Electronic isomerism in a heterometallic nickel–iron–sulfur cluster models substrate binding and cyanide inhibition of carbon monoxide dehydrogenase

The nickel–iron carbon monoxide dehydrogenase (CODH) enzyme uses a heterometallic nickel–iron–sulfur ([NiFe4S4]) cluster to catalyze the reversible interconversion of carbon dioxide (CO2) and carbon monoxide (CO). These reactions are essential for maintaining the global carbon cycle and offer a route towards sustainable greenhouse gas conversion but have not been successfully replicated in synthetic models, in part due to a poor understanding of the natural system. Though the general protein architecture of CODH is known, the electronic structure of the active site is not well-understood, and the mechanism of catalysis remains unresolved. To better understand the CODH enzyme, we have developed a protein-based model containing a heterometallic [NiFe3S4] cluster in the Pyrococcus furiosus (Pf) ferredoxin (Fd). This model binds small molecules such as carbon monoxide and cyanide, analogous to CODH. Multiple redox- and ligand-bound states of [NiFe3S4] Fd (NiFd) have been investigated using a suite of spectroscopic techniques, including resonance Raman, Ni and Fe K-edge X-ray absorption spectroscopy, and electron paramagnetic resonance, to resolve charge and spin delocalization across the cluster, site-specific electron density, and ligand activation. The facile movement of charge through the cluster highlights the fluidity of electron density within iron–sulfur clusters and suggests an electronic basis by which CN− inhibits the native system while the CO-bound state continues to elude isolation in CODH. The detailed characterization of isolable states that are accessible in our CODH model system provides valuable insight into unresolved enzymatic intermediates and offers design principles towards developing functional mimics of CODH.


Introduction
8][9] The carbon monoxide dehydrogenase (CODH) enzyme uses a cuboidal [NiFe 4 S 4 ] cluster, known as the C-cluster, 10 to catalyze the reversible reduction of carbon dioxide (CO 2 ) to carbon monoxide (CO) at ambient temperatures and pressures. 11The enzyme performs this difficult interconversion with high turnover rates, low overpotential, and perfect selectivity, 7,[12][13][14] a set of characteristics that has yet to be replicated in any synthetic catalyst. 15,16Thus, understanding the enzymatic mechanism and elucidating key intermediates in the catalytic cycle has become an active area of research, with the overarching goals of applying these principles to the design of future synthetic catalysts for global carbon cycling and conversion.
However, CODH is a large and complex homodimeric protein with two additional [Fe 4 S 4 ] clusters per subunit required for electron transfer, complicating attempts to characterize intermediates within the catalytic cycle. 17,18At this point in time, few CODH states can be considered well-understood (Fig. 1).The C red1 , C red2, and "C red2-CO 2 " states have been isolated and characterized spectroscopically and crystallographically, 10,19 while the C red1 and C red2 states have been further analyzed using electrochemical techniques. 13,20However, the electronic structure of the C red2 state is still unknown, as two isoelectronic species have been proposed (Fig. 1). 7,21An EPR study indicating the presence of an EPR-silent state (C int ) between C red1 and C red2 has further convoluted the proposed mechanism, indicating two potential pathways to the "C red2-CO 2 " state. 22Moreover, despite its importance, no CO-bound state of CODH has ever been isolated or observed.While an early structure of COexposed Methanosarcina barkeri CODH found coordination of a diatomic ligand at the nickel site, the identity of this ligand has since been attributed to a bound formyl species. 23Considering the high reactivity of CODH towards CO oxidation, 12,20 understanding this intermediate will provide valuable insight into the electronic basis for catalysis.
One strategy to elucidate structure-function relationships of complex metalloenzymes includes the development and study of model systems that replicate key structural motifs. 246][27][28] While modelling the heterometallic iron-sulfur cubane of CODH has been achieved in select systems, [29][30][31][32][33] and some of these model compounds also bind cyanide (CN − ), a known inhibitor of CODH, 29 none of the aforementioned structural models have demonstrated the propensity to bind CO 2 or CO. 7,34 Furthermore, these synthetic models are generated in aprotic organic solvents and typically utilize p-accepting ligands such as phosphines, which have signicantly different bonding characteristics than the p-donating cysteine thiolate ligands of the native system.6][37][38][39] Inspired by other model systems, we sought to develop a protein-based model of the C-cluster within a ferredoxin scaffold.By reproducing the iron-sulfur-cluster moeity, we hypothesized that we could access reactivity analogous to CODH and use spectroscopy to identify key electronic motifs promoting reactivity.
Our initial work has shown that incorporation of Ni II into the site-differentiated iron-sulfur cluster of the Pyrococcus furiosus (Pf) ferredoxin (Fd) produced a stable, heterometallic [NiFe 3 S 4 ] 2+ cluster in Fd (NiFd ox ) capable of rapid, reversible electron transfer. 40,41The reduced [NiFe 3 S 4 ] + Fd (NiFd red ) displayed the ability to bind both CN − and CO, as evidenced by EPR spectroscopy.The NiFd-CO species serves as the rst model for the CO-bound state of CODH.Herein, we use an array of complementary spectroscopic techniques to resolve the electronic structures of the NiFd red , NiFd ox , NiFd-CN, and NiFd-CO species, including variable-temperature electron paramagnetic resonance (EPR) spectroscopy, resonance Raman (rR) spectroscopy, X-ray absorption spectroscopy (XAS), and computational modelling.This study reveals distinct changes in the system upon reduction and binding of CO and CN − , suggesting a rearrangement of electron density within the cuboidal heterometallic iron-sulfur cluster depending on the identity of the ligand.The relationship between the NiFd-CO species and the "untrappable" C red1-CO state in CODH is discussed, along with the role of electronic isomerization in driving inhibition of CODH activity by CN − .As the C red1-CO state has yet to be observed in the native system, the in-depth characterization of the NiFd-CO species presented here provides valuable insight into the spin state, geometry, and complex electronic structure of this elusive C red1-CO intermediate.Given the high catalytic rates and full reversibility of native CODH towards CO oxidation and CO 2 reduction, 12,20 elucidating the structure of this essential species will facilitate the development of functional models that reproduce this reactivity, with long-term implications in environmental remediation and energy storage.

