Dynamic electrochemical investigations of hydrogen oxidation and production by enzymes and implications for future technology

Abstract

This tutorial review decribes studies of hydrogen production and oxidation by biological catalysts—metalloenzymes known as hydrogenases—attached to electrodes. It explains how the electrocatalytic properties of hydrogenases are studied using specialised electrochemical techniques and how the data are interpreted to allow assessments of catalytic rates and performance under different conditions, including the presence of O₂, CO and H₂S. It concludes by drawing some comparisons between the enzyme active sites and platinum catalysts and describing some novel proof-of-concept applications that demonstrate the high activities and selectivities of these ‘alternative’ catalysts for promoting H₂ as a fuel.

---

The authors

From left to right: Gabrielle Goldet graduated with an MChem from Durham University in 2004 and is now completing her DPhil with Fraser Armstrong at St John’s College, Oxford. Dr Kylie Vincent is Royal Society University Research Fellow and RCUK Academic Fellow in the Inorganic Chemistry Laboratory at the University of Oxford, and is a Fellow of Jesus College, Oxford. She is a graduate of the University of Melbourne, Australia, where she completed a BA/BSc (Hons), and a PhD jointly with Stephen Best (Melbourne) and Chris Pickett (Norwich, UK). She carried out postdoctoral work with Fraser Armstrong from 2002–2007 and was a Research Fellow at Wadham College during this time. Her research interests include the application of electrochemical and spectroelectrochemical methods to biological systems, in particular enzymes involved in energy cycling. James Cracknell graduated with an MChem from Corpus Christi College, University of Oxford in 2005, and is now completing his DPhil with Fraser Armstrong at St John’s College, Oxford. Fraser Armstrong is Professor of Chemistry at Oxford University and is Fellow of St. John’s College. His interests are in biological redox chemistry, in particular the application of dynamic electrochemical techniques in studies of complex electron-transfer and catalytic reactions in proteins, and most recently the mechanisms and exploitation of biological hydrogen cycling. He was elected a Fellow of the Royal Society in 2008. Dr Alison Parkin completed her doctoral studies under the supervision of Fraser Armstrong in 2008 and is now a Junior Research Fellow at Merton College, Oxford. Natalie Belsey graduated with an MChem from Lincoln College, University of Oxford in 2005 and is now completing her DPhil with Fraser Armstrong at Lincoln College, Oxford. Dr Erwin Reisner performed his doctorate work as a joint project between the University of Vienna (Professor Bernhard Keppler) and the Technical University of Lisbon (Professor Armando Pombeiro). He then joined the laboratory of Professor Stephen Lippard at MIT as a Schrödinger Postdoctoral Fellow. Currently, he is a Research Assistant with Fraser Armstrong interested in solar H2 production. Annemarie Wait graduated with an MChem from St Hilda’s College, University of Oxford in 2007, and is now completing her DPhil with Fraser Armstrong at Merton College, Oxford.

Introduction

One of the most promising directions for chemistry today is the development of ways to capture and store renewable energy (meaning, most obviously, sunlight). Hydrogen is one answer as there are infinite resources of water available to be ‘energised’ to form H₂ (energy content over 100 kJ per gram) using direct solar energy or surplus electricity. Generation of H₂ from water and its reoxidation are (at first glance) simple reactions, but H₂ is a rather stable and inert molecule (the H–H bond energy is 436 kJ mol^–1) and it is difficult to produce efficiently. In practice, many aspects of the clean and efficient production, storage and oxidation of H₂ are totally unresolved at the molecular level. Regarding almost insurmountable scale-up demands, we should be reminded that the nuclear age grew out of the serendipitous detection of trace BaSO₄ precipitates in neutron-irradiated U(VI) solutions by Meitner and Hahn in 1938. In this tutorial review we describe how to extract detailed information on how the H₂ molecule is manipulated very efficiently in biology, at the active sites of metalloenzymes. In ‘learning from biology’ we believe it would be hard to find a better example of a ‘bottom up’ approach (from the molecular level) to such a large and daunting problem.

As a fuel, H₂ was familiar as a 50% component of coal gas, once piped to homes in many industrialised nations. More than 99% of the world’s H₂ is still produced by steam reforming of fossil fuels, and most is used directly by the chemical industry, rather than as a fuel. Currently, production of H₂ by electrolysis at Fe and Ni electrodes is costly in terms of energy as it involves high temperatures and a voltage of 2 V or more. Most fuel cells operating at temperatures below 100 °C use Pt as electrocatalysts, but Pt is not an unlimited resource and even the best efforts to maximise the efficiency of its use will not be enough. Neither is Pt the perfect catalyst for H₂ cycling, because it is poisoned by H₂S and CO which are contaminants of H₂ from steam reforming. Alternative catalysts are therefore highly desirable, both to produce and oxidise H₂, and one important question is: ‘what can we learn from biology?’. The reason for the interest in biology is that H₂ is cycled and used efficiently by microbial organisms, in processes such as photosynthesis and respiration. The enzymes responsible are known as hydrogenases¹ and they were first discovered in the 1930s. Hydrogenases in certain organisms must have high affinity for H₂ and scavenge it from the atmosphere or local zones, whereas other hydrogenases are more active in producing H₂. Hydrogenases are found in pathogens, and H₂ may be a useful disease marker.²

Although hydrogenases are among the most ancient of enzymes, they represent a ‘new’ class of catalyst—complexes of abundant first-row transition elements that are potentially as active as Pt. The structures of the main classes of hydrogenases were solved in the 1990s,³ and we describe these briefly below. In our studies^4–6 of hydrogenases by electrochemical methods, our aim is to develop a good understanding at the atomic-mechanistic level. We measure their performance as catalysts, not only in simple terms of rates but also in terms of their abilities to fend off inactivation by molecules such as O₂, CO, H₂S, and even H₂ itself which is often an inhibitor of its own production. The information obtained is helpful in assessing the viability of organisms for ‘H₂ farms’, new alternative catalysts for H₂ production and fuel cells, and niche applications for the enzymes themselves. We and others have studied hydrogenases from a range of bacterial sources, and Table 1 summarises details and abbreviations used in this review.

Table 1 Key details of selected hydrogenases and the abbreviations used in this review

Bacterium	Type of organism	Hydrogenase studied by PFV (Abbreviation)
Allochromatium vinosum	Purple photosynthetic sulfur metabolizer, facultative anaerobe	Membrane-bound [NiFe]-hydrogenase
		(Av [NiFe]-MBH)
Clostridium acetobutylicum	Fermentative anaerobe	HydA—[FeFe]-hydrogenase
		(Ca [FeFe]-H)
Desulfomicrobium baculatum	Sulfate reducer, facultative anaerobe	Periplasmic [NiFeSe]-hydrogenase
		(Db [NiFeSe]-H)
Desulfovibrio desulfuricans	Sulfate reducer, facultative anaerobe	Periplasmic [FeFe]-hydrogenase
		(Dd [FeFe]-H)
Desulfovibrio fructosovorans	Sulfate reducer, facultative anaerobe	Periplasmic [NiFe]-hydrogenase
		(Df [NiFe]-H)
Desulfovibrio gigas	Sulfate reducer, facultative anaerobe	Periplasmic soluble [NiFe]-hydrogenase
		(Dg [NiFe]-H)
Desulfovibrio vulgaris Miyazaki F	Sulfate reducer, facultative anaerobe	Membrane-bound [NiFe]-hydrogenase
		(Dv [NiFe]-MBH)
Ralstonia eutropha H16	Knallgas bacterium, aerobe	Membrane-bound [NiFe]-hydrogenase
		(Re [NiFe]-MBH)
Ralstonia metallidurans CH34	Knallgas bacterium, aerobe	Membrane-bound [NiFe]-hydrogenase
		(Rm [NiFe]-MBH)

---

Key features of the active sites of hydrogenases are shown in Fig. 1, which includes the structure of an entire enzyme molecule.³ The two main classes of enzyme are known as [FeFe]- or [NiFe]-hydrogenases, depending on the metal content of the active site. Some, known as membrane-bound hydrogenases (MBHs), are components of energy chains, although they are easily solubilised by detergents or by loss of the hydrophobic membrane anchor domain. The minimum common factor in hydrogenases studied to date is a [Fe(CO)(CN)(RS)] unit, bridged to either Ni or a second Fe by thiolate ligands. These fragile active sites are protected by the surrounding protein and serviced by a chain of Fe–S clusters that transfer electrons to and from the enzyme surface. It is widely accepted that H₂oxidation and production involve heterolytic mechanisms in which H^– and H⁺ are stabilised by metal coordination and Brønsted base, respectively. Different cycles may be used, depending on the direction of catalysis.^4,5

Fig. 1 Hydrogenases and their active sites and relays. Left: Ribbon representation of the X-ray determined structure of Dg [NiFe]-H (PDB code 1YQ9). Right: The structures of the active sites. A shows the active site of Db [NiFeSe]-H in its active state (PDB code 1CC1), B shows the active site of Dg [NiFe]-H in the oxidised inactive ‘Ni-A’ state (PDB code 1YQ9) in which an inhibitory peroxide is present in the bridging position, and C shows the active site (the ‘H-cluster’) of Dd [FeFe]-H (PDB code 1HFE) in its active form.

