New Approaches to an Old Problem : Original Techniques in [ FeFe ]-hydrogenase Research

Hydrogenases are redox enzymes catalyzing the conversion of protons (H) and molecular hydrogen (H2). Based on the composition of the active site cofactor, the monometallic [Fe]hydrogenase is distinguished from the bimetallic [NiFe]or [FeFe]-hydrogenase. The latter has been reported with particularly high turnover activities for both H2 release and H2 oxidation, notably at neutral pH, ambient temperatures, and negligible electric overpotential. Due to these properties, [FeFe]-hydrogenase represents the “gold standard” in enzymatic hydrogen turnover. Understanding hydrogenase chemistry is crucial for the design of transition metal complexes that may serve as sustainable proton reduction or H2 oxidation catalysts, e.g. in electrolytic devices or fuel cells. Even 20 years after the first crystal structures of [FeFe]-hydrogenase have been published, several aspects of biological hydrogen turnover are heatedly discussed. In this perspective, we give an overview on how the diversity of naturally occurring and artificially prepared, semisynthetic [FeFe]-hydrogenases deepens our understanding of hydrogenase chemistry. In parallel, we cover new results from biophysical techniques that go beyond the scope of conventional electrochemistry, X-ray diffraction, EPR, and FTIR spectroscopy. Taking into account both proton transfer and electron transfer as well as the notorious sensitivity of [FeFe]hydrogenases towards carbon monoxide, the discussion further touches upon the molecular proceedings of biological hydrogen turnover. Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 12 April 2020 doi:10.20944/preprints202004.0177.v1 © 2020 by the author(s). Distributed under a Creative Commons CC BY license.


Introduction
Molecular hydrogen (H2) is widely recognized as a green energy carrier with the potential to replace fossil fuels. 1-3 Considering our interest in a 'hydrogen society', it is important to realize that evolution has already developed an efficient economy based on H2. 4 Iron-sulfur enzymes called hydrogenases are central to the hydrogen metabolism of numerous microorganisms as they catalyze the reversible conversion between protons (H + ) and H2. 5 Depending on the nature of the metal cofactor, hydrogenases are divided into three main classes denoted as [Fe]-, [NiFe]-, or [FeFe]-hydrogenases. The latter are particular efficient catalysts with H2 evolution activities surpassing 10.000 H2/s. 6 Phylogenetically, there are indications that [FeFe]hydrogenases could be the most recent class as archaea rely on [Fe]-and [NiFe]-hydrogenase whereas certain unicellular plants encode for [FeFe]-hydrogenase exclusively. [7][8][9][10][11] All [FeFe]hydrogenases share a biologically unique cofactor, the hydrogen-activating 'H-cluster'. This organometallic moiety consists of a [4Fe-4S] cluster connected to a diiron site via a bridging cysteine ( Figure 1). The low valence iron ions of the diiron site are bridged by an azadithiolate ligand (adt) and coordinated by the π-accepting, strong-field ligands cyanide (CN -) and carbon monoxide (CO). 12,13 Details of the electronic structure and potential catalytic intermediates are discussed in Chapter 3. The biosynthesis of the H-cluster is a complex process that depends on a minimum of three specific, auxiliary enzymes. 7 Through the activity of two radical SAM enzymes (HydE and HydG), a pre-catalyst is assembled on a scaffold protein (HydF) and transferred to the hydrogenase apo-protein, generating the complete H-cluster. [14][15][16][17]  Stephenson & Stickland were the first to report enzymatic hydrogen activation in 1931 18 and the efforts aimed at unravelling Nature's design principles for biological hydrogen conversion have accelerated quite significantly over the past 20 years. In this perspective, we cover the state-of-the-art with regards to [FeFe]-hydrogenase with an emphasis on recent insights gained by studying 'new' hydrogenases, derived from both biodiversity and artificial enzymes featuring synthetically modified cofactors (Chapter 2). Furthermore, we review trends in the spectroscopic analysis of [FeFe]-hydrogenase including transient approaches and whole-cell spectroscopy (Chapter 3). Our perspective will close on a brief note on the catalytic mechanism of [FeFe]-hydrogenase (Chapter 4). [FeFe]-hydrogenases are widely used as inspiration for the design of molecular catalysts and it is crucial to understand their catalytic design principles. However, only a fraction of the known [FeFe]-hydrogenases are characterized to date and our knowledge on how the reactivity of the H-cluster is influenced by the protein fold in general and the active site pocket in particular is fragmentary at best. Truncation and mutation studies have been a powerful tool in this context but typically result in a loss of function. 19 In order to achieve a deeper understanding of biological hydrogen turnover, the biodiversity of [FeFe]-hydrogenases needs to be explored.

