Electrochimica Acta 45 (2000) 2623 – 2645 www.elsevier.nl/locate/electacta
Recent developments in faradaic bioelectrochemistry Fraser A. Armstrong a,*, George S. Wilson b a
Inorganic Chemistry Laboratory, Uni6ersity of Oxford, South Parks Road, Oxford OX1 3QR, UK b Department of Chemistry, Uni6ersity of Kansas, Lawrence, KS 66045, USA
Papers received in Newcastle, 20 December 1999
Abstract Progress in the area of faradaic bioelectrochemistry over the last decade has been reviewed. Electrochemical and spectroscopic studies of the protein–electrode interface are discussed along with the use of electron transfer promoters, self-assembled monolayers (SAMs), and surfactant films. Voltammetric techniques are described that permit rapid, in-situ measurement of formal potentials while also providing kinetic resolution on the sub-millisecond time scale. Some enzymes, especially those of analytical interest, must be coupled or ‘wired’ to the electrode using mediators in order to achieve rapid electron transfer. Supramolecular structures designed to facilitate electron transfer and to immobilize co-factors are described. The use of electrochemically-based biosensors for in-vivo measurements is documented. © 2000 Elsevier Science Ltd. All rights reserved. Keywords: Protein voltammetry; Wired enzymes; In vivo electrochemistry; Chemically modified electrodes; Electron transfer kinetics
1. Introduction More than 200 years have passed since the death of Luigi Galvani, who could certainly be considered a father of bioelectrochemistry, so the field is hardly a new development. Because of the broad range of activities we have decided to focus only on faradaic bioelectrochemistry, meaning that important biological phenomena such as membrane potentials and currents resulting from ion transport across membranes will not be discussed. Instead, we focus on three areas that have been especially prominent in the 1990s. First we note the extensive use of voltammetric techniques to extract essential physicochemical data concerning the kinetics and energetics of protein redox reactions. Second, we consider fundamental and practical considerations in linking or ‘wiring’ otherwise electroinactive enzymes to electrodes so that they can efficiently transport elec* Corresponding author. Tel.: + 44-1865-270841; fax: +441865-272690. E-mail address:
[email protected] (F.A. Armstrong)
trons and finally, we document the development of biosensors to be used especially in monitoring events occurring about single cells, in tissue slices or in intact brain. The use of voltammetric methods to study redox proteins and their active sites has gained increasing attention. Following the early pioneering studies of the 1970s and 1980s, voltammetric methods have become routine tools for determining formal potentials of redox centers in small proteins. It is now well established that, with suitable electrodes, small electron carriers such as cytochromes, ferredoxins and blue Cu proteins give reversible, electrochemistry without the need for small mediators. Progress is being made with increasingly complex systems, including multi-centered enzymes, and it is becoming acknowledged that the electrochemical response of active sites provides a useful diagnostic signal, in many ways analogous to that provided by various spectroscopic methods. The electrode replaces physiological redox partners, providing instead a driving force that is variable in the potential and time domains to energize reactions and a sensor to measure the response.
0013-4686/00/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 3 - 4 6 8 6 ( 0 0 ) 0 0 3 4 2 - X
2624
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
2. The protein–electrode interface Little more than 20 years ago, the idea that dynamic electrochemical methods could be applied to protein molecules as large as cytochrome c (let alone much larger enzymes) was regarded with scepticism. The notable pioneers of this era were the groups led by Niki (voltammetry of cytochrome c3 on mercury) [1], Hill (Au modified with a monolayer of organic ‘promoter’) [2] and Kuwana (cytochrome c at a metal oxide electrode) [3]. In each case it was established that important information on a protein’s redox properties could be obtained from experiments in which direct electron exchange took place between the electrode and the active site. Since that time, many important developments have taken place in the design and structural characterization of electrode surfaces that are active with proteins [4]. The misleading dogma that protein adsorption always presents an undesirable problem has given way to more constructive ideas in which the adsorbed state actually provides the key to studying and understanding these complex systems, many of which are intimately associated with interfaces (e.g. membranes) in their natural environment. Accordingly, efforts have been made to achieve strong adsorption of proteins on electrodes modified in such a way as to preserve native properties or provide a well-defined environment in which electron-transfer or spectroscopic
features can be probed. The various types of protein – electrode interface currently under study are depicted in Fig. 1. Interactions between protein and electrode may be tailored to be weak or strong. Weak interactions might ideally give rise to diffusion controlled voltammetry of solution species, whereas with strong interactions the experiment may address just a stable protein film, ideally a monolayer. Following the pioneering work by the groups of Hill and Taniguchi on the modification of gold electrodes with organic adsorbates that yield diffusion-controlled electrochemistry of cytochrome c [5,6], further efforts have been made to understand how these ‘promoters’ work. In particular, do they provide transient interaction sites for the protein or do they prevent blocking by contaminants? It appears that both these factors are important. Lateral interactions among the adsorbate molecules are important and lead to self assembled monolayers (SAMs), the X-functionality (nowadays usually a thiol) being bound to the metal (as thiolate) and the Y-functionality interacting with the solution and protein molecule. Selectivity of modified electrodes for particular proteins can be tuned by changing the functionality Y; thus whereas cytochrome c prefers acidic groups such as COO(H) which can interact more favorably with the lysine-dominated region around the exposed heme edge, acidic proteins such as plastocyanin or ferredoxins respond best if Y is
Fig. 1. Cartoon depicting the various types of electrode surfaces for which protein voltammetry is commonly observed. Shown are a metal electrode modified with an ‘XY’ SAM, a metal oxide electrode, a pyrolytic graphite ‘edge’ electrode often used in conjunction with mobile co-adsorbates such as aminocyclitols, and an electrode modified with a surfactant layer in which protein molecules are embedded.
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
a basic group such as NH2 (H+) [7,8]. Although it has often been assumed that the adsorbate itself is stable, recent work has shown that 4-mercaptopyridine (or bis(4-pyridyl) disulfide, which adsorbs as 4-mercaptopyridine decomposes slowly at gold to give adsorbed elemental sulfur, thereby inactivating the surface for protein interaction [9]. The loss of activity gives changes in waveform similar to those expected on the basis of the ‘microscopic model’, as active areas of the electrode become progressively blocked [10]. The microscopic model accounts for the rounded and drawn-out shapes of voltammograms attributed to ‘irreversible’ electrochemistry, in terms of electron transfer that is in fact reversible but occurs only at isolated active zones on the electrode, i.e. those sites that are suitably functionalized to interact with the protein molecule. It has found fairly wide acceptance, with support stemming from studies on electrodes coated with lipids and presenting only ‘pinholes’ to allow interaction with proteins such as cytochrome c [11]. The effect of blocking protein interaction sites has also been examined using mixed SAMs comprising active ‘X–Y’ adsorbates diluted with non-active co-adsorbates that lack the appropriate ‘Y’-functionality [12]. With solutions of cytochrome c, the classical, peak shaped voltammograms which are exhibited at gold electrodes modified with just 3-mercaptopropionic acid or bis(4pyridyl) disulfide (which is bound as 4-mercaptopyridine) become increasingly sigmoidal as n-alkylthiols are added to the monolayer. Interestingly, with 3-mercaptoethanoic acid, the effect is much more marked as n is increased from 3 to 18, whereas with bis(4-pyridyl) disulfide it is the shorter chain n-alkylthiols that are more effective in blocking the surface. It is proposed that this is due to the way the co-adsorbates interdisperse on the surface, with interaction between bis(4pyridyl) disulfide and long chain n-alkylthiols being sufficiently weak so as to leave open large areas of adsorbed bis(4-pyridyl) disulfide suitable for cytochrome c interaction. Relevant to this are recent studies on how stabilities of n-alkyl SAMs depend on chain length; notably shorter chains produce a less stable layer because the hydrophobic effect is smaller [13]. Further refinements in the analysis and interpretation of diffusion-controlled protein voltammograms have been reported. Honeychurch and Rechnitz have considered the influence on the appearance of cyclic voltammograms arising from the requirement that for SAM-modified electrodes, electron exchange must occur at the outer extremity of the SAM and not the electrode surface itself [14]. For cytochrome c solution voltammetry at a bis(4-pyridyl) disulfide-modified gold electrode, the peak separation is always greater than 57 mV; further, the diffusion coefficient is smaller than that obtained from studies at simple electrodes such as
2625
bare gold or indium oxide, or at a rotating modified gold electrode. Assuming that the surface is not blocked so as to generate inactive sites, these observations are predicted by a model that takes into account the effect of the potential drop that occurs across the adsorbed layer. Important advantages are gained if proteins can be immobilized at the electrode surface, giving an electroactive film that is ideally of monolayer coverage. Armstrong and co-workers have termed this approach ‘protein film voltammetry’ to encompass a variety of different configurations [15]. Without the need to have protein molecules free in bulk solution, minuscule sample quantities can be used, and there is greatly increased resolution of thermodynamic and kinetic information. Referring back to Fig. 1, immobilization can be achieved in several ways. Most simply, the electrostatics can be optimised. For example, at low ionic strength, cytochrome c adsorbs strongly at gold electrodes modified with v-carboxylate alkanethiol SAMs, as expected if this is governed by electrostatic interactions involving the lysine residues surrounding the exposed heme edge [16]. Stable hydrophilic electrode materials such as metal oxides or polished carbon (glassy carbon or pyrolytic graphite edge-PGE, which has acidic CO functionalities [17] also have a good capability for electrostatically controlled adsorption of proteins without denaturation or loss of biochemical activity [15,18,19]. Proteins have even been shown to be electroactive when immobilized in carbon nanotubes [20]. For the more acidic proteins, adsorption may require or be enhanced by the inclusion of co-adsorbates such as neomycin, polymyxin or other polyamines. Although the structures of the resulting electrode surfaces are not well understood, carbon and metal oxides offer certain advantages over SAM-modified metal electrodes: for example, pyrolytic graphite provides a very wide potential window, while metal oxides are optically transparent and permit UV-visible spectral changes to be observed as the surface-bound protein is redox-cycled [21]. Protein molecules may also be attached to electrodes by covalent bonding. Dong and co-workers have reported the use of carbodiimide coupling to link the normally membrane-bound enzyme cytochrome c oxidase to a Au/3-mercaptopropionic acid SAM [22]. The ‘chemisorbed’ enzyme shows reversible voltammetry due to one or more active sites, and can transfer electrons on to cytochrome c in solution. Hill and co-workers achieved adsorption of Azurin on gold by engineering a cysteine residue into the structure, The resulting AuS bond orients the enzyme for electron exchange and restricts its lateral movement on the electrode [23]. Various spectroscopic studies have been undertaken to examine the structural integrity of proteins adsorbed at electrodes. For heme proteins, the Raman spectrum
2626
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
