Hydrogenases and Oxygen

View Article Online / Journal Homepage / Table of Contents for this issue Chemical Science Dynamic Article LinksC<

Cite this: Chem. Sci., 2012, 3, 1739 www.rsc.org/chemicalscience PERSPECTIVE Hydrogenases and oxygen

Martin Tillmann Stiebritz*a and Markus Reiher*b

Received 23rd December 2011, Accepted 3rd February 2012 DOI: 10.1039/c2sc01112c

Hydrogenases represent a heterogeneous group of enzymes consisting of three evolutionary unrelated classes, i.e. [NiFe], [FeFe] and [Fe] hydrogenases. They allow the uni-cellular organisms in which they are expressed to use hydrogen as energy source or to reduce protons as a sink for excess reduction equivalents. Because of this capability there is growing interest in exploiting these enzymes in the ﬁeld of sustainable energy generation. However, most hydrogenases which are appealing in this context are reversibly or irreversibly inhibited by dioxygen. Here, we summarize the current picture of oxygen- induced inhibition of the different classes of hydrogenases and discuss possible avenues that might lead to tailored oxygen-robust enzyme variants.

1 Introduction discuss the current status of knowledge about the mechanisms of

1–5 O2 induced inhibition in order to clarify and evaluate strategies Hydrogenases catalyze the reversible oxidation of molecular for obtaining O -tolerant enzyme variants. + À 2 hydrogen, H2 # 2H +2e , and it is exactly this capability that makes them particularly interesting for sustainable energy production—either by directly exploiting their catalytic activity 2 Hydrogenase classes and their structural in electrochemical devices or by biomimetic complex-design characteristics inspired by their active sites. Considerable progress in the latter 6 2.1 Genetical and physiological classiﬁcation ﬁeld has recently been achieved with the prospect of becoming independent from enzyme based systems. However, despite the Downloaded by ETH-Zurich on 25 March 2013 Physiologically, hydrogenases are involved in anaerobic energy

remarkable turnover rates observed, the over-potentials neces- conversion processes that are accompanied by H2 formation or sary to achieve catalysis are still signiﬁcantly larger for synthetic consumption. Based on sequence similarity and active site

Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C 6 catalysts. On the other hand, the technical application of composition the three phylogenetically unrelated classes of hydrogenases is hampered by the fact that, to a variable extent, [FeFe], [NiFe] and [Fe] hydrogenases can be distinguished.

these enzymes are rather sensitive towards molecular oxygen, Whereas enzymes that are catalytically active in H2 conversion which is presumably due to their evolutionary origin in anaerobic are distributed over unicellular members of the three domains living organisms. Any development exploiting their catalytic eubacteria, archaea and eukarya, hydrogenase-like genes are also capabilities has therefore to face and to cope with the oxygen- found in higher eukaryotes and even in humans, where their sensitivity problem of hydrogenases. protein products are presumably involved in the assembly of Recent years have seen signiﬁcant progress in the mechanistic cytosolic iron–sulfur proteins.7–10 [NiFe] hydrogenases are het- understanding of how the active sites of different hydrogenases erodimeric proteins with the a subunit carrying the active site interact with molecular oxygen and reactive species formed whereas the b subunit contains Fe–S clusters involved in electron thereof. Hydrogenase chemistry has already been reviewed in the transport and facilitates protein complex formation and sub- literature, but this has not yet been done in view of the reactivity cellular localization. In terms of their physiological roles [NiFe] of all different classes of hydrogenases against reactive oxygen enzymes display the largest variability amongst the convergently species. Here, we pursue this goal in order to deduce general evolved hydrogenases. Based on homology they are grouped into reactivity principles that may eventually also be applied to other four classes as reviewed by Vignais et al.1–5 Group 1 comprises metallo-enzymes. After giving a brief overview of the various membrane-bound so-called uptake [NiFe] hydrogenases, that

hydrogenase classes and the composition of their active sites, we allow cells to use H2 as an energy source by forming a proton motive force. They occur in bacteria and archaea. Based on sequence similarity the uptake hydrogenases of cyanobacteria aLaboratorium fur€ Physikalische Chemie, ETH Zurich, Wolfgang-Pauli- and the cytoplasmic hydrogen sensors that in bacteria regulate Strasse 10, 8093 Zurich,€ Switzerland. E-mail: [email protected]. hydrogenase gene expression in response to environmental H2 ethz.ch; Fax: +41 44 633 15 94; Tel: +41 44 633 38 66 concentration form group 2. Group 3 contains bidirectional het- bLaboratorium fur€ Physikalische Chemie, ETH Zurich, Wolfgang-Pauli- Strasse 10, 8093 Zurich,€ Switzerland. E-mail: [email protected]. eromultimeric cytoplasmic [NiFe] hydrogenases from bacteria ethz.ch; Fax: +41 44 633 15 94; Tel: +41 44 633 43 08 and archaea. Here, the basic hydrogenase dimer is associated

This journal is ª The Royal Society of Chemistry 2012 Chem. Sci., 2012, 3, 1739–1751 | 1739 View Article Online

with other subunits speciﬁc for soluble cofactors such as NAD— composition the common feature is the presence of Fe and/or Ni because these enzymes operate in a reversible fashion the as central ions coordinated by strong ligands, which due to their cofactors can be re-oxidized under anaerobiosis. The fourth toxicity do normally not occur in biological systems—namely group consists of energy-conserving, membrane-associated CO and CNÀ. hydrogenases which are found in bacteria and archaea. Structures of [NiFe] hydrogenases have mainly been solved for [FeFe] hydrogenases show the highest catalytic activity in the enzymes from sulfate-reducing bacteria of the Desulfovibrio hydrogen producing reaction regime, unlike [NiFe] hydrogenases genus15,17–22 and recently for enzymes from the purple sulfur 3 23 which are generally more biased toward H2 oxidation. This bacterium Allochromatium vinosum, the Knallgas bacterium makes them particularly interesting for sustainable energy Ralstonia eutropha24 and the hydrogen oxidizer Hydrogenovibrio production. With exceptions, hydrogen formation appears to be marinus.25 Therefore, all structures available belong to group 1 of the major physiological role of this class of enzymes which the class of [NiFe] hydrogenases. mainly occur as monomeric proteins in anaerobic prokaryotes The active site of [NiFe] hydrogenases, which is located in the (Clostridium spec., Desulfovibrio spec.), in the chloroplasts of large subunit of the enzyme, consists of a binuclear metal cluster unicellular algae (e.g. Chlamydomonas spec.) and in hydro- formed by Ni and Fe (see Fig. 1, middle). The Ni center is genosomes of anaerobic protozoa and certain fungi.2 Beside the coordinated by the sulfur groups of four cysteine residues, from active site, [FeFe] hydrogenases can contain varying numbers of which it shares two with the neighboring Fe atom, that addi- iron-sulfur (Fe–S) clusters, presumably involved in electron tionally carries two CNÀ and one CO ligands. In a special transport. More detailed information can be found in a number variant, the [NiFeSe] enzyme (see below) from Desulfomicrobium of excellent reviews about hydrogenases.1–4 [Fe] hydrogenase—or baculatum, one of the Ni-coordinating cysteine residues is Hmd (5,10-methylenetetrahydromethanopterin dehydroge- replaced by selenocysteine. nase)—is only found in the group of methanogenic arch- The active site is separated by a distance of approx. 6 A from

aebacteria and catalyzes the heterolytic cleavage of H2 upon a regular [Fe4S4] cluster that resides in the small subunit and which a hydride ion is transferred to the cofactor 5,10-meth- which is most probably involved in electron transfer. This enyltetrahydromethanopterin.11 This pathway provides reduc- proximal Fe–S cluster is complemented by two further Fe–S tion equivalents for converting carbon dioxide into methane.12,13 clusters. In oxygen-sensitive [NiFe] hydrogenases, the medial

Interestingly, the expression of this enzyme is tightly regulated cluster is a one-Fe ion deprived [Fe3S4] cluster and the distal one and only occurs when Ni, which is needed to build the active sites a regular [Fe4S4] cubane, in which, however, one Fe is coordi- of [NiFe] hydrogenases, is scarce.12 nated by a histidine instead of a cysteine residue. Crystal structures of [FeFe] hydrogenases have so far only been solved for the enzymes from Clostridium pasteur- 2.2 Protein structures and active sites ianum,14,26,27 Desulfovibrio desulfuricans28 and for the enzyme Fig. 1 gives an overview over the active sites of the three classes of from the unicellular green alga Chlamydomonas reinhardtii.29 The

Downloaded by ETH-Zurich on 25 March 2013 hydrogenases and the protein structures of typical representa- latter however, though properly folded, does not contain the tives for which crystal structures have been solved. A comparison catalytically active subsite, because the protein was isolated from of the overall reactions catalyzed by the enzymes and their a strain deﬁcient in maturation factors involved in active site

Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C 29 inhibition by O2 can be found in Fig. 2. In terms of active site synthesis. Whereas the bacterial enzymes feature additional

Fig. 1 Protein structures and active sites of structurally characterized representatives of the three classes of hydrogenases. Left: [FeFe] hydrogenase, protein model based on the crystal structure of Clostridium pasteurianum hydrogenase (PDB entry 3C8Y14). The Lewis structure depicts the H cluster. Middle: [NiFe] hydrogenase, protein model generated from the crystal structure of the Desulfovibrio gigas enzyme (PDB entry 2FRV15). The large subunit is depicted in dark grey, the small subunit in light grey, respectively. The Lewis structure shows the active site. Right: [Fe] hydrogenase, protein model derived from the crystal structure of the Methanocaldococcus jannashii enzyme (PDB entry 3H65,16). The Lewis structure depicts the active site without the cofactor 5,10-methenyltetrahydromethanopterin. In the enzyme the pyridine ligand of the central iron atom is linked to a guanosine- monophosphate moiety (GMP).

