Use of Metallopeptide Based Mimics Demonstrates That The Metalloprotein Nitrile Hydratase Requires Two Oxidized Cysteinates For Catalytic Activity

Abstract

Nitrile hydratases (NHases) are non-heme Fe^III or non-corrin Co^III containing metalloenzymes that possess an N₂S₃ ligand environment with nitrogen donors derived from amidates and sulfur donors derived from cysteinates. A closely related enzyme is thiocyanate hydrolase (SCNase), which possesses a nearly identical active-site coordination environment as CoN-Hase. These enzymes are redox inactive and perform hydrolytic reactions; SCNase hydrolyzes thiocyanate anions while NHase converts nitriles into amides. Herein an active CoNHase metallopeptide mimic, [Co^IIINHase-m1] (NHase-m1 = AcNH-CCDLP-CGVYD-PA-COOH), that contains Co^III in a similar N₂S₃ coordination environment as is found in CoNHase/SCNase. [Co^IIINHase-m1] was characterized by ESI-MS, GPC, Co K-edge X-ray absorption spectroscopy (Co-S: 2.21 Å; Co-N: 1.93 Å), vibrational, and optical spectroscopies. We find that [Co^IIINHase-m1] will perform the catalytic conversion of acrylonitrile into acrylamide with up to 58 turnovers observed after 18 hours at 25 °C (pH 8.0). FTIR data used in concert with calculated vibrational data (mPWPW91/aug-cc-TZVPP) demonstrates that the active form of [Co^IIINHase-m1] has a ligated SO₂ (ν = 1091 cm⁻¹) moiety and a ligated protonated SO(H) (ν = 928 cm⁻¹) moiety; when only one oxygenated cysteinate ligand (i.e. a mono-SO₂ coordination motif) or the bis-SO₂ coordination motif are found within [Co^IIINHase-m1] no catalytic activity is observed. Calculations of the thermodynamics of ligand exchange (B3LYP/aug-cc-TZVPP) suggest that the reason for this is that the SO₂/SO(H) equatorial ligand motif promotes both water dissociation from the Co^III-center and nitrile coordination to the Co^III-center. In contrast, the under- or overoxidized motifs will either strongly favor a five coordinate Co^III-center or strongly favor water binding to the Co^III-center over nitrile binding.

Introduction

Nitrile hydratases (NHases) are an industrially important class of redox-inactive non-heme Fe^III or non-corrin Co^III metalloenzymes that effect the conversion of nitriles into amides (Figure 1a).^1–5 A structurally related class of metalloenzymes are thiocyanate hydrolases (SCNases), which converts the SCN^– anion into NH₃ and SCO (Figure 1b).6–8 X-ray crystal structures are available for both the cobalt (CoNHase) and the iron forms (FeNHase) of NHase as well as SCNase, who’s active-site structure is nearly identical to that of CoNHase.^7–14 In all three metalloenzymes the M^III ion is ligated by three cysteinate sulfurs in a mer configuration and two amidate nitrogen ligands derived from the protein backbone. One of the cysteinate sulfur is trans to a site that has been speculated to be occupied by a water (or hydroxide) molecule or NO in the inactivated form, but could also be a vacant site in the absence of substrate.^9,10,12 An interesting structural aspect of the NHase and SCNase active site concerns the two cysteinate sulfurs that are contained within the metal-centers equitorial plane; cystallographic analysis has suggested that these are post-translationally oxygenated to the corresponding sulfenate/sulfenic acid (CysSO^–/CysSO-H) and a sulfinate (Cys–SO₂). To our knowledge these two classes of metalloenzymes represent the only metalloenzymes containing this oxidized cysteinate coordination motif.

Figure 1.

A) Reaction catalyzed by nitrile hydratases and B) reaction catalyzed by thiocyanate hydrolase. Note the similarity in purported active-site structures.

Several X-ray structures have been reported for both NHase and SCNase with varying degrees of equatorial cysteinate oxygenation.^7–14 From these studies it has been suggested that insertion of the M^III ion into the active-site and formation of the sulfinate proceeds nearly simultaneously. Sulfenate formation, at least in recombinant SCNase, appears to occurs at a later time.⁹ In addition to X-ray crystallographic studies, the formulation of the post-tranlationally modified cysteinates as one sulfenate and one sulfinate ligated to the M^III ion has also been supported by spectroscopic (sulfur K-edge X-ray absorption spectroscopy and FTIR spectroscopy)^15,16 and mass spectrometry studies.^17–20 We note that the sulfur K-edge X-ray absorption and FTIR spectroscopic studies have suggested that the sulfenate is likely protonated or contained in a strong hydrogen-bonding network.^15,16

Mass spectrometry studies have provided fairly convincing evidence that at least one of the cysteinates must be post-translationally modified to a CysSO₂ ligand for NHases to display activity; combined LC-MS/activity studies have directly correlated enzymatic activity with cysteinate oxygenation.¹⁷ Furthermore, when FeNHase is prepared under anaerobic conditions, where cysteinate oxidation would be unlikely, no enzymatic activity is observed.¹⁷ Upon exposure of anaerobically prepared FeNHase to air the enzyme becomes activated suggesting a cysteinate oxidation event may be promoting activity. In fact, most industrial processes that utilize NHase involve maintaining the enzyme under constant aerobic or oxidizing conditions in order to maintain optimal NHase activity.^21–24 There has also been one combined X-ray crystallography/activity study involving SC-Nase that has produced some evidence that the sulfinate/sulfenate modification is necessary for full enzymatic activity; modification of only one cysteinate to the sulfinate and the bis-sulfinate modification each produces lowered enzymatic activity.⁹ For NHase it is not so clear cut, with studies suggesting that over-oxidation can either lead to enzymatic deactiviation or produce a fully active metalloenzyme.^14,25 When 2-cyano-2-propyl hydroperoxide (CPx) is added to active FeNHase it will strongly inhibit the metalloenzyme. As CPx is capable of oxidizing sulfenates to sulfinates it was suggested that the lose of activity is likely due to the oxidation of the CysSO ligand to a CysSO₂ ligand.25 However, CPx is also capable of reacting with the Fe^III metal-center generating an Fe^II ion in the process. This metal-based reduction event would also lead to enzymatic deactivation. LC-MS studies also indirectly provides evidence of a sulfenate in the active form of NHase. All of the mass spectrometry studies that can identify the “Cys-SO” ligand in the active form of NHase rely on a trypsin digests. In these studies the Cys-SO is never observed by MS in the sulfenate form, but in the sulfinate form.^17–20 There are three possible explanations for this observation. One is that the inherently reactive Cys-SO moiety readily oxidizes to the corresponding Cys-SO₂ group under the experimental conditions used to detect it (i.e. trypsin digest followed by MS/MS analysis). Another possibility is that the cysteinate remains completely unmodified in the active form of NHase and that the oxidation of the unmodified equatorial Cys to the Cys-SO(H) or CysSO₂ form results in enzymatic inactivity. The oxidized cysteinate observed by ESI-MS may therefore be an artifact of the experiment. A third possibility is that the fully oxidized bis-CysSO₂ form of NHase is required for activity, and that the supposition that the underoxidized sulfenate is present at the active-site of the active form of NHase is incorrect. This third possibility is supported by limited structural/mechanistic considerations resulting from a recent crystallographic study of CoNHase.¹⁴

Much of the ambiguity concerning the requirement of one vs. two oxygenated cysteinates coordinated to the M^III ion at the NHase and SCNase active-sites and the exact extent of cysteinate oxygenation stems form the lack of studies that are capable of directly correlating enzymatic activity with the extent of cysteinate oxygenation. From a biochemical perspective this is exceptionally difficult due to the inherent lack of fidelity with concern to cysteinate-oxygenation. This is exasperated by the difficulty of directly monitoring cysteinate-oxygenation as a function of enzymatic activity owing to the fact that the available techniques for doing so offer either nearly insurmountable signal-to-noise/overlapping signal problems (e.g. protein FTIR), or the introduction of additional artifacts (e.g. trypsin digestions).