On the Diversity of [FeFe]-hydrogenase
Several studies of [FeFe]-hydrogenase biodiversity have been published. [20][21][22] Unfortunately, none of these studies cover the full range of diversity and naming conventions are inconsistent.
Herein, we provide a comprehensive summary of all [FeFe]-hydrogenase sub-classes identified to-date, organized based on their domain structure and following the naming convention put   [FeFe]-hydrogenases can be separated into four distinct phylogenetic groups A -D (Figure 2). Group A consists of prototypical and bifurcating [FeFe]-hydrogenases. In nature, prototypical [FeFe]-hydrogenases perform hydrogen turnover using ferredoxin as redox partner while bifurcating [FeFe]-hydrogenases perform the same reaction using both ferredoxin and NAD(H) as electron donor/acceptor. [23][24][25] This group comprises the best characterized enzymes.
Moreover, the catalytically most active enzymes have been found in Group A such as the [FeFe]-hydrogenases from Chlamydomonas reinhardtii (CrHydA1, sub-class M1), 26 Desulfovibrio desulfuricans (DdHydAB, often referred to as DdH, sub-class M2(d)), 27 and Clostridium pasteurianum and acetobutylicum (CpHydA1 and CaHydA1, referred to as CpI and CaI, sub-class M3). 28 Group B is phylogenetically distinct although its three sub-classes show similar amino acid motifs around the H-cluster as Group A [FeFe]-hydrogenases. As no representative example of Group B has been characterized so far any distinct differences between Group A and Group B [FeFe]-hydrogenase remain to be discovered. Group C has been classified as 'sensory' based on the presence of a PAS domain ( Figure 2). 8,22 Moreover, genes encoding Group C enzymes are commonly found upstream from known H2-producing [FeFe]-hydrogenases. The biochemical characterization of the M2f-type [FeFe]-hydrogenase from Thermotoga maritima further supports the notion of a sensory function as it shows modest catalytic rates and an apparent high sensitivity towards H2. 29 The closely related subclass M2e (Group D) have a similar gene localization and domain structure to M2f-type enzymes, suggestive of a similar physiological function ( Figure 2). However, the lack of a PAS domain in combination with several differences in the active-site amino acids makes their biological function unclear at the moment. 8 One enzyme from this sub-class, derived from Thermoanaerobacter mathranii, has recently been partly characterized but further investigation is needed. 30