is a sensitive probe of spin state and hence of any disruption to the active site. Surface-enhanced resonance Raman spectroscopy (SERRS) studies show that oxidized and reduced cytochrome c remain in their native conformations when adsorbed at a silver electrode coated with a monolayer of bis(4-pyridyl) disulfide or 4-mercaptopyridine, the depth of which is short enough to allow resonance transfer to the protein [24]. By contrast, at a bare silver electrode, the spin-state marker bands of adsorbed cytochromes are shifted, signalling conformational changes that may be relevant to function. Exploiting this further, Hildebrandt and co-workers have reported an application of time-resolved SERRS to measure the kinetics of electron transfer and coupled conformational changes in a bacterial counterpart of cytochrome c, cytochrome c552, adsorbed at a silver surface [25]. Techniques such as ellipsometry and quartz crystal microbalance (QCM) measurements are widely used to quantitate the progress of adsorption of proteins and supporting nanostructure on electrode surfaces [26]. The question of how protein molecules are oriented at the electrode has provoked some interesting experiments, such as that reported by Wilson and coworkers who used antibodies to study the potentialdependence of orientation of adsorbed cytochrome c3 [27]. Orientation distributions of molecules adsorbed on various surfaces have been studied by a combination of absorption linear dichroism and emission anisotropy [28]. The results indicate that for protein adsorption at SAMs there is a quite a broad distribution of orientations, so that the protein molecules are rather disordered. This is relevant to the problem of peak broadening due to dispersion, noting that, ideally, all the molecules in a monolayer should behave identically. This aspect has been identified in different kinds of experiment and modeled [29,30]. In cyclic voltammetry, reversible electrochemistry for an immobilized sample is characterized by a peak half-height width of 90.6/n mV at 25°C. This value varies with temperature (width 8T/K), or interactions between sites (cooperative or anticooperative). However it can also be increased by dispersion, i.e. inhomogeneity in thermodynamics or kinetics. Another way to prepare electroactive films of protein molecules is to immobilize them in surfactant films, which resemble lipid bilayers and multilayers. Rusling and co-workers have developed procedures for preparing surfactant films into which proteins are embedded. These may be multi-layer assemblies within which protein molecules diffuse, but do not escape to the bulk aqueous phase [31]. The surfactants are water-insoluble, with two long-chain alkyl chains; examples being didodecyldimethylammonium bromide (DDAB), dimyristoyl-phosphatidyl-cholate (DMPC) and dihexadecylphosphate (DHP). Films containing
protein molecules are obtained by several methods. The simplest way is to evaporate a chloroform solution of surfactant onto the electrode, which is typically pyrolytic graphite, gold or platinum. The electrode is then placed in a solution of the protein which then diffuses into the film. Another method is to mix an aqueous vesicle dispersion of surfactant with protein solution then evaporate this onto the electrode. These procedures give multiple layers, and for gold or platinum deposited on a quartz crystal, the depth of the layer can gauged by QCM measurements. Initial studies were carried out with myoglobin, a heme-containing oxygen binding protein that displays only slow electron transfer at bare electrode surfaces [32]. Rusling’s group found that myoglobin (Mb) gives reversible voltammetry when contained in a DDAB film at basal plane graphite, a discovery that has been elegantly exploited (see below) by Farmer and co-workers [33]. The approach was subsequently extended to cytochrome P450, enabling direct detection of redox transitions of the heme group [34]. Reduction potentials are sensitive to the nature of the surfactant, i.e. whether DDAB or DMPC are used, and also (as expected) to the presence of CO. The enzyme film also catalyzes reduction of O2 and reductive dechlorination of trichloroacetic acid. These films behave electrochemically as thin layers within which the protein molecules diffuse freely, although the reason for this high mobility is not clear. As liquid crystals, it is easy to envisage a very dynamic medium with alkyl chains moving around like a solvent. There is substantial spectroscopic evidence to suggest that these proteins retain most if not all of their native structure when immobilized in these surfactant films. Rusling and co-workers have employed UV-visible, EPR and reflectance FT-IR to establish the integrity of myoglobin and cytochrome P450 in films deposited on suitable substrates [35]. Amide IR bands, which reflect backbone CO stretching and are sensitive to protein conformation, do not alter significantly when Mb is transferred from aqueous solution to the DDAB film. Boussaad and Tao have used atomic force microscopy (AFM) to examine the interactions of proteins (myoglobin or cytochrome c) and DDAB with a pristine, highly-ordered pyrolytic graphite (HOPG) surface [36]. Cytochrome c adsorbs randomly, rapidly giving a complete layer. By contrast, adsorption of Mb from solution proceeds slowly, but eventually results in chain-like structures about 60 nm long consisting of blobs about 6 nm wide. At this point in time, voltammetry reveals a reversible redox couple with a potential close to that expected for myoglobin. If the electrode is coated with DDAB, adsorption of Mb is much more random, suggesting that strong DDAB – protein interactions break up the
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
inter-protein interactions responsible for aggregation at the bare electrode. The corresponding voltammetry is less reversible than at the bare electrode, most likely reflecting the increased distance over which electron transfer must occur. The structure of the DDAB layer was also found to be potential dependent, with the liquid crystal phase only being favored above 0 V. These results are somewhat at variance with previous conclusions that Mb is inactive at a bare graphite electrode, but nevertheless provide an alternative and revealing perspective. Rusling, Lvov and co-workers have also examined the properties of films prepared using an alternating layer-by-layer adsorption strategy [37]. Stacked bilayers can be formed starting from a layer of mercaptopropanesulfonic acid on gold or quartz, followed by successive layers of protein (Mb, cytochrome P450), polyanion/DNA, protein, etc. The progress of assembly can be monitored using a QCM. These films do not support protein mobility and most of the resulting electrochemical activity stems from proteins closest to the electrode surface. Using this method, it has recently proved possible to detect active sites in a photosynthetic reaction center [38]. The scope for developing membrane-mimetic surfaces appears very promising. Salamon and co-workers reported studies on two membrane bound proteins, spinach chloroplast cytochrome f and beef heart mitochondrial cytochrome c oxidase integrated into a phosphatidylcholine bilayer coated on a tin-doped indium oxide electrode [39]. Voltammetric signals due to reversible redox couples were obtained in each case, one from cytochrome f at 365 mV and two from cytochrome c oxidase, at 250 and 380 mV. These values are quite close to reported values, the latter corresponding to the CuA and cytochrome a electron-transfer centers of the terminal oxidase, although no catalytic activity was described. The shapes of the voltammograms suggest that these proteins are immobilized in the bilayer. Hawkridge and co-workers have developed an interesting way to immobilize cytochrome c oxidase, based on the formation of a submonolayer of octadecylmercaptan (OM) on a metal surface (Ag deposited on Au) [40]. This serves as the template/anchor for assembling a bilayer that is completed by amphiphiles (L-phosphatidylethanolamine or L-phosphatidylcholine). The cytochrome c oxidase/membrane mimetic electrode is prepared by entrapping an aliquot of solution containing detergent (deoxycholate)-solubilised enzyme and amphiphile against the OM-modified electrode surface with dialysis membrane, the principle being that as the detergent dialyzes out, the enzyme becomes incorporated into the OM/amphiphile bilayer. The immobilized enzyme exhibits voltammetry in which a broad oxidative wave observed at low scan rates is joined by a
2627
reductive wave as the scan rate is increased. The presence of a bilayer is supported by QCM measurements, while the catalytic enhancement that is observed with cytochrome c in solution suggests that the enzyme is oriented with its binding site for this natural partner directed away from the electrode. However, the charge that is passed during potential sweeps is far greater than expected for a monolayer, prompting the suggestion that the excess charge is capacitative and arises from a redox-coupled conformational change that increases ion transport in the lipid bilayer. This is an interesting idea, since redox-driven conformational changes are believed to be functionally important in this enzyme’s role as a proton pump.
3. What factors determine ET rates at protein – electrode interfaces? The most revealing experiments in this area have been carried out with SAM-modified electrodes since these provide the best defined surfaces. As with nonprotein systems, it is expected that ET rates for proteinSAM-electrode systems depend on a donor-acceptor distance, and indeed electrochemical studies with functionalized (X – Y) n-alkanethiol SAMs have confirmed that rates depend on the length of the spacer. Bowden and co-workers used cyclic voltammetry to determine electron exchange rate constants (k0) between cytochrome c and gold across medium-length HS(CH2)n COOH SAMs [16], whereas Niki and coworkers studied chain lengths as short as n =2 using ac potential-modulated UV-visible electroreflectance [41]. With this method, ET rates are obtained from the frequency dependence of the reflected light, and it is worth noting that it can be applied to systems such as ‘blue’ Cu proteins which do not have such intense optical transitions as hemes [42]. With the longer spacers 9 BnB 11, rate constants depend exponentially on chain length, consistent with a through-bond tunneling mechanism and a decay factor of 1.1 per CH2− group. This result is in excellent agreement with results obtained for simple redox species attached to alkanethiol SAMs [43,44]. As the chain length is shortened, rate constants do not increase as expected but instead approach a limiting value; accordingly it is proposed that electron exchange becomes gated by changes in conformation/orientation of the protein. The implication is that the most stable orientation does not correspond with the most favorable orientation for electron transfer, so that some rate-limiting adjustment is required. However, studies carried out on n-alkyl SAMs terminated with a covalently attached small-molecule redox couple have indicated that complications arise here also as the chain length is shortened [45]. Whatever the reason for kinetic limitations, it is clear that rate con-
2628
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
stants well in excess of 1000 s − 1 are readily attainable with proteins whose active sites lie quite close to the surface that contacts the electrode. Rate constants of this magnitude are observed also in studies of proteins adsorbed at graphite electrodes [46]. Evidence that the rates of electron transfer depend in a very subtle way on how the protein interacts with the electrode is provided by Bowden and co-workers, who have reported an intriguing result obtained for the voltammetry of horse heart and yeast cytochrome c adsorbed at mixed SAMs [47]. With horse heart cytochrome c, the ET rate constant obtained using a mixed SAM derived from equimolar amounts of HS(CH2)10COOH and HS(CH2)8OH is about 5-fold faster than one comprised only of HS(CH2)10COOH. By contrast, with yeast cytochrome c, the pure HS(CH2)10COOH SAM gives only a very low rate, but there is a 2500-fold rate enhancement when switching to the mixed HS(CH2)10COOH/HS(CH2)7OH monolayer. The two proteins have similar ET reorganisation energies and surface charge distribution, but there are large differences in amino acid composition that could affect the ET coupling to the electrode. Continuing this theme, the question of why the electrochemical rates observed for myoglobin depend so much on the electrode environment is unclear. Here we are comparing heterogeneous rate constants (cm s − 1) among different studies in which myoglobin is diffusing in the aqueous or surfactant film phase. Rusling and co-workers have suggested that the high activity observed in surfactant films is due, at least in part, to inhibition of adsorption of deactivating contaminants [48]. Another possibility concerns the nature of the ligation to iron and how this affects the ET reorganisation energy. Hawkridge and co-workers showed that the CN−-ligated form displays much more reversible voltammetry than the native aquo-Fe(III) form at an indium tin oxide electrode [49]. Variations also exist between myoglobins from different species; the protein from horse heart shows a higher rate constant than the recombinant protein from sperm whale [50]. Armstrong’s group later showed that a significant increase in the rate constant for myoglobin electrochemistry at a pyrolytic graphite edge electrode could be achieved simply by replacing the distal histidine (H64) by amino acids such as glycine or leucine [51]. The structures of many of these H64 mutants have been crystallographically characterized. Histidine-64 stabilizes the H2O ligand that is bound to Fe(III) but not Fe(II), and also links this coordinated H2O to bulk solvent water via a chain of H-bonds; the mutants break this connection. Consequently, placing myoglobin in the more hydrophobic environment of a surfactant film can disrupt these interactions and lower the reorganization energy for electron transfer.