Isotopic labelling indicates spin is delocalized across Ni and Fe
Reduction of the EPR-silent nickel-incorporated NiFd ox species to the NiFd red state gives a characteristic X-band EPR spectrum spanning ∼300 mT, with sharp peaks at g app = 5.7 and g app = 4.3 as well as broad peaks around g app ∼ 2.7 and g app ∼ 1.9 that are consistent with the formation of a single S = 3/2 species.This species had previously been simulated with spin Hamiltonian parameters of g iso = 2.0 and jE/Dj = 0.16 (Fig. 2A).Introduction of CN − results in quantitative conversion to a new S = 3/2 species, with peaks at g app = 4.35, 3.9, 1.92 that had been simulated with spin Hamiltonian parameters of g iso = 2.0 and jE/Dj = 0.07 (Fig. 2B).A more complicated spectrum is obtained upon exposure to CO, as conversion to the new species is not quantitative.We attribute the residual NiFd red to the limited solubility of CO in aqueous solutions and weak binding affinity of NiFd red towards CO.However, the spectra are dominated by two new apparent sets of signals-one set starting around g app ∼ 4 that would be consistent with an S = 3/2 species, and one set around g app ∼ 2 that would be consistent with an S = 1/2 species (Fig. 2C).These signals were assumed to derive from a spincoupling scheme within the cluster that involved all 4 metal centers with local high spin, analogous to conventional iron sulfur clusters.This intracluster coupling was implied to change upon ligand binding, giving rise to distinct spectra.
To test this hypothesis and obtain element-specic information on the spin projection across each center in the EPRactive NiFd red , NiFd-CN, and NiFd-CO states, we incorporated 61 Ni (I = 3/2) and 57 Fe (I = 1/2) into the cluster.As previously observed for homometallic iron-sulfur clusters and other heterometallic clusters, 57 Fe labeling resulted in global broadening of all species, consistent with a signicant degree of spin delocalized across the Fe centers (Fig. S1 †). 42,43The nickel contributions were more distinct across the NiFd species.In all cases, the broad linewidth precluded observation of well-dened hyperne peaks from the I = 3/2 nucleus, which is typical of high-spin systems. 40,44A modest, 0.4 mT line broadening was observed for the g app = 5.7 feature of NiFd red upon 61 Ni incorporation (Fig. 2A), in good agreement with previously reported results. 40Similarly, upon isotopic labelling, the NiFd-CN and NiFd-CO exhibit 0.7 and 0.4 mT line broadening at the g app = 4.35 and 3.65 features, respectively (Fig. 2B and C).Notably, there is no substantial broadening (<0.1 mT) of the feature at g app = 2.05 for the NiFd-CO species (Fig. S2 †).The measurable but small line broadening observed for all species indicates that spin density is distributed across the iron and nickel centers in the cluster, rather than being localized on a single ion, and highlights the integral role of nickel in the overall spin-coupling scheme of the cluster.Work is ongoing to obtain increased resolution of the electronic hyperne coupling to 57 Fe and 61 Ni nuclei using variable-eld Mössbauer and pulsed, high-frequency EPR techniques but is beyond the scope of this work.