1740 | Chem. Sci., 2012, 3, 1739–1751 This journal is ª The Royal Society of Chemistry 2012 View Article Online Downloaded by ETH-Zurich on 25 March 2013 Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C

Fig. 2 Basic reactions catalyzed by the three classes of hydrogenases and their response to O2 exposure. Upper panel: The H cluster of [FeFe] hydrogenases is most active in the hydrogen producing reaction regime. There is experimental and theoretical evidence that O2 exposure leads primarily 61–63 to the formation of reactive oxygen sepecies (ROS) at the H cluster followed by breakdown of the [Fe4S4] cubane (upper pathway) and/or 64 61,62 decomposition of the 2[Fe]H subsite (lower pathway). After prolonged exposure, dioxygen terminally leads to the total breakdown of the H cluster (not shown). Middle panel: [NiFe] hydrogenases are mostly efﬁcient H2 oxidizers. O2 leads to the formation of up to two inactive oxygen adducts that differ in their reactivation behavior. Ni-B is readily reactivated whereas the reactivation of the Ni-A state takes up to several hours to complete.59,60 Note À that O2-tolerant [NiFe] hydrogenases known today do not form the Ni-A state. This inactivated state was originally characterized as carrying a OOH moiety in a briding position between Ni and Fe. This assignment is currently under debate. Lower panel: [Fe] hydrogenases catalyze H2 splitting by using 3 the cofactor 5,10-methenyltetrahydromethanopterin that acts as an acceptor for a hydride ion. This activity is not affected by O2 as it does not bind to the metal center.

Fe–S clusters the algal protein carries only the active site— thiolate groups of a bidentate ligand. There has been an ongoing commonly called the H cluster. The H cluster consists of debate about the identity of the bridgehead-group of this

a binuclear component, the 2[Fe]H subsite, that is connected to compound, which according to recent experimental, as well as a regular [Fe4S4] cubane by a shared cysteinyl sulfur atom (see theoretical investigations, most probably corresponds to a NH- Fig. 1, left). The Fe center next to the cubane is called the moiety.30,31

proximal (Fep) Fe center and the second one consequently the The currently available crystal structures of [Fe] hydrogenase distal Fe center (Fep). Both Fe atoms are coordinated each by were all solved for the enzyme from Methanocaldococcus janna- aCOandaCNÀ ligand and connected with each other by the shii.16,32–34 Initially, the enzyme was thought not to carry any

This journal is ª The Royal Society of Chemistry 2012 Chem. Sci., 2012, 3, 1739–1751 | 1741 View Article Online

transition-metal ions.35–37 Later it became apparent that its resonance (EPR) spectroscopy has been an invaluable method to actives site contains a single Fe ion but no additional Fe–S identify different paramagnetic states and to gain structural clusters,32,38 because of which it is also called Fe–S cluster-free information about the active site.52 Structural details about the hydrogenase. The catalytically active Fe center is linked via the ligand environment of the Ni and Fe atoms and the study of thiolate group of a cysteine residue to the protein (see Fig. 1, EPR-silent states of the enzyme have been carried out by Four- right). Apart from two CO ligands the central Fe is coordinated ier-transform infrared (FTIR) spectroscopy and spectroelec- by the N atom and acyl group of a pyridinol ligand. Originally, trochemistry.4,53,54 In general, electrochemical methods, the acyl group was believed to be a carboxyl moiety which was especially protein-ﬁlm voltammetry (PFV) allowed for the study corrected after a better reﬁned crystal structure was available.16 of redox-transitions of the enzymes.3 These experimental endeavors have been complemented by theoretical studies based 3 Theory complements experiment on quantum mechanical calculations on active site models or combined quantum mechanical/molecular mechanical studies on 55 In this work, we aim at a converging picture of O2 inhibition of the whole protein. These matters have been excellently reviewed hydrogenases based on experimental and theoretical results. in the recent literature.3,4,53,55–57 Our aim here, however, is to look

Major breakthroughs have been achieved in the understanding at all known types of hydrogenases in the context of O2-induced of hydrogenase chemistry by sophisticated experimental tech- inhibition and to discuss general mechanistic principles that can niques. However, structural, energetic and spectroscopic infor- be deduced from there. mation about important intermediates and transition states As-isolated preparations of [NiFe] hydrogenases are usually needed to unravel regular enzymatic or inhibition mechanisms catalytically inactive and need to be reactivated ﬁrst. The degree are often not easily accessible by experiment. Here, quantum of inactivation, however, differs particularly between ‘‘standard’’ chemistry is a valuable complement, because it allows for the oxygen-sensitive and oxygen-tolerant hydrogenases (see below).3 analysis of reaction pathways and intermediates at atomic Anaerobic, as well as aerobic inactivation produces a para- detail.39 However, the theoretical approach to bioinorganic magnetic form, referred to as the Ni-B state, which can be 58 reaction mechanisms is not without pitfalls as we have recently reactivated within seconds by H2. In oxygen-sensitive hydrog- elaborated in ref. 40, and requires input from experiment as it enases O2 exposure results in a further inactive paramagnetic heavily relies on initial structural information about the chemical species, the Ni-A state, which shows a much slower reactivation composition of the reactive species. The principal structural kinetics that takes up to several hours to complete.59,60 Therefore, composition is difﬁcult to ‘derive’ from theory (as, e.g., high- Ni-A is usually called the unready and Ni-B the ready state. Both lighted by theoretical studies on active sites whose structures states are EPR-active in oxygen-sensitive variants with Ni in were later proven to be incorrect). Molecular structures can be formal +III and Fe in formal +II oxidation states. One-electron optimized, but the principal composition of relevant atoms and reduction of Ni-A results in an EPR-silent state, Ni-SU (silent

functional groups in the active site is in general required as input. unready), whereas reduction of Ni-B delivers the state (Ni-SIr)I 57

Downloaded by ETH-Zurich on 25 March 2013 As a consequence, the interdependence of experiment and theory which is in protonation equilibrium with (Ni-SIr)II. It appears usually leads to an iterative process in which facts and knowledge that the rate determining step of the slow reactivation of Ni-A about a mechanism in atomic detail are established. involves the conversion of the silent unready Ni-SU form into

Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C 4 Due to the size of hydrogenases—or enzymes in general— (Ni-SIr)II. (Ni-SIr)II is converted into a further EPR-silent quantum chemical methods applied are of approximate nature. species, Ni-SIa, which is catalytically active and which forms the Approximations are unavoidable in the set-up of the structural paramagnetic state Ni-C upon coupled proton/electron transfer. model of limited size as well as in the choice of the theoretical The catalytic cycle of the enzyme presumably involves oscilla- 57 approach. As a result, calculated observables are affected by tion between the states Ni-SIa, Ni-C and Ni-R, which results inaccuracies, so that a thorough interpretation of reaction from Ni-C by a further coupled H+/eÀ transfer. Here, it should be energies and barrier heights requires some expertise, especially noted that [NiFe] hydrogenases are also reversibly inhibited when open-shell transition-metal clusters are to be described.41–43 by CO. Interestingly, CO speciﬁcally interacts with state

However, even if calculated reaction energies and barrier heights Ni-SIa. These matters have been reviewed in a recent article by can be affected by non-negligible inaccuracies, the structural Pandelia et al.52 parameters (bond lengths and angles) obtained are quite reliable The mechanistic details of oxygen-induced enzyme inactiva- and provide invaluable insights into active sites. Moreover, tion is still not completely resolved but major progress has been spectral signatures of molecular structures can be obtained achieved in recent years. For the two inactive oxidized forms, Ni- directly by calculations that help assign these structures to signals A and Ni-B, crystal structures have been solved21,22 which clar- and spectra observed in experiment.39 iﬁed to some extent the debate about the structure of these compounds. For the Ni-A state of [NiFe] hydrogenases from sulfate- 4O2-induced inhibition of [NiFe] hydrogenases: a converging picture? reducing Desulfovibrio species it turned out that electron density found in a bridging position between Ni and Fe can best be For [NiFe] hydrogenases, which, in contrast to [FeFe] hydroge- explained by a hydroperoxide species.21,22 Remarkably, in the À nases, are not irreversibly destroyed by O2, a rather detailed reﬁned structure, OOH is bound in a side-on fashion to the Ni picture is available to date. X-Ray crystal structures have been atom (see Fig. 2, middle panel). However, this assignment is solved for the inactive and the oxidized forms of the enzyme from a matter of ongoing debate. A recently measured crystal struc- different bacterial sources.15,17–20,22,25,44–51 Electron paramagnetic ture of the related enzyme from the purple photosynthetic

1742 | Chem. Sci., 2012, 3, 1739–1751 This journal is ª The Royal Society of Chemistry 2012 View Article Online

bacterium Allochromatium vinosum could more reliably be the removal of the oxygen-based species that are blocking the reﬁned by assuming a mono-oxo species (OHÀ) in the bridging active site, irrespective of their exact chemical composition. position,23 a result that is also in accordance with ENDOR Jayapal et al. studied this scenario computationally and found (electron nuclear double resonance) spectroscopy experiments that a bridging hydroperoxo-species which was assumed to performed on the enzyme from Desulfovibrio vulgaris MF.57 describe the Ni-A state, can most effectively be removed by The assignment of the Ni-B state is less controversial in the reduction and protonation events that lead to water abstraction literature and it is assumed that the same bridging position most and leave a m-oxo-bridge between Ni and Fe.71 This form is