4. Studies of protein redox reactions using voltammetric techniques An obvious advantage of voltammetric methods over the more conventional and widespread use of mediated potentiometric titrations is that a large amount of data can be accumulated in a short time-an obvious asset in such laborious tasks as measuring the variation of reduction potentials with pH. Freedom from the dependence on chemical titrants and mediators or the need for a characteristic spectral feature further facilitates experiments that are not readily undertaken with traditional methods. Some of these aspects will now be described before dealing with some of the more recent developments in which the protein is interrogated as a film on the electrode. First, unlike potentiometric studies, there is no requirement for a distinctive and unambiguous change in some spectroscopic parameter-for example, an EPR signal that may be observable only at cryoscopic temperatures and thus require rapid freezing. A problem with freeze quenching is that re-equilibration of electrons can occur during the freezing process, particularly if quenching from elevated temperatures [52]. By contrast, direct electrochemical methods allow in situ measurements of reduction potentials. This aspect is exemplified in measurements of the temperature dependence of the reduction potential of the ferredoxin from Pyrococcus furiosus [52 – 54]. This organism is a hyperthermophile found naturally near deep sea hydrothermal vents at temperatures exceeding 100°C; an understanding of its electron-transport chains thus requires knowledge of the thermodynamic properties of individual components measured under these extreme conditions. Earlier potentiometric titrations, monitored by EPR spectroscopy after freezing equilibrated samples, had shown a non-linear variation of reduction potential at high temperature, with a discontinuity occurring at 80°C [53]. A change in conformation was proposed, and this was discussed in terms of its relevance for an organism that grows at these elevated temperatures. However, recent in situ voltammetric studies (at a glassy carbon electrode) have revealed that no such discontinuity occurs [52,54]. A second example is the ability to measure redox thermodynamics at high pressures. The variation of reduction potentials with pressure yields V 0, the change in molar volume between oxidized and reduced forms, and hence the difference in compressibility ( 8 #V 0/ · #P). Sligar and co-workers used a high-pressure voltammetric cell to study cytochrome c up to pressures of 5 kbar [55]. Over the range from ambient to 5 kbar, the potential increased by 76 mV, enabling the authors to quantify independent observations that cytochrome c(II) has a more compact structure than cytochrome c(III). Smith and coworkers have applied voltammetric
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
Fig. 2. Voltammograms of films of 7Fe ferredoxins on a PGE electrode at 0°C. Azotobacter 6inelandii (scan rate 20 mV s − 1, pH 7.0). Desulfo6ibrio africanus (scan rate 191 mV s − 1, pH 7.0) and Sulfolobus acidocaldarius (scan rate 10 mV s − 1, pH 7.4). In each case an electroactive coverage of approximately one monolayer is obtained in the presence of polymyxin as co-adsorbate. Signals A%, B% and C% refer to the redox couples [3FE–4S] + /0, [4FE – 4S]2 + / + and [3FE–4S]0/2, respectively.
methods to study both high-temperature and high-pressure effects on rubredoxins (Rd, active site =Fe(SR)4) from a mesophile (Clostridium pasteurianum) and hyperthermophile (P. furiosus) [56]. The effects are appreciable, for example the reduction potential of P. furiosus Rd decreases from 31 mV at 25°C to −93 mV at 95°C, while reduction potentials increase slightly with pressure. Highly reducing species with reduction potentials that are too negative to access with conventional chemical titrants such as dithionite can be studied by voltammetric methods. The result of this is that there are now many examples of proteins containing a [4Fe–4S]2 + / + cluster with reduction potentials more negative than −600 mV. In particular, voltammetry has made possible the extensive characterisation of crystallographically-defined mutants of the 7Fe ferredoxin from Azotobacter 6inelandii in which the environments of the [3Fe–4S] and [4Fe–4S] cluster are altered systematically [57,58]. The [4Fe–4S] cluster in this protein has a
2629
very negative reduction potential and was once thought to be of the ‘HiPIP’ type ([4Fe – 4S]3 + /2 + ). Until recently, the so-called ‘clostridial 8Fe’ ferredoxins were the only well studied class of small proteins containing two [4Fe – 4S] clusters, and invariably these have very similar reduction potentials, at around − 400 mV. Through voltammetric methods, and again through the ability to probe active sites with very negative reduction potentials, it has recently become clear that there is a distinct class of naturally occurring 8Fe ferredoxins in which one of the [4Fe – 4S] clusters has a much lower potential which prevents its reduction by dithionite [59,60]. Even when very negative potentials do not pose a problem, iron-sulfur clusters are often difficult to study by conventional methods, their broad and uncharacteristic absorption spectra providing only poor signatures for real-time measurements at ambient temperature. The situation is particularly unsatisfactory if dealing with labile, air-sensitive species. Voltammetry provides an attractive solution to these problems, a good example being the ability to engage and monitor reactions occurring at [3Fe – 4S] clusters. As shown in Fig. 2, [3Fe – 4S] clusters display two signals, one due to the well-studied [3Fe – 4S] + /0 couple and one at very negative potential which integrates to double intensity and is assigned to the [3Fe – 4S]0/2 − couple [61,62]. The latter signal has the narrow width expected for cooperative two-electron transfers and the variation of reduction potentials with pH shows that 2 – 3 H+ are also transferred. Proton transfer is not a widely acknowledged property of Fe – S clusters, but [3Fe – 4S] clusters appear to be an interesting exception. The two signals serve as a diagnostic feature of the presence of [3Fe – 4S] clusters in proteins, and this has been exploited in studies of the potential-dependent transformations to cubane adducts [4Fe – 4S] or [M3Fe – 4S] that are observed for the ferredoxins from Desulfo6ibrio africanus and P. furiosus [62 – 65]. These redox-coupled reactions (see below) are conveniently induced and monitored in proteins adsorbed at a pyrolytic graphite edge electrode. Several novel clusters have been ‘synthesised’ in this way, including those incorporating Tl, Cu and Pb. The resulting [M3Fe – 4S] clusters undergo reversible ligand exchange, which presumably occurs at the new subsite M [66]. Coupling between electron transfer and chemical/ conformational changes is conveniently depicted in terms of a square-scheme such as that shown (Scheme 1), in which the species are interconnected by electrochemical (E) and chemical (C) transformations [66]. The latter include protein/prosthetic group conformational changes, and transfer of ions (including protons). Whereas equilibrium methods like potentiometry reveal only the thermodynamics of these systems, dy-
2630
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
namic methods provide kinetic information. In biology, electron transfer drives chemical reactions; equally important, perhaps, are ‘gated’ electron-transfer reactions, in which electron-transfer rates are determined (gated) by the preceding chemical process. Voltammetry is a powerful tool for resolving and quantifying these complex reactions, and several examples involving cytochromes and Fe–S clusters have recently been reported. The redox properties of cytochromes depend greatly on the nature of the axial ligation to iron. In particular, since Fe(II) may prefer different ligands to Fe(III), electron transfer can drive changes in axial coordination and vice versa, and the resulting kinetics can be observed and deconvoluted by electrochemical methods. Studies by Haladjian and co-workers [67], then Barker and Mauk [68] revealed the ligand interchange kinetics accompanying redox cycling of mitochondrial cytochrome c. Under mild alkaline conditions, the axial methionine ligand to Fe(III) is easily replaced by competing ligands, and cyclic voltammograms recorded on appropriate timescales (i.e. scan rates of the order of a few volts per second) reveal the interconversion kinetics. Interestingly, this type of problem can be tackled by other electrochemically-based methods, notably Hawkridge and co-workers used double potential step chronoellipsometry, in which circular dichroism is monitored through an optically transparent indium oxide electrode [69]. Feinberg and co-workers have recently studied a mutant form of yeast cytochrome c, in which phenylalanine-82, situated close to Met-80, is replaced by histidine [70]. This directs the formation of a welldefined Fe(III) state in which His-82 replaces the original Met-80 ligand present in Fe(II). The results are depicted in Fig. 3. Recently Mauk and co-workers have been able to detect states formed in mutants in which one or more of the lysine residues that normally replace Met-80 in the Fe(III) form are replaced by alanine [71]. Even more complex transformations occur in the di-heme enzyme cytochrome c peroxidase from P. denitrificans. Here, cyclic voltammetry displays how reduction of one heme induces a change at the other, and
Scheme 1.
activates the enzyme to catalyze peroxide reduction [72]. The redox-coupling experiments outlined above involved free protein molecules undergoing reversible diffusion-controlled electrode reactions. Alternatively, coupled chemistry can be studied in adsorbed proteins. Immobilising the protein sample creates important advantages in terms of sample size and response time, and even cyclic voltammetry can provide kinetic resolution on the millisecond timescale or lower, and allow the detection and measurement of electron-transfer gating [73]. Many redox reactions in proteins are coupled to proton transfer, and indeed this forms the basis of energy conservation whereby transfer of electrons down a thermodynamic gradient is used to pump protons uphill so that energy is stored as a proton gradient. Some of the details of how this can occur at an active site have been revealed by a study of the [3Fe – 4S] cluster in A. 6inelandii 7Fe ferredoxin [73]. One-electron reduction (to give [3Fe – 4S]0) is coupled to the uptake of a single proton: this further example of proton transfer at Fe – S clusters is significant because high-resolution crystal structure studies on the different oxidation and protonation states reveal (a) that the cluster is buried with no access for water molecules, and (b) that a carboxylate group from an aspartate (D15) is located close to the cluster on the protein surface. The mechanism of proton transfer between the cluster and solvent has been determined using protein film voltammetry in conjunction with site-directed mutagenesis. As indicated in Fig. 4, data analysis procedures (‘trumpet plots’) compare the scan rate dependence of voltammetric peak positions to those expected for suitable kinetic models. With the D15N mutant, proton transfer is very slow, and oxidation of [3Fe – 4S]0 is gated by release of H+. The native protein exhibits much faster proton transfer, and rate constants vary with pH in accordance with the proton being transferred by the D15 carboxylate. As mentioned above, [3Fe – 4S] clusters are unique in exhibiting further redox chemistry in the form of a two-electron/two-proton reduction of [3Fe – 4S]0 to yield a novel ‘hyper-reduced’ state in which all three iron atoms are formally 2+ [61]. The physiological significance of this reaction is unclear. However the achievement of formal oxidation levels ranging from all-Fe(III) to all-Fe(II) is unprecedented for Fe – S clusters, as is the coupling to multiple proton transfer. The process occurs at potentials of around −700 mV at pH and may involve some rearrangement of the cluster environment. Recent experiments using fast scan cyclic voltammetry reveal that the hyper-reduced 2-state can undergo a very fast and reversible two-electron/twoproton oxidation, yielding a state that differs from the normal ‘0’ level into which it converts within a second
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
2631
Fig. 3. Top. Experimental cyclic voltammogram (circles) and simulation (solid line) for Phe82His yeast Iso-1-cytochrome c (in solution) obtained using a gold electrode modified with a monolayer of bis(4-pyridyl) disulfide. Scan rate 50 mV s − 1. Bottom. Square scheme showing redox-coupled interchange of Met-80 and the residue His-82 that is introduced. The voltammogram is simulated using an equilibrium constant KR/R% = 3.6× 105 (i.e. R favored) and k (R% R)=10 s − 1. Adapted from original figure in reference [70], with permission.
2632
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
Fig. 4. Plots of peak position as a function of scan rate (logarithmic scale) for proton-coupled reduction and reoxidation of the buried [3FE – 4S] + /0 cluster of the D15N mutant of Azotobacter 6inelandii ferredoxin, in which proton transfer is retarded. The detailed kinetics and energetics can be analyzed by simulation of the shapes of the ‘trumpet plots’. In this case the black diamonds correspond to pH 8.34 (pH pK, thus no protonation occurs) while the gray diamonds correspond to pH 5.50 (pH pK). The solid line is a simulation with pK = 6.9, and the proton release rate koff is 2.5 s − 1. The shapes are explained as follows using the scheme shown above. At pH\ pK, no proton transfer occurs and the peaks separate in an approximately symmetric fashion. By contrast, at pH pK, the reduced cluster becomes protonated and three regimes are observed: (a) at low scan rate, the peaks lie close together at the formal reduction potential which is at a more positive value because protonation is coupled to electron transfer; (b) at intermediate scan rates, electron transfer to the cluster is followed by proton transfer (E1CR), but re-oxidation is prevented because the proton ‘off’ rate is too slow and gates the reaction, so that no return peak is observed; (c) at fast scan rates the electron is ‘re-called’ from the reduced cluster before protonation can occur, thus giving peaks that coincide in position with the high pH experiment, i.e. protonation is decoupled (E1 only). Reference [46].
[73]. These results are supported by recent experiments carried out at −85°C in which the protein coated electrode is transferred to a cryosolvent (70% methanol), enabling this chemistry to be observed at slow scan rates [74]. The results suggest that the cluster is performing disulfide chemistry, although its physiological relevance is unclear.