Variable temperature CW-EPR suggests the presence of a single NiFd-CO species
To further evaluate the origin of the distinct EPR signals at g app ∼ 4 and g app ∼ 2 in NiFd-CO, variable temperature (VT) CW-EPR spectra were measured and compared to samples of NiFd-CN (Fig. 3).The VT-EPR lineshapes of NiFd-CN do not change signicantly from 5.5 K to 40 K, suggestive of a single S = 3/2 species (Fig. 3A).On the other hand, the VT-EPR lineshapes of NiFd-CO display unusual behavior.The feature centered at g app = 2.05 broadens substantially and decreases in intensity from 5.5 to 15.0 K, counter to what is typically observed in conventional mononuclear or spin-coupled S = 1/2 systems (Fig. S3-S5 †).By 20 K, the signal centered at g app = 2.05 is completely gone, while the low-eld signal at g app z 4 can still be observed.Comparison of the individual signal intensities to the total integrated intensity across the temperature range suggests the g app ∼ 2 species converts into the g app ∼ 4 species at higher temperatures, rather than arising from independent signals (Fig. S6 and S7 †).
The power saturation behavior of the NiFd red , NiFd-CN, and NiFd-CO was assessed from 5.5 to 15.0 K to estimate the energetic spacing of excited states.The P 1/2 values of the NiFd-CO features were consistently larger than those found for the other NiFd species, suggestive of faster relaxation rates.Both the NiFd red and NiFd-CN thus appear to be well-isolated S = 3/2 spin systems (Fig. S8-S14 †), while the NiFd-CO shows much faster relaxation.Collectively, the temperature-dependent lineshapes, Curie-corrected intensities, and relaxation properties suggest the NiFd-CO species possesses an S = 1/2 ground spin state with a low-lying S = 3/2 excited state. 45,46n CO of NiFd-CO suggests a large degree of ligand activation In prior work, Fourier transform infrared (FTIR) spectroscopy was used to probe the n CN mode frequency, which was observed at 2050 cm −1 and suggested a strong degree of s donation from the CN − ligand and only weak p back-bonding from nickel. 41owever, we were unable to observe an IR band for the NiFd-CO sample from 1800 to 2200 cm −1 that would be indicative of a bound CO ligand (Fig. S15 †).The lack of signal was attributed to the low binding affinity of CO to NiFd red and limited CO solubility in aqueous solutions, which prevents measurements at high protein concentrations.Because we were unable to acquire vibrational information using FTIR spectroscopy, resonance Raman spectroscopy was employed.
In addition to revealing the n CO mode frequency, this technique provides information on structure by resolving the lowfrequency cluster vibrational modes.The resonance Raman spectrum of reduced [Fe 3 S 4 ] 0 Fd exhibits a strong vibrational band centered at 352 cm −1 , which is attributed to the symmetric Fe-S bridging stretching modes (Fig. 4).Weaker vibrational bands can be seen arising from the Fe-S terminal modes. 47Upon incorporation of the Ni center into the cluster, the dominant vibrational band shis to 333 cm −1 .This bathochromic frequency shi is consistent with incorporation of a fourth metal center into the cluster. 47Additional weak bands can be observed for all other NiFd species (Fig. 4).The Raman spectrum does not change signicantly upon CN − binding, with the major vibrational band from the M-S bridging mode remaining at 333 cm −1 .In contrast, CO binding to the NiFd red cluster signicantly changes the low frequency region of the Raman spectrum.The dominant vibrational band representing the M-S bridging mode shis to 342 cm −1 .The bands at 365 and 386 cm −1 shi slightly to lower energy in the presence of 13 C-labeled CO, consistent with CO displacement coupling into these modes.As the shis are substantially lower than those estimated for a local Ni-C or Fe-C oscillator, the CO motion in those bands must be coupled into other vibrational modes.Importantly, an additional isotopically sensitive band at 1964 cm −1 is present in the NiFd-CO species, which shis to 1921 cm −1 for NiFd-13 CO (Fig. 4 and S16 †).This shi is nearly exactly as calculated for a local C-O oscillator, suggesting it corresponds to the n CO mode of the NiFd-CO species.This vibrational frequency is signicantly lower than that of free CO (2170 cm −1 ), indicative of signicant p back-donation and activation of the CO bond.
Near edge X-ray absorption spectra of NiFd species highlight electronic structure changes localized at the Ni center In order to obtain element-specic information on the electronic and geometric structure of the different NiFd species, both Ni and Fe K-edge X-ray absorption spectroscopies were used (Fig. 5).The edge transition energy of NiFd ox occurs at 8343.8 eV, accompanied by a low-intensity, pre-edge feature located at 8333 eV.9][50][51] Upon reduction of the cluster to the NiFd red state, the edge position shis by −4.2 eV to 8339.6 eV, consistent with an increase of electron density at the nickel center.We note that making formal and physical oxidation state assignments for Ni ions using edge positions alone is Fig. 4 Resonance Raman spectra of [Fe 3 S 4 ] 0 Fd (gray), NiFd red (green), NiFd-CN (orange), and NiFd-CO (blue).Samples were collected at 77 K using an excitation wavelength of 407 nm, P = 8 mW.Residual features corresponding to buffer are indicated with an *.Bands arising from buffer, DT, and quartz were subtracted after collection.(Inset) High frequency region of the resonance Raman spectra of NiFd-CO prepared with natural abundance CO (dark blue) and 13 CO (light blue), shown offset from the difference spectrum (gray).The band at 1906 cm −1 is present in both samples and independent of the CO isotope.challenging, as the edge position is not only dependent on the physical oxidation state but is also impacted strongly by metalligand covalency and geometry. 48,51,52The pre-edge transition of NiFd red occurs at a very similar energy and intensity as that of the oxidized cluster, suggesting the Ni center remains in a tetrahedral (or distorted tetrahedral) geometry.Interestingly, the XAS spectrum of NiFd-CN shows signicant changes when compared to NiFd red .The edge position shis higher in energy to 8342.2 eV, appearing at a similar energy as NiFd ox .More importantly, the cyanide-bound cluster exhibits an intense preedge feature at 8336 eV and a very low intensity pre-edge feature at 8332.4 eV, suggesting a change in the geometric structure.Previous studies on model nickel compounds have also observed an intense pre-edge feature around 8336 eV, which is suggested to derive from a 1s-4p transition induced by substantial mixing of the d and p orbitals in a low-spin, square planar geometry. 48,49On the other hand, CO binding to NiFd red shis the edge energy only slightly, to 8339.8 eV, with the appearance of a similarly intense pre-edge feature at 8334.8 eV.The second pre-edge feature observed in the spectrum at 8333 eV belongs to residual NiFd red that is not bound to CO.The relative edge shi between NiFd-CO and NiFd red suggests only a modest decrease in electron density at the Ni center upon CO binding.

