The Hydricity and Reactivity Relationship in [ FeFe ]-hydrogenases

Reactivity of transition metal catalysts is controlled by covalent and non-covalent interactions that tune thermodynamic properties including hydricity. Hydricity is critical to catalytic activity and for modulating the reduction or oxidation of chemical compounds. Likewise, enzymes can employ transition metal cofactors and use metal-hydride intermediates tuned by protein frameworks to selectively control reactivity. One example, the [FeFe]-hydrogenases, catalyze reversible H 2 activation with H 2 oxidation to H + reduction ratios spanning ~10 7 in rate, offering a model to determine the extent that hydricity controls reactivity. To address this question, the hydricity of the catalytic H cluster of two [FeFe]-hydrogenases, CpI and CpII, were compared. We show that for CpI, the higher rates of H + reduction correspond to a more hydridic H cluster, whereas CpII, which strongly favors H 2 oxidation, has a less hydridic H cluster. The results demonstrate that enzymes manipulate metal cofactor hydricity to enable an extraordinary range of chemical reactivity.


Introduction
Metal-ligand complexes have fundamental roles in reduction-oxidation reactions, where covalent and non-covalent interactions strongly in uence the metal-ligand reactivity. 1 One class of metal-ligand complexes, the transition metal hydrides, functions as intermediates in a broad number of chemical transformations. 2,3 For example, in small molecule activation reactions, hydricity of a catalytic metal site strongly in uences the control of turnover rates and end-product speci cities. 4 This effect of metal hydricity on a reduction-oxidation reaction is exempli ed by control of H 2 reactivity in Ni-diphosphine complexes. 5 In these series of compounds, a more hydridic Ni-site strongly favors reduction of protons to form H 2 , whereas weakly hydridic Ni sites favor H 2 binding and activation. 6 A caveat is that proton donoracceptor pKa in uences the reactivity trends, and matching of hydricity to pKa provides a means to further tune the end-product preferences of metal hydride based catalysts and their chemical reactions. 7,8 Transition metals are critical to biological energy transformation reactions and are found in many enzyme active sites, including the FeMo cofactor in nitrogenase, leading to reduction of dinitrogen to ammonia, 9 the tungsten-or molybdenum-containing formate dehydrogenases, which reversibly reduce CO 2 to formate, 10 methyl coenzyme reductase, 11 which catalyzes the reduction of CH 3 -S-CoM to methane, and hydrogenases, which catalyze reversible H 2 activation chemistry. 12,13 For the latter, striking differences in preferences for H 2 oxidation or H 2 production reactivity for a series of three unique [FeFe]hydrogenases from Clostridium pasteurianum has been shown. 12,14 Each [FeFe]-hydrogenase incorporates a organometallic, iron-sulfur rich H cluster (Fig. 1), with additional iron-sulfur clusters, or F clusters, that can provide electron transfer functions. 15 The H cluster is composed of a [4Fe-4S] cubane linked by a protein cysteine thiolate ligand to a organometallic, diiron subsite ([2Fe]). The [2Fe] subsite has pairs of terminal CO and CN (t-CO and t-CN) ligands, a bridging CO (µ-CO) of the Fe atom pair, and a bridging azadithiolate ligand (Fig. 1). 16 for H 2 oxidation, with a 10 4 -fold higher reaction rate than proton reduction whereas C. pasteurianum [FeFe]-hydrogenase I, or CpI, has high rates of both proton reduction and H 2 oxidation, or more neutral reactivity. 12,26 The catalytic site of CpII also differs from CpI due to several changes in nearby noncovalent amino-acid interactions that are proposed to create a more hydrophobic environment around the catalytic site H cluster. 12 In this study, the biophysical properties of the CpII catalytic intermediates were compared to CpI in order to determine if there are underlying electronic properties that change H cluster thermodynamics to account for differences in reactivity. The outcomes clearly demonstrate that the CpII H cluster electronic structure is unique among [FeFe]-hydrogenases, and that the reactivity differences result from substantial changes in the H cluster hydricity that tune reactivity to favor H 2 oxidation.

