius of, 288, 290, 293, 297
reserpine, inhibition of vesicular
monamine transporter, 319
release, 283, 315
size distribution, 287, 297

Index
[Vesicles]
in transporters, 287, 318
uptake, 318
Vesicular monamine transporter
(VMAT1), 318
inhibition, 319
release modulation by, 321

539
Vesicular stomatitis virus, detection by
impedance spectroscopy, 90
Voltammetry, advantages for redox protein studies, 148149
X-ray diffraction, of myoglobin-bilayer
lms, 207

You might also like