likely is occupied by a hydroxyl group (see Fig. 2, middle panel). reducible by H2 which finally leads to removal of a second water However, the exact chemical structure as well as the conversion molecule and free coordination sites at Fe and Ni. In case of Ni- mechanisms between the two forms have not yet been unam- B, reduction and protonation of the m-hydroxo bridge would biguously identified.52 A number of density functional theory presumably produce a weakly bound water molecule, which can 57 (DFT) studies have been carried out on this subject to comple- easily be removed to form the active Ni-SIr state. ment the experimental available data but it turned out that especially modeling of the Ni-A is particularly difficult, i.e. 5 [FeFe] hydrogenase: the search for a mechanism of theoretical predictions could not reliably reproduce the experi- O -induced inhibition mental assigned arrangements of the assumed peroxide ligand 2 55 and the corresponding Ni–O bond distances. These complica- There is compelling evidence that the process by which O2 tions, however, could also imply that the available structural inactivates the active site of [FeFe] hydrogenases differs information for this compound is not complete. A broad and substantially from the reversible inhibition observed for [NiFe] detailed overview about this subject was given by Siegbahn hydrogenases. These enzymes are generally irreversibly 55 3,53 et al. destroyed by O2. However, differences in sensitivity toward The theoretical studies on the oxidation products of the active dioxygen have been reported dependent on the species back- site of [NiFe] hydrogenase show indeed some of the subtleties ground.72 For [FeFe] hydrogenases much less is known about the involved in achieving agreement between DFT calculations on molecular events that finally destroy enzyme catalysis (see Fig. 2, active site models and structural experimental data. Discrep- top panel). Unlike in the case of [NiFe] hydrogenase, no crystal ancies do not necessarily imply that the computational method is structure of inactive oxidation products could be solved up to flawed but often demonstrate the limitations of the model chosen now. The structures available, which most likely correspond to 40 for calculation. However, also experimental results are not free an oxidized state, called Hox, feature an oxygen species bound from shortcomings, especially, when compounds are involved to the distal Fed atom of the 2[Fe]H subsite, which most probably that are rather sensitive with respect to radiation damage, which corresponds to a water molecule or an OHÀ group (see cannot be excluded in structures solved by means of X-ray Fig. 1).14,26 crystallography. Soderhjelm€ and Ryde reinvestigated the avail- Recently, sophisticated electrochemical and spectroscopic

Downloaded by ETH-Zurich on 25 March 2013 able solved crystal structures of the oxidized states Ni-A and studies have shed some light on possible reaction pathways that 21,22 Ni-B from the Desulfovibrio enzymes by applying a quantum are involved in H cluster destruction induced by O2. Using refinement scheme that allows to complement the regular X-ray protein film voltammetry Baffert et al. were able to show that Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C refinement procedure by treating parts of the structure—here, destruction of the active site is preceded by the formation of an 72 differently sized parts of the active site region—quantum O2 adduct to the H cluster. Furthermore, this analysis mechanically.65 The unambiguous assignment of the Ni-A and demonstrated that carbon monoxide, a reversible inhibitor of Ni-B structure was still difficult and the authors were able to [FeFe] hydrogenases, protects the active site from irreversible

show that the experimental structures represent a mixture of destruction by O2—after removal of the inhibitor CO, the different states, including forms where the sulfur groups of the enzymes studied resumed catalytic activity.72,73 Because a crystal ligating cysteine residues are oxidized. structure has been solved for a CO-inhibited [FeFe] hydrogenase Partly because of these discrepancies and despite the structural from C. pasteurianum27 featuring an additional CO molecule

characterization of the inactive states Ni-A and Ni-B, the coordinated to the distal Fed atom, the data suggest this center to mechanism of O2-induced inhibition of [NiFe] hydrogenases is be the ﬁrst site of attack for an O2 molecule approaching the not settled. Experimental evidence exists, however, for a direct active site.

contact of O2 with the active site followed by formation of Ni-A These observations are in line with our independent theoretical and Ni-B.66–69 Especially the structure of state Ni-A has been investigations into the degradation mechanism63,64 which show

challenged recently for the periplasmic hydrogenase of Desulfo- that the coordination of O2 to the Fed atom is energetically vibrio gigas24 The authors could not conﬁrm a second oxygen favored. We furthermore demonstrated that the coordination of

atom corresponding to a bridging hydro-peroxide species. O2 can be followed by exothermic transfer of two protons to the However, a second oxygen atom was found on a terminal distal O atom of the bound O2 species that leads to splitting of sulfenate group which belongs to a oxygenated cysteine species at the O–O bond and water release and leaves a highly oxidized 70 64 74 the active site Ni, which corroborates the theoretical ﬁndings by Fed–O center, a result that was also reported by Dogaru et al. Soederhjelm and co-workers.65 This compound can either be protonated or it could undergo As already mentioned, oxygen-induced inactivation of [NiFe] a recombination reaction in which the O ligand may react with 64 hydrogenases is mainly reversible, i.e. does not lead to a CO moiety under formation of CO2 (see Fig. 3). Recent a destruction of the active site. Based on the available structural investigations into the kinetics of the degradation process have and mechanistic information, the reactivation process requires shown that this process is feasible with an activation barrier of

This journal is ª The Royal Society of Chemistry 2012 Chem. Sci., 2012, 3, 1739–1751 | 1743 View Article Online

around 10 kcal molÀ1.75 The emerging species can favorably or is reduced to form reactive oxygen species (ROS) such as

coordinate a second O2 molecule and this mechanism therefore superoxide or hydrogen peroxide that after dissociation reach the accounts for the irreversibility of O2-induced inhibition by nearby cubane via a diffusive pathway and react in a way that breakdown of the ligand environment of the H cluster. Notably, destroys cluster integrity.62,76 The hypothesis that the coordina-

the formation of a Fed–O compound followed by CO2 abstraction of O2 to the H cluster involves the formation of ROS was tion was also observed when additional electron transfer was already established by DFT calculations64 as the O–O bond 64 considered, i.e. without forming a highly oxidized cluster. length of the O2 adduct corresponds to a superoxide moiety The reaction of O2 with the H cluster was also investigated in which was further conﬁrmed by local spin analysis revealing one 74 64 a QM/MM study by Dogaru et al. They found spontaneous unpaired spin at the bound O2. Our work suggests that both coordination of O2 to the Fed atom in the liquid phase and ﬁnally superoxide and hydrogen peroxide can be formed at the 2[Fe]H 74 formation of a hydroxy group bound to Fed. It is however not subsite in an autocatalytic fashion provided the active site is likely, that the irreversibility of O2-induced inactivation can be supplied by electrons and re-reduction of the H cluster can occur explained by the formation of this product alone because its (see Fig. 4). Therefore, ROS formation would represent a path- formation does not disrupt the integrity of the catalytic site (see ological catalytic activity of the H cluster.63 Furthermore, the Fig. 3). Furthermore, it does not explain why the inactivated study shows that in all relevant redox states the cubane can be form cannot reductively be reactivated in electrochemical attacked by a OOH radical, which leads to opening of the cluster.

experiments, as is the case for the inhibited forms of [NiFe] H2O2 preferably reacts with the sulfur groups of the bidentate hydrogenases (see previous section). ligand and the anchoring cysteine residues. Even though oxida-

O2-induced destruction of the [2Fe]H subsite was conﬁrmed by tive damage of biomimetic complexes is not broadly studied in time-resolved X-ray absorption spectroscopy on the [FeFe] the literature, similar sulfoxygenation reactions of active site hydrogenase from the unicellular green alga Chlamydomonas models of [FeFe] and [NiFe] hydrogenases have already been reinhardtii.62 However, apparently this event takes place on described, as reviewed by Darensbourg and Weigand.78 In our longer timescales and the authors identiﬁed destruction of the study it became apparent that the degradation process may 62,76 63 [Fe4S4] cubane as the primary process of oxidative inhibition. involve a number of competing oxidation reactions and the In the experiments, changes in characteristic extended X-ray theoretical results suggest that there might be a competition

absorption ﬁne structure (EXAFS) signals that are indicative for between destruction of the [2Fe]H subsite, as shown in Fig. 3, and Fe–Fe distances in the [Fe4S4] cubane occurred while the ROS production followed by cubane decomposition reactions. distances in the 2[Fe]H subsite still remained unchanged, which For the oxygen-inhibited H-cluster, experimental data that can would imply that up to this point the 2[Fe]H subsite remains directly be conﬁrmed computationally (i.e. structures of inter- intact.62,76 This study also suggests that upon destruction the mediate or terminal oxidation products), are not available yet.

[Fe4S4] cubane is transformed into a [Fe3S4] cluster by losing one Therefore, it is crucial to reproduce the measured time-resolved Fe2+ ion, a mechanism discussed for other enzymes containing EXAFS spectra in order to validate theoretical predicted reac- 77 À

Downloaded by ETH-Zurich on 25 March 2013 cubane clusters such as aconitase. tion steps. Due to the numerous CO and CN ligands present at For the H cluster the working hypothesis of the authors is that the H cluster FTIR spectroscopy would be a powerful means for

O2 binds to the distal iron atom and either retrieves electrons the identiﬁcation and veriﬁcation of theoretically predicted Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C from the cubane, a process that would lead to oxidative damage, oxidation products and intermediates, since vibrational spectra

Fig. 3 Possible reaction mechanism of O2 with the 2[Fe]H subsite of the H cluster. Based on theoretical results and experimental hints O2 can replace H2O coordinating the distal Fed ion. Subsequent two-fold protonation leads to water abstraction. The terminal oxygen can be protonated or undergo a intramolecular reaction by forming CO2. The resulting product can favorably be coordinated by a second O2 molecule. This mechanism accounts for the oxidative and irreversible breakdown of the ligand environment of the H cluster.64