Dynamic electrochemical techniques in hydrogenase research

Equipment

In protein film voltammetry (PFV), a minuscule sample of protein is adsorbed directly on an electrode to give a mono- or submonolayer film.^6–8 In this state the enzyme’s catalytic properties are fully expressed, and facile, reversible electron transfer occurs with the electrode. The enzyme’s catalytic activity is now controlled by the electrode potential and simultaneously monitored through the current that is observed.^5,7,8 The steady-state current is a direct measure of the catalytic turnover rate at any given electrode potential. Because the enzyme is immobilised on the electrode, the environment (a few mL of solution) is easily switched from one composition to another. Precise hydrodynamic control is achieved by rotating the electrode at variable high speeds. Rotation draws substrate to the electrode and it is easy to see if a reaction is diffusion controlled; likewise, products are swept away, allowing product inhibition to be measured and controlled. Contrary to expectation, in our experience, rotating at high speeds does not normally cause problems due to increased instability or loss of enzyme from the electrode surface. The technique is sensitive to trace reagents (<nM) because the electrochemical response at a typical electrode (area <0.1 cm²) reflects the action of much less than a picomole of active enzyme. For experiments on hydrogenases (and other enzymes acting on gaseous substrates) we use an all-glass electrochemical cell, fitting snugly against the electrode rotator.⁶ A precise mixture of gases can be passed through and solutions can be injected through a septum; gases can subsequently be removed by flushing with another gas. The experimental reference electrode we have used is the saturated calomel electrode (SCE): potentials are quoted relative to the standard hydrogen electrode (SHE) by adding +242 mV at 25 °C. The sealed apparatus, illustrated in Fig. 2, gives much improved results compared to open cells in which gases are introduced simply by bubbling into the solution.

Fig. 2 A schematic diagram of the apparatus used for protein film voltammetry experiments. The cell fits tightly against an electrode rotator and is equipped with gas flow controllers for providing accurate mixtures of gases.

So far, most PFV studies on hydrogenases have used a carbon electrode, particularly pyrolytic graphite cut so that the ‘edge’ surface rather than the hydrophobic basal plane contacts the solution. This is known as a PGE electrode. Polishing produces a rough, oxidised surface that is effective for non-covalent hydrogenase adsorption. In some cases, polyions such as polymyxin (a polycation) pre-adsorbed on the electrode or mixed with enzyme solution, increase both current and stability.⁹ Efforts are being made to establish procedures for achieving more permanent, covalent links between hydrogenases and electrode surfaces.¹⁰Hydrogenase electrocatalysis has been observed on carbon microelectrodes, nanotubes and even Au modified with polymyxin.⁹ Semiconductors such as TiO₂ have also been successfully tested and are now under investigation for photo H₂ production.¹¹

Interpreting data

For hydrogenases, two main types of experiment are employed to measure catalytic electron transport. The first of these is cyclic voltammetry, which involves cycling the electrode potential between two limiting values and monitoring the resulting current. This is useful for gaining an overall picture of the catalytic activity of the enzyme in each direction and for identifying potentials at which there is an interesting change in catalytic current, caused, for example, by a transition between active and inactive states. The second method is chronoamperometry, in which current is monitored as a function of time, after initiating the reaction, for instance by stepping the electrode potential to a particular value or changing the composition of the gas entering the cell.

Oxidation processes produce a positive current whereas reduction processes give a negative current. The voltammetry of a hydrogenase usually shows three distinct regions, as shown in Fig. 3. At low potentials, the region of negative current (zone 1) corresponds to H⁺ reduction (H₂ production) whereas at higher potential there are typically two regions of positive current (zones 2 and 3) each corresponding to net H₂oxidation activity. Under conditions where the enzyme catalyses both H₂oxidation and H₂ production, the current trace cuts the zero-current line cleanly at the thermodynamic potential of the 2H⁺/H₂ couple. This observation shows that hydrogenases operate ‘reversibly’ and require only the smallest overpotential to work in either direction (as with Pt, see later). The magnitudes of the currents achieved, once a sufficiently high overpotential is applied, reflect the inherent activities of the active sites. Taking pH and H₂ partial pressure into consideration, comparisons of the currents achieved in zones 1 and 2 quantify the catalytic bias of a hydrogenase for catalysing H₂ production relative to H₂oxidation. At high potential, zone 3, hydrogenases undergo oxidative inactivation;^5,8,12,13 and although this occurs over a large range of rates (depending on the hydrogenase) it is usually reversible and results in an oxidised ‘resting’ state. This is considered in greater detail below.

Fig. 3 Cyclic voltammograms comparing the catalytic activities, catalytic bias and extent of anaerobic inactivation of different hydrogenases as a function of potential. Experiments were conducted on films of Dd [FeFe]-H (10 °C, pH 6.0, scan rate 50 mV s⁻¹), Rm [NiFe]-MBH (30 °C, pH 5.5, scan rate 20 mV s⁻¹) and Db [NiFeSe]-H (25 °C, pH 6.0, scan rate 10 mV s⁻¹) under Ar (grey lines) and H₂ (black lines). The electrode was rotated at a constant rate ≥2500 rpm in each case. The results show that the [FeFe]-H is the most biased towards H₂ production and that the [NiFeSe]-H is the least prone to anaerobic inactivation.

In addition to catalysis, it is sometimes possible to observe discrete ‘non-turnover’ signals due to reversible electron transfer at redox centres. This requires a high electroactive coverage (above a few picomoles per square cm) and an inhibitor (such as CO) to block catalytic electron flow. Non-turnover signals consist of a pair of peaks, corresponding to oxidation and reduction, centred at the reduction potential of the redox centre, in this case, one or more of the FeS-clusters of the intramolecular relay.¹⁴ The signals can be analysed to obtain estimates of electron-transfer rates and electroactive coverage, the latter being necessary to convert current into turnover frequency per molecule (k_cat).⁵

Some non-idealities must be mentioned. First, long-term stability is rarely achieved. The problem we refer to as ‘film loss’ involves enzyme molecules desorbing from the electrode or even unfolding, and it is a major reason for efforts that are underway to attach hydrogenases to the electrode by covalent bonds.¹⁰ Second, an effect that influences the shape of voltammograms is dispersion—inhomogeneity among the environments and orientations of the enzyme molecules in the film. A range of orientations of the enzyme with respect to the electrode surface gives rise to a spread of interfacial (enzyme-electrode) electron-transfer rate constants. At sufficiently high scan rates, an ideal voltammogram should show a constant current plateau corresponding to the enzyme turning over at its maximum possible rate. In real experiments, with the electrode rotating rapidly to remove mass-transport control, the voltammograms at high overpotential rarely reach a plateau but instead show a residual slope (a linear potential dependence), as is most evident in Fig. 3 (right-hand panel). Detailed studies¹⁵ of this potential dependence for some [NiFe]-hydrogenases show that the gradient of the residual slope is related to the degree of dispersion and to the ratio between the rate of enzyme turnover (k_cat) and the (exchange) rate of interfacial electron transfer (k₀)—a high ratio leading to greater slope. Raising the temperature results in a more linear waveform because chemical transformations at the active site (k_cat) should have a higher activation energy (rate-determining at low temperature) than long-range interfacial electron transfer (k₀) (rate limiting at high temperature). In hydrogenases, intramolecular electron transfer along the FeS relay is fast, yet can be slowed down when individual FeS clusters are altered by site directed mutagenesis.¹⁶

Reactions of hydrogenases with small molecules

Quantifying the affinity of a hydrogenase for H₂

The affinity of hydrogenases for H₂ is expressed most simplistically by the Michaelis constant (). A useful electrochemical method for determining reported by Léger et al.¹⁷ is based upon the fact that when a gas is introduced into the cell solution, it can (unlike non-volatiles) be removed simply by flushing with an inert gas. The key point is that removal of the gas from solution follows a simple exponential time course. By measuring current as a function of time during the gas exchange, it is possible to extract not only K_M values but also (as described later) inhibition constants (K_I) for gaseous inhibitors such as CO.