Interesting voltammetry has been observed for the iron storage protein ferritin. This protein has a giant cavity which can accommodate over 4000 Fe(III) mineralized as a hydrated oxide. Although it is widely accepted that uptake and release of iron involves reduction to Fe(II), details of the processes involved are yet to be established. Zapien and coworkers observed that ferritin adsorbs at a tin-doped indium oxide electrode to give films of submonolayer coverage, thus providing the opportunity to control and observe iron mobilization [75]. The voltammetry is not straightforward; in the first reductive scan, two large reduction peaks appear at potentials of approximately − 100 and − 300 mV versus SHE, and correspond to two different forms of the protein. Subsequent cycles show just single oxidation and reduction peaks at an average potential of 0 mV. However, in the presence of EDTA, the voltammeric peaks vanish after only one cycle, showing that iron is released and then complexed. This establishes the capability of voltammetry to probe this complex chemistry and further studies should provide some interesting alternative perspectives on this important problem. For electron-transport enzymes, an important criterion for voltammetric experiments should be that electrons exchanged with the electrode should be translated into catalytic action. Without this, there must be doubt about whether or not the enzyme is catalytically active or able to function with the electrode as a redox partner. A theoretical analysis of electrocatalytic waveforms due to adsorbed enzymes has been described, the aim being to recognize different modes of kinetic control and extract mechanistic information [76]. The interface consisting of electrode, enzyme and electrolyte is considered as a three-resistor series through which catalytic current flows. Three limiting cases have been considered, corresponding to control by interfacial electron transfer (least useful), substrate mass transport, and enzyme catalysis (most useful). Farmer and co-workers have discovered that myoglobin entrapped in surfactant films catalyzes interesting transformations of NO and NO2− [33]. In this confined and controllable state, myoglobin provides an excellent small-protein model system for examining the NO metabolizing activities of certain heme enzymes. Some results are shown in Fig. 5. Under N2, Mb confined within a DDAB film at basal-plane graphite displays two reversible redox couples, assigned to Fe(III)/(II) and Fe(II)/(I) transformations of the active site. The protein appears to diffuse within this film, as noted also by the Rusling group. When N2 is replaced by NO, the voltammetry changes dramatically, with the appearance of a catalytic reduction peak at −0.75 V versus SCE which is dependent on the partial pressure of NO but independent of pH over the range 5.5 – 10. Controlled potential electrolysis with mass spectrometric analysis of the head gas reveals
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
that the major product is N2O. A film of pre-formed MbNO cycled under a N2 atmosphere shows a reversible couple at −0.87 V, which persists indefinitely at high pH provided the potential is cycled sufficiently fast to outrun an irreversible chemical reaction which results in the reappearance of Fe(III)/(II) and Fe(II)/(I) signals. Piecing these observations together and drawing on known reactions of NO species in solution, the authors have modeled the MbNO electrochemistry in terms of the following hypothesis and scheme (Scheme 2) the result of which is the reduction of NO to give N2O. Production of NO2 is accounted for by side reactions including reduction of the Fe(II) nitrosyl Mb which is followed by irreversible dissociation of reactive NO−. Mb-DDAB films catalyze reduction of NO2− in reactions that produce NH3, NH2OH, N2 and N2O [33].
Fig. 5. Catalytic reduction of NO by myoglobin incorporated in a surfactant film. Cyclic voltammograms in deaerated pH 7.4 phosphate buffer, 0.1 M NaBr, scan rate 500 mV s − 1: (a) graphite electrode cast with a DDAB film; (b) myoglobin incorporated into DDAB film; (c) as for (b) but solution saturated with NO. Modified from original figure in reference [33]b, with permission.
2633
Mb confined to these surfactant films is also active in oxygenation reactions. Under aerobic conditions, large catalytic reduction currents are observed, which are ascribed to the reduction of coordinated O2 to produce H2O2 [77]. However, the oxy-intermediate that is produced in this reaction is active also in oxygenating substrates such as styrene. Styrene oxidation is not rate-determining in the electrochemical mechanism but the transformations are clearly evident from glc analysis of bulk electrolysis products. For adsorbed redox couples, half-height widths vary as 1/n, while peak heights depend on n 2, where n is the number of electrons transferred in the process. The significance of this relationship is that cooperative twoelectron active sites appear much more prominent in voltammograms and can quite easily be diagnosed. This is illustrated in the voltammetry of yeast cytochrome c peroxidase, an enzyme that catalyses reduction of hydrogen peroxide by cytochrome c and which has long been a valuable system for studying oxygen activation by hemes. Adsorption of cytochrome c peroxidase (Fe(III) resting state) at a pyrolytic graphite edge electrode from dilute ice-cold 20 mM phosphate buffer gives rise to a prominent signal at + 750 mV versus SHE (pH 5.4) [78]. The signal consists of oxidation and reduction peaks that each have half widths below 80 mV, and it is therefore assigned to a cooperative twoelectron couple, i.e. oxidation of Fe(III) to a state that is equivalent to ‘Fe(V)’. The peaks transform to a sigmoidal catalytic reduction wave when H2O2 is added. Studies of the rotation rate and peroxide concentration dependence yielded a turnover number comparable to that reported from solution studies, and it was concluded that the oxidized form of the redox couple is either ‘compound I’ (Fe(IV) and radical) or a closely related species. The electron-exchange kinetics of adsorbed cytochrome c peroxidase were also measured using staircase cyclic voltammetry, although interpretation of the value of approximately 6 s − 1 is complicated by the fact that two electrons are transferred [79]. Comparisons have been made with a mutant W51F in which the distal tryptophan (not the same tryptophan that houses the radical) changed to phenylalanine. The mutant has a more positive reduction potential ( + 883 mV at pH 5.4) and higher catalytic activity, but is less stable at the electrode. The result showed that this tryptophan plays a significant role in stabilizing the Fe(IV) heme, possibly by creating favorable interactions with the bound oxide ion. The flavin group provides a further illustration of how cooperative two-electron centers may be particularly ‘visible’ to voltammetric methods. Normally, FAD or FMN cofactors in enzymes are detected and studied via their visible absorption spectrum (400 – 500 nm) or occasionally through EPR if a one-electron radical is stabilized. Flavocytochrome c3, a periplasmic fumarate
2634
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
Scheme 2.
reductase from the marine bacterium Shewanella frigidimarina, provides an interesting example of an enzyme in which the FAD is very difficult to observe because its spectrum is obscured by the intense Soret bands from four heme groups; however, it is rendered clearly visible by voltammetry. The enzyme adsorbs from dilute solution at a pyrolytic graphite edge electrode to give a dense coverage equivalent to an electroactive monolayer [80]. The resulting voltammetric signal consists of complex reduction and oxidation peaks enveloping the contributions of all redox centers. As shown in Fig. 6, the FAD cofactor is clearly visible as a sharp and prominent component of this envelope. Despite not being covalently bound, the FAD does not dissociate from the adsorbed enzyme. The adsorbed enzyme is very active: when fumarate is added, the complex envelope collapses to a sigmoidal wave the magnitude of which is very dependent on electrode rotation rate. The ability to probe activity as a function of potential makes it possible to detect even the subtlest of changes in rate that may accompany redox transformations of centers in an enzyme. This may indicate a regulatory activity for certain centers or help confirm their involvement in catalytic electron relays. An interesting example is the fumarate reductase from E. coli. This is a membrane-bound enzyme consisting of four subunits, two of which are catalytic and membrane-extrinsic (i.e. protrude from the membrane surface) while the other two are membrane-intrinsic anchors and provide a binding site for the natural electron donor menaquinol. A soluble form consisting only of the membrane-extrinsic subcomplex can be obtained, either by chemical resolution or genetic engineering. This adsorbs strongly at a rotating disk pyrolytic graphite edge electrode and yields informative voltammetry [81,82]. In the absence of fumarate, the observed ‘non-
turnover’ signals consist of (a) oxidation and reduction envelopes covering the range 0.1 to − 0.2 V, each dominated by a prominent peak (the covalently-bound FAD) which overlays the weaker features due to two Fe – S clusters, and (b) a weaker feature at −0.3 V which is assigned to a [4Fe – 4S] cluster having a singularly negative reduction potential. Addition of fumarate gives rise to an intense sigmoidal catalytic reduction wave whose amplitude is dependent on rotation rate. As mass-transport control is relaxed, i.e. at high fumarate concentrations and high rotation rates, the cata-
Fig. 6. Baseline corrected voltammogram for a film of flavocytochrome c3 (Shewanella frigidimarina) adsorbed at a PGE electrode in the presence of polymyxin as co-adsorbate. Temperature 0°C, pH 6.0. The FAD cofactor is clearly observed above the redox transitions due to the four heme groups. The different components have been resolved (gray lines within envelope) and the resulting overall simulation is overlaid on the experimental result with excellent agreement. When fumarate is added to the solution and the electrode is rotated, the signals transform into a strong catalytic wave.
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
lytic wave exhibits a second boost in current close to the potential of the [4Fe–4S] cluster. This shows that despite its low potential, this center is somehow involved in catalytic electron transfer. Mitochondrial complex II (succinate–ubiquinone oxidoreductase) is structurally similar to fumarate reductase (two membrane anchors and two membrane-extrinsic subunits containing a covalentlybound FAD and three Fe–S centers) and can also be prepared as a membrane-extrinsic soluble form (succinate dehydrogenase) which adsorbs on a pyrolytic graphite edge electrode. In this state it is catalytically active, but the response is not stable and the coverage is too low to observe signals from the active sites. However, the short-lived activity is informative [83]. Experiments conducted with a 1:1 mixture of fumarate and succinate produce a characteristic current response: successive cycles show succinate oxidation at high potential and fumarate reduction at low potential, each decreasing in amplitude but crossing at isosbestic potentials in each direction. The average values of these potentials (at which there is no net reaction) corresponds to the formal reduction potential of fumarate. Catalytic waveforms at different pH values are determined by computing difference voltammograms, i.e. subtracting an early cycle from a later one. The resulting profiles show: (a) that at pH values above 7.4 the enzyme is actually biased to operate in the direction of fumarate reduction, and (b) that although fumarate reductase activity is high, it is limited to just a narrow region of potential. As the potential is made more negative, intending to drive fumarate reduction faster, the activity shuts down. Because this resembles negative resistance, the behavior was termed the ‘tunnel diode’ effect [81]. The decrease in activity can be fitted to a two-electron reaction that occurs close to the potential region expected for the FAD. This suggests that in its FAD-reduced form, the enzyme favours a less-active conformation which is avoided when the steady-state potential is maintained at a higher value. In support of this, non-electrochemical assays showed a negative order in viologen concentration, i.e. the rate increases as the reductant is consumed [84]. Succinate dehydrogenase from E. coli gives the same result as the beef heart enzyme despite there being a significant difference in the isoelectric point and only 50% amino acid similarity (mostly confined to those residues believed to lie in the active site) [85]. These results argue against, but do not rule out the possibility, that the ‘tunnel diode’ effect is due instead to a potential dependent re-orientation. Armstrong and co-workers used this system to explore how voltammetry can be used to observe and resolve H/D isotope effects in electron-transport catalysis [86]. Hydrogenases have featured in several reports, and useful information has been obtained. An early example of faradaically monitored catalytic activity was re-
2635
ported by Bianco and Haladjian, but details on the enzyme properties were not revealed [87]. More recently, the all-Fe enzyme from Megasphaera elsdenii has been studied [88]. The enzyme adsorbs at a glassy carbon electrode, and from the relative catalytic currents obtained in either direction it was possible to determine electrochemically the degree to which it is biased to operate in the direction of H2 production. The NiFe enzyme from Chromatium 6inosum has recently been studied; in this case ‘non-turnover’ signals are revealed quite clearly in the CO-inhibited enzyme, which are tentatively assigned to the [3Fe – 4S] cluster and the two [4Fe – 4S] inhibited enzyme, that are tentatively assigned to the two [4Fe – 4S] clusters (each at approximately − 0.3 V) and the [3Fe – 4S] cluster at −0.03 V [89]. Catalytically, the enzyme shows a clear preference for H2 oxidation over proton reduction. Oxidation of H2 (10% atmosphere) is mass-transport controlled, as revealed by strong sigmoidal currents which are very dependent on electrode rotation rate, and the estimated turnover number is in excess of 1500 s − 1, considerably higher than values obtained by conventional dye reduction assays in solution. Significantly, the H2 oxidation activity is very high at potentials below that of the [3Fe – 4S] cluster, so that this site must remain reduced throughout turnover.