1744 | Chem. Sci., 2012, 3, 1739–1751 This journal is ª The Royal Society of Chemistry 2012 View Article Online

site. A proof of concept was provided in an indirect way by

observing an increased O2 sensitivity in H2 sensors in which these residues were mutated into those occurring in O2-sensitive hydrogenases, i.e. valine and leucine.82,83 However, the inverse approach, where in the [NiFe] hydrogenase of D. fructosovarans the respective residues were substituted by isoleucine and phenylalanine, did not result in a more oxygen-tolerant mutant, despite a signiﬁcantly reduced 84 gas diffusion rate (as measured for H2). Moreover, the analo- gous mutagenesis experiment performed with the membrane-

bound hydrogenase (MBH) from R. eutropha even increased O2 sensitivity of this enzyme.85 Theoretical investigations into the gas transport reveal that Fig. 4 The distal iron atom (green) of the H cluster as a source of this process cannot be reduced to a single diffusive pathway, reactive oxygen species (ROS). According to theoretical results ROS may which would explain the different effects of single mutations. be formed in a catalytic fashion at the H cluster that subsequently is Using a multiscale simulation approach Wang et al. could show damaged by these compounds. OOH radicals primarily attack the iron that there are at least three diffusion pathways in the D. fructo- atoms (blue) of the [Fe4S4] cubane whereas H2O2 oxidizes the sulfur sovarans enzyme.86,87 These simulations show furthermore that, atoms (red) of anchoring cysteine residues as well as the bidentate with different probabilities, small gaseous molecules can virtually ligand.63 reach any region in the protein matrix. Remarkably, substituting valine and leucine at the putative diffusion gate by methionine can conveniently be obtained by theory.79,80 Furthermore, it is residues resulted in the ﬁrst engineered oxygen-tolerant [NiFe] important to analyze how the protein environment inﬂuences the hydrogenase that remains active at higher oxygen concentra- basic reactivity of the H cluster with respect to ROS by consid- tions.84 The single mutant in which only leucine was replaced by ering larger quantum mechanical (QM) models and hybrid methionine showed 10% of the catalytic activity observed under quantum mechanical/molecular mechanical (QM/MM) anaerobiosis even at O2 concentrations similar to air-equilibrated approaches. These directions of research are currently being solutions.84 The double mutant, where both side chains were pursued in our laboratory. replaced by methionine could be reactivated by H2 under very oxidizing conditions. To a lesser extent, this reactivation could also be observed in the single mutant.84 Based on these results it is 6O2-tolerant [NiFe] hydrogenases questionable whether oxygen-resistance in H2 sensors can be [NiFe] hydrogenases appear as a rather heterogeneous class, explained by an increased selectivity against O2 diffusion to the

Downloaded by ETH-Zurich on 25 March 2013 which can be subdivided into four groups, as outlined above active site. Rather, the methionine residues appear to be directly according to composition, compartmentalization and physio- involved in the reactivation process that presumably involves logical roles. Remarkably, a number of oxygen-tolerant variants removal of a bound OOHÀ as outlined above. As proposed by Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C exist which have the common feature that the inactive and slowly Dementin et al. methionine at position 74 (Desulfovibrio fructo- reactivating Ni-A state cannot be observed. sovarans numbering) could possibly assist a reorientation of the peroxide species which facilitates its protonation and escape from the active site.84 6.1 Important role for gas diffusion channels? The high O resistance of H sensor hydrogenases (SH) was 2 2 6.2 Redox protection of the active site attributed to limited gas access to their active site. Based on X-ray structural data that were obtained under Xe-atmosphere, Interestingly, the mutations described in hydrogen sensors are a hydrophobic,45 gas-accessible channel was identiﬁed in not the only means by which natural selection is seen for

D. fructosovarans [NiFe] hydrogenase for which it was shown decreased oxygen sensitivity. Up to now, clearly O2-tolerant later that a channel network facilitates gas exchange within the [NiFe] hydrogenases, i.e. which are able to catalyze the reversible 81 enzyme. It seems plausible that the existence of similar gas oxidation of H2 in the presence of O2, have been described in four transport pathways exist in other hydrogenases as well. organisms. Remarkable examples in this context are membrane- A connection between gas diffusion and oxygen-sensitivity in bound hydrogenases (MBH) of the Knallgas bacteria R. eutropha

oxygen-tolerant H2 sensor hydrogenases was established by the and R. metallidurans which operate at ambient oxygen concen- observation that two conserved hydrophobic residues presum- trations and with trace amounts of H2, which requires an ably at the end of the gas channel next to the active site are extremely high speciﬁcity for the substrate and tolerance with 45 85 mutated in oxygen-tolerant H2 sensors. Valine and leucine— respect to O2. Indeed, these enzymes are the only true oxygen- which are usually present at these positions in O2-sensitive tolerant hydrogenases described so far. Kinetic electrochemical enzymes—are replaced by isoleucine and phenylalanine in H2 analysis suggested that this stability cannot be explained by 48 85 sensors. Because of the bulky character of these substituents a limited diffusion of O2 toward the active site. In addition to which in principle could lead to a narrowing of the gas diffusion the hydrogenases from Knallgas bacteria, the two other forms channel, these positions were interpreted as serving as a ‘‘gas interestingly occur in facultative anaerobic organisms—E. coli sieve’’,45 which determines speciﬁcity of gas access to the active and Salmonella enterica.

This journal is ª The Royal Society of Chemistry 2012 Chem. Sci., 2012, 3, 1739–1751 | 1745 View Article Online

Recent structural and spectroscopic studies on the membrane- It appears that this special redox chemistry of the cluster bound hydrogenases from Knallgas bacteria give insight into the allows MBHs to catalyze at the same time the complete reduction

mechanistic peculiarities leading to oxygen-tolerant hydrogen of O2 to water as well as the regular reversible oxidation of H2. oxidation in these enzymes. Comparative modeling already led to This implies that in the presence of oxygen there is a permanent the conclusion that the direct sequence environment of the Fe–S production of water at the active site which requires a special cluster proximal to the active site is modified by the presence of adaptation.25,51 Support for this hypothesis is provided by the two additional cysteine residues. X-Ray absorption spectroscopy presence of a hydrophilic channel present in the recent crystal revealed that the arrangement of this cluster deviates signifi- structures of the MBHs from R. eutropha and H. marinus that cantly from the regular cubane structure found in other [NiFe] have not been described for oxygen-sensitive hydrogenases. hydrogenases.24 The recently published crystal structures of the Comparison with structures of the latter reveals that the putative MBHs from Ralstonia eutropha H16 and Hydrogenovibrio water channel emerged by a single mutation that connects two 25,51 25,51,90 marinus revealed an up to then undescribed [Fe4S3] cluster hydrophilic cavities inside the protein matrix. coordinated by six cysteines (see Fig. 5). Due to the presence of One of the at least three [NiFe] hydrogenases of E. coli—Hyd- 91 two further cysteines all Fe centers of the cluster are 4-fold 1—is able to catalyze H2 oxidation in the presence of O2 and is coordinated despite the missing sulfide ion. One iron atom is highly homologue to the MBH from R. eutropha discussed coordinated by two cysteines and somewhat displaced from the above. Similarly, it features two additional cysteine residues in incomplete cubane. At the same time the second additional the vicinity of the proximal Fe–S cluster which can be deduced by cysteine replaces the missing sulfide that connects two iron homology modeling (not shown). The significance of these resi-

centers in regular [Fe4S4] clusters (see Fig. 5). The significance of dues was systematically addressed by comparing electrochemical this special cluster becomes apparent from the fact that the and EPR spectroscopical properties of both single mutants and replacement of the two additional cysteines by glycines signifi- a double mutant in which cysteines were replaced by glycines.92 cantly reduces the oxygen-tolerance of the enzyme.70 Generally, Remarkably, it turned out that the two additional cysteine resi- the electrochemical potentials of all three Fe–S clusters in MBHs dues do not equally contribute to oxygen-tolerance. Due to the are shifted toward more positive values.70,88 MBHs share this homology between the Ralstonia enzyme and Hyd-1 it appears property and the presence of two additional cysteines with the justified to establish a connection between the mutagenesis data

O2-tolerant Hase I enzyme from the hyperthermophilic Knallgas and the structural features of the proximal cluster of this enzyme. bacterium Aquifex aeolicus. Compared to oxygen-sensitive Whereas the cysteine that replaces the role of the sulﬁde ion

[NiFe] hydrogenases Hase I shows a more complex EPR spec- (compare Fig. 5) in the [Fe4S3] cubane is crucial for O2-tolerance, trum indicating spin–spin interactions between adjacent Fe–S because its replacement by glycine renders the enzyme O2- clusters—presumably between the oxidized [Fe3S4] cluster and sensitive, the mutation of the second cysteine does not abolish 70,89 92 the proximal Fe–S cluster, that most probably also corre- catalytic activity in the presence of O2. Due to the lacking sponds to the newly described special [Fe4S3] form. By measuring structural information the relevance of the architectural features

Downloaded by ETH-Zurich on 25 March 2013 EPR spectra at defined potentials Pandelia et al. were able to of the [Fe4S3] cluster for the observed enzymatic properties is not show that the proximal Fe–S cluster of Hase I can undergo two entirely clear at the moment. Furthermore, it appears that there redox transitions over a comparatively small potential range, is no direct connection between a modified proximal Fe–S cluster Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C namely 3+/2+, 2+/1+, which usually do not occur in redox- and the catalytic bias of [NiFe] hydrogenase enzymes, because, in 57 proteins. contrast to O2-tolerance, this property was not affected by The significance of additional cysteine residues in the vicinity mutating the additional cysteine residues to glycines in Hyd-1.92

of [Fe4S4] clusters has been investigated by mutagenesis experi- These fascinating results demonstrate how Nature succeeded ments.88 Interestingly, mutation of both additional cysteines to in turning an enzyme form that emerged in the context of an glycines results in an oxygen-sensitive enzyme that shows normal anaerobic lifestyle into a completely oxygen compatible variant activity under anaerobiosis.88 by variations of a well-known theme—the Fe–S cubane motif. From an engineering point of view however, this acquired robustness comes at a price, for all described oxygen-tolerant enzyme forms show a signiﬁcantly reduced catalytic activity with

complete bias for H2 oxidation.