Biomimetic NHase complexes have proven useful in understanding many of the key structural components of the NHase active-site, and how these components contribute to the electronic structure and bonding properties of the metalloenzyme.^1,5,26–28 To date several NHase mimics have been prepared that contain oxygenated thiolate ligands.^{29–36,38–40,109} Most of these complexes contain bis-sulfinate ligands; only one model compound has been reported thus far that contains both a sulfenate and a sulfinate coordinate to the metal-ion (a CoNHase model reported by Kovacs and coworkers; Chart 1a).³² Although this compound revealed a number of interesting aspects of CoNHase chemistry, including the likely order of addition of oxygen atoms to the coordinated thiolate-sulfurs, the sulfenate oxygen of this compound is coordinated to the Co^III center, and the compound does not display NHase activity. Likewise, only one NHase model compound (a CoNHase model complex reported by Mascharak and coworkers; Chart 1b) has been reported thus far that displays any NHase activity.⁴¹ This model complex, although active (18 turnovers after 4 hours), contains no oxygenated thiolates, and is reactive at relatively high temperatures (50 °C) and pH (9.5). To better correlate activity with thiolate-oxidation one needs a compound that: a) can under go sequential thiolate oxygenation, and b) displays at least modest NHase activity.

Chart 1.

Chart 1

Reported herein is a metallopeptide CoNHase mimic that displays reasonable catalytic activity under optimal conditions (approximately 60 turnovers in 18 hours at 25 °C). Furthermore, through controlling the extent of cysteinate oxygenation, we can directly correlate the extent of oxygenation with catalytic activity. It will be demonstrated that catalytic activity requires oxygenation of two cysteinate, one to a Cys-SO₂ and the other to a protonated Cys-SO(H).

Experimental

Preparation of NHase-m1

The apo-peptide NHase-m1 (AcNH-CCDLP-CGVYD-PA-COOH) was prepared in a manner similar to that previously reported for similar peptides using solid-state peptide synthesis methods (Fmoc/tBu protection strategies) on Wang resin.^42–45 The peptide was cleaved from the resin using a 95:2.5:2.5 mixture of trifluoroacetic acid(TFA):triisopropylsilane: ethanedithiol. Following evaporation of the cleavage solution the resulting crude peptide was washed with freshly distilled diethyl ether and purified by reverse-phase HPLC (gradient: 10–29% MeCN (0.1% TFA) in H₂O (0.1% TFA) over 15 min) using a Waters DeltaPrep 60 equipped with a Waters X-Bridge C-18 column (30 × 150 mm; 5 μm). Fractions containing NHase-m1 were pooled and lyophilized yielding a white fluffy solid (26.3 mg; Yield = 20.3%). Final purity was assessed by analytical HPLC using a Waters X-Bridge C-18 column (4.6 × 150 mm; 5 μm) on a Waters DeltaPrep 60 and ESI-MS on a Waters MicroMass 20 ESI mass spectrometer (positive ion mode). (Analytical HPLC - gradient: 10–65% MeCN (0.1% TFA) in H₂O (0.1% TFA) over 60 min; Retention time: 20.7 min. ESI-MS (pos. ion mode) - (M+Na)⁺ m/z calcd 1319.4; found 1320.0).

Metallation and Maturation of NHase-m1 With Cobalt

NHase-m1 was dissolved in degassed 10 mM N-ethylmorpholine buffer (NEM; pH 8.0) under an inert atmosphere yielding a final concentration of approximately 1.0 mM as assessed by the absorbance of the charge transfer band associated with the Y(9) residue (λ_max= 278 nm; ε = 1,490 M⁻¹ cm⁻¹).⁴⁶ Using Ellmans’ methods it was verified that all of the cysteine residues were free and no further treatment of the peptide solution was performed.⁴⁷ The addition of one equivalent of CoCl₂ under strictly anaerobic conditions in a Coy chamber resulted in the immediate formation of a light colored green solution (λ_max= 736 nm; ε= 425 M⁻¹ cm⁻¹; λ_max= 687 nm; ε= 510 M⁻¹ cm⁻¹; λ_max= 611 nm; ε= 405 M⁻¹ cm⁻¹; and λ_max= 340 nm; ε= 2113 M⁻¹ cm⁻¹) indicating the formation of [Co^IINHase-m1]. Exposure to air resulted in the conversion of the green solution into a brown one within ten minutes. After 3 hours the solution became a deep brown, by which time the fully functional met-allopeptide [Co^IIINHase-m1] had formed. The solution was then purged with argon and could be stored indefinitely under an inert atmosphere. Subsequent gel-permiation chromatography (GPC) experiments were performed using a Waters Protein-Pak GPC column (7.8 × 300 mm; 60 Å pore size) using a NaHCO₃(aq.) mobile phase under a positive pressure of He. GPC calibrations were made using a Water polyethyleneglycol standards kit. (Mass by GPC: calc. 1398 g mol⁻¹ found 1395 g mol⁻¹; ESI-MS (negative ion mode) m/z calc. 1398.4 found 1398; λ_max= 453 nm (sh); ε= 1350 M⁻¹ cm⁻¹; λ_max= 332 nm (sh); ε= 3255 M⁻¹ cm⁻¹)

Physical Methods

Electronic absorption spectra were obtained in air-tight quartz cuvettes using either a Varian CARY 50 or a Perkin-Elmer Lambda 750 UV-vis-NIR spectrometer. Circular dichroism spectra were obtained on a Jasco J-715 CD spectropolarimeter in circular 1 cm quartz cuvettes and represent the average of five scans. Electronic absorption and CD spectra were simultaneously deconvoluted into the minimum number of gaussian line shapes that reproduced the two spectra using in-house written procedures for the data analysis and graphing program Igor Pro (Wavemetrics; Lake Oswego, OR). We used the criteria that all peak widths were within ±500 cm⁻¹ of $11\times \sqrt{\nu }$ to ensure a meaningful fit to the absorption spectrum.

FTIR spectra were obtained on a Thermo-Nicolet 470 FTIR spectrometer with either a SMART MIRacle Attenuated Total Reflectance (ATR) accessory (as dehydrated films formed from the addition of 10 μL of sample to the ZnSe crystal) or in a liquid sample holder between CaF₂ plates. Approximately 1 mM solutions of [Co^IIINHase-m1] in 10 mM NEM (pH 8.0) were used in all experiments. A background of a 1 mM solution of the apo-peptide NHase-m1 in 10 mM NEM buffer (pH 8.0) was used for all experiments. For times course experiments solutions of [Co^IINHase-m1] were exposed to air and then the FTIR spectra were obtained after specified time intervals. In all cases the spectra represent the average of 1000 scans at 1 cm⁻¹ resolution except for the spectra following 10 min. of air exposure, which represents the average of only 160 scans at 1 cm⁻¹ resolution. Data were then normalized to a positive-signed peak in the difference spectrum at 1261 cm⁻¹, which is invariant in relative intensity between spectra. ¹⁸O₂ labeling was performed by injecting 98% enriched ¹⁸O₂ gas (ICON Isotopes; Summit, NJ) via an air tight syringe into a solution of [Co^IINHase-m1]. The IR spectrum of the ¹⁸O₂ matured metallopeptide was then recorded after 3 hours of maturation (1000 scans at 1 cm⁻¹ resolution).