The Influence of F-clusters and Protein Environment
[FeFe]-hydrogenase carries up to five additional iron-sulfur clusters including the M4 sub-class that carries an additional C-terminal cluster in the TRX domain (Figure 2). 22 These ferredoxintype or 'F-clusters' act as a molecular wire through the protein ensuring efficient electron transfer between biological redox partners and the H-cluster (Figure 3). The part of the enzyme that binds the F-clusters is commonly referred to as the 'F-domain' in variance to the 'Hdomain' that exclusively comprises the H-cluster. The F-clusters have a significant influence on the overall reactivity of the hydrogenase. This has been elucidated in the case of the two closely related M3-type enzymes from CpI and CaI as well as the sub-class M2 enzymes from 7 Megasphaera elsdenii (MeHydA) and DdH. [31][32][33][34] In the latter case, redox titrations combined with FTIR and EPR spectroscopy revealed that the redox equilibrium of the diiron site and its pKa is influenced by the oxidation state of the F-clusters. 33 Parallel studies of CaI and MeHydA have shown that the F-clusters affects the catalytic bias of the enzyme. In both cases, removal of the F-domain resulted in enzymes favoring H2 release following a significant drop in H2 oxidation rates. 31,32  In addition to the F-clusters the activity of [FeFe]-hydrogenase is influenced by the mass transfer of gases (e.g., H2, O2, or CO) and protons (H + ). Molecular dynamics simulations proposed a number of putative gas channels 35 , and site-directed mutagenesis could slow down O2 inactivation in individual studies. 36 However, experimental proof verifying a main trajectory for gas transfer has yet to be obtained. 37 There are indications that H2 diffuses more freely through the protein than bulky gases, and there might indeed be no such thing as a specific H2 channel. The proton transfer pathway in Group A [FeFe]-hydrogenase comprise a series of well-conserved amino acid and water residues that enable proton transfer between the H-cluster and the enzyme surface ( Figure 4). [38][39][40] It starts at the H-cluster with a cysteine residue (C1) responsible for proton transfer to the H-cluster and continues with a serine (S1), glutamate (E1, E2), and an arginine residue (R). 41,42 A methionine 'above' the H-cluster (M2) was discussed as hydrogen-bonding partner to the adt ligand possibly providing an alternative proton transfer trajectory. 43 However, in Group C [FeFe]-hydrogenases neither cysteine C1 nor methionine M2 are conserved. 29 Another putative proton transfer pathway was suggested to involve a conserved lysine residue (K) close to the distal iron ion. 13 Together, this implies that more investigations on possible proton transfer pathways in [FeFe]-hydrogenase are needed, in particular to understand the chemistry of the Group C and Group D enzymes.  The amino acid environment of the H-cluster is suspected to play a role beyond modulating gas access and proton transfer. For example, the CNligands of the H-cluster were modelled according to potential hydrogen-bonding partners, i.e. lysine K and a backbone contact involving a proline residue close to the proximal iron ion (Figure 4). 19 While the former remains speculative, site-directed mutagenesis of the 'P-motif' had significant impact on catalytic bias and spectroscopic properties. 44 This demonstrates how the protein modulates the electron density of the diiron site through Lewis acid/base interactions with the CNligands. At the [4Fe-4S] cluster, amino acid exchanges 45 and protonation differences 46 have been shown to affect the electrochemical properties of the H-cluster, and a similar effect was discussed for the orientation of a conserved serine (S2 in Figure 4). 47 Finally, a methionine residue below the Hcluster was suggested to promote the release of µCO into a semi-bridging or terminal position upon reduction of the diiron site (M1 in Figure 4). 29 The conservation of this methionine is a key difference between groups A/B and C/D and has important implications for the catalytic mechanism (Chapter 4).