6.3 Modiﬁcation of the ﬁrst-shell ligand sphere of the active site The Knallgas bacterium Ralstonia eutropha contains in addition

to the membrane-bound hydrogenase a H2-sensing regulatory (RH) and a soluble bidirectional hydrogenase (SH) which are also remarkably oxygen tolerant. For the former this adaptation is facilitated by mutations at the putative gas-sieve that presumably have direct effects on the active site as outlined above. The latter is composed of the heterodimeric hydrogenase module and a further heterodimeric unit that acts as a NADH dehydrogenase

Fig. 5 Structure of the special proximal [Fe4S3] cluster of the MBHs and which allows the cell to produce reduction equivalents by from R. eutropha M1651 and H. marinus.25 oxidizing molecular hydrogen.93,94 The oxygen-tolerance of this

1746 | Chem. Sci., 2012, 3, 1739–1751 This journal is ª The Royal Society of Chemistry 2012 View Article Online

enzyme can most probably be attributed to changes in the active 1%.106 A crystal structure of [NiFeSe] hydrogenase from Desul- site which, in contrast to oxygen-sensitive forms, carries two fovibrio vulgaris Hildenborough in its oxidized state revealed additional CNÀ residues, one coordinating to Ni and the other structural peculiarities of the active site that might account for 95 50 one to Fe, as shown by means of FTIR spectroscopy. The Ni- the O2 tolerance of the enzyme. Most importantly, the crystal bound CNÀ seems to be essential for oxygen-tolerance, for its structure does not show any signals that point to bridging oxygen removal turns the SH enzyme into an oxygen-sensitive form.93,96 species as observed for the oxidized states of oxygen-sensitive X-Ray absorption spectroscopy further revealed significant [NiFe] hydrogenases. Instead, the authors observe an unprece- deviations for the oxidized state of the enzyme. It appears that dented cysteine-S-dioxide form of one Ni-coordinating cysteine there is no bridging oxygen species between the Ni and the Fe but residue and for the seleno-cysteine a mixture of three confor- rather a OOHÀ moiety that solely binds to Ni. The two non- mations.50 One conformation corresponds to the reduced form of bridging cysteine residues are most probably oxidized to the enzyme and the other two contain additionally a S2À ion that sulfenates and coordinate the Ni atom via their O atoms, this is linearly connected with the Se atom of the selenocysteine. In modification was shown to be essential for enzyme activity.97 one conformation only the S2À coordinates the Ni ion and in the second both the sulfide and the selenium coordinate.50 The latter two conformations could significantly shield the Ni center and 6.4 Hydrogen-producing [NiFe] enzymes therefore impede the access of small molecule agents such as 50 The enzymes discussed so far are mainly typical H2 oxidizers. O2. The higher acidity of selenocysteine might be responsible Currently, only two [NiFe] hydrogenases have been described for the bias of this enzyme for hydrogen production107 which that show catalytic activity in the hydrogen production reaction makes it particularly interesting for sustainable energy regime and that are at the same time relatively robust with production.

respect to O2. One is a cyanobacterial bidirectional hydrogenase The catalytic preferences are also inﬂuenced by the redox and the other one the selenium variants of the uptake hydroge- potential of the Fe–S clusters present in the protein matrix. nase that are found in Desulfovibrio species. Notably, in [NiFeSe] hydrogenases the proximal Fe–S cluster is

The bidirectional [NiFe] hydrogenase from the cyanobacte- a regular [Fe4S4] cubane and not a high-potential [Fe3S4] cluster rium Synechocystis sp. PCC6803 was recently electrochemically as in oxygen-sensitive enzymes, which also contributes to a bias and spectroscopically analyzed by McIntosh et al.98 As its name toward proton reduction.

implies, the enzyme is able to catalyze both H2 oxidation and proton reduction, and shows even a small bias for the latter 98 7O2 resistance: what can be learned from mono-iron reaction. Similar to MBHs it shows a quick reactivation [Fe] hydrogenase? kinetics after O2-exposure and no Ni-A state is formed. However, it cannot perform catalysis in the presence of larger Regarding the reaction mechanism it catalyzes, mono-iron [Fe]

amounts of O2—at a O2 concentration of 1% enzyme activity is hydrogenase deviates signiﬁcantly from the two other classes of 98

Downloaded by ETH-Zurich on 25 March 2013 about 50% of the level measured for anaerobic conditions. hydrogenases discussed so far. Experimental studies have shown The bidirectional enzyme differs spectroscopically and electro- that the central Fe is always in a singlet Fe(II) state throughout chemically from oxygen-sensitive [NiFe] hydrogenases. Upon the catalytic cycle.108,109 The cofactor 5,10-methenyltetrahy- Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C interaction with O2 two states have been described which seem dromethanopterin appears to be essential for catalytic activity 110,111 not to be related to Ni-A and Ni-B because they are both EPR- and receives a hydride ion upon H2 cleavage (see Fig. 2, silent.98 Both states are rapidly reactivated. As structural lower panel). Interestingly, the enzyme is air-stable but light- information for this enzyme is not available to date the sensitive.112 Similar to the other hydrogenases it can be inhibited molecular details responsible for its oxygen tolerance are by CO but also by CNÀ. Both ligands bind with relatively low currently not known. afﬁnity to the free coordination site of the Fe center as revealed The bidirectional Synechocystis enzyme shares the existence of by different spectroscopic techniques38,112–114 (see Fig. 1). Very EPR-silent oxidized states with a special selenium-containing recent experimental results show that isocyanides are strong uptake hydrogenase that is formed in certain Desulfovibrio inhibitors of this enzyme and show very high afﬁnity to the active 115 species. This [NiFeSe] hydrogenase shows a rather high H2 site. forming activity, which is unusual for [NiFe] hydrogenase and is Despite the differences in catalysis it is interesting to ask about produced by the cells when selenium is available.99–102 Struc- the molecular principles accounting for the oxygen-insensitivity tural46,100 and spectroscopic103–105 studies of the enzyme revealed of this class of enzymes and whether they can be transferred to that one of the two terminal cysteines that solely coordinate the oxygen-sensitive hydrogenases. Ni ion in the active site (see Fig. 1) is replaced by selenocysteine. In terms of coordination chemistry the active sites of [FeFe] As for the oxygen-sensitive enzyme forms, two distinct oxidized and [Fe] hydrogenase which appear totally unrelated turn out to states have electrochemically been described in [NiFeSe] be surprisingly similar:116 strong p-acceptors (CO or CNÀ) hydrogenase from D. baculatum.106 However, similar to the arranged in the quadratical plane of the incomplete coordination Synechocystis bidirectional [NiFe] hydrogenase, both states are octahedron and in axial position to the free coordination site reactivated quickly. This reactivation however, occurs at another CO-like moiety—CO in case of [FeFe] hydrogenase and, different potentials for both states. Due to the low potential at surprisingly, an acyl ligand in the case of [Fe] hydrogenase.16,34

which the form that is obtained by O2 exposure is reactivated, the Indeed, this configuration is so unusual that the axial ligand was enzyme cannot catalyze the oxidation of H2 in the presence of O2, first assigned as a carboxyl group, which was corrected after re- 16,34,117 but H2 production is observed even at a O2 concentration of refinement of the crystal structure.

This journal is ª The Royal Society of Chemistry 2012 Chem. Sci., 2012, 3, 1739–1751 | 1747 View Article Online

The striking similarity between the ligand spheres of both to be oxygen tolerant. For in vitro use the goal is of course not active iron centers led us to assess the signiﬁcance of the single only to obtain catalytic activity for a limited time in the presence

ligand identities in a gedanken experiment: what is the effect on of O2 but rather to ensure a competitive lifetime of the catalyst. active site integrity and ligand binding strength if the inner In terms of stability the non oxygen-sensitive [NiFe] hydrog- coordination spheres if both active sites are interchanged, i.e. enases described above are suited well for biotechnological

when the first ligand shell of the Fed ion resembles that of the applications and in fact, [NiFe] hydrogenase from Desulfovibrio central Fe ion of [Fe] hydrogenase and vice versa? In order to fructosovarans has already been used as an H2 oxidizing (anode) understand its consequences for hydrogenase activity, active site catalyst in fuel cells.128–131 However, the enzymes that can be models with swapped ligands were investigated by means of described as being oxygen tolerant, i.e. which are active in the 116,118 density functional theory. The first important observation presence of O2 such as MBH from R. eutropha, show a signifi- that we made was that the process of interchanging the ligand cantly lower catalytic activity compared to oxygen-sensitive environments left the complex structures unchanged.116 enzyme forms. Moreover, they are not suited for hydrogen

O2-induced enzyme-inactivation as observed in [FeFe] hydroge- production. Some [NiFe] hydrogenases which are able to effi- nases occurs by a complicated mechanism but the exothermic ciently catalyze H2 formation, such as [NiFeSe] hydrogenases, binding of O2 to the central iron atom can be considered as the are O2 insensitive and are quickly reactivated after O2 exposure 63,64 essential first step for all subsequent reactions. Remarkably, but they cannot operate at high O2 concentrations. it turned out that exactly this step is significantly influenced by Because of these limitations [FeFe] hydrogenases, with H2 the composition of the first-shell ligand spheres of the catalyti- forming activities outperforming regular [NiFe] hydrogenases by 118 cally active metal centers. We found that O2 binding to the Fed three to four orders of magnitude, are still the most promising atom becomes less exothermic in a ligand environment resem- enzyme class if they can be transformed into oxygen-tolerant bling the active site of [Fe] hydrogenase. Conversely, binding of proteins. What has been said earlier implies the question whether

O2 to the active iron center of [Fe] hydrogenase turns from endo- principles of oxygen tolerance that have been elucidated in to exothermic upon ligand mutation.118 More speciﬁcally, [NiFe] hydrogenases can be transferred to [FeFe] hydrogenases. À substituting the CN ligand at the distal Fe atom of [FeFe] Due to the diffusion behavior of O2 that was modeled using hydrogenase by CO and replacing the corresponding CO by CNÀ a multi-scale approach86,87 it is questionable whether blocking of in [Fe] hydrogenase were sufﬁcient to induce this effect. putative gas diffusion channels will provide an universal means In this context it is important to ask how the catalytic bias of by which oxygen resistance can be accomplished in hydroge-

the active sites for H2 splitting or H2 formation is generated. This nases, because in the case of the H cluster long-term exposure to question is currently being addressed by studying the kinetics of O2 will be deleterious even at signiﬁcantly reduced diffusion enzyme catalysis in both hydrogenases under native and ligand- rates of the inhibitor, and it is pivotal that the cluster itself is exchange conditions.75 It appears that the replacement of one CO kinetically or thermodynamically protected from oxidative by CNÀ in [Fe] hydrogenase does not only affect the energetics of damage.