Cobalt K-edge X-ray absorption spectra were obtained at the National Synchrotron Light Source on beamline X3b. Samples of O₂ matured [Co^IIINHase-m1] in a 1:1 mixture of 10 mM NEM buffer (pH 8.0) and glycerol were injected between windows made from Kapton tape (3M; Minneapolis, MN, catalog no. 1205) and quickly frozen in liquid nitrogen. Data were collected at 20 K maintained by a He Displex cryostat and recorded as fluorescence spectra on a Canberra 13-element solid-state Ge detector. Total count rates were maintained under 25 kHz for all channels and a dead time correction was not applied. For edge spectra the primary hutch aperture height (PHAH) was set to 0.4 mm to obtain the maximum spectral resolution in the edge region, while for the EXAFS spectra the PHAH was set to 0.8 mm. The edge spectra the data were then collected in 5 eV steps in the pre-edge region (7609 – 7689 eV), 0.3 eV steps in the edge region (7689 – 7729 eV), and 2.0 eV steps in the near-edge region (7729 – 7909 eV). EXAFS spectra were recorded in 10 eV steps in the pre-edge region (7509 – 7689, 0.5 eV steps in the edge region (7689 – 7759 eV), and 2.0 eV steps in the near-edge region (7759 – 8009 eV), and 5 eV steps in the far-edge region (8009 – 15.5 k). The spectra represents the average of 5 scans. Prior to data averaging each detector channel of each spectrum was individually inspected. Data were analyzed using the software packages EXAFS123⁴⁸ and FEFF 8.20⁴⁹ as previously described.⁴³ All refinements are based on Fourier Filtered κ³(χ) data over the energy range of κ = 2.0 – 14.3 Å and back-transformed from r′ = 1.0 – 2.5 Å. A bond valence sum (BVS) analysis was then performed on the various statistically valid models obtained form the EXAFS data.⁵⁰ The BVS is the sum of the individual bond valences (s_i) defined by:

$${\text{s}}_{\text{i}}=exp\left[\frac{{\text{r}}_{\text{i}}-{\text{r}}_{\text{o}}}{0.37}\right]$$ (1)

$$\text{BVS}=\sum _i{\text{s}}_{\text{i}}$$ (2)

where r_i are the refined bond length from the EXAFS experiments and r_o are the reference bond lengths. Here we use the method described by Brown and Altermatt for determining the values of r_o for the various Co–L bonds, and use the following values: Co–S: 2.079 Å; Co–N: 1.759Å; Co–O: 1.687Å.⁵⁰

Nitrile Hydrolysis Studies

All aqueous solutions were prepared from water that had been passed through a three filter purifier to a final resistance of at least 18 MΩ. Nitrile hydrolysis was assessed by ¹³C-NMR spectroscopy and GC-mass spectrometry. In a typical experiment a 0.10 mM solution of [Co^IIINHase-m1] in 10 mM NEM buffer (pH 8.0) was added to a 10 mM nitrile solution in pure water (total volume 300 μL). The reaction mixture was then heated to 50 °C, except for acrylonitrile hydrolysis experiments where the solution was only heated to 25 °C to avoid unwanted polymerization of the nitrile substrate. Following 18 hours the reaction mixtures were analyzed. GC/MS analysis was performed by creating standard curves of nitrile and amide peak areas vs. the peak area of a known quantity of n-octane standard. For ¹³C-NMR analysis 300 μL of D₂O (with a DDS standard) was added, and the integrated peak areas of the quaternary carbon atoms were measured and compared to a standard curve.

Electronic Structure Calculations

All density functional theory (DFT) calculations were performed using the software package ORCA 2.6.35 written by F. N. Neese.⁵¹ Truncated computational models of the predicted [Co^IIINHase-m1] coordination sphere were constructed from a AcN-Cys-Cys-H fragment about a Co^III ion with an additional equatorial ethanethiolate ligand. A sixth ligand (water or acetonitrile) was then added to the coordination sphere about cobalt. Models containing successive thiolate oxidation were then prepared along with sulfenate protonation. This produced fifteen possible models, of which the twelve models discussed in this manuscript are outlined in Chart 2. All calculations utilized Ahlrichs’ TZVPP basis set^52–54 augmented by Dunning’s diffuse functions.^55–57 Initial geometry optimizations were performed using the local density approximation of Vosko, Wilk, and Perdew and the nonlocal gradient corrections of Beck and Perdew.^58–63 These were then further refined with the modified Perdew-Wang 91 exchange and Perdew-Wang 91 correlation functionals.^62,64,65 Bonding indices are described by using a Mayer bonding analysis.^66–68 All vibrational analyses utilized two sided displacements, the modified Perdew-Wang 91 exchange and Perdew-Wang 91 correlation functionals, and made use of the resolution of identity (RI) approximation^69–75 with the corresponding TZV/C auxiliary basis set.^54,76 Electronic spectra were calculated using Neese’s SORCI methodology on the five lowest energy spin-allowed transitions (CAS-SCF(8,8); selection threshold = 10⁻⁶ E_h; prediagonalization threshold = 10⁻⁶ E_h; natural orbital selection threshold = 10–5 E_h).77

Chart 2.

Chart 2. Computational models examined in this study: A) (Co-NHase-SO₂-L)^2- (2-L), B) (Co-NHase-SO₂/SO-L)^2- (3-L), C) (Co-NHase-SO₂/SOH-L)^1- (4-L), D) (Co-NHase-2SO₂-L)^2- (5-L). In all cases L = H₂O (top), vacant (middle, abbreviated with no ligand description), or MeCN (bottom).

Calculation of Free Energies of Ligand Displacement

The free energies of relevant stationary points were calculated using Gaussian 03⁷⁸ employing the above prescribed basis sets and Becke’s three-parameter hybrid functional for exchange along with the Lee-Yang-Parr correlation functional (B3LYP).^79–81 For these calculations a T = 298.15 K, a 1.0 M concentration, and a P = 1.0 atm were all utilized. All free energies are zero-point corrected. Energies were calculated from the thermodynamic scheme depicted in Scheme 1.⁸² The free energies of solvation (ΔG_n) were calculated using the self-consistent reaction field model with a dielectric constant of 80.37 and probe radius of 1.4 Å.⁸³

Scheme 1.

Sulfenate pK_a values were calculated according to the square-scheme displayed in Scheme 2.⁸² Here we used experimental values of −262.23 kcal mol⁻¹ for the free energy released upon transfering a proton from the gas-phase to water and −6.28 kcal mol⁻¹ for the free energy of a proton in the gas-phase.^84,85 The pK_a was then calculated according to equation (3).

Scheme 2.

$${\text{pK}}_{\text{a}}=\frac{1}{2.303RT}\mathrm{\Delta }\text{G}$$ (3)

Results and Discussion

In previous studies we have utilized the first twelve residues from the primary sequence nickel containing superoxide dismutase (NiSOD) from Streptomyces coelicolor to prepare active NiSOD biomimetic metallopeptides (apo-peptide: H₂N-HCDLP-CGVYD-PA-COOH).^42,43,45 This metallopeptide provides the nickel ion with an equatorial N₂S₂ coordination environment and an axial imidazole ligand ligand (from His(1)) to the oxidized Ni^III center (Scheme 3). It was reasoned that this peptide would make an excellent platform for the production of a NHase metallopeptide-based mimic following minor modifications to the apo-peptide. Thus, the His(1) was changed to a Cys-residue and the free N-termial amine was capped with an acetyl group yielding the peptide NHase-m1 (AcNH-CCDLP-CGVYD-PA-COOH).

