Structures, spectroscopic properties, and dioxygen reactivity of 5- and 6-coordinate nonheme iron(II) complexes: a combined enzyme/model study of thiol dioxygenases

Abstract

The synthesis of four new Fe^II(N₄S(thiolate)) complexes as models of the thiol dioxygenases are described. They are composed of derivatives of the neutral, tridentate ligand triazacyclononane (R₃TACN; R = Me, iPr) and 2-aminobenzenethiolate (abt^x; X = H, CF₃), a non-native substrate for cysteine dioxygenase (CDO). The coordination number of these complexes depends on the identity of the TACN derivative, giving 6-coordinate (6-coord) complexes for Fe^II(Me₃TACN)(abt^x)(OTf) (1: X = H; 2: X = CF₃), and 5-coordinate (5-coord) complexes for [Fe^II(iPr₃TACN)(abt^x)](OTf) (3: X = H; 4: X = CF₃). Complexes 1 – 4 were examined by UV-vis, ¹H/¹⁹F NMR, and Mössbauer spectroscopies, and density functional theory (DFT) calculations were employed to support the data. Mössbauer spectroscopy reveals that the 6-coord 1 – 2 and 5-coord 3 – 4 exhibit distinct spectra, and these data are compared with that for cysteine-bound CDO, helping to clarify the coordination environment of the cys-bound Fe^II active site. Reaction of 1 or 2 with O2 at −95 °C leads to S-oxygenation of the abt ligand, and in the case of 2, a rare di(sulfinato)-bridged complex, [Fe2^III(µ-O)(O₂S(NH₂)C₆H₃CF₃)₂](OTf)₂ (5), was obtained. Parallel enzymatic studies on the CDO variant C93G were carried out with the abt substrate, and show that reaction with O₂ leads to disulfide formation, as opposed to S-oxygenation. The combined model and enzyme studies show that the thiol dioxygenases can operate via a 6-coord Fe^II center, in contrast to the accepted mechanism for nonheme iron dioxygenases, and that proper substrate chelation to Fe appears to be critical for S-oxygenation..

Graphical Abstract

Introduction

Cysteine dioxygenase (CDO) is a nonheme iron oxygenase responsible for regulating human cysteine levels by breaking down cysteine (cys) to cysteine sulfinic acid.¹ CDO represents one member of a family of thiol dioxygenases, which includes cysteamine dioxygenase (ADO) and 3-mercaptopropionic dioxygenase (3MDO). These enzymes convert sulfur substrates to sulfinic acids using O₂ as the oxidant (Figure 1). The resting state of CDO is known to contain a high-spin ferrous center that is coordinated by three histidine residues in a facial arrangement. The evidence indicates that the other thiol dioxygenases have similar iron active sites bound by three His residues.^2–3 The structure of substrate-bound CDO with cys coordinated to the iron center was obtained from single crystal X-ray diffraction (XRD) studies,⁴ and shows the expected 3-His ligand motif as well as the cys chelated through the sulfur atom and the nitrogen atom of the amino group. It is plausible that the other thiol dioxygenases exhibit similar substrate-bound structures, although they have not yet been characterized by XRD.^{2–3, 5}

Figure 1.

Crystal Structure of CDO variant C164S (PDB: 4Z82)

The 3-His coordination environment of CDO contrasts with the structures of other nonheme iron dioxygenases, which typically contain a 2-His-1-carboxylate motif (the “facial triad”).⁶ The significance of the unusual 3-His coordination environment, and the importance of these structural differences with regard to substrate binding and O₂ activation, remains poorly understood.^7–9 A second unusual finding for cys-bound CDO comes from recent Mössbauer studies, which shows two overlapping quadrupole doublets consistent with high-spin ferrous centers, indicating two distinct types of iron sites. Jameson and coworkers proposed that the two doublets correspond to a mixture of five-coordinate (5-coord) Fe^II, with an N₄S primary coordination sphere, and a six-coordinate (6-coord) Fe^II, with a bound H₂O molecule giving rise to N₄SO coordination.¹⁰ The presence of the 6-coord Fe^II center is supported by data for a C164S variant, which has similar activity to wild-type (wt) CDO, and exhibits a 6-coord N₄SO-ligated iron center with an H₂O molecule occupying the putative O₂-binding site (Figure 1).¹¹ Similarly, the CDO variant H155A with cys was proposed to contain water as a sixth ligand on the basis of magnetic circular dichroism and computational studies.¹²

