Search

Determining the redox potentials of protein cofactors and how they are influenced by their molecular neighborhoods is essential for basic research and many biotechnological applications, from biosensors and biocatalysis to bioremediation and bioelectronics. The laborious determination of redox potential with current experimental technologies pushes forward the need for computational approaches that can reliably predict it. Although current computational approaches based on quantum and molecular mechanics are accurate, their large computational costs hinder their usage. In this work, we explored the possibility of using more efficient QSPR models based on machine learning (ML) for the prediction of protein redox potential, as an alternative to classical approaches. As a proof of concept, we focused on flavoproteins, one of the most important families of enzymes directly involved in redox processes. To train and test different ML models, we retrieved a dataset of flavoproteins with a known midpoint redox potential (Em) and 3D structure. The features of interest, accounting for both short- and long-range effects of the protein matrix on the flavin cofactor, have been automatically extracted from each protein PDB file. Our best ML model (XGB) has a performance error below 1 kcal/mol (∼36 mV), comparing favorably to more sophisticated computational approaches. We also provided indications on the features that mostly affect the Emvalue, and when possible, we rationalized them on the basis of previous studies.

The apparently simple dihydrogen formation from protons and electrons (2H+ + 2e- ↔ H2) is one of the most challenging reactions in nature. It is catalyzed by metalloenzymes of amazing complexity, called hydrogenases. A better understanding of the chemistry of these enzymes, especially that of the [NiFe]-hydrogenases subgroup, has important implications for production of H2 as alternative sustainable fuel. In this work, reactivation mechanism of the oxidized and inactive Ni-B and Ni-A states of the [NiFe]-hydrogenases active site has been investigated using density functional theory. Results obtained from this study show that one-electron reduction and protonation of the active site promote the removal of the bridging hydroxide ligand contained in Ni-B and Ni-A. However, this process is sufficient to activate only the Ni-B state. H2 binding to the active site is required to convert Ni-A to the active Ni-SIa state. Here, we also propose a reasonable structure for the spectroscopically well-characterized Ni-SIr and Ni-SU species, formed respectively from the one-electron reduction of Ni-B and Ni-A. Ni-SIr, depending on the pH at which the reaction occurs, features a bridging hydroxide ligand or a water molecule terminally coordinated to the Ni atom, whereas in Ni-SU a water molecule is terminally coordinated to the Fe atom, and the Cys64 residue is oxidized to sulfenate. The sulfenate oxygen atom in the Ni-A state affects the stereoelectronic properties of the binuclear cluster by modifying the coordination geometry of Ni, and consequently, by switching the regiochemistry of H2O and H2 binding from the Ni to the Fe atom. This effect is predicted to be at the origin of the different reactivation kinetics of the oxidized and inactive Ni-B and Ni-A states

FeFe hydrogenases catalyze H2 oxidation and formation at an inorganic active site (the "H-cluster"), which consists of a [Fe2(CO)3(CN)2(dithiomethylamine)] subcluster covalently attached to a Fe4S4 subcluster. This active site is photosensitive: visible light has been shown to induce the release of exogenous CO (a reversible inhibitor of the enzyme), shuffle the intrinsic CO ligands, and even destroy the H-cluster. These reactions must be understood because they may negatively impact the use of hydrogenase for the photoproduction of H2. Here, we explore in great detail the reactivity of the excited states of the H-cluster under catalytic conditions by examining, both experimentally and using TDDFT calculations, the simplest photochemical reaction: the binding and release of exogenous CO. A simple dyad model can be used to predict which excitations are active. This strategy could be used for probing other aspects of the photoreactivity of the H-cluster.

A comparative analysis of a series of DFT models of [NiFe]-hydrogenases, ranging from minimal NiFe clusters to very large systems including both the first and second coordination sphere of the bimetallic cofactor, was carried out with the aim of unraveling which stereoelectronic properties of the active site of [NiFe]-hydrogenases are crucial for efficient H2 binding and cleavage. H2 binding to the Ni-SIa redox state is energetically favored (by 4.0 kcal mol-1) only when H2 binds to Ni, the NiFe metal cluster is in a low spin state, and the Ni cysteine ligands have a peculiar seesaw coordination geometry, which in the enzyme is stabilized by the protein environment. The influence of the Ni coordination geometry on the H2 binding affinity was then quantitatively evaluated and rationalized analyzing frontier molecular orbitals and populations. Several plausible reaction pathways leading to H2 cleavage were also studied. It turned out that a two-step pathway, where H2 cleavage takes place on the Ni-SIa redox state of the enzyme, is characterized by very low reaction barriers and favorable reaction energies. More importantly, the seesaw coordination geometry of Ni was found to be a key feature for facile H2 cleavage. The discovery of the crucial influence of the Ni coordination geometry on H2 binding and activation in the active site of [NiFe]-hydrogenases could be exploited in the design of novel biomimetic synthetic catalysts. © 2014 American Chemical Society.