Isolation of Functional [FeFe]-hydrogenase
Historically, the isolation of [FeFe]-hydrogenase has been limited by several factors. In the 1950s, enzyme isolation was dependent on homologous expression, causing significant challenges depending on the source organism. 48 During the 1970s and 1980s, heterologous overexpression in easy-to-handle host organisms such as Escherichia coli or Pichia pastoris was developed and became the method of choice for most researchers. 49,50 The first example of an active [FeFe]-hydrogenase obtained in such a way was published by Posewitz et al. in 2004, following the successful co-expression of CrHydA1 with the auxiliary maturases HydEFG. 14 The method was then refined by using E. coli strain BL21(DE3)ΔiscR in which deletion of the gene encoding for the transcriptional negative regulator IscR stimulates FeScluster biosynthesis. This resulted in at least a ten-fold increase in yield of active [FeFe]hydrogenase compared to previous reports. 51,52 Expressing [FeFe]-hydrogenase in Clostridium acetobutylicum or Shewanella oneidensis (i.e., bacteria natively expressing the HydEFG proteins) suffered from complications in handling these organisms. [53][54][55] In 2013, it was shown that [FeFe]-hydrogenase apo-protein can be heterologously overexpressed in inactive form and artificially activated using a synthetic mimic of the cofactor (Chapter 2.2). 56,57 This breakthrough in [FeFe]-hydrogenase isolation has paved the way for a new era in hydrogenase discovery. By taking advantage of the increasing amount of sequenced genomes 58 , bioinformatics can identify any [FeFe]-hydrogenase encoding gene of interest, regardless of the source organism. The genes can then be codon-optimized for expression in common hosts and synthesized in a matter of days. This approach was recently explored when eight putative [FeFe]-hydrogenase encoding genes from a range of different sub-classes were synthesized and expressed in E. coli. 30 Notably, this was done without co-expression of HydEFG.
Outlook. Several aspects motivate discovering novel [FeFe]-hydrogenases. Firstly, mechanistic studies of [FeFe]-hydrogenase are necessary in order to understand their complex chemistry and to aid the development of efficient biomimetic catalysts. As previously mentioned, the vast majority of the characterized [FeFe]-hydrogenases originate from Group A. Although the active site architecture in this group is conserved, these enzymes still exhibit clear differences in catalytic behavior: reported rates for H2 release and H2 oxidation differ by a factor of 500 and 15.000, respectively, underscoring the influence of the non-catalytic domains on the activity of the enzyme. However, also between the closely related [FeFe]hydrogenases within the same sub-class of Group A H2 release rates differ up to 40 times, 47,59-61 suggesting that even a detailed discrimination based on differences in domain structure and active site architecture does not allow for an impeccable prediction of catalytic activity.
Clearly, aspects of the molecular determinants of [FeFe]-hydrogenase turnover kinetics are poorly understood.
Secondly, the understanding of biological hydrogen metabolism is still rather limited and from an environmental and biotechnological perspective the organisms capable of metabolizing H2 needs to be better understood. This is also relevant from a medical perspective since many of these organisms are involved in human pathogenesis. 11 Beyond hydrogen turnover activity, the bifurcating, multimeric [FeFe]-hydrogenases of Group A are interesting for the study of catalytic reactions coupled to H2 turnover, e.g. NAD(P)H and CO2 conversion. 25 Moreover, the [FeFe]-hydrogenase from Clostridium beijerinckii (CbA5H, sub-class M2c) appears to display a unique tolerance towards O2. 62 These observations underscore the potential of exploring hitherto uncharacterized sub-classes in efforts to discover enzymes with unprecedented activities and properties.
Thirdly, there is a great need to increase the toolbox of available [FeFe]-hydrogenases so that the best possible candidates can be identified with regards to catalytic performance and O2 tolerance. These enzymes can then be optimized, e.g. through directed evolution and applied in industrial H2 production or in biotechnological devices such as fuel cells. 63 In the next chapter, we will discuss modification of the active site cofactor beyond the scope of nature.