Ni K-edge EXAFS provides insight into structural changes at the Ni site of NiFd
In addition to information on electron density and local geometry, the Ni K-edge XAS spectra provide solution-phase structural information through analysis of the EXAFS region (Fig. 6).The best t to the data for the NiFd ox was obtained with a primary shell consisting of one N/O atom, three S atoms, and three Fe atoms at distances of 2.01, 2.22, and 2.65 Å, respectively (Table 1).The t for the NiFd red is similar to that of the oxidized cluster, with contributions from one N/O, three S, and three Fe atoms at distances of 1.95, 2.27, and 2.65 Å, respectively.The distance along the S scattering pathways lengthens slightly, consistent with reduction of the Ni center and in line with previously observed trends for reduction of other [Fe 4 S 4 ] clusters using Fe K-edge EXAFS. 53In both the reduced and oxidized states of NiFd, the EXAFS data suggest the nickel center retains coordination to three bridging sulde ligands and a single water/aspartate residue (Fig. S20, S21 and Tables S1, S2 †).
The NiFd-CN species showed signicant changes in the XANES region, suggestive of substantial distortion towards a square planar structure.With this information, the preferred best-t model of the experimental EXAFS data includes one O/ N, one N/C, two S, and three Fe scattering pathways at distances of 1.87, 1.88, 2.29, and 2.73 Å, respectively, implicating loss of a bridging sulde ligand.A signicantly more intense peak at R + D = 2.3 Å in the R-space data is also observed.This feature is attributed to contributions from both a Ni-Fe single scattering pathway as well as the Ni-CN multiple scattering pathways (Fig. S22 and Table S3 †).The EXAFS trace of NiFd-CO is similar to that of the NiFd-CN, including the intense peak in the R-space data at R + D = 2.3 Å.Including the parameters for residual NiFd red in a two-component t to the EXAFS k-and R-space traces, the best t to the data includes one C/N, three S, and three Fe pathways at distances of 1.80, 2.29, and 2.73 Å, respectively, suggesting that the three sulde bridges remain intact with loss of the carboxylate ligand (Fig. S23 and Table S4 †).4][55] The Fe edge positions of the four isolated forms of NiFd also lie within a narrow range, with values at ∼7119 eV for the NiFd ox , NiFd red , NiFd-CN, and NiFd-CO, respectively (Fig. 5).While the differences are small (<1 eV), the edge position does follow a general trend across the series, where NiFd-CN < NiFd-CO z NiFd red < NiFd ox .Similarly, the pre-edge feature at 7113.2 eV does not change signicantly when the cluster is reduced or bound to the small ligands.The insensitivity of the Fe K-edge XANES features across all species suggests that the electronic structure perturbations introduced by oxidation state changes and binding of small molecules are predominantly localized to the Ni center.
DFT geometry optimization suggests substrate binding can occur with minimal perturbation to protein structure At present, there are no published X-ray crystal structures of the WT Pf Fd in the [MFe 3 S 4 ] state, and we were not able to obtain crystals of NiFd in any oxidation state. 56,57To gain some information on the interaction between the metallocofactor and protein environment, density functional theory calculations were used.A computational model was constructed that included the ligating residues and secondary coordination sphere (Fig. 7).Geometry optimizations were performed for the cluster assuming local and global high spin congurations, which overestimates bond lengths but allows us to qualitatively consider general trends across redox state changes and upon CO binding (Tables S7 and S8 †). 58This method was not deemed appropriate for calculations on the NiFd-CN state because of the XAS evidence in favor of a low-spin Ni II center (vide supra).Minimal structural perturbations were observed upon cluster reduction from the NiFd ox to the NiFd red state, with all metal centers retaining a tetrahedral geometry (Fig. S25-27 and Table S7 †).Two conformations of the Asp14 ligand could be optimized for the CO-bound state, one in which the aspartate remains ligated and a Ni-sulde bond breaks, and another in which the aspartate rotates away from the cluster (Fig. S28-S30 and Table S8 †).The EXAFS data and analyses led us to favor the latter geometry, in which the aspartate coordination to the cluster is broken (Fig. S29 and S30 †), which preserves tetrahedral geometry at the nickel center and gives a near-linear Ni-C-O bond of 170°.No structures converged in which CO was bound to any of the Fe centers.Preliminary validation of the optimized geometries was supported through comparison of the calculated and experimental Ni K-edge XAS pre-edge energies and intensities (Fig. 7).The calculated pre-edge transitions of NiFd ox and NiFd red exhibit similar pre-edge energies and intensities while the NiFd-CO pre-edge transition is shied to higher energy.In all cases, the dominant orbitals contributing to the transitions are delocalized across the Ni and the three Fe centers.The general agreement between the calculated and experimental data supports the premise that CO binding and redox state changes can occur without inducing large structural perturbations to the protein scaffold.