5. Mediated protein electron transfer Most of the proteins discussed above have as their biological function the transfer of electrons to other proteins. They typically function in ordered structures such as mitochondria and the redox active centers are generally accessible to the outer surface of the protein. In contrast, the redox reactions catalyzed by a class of enzymes known as oxidoreductases involve small molecules. Thus one of the two required substrates (redox couples) for the enzyme will be designated the substrate and other the co-factor or co-substrate. Because the catalytic center of the enzyme is buried well within the protein, communication between this center and an electrode serving as a source or sink for electrons is frequently poor or non-existent. Because oxidoreductases selectively catalyze reactions of analytical importance, it is of interest to efficiently couple the enzyme electron transfer to the electrode, which then effectively serves as one of the substrates. With a few exceptions such as horseradish peroxidase (HRP) [90,91], heterogeneous electron transfer rates for enzymes are vanishingly low so that the enzyme redox activity must be coupled to the electrode using mediators whose redox states are alternatively recycled at the electrode, at adjacent mediators, and the enzyme active site. If successfully accomplished, the rate of recycling of the redox state of the enzyme then becomes propor-
2636
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
tional to the rate of the enzymatic reaction and therefore to the concentration of the substrate of interest. Such a device is known as an enzyme electrode and has been widely employed as an analytical tool, especially for the determination of glucose and lactate. There are two broad classes of oxidoreductases: oxidases that use oxygen as a co-substrate and dehydrogenases that typically use NAD or NADP as a co-factor. It should be said at the outset that critical properties of the enzyme electrode as a practical device (sensitivity, stability, size, selectivity and the medium in which measurements will be made, etc.) will determine whether the device is useful or not. One important goal is to make the sensor ‘reagentless’, meaning that nothing has to be added to the medium for proper functioning. This requirement is a major reason for the overwhelming popularity of oxidases, because many media have sufficient oxygen present that none need be added. The reaction sequence is: Sred +Eox ? Sox +Ered
(1)
S%ox +Ered ? S%red +Eox
(2)
If, for example, glucose oxidase (GOx) is the enzyme employed, then Sred, Sox, S%ox and S%red correspond to glucose, gluconic acid, oxygen, and hydrogen peroxide, respectively. The goal is to make reaction 1 rate limiting, and this necessarily requires a large excess of oxygen (S%ox). The overall rate of the reaction is monitored either by following the rate of consumption of oxygen (requires two measurements: one in the presence of enzyme, one in its absence) or the formation of hydrogen peroxide. The advantage of the former method is the selectivity that can be achieved using a gas permeable membrane, the disadvantage being the added complexity of the device. In the latter case an applied potential of around 600 mV versus AgCl/Ag reference electrode is needed, and in this region numerous endogenous species such as ascorbic acid, uric acid, and a variety of biogenic amines are electroactive. A major and largely successful solution to this problem is the use of permselective membranes to exclude interfering species. The electrochemistry of hydrogen peroxide at physiological pH at a platinum electrode is actually quite complicated and has been the subject of recent study [92,93]. The rate of electron transfer depends on the potential-dependent formation of Pt(II) sites on the electrode surface. Mediated electron transfer has therefore attracted considerable attention not only for reasons of fundamental interest but also to make a more practical enzyme electrode. Mediators should be distinguished from electrode modifiers which are not themselves electroactive at the potential of the protein electron transfer reaction. If reaction 2 above could be replaced by one involving coupling or ‘wiring’ to the electrode, then, in
principle, the sensor response would no longer be sensitive to fluctuations in S%ox. This condition can only be met if the rate of electron transfer via the ‘wired’ route is high compared to the rate of reaction of the enzyme with endogenous oxygen. Most mediator-based sensors involving oxidases do not meet this requirement and exhibit parasitic effects of oxygen on their response which in some cases are as large as 70% [94]. Willner’s group has elegantly addressed the question of whether it is possible to ‘wire’ an enzyme with sufficently high efficiency to prevent competition with oxygen [95]. This objective was apparently met by removing the non-covalently bound FAD redox center from the enzyme (GOx) and reconstituting the enzyme on a tether consisting of cystamine chemisorbed on a gold surface, a pyrroloquinoline quinone (PQQ) link, and FAD. The transfer of electrons from the FAD to the PQQ and thence to the electrode was sufficiently rapid that a high turnover number (maximum rate for the enzymatic reaction) was obtained (900 9 150 s − 1), which compares favorably with a value of about 1000 s − 1 for the enzyme in solution. There is only a monolayer of enzyme on the electrode surface, thus aiding efficient communication. Such a device might have, however, only limited stability because of the small amount of enzyme present. Heller’s group has examined a variety of approaches to ‘wiring’ of enzymes with the goal of finding a mediator that reacts rapidly with the enzyme, but has also a low potential. In the case of oxidases, the formal potential of the flavin redox center is low enough that a relatively weak oxidant with a potential in the range of +30 to −100 mV versus AgCl/Ag reference would be sufficient to drive reaction 2 to completion without at the same time effecting the electrocatalytic oxidation of species such as ascorbate. The approach has involved the covalent attachment of redox centers such as [Os(bpy)2Cl] + /2 + [96] or [Os(4,4%dimethoxy-2,2% bipyridine)2Cl] + /2 + (E o% = 0.035 V versus SCE [96] to polymeric hydrogels. The low potential of the mediator reduces but does not eliminate the electrocatalytic oxidation of ascorbate. Although the hydrogels are somewhat fluidic, the redox centers are still not mobile enough to compete favorably with freely diffusing oxygen. Some success in reducing oxygen dependence has been achieved by employing membranes that will ‘salt out’ oxygen so that it does not reach the enzyme layer [97]. A number of other mediators have been tried including ferrocene derivatives [98], phenoxazines, phenothiazines, and Wurster’s salts [99] and the conclusions are consistent: mediated systems frequently show poor stability when operated continuously for extended (hours – days) periods of time. In general, it appears that the demise of the mediator itself, through leaching or chemical degradation, is the primary problem [100]. A solution has been the dissolution of the mediator in carbon
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
paste to provide a largely renewable supply. Wang has proposed this strategy for ‘storing’ oxygen [101]. This approach is acceptable provided that there is no problem with soluble mediator leaking into the sample. On the other hand, electrochemically-based ‘fingerstick’ strips which aid diabetic patients in monitoring blood glucose use non-covalently immobilized mediators such as ferrocene and ferricyanide, but such devices need only to give a reliable response for perhaps 30 s [102]. Enzymes have also been adsorbed on or trapped within electrodeposited conducting polymers [103,104]. One mediator system that yields surprisingly stable sensors is the so-called organic salt electrode. This is based on the use of an electrode coated with the organic salt tetrahydrofulvalene (TTF) tetracyanoquinodimethane (TCNQ). First reported by Kulys [105] in the 1980s it became the focus of significant study. It is thought to mediate electron transfer with GOx, resulting from the corrosion of the electrode to produce a layer of soluble cationic TTF+ close to the electrode surface that can reoxidize the enzyme according to reaction 2 [106]. It has been suggested that TTF+ is also not stable [99], thus the apparent sensor stability is probably due to the replenishment by corrosion of the TTF+ available for electrocatalysis. Khan [107] has investigated in some detail the importance of the morphology of the organic electrode surface in defining the sensitivity and stability of a glucose sensor constructed from this material. A high rate of mediator efficiency was achieved due in part to the nature of the TTF+TCNQ− salt deposit that was dendritic and shaped like a tree. This could have caused efficient corrosion with the result that the sensor showed very little dependence on oxygen. Ascorbate is a potential interferent to the operation of sensors using this system, but in this case the interference was about 1% at 10 mM glucose. The apparent Michaelis constants for glucose, that are again indicative of efficient mediation, were in the range of 20–60 mM, depending on the applied potential. Another approach to the facilitation of electron transfer is the use of colloidal gold particles [91]. HRP is immobilized on gold particles (30 nm diameter) which are then deposited on an electrode surface. By comparison, efficient coupling of the enzyme to a smooth gold electrode is not observed, due no doubt to the ability of the colloidal gold particles to interact intimately with the protein. The authors postulate that only the monolayer of gold particles in direct contact with the electrode is really effective and that the addition of mediators to ‘connect’ subsequent layers does not appear to have any effect. Hydrogen peroxide produced by reaction 2 above oxidizes the HRP and a potential of 0 V versus AgCl/Ag is required to reduce the HRP to its native state. The advantage of this arrangement is that the applied potential is in a range
2637
where few endogenous species in biological fluids are electroactive, the disadvantage is that HRP can nonspecifically catalyze the oxidation of various organic molecules present in the medium. Direct electron transfer of HRP has also been demonstrated at suitably prepared carbon electrodes [90]. Several fundamental studies have enhanced understanding of electron transfer between mediators and enzymes: investigation of single electron donors and acceptors (largely ferrocene derivatives) shows that their reactivities do not follow simple outer sphere electron transfer [108]. Mikkelsen and co-workers [109] have developed the theory for intramolecular electron transfer involving the multiple attachment of ferrocene derivatives to GOx. The highest rate of intramolecular electron transfer was observed for a ferrocene carboxylic acid (0.9 s − 1). If the rate of reaction with oxygen with the enzyme is assumed to be 2 ×106 M − 1 s − 1, then the ‘relay rate’ would have to be ]5× 103 s − 1 if the sensor is placed in an air saturated solution ([O2] =240 mM). The observed rates are far short of this limit, again emphasizing the influence of oxygen on the relay process. The theory for electron transfer to enzymes through mediator layers has also been developed [110,111]. There are over 300 enzymes that employ either NAD(P) or NAD(P)H as a co-factor, and for this reason the issues surrounding both the immobilization of such enzymes (dehydrogenases) and the recycling of the NAD/NADH couple become important. Many electrochemical studies of NADH oxidation have been carried out following the classical work of Elving and co-workers in the 1970s [112]. The objective has been to find a modified electrode that will reduce the overpotential for NADH oxidation and prevent or minimize adsorption of oxidation products on the electrode surface. A number of mediators including phenoxazine derivatives [113], catechols [114] bipyridines, metal complexes [115], and hydroxybenzaldehydes [116] have been used for this purpose. A significant advantage of the latter materials, especially 3,4 dihydroxybenzaldehyde, is the property of improved stability and the possibility of electrodeposition to form a chemically modified electrode. The pseudo first order rate constants for reaction with NADH are in the range of 2.6×103 M − 1 s − 1. The above functionalities have been attached to polymers [117] that are then deposited on the electrode surface. There has been considerable discussion over the years as to whether the oxidation of NADH involves hydride transfer (one step) [118] or whether instead the process involves three steps (electron/proton/electron) but within one complex [119]. The goal of the chemically-modified electrode is to lower the barrier to electron transfer and prevent the dimerization of the one-electron oxidation product. The quinone-like structures offer the possibility of meeting
2638
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
Fig. 7. Schematic of Integrated lactate dehydrogenase enzyme electrode. Reference [122]a with permission.
such requirements. When NAD is used as a co-factor it is generally soluble and freely diffusing. Numerous attempts have been made to immobilize both the enzyme and the co-factor without notable success. However, Willner’s group has developed a scheme whereby the NAD is first immobilized via a PQQ-containing tether that serves, in effect, as a template for lactate dehydrogenase binding. After the enzyme is in place, it is then cross-linked with glutaraldehyde to stabilize the layer. The ordered structure is shown schematically in Fig. 7. This is the first clear example of a fully integrated and ‘reagentless’ system where NAD is a cofac-
tor. The stability is moderate and the apparent KM for lactate is about 3.6 mM. [120]. The oxidation of NADH can be carried out bioelectrocatalytically using an enzyme (diaphorase), and a variety of quinones or flavins have been employed as mediators [121].