Results And Discussion
CpII Resting State Electronic Structure. To identify possible differences in the electronic structure of the H cluster in CpII compared to other [FeFe]-hydrogenases, EPR and FTIR measurements were initially carried out on the well-characterized resting state of the enzyme, H ox . For CpII, H ox has a unique S = 1/2 EPR signal, with g-values of 2.08, 2.03, and 2.00 ( Figure S1). However, the T opt value was 40 K, 27 and higher than the T opt average of 15 K for other [FeFe]-hydrogenases with differing reactivity pro les (Table S1).
The difference in the relaxation property of the signal suggests that the H cluster in CpII has more [2Fe-2S] cluster character 28 compared to [FeFe]-hydrogenases with more neutral reactivity such as CpI, suggesting that CpII may have subtle differences in its electronic structure or distribution of spin on [2Fe]. 22 The corresponding FTIR spectra (Fig. 2) of the resting state CpII (H ox ) has vCN bands at 2082 and 2069 cm − 1 , and terminal vCO bands at 1969 and 1944 cm − 1 , however the vCO band of the µ-CO that bridges the diiron sub-site Fe atoms was at 1752 cm − 1 , or ~ 50 cm − 1 downshifted compared to the FTIR spectra of H ox for CpI (Table 1) and other [FeFe]-hydrogenases (Table S2). This signi es an increase of π back-bonding from Fe D →µ-CO in CpII, which is further illustrated by the differences in the FTIR spectrum of H ox sample treated with CO (H ox -CO). CO is a π-acceptor ligand that terminally binds at the ligand exchangeable site of the Fe D atom (Fig. 1). 21 For CpII, the H ox -CO form has an exogenous vCO band at 2023 cm − 1 , which is upshifted by 6 cm − 1 relative to same band at 2017 cm − 1 in CpI H ox -CO (Fig. 2, Table 1). Thus, the downshift of µ-CO frequency in CpII (greater π back-bonding from Fe D into µ-CO) compared to CpI is also matched by an upshift in t-CO frequency (less π back-bonding from Fe D into t-CO) in CpII relative to CpI (Table 1), owing to differences in the underlying H cluster electronic structures between the two enzymes. Collectively, the EPR and FTIR properties of resting state CpII indicate differences in electronic structure compared to CpI, which is likely to affect the properties of catalytic intermediates.  Figure S2, Table S3). 26 As potentials became more negative, CpII formed only H sred , and there was no equivalent H hyd signal 29,36 even at potentials as low as -625 mV ( Figure S2). The addition of the natural substrate, H 2 and poising at -490 mV, led to higher enrichments of H sred (Table S3) (Fig. 1). 29 The corresponding FTIR spectra of CpII were measured as it is possible to observe a more complete pro le of the reduced state populations due to all the reduced states having unique IR spectral signatures (Table S2). This is evident as collective downshifts of t-CO bands for both the H red and H sred states compared to H ox (Fig. 2) owing to the formal reduction of the H cluster subsites. 37,38 Under the reducing potentials used here, the H cluster of CpII predominantly equilibrates into the 1-electron reduced H red state, with a smaller population of the 2-electron reduced H sred state being observed at -625 mV (Fig. 3A), consistent with the EPR results (  (Fig. 3A), consistent with the EPR results from Figure S2, Table S3. compared to the resting state spectrum (Fig. 3). Further reduction of CpII C→S to more reducing potentials led to a more enriched formation of H sred compared to CpII (Fig. 3A). 38 Likewise, the corresponding EPR spectra recorded at 5 K showed an overall weak signal that increased in intensity at lower reduction potentials due to the increased presence of both H sred 34,35 and reduced F clusters (Fig. 3B, Figure S3, to the trans-effect of the terminally bound hydride on the adjacent Fe D atom (Fig. 1). 38 (Fig. 4) led to a downshift of the vCO bands owing to binding and activation of H 2 or D 2 accompanied by reduction of the resting state H cluster (Fig. 1). The FTIR spectra indicate CpII is mainly poised in the H red state with terminal vCO at 1918 and 1889 cm − 1 , and a µ-CO band at 1730 cm − 1 , with a smaller population of H sred (Fig. 4, top panel). A clear lack of an H/D isotope sensitive µ-CO band, the de ning feature of the H hyd spectrum 36 (Table S2), strongly supports that the H cluster of CpII is tuned to favor H sred in the 2-electron reduced state over H hyd , and again consistent with enrichment of H sred in CpII C→S . Due to the lack of an observable "H hyd " state, the H/D exchange activity of CpII was measured in order to determine whether the catalytic mechanism involves heterolytic H 2 activation. 42 In reactions under H 2 in D 2 O, puri ed CpII co-evolved both HD and D 2 , (Figure S4), con rming the catalytic mechanism of CpII involves formation of a H hyd state, which due to low hydricity favors H 2 oxidation ( Figure S5).
CpII C → S Reactivity has Increased Bias Towards H 2 oxidation. The spectroscopic properties of CpII C→S demonstrate a change in the proton relay that leads to a greater stabilization of H sred, compared to CpII.
Based on the observed scaling relationship between H cluster hydricity and enzymatic reactivity in CpII versus CpI, where H 2 oxidation rates are favored by a less hydridic H cluster, the reactivity of CpII C→S is predicted to further shift towards H 2 oxidation compared to CpII. The reactivity ratio of CpII C→S for H 2 oxidation-to-proton reduction is 10 4 versus 10 3 for CpII (Table S4), a difference of 10-fold in favor of H 2 oxidation. Thus, a change in the hydricity of the H cluster leads to more favorable formation of H sred over H hyd and favors H 2 oxidation over proton reduction. This effect is accentuated for CpII by a shift in the pKa landscape of proton transfer, where the electronic structure leads to a more acidic Fe D relative to the proton relay Cys residue (see Fig. 1). This difference is further magni ed in CpII C→S where -SH to -OH creates an even larger difference in the pKa between the proton donor and hydride binding sites.