Discussion
A high density of states is present in the NiFd-CO species Given the importance of the C red1-CO state in CODH, we sought to obtain detailed information on the electronic and geometric structure of the CO-bound state(s) of NiFd.Through a holistic analysis of the results from several complementary spectroscopic studies, as reported above, we propose that the NiFd-CO species possesses a ground spin state of S = 1/2 with a low-lying, S = 3/2 excited state within a cuboidal [NiFe 3 S 4 ] + -CO cluster.This hypothesis derives from several observations: (1) the temperature-dependent EPR spectra reveal that the g app = 3.65 feature relaxes more slowly than the g app = 2.05 feature (Fig. 3, S8 and S9 †), in contrast to the expected behavior for a standard S = 1/2 species; 59 even the spin-coupled S = 1/2 [Fe 4 S 4 ] + Pf Fd species relaxes more slowly than the high-spin component (Fig. S3 †). 45Moreover, the integrated intensities of the features belonging to the NiFd-CO display a linear relationship as a function of 1/T between the temperatures of 6.5-30 K (Fig. S7 †), indicating that these two features belong to the same species, with the S = 1/2 feature interconverting to the S = 3/2 feature at higher temperatures.(2) The small but demonstrable EPR line broadening observed for the 61 Ni and 57 Fe labelling suggest both Ni and Fe contribute to the overall spin of the cluster, supporting an integral role of Ni in the cluster and spin scheme.(3) The high-frequency region of the resonance Raman spectrum of NiFd-CO displays a single isotopically sensitive band at 1964 cm −1 , consistent with the presence of a single species.The low-frequency rR bands are consistent with a cubane-like system and show small shis upon 13 CO incorporation, providing further evidence for a CO-bound nickeliron-sulfur cluster.Considering the XANES, EXAFS, resonance Raman, and EPR data together, we propose a cuboidal structure for NiFd-CO, with tetrahedral coordination at all metals and CO bound to the nickel center.The presence of a low-lying excited state in NiFd-CO may facilitate multistate reactivity in this model system and, by extension, native CODH (vide infra).