6. Preparation of ordered protein layers The desire to couple electron transfer from an enzyme layer to an electrode has additionally focused attention on the ordered immobilization of protein. The
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
strategy is to build up successive layers of enzyme so that the resulting structure can be studied systematically. Four approaches have been employed: (1) Building of successive layers of enzyme by reaction of the enzyme with bifunctional linkers (tethers) [122]; (2) Use of biotinylated enzyme linked in successive layers using avidin [122–124]; (3) Linking of successive layers using electrostatic immobilization [110]; (4) Linking of successive layers using antibodies [125]. Fundamental studies based on these ordered to quasi-ordered structures have employed freely diffusing mediators as well as mediators covalently attached to the enzyme [126]. The anchoring of the tether to the electrode can block the surface thus reducing the effective electrode surface area available for amperometric detection of substrates. This can be serious problem especially when using microelectrodes. As a solution to this problem, Kuhr and co-workers [127] have developed techniques for preparing nanostructure domains (photopatterned lines about 5 mm wide) on carbon electrodes in which alternate lines are ‘bare’ electrode and tether anchor (biotin).
7. In-vivo electrochemistry The advances in electrochemical detection of species in biological milieu in the 1990s were based on a desire to ask more detailed questions about the real-time fluctuations of biologically relevant analytes. This necessitated the development of sensors with dimensions in the 1–10 mm range and with response times (90% maximum response) of a few seconds or less. Wightman and co-workers [128] observed current transients at microelectrodes that they attributed to the single vesicular release of epinephrine and norepinephrine from isolated adrenal medullary cells. The signal, obtained under constant potential amperometric conditions, is manifested as a series of spikes ranging from 0.2 to 2 pcoulombs (1–10 attomol catecholamine detected). One micron diameter electrodes were used to monitor exocytosis from a vesicle whose diameter is B200 nm. This work established that it was possible to directly monitor and time resolve secretion events at the single-cell level. Although the quantal hypothesis for neurotransmitter release is well accepted, caution must be taken in interpretation of results. The actual time-dependent events are convoluted by diffusion from the vesicle to the point of observation by the sensor. Catecholamine exocytosis in PC12 cells was monitored by Chen, Luo and Ewing [129]. Using a carbon electrode modified with ruthenium oxide/cyanoruthenate developed by Cox and Gray [130] Kennedy and co-workers [131] were able to detect exocytosis of insulin from beta cells. The sensor is not very specific for insulin, which it does, however, oxidize rapidly, but because there are few if
2639
any other proteins present, selectivity is not a major problem. In passing from the cell culture to the brain slice or intact brain as the measurement medium, several issues arise that must be considered. First, the insertion of the sensor into tissue creates tissue damage, which is minimized as the diameter of the sensor decreases. Implantation can result in the creation of a liquid layer immediately surrounding the sensor that has diffusional and other properties different from the tissue itself. Second, species diffuse through tissue necessarily more slowly, with the result that diffusion coefficients can be a factor of 3 – 4 smaller in tissue than in aqueous solution. This issue has been dealt with in the context of transport of choline in rat brain [132]. Finally, the sensor may lose sensitivity compared to its in-vitro value as a result of its contact with biological fluid. The observed signal due to neurotransmitter release may be distorted due to the finite response time of the sensor. If the response time is known it is possible to deconvolute the signal, taking the sensor properties into account [133]. Electrochemical measurements in the brain have been recently reviewed [134]. Much of the research involving in-vivo electrochemistry has exploited the electroactivity of small molecules such as catecholamines. Many, if not most, relevant species, are not electroactive, and therefore they must be determined indirectly. Because of its importance to the treatment of diabetes, glucose sensors have become the ‘holy grail’ in the biosensor area. There are many approaches being taken to the monitoring of blood glucose, but at present electrochemical detection appears the most tractable. Although the electrochemically-based glucose sensor has been known since the 1960s [135], it was not until the mid-1980s that feasibility of in-vivo monitoring was demonstrated [136]. This approach [137] has involved a needle sensor (enzyme electrode of about 250 mm diameter) that can be implanted in the subcutaneous tissue of the forearm or abdomen for periods of 4 – 7 days, after which they are removed and replaced. Such a sensor is shown in Fig. 8. The sensing element is built on a Pt – Ir wire substrate, followed by a permselective membrane to exclude electroactive interferences such as ascorbate, urate, and acetaminophen. Next the enzyme is immobilized, and this is followed by an external membrane having the property of being highly permeable to oxygen while exhibiting only limited permeability to glucose. These properties make it possible to prepare a sensor with a linear response to glucose over the required range (2 – 20 mM glucose) and response characteristics largely independent of oxygen partial pressure. The short-term implants give useful results but may need to be calibrated one to two times per day due to limited stability while the long-term implants currently suffer from destruction of the implanted instrument package by the remarkably corrosive elements within the tissue. An
2640
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
important alternative approach might involve a chronically implanted device that will be expected to function reliably for 6–12 months [138]. Other sensors have also been developed for brain studies including glucose [139], glutamate [140] and lactate [141]. The sensors implanted in the brain must be smaller, faster in response than those in the subcutaneous tissue, and free of interferences. For example, the glutamate sensor must detect 1 mM glutamate in the presence of 300–400 mM ascorbate. The latter electroactive interferent can be removed in two ways: with a permselective membrane or by using ascorbate oxidase, an enzyme that reduces oxygen to water and not peroxide, thus making it useful for ascorbate elimination. Other tissues have also been examined as the monitoring of lactate release in cardiac tissue has been demonstrated [142]. If oxidases are employed, it is essential to establish that low oxygen levels do not affect sensor response. Progress in sensor design and methodology has emphasized the value of time-dependent data that an electrochemical sensor can provide. Of perhaps even greater importance is the possibility of obtaining simultaneously data on different analytes. By monitoring glucose, lactate and oxygen simultaneously in the rat brain, it was possible to demonstrate the elevation of lactate levels that might otherwise have been confused with ischemia (oxygen deficiency) [141]. Multiple analyte measurements have also been reported, some combining optical and electrochemical detection [143].
Electrochemical detection has proven extremely valuable in monitoring inorganic species that play an important role in biological systems. These include oxygen, hydrogen peroxide, superoxide, and nitric oxide. Oxygen is ideally suited to electrochemical detection and such methodology was reported in the physiology literature as early as 1942 [144]. There is currently significant interest in oxygen in studies of oxidative stress [145]. For such studies it is necessary to know not simply concentrations, but fluxes in or out of cells. One such approach is the ‘self-referencing’ electrode [146]. The concept is to make measurements of the current due to oxygen at points outside a cell in culture that fall within the cell concentration gradient. By physically moving the 1 – 3 mm diameter electrode at about 0.3 Hz over a distance of 5 mm, two values of the oxygen concentration can be measured. From this information and an electrode calibration, the concentration gradient can be calculated. It is presumed that electrode motion does not disturb the gradient. The movement of the electrode has another advantage: it allows electrode noise to be eliminated through phase sensitive detection. Electrochemistry has provided some of the greatest excitement of the decade due to the possibility of measuring nitric oxide, voted the Molecule of the Year by Science. This very reactive species cannot be easily monitored by alternative methods. Malinski [147] reported on an electrode modified by a Ni(II) porphyrin. Although the oxidative electro-
Fig. 8. Schematic of implantable glucose sensor used for continuous monitoring in humans (see reference [137]).
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
chemistry is quite complicated, it appears that an anodic current proportional to NO can be obtained to detection limits in the range of about 10 nM. The biochemistry of this versatile molecule has been discussed in the context of electrochemical measurements [148]. Using a 30 mm platinum disk electrode and double differential pulse amperometry, it was possible to selectively measure hydrogen peroxide in the rat striatum in the presence of dopamine, ascorbic acid, and uric acid with a detection limit of 0.03 mM [149]. A two-enzyme system (superoxide dismutase and catalase) was used as part of a 50 mm disk microelectrode to measure hyperoxia (superoxide) in the striatum of a freely moving rat [150]. Thus far relatively little attention has been paid to acquisition of detailed understanding of how sensor performance is affected by contact with tissue and how the tissue is affected by the presence of the sensor. It is understood that the initial implantation of sensors can destroy capillaries and surrounding cells and this invasion triggers the wound healing process the first step of which involves protein adsorption on the sensor. With this in mind, membrane materials have been developed with ‘water-like surfaces’, that tend to minimize adsorption [151]. Biocompatibility studies on biosensors have been limited, and much needs to be learned about the interfacial reactions which occur.
2641
purine (adenosine and inosine) release during cardiac ischemia. To accomplish this detection it was necessary to employ a cascade of three enzymes: adenosine deaminase, nucleotide phosphorylase, and xanthine oxidase. By determining the charge generated from the oxidation of peroxide, the final product of the serial enzymatic reactions, femtomole levels of adenosine could be detected. The detection limit is about 120 amol. The technology lends itself easily to rapid screening techniques as employed within the pharmaceutical industry. Its virtue, the small volume of liquid, may produce concentrations of the analyte significantly higher than physiological levels, thus leading to experimental artifacts. Similarly, the effect of incubation of a single cell in elevated enzyme levels may also affect cell behavior. Presumably many, if not most, of these difficulties can be resolved by appropriate control experiments. Although the electrochemistry of nucleic acids has been studied for many years [157], there has been increased interest in oligonucleotide detection for diagnostic purposes [158]. The technique of oscillographic polarography has been revisited to permit the selective detection of pyrimidine-based nucleotide using a copper electrode and decoupling of the charging and faradaic current in the frequency domain [159].
9. Bioelectrochemistry in the new millennium 8. New electrochemical methodology as applied to bioelectrochemistry New technology had a significant impact on bioelectrochemistry during the 1990s. Bard’s group extended scanning tunneling microscopy to the use of the probe as an electrode (scanning electrochemical microscopy (SECM)) to detect the activity of glucose oxidase on a non-conducting surface [152]. The technique has subsequently been extended to other applications such as the detection of guard cells on a plant leaf [153] or the detection of carcinoembryonic antigen (CEA) microspotted on a plate and reacted with a CEA– horseradish peroxidase conjugate [154]. The method can detect as few as 100 molecules in a 20 mm spot. SECM has also been used to pattern active enzyme on a surface. This is accomplished by deactivating enzyme by oxidation of Cl− or Br− at the enzyme surface [155]. With increased emphasis on measurements involving single cells, it has proven useful to exploit microfabrication and micromachining techniques to produce wells containing electrodes that can accommodate single cells. Using this format, Bratten and co-workers [156] developed a 200 mm diameter well with electrodes in the bottom, with a depth of 20 mm and a total volume of 600 pL. The objective was to measure
The advances of the last 10 years have been significant and have focused strongly on the understanding of the interface between the electrode and the molecules reacting with it. We can certainly expect nanotechnology to heavily influence developments. As the interfacial chemistry is better understood it will be possible to more effectively promote biocatalysis and to further develop biochemical fuel cells [160]. Several groups have reported on the development of microelectrochemical enzyme transistors [161,162]. Amplification is achieved because peroxide generated by an enzymatic reaction alters the conductivity of a poly(aniline) film [162]. Lower detection limits are possible than with conventional sensors because the products of the reaction can accumulate. Miniaturized bioelectronic devices are also expected to proliferate. As more sensors are developed, it will be increasingly possible to simultaneously monitor several key analytes in real time. Finally, we can expect to see bioelectrochemical techniques used by a growing range of scientists.
Acknowledgements FAA acknowledges financial support from the UK Research Councils (EPSRC, BBSRC), Wellcome Trust,
2642
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
Leverhulme Trust, Royal Society and St. John’s College, Oxford. The partial support of the National Institutes of Health, (US) to GSW, grants DK30718 and DK55297, is gratefully acknowledged.