Results of a typical experiment to determine for Re [NiFe]-MBH are shown in Fig. 4. A PGE electrode modified with the enzyme was placed in a sealed glass electrochemical cell containing 2 mL buffer solution. The electrode, rotating at high speed, was poised at a H₂-oxidising potential and the cell was flushed with N₂ to obtain a ‘background’ current. At time t = 0, 2 mL of H₂-saturated buffer (of identical composition and temperature to that already in the cell) was injected into the cell, causing a steep rise in current due to H₂oxidation. The H₂ concentration in the solution immediately begins to fall from its initial concentration of 0.4 mM as N₂ is flushed through the headspace of the cell (the H₂ concentration as a function of time is also indicated in Fig. 4), and the current decreases in a sigmoidal manner. The early part of the trace corresponds to the activity remaining fairly constant so long as the concentration of H₂ is close to the level needed to saturate the active site; thus, even at the qualitative level, this experiment provides a rapid appraisal of the magnitude of .

Fig. 4 Experiments to measure for wild-type Re [NiFe]-MBH at a potential of –108 mV vs.SHE. At time = 0 s, 2 mL of H₂-saturated buffer (pH 5.5, 30 °C) was injected into the cell, which originally contained 2 mL of N₂-saturated electrolyte of the same composition and temperature. Current vs. time traces are shown (bold trace), and overlaid on these are the H₂ concentration which decreases exponentially (thin trace) and the sigmoidal fit (dotted trace). The electrode was rotated at a constant rate of 4500 rpm. The extended ‘plateau region’ at the start of the experiment typifies an enzyme with a high affinity for H₂. Figure reproduced from ref. 18 with permission (Copyright 2008, American Society for Biochemistry and Molecular Biology).

For a quantitative analysis, Léger and co-workers derived eqn (1) for the sigmoidal variation of current i(t) with time t following injection of H₂ into the solution to give an initial concentration C_H2(0). The initial current is i_max and τ is the time constant for exponential gas removal (τ is typically in the range 30–100 s, depending on solution volume, rate of gas flow, and electrode rotation rate, which are constants for each experiment).

This equation was originally solved in its linear (logarithmic) form, but we have instead used linear regression analysis to fit data. Fig. 4 includes a best fit to the experimental data, allowing calculation of , which, from the average of multiple experiments, was determined as 6.1 µM at 30 °C, decreasing to 1.2 µM at 20 °C.¹⁸ These values are consistent with the requirement for hydrogenases from organisms that sequester low levels of H₂ to have low values.

A number of factors, both experimental and fundamental, may cause the current vs. time trace to deviate from the sigmoidal shape. Trace levels of H₂ diffusing back from the cell side arm or the glovebox itself hinder measurement of very low values because the assumption is that the H₂ level in the cell tends to zero over time. Substrate transport limitation, unaccounted for in the assumptions, becomes unavoidable at extremely low levels of H₂, even at high rotation rates and a low density of enzyme on the electrode. Care must also be taken to avoid measurements in a potential region corresponding to zone 3 (Fig. 3) because the current will decrease as the active site converts to an inactive form, and this transformation may become faster with decreasing H₂ concentrations. The method is not useful for measuring high values of because if the initial H₂ level is not sufficiently high relative to the current drops as soon as H₂ begins to be removed, instead of following a sigmoidal trace, preventing determination of i_max.

We have adapted Léger’s method to measure values for a series of mutants of Re [NiFe]-MBH and for the analogous Rm [NiFe]-MBH.¹⁸ All mutations of residues close to but not coordinating the [NiFe]-center have so far resulted in increases in , consistent with the idea that these enzymes have become optimised for H₂ scavenging. Similar results have recently been reported by Léger and co-workers for variants of Df [NiFe]-H.¹⁹

High potential anaerobic inactivation

We consider next how to study the anaerobic inactivation reactions that are often manifest in the most oxidising region (zone 3) of the voltammograms. Reversible, anaerobic inactivation at high potential has been studied in detail for [NiFe]-hydrogenases, in particular Av [NiFe]-MBH.¹² Anaerobic oxidation of [NiFe]-hydrogenases produces a Ni(III) state known as Ni-B, in which a hydroxide ligand is coordinated to the Ni in a bridging position with respect to the Fe(II).^12,20 This state, which can also be formed upon reaction with O₂, is rapidly re-activated by H₂, hence it is also known as the ‘Ready’ (or Ni_r*) state. The [FeFe]-hydrogenases also undergo anaerobic inactivation, in this case resulting in a state known as H_ox^inact in which the H-cluster valences are assigned as Fe(II)Fe(II)-[4Fe–4S]_ox.²¹ Again, H_ox^inact can be reactivated by H₂. Based on our studies of Av [NiFe]-MBH and several other [NiFe]-hydrogenases, and the Dd [FeFe]-H, we take the following observations as a standard against which other hydrogenases can be compared. Some main comparisons are illustrated by the results shown in Fig. 5.

Fig. 5 Comparison of anaerobic inactivation of hydrogenases under different conditions (the direction of potential cycling is indicated by the arrows); Re [NiFe]-MBH under atmospheres of 100% H₂ and 10% H₂ in N₂ at pH 5.5 (30 °C, electrode rotation rate 4500 rpm, 10 mV s^–1), Av [NiFe]-MBH under 100% H₂ at pH 6.0 and pH 9.0 (50 °C, electrode rotation rate 2500 rpm, 0.3 mV s^–1), Dd [FeFe]-H under 100% H₂ at pH 4.8 and pH 8.4 (10 °C, electrode rotation rate 2500 rpm, 0.3 mV s^–1). For Re [NiFe]-MBH, the response of an unmodified electrode recorded under the same conditions is overlaid. The hysteresis in zone 3 which is clearly evident in some experiments indicates where reactivation is much faster than inactivation.

A unique feature of electrochemical methods is the ability to resolve different kinetic components of complex redox reactions. Cyclic voltammetry is supported here by chronoamperometry experiments in which transformations are initiated by stepping the electrode potential between different regions. This separates the potential domain from the time domain and allows us to distinguish between reaction steps that respond directly to electrochemical driving force and those that are controlled by other factors. Anaerobic inactivation, which is usually much slower than re-activation, is described in electrochemists’ terms as a ‘CE’ reaction: here, an initial chemical reaction or rearrangement (C), which may be coordination of OH^– or H₂O, is followed by an electron transfer (E) that traps the product. Viewed by chronoamperometry measurements, the rate of decrease in H₂oxidation current does not depend on the electrode potential, in other words the chemical process controls the rate. Conversely, re-activation is an ‘EC’ reaction: the initial step is electron addition to give a reduced state that undergoes a chemical reaction to produce either active enzyme or a precursor that is finally activated by the binding of H₂. Again, viewed by chronoamperometry, the rate of increase in H₂oxidation activity now depends sharply on the electrode potential. As evident from Fig. 5, when comparing scans in the oxidising and reducing directions, re-activation is usually the better defined reaction in terms of reduction potential (this is sometimes called the ‘switch potential’ and it is obtained from the minimum of the first derivative of the current-potential plot).¹² The entire cycle of inactivation and re-activation is a ‘CEEC’ sequence.

We have summarised the electrochemical interpretations of reversible anaerobic inactivation in Scheme 1. The electron-transfer (E) reactions are the simplest to assign: as determined for Av [NiFe]-MBH, formation of Ni-B involves oxidation of Ni(II) to Ni(III), the Fe remaining as Fe(II). For [FeFe]-hydrogenases, inactivation involves oxidation of the H-cluster to a state H_ox^inact in which the binuclear center is formally Fe(II)Fe(II). The nature of the chemical steps is unclear. Our benchmark systems usually show a strong dependence on pH, with the [NiFe]- systems inactivating more rapidly at higher pH. Experiments on Dd [FeFe]-H have shown the opposite behaviour, i.e. more rapid inactivation at lower pH.²² For [NiFe]-hydrogenases, inactivation is also more extensive at lower H₂ levels.^5,12 This has not yet been established for the [FeFe]-hydrogenases.

Scheme 1 Simplest electrochemical representations of the conversions between active and inactive states of hydrogenases in the presence of H₂. Note that in the case of [FeFe]-hydrogenases, we consider the active site to be just the [FeFe] unit and neglect redox changes at the FeS cluster.