Artificial Maturation and Organometallic Variants
The biosynthesis of the H-cluster proceeds via a readily isolatable intermediate containing the [4Fe-4S] cluster but lacking the diiron site. 64,65 This assembly line could be hijacked through the introduction of synthetic analogues of the diiron site ('artificial maturation') enabling the preparation of semi-synthetic [FeFe]-hydrogenases ( Figure 5). 56,57 More recently, this strategy has been utilized to identify possible intermediates in the assembly of the pre-catalyst. [66][67][68] Arguably, the main impact of artificial maturation to-date has been the simplified preparation of the active enzyme 60,69 including site-selective isotopologues, [70][71][72] which also facilitates rapid screening protocols. 30 Another important aspect is the preparation of [FeFe]-hydrogenases in which the diiron site is synthetically modified in order to generate non-natural H-clusters and enzymes with new properties. 73-76 instead of the natural adt ligand can be incorporated into the apo-protein with good yields. Such variants showed very specific turnover characteristics and allowed locking the H-cluster in specific oxidation states (Chapter 3.1). 56,57 Numerous other organometallic variants have been reported, documenting modifications all across the H-cluster ( Figure 5). The iron ions have been replaced by ruthenium, resulting in the formation of a [RuRu] bridging hydride species that appears highly stable. 77 The chalcogens have been changed from sulfur to selenium in both the [4Fe-4S] cluster as well as the diiron site. 78,79 The latter variant showed a shift in catalytic bias towards H2 release, but also suffered from a significant decrease in cofactor stability.
Modifications of the diatomic ligands revealed that monocyanide variants of the H-cluster retain remarkable residual activity both in vitro and in whole cells. [79][80][81][82][83] Outlook: The preparation of organometallic mutants will undoubtedly continue contributing to our mechanistic understanding of [FeFe]-hydrogenase. The preparation of modified Hclusters outcompeting the native cofactor remains a significant challenge, although the Selenium analogue appears to improve H2 release activity. 78,79 It is noteworthy that synthetic modification of the diiron site have a dramatic effect on the reactivity towards known inhibitors. 83 Thus, it arguably provides a convenient route towards more stable catalysts. In parallel to the enzymology aspect, the generation of highly active, semi-synthetic [FeFe]hydrogenases is of particular relevance in the context of designing small molecule systems.
Since the structural elucidation of the H-cluster two decades ago, 12,13 the design and characterization of synthetic [FeFe] complexes has been a highly active research field, and several comprehensive reviews have been published on the topic. [84][85][86] Historically, this work has been critical, e.g. in assigning the nature of the diatomic ligands, identifying the nature of the bridgehead atom, highlighting the importance of acid/base residues in the vicinity of the

Novel Methods
The electronic structure of the H-cluster has been investigated by continuous-wave and pulsed EPR spectroscopic techniques as well as Mössbauer spectroscopy and nuclear inelastic scattering (NIS). Similarly, the CO and CNligands of the H-cluster provide exquisite spectroscopic handles for vibrational spectroscopy and FTIR spectroscopy has arguably become the method of choice in the field. Albeit not novel methods, they are continuously employed in the identification of new states of potential catalytic relevance. In the following section, we will outline how these methods have provided a foundation for our current mechanistic understanding, before describing how they are used in new ways to provide even more detailed insight into H-cluster chemistry. In addition, recent data from synchrotron methods is discussed.

Electronic Properties of the H-cluster
The paramagnetic resting state Hox that gives rise to a rhombic signal (g=2.10, 2.04, 2.00; see   Figure 7 depicts the electronic configuration of key H-cluster intermediates.

Vibrational Properties of the H-cluster
The CO and CNligands preserve the low-spin character of the diiron site and couple the Hcluster to the protein environment. 44 cluster most likely concerted with a protonation event at the same site. 108 species. [101][102][103][104] This results in a mean upshift of the CO/CNsignature by ~20 cm -1 relative to Hox, which is comparable to Htrans. 110  (iv) Note the inversion of band frequencies. (B) Site-selective 13 100,107,110 Here, site-selective 13  Hred and Hsred are discussed in the next chapter.

