Professional Documents
Culture Documents
Electroanalytical Methods
for Biological Materials
edited by
Anna Brajter-Toth
University of Florida
Gainesville, Florida
James Q. Chambers
University of Tennessee
Knoxville, Tennessee
TM
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
Contents
Preface
Contributors
Part I
iii
ix
Electrochemistry of DNA
1
27
43
109
143
vii
viii
Contents
195
233
255
279
Part IV Bioelectroanalysis
10. Milestones of Electrochemical Immunoassay at Cincinnati
C. Ajith Wijayawardhana, H. Brian Halsall, and
William R. Heineman
329
367
399
417
439
461
491
523
Contributors
Contributors
Contributors
xi
xii
Contributors
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
II.
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.
A.
1.
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]).
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.
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
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.
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
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].
10
2.
Kelley
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
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.
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
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.
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
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.
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
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.
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
C.
17
18
D.
Kelley
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.)
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.
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.
21
5.
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
23
6.
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.
25
26
Kelley
39.
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
Figure 1 Major processes involved in any biosensor system: analyte recognition, signal transduction, and readout.
28
Wang
29
II.
30
Wang
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-
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
33
trodes (similar to those use for self-testing of glucose), and hence obviate the need
for regeneration [13].
C.
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.
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.)
35
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
3.
36
Wang
(1)
Ru(bpy)3+3 + G B Ru(bpy)3+2 + G+
(2)
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.)
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
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
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.
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.
45
(1)
(2)
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.
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-
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)
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-
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).
IV.
53
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.
55
56
Willner et al.
Figure 6 Enzyme-amplied analysis of DNA in a redox hydrogel, (4), crosslinked with (5).
57
59
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
63
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
67
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
69
70
Willner et al.
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.
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.
73
74
Willner et al.
75
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.
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.
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.
A.
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)
83
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.
85
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.
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.
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.
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
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.
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.
93
(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
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.
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
(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.
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.
98
Willner et al.
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.
101
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
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
103
VII.
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.
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).
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
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
112
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.
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
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.
116
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
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
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
122
Direct Electrochemistry
123
Mb(II)(CN)
Mb(III)CN,
7 Mb(II) +
Mb(II)CN,
CN
(3)
(4)
124
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.
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
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.
128
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
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
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
Direct Electrochemistry
133
134
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
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
Direct Electrochemistry
137
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
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.
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.
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
142
74.
75.
76.
77.
78.
5
Voltammetric Investigations
of Iron-Sulfur Clusters in Proteins
Fraser A. Armstrong
Oxford University, Oxford, England
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
145
Cluster types
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
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
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
II.
A.
Why Voltammetry?
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.
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.
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.
151
152
Armstrong
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
153
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.
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
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
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
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
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.
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
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
163
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.
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)
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
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.
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 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.)
170
Armstrong
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.
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
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-
175
176
Armstrong
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.)
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
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
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.)
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.
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.
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
183
184
Armstrong
185
186
Armstrong
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
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.)
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.
191
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).
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
193
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
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.
197
198
increases the value of v while not changing aolc very much compared to a molecule with equivalent single chain length.
B.
Myoglobin
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
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-
201
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
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.)
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.
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
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
204
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
205
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
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
207
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
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.
209
210
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.
211
III.
POLYION-PROTEIN FILMS
A.
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
212
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
213
214
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
215
POLYCATIONS
poly(diallydimethylamine)
(PDDA)
POLYANIONS
216
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
(2)
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
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
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
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
221
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)
222
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-
223
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
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
225
226
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.
227
V.
228
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.
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.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
231
7
Electrochemistry of Peroxidases
Tautgirdas Ruzgas, Annika Lindgren, and Lo Gorton
Lund University, Lund, Sweden
I.
INTRODUCTION
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)
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.
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.
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.
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.
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.
Electrochemistry of Peroxidases
241
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
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.
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.
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
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.
Electrochemistry of Peroxidases
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
249
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
AT
256
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
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
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
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
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.
262
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
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.
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
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-
265
(1)
266
to remove the slow diffusion caused by Naon; however, this approach introduces noise artifacts.
IV.
A.
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].)
268
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
269
270
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.
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.
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.
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
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]).
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
274
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.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
275
276
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
277
278
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
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
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-
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
II.
TECHNIQUES
A.
283
(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
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
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].)
287
B.
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
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
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
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.
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-
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
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
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.
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-
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
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].
298
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-
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
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
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
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].)
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.
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
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.
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,
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
309
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
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.
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
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
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
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].)
D.
315
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
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].)
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
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
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
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
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
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
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.
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.
327
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.
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
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
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
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
Analyte
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
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+.
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.)
337
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.
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
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)
19
71
72
73
74
18
75
76
76
33
77
Reference
340
Wijayawardhana et al.
(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.
341
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.
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-
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.
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.
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.
D.
347
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.
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.
349
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
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.
352
Wijayawardhana et al.
Figure 13 Schematic of the overall sandwich immunoassay based on paramagnetic beads. (From Ref. 3.)
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
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.
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,
357
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.
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)
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
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.
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.
362
Wijayawardhana et al.
363
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.
365
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
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.
A.
Introduction
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.
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.
371
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)
(2)
373
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)
(4)
(5)
374
Sahlin et al.
(6)
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)
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.
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
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.
(9)
(10)
378
2.
Sahlin et al.
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)
(12)
RSSR 3 2RSads
(13)
(14)
RSR 3 RSRads
(15)
(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,
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.
381
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.
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
383
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.
384
Sahlin et al.
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].
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.
8.
Biuret Reaction
387
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-
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
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
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.
(17)
(18)
(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
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
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).
395
396
Sahlin et al.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
397
398
Sahlin et al.
138.
139.
140.
141.
142.
143.
144.
145.
146.
147.
148.
149.
150.
151.
152.
153.
154.
155.
156.
157.
158.
159.
160.
161.
162.
163.
164.
165.
166.
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
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
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
ELECTRODE CONFIGURATION
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
406
associated with disk microelectrodes is the small area available for the dual role
of enzyme immobilization and electron transfer.
IV.
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.
408
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
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.
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-
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
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.
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.
415
13
Electrocatalytic Determination
of Biochemical Compounds
James A. Cox and Long Cheng
Miami University, Oxford, Ohio
I.
INTRODUCTION
418
II.
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-
419
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
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).
421
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
423
424
425
426
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.
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
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.
(1)
430
[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
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
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.
433
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.
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.
METALLODENDRIMERS AS ELECTROCHEMICAL
OXIDATION CATALYSTS
435
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
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.
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.
438
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
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
439
440
Campbell et al.
(1)
(2)
2 Os2+ 3 2 Os3+ + 2e
(3)
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
443
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.
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)
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
447
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.
449
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.
451
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.
453
(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.
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
(8)
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.
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.
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.
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.
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
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
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.
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
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.
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
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.
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
Capillary Electrophoresis/Electrochemistry
479
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.
Capillary Electrophoresis/Electrochemistry
483
Capillary Electrophoresis/Electrochemistry
485
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
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.
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
491
492
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.
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].
C.
493
494
495
496
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
497
Derivatization
498
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.)
499
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
500
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.
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.
502
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
bCalculated
503
III.
504
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
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.
506
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.
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.
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].
508
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
509
510
511
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.
512
Current Amplier
Fused Silica
Needle
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
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.
514
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.
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.
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
516
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.
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.
518
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.
519
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.
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.
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.
Index
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