Ligand binding triggers electron redistribution throughout the NiFd cluster: formal oxidation state assignments
The electronic structure of iron-sulfur clusters is of signicant interest for the study of biological electron transfer and cluster reactivity.The most common congurations of [Fe 4 S 4 ] clusters are generally considered well-understood: formally, the oxidized canonical [Fe 4 S 4 ] 2+ cluster is best described as two pairs of mixed-valent Fe 2.5+ centers that are coupled through a double exchange pathway. 60,61Similarly, the reduced cluster can be described as a single mixed valent Fe 2.5+ pair antiferromagnetically coupled to a pair of Fe 2+ centers.However, a growing body of work suggests that the formal and physical oxidation states of the iron centers are dictated by the overall bonding environment of the cluster. 32,62One such example in biological systems is found in the radical SAM intermediate, U, where formation of the organometallic species upon reductive SAM cleavage results in a state that resembles the high-potential iron-sulfur cluster proteins (HiPIPs), with an excess ferric site. 61,63,64The physical electronic structure of U is still under debate; however, models of this intermediate suggest that the unique Fe bound to the SAM fragment may adopt a local Fe III electronic conguration. 32Similar electronic rearrangement has been observed in synthetic systems, where recent work from the Suess group has shown that binding of exogenous ligands to one unique site of a model [Fe 4 S 4 ] cluster inuences the oxidation states of the other irons within the cluster. 32,62,65This isomerization is likely linked to the close electronic communication between the metal centers through double exchange pathways within the cluster.
Analogous to these prior examples in biological and synthetic [Fe 4 S 4 ] clusters, binding of exogenous ligands to NiFd impacts the electronic distribution within the [NiFe 3 S 4 ] cluster.The initial reduction event from NiFd ox to NiFd red appears to be dominantly localized on the nickel center with minimal change at the iron centers, as evidenced by the XAS spectra.The similarity in the pre-edge features of NiFd ox and NiFd red and comparison to well-dened synthetic compounds suggests a tetrahedral nickel center. 48,49,51,52While formal oxidation state assignments of metals provide only a rough connection to the physical oxidation states, particularly in systems with highly covalent bonding, the experimental evidence supports the assignment of a high-spin Ni II in the NiFd ox species and a Ni I center in the NiFd red species (Fig. 8), with a formally assigned [Fe 3 S 4 ] 0 cluster fragment for both states respectively.The highly activated CO ligand (n CO = 1964 cm −1 ) in NiFd-CO is consistent with signicant electron donation into the p* orbitals of the CO ligand and is in line with other known Ni I -CO complexes. 66,679][70][71] Considering n CO along with the XAS edge energy, which is similar to that of NiFd red , a Ni I -CO and [Fe 3 S 4 ] 0 conguration is suggested to be present in the NiFd-CO species.On the other hand, back-donation into the CN − p* orbitals is relatively small in NiFd-CN, as the n CN is observed at 2050 cm −1 , a small shi from free cyanide (n CN = 2080 cm −1 ). 41his vibrational frequency is consistent with other Ni II -CN complexes. 72The XANES spectra provide further evidence for differences in electronic structure at the nickel center.The edge position shis to higher energies relative to NiFd red or NiFd-CO and instead overlaps with that of NiFd ox , suggesting a signicant loss of electron density from the nickel center.Additionally, the intense pre-edge feature that is observed at 8336 eV derives from the 1s-4p transition and is only prevalent in low-spin, square planar Ni II species. 48,49,52Collectively, these observations suggest the presence of a low-spin, Ni II oxidation state in NiFd-CN, with a [Fe 3 S 4 ] − fragment (Fig. 8).A shi in physical oxidation state for NiFd-CN relative to NiFd red would suggest the electron density is redistributed back into the [Fe 3 S 4 ] fragment.Indeed, careful analysis of the Fe K-edge XAS spectrum does show a shi of the NiFd-CN edge position to lower energy by approximately 0.2 eV, reecting an average increase in electron density across the three iron centers (Fig. 5).Due to the highly delocalized nature of iron sulfur clusters, we would expect additional electron density to be further distributed across the bridging sulde and terminal ligands.These analyses provide direct evidence for electronic communication and uidity of electron density between the four metal centers, highlighting a potential mechanism for promoting multielectron chemistry within heterometallic clusters.