Biophysical Investigations in cristallo
Resolving the crystal structure of the [FeFe]-hydrogenase from C. pasteurianum and D.
desulfuricans by X-ray diffractometry (XRD) enabled a molecular understanding of hydrogen catalysis. 12,13 Hot on the heels of the resting state Hox, CpI was crystallized in the presence of CO [116][117][118] and DdH was crystallized in the presence of H2 119 resulting in H-cluster geometries that served as models for Hox-CO and Hred, respectively. Albeit crystal structures provide an exceptionally strong starting point for understanding [FeFe]-hydrogenase catalysis detailed insight into the structural dynamics of the H-cluster is dependent on spectroscopy and QM calculations. As discussed in the preceding chapter, alternative ligand orientations in Hox-CO, Hred, and Hsred have been proposed. The limited spatial resolution of XRD on protein crystals impedes a distinction between the diatomic ligands at the H-cluster, which is why CO and CNwere assigned according to potential hydrogen-bonding contacts with the protein fold in the oxidized state. 12,13 At the proximal iron ion, the original CO/CNassignment was found to be correct. 44 But while the crystal structure of CO-inhibited enzyme was modelled with an apical CO ligand (position 'X' in Figure 1), [116][117][118] vibrational coupling clearly suggested two equatorial carbonyls and an apical CNligand at the distal iron ion in Hox-CO. [112][113][114] This implies pronounced rotational freedom of ligands. Such dynamics are likely to play a role in the reaction with molecular oxygen 120

and the accumulation of the diiron-site reduced states
Hred and Hsred if these indeed feature a µH ligand. The crystal structure of reduced [FeFe]hydrogenase DdH showed a partial release of the µCO ligand, which was suggested to be a feature of the reduced diiron site. 119 At first glance, this interpretation is supported by room temperature IR spectroscopy as neither DdH nor CrHydA1 show a low frequency peak in the IR signature of Hred (compare Figure 8). 110,121 But the vibrational coupling observed in Hred and Hsred is not compatible with a 'semi-bridging CO' geometry and rather suggests an apical CO ligand, 115 which would explain the observed protection against external CO binding, too. 83 Interestingly, cryogenic IR spectroscopy indicated a bridging ligand for both Hred and Hsred. [122][123][124] This is in agreement with a recent study by Artz et al. comparing cryogenic CpI crystal structures solved with both synchrotron radiation and the free electron laser (XFEL) light source at Stanford. 47 The authors were able to quantify the extent of photoreduction suggesting that in the presence of strong reductant up to 50% of CpI was found in a "more reduced conformation" featuring the H-cluster in a µCO geometry. The influence of temperature is discussed in the next chapter.
Outlook. Crystallized in the presence of H2, the structure of DdH provides insight into the reduced form(s) of the H-cluster. 119 However, a pure accumulation of Hred has never been proven in cristallo and the coordinates as reported by Nicolet 99 Once Hred´ had been properly identified 46,108 , the absence of a classic µCO ligand in Hred and Hsred was largely agreed upon. The crystal structure of reduced DdH seemed to support this, as discussed in Chapter 3.2. 119 Making use of FTIR spectroscopy on H2 and dithionite-reduced hydrogenase under cryogenic conditions, three recent studies identified low-frequency bands that were not observed at ambient temperatures. For DdH, CrHydA1, and CaI these bands were assigned to a µCO ligand in Hred (~1810 ±7 cm -1 ) and Hsred (~1790 ±12 cm -1 ). [122][123][124] The experimental variance is surprisingly high, given that the µCO band is usually well conserved among Group A [FeFe]hydrogenases (Table 1). More importantly, the µCO frequencies are in the same range as Hox and Hred´ (Figure 8), which hints at a Fe(II)-Fe(I) configuration rather than Fe(I)-Fe(I). The µCO ligand is an excellent reporter of the electron density distribution across the diiron site due to his symmetric position between Fep and Fed ( Figure 1). Additionally, the µCO vibration is uncoupled from the terminal CO ligands 46 , which allows interrogating the µCO vibration for protonation and redox differences independent of H-cluster geometry. The difference between

