Hydrogenases catalyze the formation of hydrogen. The cofactor ('H-cluster') of [FeFe]-hydrogenases consists of a [4Fe-4S] cluster bridged to a unique [2Fe] subcluster whose biosynthesis in vivo requires hydrogenase-specific maturases. Here we show that a chemical mimic of the [2Fe] subcluster can reconstitute apo-hydrogenase to full activity, independent of helper proteins. The assembled H-cluster is virtually indistinguishable from the native cofactor. This procedure will be a powerful tool for developing new artificial H₂-producing catalysts.

[FeFe]-hydrogenases catalyze the reversible reduction of protons to molecular hydrogen with extremely high efficiency. The active site (“H-cluster”) consists of a [4Fe–4S]H cluster linked through a bridging cysteine to a [2Fe]H subsite coordinated by CN− and CO ligands featuring a dithiol-amine moiety that serves as proton shuttle between the protein proton channel and the catalytic distal iron site (Fed). Although there is broad consensus that an iron-bound terminal hydride species must occur in the catalytic mechanism, such a species has never been directly observed experimentally. Here, we present FTIR and nuclear resonance vibrational spectroscopy (NRVS) experiments in conjunction with density functional theory (DFT) calculations on an [FeFe]-hydrogenase variant lacking the amine proton shuttle which is stabilizing a putative hydride state. The NRVS spectra unequivocally show the bending modes of the terminal Fe–H species fully consistent with widely accepted models of the catalytic cycle.

Active [FeFe] hydrogenases can be obtained by expressing the unmaturated enzyme in Escherichia coli followed by incubation with a synthetic precursor of the binuclear [2Fe] subcluster, namely: [NEt4]2[Fe2(adt)(CO)4(CN)2] (adt = [S-CH2-NH-CH2-S](2-)). The binuclear subsite Fe2(adt)(CO)3(CN)2 is attached through a bridging cysteine side chain to a [4Fe-4S] subcluster already present in the unmaturated enzyme thus yielding the intact native "H-cluster". We present FTIR electrochemical studies of the [FeFe] hydrogenase from Chlamydomonas reinhardtii, CrHydA1, maturated with the precursor of the native cofactor [Fe2(adt)(CO)4(CN)2](2-) as well as a non-natural variant [Fe2(pdt)(CO)4(CN)2](2-) in which the bridging amine functionality is replaced by CH2. The obtained active enzyme CrHydA1(adt) shows the same redox states in the respective potential range as observed for the native system (E(ox/red) = -400 mV, E(red/sred) = -470 mV). For the Hox → Hred transition the reducing equivalent is stored on the binuclear part, ([4Fe-4S](2+)Fe(II)Fe(I) → [4Fe-4S](2+)Fe(I)Fe(I)), while the Hred → Hsred transition is characterized by a reduction of the [4Fe-4S] part of the H-cluster ([4Fe-4S](2+)Fe(I)Fe(I) → [4Fe-4S](+)Fe(I)Fe(I)). A similar transition is reported here for the CO inhibited state of the H-cluster: ([4Fe-4S](2+)Fe(I)Fe(II)CO → [4Fe-4S](+)Fe(I)Fe(II)CO). An FTIR electrochemical study of the inactive variant with the pdt ligand, CrHydA1(pdt), identified two redox states H(pdt)-ox and H(pdt)-"red". Both EPR and FTIR spectra of H(pdt)-ox are virtually identical to those of the H(adt)-ox and the native Hox state. The H(pdt)-"red" state is also characterized by a reduced [4Fe-4S] subcluster. In contrast to CrHydA1(adt), the H(pdt)-ox state of CrHydA1(pdt) is stable up to rather high potentials (+200 mV). This study demonstrates the distinct redox coupling between the two parts of the H-cluster and confirms that the [4Fe-4S]H subsite is also redox active and as such an integral part of the H-cluster taking part in the catalytic cycle.

[FeFe]-hydrogenases are to date the only enzymes for which it has been demonstrated that the native inorganic binuclear cofactor of the active site Fe2(adt)(CO)3(CN)2 (adt = azadithiolate = [S-CH2-NH-CH2-S](2-)) can be synthesized on the laboratory bench and subsequently inserted into the unmaturated enzyme to yield fully functional holo-enzyme (Berggren, G. et al. (2013) Nature 499, 66-70; Esselborn, J. et al. (2013) Nat. Chem. Biol. 9, 607-610). In the current study, we exploit this procedure to introduce non-native cofactors into the enzyme. Mimics of the binuclear subcluster with a modified bridging dithiolate ligand (thiodithiolate, N-methylazadithiolate, dimethyl-azadithiolate) and three variants containing only one CN(-) ligand were inserted into the active site of the enzyme. We investigated the activity of these variants for hydrogen oxidation as well as proton reduction and their structural accommodation within the active site was analyzed using Fourier transform infrared spectroscopy. Interestingly, the monocyanide variant with the azadithiolate bridge showed ∼50% of the native enzyme activity. This would suggest that the CN(-) ligands are not essential for catalytic activity, but rather serve to anchor the binuclear subsite inside the protein pocket through hydrogen bonding. The inserted artificial cofactors with a propanedithiolate and an N-methylazadithiolate bridge as well as their monocyanide variants also showed residual activity. However, these activities were less than 1% of the native enzyme. Our findings indicate that even small changes in the dithiolate bridge of the binuclear subsite lead to a rather strong decrease of the catalytic activity. We conclude that both the Brønsted base function and the conformational flexibility of the native azadithiolate amine moiety are essential for the high catalytic activity of the native enzyme.

Hydrogenases are enzymes that catalyze the oxidation of H2 as well as the reduction of protons to form H2. The active site of [FeFe] hydrogenase is referred to as the "H-cluster" and consists of a "classical" [4Fe-4S] cluster connected via a bridging cysteine thiol group to a unique [2Fe]H sub-cluster, containing CN(-) and CO ligands as well as a bidentate azadithiolate ligand. It has been recently shown that the biomimetic [Fe2(adt)(CO)4(CN)2](2-) (adt(2-) = azadithiolate) complex resembling the diiron sub-cluster can be inserted in vitro into the apo-protein of [FeFe] hydrogenase, which contains only the [4Fe-4S] part of the H-cluster, resulting in a fully active enzyme. This synthetic tool allows convenient incorporation of a variety of diiron mimics, thus generating hydrogenases with artificial active sites. [FeFe] hydrogenase from Chlamydomonas reinhardtii maturated with the biomimetic complex [Fe2(pdt)(CO)4(CN)2](2-) (pdt(2-) = propanedithiolate), in which the bridging adt(2-) ligand is replaced by pdt(2-), can be stabilized in a state strongly resembling the active oxidized (Hox) state of the native protein. This state is EPR active and the signal originates from the mixed valence Fe(I)Fe(II) state of the diiron sub-cluster. Taking advantage of the variant with (15)N and (13)C isotope labeled CN(-) ligands we performed HYSCORE and ENDOR studies on this hybrid protein. The (13)C hyperfine couplings originating from both CN(-) ligands were determined and assigned. Only the (15)N coupling from the CN(-) ligand bound to the terminal iron was observed. Detailed orientation selective ENDOR and HYSCORE experiments at multiple field positions enabled the extraction of accurate data for the relative orientations of the nitrogen and carbon hyperfine tensors. These data are consistent with the crystal structure assuming a g-tensor orientation following the local symmetry of the binuclear sub-cluster.