Implications for substrate and inhibitor binding in native CODH
The Ni K-edge XANES spectrum of NiFd red has remarkably similar features to previously observed spectra of reduced Rr

Chemical Science Edge Article
CODH and other synthetic [NiFe 3 S 4 ] clusters, including a weak pre-edge feature at 8333 eV. 49Oxidation of Rr CODH to the catalytically inactive C ox state does not cause signicant changes to the Ni K-edge XAS spectra, suggesting the redox state changes are not localized to the Ni center; instead, Mössbauer and Fe K-edge XAS spectroscopy have suggested the initial reduction event to activate CODH occurs at the pendant iron center, generating the resting state C red1 (Fig. 8). 17,49,73In the absence of an exogenous Fe ion, as in NiFd, the redox state changes are localized on the substrate-binding Ni center (Fig. 8), providing benchmarks for the limiting cases that might be observed throughout CODH catalysis (Fig. 5 and S24 †).More parallels are observed in comparing the EXAFS data of reduced Rr CODH and NiFd red , as both species have an overall coordination number of four and similar distances to the bridging sulde ligands.Like the spectra of Rr CODH, the NiFd red /NiFd ox do not exhibit an intense Ni-Fe scattering feature that is present in the synthetic [NiFe 3 S 4 ] cubane clusters.We have interpreted the latter observation to indicate that, in the absence of a smallmolecule ligand, the NiFd cubane is not sufficiently rigid to resolve a clear Ni-Fe scattering vector. 49The Ni-Fe scattering pathway becomes more pronounced in the NiFd-CO and NiFd-CN samples, suggesting that binding of CO or CN − rigidies the cluster (Fig. S22, S23 and Tables S3-S4 †).This may be the case for the [NiFe 3 S 4 ] subsite of CODH as well, allowing conformational exibility prior to substrate binding.
The EPR data show parallels to native CODH and other [NiFe 3 S 4 ] synthetic models.The EPR spectra of C red1 and the cyanide-inhibited state, C red1-CN , both have total ground spin states of S tot = 1/2 with g avg = 1.82 and 1.72, respectively. 17,74,75rom the Mössbauer and EPR data, this S tot = 1/2 ground state is suggested to arise from antiferromagnetic coupling between the S = 3/2 [NiFe 3 S 4 ] subsite and the S = 2 exogenous Fe subsite.The NiFd red state has a ground spin state of S = 3/2, which is directly analogous to that of the [NiFe 3 S 4 ] subsite of the C red1 state. 17,49Additionally, the electronic parameters of the NiFd red species are similar to those of synthetic cubane [NiFe 3 S 4 ] + models by Holm and coworkers, which also have ground spin states of S = 3/2. 34,49In the case of NiFd red , we postulate this spin state likely arises from the coupling of an S = 1/2 Ni I center to the [Fe 3 S 4 ] 0 fragment on the basis of the data presented here.Experiments using Mössbauer spectroscopy are currently underway to more precisely determine the local spin states and coupling scheme of Fe centers within the cluster.Additionally, like the native system, only small line-broadening is observed upon substitution of 61 Ni, 43 which may highlight the importance of electronic cooperativity and spin delocalization between the metal centers for reactivity in CODH.Binding of CN − to NiFd red preserves the S = 3/2 ground spin state but perturbs the EPR spectrum in a manner that is similar to the changes observed for synthetic cubane [NiFe 3 S 4 ] + models; the spin state of the cubane in native CODH is also suggested to remain S = 3/2 on the basis of coupling with the exogenous high-spin ferrous ion (Fig. 8).
The proposed geometry of the nickel center in NiFd-CN is also similar to that observed in the cyanide-bound structure of CODH II Ch ; however, the n CN (2110 cm −1 ) of the native enzyme is notably higher than that in the NiFd-CN state. 76We believe this discrepancy derives from the differences in hydrogen bonding environments for the two systems.In native CODH, proposed hydrogen bonding interactions with secondary sphere residues may promote greater s-donating and less p-accepting character from the CN − ligand, giving rise to higher n CN vibrational frequencies. 77hough CO and CN − are isoelectronic and oen considered interchangeable in the context of structure and bonding, it is evident that each of these ligands imparts unique behavior to the nickel-substituted iron-sulfur cluster in NiFd.It is thus worth considering the subtle distinctions in electronic structure between the two small ligands.Both CO and CN − bind to the Ni I center in the stable NiFd red state and to the nickel center in C red1 , 76,78 driving a natural comparison between the two systems across ligands.CO is a neutral ligand with strong p-accepting character, underpinning the tendency to bind to electron-rich metals, 79 while the anionic CN − ligand has greater s-donor ability and more capacity for hydrogen bonding. 80The carbon center in CN − is therefore more nucleophilic than in CO, which likely preferentially stabilizes the Ni II oxidation state and drives the redistribution of electron density across the cluster in NiFd. 62This intracluster isomerization may hint at the mechanism through which CN − inhibits turnover of CODH.When CN − binds to the nickel center, which is suggested to be the binding site from multiple structural studies, 76,78 the nucleophilicity of the carbon center may drive the nickel to adopt a formal divalent oxidation state, effectively trapping the system in a conguration that is more electron-decient at the nickel center (Fig. 8).This electron-decient state may stabilize the system in the presence of oxidants, as recent work suggests the CN-bound C-cluster is more resistant to O 2 damage than the resting state, but requires reductive activation to re-enter the catalytic cycle. 12,81On the other hand, the electron-rich Ni I center in NiFd red is stabilized upon CO binding.The large degree of activation of the substrate CO ligand suggests a similarly nucleophilic Ni I state may be present in the catalytically active C red1 and elusive C red1-CO states (Fig. 8).The temperaturedependent EPR data on NiFd-CO may also provide insight into the absence of a known C red1-CO signal.While the ground spin state of NiFd-CO is suggested to be S tot = 1/2, it is not wellisolated, as evidenced by the clear appearance of high-spin signals at cryogenic temperatures.We consider it possible that low-lying excited states of C red1-CO may be present in native CODH that could complicate EPR observation.Moreover, the presence of the exogenous Fe center in the C-cluster is expected to affect the electronic structure of the spin system.The absence of a nearby nucleophilic site in NiFd-CO likely minimizes reactivity, permitting facile observation of the CO-bound state.Access to excited states in NiFd may also indicate potential for multistate reactivity in CODH, though this discussion would require advanced computational analysis that is beyond the scope of this work. 45,82The distinct differences observed between the CN − and CO underscore the importance of understanding the interactions between enzymes and their native substrates.