Scheme 3.

Addition of one equivalent of CoCl₂ to solutions of NHase-m1 (10 mM NEM buffer, pH = 8.0) instantly forms a green solution. The UV-vis-NIR spectrum of the Co^II metallopeptide (Figure 2 and Supporting Information) in D₂O NEM buffer (10 mM, pD = 8.0) displays ligand field bands in the vis-NIR spectrum between 5,300 – 10,500 cm⁻¹ (1,890 – 952 nm) and 11,800 – 18,350 cm⁻¹ (850 – 545 nm) and more intense charge transfer bands in the UV-region at 29,325 and 36,495 cm⁻¹ (341 and 274 nm) (Figure 2). The higher-energy ligand-field bands display evidence of splitting due to spin-orbit coupling effects, consistent with tetrahedral Co^II. All of these visible absorption bands are CD active demonstrating the Co-ion is ligated to the apo-peptide forming “[Co^IINHase-m1]” (Supporting Information). Based on the overall broad-shapes of the ⁴T₁(F) ← ⁴A₂(F) and ⁴T₁(P) ← ⁴A₂(F) transitions (from T_d parentage) we can surmise that the Co^II center is contained in a mixture of ligand environments ranging from a Co^IIN₂S₂ to a Co^IIS₄ ligand environment.⁸⁶ This may be expected as the peptide NHase-m1 was not designed to hold a metal ion in a tetrahedral geometry, and thus the cobalt ion will be ligated in a non-distinct ligand environment by one or more NHase-m1 peptides. In contrast, oxidation of cobalt to the Co^III oxidation state yields one distinct ligand environment for cobalt within NHase-m1 (vide infra).

Figure 2.

Electronic absorption spectrum of Co^II ligated NHase-m1. The inset depicts an expansion of the higher-energy ligand-field transitions.

Formation of [Co^IIINHase-m1] and Subsequent Characterization

Exposure of solutions of [Co^IINHase-m1] to air results in the gradual formation of a brown solution (Supporting Information). This color change is consistent with the oxidation of the Co^II ion to a thiolate-ligated Co^III ion over the period of ten minutes.^{32,41,87–89} Despite the fact that the cobalt center is oxidized over a relatively short period of time, full maturation of the metallopeptide is only achieved after three hours of maturation under air exposure (vide infra). Following the three hour maturation time solutions of the oxidized metallopeptide are purged with argon and handled using air-free techniques.

Solutions of the [Co^IIINHase-m1] were first examined by gel-permeation chromatography (GPC) and ESI-MS. GPC data were recorded at 350 nm, where the free peptide does not absorb, and show a single peak at 7.24 min, which equates to a molecular mass of 1395 g mol⁻¹ (Supporting Information). Considering the up to 30% error associated with GPC this is consistent with both the 1343 g mol⁻¹ molecular mass of the Co^III-ligated peptide, and the Co^III-ligated peptide with additional oxygen atoms associated with the metallopeptide. ESI-MS data are most consistent with several oxygen atoms covalently attached to the metallopeptide. In order to obtain mass-spec data for [Co^IIINHase-m1] a relatively high cone-voltage was required, not unlike other metallopeptides prepared from similar peptide scaffolds.^42–45 Thus, significant fragmentation was observed in the mass spectrum of fully matured [Co^IIINHase-m1] (supporting information). The most prominent peak in the ESI-MS corresponds to NHase-m1 (minus the five protons required for Co^III ligation) with three additional oxygen atoms (m/z = 1339), while the second most prominent peak corresponds to the peptide with two additional oxygen atoms (m/z = 1323). The other two peaks observed correspond to Co-coordinated NHase-m1 with three oxygen atoms associated with the metallopeptide (m/z = 1398) followed by the metallopeptide with two oxygen atoms associated with the peptide (m/z = 1381). Thus the ESI-MS data is consistent with [Co^IIINHase-m1] containing oxidized cysteinate ligands. This is fully supported by the FT-IR data.

FTIR spectra were recorded with an ATR device on partially dehydrated films of [Co^IIINHase-m1]. To simplify the analysis of the resulting spectrum the FTIR spectrum of the apo-peptide was used as a background. The resulting FTIR data yielded peaks in the region where metal-coordinated Cys-SO₂ (1150 – 1075 cm⁻¹) and Cys-SO (1000 – 850 cm⁻¹) moities should be observed (Figure 3).¹⁶ To both confirm the presence of coordinated oxygenated cysteinates and determine their corresponding energies we matured [Co^IIINHase-m1] in the presence of ¹⁸O₂ gas. The resulting spectrum showed a disappearance of the peaks at 1091 and 928 cm⁻¹ from the ¹⁶O₂ matured spectrum and the appearance of peaks at 1052 and 889 cm⁻¹. This is consistent with what would be expected from a simple Hooke’s Law relationship of changing a S=¹⁶O harmonic oscillator to a S=¹⁸O harmonic oscillator. Furthermore, these peaks appear in the region where the Cys-SO₂ and Cys-SO asymmetric stretches in FeNHase appear following photorelease of NO.₁₆

Figure 3.

FTIR spectra of [Co^IIINHase-m1] matured in the presence of ¹⁶O₂ (red) vs.¹⁸O₂ (blue) gas.

Co K-edge X-ray absorption spectroscopy was utilized to probe the coordination environment about the Co-center of [Co^IIINHase-m1] (Figure 4). The pre-edge region displays a weak pre-edge feature that is assigned to a Co(1s → 3d) transition. This transition is formally dipole forbidden, but can gain intensity in non-centrosymmetric coordination environments through the mixing of 4p-character into the final state.⁹¹ Thus, the area under this peak can be compared with other complexes to estimate the coordination number about the cobalt center. We find that the area of this peak (6.8 eV relative to the edge height) is intermediate between a five and a six coordinate Co^III center, as was previously observed in the X-ray absorption spectrum of CoNHase (area = 6.3 eV).⁹²

Figure 4.

Top: XANES region of the X-ray absorption spectrum of [Co^IIINHase-m1] following 3 hours of maturation. Inset depicts the FF k³ data (Fourier Transformed from 2.2 – 14.2 k; back-transformed from 0.8 to 2.5 Å). Bottom: FT k³ data. In both the FT and FF k³ data the real data is represented as the solid red lines, the simulated data is depicted as the dashed blue line, and the difference spectra are depicted as the dashed green line. Best fits to the data: 3 N/O scatterers (1.93 Å σ² = 0.0062(8)Ų), 2 S scatterers (2.21 Å σ² = 0.0037(3)Ų); ε² = 0.82.