Inhibition by small molecules

The common inhibitors of hydrogenases are certain small gaseous molecules such as H₂, CO, O₂ and H₂S, and electrochemical measurements reveal exquisite details of the rates and reversibility of the reactions involved. Inhibition can be quantified from steady-state current measurements for gas mixtures under constant flow or by using the procedure we have discussed earlier for measuring K_M. Léger and co-workers derived an expression similar to eqn (1) to extract K_I from the time dependence of the change in current as inhibitor is flushed out of solution.¹⁷

Inhibition by H₂

Hydrogenases exhibit widely differing behaviour in regard to how H₂ production is inhibited by H₂. Fig. 6 compares inhibition of H₂ production by H₂ for three different hydrogenases. Product inhibition is also evident in Fig. 3 where the [NiFe]-hydrogenases show a much lower H⁺ reduction current (zone 1) in the presence of H₂ than in the presence of Ar. As a rough guide, K_I values for H₂ inhibition are of the order of µM for the [NiFe]-hydrogenases²³ but may be much higher for the [FeFe]-hydrogenases. This is not surprising given that the physiological role of most [FeFe]-hydrogenases is production of H₂ rather than its oxidation. Product inhibition is certainly one of the reasons why some hydrogenases are better H₂ producers than others.

Fig. 6 Experiments to compare the inhibition of H₂ production by H₂ and CO across a range of hydrogenases. All experiments were performed at –0.45 V vs.SHE at an electrode rotation rate of 3000 rpm. Experiments on Re [NiFe]-MBH and Dg [NiFe]-H were conducted at pH 5.5, 30 °C, whilst those on Dd [FeFe]-H were conducted at pH 6.0, 10 °C. All gas concentrations are 100% unless otherwise stated. Note the unusual case of Re [NiFe]-MBH, where H₂ production is inhibited much less by CO than by H₂.

Inhibition by CO

Carbon monoxide is usually a potent inhibitor of hydrogenases, for both H₂oxidation and H₂ production. Interesting exceptions are the [NiFe]-MBH enzymes from Ralstonia species for which CO is an extremely weak (essentially ineffective) inhibitor of H₂oxidation²⁴ and very much weaker than H₂ as an inhibitor of H₂ production.²³

Fig. 6 also compares CO and H₂ as inhibitors of H₂ production by Re [NiFe]-MBH,²³Dg [NiFe]-H and Dd [FeFe]-H. These experiments show the effects of switching between inert and inhibitor gases. For Dd [FeFe]-H and Dg [NiFe]-H, switching from N₂ to CO immediately results in almost complete loss of H⁺ reduction current, then upon switching back to N₂ there is a short lag period before the current begins to recover. In contrast, Re [NiFe]-MBH shows very much weaker inhibition by CO and recovery begins immediately upon switching the gas flow back to N₂. When N₂ is switched to 100% H₂, the greatest effect is seen for Re [NiFe]-MBH (immediate and almost complete inhibition, then a lag period upon switching back to N₂) and the smallest effect is seen for Dd [FeFe]-H. The rate of re-activation of CO-inhibited Dd [FeFe]-H is light sensitive and depends on potential.²²

The observation that H₂ and CO have such contrasting affinities for the same type of active site in different hydrogenases suggests that the active site in each case is sterically constrained so much as to be able to distinguish between the favoured binding geometries (sideways-on vs. end-on) of these ligands.

Inhibition by O₂—the question of ‘O₂ tolerance’

To enable hydrogenases or small molecular analogues to be used as H₂ cycling catalysts in technological applications, they must be capable of operating in the presence of O₂, even if this means just transient exposure to low levels. However, all hydrogenases that have so far been studied are inactivated or permanently damaged by O₂, in many cases even at trace levels. Hydrogen-cycling organisms were on Earth long before O₂ appeared in the atmosphere: as O₂ levels began to build up organisms had to adapt to survive and one way to do this was to protect the active sites of hydrogenases against irreversible damage by O₂. The ability of hydrogenases to function in the presence of O₂ is known as ‘O₂ tolerance’ and it is now an established property of some [NiFe]-hydrogenases.

In contrast, it is widely reported²⁵ that [FeFe]-hydrogenases (which are usually isolated from facultative anaerobic organisms) cannot sustain H₂-oxidation activity in the presence of even traces of O₂. Unlike CO, O₂ induces substantial changes to the active site, causing oxidation of metal ions and sulfur, and attack by reactive products of O₂reduction. Only for some [NiFe]-hydrogenases are the products of O₂ attack well established, namely Ni-B and Ni-A (also known as ‘Unready’ or Ni_u*), which contain, respectively, a OH^– ligand or the bound product (a peroxide or oxygenated cysteine-S atom) of incomplete reduction of O₂ (see Scheme 2).^26–28

Scheme 2 Representations of the products of aerobic inactivation of [NiFe]-hydrogenases. Depending on the availability of electrons, O₂ is either reduced completely or trapped as a reactive oxygen species.

For [FeFe]-hydrogenases, the products of O₂ reaction have not yet been characterised. There are major efforts to engineer greater O₂ tolerance into hydrogenases, the main aim being to develop biosystems that can sustain high H₂ production by photosynthesis or fermentation when O₂ is present. The main focus so far has been on selectively restricting the access of O₂ to the active site, the assumption being that H₂ diffuses through the enzyme using tight gas channels.³⁰ Although we agree that a diffusion filter may be operative, we believe that other factors are also important in enabling a hydrogenase to be O₂-tolerant. The experiments shown in Fig. 7 illustrate the wide spectrum of O₂ tolerance among hydrogenases.^13,22,28,29 In each case, a small aliquot of O₂-saturated solution is injected whilst H₂oxidation is monitored as the potential is scanned in the positive direction. The O₂-tolerant enzyme Re [NiFe]-MBH is only partially inhibited by O₂ although the drop in current is rapid. At higher potential the hydrogenase undergoes anaerobic inactivation, then almost full re-activation occurs on the return scan. The switch potential is about +150 mV, the same as observed for anaerobic inactivation alone. Addition of O₂ to Av [NiFe]-MBH causes rapid and complete inactivation, and (slow) re-activation occurs only when the potential on the return scan is taken below –0.1 V at which point in time all O₂ has also been removed. The Db [NiFeSe]-H is also completely inhibited by a small amount of O₂ but subsequently recovers activity (rapidly) below a potential of –0.3 V.²⁹ The [FeFe]-hydrogenase reacts more slowly with O₂, but shows only a tiny current when a more reducing potential is applied.²² Interestingly, Baffert et al. have reported that the initial reaction of O₂ with Ca [FeFe]-H is reversible³¹ and we return to this later. For the [NiFe]-hydrogenases, there is a correlation between the switch potential and O₂ tolerance in H₂oxidation, because under aerobic conditions, each enzyme must remain inactive at potentials more positive than this value.

Fig. 7 Inactivation and subsequent re-activation of various hydrogenases after reaction with O₂. Each O₂ injection involved introduction of 0.2 mL of O₂-saturated buffer into a 2 mL cell solution, resulting in ca. 90 µM O₂ in the cell solution. The constant H₂ flow rate and electrode rotation rate ensure that the O₂ is flushed out after injection, and after 300 s no O₂ remains in solution. All experiments were conducted at pH 6, 30 °C, 1 bar H₂, at an electrode rotation rate of 2500 rpm. Panels 1, 2 and 4 are reprinted, with permission, from ref. 13 (copyright 2005 American Chemical Society), and panel 3 is reprinted, with permission, from ref. 29 (copyright 2008 American Chemical Society).

Although not an O₂-tolerant hydrogenase, Av [NiFe]-MBH has been extensively studied, first by EPR and IR spectroscopy, then by protein film voltammetry.^{13,28,32–34} Reaction with O₂ produces the two inactive Ni(III) species Ni-A and Ni-B, in amounts that vary with the conditions under which O₂ reacts with the active site. Under 100% H₂ and applying a more negative electrode potential, most of the product is Ni-B, whereas when O₂ is introduced under N₂ and a more positive electrode potential, Ni-A predominates. This is easily seen in experiments (Fig. 8) that record the progress of reductive re-activation after a potential step. Re-activation is biphasic—the fast stage corresponding to reactivation of Ni-B whereas the slow phase corresponds to Ni-A. The effect of potential on the ratio of Ni-A to Ni-B produced by reaction with O₂ suggests that Ni-A results from incomplete reduction of O₂ in the active site and the formation of a ‘reactive oxygen species’ (ROS), whereas Ni-B results from complete reduction of O₂. We have also noted, for all [NiFe]-hydrogenases, that reaction with O₂ is much faster than anaerobic inactivation, showing that under O₂, Ni-B forms by a different route, i.e. direct attack by O₂.