Fe(II)-Fe(II) in Hhyd and Fe(II)-Fe(I) in
Hox is ~60 cm -1 , for example ( Figure 8). 103 In the Outlook. Thermodynamic considerations exclude H-cluster intermediates with larger structural differences from the catalytic cycle [129][130][131] , which raised the question whether Hred and Hsred can play a role in the rapid hydrogen turnover of [FeFe]-hydrogenase (Chapter 4). 132 Today, ambient and cryogenic measurements show both the presence of an additional terminal CO ligand at room temperature 110,115,119,121 and the presence of a µCO ligand at cryogenic temperature. [122][123][124] Moreover, the geometry of Hred and Hsred under ambient conditions is hotly debated, with regards to protonation sites and whether the diiron site adopts an H2inhibited, µH geometry or an active-ready, Hox-like geometry with a µCO ligand. 115 In order to understand the interconversion of room temperature and cryogenic species of the H-cluster, it will be important to address the influence of rotational freedom and protein environment as well as proton transfer and proton-coupled electron transport in the accumulation and 'shaping' of redox states as a function of temperature. The lack of compatibility between all recent models underscores the inadequate comprehension of hydrogenase catalysis in general. 132 These considerations are of utmost importance when it comes to the interpretation of crystal structures derived from conventional, cryogenic XRD and spectroscopic investigation in cristallo (Chapter 3.2). Exploiting XFEL radiation to solve the crystal structure of oxidized and reduced [FeFe]-hydrogenase under ambient conditions is an exciting prospect towards a unification of models. Here, it will be important to investigate enzyme with clear preferences for specific redox states, either natural or semi-synthetic [FeFe]hydrogenases (Chapter 2). Moreover, XFELs can be used for serial femtosecond crystallography (SFX) that allows investigating the structure of short-lived, catalytic intermediates. [133][134][135] As [FeFe]-hydrogenases are not easily activated, we discuss suitable trigger concepts in the next chapter.

Beyond Steady-state Spectroscopy
Hydrogenases have been characterized by steady-state spectroscopy (Chapters 3.1) and X-ray crystallography (Chapter 3.2). In the case of [FeFe]-hydrogenase, various operando or in situ methods facilitated recording spectral data or activity profiles as a function of electrochemical potential [136][137][138] , gas composition 83 and reactant concentration 46 , visible light irradiation, 139,140 and temperature. 122,123 While in situ methods typically provide a time resolution between hours and seconds, transient spectroscopy allows following short-lived cofactor intermediates that are difficult to stabilize under steady-state conditions. UV/vis and IR spectroscopy are well established for the analysis of transient processes in retinal-, porphyrine-, or flavine-binding proteins. [141][142][143] This is due to advantageous absorption properties (good signal-to-noise) and the ease of handling visible and infrared light. Here, laser sources are exploited to trigger and trace the natural reactivity of the chromoprotein, i.e. in a 'flash photolysis' setup comprising separate beams (as opposed to ultrafast 'pump/probe' spectroscopy). 144,145 Hydrogenases lack a dedicated chromophore thus triggering activity is not trivial. To this end, absorption of laser light can be exploited to induce changes in temperature ('T-jump'), pH, or redox potential, the latter in combination with suitable dyes. [146][147][148] With the development of tuneable quantum cascade lasers (QCL), a powerful tabletop infrared light source became available that can be exploited as broadband probing light in a flash photolysis-type set-up. 144

Greene et al. reported the first characterization of the soluble [NiFe]-hydrogenase from
Pyrococcus furiosus by QCL flash photolysis in the mid-IR. 149 The authors used either NADH or Cd nanorods (DIR, dot-in-rod) as reductant, transferring an electron to the oxidized hydrogenase upon ionization at 355 nm or 405 nm, respectively (Ni-S → Ni-C). 150 when activated at 355 nm. 155 The transient accumulation of Hsred and Hred´ (incorrectly assigned to Hred in the publication) over Hox was found to be compatible with the relatively low turnover frequency of the enzyme (~10 H2/s). 156 124 In the future, it will be exciting to see how these triggering concepts are utilized in transient UV/vis, FTIR, or X-ray spectroscopy as well as SFX diffraction experiments. Figure 9 summarizes the different trigger concepts discussed and suggested in this chapter. Here, iron-carbonyl or iron-hydride bonds are interrogated by visible light of different color.