This paper reports on the protonation of phosphine-substituted diiron diphosphido carbonyls, analogues of diiron dithiolato centers at the active sites of hydrogenase enzymes. Reaction of the diphosphines (CH2)(n)(PPhH)(2) (n = 2 (edpH(2)) and n = 3 (pdpH(2))) with Fe-3(CO)(12) gave excellent yields of Fe-2(edp)(CO)(6) (1) and Fe-2(pdp)(CO)(6) (2). Substitution of Fe-2(edp)(CO)(6) with PMe3 afforded Fe-2(edp)(CO)(2)(PMe3)(4) (3; nu(CO) 1855 and 1836 cm(-1)). Crystallographic analysis showed that 3 adopts an idealized C-2 symmetry, with pairs of phosphine ligands occupying apical-basal sites on each Fe center. Relative to that in the dithiolato complex, the Fe-Fe bond (2.7786(8) angstrom) is elongated by 0.15 angstrom. Treatment of 3 with H(OEt2)(2)BAr4F (Ar-F = C6H3-3,5-(CF3)(2)) gave exclusively the C-2-symmetric mu-hydride complex [HFe2(edp)(CO)(2)(PMe3)(4)](+). This result contrasts with the behavior of the analogous ethanedithiolate Fe-2(edt)(CO)(2)(PMe3)(4) (edt = 1,2-C2H4S2), protonation of which gives both the bridging and terminal hydride complexes. This difference points to the participation of the sulfur centers in the formation of terminal hydrides. The absence of terminal hydride intermediates was also revealed in the protonation of the diphosphine diphosphido complexes Fe-2(pdp)(CO)(4)(dppv) (4; dppv = cis-1,2-C2H2(PPh2)(2)) and Fe-2(edp)(CO)(4)(dppbz) (5; dppbz = 1,2-C6H4(PPh2)(2)). Protonation of these cliphosphine complexes afforded mu-hydrido cations with apical-basal diphosphine ligands, which convert to the isomer where the diphosphine is dibasal. In contrast, protonation of the dithiolato complex Fe-2(pdt)(CO)(4)(dppv) gave terminal hydrides, which isomerize to mu-hydrides. In a competition experiment, 4 was shown to protonate faster than Fe-2(pdt)(CO)(4)(dppv)

The transport of small ligands to active sites of proteins is the basis of vital processes in biology such as enzymatic catalysis and cell signaling, but also of more destructive ones including enzyme inhibition and oxidative damage. Here, we show how a diffusion-reaction model solved by means of molecular dynamics and density functional theory calculations provides novel insight into the transport of small ligands in proteins. In particular, we unravel the existence of an elusive, dynamically formed gas channel, which CO2 takes to diffuse from the solvent to the active site (C-cluster) of the bifunctional multisubunit enzyme complex carbon monoxide dehydrogenase/acetyl-CoA synthase (CODH/ACS). Two cavities forming this channel are temporarily created by protein fluctuations and are not apparent in the X-ray structures. The ligand transport is controlled by two residues at the end of this tunnel, His113 and His 116, and occurs on the same time scale on which chemical binding to the active site takes place (0.1-1 ms), resulting in an overall binding rate on the second time scale. We find that upon reduction of CO2 to CO, the newly formed Fe-hydroxy ligand greatly strengthens the hydrogen-bond network, preventing CO from exiting the protein through the same way that CO2 takes to enter the protein. This is the basis for directional transport of CO from the production site (C-cluster of CODH subunit) to the utilization site (A-cluster of ACS subunit). In view of these results, a general picture emerges of how large proteins guide small ligands toward their active sites

DFT/BP86/TZVP and DFT/B3LYP/TZVP have been used to investigate systematically the reaction pathways associated with the H-transfer step, which is the rate-determining step of the reaction HCOO- reversible arrow CO2 + H+ + 2e(-), as catalyzed by metalloenzyme formate dehydrogenase (FDH). Actually, the energetics associated with the transfer from formate to all H (proton or hydride) acceptors that are present within the FDH active site have been sampled. This study points to a viable intimate mechanism in which the metal center mediates H transfer from formate to the final acceptor, i.e. a selenocysteine residue. The Mo-based reaction pathway, consisting of a beta-H elimination to metal with concerted decarboxylation, turned out to be favored over previously proposed routes in which proton transfer occurs directly from HCOO- to selenocysteine. The proposed reaction pathway is reminiscent of the key step of metal-based catalysis of the water-gas shift reaction.