Fig. 8 Experiments to investigate the formation and re-activation of the Unready (Ni-A) state of Av [NiFe]-MBH. Panel A shows the time-course of re-activation at −158 mV following injection of O₂ under the conditions indicated for each trace. Note here that following exposure to O₂ at low potential, the fast phase corresponding to Ready (Ni-B) predominates, whereas following exposure to O₂ at high potential, the slow phase corresponding to Ni-A predominates. Panels B and C show the dependence of reactivation of Ni-A on electrode potential. All experiments were conducted at 45 °C, at an electrode rotation rate of 2500 rpm, and at 1 bar H₂ and pH 6 unless otherwise stated. Panel A is reprinted, with permission, from ref. 28 (copyright 2004 American Chemical Society), and Panels B and C are reprinted, with permission, from ref. 27 (copyright 2005 American Chemical Society)

Based upon X-ray structure analysis, Ni-A has been proposed to be a peroxo species,²⁶ which is consistent with the more oxidising conditions required for its formation. The electrochemistry has been studied in some detail: the kinetics of re-activation are first order and potential-dependent, reaching a limiting rate of 0.0025 s^–1 (half-life 280 s) at 45 °C. The mechanism is of the ‘EC’ type, with the ‘C’ step rate-determining but not final: the last stage it seems is achieved by the binding of H₂ or, surprisingly, CO.²⁷ For Av [NiFe]-MBH, Ni-A and Ni-B appear to re-activate at very similar potentials. Evidently a HO^– and a (putative) HOO^– confer very similar influences on the Ni(III)/Ni(II) reduction potential. In contrast, experiments on the Db [NiFeSe]-H, which has not been well characterised in any oxidised inactive form, show that the product of reaction with O₂ re-activates at a much lower potential than the species that is formed upon anaerobic inactivation.²⁹

The slow rate of reactivation of Ni-A is sufficient to prevent most [NiFe]-hydrogenases from oxidising H₂ in the presence of even trace O₂. The [NiFe]-MBH enzymes from Ralstonia sp. are unusual in that they have the ability to oxidise H₂ even in the presence of atmospheric levels of O₂. Under conditions of 20% O₂/80% H₂, significant H₂oxidation activity is retained, although the extent of O₂ inhibition is strongly potential dependent. Significantly, EPR studies show that Ni-A is not formed when these enzymes react with O₂, but signals are observed that correspond to Ni-B, the rapidly re-activating ‘resting’ state.³⁵ The reason why these enzymes are O₂ tolerant therefore is due, at least in part, to the fact that Ni-A is not formed, and in part to the high potential at which the inactive state, presumably Ni-B, is rapidly re-activated. One way to view the [NiFe]-hydrogenases is that O₂ is a substrate that competes with H₂, but causes reverse electron flow and produces a more complicated choice of reaction intermediates and rates of interconversion. This idea, that [NiFe]-hydrogenases are also oxidases (catalysing O₂reduction albeit at rates that are sometimes very slow) is represented in Scheme 3.

Scheme 3 Schematic catalytic cycles for competing H₂oxidation (left) and aerobic inactivation (right) of [NiFe]-hydrogenases.

A model for the O₂ tolerance of certain [NiFe]-hydrogenases can be proposed, based upon consideration of three distinct stages of the disruption caused by O₂ to the normal catalytic cycle:

Access. How quickly does O₂ reach the active site?

Reaction. What happens when O₂ reaches the active site?

Recovery. How quickly can the active site recover?

‘Access’ refers to the so-called ‘gas-channel effect’ in which access of H₂ and other gases to the active site is controlled by the size of channels in the protein.^19,30,36,37 These channels are likely to be dynamic and are not necessarily directly revealed in static crystal structures.³⁰ However, limiting access of O₂ to the active site is insufficient itself to confer O₂ tolerance, as some O₂ will invariably reach the active site.

‘Reaction’ involves two sequential steps. First; the binding of O₂ to the active site, which may be in direct competition with H₂. Second; reduction of O₂, either by four electrons to give Ni-B and, eventually, two H₂O molecules, or by fewer than four electrons, which will result in a reactive oxygen species and formation of Ni-A. Thus O₂ tolerance could be conferred by the availability of an additional electron source close to the active site. This raises an interesting point—an enzyme attached to an electrode (and by extension a biological membrane loaded with reduced quinol) ought to be better able to avoid ROS formation than a free enzyme molecule in solution.

‘Recovery’ involves the thermodynamics and kinetics of reductive activation. A high reduction potential for the inactive state (i.e. minimal driving force for reactivation) is advantageous and an inherently fast reduction process is essential in order for hydrogen oxidation to persist under aerobic conditions.

In summary, for O₂ tolerant H₂oxidation, access of O₂ must be slow, the reaction of O₂ must be highly selective for the formation of an easily-recovered state, and recovery from this state should be fast. Although a hydrogenase can thus survive O₂ reaction by acting as an oxidase, the extent of this reverse electron flow is never sufficient to prevent the biological action of the enzyme.

A similar model could be useful for [FeFe]-hydrogenases. However, the products of O₂ attack have yet to be characterised and most studies to date suggest that the formation of the oxidised inactive states is irreversible. The irreversible reaction of O₂ at the active site implies that limiting its access is crucial as this is the only available mechanism for restricting the rate at which O₂ can react. This draws attention again to the suggestion by Baffert et al., that for Ca [FeFe]-H the initial attack of O₂ on the active site is reversible, but the O₂-adduct undergoes further irreversible reaction.³¹

Inhibition by H₂S

Hydrogen sulfide is another potentially potent inhibitor and was the subject of earlier controversy regarding the active site structure of Dv [NiFe]-MBH.³⁸ Initial crystallographic studies on this enzyme by Ogata and colleagues showed that a sulfur atom rather than an oxygen atom was coordinated to Ni in the bridging position. Additionally, H₂S liberation was detected during the re-activation of as-purified enzyme.^39,40 A question we addressed⁴¹ was how easy it would be to introduce a sulfur ligand into the active site then remove it by reduction. Since H₂S is produced by sulfate-reducing bacteria the conditions under which it inhibits are particularly relevant.

Fig. 9 shows experiments to investigate the reaction of Dv [NiFe]-MBH with sulfide and the mechanism by which the inactive product is re-activated. Most experiments employed injection of a solution of Na₂S dissolved in buffer at pH 6: the pK for H₂S/HS^– is 6.8 so the predominant species is H₂S. Injecting the sulfide solution when the potential is more positive than +100 mV causes immediate inactivation; but if sulfide is injected at a more negative potential, no inactivation is observed until the potential has climbed above 0 V. Activity is recovered on the return scan, at a potential that is some 130 mV higher than that observed for recovery from O₂ under the same conditions. Reaction of the hydrogenase with sulfide is thus reversible, but the inactive species that is formed is stable only under quite oxidising conditions (at a positive potential). It is certain that these conditions must have determined the state of the active site when crystals of the sulfide-form of Dv [NiFe]-MBH were grown. However, since sulfide is not bound at less oxidising potentials, it is unlikely that the enzyme is inhibited in vivo, even when large amounts of H₂S are being produced.

Fig. 9 Voltammetric experiments to investigate the reaction of sulfide with a [NiFe]-hydrogenase (Dv [NiFe]-MBH). All experiments were conducted at pH 6, 45 °C, electrode rotation rate of 2500 rpm, under 1 bar H₂. The voltammograms in panel A were recorded at 3 mV s^–1 and the voltammograms in Panel B were recorded at 1 mV s^–1. Figure reproduced with permission from ref. 41 (copyright 2006 American Chemical Society).

As the pH is raised, the inactivation becomes less rapid and the switch potential for reactivation shifts to lower values. As a higher pH lowers the amount of H₂S available to attack the active site, the results suggest that the attacking species is H₂S rather than HS⁻, although HS⁻ is likely to be the ligand actually coordinated. Another interesting observation is that active hydrogenase reacts with sulfide, but the inactive ‘resting states’ Ni-B and Ni-A do not. This supports the proposal that the site of reaction is the active site rather than one of the FeS clusters. Ogata and colleagues reported that Ni-A is formed in particularly high yield if the enzyme is first exposed to sulfide before reacting with O₂.⁴⁰ This ‘new route’ to Ni-A was thus tested electrochemically. The green trace in Fig. 9B shows the reactivation of hydrogenase in which the sulfide adduct was subsequently treated with O₂ while holding the electrode potential at a positive value; from its much lower re-activation potential, the sulfide adduct is no longer present. Chronoamperometric studies showed that the product behaves like ‘pure’ Ni-A (i.e., there is a single phase having the same rate of reactivation as Ni-A generated from serial O₂ injections, cf.Fig. 8) rather than the mixtures of Ni-A and Ni-B that are always formed upon direct reaction with O₂. One possibility is that the sulfide entity is quite easily released from a state that is otherwise very electron-poor, and we have seen above, with Av [NiFe]-MBH, how more oxidising conditions favour formation of Ni-A over Ni-B.