The EXAFS region of the Co K-edge X-ray absorption spectrum is also consistent with what had been previously observed for CoNHase.^92,93 We obtain a best-fit to the EXAFS data using a five coordinate model with the Co-center coordinated by two S-scatterers at 2.21 Å and three N/O scatterers at 1.93 Å. These data could also be refined for a five coordinate model with three S-scatteres at 2.22 Å and two N/O scatterers at 1.97 Å with only a moderate increase in both the error to the data refinement (ε² = 0.82 vs. 0.93) and the S Debey-Waller parameter (σ² = 0.0037(3)Ų vs. 0.0086(5)Ų). Similar to CoNHase, we also find a statistically valid fit to the EXAFS data for six coordinate models. When the EXAFS data for [Co^IIINHase-m1] is modeled as an S₂(N/O)₄ environment there is a slight modification to the refined bond-lengths with two S-scatterers at 2.24 Å and four N/O scatterers at 1.93 Å. Despite the slight increase in the error of the EXAFS fit (ε² = 0.82 vs. 1.06) there is a dramatic increase in the refined Debey-Waller parameter for the N/O scatterers (σ² = 0.0062(8)Ų vs. 0.022(5)Ų), making this six-coordinate model unlikely. Likewise, the S₃(N/O)₃ model yielded a valid fit to the data, but the Debey-Waller factors for both the S and N/O scatterers became unrealistically large (~ 0.015 Ų). Of all four models considered above the S₃(N/O)₂ model yielded a bond valence sum that was most consistent with Co(III) (3.19 vs. an optimal value of 3.00)⁵⁰ with the other refinements yielding bond valence sums that ranged from 3.30 – 3.70. Thus, the S₃(N/O)₂ is the most consistent formulation for the coordination environment of [Co^IIINHase-m1]. It should also be noted that overall the EXAFS data for [Co^IIINHase-m1] compare well with fits to the EXAFS data originally recorded for CoNHase.⁹³ Thus, it appears that the coordination environment about [Co^IIINHase-m1] is similar to the coordination environment about CoNHase/SCNase (Figure 1).

The electronic absorption spectrum of [Co^IIINHase-m1] is consistent with what has been previously observed for CoNHase and other low-spin thiolate-ligated Co^III complexes (Figure 5).^{32,41,87–90} The spectrum is relatively featureless with a gradual increase in absorbance from 11,500 to 20,000 cm⁻¹ (870 – 500 nm), shoulders at 22,125 (452 nm) and 28,735 cm⁻¹ (348 nm), and a defined peak at 35,460 ⁻¹ (282 nm). The CD spectrum (Figure 5) of the oxidized Co-metallopeptide is also consistent with CoNHase, however, the signs of the resulting low energy bands are different than observed in the metalloenzyme.⁹⁴ This can be largely rationalized by the fact that the Co^III centers contained within NHase-m1 vs. NHase are in different chiral environments thus leading to different CD properties.

Figure 5.

CD (top) and electronic absorption spectra (bottom) of [Co^IIINHase-m1] obtained in 10 mM NEM buffer at a pH = 8.0. The black solid spectrum represents the experimental data, the dashed red peaks represent the best-fit of the absorption and CD spectra decovolved into Gaussian line shapes, and the dashed blue spectrum represents the best-fit of the absorption and CD spectrum resulting from the sum of the Gaussian line shapes.

Further information concerning the ligand environment about Co^III was gained from an analysis of the CD and absorption bands in the ligand field region of the corresponding spectra, which allows for the simultaneous deconvolution of the spectra into Gaussian line shapes (Table 1, OFigure 5). From these data it is instantly apparent that the complex cannot be described as arising from symmetry; values obtained for the Racha B parameter and 10D_q are unreasonably small for low-spin Co^III. Instead, the data is nicely described for a complex arising from D₄ (or C₄) parentage. For transitions to be CD active they must be both electronically and magnetically dipole allowed. The anisotropy factor, g (g = |Δε|/ε), is a sensitive measure for judging if a transition is magnetically dipole allowed; if g is greater than 0.01 the transition can be considered magnetically dipole allowed.⁹⁵ When g is calculated for the four lowest energy transitions they are all consistent with magnetically dipole allowed transitions with one exception, the lowest energy transition; this transition has a g = 0.004 while the remaining three have g > 0.01 (Table 1). This situation is best described by a ligand environment derived from D₄ (or C₄) parentage. In the case of states derived from D₄ parentage the lowest energy transition, the ¹B₂ ← ¹A₁ transition, is magnetically dipole forbidden, while the other three next highest energy transitions are magnetically dipole allowed (i.e. the ¹E ← ¹A₁, ¹A₂ ← ¹A₁, and ¹E ← ¹A₁). Thus, the CD and absorption data are most consistent with the Co^III center contained within either a square-pyramidal five-coordinate or tetragonally distorted six-coordinate ligand environment. We will argue that the five-coordinate species is most likely (vide infra, vide supra).

Table 1.

Spectral Parameters Obtained from the Simultaneous Deconvolution of the Electronic Absorption and CD spectra of fully matured [Co^IIINHase-m1].

Peak	Energy (cm−1)	ε(M−1 cm−1)	Δε(M−1 cm−1)
1	12,602	30	−0.12
2	15,766	80	8.13
3	17,794	140	−2.50
4	19,135	235	−2.51
5	20,842	825	−4.79
6	22,813	685	3.03
7	24,621	1,370	2.46
8	26,663	1,630	2.98
9	29,015	2,440	3.19
10	31,686	2,380	−4.36
11	33,744	2,490	−4.08
12	34,813	N/Aa	−4.83

^aThis peak could not be deconvoluted in the electronic absorption spectrum due to an intense feature at ~ 38,000 cm⁻¹.

Nitrile Hydrolysis By [Co^IIINHase-m1]

Nitrile hydrolysis studies were performed in aqueous buffer (10 mM NEM; pH 8.0) by the addition of 1 eq. of [Co^IIINHase-m1] to 100 eq. of nitrile (10 mM nitrile was used). Reactions were run for 18 hours at 25 or 50 °C in sealed NMR tubes under an atmosphere of argon. The nitriles that were investigated included: acetonitrile, 3-cyanopyridine, and acrylonitrile. In the case of 3-cyanopyridine no nitrile hydrolysis was observed under any conditions, while for acetonitrile only trace amounts of acetamide could be detected by GC/MS after 18 hours of reaction time at 50 °C. Analysis of the acetonitrile reaction mixture indicated that approximately 0.6 equivalents of acetonitrile per [Co^IIINHase-m1] was converted to acetamide after 18 hours of incubation. Prolonged reaction times yield no additional products. In contrast to these two nitriles, the sterically unencumbered and activated nitrile acrylonitrile was readily converted into acrylamide; up to 58 turn overs of acrylonitrile to acrylamide was observed after 18 hours of reaction time. Control experiments of buffer, 0.1 mM [Co^III(NH₃)₆](3Cl⁻), or the NHase-m1 apo-peptide yielded no evidence of nitrile hydrolysis.

The hydrolysis of acrylonitrile to acrylamide is highly dependent upon the maturation time of [Co^IIINHase-m1] (i.e. the exposure time of the metallopeptide solution to air prior to purging the solution with argon). If the metallopeptide solution is exposed to air for only 10 min and then purged with argon no acrylonitrile hydrolysis is observed. As one extends the maturation time the amount of nitrile hydrolysis dramatically increases, with an optimal maturation time of 3 hours observed for [Co^IIINHase-m1]. Following 3 hours of air maturation there is a precipitous decrease in acrylonitrile hydrolysis, with no acrylonitrile hydrolysis observed following 5 hours of maturation (Table 2). This strongly suggests that there is a slow activation process followed by a more rapid deactivation process effected by O₂.

Table 2.

Turnover Numbers Following 18 Hours of Reaction Time Per [Co^IIINHase] as a Function of Air Exposure Time

Air Exposure Time (min.)	Turnover Number
0	NDa
10	NDa
60	12(2)
120	36(2)
180	58(4)
240	17(1)
300	NDa
360	NDa

^aND: nitrile hydrolysis was not detected

The reduced catalytic activity of [Co^IIINHase-m1] compared to the NHase metalloenzymes can be readily rationalized. It is reasoned that the mechanism of nitrile hydrolysis likely involves hydroxide (or water) attack of a M^III-bound nitrile.^96–98 The active-site of [Co^IIINHase-m1] is stericly crowded owing to the peptide loop comprised of residues Cys(2) to Cys(6).^99,100 This makes it difficult, if not impossible, for all but the most stericly unencumbered nitriles to coordinate to the metal-center of [Co^IIINHase-m1]. Therefore 3-cyanopyridine likely will not be catalytically hydrolyzed by [Co^IIINHase-m1] because it is incapable of coordinating to the Co-center, as is observed. In addition the active-site of this minimized CoNHase mimic has removed many of the hydrogen bonding residues from about the NHase active-site that are likely vital for catalysis.^8,10–14 Therefore only activated nitriles that are already prone to nucleophilic attack will be effectively hydrolyzed by [Co^IIINHase-m1], explaining why acrylonitrile is converted to acrylamide catalytically while acetonitrile is not hydrolyzed effectively.