Edge Article
Chemical Science

Conclusions
A nickel-substituted [NiFe 3 S 4 ] ferredoxin (NiFd) protein was investigated as a model of the cubane subsite of carbon monoxide dehydrogenase (CODH).The NiFd system was prepared in several oxidation and substrate-bound states to mimic key states of native CODH.Through the use of EPR, rR, K-edge XANES, and EXAFS spectroscopies, we have assigned the redox processes observed to be formally centered on nickel and characterized the NiFd-CO species as a Ni I -CO nickel center that has tetrahedral coordination.Cyanide binding induces redox isomerization, resulting in a low-spin, square planar nickel center and increased electron density spread across the iron atoms in the cluster.This is in line with phenomena previously observed within synthetic iron-sulfur clusters, providing evidence for the ability of heterometallic cubane clusters to shuffle electron density around within the cluster.Furthermore, extrapolating the results found using this proteinbased model, we have postulated that similar phenomena may occur in native CODH, thus providing insight into the mechanism of inhibition by CN − in the native system.Ultimately, the results found from this study highlight the utility of developing protein-based models of enzymes.

Fig. 1
Fig. 1 Proposed catalytic cycle for CO oxidation and CO 2 reduction at the C-cluster of CODH showing two different hypothesized structures for C red2 .Amino acid numbering is from CODH II Ch .Structures in gray indicate intermediate states that are not well-characterized.

Fig. 5
Fig. 5 (A) Ni K-edge XANES of the four isolated forms of NiFd.(Inset) Zoom in on the pre-edge region.(B) Fe K-edge XANES of the four isolated forms of NiFd.(Inset) Zoom in on the pre-edge region.(C) Derivative of the Ni K-edge XANES from the traces in A. (D) Derivative of the Fe K-edge XANES from the traces in B.
Fe K-edge XANES suggest minor changes at Fe across cluster oxidation states The changes in the edge and pre-edge positions of the different NiFd species are signicantly less pronounced in the Fe K-edge XAS spectra.The Fe K-edge XANES of the [Fe 3 S 4 ] Fd and the [Fe 4 S 4 ] Fd species were measured for comparison (Fig. S24 †) and observed to follow the trends previously reported for other

Fig. 6
Fig. 6 Ni K-edge EXAFS of the four isolated forms of NiFd.(A) Comparison of experimental Fourier transform (FT) EXAFS data (solid) for the four different forms of NiFd overlaid with the best fit (gray).(B) Comparison of experimental k 3 EXAFS data (solid) for the four different forms of NiFd overlaid with the best fit (gray).

Fig. 7
Fig. 7 Experimental pre-edge Ni K-edge XANES spectra of NiFd red (green), NiFd ox (black), and NiFd-CO (blue) with calculated TD-DFT contributions using the DFT geometry-optimized structures of the NiFd species (inset).Dominant contributing orbital to the indicated transition for each species shown at an isosurface value of 0.03 with distribution over Ni and the Fe atoms indicated.

Fig. 8
Fig. 8 Proposed electronic and geometric structures of the (A) NiFd ox , NiFd red , NiFd-CO, and NiFd-CN species with key experimental metrics indicated that have been obtained from this work, and (B) analogous C ox , C red1 , C red1-CO , and C red1-CN states of CODH with spectroscopic metrics obtained from ref. 17, 75 and 76.