Similar experiments have shown that both Dg [NiFe]-H and Df [NiFe]-H, also both from sulfate-reducing bacteria, form sulfide adducts, and Av [NiFe]-MBH also reacts with sulfide but to a much lesser extent than the Desulfovibrio species. In addition to the [NiFe]-containing hydrogenases, we have observed sulfide inhibition for Db [NiFeSe]-H and Dd [FeFe]-H. However, the O₂-tolerant Re [NiFe]-MBH shows no reaction with sulfide over a wide range of conditions. As with CO, this lack of inhibition is informative, but only once structural information becomes available will an explanation be found.

H₂ production in the presence of O₂

Concepts behind electrochemical measurements of H₂ production in the presence of O₂

Microbes produce H₂ not only by anaerobic fermentation, allowing its generation from organic waste, but also in photosynthetic processes that allow it to be generated from sunlight. Microalgae and cyanobacteria are attracting considerable interest regarding their possible exploitation in ‘H₂ farms’;^25,42 however, they carry out oxygenic photosynthesis which poses a problem because the O₂ released by Photosystem II must be prevented from inactivating the hydrogenase. Hydrogenases with the ability to catalyse H₂ production under aerobic conditions have thus become the ‘Holy Grail’ of the effort to photosynthesise H₂. We have already discussed O₂ tolerance in the context of H₂oxidation, and we pointed out that aside from ‘designing’ enzymes to physically restrict access of O₂ to the active site, a chemical tolerance to O₂ can also be conferred upon the active site.

Several studies have been carried out to evaluate the ability of hydrogenases to carry out sustained H₂ production in the presence of O₂.^23,29,31,32 This type of experiment poses a problem for conventional methods because O₂ reacts rapidly with (and consumes) the low-potential electron donors that provide the electrons for H₂ production; but this need not be a problem when the hydrogenase is attached to an electrode. Our experiments begin by forming a film of hydrogenase and setting the electrode potential at a value below that for the reversible 2H⁺/H₂ couple: we are thus examining zone 1 of the voltammograms shown in Fig. 3. A steady-state reduction current is observed which corresponds to H₂ production. When O₂ is introduced, two things occur regardless of whether or not the hydrogenase is inactivated; first, there should be an increase in reduction current due to O₂reduction at the graphite electrode; second, this consumes some of the O₂ and may produce partially-reduced oxygen species. The problem of monitoring H₂ production when O₂ is present is solved by introducing a gaseous inhibitor of hydrogenase and measuring the changes in reduction current as it first attacks and is then flushed from the cell. The actual O₂ concentration experienced by the hydrogenase at any particular electrode potential is estimated by carrying out control experiments on O₂reduction at a bare PGE rotating disc electrode.

Hydrogen production in air by the ‘O₂-tolerant’ [NiFe]-MBH enzymes from aerobes (Ralstonia)

Although the [NiFe]-MBH enzymes from Ralstonia (Re and Rm) are very poor at catalysing H₂ production (see Fig. 3 and 5), part of this is due to product inhibition and these hydrogenases can produce H₂ provided it is continually removed. With this fact in mind, the high O₂ tolerance of these hydrogenases suggested them as candidates for investigations of H₂ production in the presence of O₂. An experiment demonstrating H₂ production when air is flushed through the cell is shown in Fig. 10.²³ Because CO is such a poor inhibitor of Rm [NiFe]-MBH, we used H₂ as the inhibitor (to quantify its own production). The experiment commences with N₂ flowing through the headspace of the cell. The current due to H⁺ reduction at –0.45 V drops immediately when 10% H₂ is introduced into the cell then recovers as the inhibitor is removed from the headspace. When 21% O₂ is flushed through the headspace, there is a large increase in current resulting from O₂reduction at the electrode. From control experiments we estimated that under these reducing conditions approximately 14% O₂ survives at the electrode surface to be ‘sensed’ by the enzyme. However, when H₂ is introduced into the headgas (i.e., the composition becomes 21% O₂, 10% H₂, 69% N₂), a further decrease in current is recorded. This decrease is similar in magnitude to that prior to O₂ introduction. This result shows that hydrogenase-catalysed H₂-production must have contributed to the total current prior to introduction of the inhibitor. When the headgas is restored to 21% O₂ in N₂, the increase in reduction current signifies recovery of enzyme-catalysed H₂ production. When O₂ is flushed out, the current corresponding solely to enzymatic H₂ production is re-established. Finally, a repeat of the initial inhibition phase is performed to confirm the activity of the enzyme. The surprising significance of these results is that for Rm [NiFe]-MBH, O₂ behaves as a much weaker inhibitor than H₂.

Fig. 10 H₂ production by Rm [NiFe]-MBH in the presence of 21% O₂. Experimental conditions: pH 5.5, potential –0.45 V vs.SHE, electrode rotation rate 3000 rpm, 40 °C, N₂ is the carrier gas throughout the experiment and the concentrations of H₂ and O₂ indicated are introduced into the head gas. Figure reproduced with permission from ref. 23 (copyright 2008 American Chemical Society).

Hydrogen production by a [NiFeSe]-hydrogenase in the presence of low-level O₂

The Db [NiFeSe]-H is a much better H₂ producer than the [NiFe]-hydrogenases we have investigated.²⁹ We saw in Fig. 7 that this enzyme cannot be an O₂-tolerant H₂ oxidiser because although it recovers from O₂ attack, it only does this rapidly at a potential that is almost as negative as the 2H⁺/H₂ couple. But the rapidity of this re-activation, determined in potential-step experiments, suggested that the hydrogenase could instead be a H₂-producer in the presence of O₂. The experiment confirming this proposal is shown in Fig. 11.²⁹ Note that when O₂ is introduced into the cell the reduction current decreases rapidly as the enzyme is immediately inhibited before levelling off as the O₂ equilibrates with the cell solution. When O₂ is removed from the cell, the current initially decreases slightly (due to decreased O₂-reduction at the electrode, circled) but then increases as hydrogenase-catalysed H₂ production recovers.

Fig. 11 H₂ production in the presence of O₂ by Db [NiFeSe]-H. Experimental conditions: electrode rotation 3000 rpm, –0.45 V vs.SHE, pH 6.0, 30 °C, N₂ is the carrier gas throughout the experiment and the concentrations of CO and O₂ indicated are introduced into the head gas. Figure reproduced, with permission, from ref. 29 (copyright 2008 American Chemical Society).

Demonstrations of possible future applications of hydrogenases

Comparing hydrogenases with platinum

It is very timely to compare the electrocatalytic activities of hydrogenases with the activity of a Pt electrocatalyst. So far, in simple experiments, and based on estimates of enzyme coverage, the H₂ production rate by Ca [FeFe]-H (HydA)⁴³ and the H₂oxidation rates of Av [NiFe]-MBH⁴⁴ and Db [NiFeSe]-H⁴⁵ all seem to compare quite favourably with Pt at modest temperatures and neutral pH. However, it matters whether our comparisons relate to practical or fundamental considerations. The H₂ pressures provided to a normal fuel cell are much higher than the levels used in experiments on hydrogenases, and very much higher than the enzymes would experience in an organism. It is very likely that evolutionary pressure on H₂ oxidising enzymes favoured a high affinity for H₂ (a low ) and ability to deal with inhibitors such as CO and O₂, which would be more important than simply having a high turnover frequency (k_cat). The H₂ oxidising hydrogenases might then be superior to Pt only when compared at low H₂ pressures (too low to be useful in normal fuel cells) and where a membrane is undesirable (e.g. in miniaturised fuel cells). Another practical disadvantage is that the ‘footprint’ of an enzyme molecule on an electrode is orders of magnitude larger than the tiny cluster of Pt atoms on which H₂ is produced or oxidised. However, at the fundamental level, the activity per active site of enzyme could be much higher than for the minimum catalytic Pt unit. Generally speaking, enzymes are restricted to operate within a narrow range of temperature and pH and they are usually considered rather unstable. Against this, neither are Pt fuel cell electrodes stable at the atomic level (high Pt loadings are needed to counteract loss from the support) and Pt is particularly susceptible to inactivation by H₂S and CO. In contrast the activity of hydrogenases is finely tuned by the ligand and supramolecular environment around the active site and inhibition by these gases (and O₂) can be prevented or minimised.