References [1] K. Niki, T. Yagi, H. Inokuchi, K. Kimura, J. Am. Chem. Soc. 101 (1979) 3335. [2] M.J. Eddowes, H.A.O. Hill, J. Chem. Soc. Chem. Commun. 771 (1977). [3] P. Yeh, T. Kuwana, Chem. Lett. 1145 (1977). [4] F.A. Armstrong, Struct. Bond. 72 (1990) 137. [5] P.M. Allen, H.A.O. Hill, N.J. Walton, J. Electroanal. Interface Electrochem. 178 (1984) 69. [6] Y. Taniguchi, K. Toyosawa, H. Yamaguchi, K. Yasukouchu, J. Chem. Soc. Chem. Commun. 1032 (1982). [7] H.A.O. Hill, D.J. Page, N.J. Walton, D.J. Whitford, J. Electroanal. Interface Electrochem. 187 (1987) 315. [8] L.S. Wong, V.L. Vilker, W.T. Yap, Y. Reipa, Langmuir 11 (1995) 4818. [9] B.D. Lamp, D. Hobara, M.D. Porter, K. Niki, T.M. Cotton, Langmuir 13 (1997) 736. [10] A.M. Bond, Anal. Proc. 29 (1992) 132. [11] Z. Salamon, G. Tollin, Bioelectrochem. Bioenerg. 26 (1991) 321. [12] Y. Sato, F. Mizutani, J. Electroanal. Chem. 438 (1997) 99. [13] K.J. Stevenson, M. Mitchell, H.S. White, J. Phys. Chem. 102 (1998) 1235. [14] M.J. Honeychurch, G.A. Rechnitz, J. Phys. Chem. 101 (1997) 7472. [15] F.A. Armstrong, J. Hirst, H.A. Heering, Chem. Soc. Rev. 26 (1997) 169. [16] S. Song, R.A. Clark, E.F. Bowden, M.J. Tarlov, J. Phys. Chem. 97 (1993) 6564. [17] F.A. Armstrong, P.A. Cox, H.A.O. Hill, V.J. Lowe, B.N. Oliver, J. Electroanal. Chem. Interface Electrochem. 217 (1987) 331. [18] J.L. Willit, E.F. Bowden, J. Phys. Chem. 94 (1991) 8241. [19] A. Manjaqui, J. Haladjian, P. Bianco, Electrochim. Acta 35 (1990) 177. [20] J.J. Davis, M.L.H. Green, H.A.O. Hill, Y.C. Leung, P.J. Sadler, A.V. Xavier, S.C. Tsang, Inorg. Chim. Acta 272 (1998) 261. [21] M. Collinson, E.F. Bowden, Anal. Chem. 64 (1992) 1470. [22] J. Li, G. Cheng, S. Dong, J. Electroanal. Interface Electrochem. 416 (1996) 97. [23] J.J. Davis, C.M. Halliwell, H.A.O. Hill, G.W. Canters, M.C. van Amsterdam, M.P. Verbeet, New J. Chem. 22 (1998) 1119. [24] D. Hobara, K. Niki, C. Zhou, G. Chumanov, T.M. Cotton, Colloids. Surf. B 93 (1994) 241. [25] S. Lecomte, H. Wackerbarth, T. Soulimane, G. Buse, P. Hildebrandt, J. Am. Chem. Soc. 120 (1998) 7381. [26] V. Reipa, A.K. Gaigalas, V.L. Vilker, Langmuir 13 (1997) 3508.
[27] K.L. Egodage, B.S. DeSilva, G.S. Wilson, J. Am. Chem. Soc. 119 (1997) 5295. [28] P.L. Edmiston, J.E. Lee, S.-S. Cheng, S.S. Saavedra, J. Am. Chem. Soc. 119 (1997) 560. [29] T.M. Nahir, E.F. Bowden, J. Electroanal. Chem. 410 (1996) 9. [30] R.A. Clark, E.F. Bowden, Langmuir 13 (1997) 559. [31] J.F. Rusling, Acc. Chem. Res. 31 (1998) 363. [32] J.F. Rusling, A.-E.F. Nassar, J. Am. Chem. Soc. 115 (1993) 11891. [33] (a) R. Lin, M. Bayachou, J. Greaves, P.J. Farmer, J. Am. Chem. Soc. 119 (1997) 12689; (b) M. Bayachou, R. Lin, W. Cho, P.J. Farmer, J. Am. Chem. Soc. 120 (1998) 9888. [34] Z. Zhang, A.-E.F. Nassar, Z. Lu, J.B. Schenckman, J.F. Rusling, J. Chem. Soc. Faraday Trans. 93 (1997) 1769. [35] A.-E.F. Nassar, Z. Zhang, N. Hu, J.F. Rusling, T.F. Kumosinski, J. Phys. Chem. B 101 (1997) 2224. [36] S. Boussaad, N.J. Tao, J. Am. Chem. Soc. 120 (1999) 4510. [37] Y.M. Lvov, Z. Lu, J.B. Schenckman, X. Zu, J.F. Rusling, J. Am. Chem. Soc. 120 (1998) 4073. [38] J. Kong, Z. Lu, Y.M. Lvov, R.Z.B. Desamero, H.A. Frank, J.F. Rusling, J. Am. Chem. Soc. 120 (1998) 7371. [39] Z. Salamon, J.T. Hazzard, G. Tollin, Proc. Natl. Acad. Sci. USA 90 (1993) 6420. [40] J.D. Burgess, M.C. Rhoten, F.M. Hawkridge, Langmuir 14 (1998) 2467. [41] Z.Q. Feng, S. Imabayashi, T. Kakiuchi, K. Niki, J. Chem. Soc. Faraday Trans. 93 (1997) 1367. [42] A.K. Gaigalus, G. Niaura, J. Colloid Interface Sci. 193 (1997) 60. [43] C.E.D. Chidsey, Science 251 (1991) 919. [44] H.O. Finklea, L. Liu, M.S. Ravenscroft, S. Punturi, J. Phys. Chem. 100 (1996) 18852. [45] J.F. Smalley, S.W. Feldberg, C.E.D. Chidsey, M.R. Linford, M.D. Newton, Y-P. Liu, J. Phys. Chem. 99 (1995) 13141. [46] J. Hirst, J.L.C. Duff, G.N.L. Jameson, M.A. Kemper, B.K. Burgess, F.A. Armstrong, J. Am. Chem. Soc. 120 (1998) 7085. [47] A.E. Kasmi, J.M. Wallace, E.F. Bowden, S.M. Binet, R.J. Linderman, J. Am. Chem. Soc. 120 (1998) 225. [48] A.-E.F. Nassar, W.S. Willis, J.F. Rusling, Anal. Chem. 67 (1995) 2386. [49] B.C. King, F.M. Hawkridge, B.M. Hoffman, J. Am. Chem. Soc. 114 (1992) 10603. [50] M. Tominaga, T. Kumagai, S. Takita, I. Taniguchi, Chem. Lett. 1771 (1993). [51] B.R. Van Dyke, P. Saltman, F.A. Armstrong, J. Am. Chem. Soc. 118 (1996) 3490. [52] P.L. Hagedoorn, M.C.P.F. Driessen, M. Van den Bosch, I. Landa, W.R. Hagen, FEBS Lett. 440 (1998) 311. [53] J.-B. Park, C. Fan, B.M. Hoffman, M.W.W. Adams, J. Biol. Chem. 266 (1991) 19351.
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645 [54] P.S. Breraton, M.F.J.M. Verhagen, Z.H. Zhou, M.W.W. Adams, Biochemistry 37 (1998) 7351. [55] M.T. Cruanes, K.K. Rodgers, S.G. Sligar, J. Am. Chem. Soc. 114 (1992) 9660. [56] L.D. Gille`s de Pe´lichy, E.T. Smith, Biochemistry 38 (1999) 7874. [57] S.E. Iismaa, A.E. Va´zquez, G.M. Jensen, P.J. Stephens, J.N. Butt, F.A. Armstrong, B.K. Burgess, J. Biol. Chem. 266 (1991) 21563. [58] B. Shen, D.R. Jollie, C.D. Stout, T.C. Diller, F.A. Armstrong, C.M. Gorst, G.N. La Mar, P.J. Stephens, B.K. Burgess, J. Biol. Chem. 269 (1994) 8564. [59] H.S. Gao-Sheridan, H.R. Pershad, F.A. Armstrong, B.K. Burgess, J. Biol. Chem. 273 (1998) 5514. [60] P. Kyritsis, O. Hatzfeld, T.A. Link, J.-M. Moulis, J. Biol. Chem. 273 (1998) 15404. [61] J.L.C. Duff, J.L.J. Breton, J.N. Butt, F.A. Armstrong, A.J. Thomson, J. Am. Chem. Soc. 118 (1996) 8593. [62] S.E.J. Fawcett, D. Davis, J.L. Breton, A.J. Thomson, F.A. Armstrong, Biochem. J. 335 (1998) 357. [63] J.N. Butt, F.A. Armstrong, J. Breton, S.J. George, A.J. Thomson, E.C. Hatchikian, J. Am. Chem. Soc. 113 (1991) 6663. [64] J.N. Butt, A. Sucheta, F.A. Armstrong, J. Breton, A.J. Thomson, E.C. Hatchikian, J. Am. Chem. Soc. 113 (1991) 8948. [65] J.N. Butt, S.E.J. Fawcett, J. Breton, A.J. Thomson, F.A. Armstrong, J. Am. Chem. Soc. 119 (1997) 9729. [66] J.N. Butt, A. Sucheta, F.A. Armstrong, J. Breton, A.J. Thomson, E.C. Hatchikian, J. Am. Chem. Soc. 115 (1993) 1413. [67] J. Haladjian, R. Pilard, P. Bianco, P.-A. Serre, Bioelectrochem. Bioenerg. 9 (1982) 91. [68] P.D. Barker, A.G. Mauk, J. Am. Chem. Soc. 114 (1992) 3619. [69] X. Yuan, S. Sun, F.M. Hawkridge, J.F. Chlebowski, I. Taniguchi, J. Am. Chem. Soc. 112 (1990) 5380. [70] B.A. Feinberg, X. Liu, M.D. Ryan, A. Schejter, C. Zhang, E. Margoliash, Biochemistry 37 (1998) 13091. [71] F.I. Rosell, J.C. Ferrer, A.G. Mauk, J. Am. Chem. Soc. 120 (1998) 11234. [72] H. Lopes, G.W. Pettigrew, I. Moura, J.J.G. Moura, J. Biol. Inorg. Chem. 3 (1998) 632. [73] J. Hirst, G.N.L. Jameson, J.W.A. Allen, F.A. Armstrong, J. Am. Chem. Soc. 120 (1998) 11994; J. Hirst, F.A. Armstrong, Anal. Chem. 70 (1998) 5062. [74] J.P. McEvoy, F.A. Armstrong, J. Chem. Soc. Chem. Commun. (1999) in press. [75] R.J. Cherry, A.J. Bjornsen, D.C. Zapien, Langmuir 14 (1998) (1971). [76] H.A. Heering, J. Hirst, F.A. Armstrong, J. Phys. Chem. B 102 (1998) 6889. [77] A. Onuoha, X. Hu, J.F. Rusling, J. Am. Chem. Soc. 119 (1997) 3979. [78] M.S. Mondal, D.B. Goodin, F.A. Armstrong, J. Am. Chem. Soc. 120 (1998) 6270. [79] H.A. Heering, M.S. Mondal, F.A. Armstrong, Anal. Chem. 71 (1999) 174.