The DFTB and DFT methods are applied to the study of different forms of the [Cu(HGGG)(Py)] complex in water, with the aim of identifying the most stable isomer. The DFTB calculations were possible thanks to a careful parametrization of the atom–atom repulsive energy terms for Cu–H, Cu–C, Cu–N, and Cu–O. The speciation process is carried out by computing different DFTB-steered molecular dynamics (SMD) trajectories, each of which ends in a well-defined different form. The last frame of each trajectory is subjected to geometry optimization at both DFTB and DFT levels, leading to a different isomer. From the corresponding energy values, a rank of relative stability of the isomers can be established. The computational protocol developed here is of general applicability to other metal–peptide systems and represents a new powerful tool for the study of speciation of metal-containing systems in water solution, particularly useful when the full characterization of the compound cannot be carried out on the basis of experimental results only

Carbon monoxide is often described as a competitive inhibitor of FeFe hydrogenases, and it is used for probing H2 binding to synthetic or in silico models of the active site H-cluster. Yet it does not always behave as a simple inhibitor. Using an original approach which combines accurate electrochemical measurements and theoretical calculations, we elucidate the mechanism by which, under certain conditions, CO binding can cause permanent damage to the H-cluster. Like in the case of oxygen inhibition, the reaction with CO engages the entire H-cluster, rather than only the Fe2 subsite.

The photochemistry originated by the Fe2(CO)6S 2 low-lying excited states is investigated using density functional theory (DFT), time-dependent density functional theory (TDDFT), and the quantum theory of atoms in molecules (QTAIM) methods.The 11A1 excitation is the most intense among the low-lying excited states computed at the TDDFT level, which is assigned to the 449 nm metal-to-ligand charge transfer (MLCT) band observed experimentally. We then investigated the nine excited states in a range of ±35 nm centered on the 11A1 excitation energy, which reproduces the range of wavelengths covered by a recent 450 ± 35 nm low-energy laser photolysis experiment. The results presented in this paper suggest that the 450 nm photochemistry recently investigated proceeds mainly through the 11B2 lowest energy singlet excited state. The comparison between tetrahedrane ground state and 11B2 vertical excited state QTAIM atomic net charges evidenced the Fe→S MLCT character but also a significant CO→S ligand-to-ligand charge transfer (LLCT). Geometry relaxation of the nine excited state structures shows a S-S bond distance elongation that reaches the highest value for the 11B2 state. Moreover, during geometry optimization of the 11B2 state, the HOMO/LUMO crossing occurs, favoring the formation of the Fe-Fe butterfly isomer upon decay to the ground state, in agreement with experimental findings. Delocalization indexes allow us to describe the shift of the bonding electron density along the 1 1B2 photochemical reaction path that connects the tetrahedrane ground state to the Fe-Fe butterfly intermediate. Along this path, the S-S bond is progressively weakened until its breaking in the Fe-Fe butterfly isomer, while the Fe-Fe bond is only partially weakened. The S atom is progressively reduced with a total increasing of its negative charge by 0.211 electron, leading to the Fe-Fe butterfly intermediate suitable for oxidative addition. In light of the results obtained, a mechanism of the p

The active site of the [FeFe] hydrogenases contains two Fe ions bound to one Cys ligand, three CO molecules, two CN- ions, and a dithiolate ligand. The nature of the last of these has been much discussed, and it has been suggested that it contains C, N, or O as the bridgehead atom. Most experimental studies indicate a N atom, whereas a recent density functional theory (DFT) study of a crystal structure indicated an O atom. Here, we performed quantum refinement on the same crystal structure with five different models of the dithiolate ligand X(CH2S-)2, with X = CH 2, NH2+, NH (two conformations), or O; we found that structures with a N bridgehead atom actually provide the best fit to the raw crystallographic data. Quantum refinement is standard crystallographic refinement in which the molecular mechanics force field normally used to supplement the experimental raw data to give a more chemical structure is replaced by more accurate DFT calculations for the active site. Thereby, we obtain structures that are an ideal compromise between DFT and crystallography.