Identification of the Active Form of [Co^IIINHase-m1]

From the above acrylonitrile hydrolysis experiments it seems a reasonable suposition that the active form of [Co^IIINHase-m1] is obtained only after relatively extensive oxidation by O₂. This could be explained by either the slow oxidation of Co^II to Co^III or a slow sulfur based oxidation process. Analysis of the NIR region of the electronic absorption spectrum of [Co^IIINHase-m1] following 10 min of air maturation shows the complete disappearance of the low energy Co^{II 4}T₁(F) ← ⁴A₂(F) transition demonstrating the complete oxidation of Co^II to Co^III in this time frame. This suggests that slow sulfur based oxidation may be the reason for the low activity of the not fully matured [Co^IIIN Hase-m1].

To quantify the changes in sulfur based oxygenation of [Co^IIINHase-m1] we examined the FTIR spectrum of the metallopeptide as a function of air exposure time (Figure 6). After 10 min of air maturation only the bands corresponding to a sulfinate species are present, with bands observed between 1120 – 1060 cm⁻¹. Over the course of 3 hours the band corresponding to the sulfenate begins to appear at 928 cm⁻¹, with the maximum in absorbance of this transition reached after approximately 3 hours. Following three hours of maturation the band corresponding to the sulfenate disappears, and is completely absent after 5 hours [Co^IIINHase-m1] of air exposure. Concurrent with the disapperance of the sulfenate stretch is the appearance of additional stretching modes at 1073 and 1127 cm⁻¹, suggesting the formation of a bis-sulfinate coordinated Co^III center. A bis-sulfinate species appears to be the end sulfur-based oxygenation product of Co^III and Fe^III coordination compounds reminiscent of NHase; such products are observed in both SCNase, NHase, and small molecule Co- and Fe-NHase biomimetic complexes.^1,9,14,36 Furthermore, there is good experimental evidence that the bis-sulfinate form of SCNase is catalytically inactive,⁹ which is what we observe in our NHase biomimetic compound. From our data it thus appears that the active form of [Co^IIINHase-m1] has one sulfinate and one sulfenate coordinated to the Co^III center (Scheme 4).

Figure 6.

Infrared spectra of [Co^IIINHase-m1] recorded after 10 minutes (blue), 1 hour (green), 3 hours (red), and 5 hours (purple) of exposure to air.

Scheme 4.

Comparison of The Experimental Spectroscopic Data With Computationally Derived Spectroscopic Data

To further probe the identity of the active species a series of computational models were constructed based on the predicted active-site structure of [Co^IIINHase-m1]. The electronic absorption, CD and IR spectra of these complexes were then calculated and compared with the available experimental data for [Co^IIINHase-m1] described above. All of the models involved coordinating a Co^III ion to an AcN-Cys-Cys-H bis-peptide, an equatorial ethanethiol, and an additional axial ligand. Subsequent sulfur-based oxidation and sulfenate protonation was then examined. Of the fifteen computational models considered, stationary points along the potential energy surface could only be located for thirteen; in the case of the computational model with no oxygenated sulfurs ((Co-NHase)²⁻; 1) only the five coordinate complex was found to be a stationary point along the potential energy surface. For the case of MeCN and H₂O coordinated 1 all reasonable six coordinate structures did not correspond to stationary points (i.e. several imaginary vibrational modes were obtained in the IR spectra). For the oxygenated computational models, both five coordinate, MeCN, and H₂O bound forms could be successfully located. Structures of the computational models and their relevant metrical parameters are displayed in Chart 2 and Table 3, respectively. Surprisingly, all of the five coordinate computational models investigated were found to posses singlet ground states; higher spin-states (triplet and quintet spin states) were found to be ~4,000 cm⁻¹ higher in energy than the low-spin complexes. This is a direct result of sulfur-based oxygenation, which renders the sulfur π based AOs unavailable for bonding with the cobalt π-type AOs. As a consequence the strong σ bonding along the molecular x, y, and z axis from the anionic ligands significantly destabilizes the Co(3d_x2-y2) and Co(3d_z2) type orbitals while relatively poor π-donation from the amides to the Co(3d(π)) orbitals leads to only a small destabilization of these d-type orbitals (Figure 6). It should be noted that these five coordinate complexes also yielded Co-S and Co-N bond lengths that best matched the EXAFS data obtained for [Co^IIINHase-m1], with Co-S bond lengths ranging between 2.21 – 2.26 Å and Co-N bond lengths ranging between 1.90 – 1.98 Å (Table 3).

Table 3.

Co^III Bond Lengths Obtained Form The Electronic Structure Calculations For Selected Computational Models (mPWPW91/aug–cc–TZVPP).

Model	Co–S Åa	Co–S Åb	Co–S Å	Co–N Åc	Co–N Å	Co–O Å
2	2.220	2.170	2.223	1.975	1.950	N/A
2–H2O	2.229	2.177	2.233	1.989	1.954	2.106
3	2.228	2.167	2.235	1.980	1.969	N/A
3–H2O	2.227	2.216	2.283	1.996	1.976	2.070
4	2.219	2.159	2.243	1.965	1.888	N/A
4–H2O	2.265	2.191	2.270	1.992	1.918	2.114
5	2.229	2.119	2.232	1.984	1.941	N/A
5–H2O	2.282	2.241	2.292	2.003	1.972	2.080

^athiolate trans to the water or vacant–site;

^bsulfinate;

^camidate nitrogen trans to the sulfinate.

IR spectra were calculated for all stationary points with both ¹⁶O₂ and ¹⁸O₂ labeled oxygenated sulfurs (Table 4). Of these an excellent match between the experimental data obtained for [Co^IIINHase-m1] and the simulated vibrational results calculated was obtained for the five-coordinate protonated-sulfenate/sulfinate ligated computational model 4 ((Co-NHase-SO₂/SO(H))¹⁻) (Figure 8, Table 4). The calculated spectrum displays ¹⁶O₂=S and H-¹⁶O=S stretches at 1093 and 930 cm⁻¹, respectively, that red-shift to 1056 and 881 ⁻¹ upon ¹⁸O substitution. The calculated IR spectrum for the corresponding six-coordinate water ligated 4-H₂O ((Co-NHase-SO₂/SO(H)-H₂O)^1-) displays a ¹⁶O₂=S stretch at 1154 cm⁻¹ and an H-¹⁶O=S stretch at 1051 cm⁻¹, which are both much to high in energy to be consistent with the experimental data. The unprotonated complex 3-H₂O is an even poorer match to the experimental data with a ¹⁶O₂=S stretchs at 1232/1073 cm⁻¹ and an ¹⁶O=S stretch at 1179 cm⁻¹. We note that the calculated IR spectrum for the five coordinate mono-sulfinate complex 2 ((Co-NHase-SO₂)^2-) matches well with the IR data obtained for [Co^IIINHase-m1] following 10 min of maturation, while the bis sulfinate computational model 5-H₂O ((Co-NHase-2(SO₂)-H₂O)^2-) has 16O₂=S stretches that match well with the IR data obtained for [Co^IIINHase-m1] following 5 hours of air exposure. These data are summarized in Table 4.