Biophysical Investigations and Electrochemistry in vivo
Mössbauer, EPR, and FTIR spectroscopic measurements with a good signal-to-noise ratio demand purified protein samples, often in high concentration and/or larger quantities.
However, the properties of purified hydrogenase may deviate from the enzyme in its native environment, and several aspects of hydrogen turnover were first observed in vivo. 160   The paramagnetic states Hox and Hox-CO give relatively unperturbed signals in whole-cell EPR, comparable with purified enzyme (Figure 10). 164 (Figure 10).
Outlook. The first generation of whole-cell biophysical investigations agrees with most measurements performed on purified enzyme. Recombinant hydrogenase was found to react surprisingly fast to changes in buffer, redox potential, pH, and gas composition. H-cluster states Hox, Hred, and Hhyd could be detected in vivo, even under mildly alkaline conditions. 111 However, elusive species like Hred´ and Hsred were barely observed. This is indicative of oxidation and docking to the energy metabolism of host cells, i.e. via the bacterial ferredoxin as demonstrated by Barstow et al. before. 165 In the future, it will be exciting to study the physiological involvement of hydrogenase in recombinant bacteria and plant cells, e.g. in the context of bifurcation or photosynthesis. 162,166 Using suitable dyes, fluorescence microscopy could give further insight in pH changes or H2 release of single cells. [167][168][169] Methodically, EPR and FTIR spectroscopy proved to be powerful tools for the analysis of recombinant whole cells.
It may be possible to optimize the chemical sensitivity and spatial resolution of near-field techniques to reveal the intracellular location of hydrogenase based on the CO/CNsignature of the H-cluster. [170][171][172][173] Film electrochemistry on intact E. coli cells suffered from low catalytic currents; however, introducing electron relays in the outer membrane may establish electrophysiological measurements of hydrogenase activity.

The Catalytic Mechanism
In Chapter 2. Based on the wealth of biochemical, structural, and spectroscopic data, different catalytic cycles have been suggested. These are subject to constant evolution; therefore, we refrain from reviewing any details. Figure 11   The 5-step model in Figure 11 is based on the interpretation of Hred and Hsred as H-cluster intermediates with an open coordination site at the distal iron ion, not unlike Hox or Hred´. As interpretation of the diiron site geometry in Hred and Hsred. 132 Here, the presence of a terminal CO ligand at the distal iron ion alongside stabilization of charge via a bridging hydride species puts Hred and Hsred in the position of 'H2-inhibited' intermediates. 83 The 3-step model in Figure 11 starts with the conversion of Hox into Hred´, including reduction of the

Concluding Remarks
We only just scratched the surface of hydrogenase biodiversity. Moving forward, investigating novel [FeFe]-hydrogenases will help understanding the molecular proceedings of hydrogen turnover. Even less comprehensive is our knowledge about how amino acid patterns shape proton transfer, gas channels, and the dynamic geometry of the H-cluster. While organometallic variants are not likely to enhance catalytic efficiency, interfering with nature will facilitate our comprehension of the inner workings of [FeFe]-hydrogenase. Future experiments must facilitate connections between data collected under ambient or cryogenic conditions, on isolated enzyme or whole cells, and in-between species. Critically, steady-state measurements must also be complemented with transient spectroscopy studies and high-level QM calculations to verify the catalytic relevance of identified states. Fortunately, the community is not shy tackling such problems with novel techniques, e.g. two-dimensional infrared spectroscopy (2DIR), 174 Fourier-transformed alternating current voltammetry (FTacV) 175 , or scattering-type scanning near-field optimal microscopy (sSNOM). 111 More will follow, from the biological, chemical, and physical corners of the community, and it will be exciting to participate in the progress.

Acknowledgements
We would like to thank Prof. Dr. David R. Britt and Dr. Guodong Rao who provided the EPR spectra of Hox and HoxH. The work presented in this article is supported by Novo Nordisk