A hydrogen fuel cell running on a dilute mixture of H₂ in air

The high H₂ affinity and O₂-tolerance of Rm [NiFe]-MBH⁴⁶ was demonstrated through its use as the anode catalyst in a membraneless fuel cell, generating power from an atmosphere of 3% H₂ in air.^47,48 As shown schematically in Fig. 12A, the hydrogenase-modified anode was placed in a pH 5 solution close to a cathode modified with laccase from Trametes versicolor. Laccase is a ‘blue’ multi-copper enzyme which catalyses four-electron reduction of O₂ to H₂O at a potential within 100–200 mV of the thermodynamically reversible value. The power produced was sufficient to allow a digital watch to run for over 24 hours (Fig. 12B), and a maximum power output of ca. 5 µW cm^–2 was recorded (Fig. 12C).

Fig. 12 H₂oxidation in air and power generation in a membraneless H₂/O₂fuel cell. Panel A: Schematic diagram summarising the reactions taking place in the fuel cell: desirable reactions are shown in blue whilst undesirable reactions are shown in red. Panel B: Using the fuel cell to generate useful power. Panel C: Power vs. applied resistance curves show that power is not produced when the O₂-tolerant Rm [NiFe]-MBH was replaced with the O₂-sensitive Av [NiFe]-MBH, or when either enzyme was omitted. Experimental conditions were pH 5, 25 °C. Panel D: Cyclic voltammograms of Rm [NiFe]-MBH under conditions of 1% H₂ in N₂ (dotted) and 1% H₂ in air (bold); the response of a blank electrode in 1% H₂ in air is also shown (dashed). All pH 5.5, 30 °C, scan rate 2 mV s⁻¹. Panels A and B are reprinted with permission from ref. 47, reproduced by permission of the Royal Society of Chemistry, and Panel D is reprinted with permission from ref. 46 (copyright 2008 American Chemical Society).

Although the absolute amount of power generated from this setup was small, the experiment stands as a proof of concept, demonstrating what is possible under the most disadvantageous conditions with a sufficiently selective catalyst. For practical purposes, the device is perhaps better employed as a H₂ sensor. Additionally, these experiments are valuable as an alternative to ‘conventional' electrochemical experiments in probing the reactions of hydrogenases, providing probably the closest way in which we can explore how certain enzymes would behave in vivo.

At low loads, the power generated by the fuel cell drops sharply (Fig. 12C), reflecting the high-potential inactivation process at the hydrogenase, which could be ‘re-activated' only by application of open-circuit conditions for a short period, presumably in order to reduce the hydrogenase through enforcement of a low electrode potential. Fig. 12D shows a voltammetric experiment carried out in an atmosphere containing low-level H₂ in air, and demonstrates the narrow potential region in which a net oxidation current can be observed.⁴⁶ A similar effect would presumably also occur in vivo if the cell redox potential was to become too positive.

Coupling H₂oxidation to catalysis by a second enzyme on conducting particles

Pyrolytic graphite is easily ground into particles, which produces some interesting possibilities. We are developing the concept of attaching complementary pairs of redox enzymes to particles of graphite and other conducting materials, in order to create selective heterogeneous redox catalysts.⁴⁹ Electrons required by one enzyme (the electron acceptor enzyme) are supplied, via conduction, from the reaction catalysed by the electron-donor enzyme. Hydrogenases are the obvious candidates as electron-donor enzymes, making it possible to carry out enzymatic reductions simply by placing a suspension of particles under a H₂ atmosphere. The particles can be easily separated by centrifugation and stored between use. The concept is illustrated in Fig. 13.

Fig. 13 A schematic representation of catalytic electron flow through a conducting particle to which are attached a hydrogenase and another enzyme capable of catalysing a reduction process using electrons produced through H₂oxidation.

As a proof of concept, we carried out experiments using Av [NiFe]-MBH as the electron-donor enzyme and the respiratory nitrate reductase from E. coli (NarGHI) as electron-acceptor enzyme. The reaction product, nitrite, was easy to assay using the colorimetric Griess reagent and quantitative conversions were observed.⁴⁹

Photochemical H₂ production

With the goal of achieving photocatalytic H₂ production, we have investigated the electrochemistry of hydrogenases adsorbed on TiO₂-ITO electrodes (ITO = indium-doped tin oxide). Experiments with Db [NiFeSe]-H which is both a good H₂ producer and reasonably O₂ tolerant in H₂ production²⁹ (see above) showed that it displayed excellent electrocatalytic voltammetry with very promising stability.¹¹

This allowed us to develop a rational basis for the interpretation of earlier reports of photo-H₂ production by hydrogenases, including highly O₂-sensitive enzymes, adsorbed on TiO₂ upon UV band-gap irradiation.⁵⁰ To achieve visible light-driven water reduction, we adsorbed Db [NiFeSe]-H on TiO₂nanoparticles to which we had attached the photosensitizer [Ru^II(bipy)₂(4,4′-(PO₃H₂)₂-bipy)]Br₂ (RuP; bipy = 2,2′-bipyridine). We then suspended the particles in a buffer solution containing EDTA/Tris as sacrificial electron donor (Fig. 14). Visible light irradiation (λ > 420 nm) excites the RuP photo-sensitizer, which injects electrons into the conduction band of TiO₂ and on to the hydrogenase, resulting in efficient and catalyticH⁺ reduction. The O₂-tolerance of Db [NiFeSe]-H for H₂ production may allow this to be extended to include a water oxidation catalyst thereby leading to photochemical water splitting, resembling artificial photosynthesis.

Fig. 14 (A) Schematic representation of visible light-driven H₂ production with Db [NiFeSe]-H attached on ruthenium-dye sensitized TiO₂nanoparticles, in the presence of a sacrificial electron donor D. (B) Time-dependent H₂ evolution from bio-photocatalytic experiments (λ > 420 nm) with this system at 25 °C in buffered neutral water (squares) measured by gas chromatography. The turnover numbers (TON) are defined as mol H₂ produced per mol of hydrogenase (shown in thousands) and a control experiment (no dye) is also shown (circles). Chem. Commun., DOI: 10.1039/b817371k. Reproduced by permission of the Royal Society of Chemistry.

Outlook

Protein film voltammetry is revealing a wealth of important information on hydrogenases, and complementing structural and spectroscopic studies on this class of enzyme. Electrochemical methods provide a sharp tool with which to navigate an extremely complex subject. The breadth of experiments now being conducted suggests that ‘Protein Film Electrochemistry’ (PFE) might be a more suitable term.

In order to make further progress in determining the mechanisms of O₂ tolerance it is essential to obtain structural and spectroscopic data for the relevant states of O₂-tolerant hydrogenases. The tiny sample requirements and high resolution of kinetic and thermodynamic factors make PFV studies very valuable for comparing different enzymes and pinpointing even the subtlest of differences that should eventually be traceable at the atomic level.

A significant observation, when comparing hydrogenases with model complexes that have been prepared so far, is that hydrogenases, like platinum, catalyse reversible H₂ cycling without any overpotential. This is even true of those hydrogenases from Ralstonia, that are strongly inhibited by H₂—take away the H₂ and they will produce H₂ with minimal overpotential. This fact sets an important benchmark for the development of alternative catalysts to platinum for fuel cells. The intrinsic activity of the hydrogenase active site is very high and is comparable with that of Pt. Despite disadvantages such as the low practical power density (due to the voluminous nature of enzymes) it is possible that hydrogenases could be used in niche applications, including sensors.

Could an isolated active site work without the protein ‘casing’? It would be highly desirable to have molecular catalysts that were almost as small as the active sites, because enzymes have limited stability, are expensive to isolate, and take up so much space on an electrode that we would need hundreds of monolayers to achieve acceptable current densities. The difficulty of synthesising small model analogues that have all their component atoms ‘in the right place’ cannot be overstated. The coordination spheres within either the [NiFe] and [FeFe] classes probably differ little between enzymes that otherwise display wide variations in O₂ tolerance and catalytic bias for H₂ production vs.oxidation; so the protein environment plays a very important role in tuning the activity and deterring inhibitors.

Assuming that in the right context hydrogenases will be applied in new electrochemical technologies, it will be necessary to establish reliable procedures to achieve strong and robust attachment to their electrode or other conducting support. Achieving this goal would be useful also in particle technologies such as the coupling of hydrogenases to other enzymes or in developments along the lines of the photo H₂ production we described previously.

Biological production of H₂ on a large scale brings certain practical requirements that need to be addressed at the molecular as well as the microbial level. The hydrogenases should be good H₂ producers and for photosynthetic H₂ production the enzymes must be able to sustain activity in the presence of O₂. The electrochemical studies are providing important new data and generating clearer insight as to how this might be achieved.

Acknowledgements

Research on the electrocatalytic properties of hydrogenases carried out in the Armstrong group has been supported by grants from the BBSRC (43/E16711 and BB/D52222X/1) and EPSRC (Supergen and quota awards). K. A. V. is a Royal Society Research Fellow.