Table 4.

Vibrational data obtained for [Co^IIINHase-m1] and selected computational models for the sulfinate and sulfenate S=O stretching energies (mPWPW91/aug–cc–TZVPP).

Compound	S=16O2 (cm−1)	S=18O2 (cm−1)	S=16O (cm−1)	S=18O2 (cm−1)
[CoIIINHase-m1]	1091	1052	928	889
2	1098	1060	N/A	N/A
	1023	982
3	1081	1056	1044	1005
3–H2O	1232	1209	1179	1131
	1073	1028
4	1093	1056	930	881
4–H2O	1154	1121	1051	1010
	1126	1094
5–H2O	1131	1096	N/A	N/A
	1120	1087
	1082	1045
	1064	1033

Figure 8.

Simulated IR spectrum of 4 with ¹⁶O (red) and ¹⁸O (blue) labeling of the sulfinate and sulfenate. The sticks represent the individual transitions and the spectra was generated by applying Gaussian line shapes (50 cm⁻¹ peak width) to the transitions and summing them together.

Comparisons were also made between the experimental electronic absorption and CD spectra obtained for [Co^IIINHase-m1] with the simulated spectra of 3-H₂O, 4, 4-H₂O, and 5-H₂O (Figure 9 and supporting information). The aqua-complexes of 3 and 5 as well as five-coordinate 4 were examined as these represent minima along the potential energy surface (vide infra), while 4-H₂O was examined for comparison with the other six-coordinate aqua-complexes. All excited state calculations employed Neese’s spectroscopically oriented configurational interaction (SORCI) methodology⁷⁷ where the five lowest energy transitions were examined. Unlike time-dependent DFT, the SORCI method is a true multiconfigurational ab initio method, and therefore takes into account the multiconfigurational nature of the excited and ground states. Although this is most important for open shell systems, we and others have found that it is an economical method that is much more accurate in reproducing the electronic spectra of closed shell systems than other more widely used excited state methods.^77,101–103 A full account of the theory behind the SORCI method can be found in reference 77. Briefly, in the SORCI method the first order active-space is divided into strongly and weakly interacting perturbing configurations according to user defined cutoffs, the later are treated variationally while the former are treated using second-order Møller-Plesset theory. Results obtained by the SORCI method generally agree well with more expensive techniques at a fraction of the computational cost making it attractive for larger systems.

Figure 9.

Simulated CD spectra for 2-H₂O (A), 3-H₂O (B), 4-H₂O (C), and 4 (D). Simulated spectra are sums of Gaussian lineshapes applied to the individual transitions (red sticks) using a 1500 cm⁻¹ peak width.

In this study we used a moderately–large sized CAS(8,8) active space to examine the first five spin allowed transitions of the four computational models outlined above (Figure 9, supporting information). The simulated and experimental electronic absorption data are rather indistinct, as would be expected for low-spin Co^III. In contrast the simulated and experimental CD spectra are rather illuminating. For all six coordinate computational models considered the simulated CD spectra do not match that of the experimental data well in terms of both transition energies, rotational strengths and signs. For these, the lowest energy transition corresponds to a nominal Co(3d(π)) → Co(3d_x2-y2) transition, which occur at relatively high energies (~17,000 cm⁻¹) when compared to the experimental data. Although a ~5,000 cm⁻¹ blue-shift in calculated vs. experimental data may be expected for methods such as TD-DFT the SORCI methodology is not typically this inaccurate. Furthermore, the calculated Co(3d(π)) → Co(3d_x2-y2) transition has a large rotational strength, and is negatively signed, which is inconsistent with the experimental data. The next two transitions for the six-coordinate models considered are weak and positively signed, while the forth and fifth are stronger than the first and are negatively signed. This leads to an overall low energy region of the CD spectrum that does not match the experimental data well for any of the computationally derived aqua-species. It should be noted that the CD spectrum of [Co^IIINHase-m1] following extended O₂ exposure matches well with these six-coordinate simulated CD spectra (supporting information).

In contrast, the simulated spectrum for five-coordinate 4 is an excellent match to the experimental data. This displays a weak negative signed CD band at 12,857 cm⁻¹ corresponding to a nominal Co(3d_xy) → Co(3d_z2) transition, two strong positively signed transitions corresponding to Co(3d(π)) → Co(3d_z2) transitions at ~15,500 cm⁻¹, an intense negatively signed Co(3d_xy) → Co(3d_x2-y2) transition at 17,619 cm⁻¹ and an intense negatively signed Co(3d(π)) → Co(3d_x2-y2) transition at 19,866 cm⁻¹. Thus, the low energy region of the computationally derived CD spectrum for 4 matches the overall line shape, transition signs and energies determined experimentally for [Co^IIINHase-m1]. Also, the calculated g values compare well with those determined experimentally (supporting information). The CD spectrum of the five coordinate unprotonated species was also investigated and produced a reasonable match to the experimentally derived CD data, albeit with minor (~500 cm⁻¹) perturbations to the transition energies. However, we note that the calculated pK_a of the sulfenate H-OS proton is likely protonated under the conditions we examined for this study.¹⁰⁴ We reason that the hydrogen bonding between the H–O=S proton and the adjacent sulfinate oxygen raises the sulfenic acid pK_a above the typical range of 3 – 7¹⁰⁵ in a manner similar to how proton sponge behaves.

Energetics of Ligand Displacement From the Co^III Center as a Function of Sulfur Oxygenation

It has been speculated that nitrile hydrolysis by NHase and thiocyanate hydrolysis by SCNase occurs through substrate coordination to the M^III center of the metalloenzyme’s active-site.^7,96–98 Due to the sterics about the metal-center in these metalloenzymes such a process must occur through a dissociative mechanism where water (if bound) will dissociate from the M^III center and the substrate will coordinate to the vacant site. We therefore examined the energetics of the first two steps of the catalytic process: water dissociation from the Co^III center of the computational [Co^IIINHase-m1] models followed by nitrile (MeCN in this case) association.

Figure 10 displays the computationally derived (B3LYP/aug-cc-TZVPP) free energies associated with each step. Here all energies are zero-point corrected, and are normalized to the corresponding aqua-complexes for comparative purposes. The mono-sulfinate complex shows a significant energy minima for five-coordinate 2, which lies 13.45 kcal mol⁻¹ lower in energy than the aqua-complex 2-H₂O and 13.52 kcal mol⁻¹ lower in energy than the MeCN ligated complex 2-MeCN. Thus, 2 would be expected to remain five-coordinate in solution as the equilibrium for either water or nitrile ligation would lie significantly to the left. In contrast, water ligated sulfenate/sulfinate 3-H₂O is energetically favored due to strong hydrogen bonding to the oxygenated sulfurs; dissociation of the water forming 3 is 12.18 kcal mol⁻¹ uphill, while MeCN ligation is 13.33 kcal mol⁻¹ higher in energy than the aqua-complex. Protonation of the sulfenate oxygen, forming 4, significantly destabilizes water hydrogen bonding to the oxidized sulfurs, and thus destabilizes water ligation. Thus, five coordinate 4 is 14.03 kcal mol⁻¹ lower in energy than 4-H₂O. Furthermore, ligation of MeCN to the metal-center forming 4-MeCN is lower in energy than the five coordinate species, resting 16.05 kcal mol⁻¹ below 4-H₂O. Full oxygenation to the bis-sulfinate ligated compound 5 yields a highly favored aqua-species due to strong hydrogen bonding between the two sulfinate moieties; formation of the five coordinate compound 5 is 29.02 kcal mol⁻¹ uphill from the aqua species, while the MeCN compound 5-MeCN is 9.44 kcal mol⁻¹ uphill from the aqua-species. Thus it would appear on the basis of the available spectroscopic and computational evidence that the Co^III center of the active-form of [Co^IIINHase-m1] is five coordinate in aqueous solution. It would also displays a small, but energetically favorable, ability to coordinate nitrile substrates over remaining a five-coordinate species.