Downloaded by ETH-Zurich on 25 March 2013 O2 coordination but also inverts the rate determining dihydrogen It appears that the only way to arrive at a oxygen-tolerant cleavage step in a way that the opposite reaction direction [FeFe] hydrogenase consists in preventing the formation of ROS becomes favored.75 at the active site without at the same time sacrificing regular Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C catalytic activity. This aim might be difficult to achieve. As 8 Engineered oxygen-tolerant hydrogenases? outlined above it has been shown that the primary ligand sphere of the distal Fe ion of the H cluster can be changed in a way to 118 Hydrogenases are highly attractive for a hydrogen-based energy render O2 coordination less favorable. Studies, as to whether economy.119 In principle, they can be used in electrochemical this approach can be generalized and how the reaction steps of devices for the direct formation of hydrogen as well as catalysts the regular catalytic mechanism are influenced by these changes 75 for H2 oxidation in fuel cells. are currently being performed. Of course, the biosynthesis of For the purpose of hydrogen production it was shown that the H cluster is a complicated process that involves a highly hydrogenases can be exploited both in vitro by combining them sophisticated maturation machinery (for a recent review see ref. with appropriate electron donors such as photosystem I or in 132) and therefore it might be asked whether any modification of vitro by heterologous expression in suitable host organsims.119,120 the active site can easily be introduced. Due to their elaborate One problem that emerges in the former approach is the still molecular biological toolbox, proteins are far more amenable to limited stability of enzyme-based catalysts which would have to directed changes and engineering attempts than their metal resist intensive radiation for prolonged time spans in order to be clusters, mainly because of our still limited insight into their suited and competitive for practical solutions in sunlight energy biosynthesis. conversion. The advantage of self-repair and self-renewing It is therefore important to study whether oxygen tolerance can associated with expressing hydrogenase constructs in vivo comes in principle be accomplished by changes in the protein environ- with the complication that the yield is still lower and appropriate ment of the H cluster. Although generally considered as being

re-wiring of existing electron transfer pathways already existing irreversibly inhibited by O2, PVM experiments have shown that in the cell is required. Despite these complications signiﬁcant the O2 concentration that causes inhibition, varies dependent on progress has been achieved in both areas of research over the past the species background of the particular [FeFe] hydrogenase few years.121–127 studied.72,76,133 It remains to be shown whether these variations Irrespective of what developmental avenue is followed, foretell the possibility for successful protein engineering attempts. whenever oxygen exposure is unavoidable the enzymes used need Therefore, mutational screening analysis appears to be very

1748 | Chem. Sci., 2012, 3, 1739–1751 This journal is ª The Royal Society of Chemistry 2012 View Article Online

promising in this context.134–137 Complementing to these studies, References a more systematic approach inevitably requires a detailed 1 P. M. Vignais and A. Colbeau, Curr. Issues Mol. Biol., 2004, 6, 159– understanding of the inactivation mechanisms. Of special interest 188. to us is here the question why the reaction of the [NiFe] active sites 2 P. M. Vignais and B. Billoud, Chem. Rev., 2007, 107, 4206–4272. with O2 stops with the formation of oxygen adducts such as Ni-A 3 K. A. Vincent, A. Parkin and F. A. Armstrong, Chem. Rev., 2007, and/or Ni-B, without compromising the active site architecture, 107, 4366–4413. 4 A. L. D. Lacey, V. M. Fernandez, M. Rousset and R. Cammack, whereas in [FeFe] hydrogenases there is compelling evidence that Chem. Rev., 2007, 107, 4304–4330. the inactivation process leads to the irreversible destruction of the 5 P. M. Vignais, Results Problems Cell Differentiation,2008,45, 223–252. H cluster61,63,64,72,76,133 which is preceded by the formation of 6 M. L. Helm, M. P. Stewart, R. M. Bullock, M. R. DuBois and reactive oxygen species (ROS). Here, it is especially interesting to D. L. DuBois, Science, 2011, 333, 863–866. 7 J. Huang, D. Song, A. Flores, Q. Zhao, S. M. Mooney, L. M. Shaw investigate the hypothesis that the active sites of oxygen-tolerant and F. S. Lee, Biochem. J., 2007, 401, 341–352. [NiFe] hydrogenases are able to completely reduce O2 to water, in 8 D. Song and F. S. Lee, J. Biol. Chem., 2008, 283, 9231–9238. order to avoid the formation of inhibited forms, and to study the 9 D. Song, Z. Tu and F. S. Lee, J. Biol. Chem., 2009, 284, 35297– 35307. role of the newly described Fe4S3 cluster. The next question is 10 D. Song and F. S. Lee, J. Biol. Chem., 2011, 286, 15797–15805. whether this principle can be transferred to [FeFe] hydrogenases, 11 A. R. Klein, G. C. Hartmann and R. K. Thauer, Eur. J. Biochem., especially without reducing the catalytic activity of the enzymes. 1995, 233, 372–376. The missing correlation between oxygen tolerance and catalytic 12 R. K. Thauer, A.-K. Kaster, M. Goenrich, M. Schick, T. Hiromoto and S. Shima, Annu. Rev. Biochem., 2010, 79, 507–536. bias implies that there might be the possibility to arrive at an O2- 13 M. J. Corr and J. A. Murphy, Chem. Soc. Rev., 2011, 40, 2279–2292. tolerant enzyme form that shows at the same time the desired H2 14 A. S. Pandey, T. V. Harris, L. J. Giles, J. W. Peters and producing activity. R. K. Szilagyi, J. Am. Chem. Soc., 2008, 130, 4533–4540. 15 A. Volbeda, M. H. Charon, C. Piras, E. C. Hatchikian, M. Frey and In fact, it appears that the avoidance of incomplete reduced J. C. Fontecilla-Camps, Nature, 1995, 373, 580–587. products of O2 is the mechanistic solution achieved in [NiFe] 16 T. Hiromoto, E. Warkentin, J. Moll, U. Ermler and S. Shima, hydrogenases to remain functional in the presence of molecular Angew. Chem., Int. Ed., 2009, 48, 6457–6460. oxygen. In [FeFe] hydrogenases however, the process of ROS 17 A. Volbeda, E. Garcin, C. Piras, A. L. de Lacey, V. M. Fernandez, E. C. Hatchikian, M. Frey and J. C. Fontecilla-Camps, J. Am. Chem. formation could even be an intrinsic pathological catalytic Soc., 1996, 118, 12989–12996. activity of the H cluster.63 When ROS have already been formed 18 Y. Higuchi, T. Yagi and N. Yasuoka, Structure, 1997, 5, 1671–1680. inactivation of the H cluster is most probably unavoidable. 19 Y. Higuchi, H. Ogata, K. Miki, N. Yasuoka and T. Yagi, Structure, Therefore, parallel to search for modiﬁcations of the [Fe S ] 1999, 7, 549–556. 4 4 20 H. Ogata, Y. Mizoguchi, N. Mizuno, K. Miki, S. Adachi, subcluster it is important to identify mutations that interfere N. Yasuoka, T. Yagi, O. Yamauchi, S. Hirota and Y. Higuchi, J. with O2 coordination to the active site. Based on a QM/MM Am. Chem. Soc., 2002, 124, 11628–11635. study Dogaru et al. proposed mutations that render the coor- 21 H. Ogata, S. Hirota, A. Nakahara, H. Komori, N. Shibata, T. Kato, 138 K. Kano and Y. Higuchi, Structure, 2005, 13, 1635–1642. dination reaction endergonic. The authors propose to delete 22 A. Volbeda, L. Martin, C. Cavazza, M. Matho, B. W. Faber,

Downloaded by ETH-Zurich on 25 March 2013 (!) the residues Thr157 and Ser202 (sequence numbering of W. Roseboom, S. P. J. Albracht, E. Garcin, M. Rousset and Desulfovibrio desulfuricans [FeFe] hydrogenase) or to mutate J. C. Fontecilla-Camps, JBIC, J. Biol. Inorg. Chem., 2005, 10, them into alanine side chains. It seems however questionable, 239–249.

Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C 23 H. Ogata, P. Kellers and W. Lubitz, J. Mol. Biol., 2010, 402, 428– whether single amino acid substitutions are sufﬁcient to achieve 444. the goal of oxygen-tolerance. Especially these two residues for 24 J. Fritsch, S. Loscher,€ O. Sanganas, E. Siebert, I. Zebger, M. Stein, example show strict conservation over several Desulfovibrio and M. Ludwig, A. L. D. Lacey, H. Dau, B. Friedrich, O. Lenz and M. Haumann, Biochemistry, 2011, 50, 5858–5869. Clostridium species and are also conserved in the enzymes from 25 Y. Shomura, K.-S. Yoon, H. Nishihara and Y. Higuchi, Nature, Chlamydomonas reinhardtii. Single substitutions will therefore 2011, 479, 253–256. most probably disrupt catalysis or will even prevent proper 26 J. W. Peters, W. N. Lanzilotta, B. J. Lemon and L. C. Seefeldt, folding of the protein. The conservation of amino acids Science, 1998, 282, 1853–1858. 27 B. J. Lemon and J. W. Peters, Biochemistry, 1999, 38, 12969–12973. however, does not necessarily imply that a substitution is 28 Y. Nicolet, C. Piras, P. Legrand, C. E. Hatchikian and principally incompatible with a certain protein fold but merely J. C. Fontecilla-Camps, Structure, 1999, 7, 13–23. reﬂects the spectrum of possibilities in a limited sequence space 29 D. W. Mulder, E. S. Boyd, R. Sarma, R. K. Lange, J. A. Endrizzi, explored in the organisms where homologue genes occur. This J. B. Broderick and J. W. Peters, Nature, 2010, 465, 248–251. 30 A. Silakov, B. Wenk, E. Reijerse and W. Lubitz, Phys. Chem. Chem. means however, that for applying substitutions at a certain Phys., 2009, 11, 6592–6599. sequence position one has to take the identity and conforma- 31 U. Ryde, C. Greco and L. D. Gioia, J. Am. Chem. Soc., 2010, 132, tions of all neighboring amino acids into account—a problem 4512–4513. 32 O. Pilak, B. Mamat, S. Vogt, C. H. Hagemeier, R. K. Thauer, that lies at the very heart of computational protein design. The S. Shima, C. Vonrhein, E. Warkentin and U. Ermler, J. Mol. Biol., challenge here is to prevent the energetically favorable forma- 2006, 358, 798–809. tion of O2 adducts and subsequent reactions without compro- 33 S. Shima, O. Pilak, S. Vogt, M. Schick, M. S. Stagni, W. Meyer- mising the regular catalytic activity of the metal clusters Klaucke, E. Warkentin, R. K. Thauer and U. Ermler, Science, 2008, 321, 572–575. involved. 34 T. Hiromoto, K. Ataka, O. Pilak, S. Vogt, M. S. Stagni, W. Meyer- Klaucke, E. Warkentin, R. K. Thauer, S. Shima and U. Ermler, FEBS Lett., 2009, 583, 585–590. Acknowledgements 35 C. Zirngibl, W. V. Dongen, B. Schworer,€ R. V. Bnau, M. Richter, A. Klein and R. K. Thauer, Eur. J. Biochem., 1992, 208, 511–520. Financial support by the Schweizer Nationalfonds SNF (proj. nr. 36 G. C. Hartmann, A. R. Klein, M. Linder and R. K. Thauer, Arch. 200021L_138536/1) is gratefully acknowledged. Microbiol., 1996, 165, 187–193.