In this work a density functional theory study of the CO addition reaction to FeIFeI and FeIFeII models of the active site of [FeFe] hydrogenases is presented. A series of model complexes, ranging from simple diiron model complexes of the binuclear [2Fe]H subcluster to the full H-cluster, have been investigated. For each system, the thermodynamic parameters for the CO adduct formation, a reaction that mimics the enzyme CO inhibition, were computed. Parallel to the investigation of the CO addition reaction, the structural features of the various FeIFeI and FeIFeII species have been evaluated, with particular attention to the issue of the ligand arrangement as a function of the redox state. CO affinity depends on the redox state of the model and the chemical nature of its ligands. FeIFeII species are more favorable to form the CO adducts than the reduced FeIFeI species. According to the computed free energies and enthalpies for the CO adduct formation from Fe2(pdt)(CO)5L models, the CO affinity follows the ligand sequence L = SCH3- > CN- > PPh3 > CO (FeIFeI) and L = CO > CN- > PPh3 > SCH3- (FeIFeII). As the models become more similar to the H-cluster, the CO affinity increases, although the FeIFeI CO -inhibited H-cluster is not stable. The bonding properties of the models considered have been investigated by means of the quantum theory of atoms in molecules approach. Upon CO addiction, the new Fe-C bond is formed to the detriment of the Fe-Fe bonds and, to a lesser extent, the Fe-S bonds. Regarding the FeIFeII systems investigated, the spin density is initially localized on the rotated Fe atom, and the formation of the CO adducts results in a delocalization of the spin density. Consequently, the FeIFeII CO-inhibited forms are better described as (Fe+1.5)2.

This study probes the impact of electronic asymmetry of diiron(I) dithiolato carbonyls. Treatment of Fe-2(S2Cn,H-2n)(CO)(6-x)(PMe3)(x). compounds (n = 2, 3; x = 1, 2, 3) with NOBF4 gave the derivatives [Fe-2(S2CnH2n)(CO)(5-x)(PMe3)(x)(NO)]BF4, which are electronically unsymmetrical because of the presence of a single NO+ ligand. Whereas the monophosphine derivative is largely undistorted, the bis(PMe3) derivatives are distorted such that the CO ligand on the Fe(CO)(PMe3)(NO)(+) subunit is semibridging. Two isomers of [Fe-2(S2C3H6)(CO)(3)(PMe3)(2)(NO)]BF4 were characterized spectroscopically and crystallographically. Each isomer features electron-rich Fe(CO)(2)PMe3 and electrophilic Fe(CO)(PMe3)(NO)(+) subunits. These species are in equilibrium with an unobserved isomer that reversibly binds CO (Delta H = -35 kJ/mol, Delta S = -139 J mol(-1) K-1) to give the symmetrical adduct [Fe-2(S2C3H6)(mu-NO)(CO)(4)(PMe3)(2)]BF4. In contrast to Fe2(S2C3H6)(CO)4(PMe3)2, the bis(PMe3) nitrosyl complexes readily undergo CO substitution to give the (PMe3)(3) derivatives. The nitrosyl complexes reduce at potentials that are similar to 1 V milder than their carbonyl counterparts. Results of density functional theory calculations, specifically natural bond orbital analysis, reinforce the electronic resemblance of the nitrosyl complexes to the corresponding mixed-valence diiron complexes. Unlike other diiron dithiolato carbonyls, these species undergo reversible reductions at mild potentials. The results show that the novel structural and chemical features associated with mixed-valence diiron dithiolates (the so-called H-ox models) can be replicated in the absence of mixed-valency by the introduction of electronic asymmetry.

Condensation of Fe2(SH)2(CO)6, acetaldehyde, and (NH4)2CO3 affords the methyl-substituted azadithiolate Fe2[(SCHMe)2NH](CO) 6 (1). The complex exists mainly ∼95%) as the meso diastereomer, but the d,l diastereoisomers could be detected. DFT calculations predict that the meso isomer would be 2.5 kcal/mol more stable than the d,l isomer due to conventional nonbonding interactions between the methyl groups and the ring hydrogen atoms. Crystallographic analysis of meso-1 confirms that the two methyl groups are equatorial, constraining the diferraazadithiolate bicycle to a conformation that desymmetrizes the diiron center. The lowered symmetry is confirmed by the observation of two 13C NMR signals in the FeCO region under conditions of fast turnstile rotation at the Fe(CO)3 groups. The pKa value of the amine in 1 is 7.89 (all pKa's determined in MeCN solution), which is similar to a redetermined value for Fe2[(SCH2)2NH](CO)6 (2, pK a = 7.98) and only slightly less basic than the tertiary amine Fe2[(SCH2)2NMe](CO)6 (pKa = 8.14). Substitution of 1 with PMe3 proceeded via the intermediacy of two isomers of Fe2[(SCHMe)2NH](CO)5(PMe 3), affording Fe2[(SCHMe)2NH](CO) 4(PMe3)2 (3). 31P NMR spectra confirm that the two PMe3 ligands in 3 are nonequivalent, consistent with the desymmetrizing effect of the dithiolate. The pKa value of the amine in 3 was found to be 11.3. Using triphenylphosphine, we prepared Fe2[SCHMe)2NH](CO)5(PPh3) as a single regioisomer. © 2008 American Chemical Society.