Figure 10.

Calculated free energies of ligand exchange normalized to the aqua species X-H₂O. The energetics of models 2 (black), 3 (blue), 4 (red), and 5 (green) are depicted.

Analysis of the Mayer bond order of the aqua-species reveals an interesting trend in bond order as the equatorial thiolate ligands to Co^III ions are oxidized (Figure 11).^66–68 Not surprisingly, as the sulfurs become oxidized the Co-S bond order decreases. This is compensated for not by an increase in the bond orders of the amide nitrogens trans to the oxidized thiolates, but instead by the ligands along the axial bonding vectors. In other words, as the bonding decreases within the xy plane it is compensated for by an increase in bonding along the z-axis. Thus, a stronger Co^III-OH₂ bond is made as the thiolates become oxidized. The only exception to this trend is the protonated sulfenate complex 4-H₂O, which has a significantly reduced Co^III-O bond-order when compared to unprotonated 3-H₂O (0.392 vs. 0.513). This is a direct result not of an intrinsic electronic destabilization of the aqua-ligand in 3-H₂O resulting from a change in the (H)O=S electronics, but instead results from the disruption of the water-O=S hydrogen bonding interactions. This elongates the Co^III-OH₂ bond in 4-H₂O vs. 3-H₂O, which in turn reduced the Co^III-O bond order.

Figure 11.

Mayer bond orders (mPWPW91/aug-cc-TZVPP) calculated for the aqua-species X-H₂O (X = 2, 3, 4, and 5).

Relevance to NHase and SCNase

NHases and SCNases are unusual metalloenzymes in that they possess post-translationally modified cysteinate residues coordinated to the active-site metal-ion.¹ It has been widely speculated that these residue modifications are required for catalytic activity, however, the extent of cysteinate oxidation and the reason why NHase and SCNase requires cysteinate oxidation to be catalytically active is still debated.^{1,9,13,14,106–108} Using metallopeptide based CoNHase mimics in concert with computational models we have demonstrated that a Co^III ion contained within a CoNHase-like ligand environment requires the oxidation of one of the cysteinates to a sulfinate and another to a sulfenate for the metallopeptide to display NHase activity. It is likely that the sulfenate is protonated under aqueous conditions. Considering the large number of waters and other proton donating groups within the active-site of NHase and SCNase we suggest that a similar situation exists in these metalloenzymes (e.g. suflenate protonation or strong Arg hydrogen bonding), as has been supported by computational, FTIR and XAS studies.^13,15,16,106 We note that the thiolates located in the equatorial plane of [Co^IIINHase-m1] are the likely candidates for oxidation. A computational analysis suggests that the oxidation of the axial thiolate, as has been observed experimentally in Fe^III and Co^III complex with similar ligand sets,^34,109 leads to the sulfienate dissociation stabilizing a four-coordinate Co^IIIN₂S₂ species (supporting information), experimentally derived structures of which have been previously reported.^31,110

A number of biological inferences can be made from this work, especially when comparisons are made to results from the biochemical literature. First, there are no high resolution crystallographic structures of either NHase or SCNase that lack equatorial cysteinate oxidation. It has been shown from biochemical studies that a Co-chaperone, NhlAE, is responsible for cobalt insertion into apo-NHase through a unique “subunit-swapping” mechanism.^111,112 NhlAE contains three subunits, the α-subunit of CoNHase that contains Co^III (in an identical ligand environment as CoNHase) and two NhlE subunits. It was suggested that the NhlE subunits of NhlAE are responsible for both Co-insersion into the α-subunit and the oxidation of the cysteinates. Furthermore, it was suggested that the origin of the oxidized cysteine oxygen-atoms was bulk water.¹¹² Our study cannot comment on the ability of the NhlE subunit of NhlAE to insert cobalt (likely introduced as Co^II based on biochemical studies)^112,113 into the NHase α-subunit. However, our study makes it apparent that oxidation of Co^II to Co^III by O₂ in a ligand environment reminiscent of NHase is facile. Also, using ¹⁸O₂ gas we have demonstrated that the oxygen atoms of SO₂ and SO within a NHase-like coordination motif can easily be derived from O₂. Furthermore, our data suggests that the oxidation of the cysteinates is regiospecific; only the equatorial cysteinates appear to be oxidized to sulfinate/sulfenates. It is thus reasonable to suggest that a similar situation likely exists in both Fe/CoNHase and SCNase, as has been suggested in recent work on sulfur oxidation by isopenicillin-N-synthase.¹¹⁴ Also, it appears that the oxidation of one cysteinate to a sulfinate occurs rapidly after cobalt-insersion, while the second oxidation appears to be considerably slower. Such a finding is consistent with other studies of Co^III contained in a NHase-like environment.^9,32,33

Another important finding of this study is that catalytic activity of [Co^IIINHase-m1] is only achieved when the equatorial cysteinates are contained in the SO₂ and SO(H) oxidation states. This has been speculated for NHase ever since the first high-resolution crystal-structure of FeNHase appeared in 1998,¹³ however it has never been unambiguously demonstrated. In a previous combined crystallographic/activity study using SCNase the authors suggested that the SO(H) modification is key to SCNase activity; the further oxidation of the Cys-SO ligand to a Cys-SO₂ ligand dramatically reduced (but did not eliminate) the activity of the metalloenzyme.⁹ Our evidence strongly suggests that the same occurs in NHase, where the mono and bis-SO₂ forms of our models were inactive. Such a finding is at odds with data suggesting the fully oxygenated metalloenzyme is active, which was based on pure crystallographic evidence.¹⁴

Lastly our study suggests why the SO₂ and SO(H) modifications are necessary. The above presented computational results suggest that when the cysteinates are unoxidized, or only oxidized to the mono-SO₂ form, neither substrate nor water can coordinate to the Co^III center. Oxidation of the other equatorial cysteinate to an SO or SO(H) ligand reduces the electron density being donated to the Co-center by the thiolates, thus allowing for the ligation of an axial ligand. Protonation of the sulfenate to an SO(H) ligand appears to be key as it significantly destabilizes water-binding via disruption of the water-O=S hydrogen bonding network allowing for facile water dissociation and substrate coordination. Full oxygenation of the SO to an SO₂ ligand so strongly favors water binding that a five-coordinate species cannot be generated. A similar situation is observed in the SCNase crystal structure where the fully oxidized bis-SO₂ form of SCNase shows a six-coordinate Co^III ion with a water ligand.⁹ In the structures of SCNase with the mono-SO₂ ligand or the mono-SO₂ and SO(H) ligand the Co^III center is five coordinate with the site trans to the unmodified axial cysteinate remaining vacant. Other support for this supposition comes from additional computational and synthetic work on NHase-like metal centers.^15,39,115 We note that it remains to be seen, both on the basis of this study and others, if the SO(H) ligand plays a more intimate role in the NHase catalytic-cycle (such as proton shuttling to the substrate)^115,116 beyond the tuning of the metal-center electronics.

Figure 7.

Energy level diagram of the Co(3d) orbitals depicting the occupied (blue) and unoccupied (red) orbitals. Iso-surface plots of the corresponding molecular orbitals were generated with the molecular graphics program Molkel.