This journal is ª The Royal Society of Chemistry 2012 Chem. Sci., 2012, 3, 1739–1751 | 1749 View Article Online

37 A. Berkessel, Curr. Opin. Chem. Biol., 2001, 5, 486–490. 73 B. Bennett, B. J. Lemon and J. W. Peters, Biochemistry, 2000, 39, 38 M. Korbas, S. Vogt, W. Meyer-Klaucke, E. Bill, E. J. Lyon, 7455–7460. R. K. Thauer and S. Shima, J. Biol. Chem., 2006, 281, 30804–30813. 74 D. Dogaru, S. Motiu and V. Gogonea, Int. J. Quantum Chem., 2009, 39 Atomistic Approaches in Modern Biology, ed. M. Reiher, Springer- 109, 876–889. Verlag, Berlin, 2007, vol. 268. 75 A. R. Finkelmann, M. T. Stiebritz and M. Reiher, in preparation. 40 M. Podewitz, M. T. Stiebritz and M. Reiher, Faraday Discuss., 2011, 76 S. T. Stripp, G. Goldet, C. Brandmayr, O. Sanganas, K. A. Vincent, 148, 119–135. M. Haumann, F. A. Armstrong and T. Happe, Proc. Natl. Acad. Sci. 41 M. Reiher, Faraday Discuss., 2007, 135, 97–124. U. S. A., 2009, 106, 17331–17336. 42 M. Reiher, Chimia, 2009, 63, 140–145. 77 D. H. Flint, J. F. Tuminello and M. H. Emptage, J. Biol. Chem., 43 K. H. Marti and M. Reiher, Phys. Chem. Chem. Phys., 2011, 13, 1993, 268, 22369–22376. 6750–6759. 78 M. Y. Darensbourg and W. Weigand, Eur. J. Inorg. Chem., 2011, 44 J. C. Fontecilla-Camps, M. Frey, E. Garcin, C. Hatchikian, 994–1004. Y. Montet, C. Piras, X. Vernde and A. Volbeda, Biochimie, 1997, 79 J. W. Tye, M. Y. Darensbourg and M. B. Hall, J. Comput. Chem., 79, 661–666. 2006, 27, 1454–1462. 45 Y. Montet, P. Amara, A. Volbeda, X. Vernede, E. C. Hatchikian, 80 L. Yu, C. Greco, M. Bruschi, U. Ryde, L. D. Gioia and M. Reiher, M. J. Field, M. Frey and J. C. Fontecilla-Camps, Nat. Struct. Inorg. Chem., 2011, 50, 3888–3900. Biol., 1997, 4, 523–526. 81 F. Leroux, S. Dementin, B. Burlat, L. Cournac, A. Volbeda, 46 E. Garcin, X. Vernede, E. C. Hatchikian, A. Volbeda, M. Frey and S. Champ, L. Martin, B. Guigliarelli, P. Bertrand, J. Fontecilla- J. C. Fontecilla-Camps, Structure, 1999, 7, 557–566. Camps, M. Rousset and C. Leger, Proc. Natl. Acad. Sci. U. S. A., 47 Y. Nicolet, A. L. de Lacey, X. Vernde, V. M. Fernandez, 2008, 105, 11188–11193. E. C. Hatchikian and J. C. Fontecilla-Camps, J. Am. Chem. Soc., 82 O. Duche, S. Elsen, L. Cournac and A. Colbeau, FEBS J., 2005, 272, 2001, 123, 1596–1601. 3899–3908. 48 A. Volbeda, Y. Montet, X. Vernde, E. C. Hatchikian and 83 T. Buhrke, O. Lenz, N. Krauss and B. Friedrich, J. Biol. Chem., J. C. Fontecilla-Camps, Int. J. Hydrogen Energy, 2002, 27, 1449–1461. 2005, 280, 23791–23796. 49 H. Ogata, W. Lubitz and Y. Higuchi, Dalton Trans., 2009, 7577– 84 S. Dementin, F. Leroux, L. Cournac, A. L. de Lacey, A. Volbeda, 7587. C. Lger, B. Burlat, N. Martinez, S. Champ, L. Martin, 50 M. C. Marques, R. Coelho, A. L. D. Lacey, I. A. C. Pereira and O. Sanganas, M. Haumann, V. M. Fernandez, B. Guigliarelli, P. M. Matias, J. Mol. Biol., 2010, 396, 893–907. J. C. Fontecilla-Camps and M. Rousset, J. Am. Chem. Soc., 2009, 51 J. Fritsch, P. Scheerer, S. Frielingsdorf, S. Kroschinsky, B. Friedrich, 131, 10156–10164. O. Lenz and C. M. T. Spahn, Nature, 2011, 479, 249–252. 85 M. Ludwig, J. A. Cracknell, K. A. Vincent, F. A. Armstrong and 52 M.-E. Pandelia, H. Ogata, L. J. Currell, M. Flores and W. Lubitz, O. Lenz, J. Biol. Chem., 2008, 284, 465–477. Biochim. Biophys. Acta, Bioenerg., 2010, 1797, 304–313. 86 P. Wang, R. B. Best and J. Blumberger, Phys. Chem. Chem. Phys., 53 W. Lubitz, E. Reijerse and M. van Gastel, Chem. Rev., 2007, 107, 2011, 13, 7708–7719. 4331–4365. 87 P. Wang, R. B. Best and J. Blumberger, J. Am. Chem. Soc., 2011, 54 M.-E. Pandelia, H. Ogata, L. J. Currell, M. Flores and W. Lubitz, 133, 3548–3556. JBIC, J. Biol. Inorg. Chem., 2009, 14, 1227–1241. 88 T. Goris, A. F. Wait, M. Saggu, J. Fritsch, N. Heidary, M. Stein, 55 P. E. M. Siegbahn, J. W. Tye and M. B. Hall, Chem. Rev., 2007, 107, I. Zebger, F. Lendzian, F. A. Armstrong, B. Friedrich and 4414–4435. O. Lenz, Nat. Chem. Biol., 2011, 7, 310–318. 56 S. P. Best, Coord. Chem. Rev., 2005, 249, 1536–1554. 89 M.-E. Pandelia, W. Nitschke, P. Infossi, M.-T. Giudici-Orticoni, 57 M.-E. Pandelia, P. Infossi, M. T. Giudici-Orticoni and W. Lubitz, E. Bill and W. Lubitz, Proc. Natl. Acad. Sci. U. S. A., 2011, 108, Biochemistry, 2010, 49, 8873–8881. 6097–6102.