References

1. P. M. Vignais and B. Billoud, Chem. Rev., 2007, 107, 4206–4272.
2. R. J. Maier, Biochem. Soc. Trans., 2005, 33, 83–85.
3. J. C. Fontecilla-Camps, A. Volbeda, C. Cavazza and Y. Nicolet, Chem. Rev., 2007, 107, 4273–4303.
4. C. Léger, A. K. Jones, W. Roseboom, S. P. J. Albracht and F. A. Armstrong, Biochemistry, 2002, 41, 15736–15746.
5. K. A. Vincent, A. Parkin and F. A. Armstrong, Chem. Rev., 2007, 107, 4366–4413.
6. K. A. Vincent and F. A. Armstrong, Inorg. Chem., 2005, 44, 798–809.
7. C. Léger, S. J. Elliott, K. R. Hoke, L. J. C. Jeuken, A. K. Jones and F. A. Armstrong, Biochemistry, 2003, 42, 8653–8662.
8. C. Léger and P. Bertrand, Chem. Rev., 2008, 108, 2379–2438.
9. F. J. M. Hoeben, I. Heller, S. P. J. Albracht, C. Dekker, S. G. Lemay and H. A. Heering, Langmuir, 2008, 24, 5925–5931.
10. O. Rüdiger, J. M. Abad, E. C. Hatchikian, V. M. Fernandez and A. L. De Lacey, J. Am. Chem. Soc., 2005, 127, 16008–16009.
11. E. Reisner, J. C. Fontecilla-Camps and F. A. Armstrong, Chem. Commun. 10.1039/b817371k.
12. A. K. Jones, S. E. Lamle, H. R. Pershad, K. A. Vincent, S. P. J. Albracht and F. A. Armstrong, J. Am. Chem. Soc., 2003, 125, 8505–8514.
13. K. A. Vincent, A. Parkin, O. Lenz, S. P. J. Albracht, J. C. Fontecilla-Camps, R. Cammack, B. Friedrich and F. A. Armstrong, J. Am. Chem. Soc., 2005, 127, 18179–18189.
14. H. R. Pershad, J. L. C. Duff, H. A. Heering, E. C. Duin, S. P. J. Albracht and F. A. Armstrong, Biochemistry, 1999, 38, 8992–8999.
15. C. Léger, A. K. Jones, S. P. J. Albracht and F. A. Armstrong, J. Phys. Chem. B, 2002, 106, 13058–13063.
16. S. Dementin, V. Belle, P. Bertrand, B. Guigliarelli, G. Adryanczyk-Perrier, A. L. De Lacey, V. M. Fernandez, M. Rousset and C. Léger, J. Am. Chem. Soc., 2006, 128, 5209–5218.
17. C. Léger, S. Dementin, P. Bertrand, M. Rousset and B. Guigliarelli, J. Am. Chem. Soc., 2004, 126, 12162–12172.
18. M. Ludwig, J. A. Cracknell, O. Lenz, K. A. Vincent and F. A. Armstrong, J. Biol. Chem. DOI:10.1074/jbc.M803676200.
19. F. Leroux, S. Dementin, B. Burlat, L. Cournac, A. Volbeda, S. Champ, L. Martin, B. Guigliarelli, P. Bertrand, J. C. Fontecilla-Camps, M. Rousset and C. Léger, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 11188–11193.
20. F. A. Armstrong and S. P. J. Albracht, Philos. Trans. R. Soc. London, Ser. A, 2005, 363, 937–954 ; discussion 1035–1040.
21. W. Lubitz, E. Reijerse and M. van Gastel, Chem. Rev., 2007, 107, 4331–4365.
22. A. Parkin, C. Cavazza, J. C. Fontecilla-Camps and F. A. Armstrong, J. Am. Chem. Soc., 2006, 128, 16808–16815.
23. G. Goldet, A. F. Wait, J. A. Cracknell, K. A. Vincent, M. Ludwig, O. Lenz, B. Friedrich and F. A. Armstrong, J. Am. Chem. Soc., 2008, 130, 11106–11113.
24. K. A. Vincent, J. A. Cracknell, O. Lenz, I. Zebger, B. Friedrich and F. A. Armstrong, Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 16951–16954.
25. M. L. Ghirardi, M. C. Posewitz, P.-C. Maness, A. Dubini, J. Yu and M. Seibert, Annu. Rev. Plant Biol., 2007, 58, 71–91.
26. A. Volbeda, L. Martin, C. Cavazza, M. Matho, B. W. Faber, W. Roseboom, S. P. J. Albracht, E. Garcin, M. Rousset and J. C. Fontecilla-Camps, J. Biol. Inorg. Chem., 2005, 10, 591–591.
27. S. E. Lamle, S. P. J. Albracht and F. A. Armstrong, J. Am. Chem. Soc., 2005, 127, 6595–6604.
28. S. E. Lamle, S. P. J. Albracht and F. A. Armstrong, J. Am. Chem. Soc., 2004, 126, 14899–14909.
29. A. Parkin, G. Goldet, C. Cavazza, J. C. Fontecilla-Camps and F. A. Armstrong, J. Am. Chem. Soc., 2008, 130, 13410–13416.
30. J. Cohen, K. Kim, P. Kimg, M. Seibert and K. Schulten, Structure, 2005, 13, 1321–1329.
31. C. Baffert, M. Demuez, L. Cournac, B. Burlat, B. Guigliarelli, P. Bertrand, L. Girbal and C. Léger, Angew. Chem., Int. Ed., 2008, 47, 2052–2054.
32. B. Bleijlevens, F. A. Broekhuizen, A. L. Lacey, W. Roseboom, V. M. Fernandez and S. P. J. Albracht, J. Biol. Inorg. Chem., 2004, 9, 743–752.
33. B. Bleijlevens, B. W. Faber and S. P. J. Albracht, J. Biol. Inorg. Chem., 2001, 6, 763–769.
34. S. J. George, S. Kurkin, R. N. F. Thorneley and S. P. J. Albracht, Biochemistry, 2004, 43, 6808–6819.
35. F. Lendzian, personal communication.
36. Y. Montet, P. Amara, A. Volbeda, X. Vernede, E. C. Hatchikian, M. J. Field, M. Frey and J. C. Fontecilla-Camps, Nat. Struct. Biol., 1997, 4, 523–526.
37. J. Cohen, K. Kim, M. Posewitz, M. L. Ghirardi, K. Schulten, M. Seibert and P. King, Biochem. Soc. Trans., 2005, 33, 80–82.
38. Y. Higuchi, H. Ogata, K. Miki, N. Yasuoka and T. Yagi, Structure, 1999, 7, 549–556.
39. M. Higuchi and T. Yagi, Biochem. Biophys. Res. Commun., 1999, 255, 295–299.
40. H. Ogata, S. Hirota, A. Nakahara, H. Komori, N. Shibata, T. Kato, K. Kano and Y. Higuchi, Structure, 2005, 13, 1635–1642.
41. K. A. Vincent, N. A. Belsey, W. Lubitz and F. A. Armstrong, J. Am. Chem. Soc., 2006, 128, 7448–7449.
42. A. Melis and T. Happe, Plant Physiol., 2001, 127, 740–748.
43. M. Hambourger, M. Gervaldo, D. Svedruzic, P. W. King, D. Gust, M. Ghirardi, A. L. Moore and T. A. Moore, J. Am. Chem. Soc., 2008, 130, 2015–2022.
44. A. K. Jones, E. Sillery, S. P. J. Albracht and F. A. Armstrong, Chem. Commun., 2002, 866–867.
45. A. A. Karyakin, S. V. Morozov, E. E. Karyakina, N. A. Zorin, V. V. Perelygin and S. Cosnier, Biochem. Soc. Trans., 2005, 33, 73–75.
46. J. A. Cracknell, K. A. Vincent, M. Ludwig, O. Lenz, B. Friedrich and F. A. Armstrong, J. Am. Chem. Soc., 2008, 130, 424–425.
47. K. A. Vincent, J. A. Cracknell, J. R. Clark, M. Ludwig, O. Lenz, B. Friedrich and F. A. Armstrong, Chem. Commun., 2006, 5033–5035.
48. J. A. Cracknell, K. A. Vincent and F. A. Armstrong, Chem. Rev., 2008, 108, 2439–2461.
49. K. A. Vincent, X. Li, C. F. Blanford, N. A. Belsey, J. H. Weiner and F. A. Armstrong, Nat. Chem. Biol., 2007, 3, 761–762.
50. P. Cuendet, K. K. Rao, M. Grätzel and D. O. Hall, Biochimie, 1986, 68, 217–221.

---

Footnote

† Part of the renewable energy theme issue.

---