Conclusions
Redox enzymes exhibit broad preferences for catalyzing the reduction or oxidation of a chemical substrate. For hydrogenases, several mechanisms have been proposed to account for differences reversibility of H 2 activation as a preference for H + reduction or H 2 oxidation. Due to the fact that hydrogenases share a common catalytic site cofactor and use similar redox intermediates, reactivity differences have been mainly attributed to differences in accessory clusters [43][44][45] and extended proton transfer pathways. [46][47][48][49] A less obvious property of enzymes that is known to profoundly affect reactivity of transition metal complexes is the thermodynamic hydricity. The understanding of hydricity and reactivity is embodied in scaling relationships that directly account for observed rate differences of reduction-oxidation reactions.
Model compounds control broad reactivity ranges using primary and secondary sphere chemistry. 5,33 This relationship is exempli ed for CpI and CpII in Fig. 5 where the dramatic difference in reactivity is a direct outcome of the thermodynamic hydricity. This result now explains why manipulation of [FeFe]hydrogenase catalytic site microenvironments 29,36 can be modeled as changes in hydricity by computational analysis. 33 What these studies reveal is that tuning of the H cluster hydricity and pKa regimes provides for kinetic control over reactive intermediates in the heterolytic exchange of H 2 , H + , and electrons ( Figure S5). 50  The organic framework of Ni-phosphines, diiron-organometallic complexes have been extensively modi ed to understand how hydricity, pKa, and redox potential of metal sites determine catalytic properties. [53][54][55] Metal site sterics, solvation networks, electronic structure, hydrogen bonding and electrostatics contribute to the hydricity, by scaling relationships. It is possible that the subtle changes in the H cluster microenvironment of CpII versus CpI exacts a similar range of control. Model complexes have an additional layer of control through changing the solvent system or using rare-earth metals. However, the broad range of reactivities in [FeFe]-hydrogenases demonstrates that reactivity may be su ciently tuned using earth-abundant materials and aqueous solvent.

Summary
We have conducted a thorough analysis of the catalytic site electronic structure and relationship to the H 2 activation mechanism of [FeFe]-hydrogenase CpII. These changes underlie and determine the hydricity to favor the formal 2-electron reduced catalytic site (lower hydricity) over a hydride bound intermediate, and reactivity towards H 2 oxidation. Changing the proton-donor amino acid from acidic to basic, further shifts reactivity towards H 2 oxidation by introducing a barrier to protonation of the hydride bound H cluster.
These changes are likely elicited by subtle differences in the amino acids that comprise the microenvironment of catalytic site H cluster, establishing a template for how the natural chemistry of redox enzymes can be utilized to optimize organometallic site reactivity. It may also make possible the implementation of rational design to reengineer redox enzymes and their reactivities for optimized production of desired compounds.