Downloaded by ETH-Zurich on 25 March 2013 58 M. Teixeira, I. Moura, A. V. Xavier, J. J. Moura, J. LeGall, 90 Y. Shomura, K. Hagiya, K.-S. Yoon, H. Nishihara and Y. Higuchi, D. V. DerVartanian, H. D. Peck and B. H. Huynh, J. Biol. Chem., Acta Crystallogr., Sect. F: Struct. Biol. Cryst. Commun., 2011, 67, 1989, 264, 16435–16450. 827–829. Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C 59 V. M. Fernandez, E. C. Hatchikian and R. Cammack, Biochim. 91 M. J. Lukey, A. Parkin, M. M. Roessler, B. J. Murphy, J. Harmer, Biophys. Acta, Protein Struct. Mol. Enzymol., 1985, 832, 69–79. T. Palmer, F. Sargent and F. A. Armstrong, J. Biol. Chem., 2009, 60 V. M. Fernandez, E. C. Hatchikian, D. S. Patil and R. Cammack, 285, 3928–3938. Biochim. Biophys. Acta, Gen. Subj., 1986, 883, 145–154. 92 M. J. Lukey, M. M. Roessler, A. Parkin, R. M. Evans, R. A. Davies, 61 S. Stripp, O. Sanganas, T. Happe and M. Haumann, Biochemistry, O. Lenz, B. Friedrich, F. Sargent and F. A. Armstrong, J. Am. 2009, 48, 5042–5049. Chem. Soc., 2011, 133, 16881–16892. 62 C. Lambertz, N. Leidel, K. G. V. Havelius, J. Noth, P. Chernev, 93 E. V. der Linden, T. Burgdorf, M. Bernhard, B. Bleijlevens, M. Winkler, T. Happe and M. Haumann, J. Biol. Chem., 2011, B. Friedrich and S. P. J. Albracht, J. Biol. Inorg. Chem., 2004, 9, 286, 40614–40623. 616–626. 63 M. K. Bruska, M. T. Stiebritz and M. Reiher, J. Am. Chem. Soc., 94 T. Burgdorf, O. Lenz, T. Buhrke, E. van der Linden, A. K. Jones, 2011, 133, 20588. S. P. J. Albracht and B. Friedrich, J. Mol. Microbiol. Biotechnol., 64 M. T. Stiebritz and M. Reiher, Inorg. Chem., 2009, 48, 7127–7140. 2005, 10, 181–196. 65 P. Soderhjelm€ and U. Ryde, THEOCHEM, 2006, 770, 199–219. 95 R. P. Happe, W. Roseboom, G. Egert, C. G. Friedrich, C. Massanz, 66 J. W. van der Zwaan, J. M. Coremans, E. C. Bouwens and B. Friedrich and S. P. Albracht, FEBS Lett., 2000, 466, 259–263. S. P. Albracht, Biochim. Biophys. Acta, Protein Struct. Mol. 96 B. Bleijlevens, T. Buhrke, E. van der Linden, B. Friedrich and Enzymol., 1990, 1041, 101–110. S. P. J. Albracht, J. Biol. Chem., 2004, 279, 46686–46691. 67 M. Carepo, D. L. Tierney, C. D. Brondino, T. C. Yang, 97 T. Burgdorf, S. Loscher,€ P. Liebisch, E. V. der Linden, A. Pamplona, J. Telser, I. Moura, J. J. G. Moura and M. Galander, F. Lendzian, W. Meyer-Klaucke, S. P. J. Albracht, B. M. Hoffman, J. Am. Chem. Soc., 2002, 124, 281–286. B. Friedrich, H. Dau and M. Haumann, J. Am. Chem. Soc., 68 S. E. Lamle, S. P. J. Albracht and F. A. Armstrong, J. Am. Chem. 2005, 127, 576–592. Soc., 2004, 126, 14899–14909. 98 C. L. McIntosh, F. Germer, R. Schulz, J. Appel and A. K. Jones, 69 S. E. Lamle, S. P. J. Albracht and F. A. Armstrong, J. Am. Chem. J. Am. Chem. Soc., 2011, 133, 11308–11319. Soc., 2005, 127, 6595–6604. 99 F. M. A. Valente, A. S. F. Oliveira, N. Gnadt, I. Pacheco, 70 O. Lenz, M. Ludwig, T. Schubert, I. Brustel,€ S. Ganskow, T. Goris, A. V. Coelho, A. V. Xavier, M. Teixeira, C. M. Soares and A. Schwarze and B. Friedrich, ChemPhysChem, 2010, 11, 1107– I. A. C. Pereira, JBIC,J. Biol. Inorg. Chem., 2005, 10, 667–682. 1119. 100 F. M. A. Valente, C. C. Almeida, I. Pacheco, J. Carita, L. M. Saraiva 71 P. Jayapal, M. Sundararajan, I. H. Hillier and N. A. Burton, Phys. and I. A. C. Pereira, J. Bacteriol., 2006, 188, 3228–3235. Chem. Chem. Phys., 2006, 8, 4086–4094. 101 F. M. A. Valente, C. C. Almeida, I. Pacheco, J. Carita, L. M. Saraiva 72 C. Baffert, M. Demuez, L. Cournac, B. Burlat, B. Guigliarelli, and I. A. C. Pereira, J. Bacteriol., 2006, 188, 3228–3235. P. Bertrand, L. Girbal and C. Leger, Angew. Chem., Int. Ed., 2008, 102 F. M. A. Valente, P. M. Pereira, S. S. Venceslau, M. Regalla, 47, 2052–2054. A. V. Coelho and I. A. C. Pereira, FEBS Lett., 2007, 581, 3341–3344.

1750 | Chem. Sci., 2012, 3, 1739–1751 This journal is ª The Royal Society of Chemistry 2012 View Article Online

103 S. H. He, M. Teixeira, J. LeGall, D. S. Patil, I. Moura, J. J. Moura, 122 M. Ihara, H. Nishihara, K.-S. Yoon, O. Lenz, B. Friedrich, D. V. DerVartanian, B. H. Huynh and H. D. Peck, J. Biol. Chem., H. Nakamoto, K. Kojima, D. Honma, T. Kamachi and I. Okura, 1989, 264, 2678–2682. Photochem. Photobiol., 2006, 82, 676–682. 104 M. K. Eidsness, R. A. Scott, B. C. Prickril, D. V. DerVartanian, 123 M. Winkler, S. Kuhlgert, M. Hippler and T. Happe, J. Biol. Chem., J. Legall, I. Moura, J. J. Moura and H. D. Peck, Proc. Natl. Acad. 2009, 284, 36620–36627. Sci. U. S. A., 1989, 86, 147–151. 124 C. E. Lubner, P. Knorzer,€ P. J. N. Silva, K. A. Vincent, T. Happe, 105 O. Sorgenfrei, E. C. Duin, A. Klein and S. P. Albracht, J. Biol. D. A. Bryant and J. H. Golbeck, Biochemistry, 2010, 49, 10264– Chem., 1996, 271, 23799–23806. 10266. 106 A. Parkin, G. Goldet, C. Cavazza, J. C. Fontecilla-Camps and 125 C. E. Lubner, A. M. Applegate, P. K., A. Ganago, D. A. Bryant, F. A. Armstrong, J. Am. Chem. Soc., 2008, 130, 13410–13416. T. Happe and J. H. Golbeck, Proc. Natl. Acad. Sci. U. S. A., 2011, 107 J. C. Fontecilla-Camps, A. Volbeda, C. Cavazza and Y. Nicolet, 108, 20988. Chem. Rev., 2007, 107, 4273–4303. 126 D. C. Ducat, G. Sachdeva and P. A. Silver, Proc. Natl. Acad. Sci. 108 X. Wang, Z. Li, X. Zeng, Q. Luo, D. J. Evans, C. J. Pickett and U. S. A., 2011, 108, 3941–3946. X. Liu, Chem. Commun., 2008, 3555–3557. 127 B. Barstow, C. M. Agapakis, P. M. Boyle, G. Grandl, P. A. Silver 109 Y. Guo, H. Wang, Y. Xiao, S. Vogt, R. K. Thauer, S. Shima, and E. H. Wintermute, J. Biol. Eng., 2011, 5,7. P. I. Volkers, T. B. Rauchfuss, V. Pelmenschikov, D. A. Case, 128 K. Vincent, J. Cracknell, A. Parkin and F. Armstrong, Dalton E. E. Alp, W. Sturhahn, Y. Yoda and S. P. Cramer, Inorg. Chem., Trans., 2005, 3397–3403. 2008, 47, 3969–3977. 129 K. Vincent, J. Cracknell, O. Lenz, I. Zebger, B. Friedrich and 110 X. Yang and M. B. Hall, J. Am. Chem. Soc., 2008, 130, 14036–14037. F. Armstrong, Proc. Natl. Acad. Sci. U. S. A.,2005,102, 16951–16954. 111 X. Yang and M. B. Hall, J. Am. Chem. Soc., 2009, 131, 10901–10908. 130 K. A. Vincent, J. A. Cracknell, J. R. Clark, M. Ludwig, O. Lenz, 112 E. J. Lyon, S. Shima, R. Boecher, R. K. Thauer, F.-W. Grevels, B. Friedrich and F. A. Armstrong, Chem. Commun., 2006, 5033– E. Bill, W. Roseboom and S. P. J. Albracht, J. Am. Chem. Soc., 5035. 2004, 126, 14239–14248. 131 J. Cracknell, K. Vincent and F. Armstrong, Chem. Rev., 2008, 108, 113 S. Shima, E. J. Lyon, R. K. Thauer, B. Mienert and E. Bill, J. Am. 2439–2461. Chem. Soc., 2005, 127, 10430–10435. 132 D. W. Mulder, E. M. Shepard, J. E. Meuser, N. Joshi, P. W. King, 114 A. M. Royer, M. Salomone-Stagni, T. B. Rauchfuss and W. Meyer- M. C. Posewitz, J. B. Broderick and J. W. Peters, Structure, 2011, 19, Klaucke, J. Am. Chem. Soc., 2010, 132, 16997–17003. 1038–1052. 115 S. Shima and K. Ataka, FEBS Lett., 2011, 585, 353–356. 133 G. Goldet, C. Brandmayr, S. T. Stripp, T. Happe, C. Cavazza, 116 M. T. Stiebritz and M. Reiher, Inorg. Chem., 2010, 49, 5818–5823. J. C. Fontecilla-Camps and F. A. Armstrong, J. Am. Chem. Soc., 117 S. Shima, Seikagaku, 2008, 80, 846–849. 2009, 131, 14979–14989. 118 M. T. Stiebritz, A. R. Finkelmann and M. Reiher, Eur. J. Inorg. 134 J. A. Stapleton and J. R. Swartz, PLoS One, 2010, 5, e10554. Chem., 2011, 1163–1171. 135 J. A. Stapleton and J. R. Swartz, PLoS One, 2010, 5, e15275. 119 B. Friedrich, J. Fritsch and O. Lenz, Curr. Opin. Biotechnol., 2011, 136 S. A. Bingham, P. R. Smith and J. R. Swartz, Int. J. Hydrogen 22, 358–364. Energy, 2012, 37, 2965. 120 M. Winkler, S. Kawelke and T. Happe, Bioresour. Technol., 2011, 137 P. Knorzer,€ A. Silakov, C. E. Foster, F. A. Armstrong, W. Lubitz 102, 8493–8500. and T. Happe, J. Biol. Chem., 2011, 287, 1489–1499. 121 M. Ihara, H. Nakamoto, T. Kamachi, I. Okura and M. Maeda, 138 D. Dogaru, S. Motiu and V. Gogonea, Int. J. Quantum Chem., 2009, Photochem. Photobiol., 2006, 82, 1677–1685. 110, 1784–1792. Downloaded by ETH-Zurich on 25 March 2013 Published on 23 February 2012 http://pubs.rsc.org | doi:10.1039/C2SC01112C

This journal is ª The Royal Society of Chemistry 2012 Chem. Sci., 2012, 3, 1739–1751 | 1751