2643
[80] K.L. Turner, M.K. Doherty, H.A. Heering, F.A. Armstrong, G.A. Reid, S.K. Chapman, Biochemistry 38 (1999) 3302. [81] A. Sucheta, B.A.C. Ackrell, B. Cochran, F.A. Armstrong, Nature 356 (1992) 361. [82] H.A. Heering, J.H. Weiner, F.A. Armstrong, J. Am. Chem. Soc. 119 (1997) 11628. [83] J. Hirst, A. Sucheta, B.A.C. Ackrell, F.A. Armstrong, J. Am. Chem. Soc. 118 (1996) 5031. [84] B.A.C. Ackrell, F.A. Armstrong, B. Cochran, A. Sucheta, T. Yu, FEBS Lett. 326 (1993) 92. [85] H.R. Pershad, J. Hirst, B. Cochran, B.A.C Ackrell, F.A. Armstrong, Biochim. Biophys. Acta (1999) in press. [86] J. Hirst, B.A.C. Ackrell, F.A. Armstrong, J. Am. Chem. Soc. 119 (1997) 7434. [87] P. Bianco, J. Haladjian, J. Electrochem. Soc. 139 (1992) 2428. [88] J.N. Butt, M. Filipiak, W.R. Hagen, Eur. J. Biochem. 245 (1997) 116. [89] H.R. Pershad, J.L.C. Duff, H.A. Heering, E.C. Duin, S.P.J. Albracht, F.A. Armstrong, Biochemistry 38 (1999) in press. [90] E. Cso¨regi, L. Gorton, G. Marko-Varga, Anal. Chim. Acta 273 (1993) 59. [91] J. Zhao, R.W. Henkens, J. Stonehuerner, J.P. O’Daly, A.L. Crumbliss, J. Electroanal. Chem. 327 (1992) 109. [92] Y. Zhang, G.S. Wilson, J. Electroanal. Chem. 345 (1993) 253. [93] S.B. Hall, E.A. Khudaish, A.L. Hart, Electrochim. Acta 43 (1997) 579; S.B. Hall, E.A. Khudaish, A.L. Hart, Electrochim. Acta 43 (1998) 2015; S.B. Hall, E.A. Khudaish, A.L. Hart, Electrochim. Acta 44 (1999) 2455. [94] G. Kenausis, C. Taylor, I. Katakis, A. Heller, J. Chem. Soc. Faraday Trans. 92 (1996) 4131. [95] I. Willner, V. Heleg-Shabtai, R. Blonder, A.F. Bu¨chmann, A. Heller, J. Am. Chem. Soc. 118 (1996) 10321. [96] R. Rajagopalan, A. Aoki, A. Heller, J. Phys. Chem. 100 (1996) 3719. [97] T.J. Ohara, R. Rajagopalan, A. Heller, Anal. Chem. 66 (1994) 2451. [98] H. Bu, A.M. English, S.R. Mikkelsen, Anal. Chem. 68 (1996) 3951. [99] J. Kulys, T. Buch-Rasmussen, K. Bechgaard, V. Razumas, J. Kazlauskaite, J. Marcinkeviciene, J.B. Christensen, H.E. Hansen, J. Mol. Catal. 91 (1994) 407. [100] P.N. Bartlett, S. Booth, D.J. Caruana, J.D. Kilburn, C. Santamaria, Anal. Chem. 69 (1997) 734. [101] J. Wang, F. Lu, J. Am. Chem. Soc. 120 (1998) 1048. [102] T.P. Henning, D.D. Cunningham, in: G. Ramsay (Ed.), Commercial Biosensors: Applications to Clinical, Bioprocess, and Environmental Samples, ch. 1, Wiley, New York, 1998. [103] P.N. Bartlett, J.M. Cooper, J. Electroanal. Chem. 362 (1993) 1. [104] W. Schuhmann, in: T. Cass, F.S. Ligler (Eds.), Immobilized Biomolecules in Analysis: A Practical Approach, ch. 11, Oxford University Press, New York, 1998.
2644
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645
[105] N.K. Cenas, J. Kulys, J. Bioelectrochem. Bioenerg. 8 (1981) 103. [106] B.S. Hill, C.A. Scolari, G.S. Wilson, J. Electroanal. Chem. 252 (1988) 125. [107] G.F. Khan, M. Ohwa, W. Wernet, Anal. Chem. 68 (1996) 2939. [108] C. Bourdillon, C. Demaille, J. Moiroux, J-M. Save´ant, J. Am. Chem. Soc. 115 (1993) 2. [109] A. Badia, R. Carlini, A. Fernandez, F. Battaglini, S.R. Mikkelsen, A.M. English, J. Am. Chem. Soc. 115 (1993) 7053. [110] J. Hodak, R. Etchenique, E.J. Calvo, K. Singhal, P.N. Bartlett, Langmuir 13 (1997) 2708. [111] C. Danilowicz, R. Etchenique, E.J. Calvo, L. Diaz, Anal. Chem. 68 (1996) 4186. [112] J. Moiroux, P.J. Elving, Anal. Chem. 50 (1978) 1056. [113] B. Persson, L. Gorton, J. Electroanal. Chem. 292 (1990) 115. [114] H. Jaegfeldt, T. Kuwana, G. Johansson, J. Am. Chem. Soc. 105 (1983) 1805; C. Ueda, D.C.S. Tse, T. Kuwana, Anal. Chem. 54 (1982) 850. [115] Q. Wu, M. Maskus, F. Pariente, F. Tobalina, V.M. Ferna´ndez, E. Lorenzo, H.D. Abrun˜a, Anal. Chem. 68 (1996) 3688. [116] F. Pariente, F. Tobalina, M. Darder, E. Lorenzo, H.D. Abrun˜a, Anal. Chem. 68 (1996); E. Lorenzo, F. Pariente, L. Herna´ndez, F. Tobalina, M. Darder, Q. Wu, M. Maskus, H.D. Abrun˜a, Biosensors Bioelectron. 13 (1998) 319. [117] B. Persson, H.L. Lan, L. Gorton, Y. Okamoto, P.D. Hale, L.I. Boguslavsky, T.A. Skotheim, Biosens. Bioelectron. 8 (1993) 81; B. Persson, H.S. Lee, L. Gorton, T. Skotheim, P. Bartlett, Electroanalysis 7 (1995) 935. [118] B.W. Carlson, L.L. Miller, J. Am. Chem. Soc. 107 (1985) 479. [119] A. Anne, J. Moiroux, J-M. Save´ant, J. Am. Chem. Soc. 115 (1993) 10224. [120] A. Bardea, E. Katz, A.F. Bu¨ckmann, I. Willner, J. Am. Chem. Soc. 119 (1997) 9114. [121] Y. Ogino, K. Takagi, K. Kano, T. Ikeda, J. Electroanal. Chem. 396 (1995) 517. [122] (a) I. Willner, E. Katz, B. Willner, Electroanalysis 9 (1997) 965; (b) W. Jin, F. Bier, U. Wollenberger, F. Scheller, Biosens. Bioelectron. 10 (1995) 823. [123] P. Pantano, T.H. Morton, W.G. Kuhr, J. Am. Chem. Soc. 113 (1832) 1991. [124] N. Anicet, C. Bourdillon, J. Moiroux, J-M. Save´ant, J. Phys. Chem. B 102 (1998) 9844. [125] C. Bourdillon, C. Demaille, J. Moiroux, J-M. Save´ant, Acc. Chem. Res. 29 (1996) 529. [126] N. Anicet, A. Anne, J. Moiroux, J-M. Save´ant, J. Am. Chem. Soc. 120 (1998) 7115. [127] N. Dontha, W.B. Nowall, W.G. Kuhr, Anal. Chem. 69 (1997) 2619; W.B. Nowall, D.O. Wipf, W.G. Kuhr, Anal. Chem. 70 (1998) 2601; W.B. Nowall, N. Dontha, W.G. Kuhr, Biosens. Bioelectron. 13 (1998) 1237. [128] R.M. Wightman, J.A. Jankowski, R.T. Kennedy, K.T. Kawagoe, T.J. Schroeder, D.J. Leszczyzyn, J.A. Near, E.J. Dilberto Jr., O.H. Viveros, Prod. Natl. Acad. Sci. USA 88 (1991) 10754.
.
[129] T.C. Chen, G. Luo, A.G. Ewing, Anal. Chem. 66 (1994) 3031. [130] J.A. Cox, T.J. Gray, Anal. Chem. 61 (1989) 2462. [131] L. Huang, H. Shen, M.A. Atkinson, R.T. Kennedy, Proc. Natl. Acad. Sci. USA 92 (1995) 9608. [132] Q. Xin, R.M. Wightman, Brain Res. 776 (1997) 126. [133] M.A. Bunin, R.M. Wightman, Methods Enzymol. 296 (1998) 689. [134] R.D. O’Neill, J.P. Lowry, M. Mas, Crit. Rev. Neurobiol. 12 (1998) 69. [135] L.C. Clark Jr., C. Lyons, Ann. New York Acad. Sci. 102 (1962) 29; S.J. Updike, G.P. Hicks, Nature 214 (1967) 986. [136] M. Shichiri, Y. Yamasaki, N. Hakui, H. Abe, Lancet 2 (1982) 1129. [137] V. Poitout, D. Moatti-Sirat, G. Reach, Y. Zhang, G.S. Wilson, F. Lemonnier, J-C. Klein, Diabetologia 36 (1993) 658. [138] B.J. Gilligan, R.K. Rhodes, M.C. Shults, S.J. Updike, Diabetes Care, 17 (1994) 882; J.C. Armour, J.Y. Lucisano, B.D. McKean, D.A. Gough, Diabetes 39 (1990) 1519. [139] Y. Hu, G.S. Wilson, J. Neurochem. 68 (1997) 1745. [140] Y. Hu, K.M. Mitchell, F.M. Albahadily, E.K. Michaelis, G.S. Wilson, Brain Res. 659 (1994) 117; M.R. Ryan, J.P. Lowry, R.D. O’Neill, Analyst 122 (1997) 1419. [141] Y. Hu, G.S. Wilson, J. Neurochem. 69 (1997) 1484. [142] S.A.M. Marzouk, V.V. Cosofret, R.P. Buck, H. Yang, W.E. Cascio, S.S.M. Hassan, Anal. Chem. 69 (1997) 2646. [143] Q. Xin, R.M. Wightman, Anal. Chem. 70 (1998) 1677. [144] P.W. Davies, F. Brink, Rev. Sci. Instr. 13 (1942) 524. [145] S. Arbault, P. Pantano, J.A. Jankowski, M. Vuillaume, C. Amatore, Anal. Chem. 67 (1995) 3382. [146] S.C. Land, D.M. Porterfield, R.H. Sanger, P.J.S. Smith, J. Exp. Biol. 202 (1999) 211. [147] Z. Taha, F. Kiechle, T. Malinski, Biochem. Biophys. Res. Commun. 188 (1992) 734. [148] T. Malinski, S. Mesaros, P. Tomboulian, Methods Enzymol. 268 (1996) 58. [149] H. Yokoyama, N. Tsuchihashi, N. Kasai, T. Matsue, I. Uchida, N. Mori, H. Ohya-Nishiguchi, H. Kamada, Biosens. Bioelectron. 12 (1997) 1037. [150] H. Yokoyama, N. Kasai, T. Matsue, N. Tsuchihashi, I. Ushida, T. Ogata, N. Mori, H. Ohya-Nishiguchi, H. Kamada, J. Brain Sci. 23 (1997) 181. [151] C.A.P. Quinn, R.E. Connor, A. Heller, Biomaterials, 18 (1997) 1665; K. Ishihara, H. Oshida, Y. Endo, T. Ueda, A.Watanabe, N. Nakabayashi, J. Biomed. Mater. Res. 26 (1992) 1543. [152] D.T. Pierce, P.R. Unwin, A.J. Bard, Anal. Chem. 64 (1992) 1795. [153] M. Tsionsky, Z.G. Cardon, A.J. Bard, R.B. Jackson, Plant Physiol. 113 (1997) 895. [154] H. Shiku, T. Matsue, I. Uchida, Anal. Chem. 68 (1996) 1276. [155] H. Shiku, T. Takada, H. Yamada, T. Matsue, I. Uchida, Anal. Chem. 67 (1995) 312.
F.A. Armstrong, G.S. Wilson / Electrochimica Acta 45 (2000) 2623–2645 [156] C.D.T. Bratten, P.H. Cobbold, J.M. Cooper, Anal. Chem. 70 (1998) 1164. [157] E. Palecˇek, Electroanalysis 8 (1996) 7. [158] S.R. Mikkelsen, Electroanalysis, 8 (1996) 15; J. Wang, G. Rivas, X. Cai, E. Palecˇek, P. Nielsen, D. Grant, M. Ozsoz, M. Flair, Anal. Chim. Acta 347 (1997) 1.
.
2645
[159] P. Singhal, W.G. Kuhr, Anal. Chem. 69 (1997) 3552. [160] I. Willner, E. Katz, F. Patolsky, A.F. Bu¨ckmann, J. Chem. Soc. Perkin Trans. 2 (1998) 1817. [161] T. Matsue, M. Nishizawa, T. Sawaguchi, I. Uchida, J. Chem. Soc. Chem. Commun. (1991) 1029. [162] P.N. Bartlett, P.R. Birkin, J.H. Wang, F. Palmisano, G. De Benedetto, Anal. Chem. 70 (1998) 3685.