The presence of a 6-coord Fe^II center with H₂O occupying the O₂ binding site in substrate-bound, wt-CDO contradicts the accepted mechanism for O₂ activation by nonheme Fe enzymes.^13–14 In this mechanism, the initial resting state of a 6-coord Fe^II center ligated by the 2-His-1-carboxylate facial triad and three H₂O molecules converts completely to a 5-coord site upon binding substrate/co-substrate, releasing all H₂O ligands. The enzyme reacts with O₂ only after loss of H₂O and formation of the 5-coord iron center (Scheme 1).¹³ It is therefore of significant interest to determine if wild-type CDO is a rare example of a nonheme Fe dioxygenase that does not strictly follow these structural changes at the iron center.¹⁵ One possibility is that 5-coord and 6-coord Fe^II sites are in equilibrium following coordination of cys, and it is the 5-coord site that reacts with O₂. However, it is also possible that O₂ displaces the H₂O molecule via direct attack on the 6-coord center. In addition to their unique active site structure, the thiol dioxygenases also exhibit a high degree of substrate specificity.^16–17 For example, mammalian CDO will bind the non-native substrate homocysteine at the iron center, but will not oxygenate this substrate.¹⁷ A more recent study was carried out on the reactivity of MmCDO and Av3MDO (Mm: Mus musculus; Av: azotobacter vinelandii) with the non-native substrate 2-aminothiophenol (or if deprotonated, 2-aminobenzenethiolate (abt)). This substrate could not be S-oxygenated, but in the presence of both O₂ and alcohol (MeOH, EtOH, BnOH), benzothiazole derivatives were formed. It was proposed that activation of O₂ by the Fe center leads to oxidation of the alcohol solvent to aldehyde, which then reacts with abt to give the benzothiazole derivative. However, no direct information on the binding or structure of abt to the active site of CDO or Av3MDO was obtained.¹⁸

Scheme 1.

Proposed Mechanisms of O₂ Activation by αKG-Dependent Iron Dioxygenases (top) and CDO (bottom)

To date there are no Fe^II thiol dioxygenase model complexes that utilize the non-native abt substrate.¹⁹ Herein we describe four new Fe^II(N₄S(thiolate)) complexes that incorporate abt and a CF₃-substituted derivative and include the same triazacyclononane (TACN) scaffold. These complexes form a series of closely related 5-coord and 6-coord Fe^II complexes in which the N₄S ligand set is held constant, but where the sixth site is either unoccupied, or contains counterion or solvent. The structural and spectroscopic properties of these complexes, including both solid-state and solution Mössbauer spectra, as well as density functional theory (DFT) calculations, provide a new body of comparative data for analysis of the 5-coord/6-coord states proposed for cys-bound CDO.

The O₂ reactivity of these complexes is also described. Previously we demonstrated that certain Fe^II(N₄S(thiolate)) complexes react with O₂ to give selective S-oxygenated products, and in one case formation of an iron-sulfinate complex was seen, similar to the thiol dioxygenases.^20–21 Limberg and Fiedler have also prepared Fe^II models of CDO and ADO using trispyrazolylborate or tris(imidazolyl)phosphine ligands, and cysteine ester or cysteamine as the sulfur source.^22–25 These complexes also reacted with O₂, ultimately yielding organic sulfinate products, but the putative iron-sulfinate complexes were not crystallographically characterized. In the current work, it is shown that the bound abt ligands are S-oxygenated upon addition of O₂ to the Fe^II model complexes. In the case of the CF₃-substituted derivative, a dioxygenated sulfinate-iron complex is definitively characterized by XRD, and these results show that thiol dioxygenase activity does occur for Fe^II(abt) model complexes.

Parallel studies on CDO with the non-native abt and abt^CF3 substrates were performed for direct comparison with the model complex studies, and as a comparative study to the earlier CDO/abt work of Pierce and coworkers.¹⁸ A C93G variant of CDO was employed because it is unable to form the thioether crosslink between tyrosine 157 and cysteine 93, and yet has comparable reactivity to wild-type.²⁶ However, unlike wt-CDO, which exists as a ~1:1 mixture of crosslinked and uncrosslinked protein, C93G is present in only one form, thereby simplifying all binding and kinetic studies. In contrast to the model complexes, S-oxygenation does not occur with the enzyme, and only trace amounts of benzothiazole are detected. The first Mössbauer studies on CDO in the presence of the abt derivatives addresses the question of substrate binding to the active site. A rare combination of both model complexes and enzymatic studies are presented here in one report, and together the results point to a critical role for proper substrate binding in controlling thiol dioxygenase activity, as well as providing new insights regarding possible mechanisms for nonheme iron dioxygenases.

Results and Discussion

Synthesis and Structural Analysis.

The triazacyclononane (TACN) ligand was selected as a neutrally charged, tridentate nitrogen donor to model the neutral 3-His binding motif of the thiol dioxygenases. This ligand is well-known to give stable iron complexes in a range of oxidation states,^27–29 and the steric properties of the ligand can be tuned by the substituents attached to the amine donors.^30–32 The methyl-substituted Me₃TACN is known to favor 6-coordinate ferrous complexes, whereas the isopropyl-substituted iPr₃TACN is only known to give 5-coordinate Fe^II complexes. The lower coordination number favored by the iPr derivative was previously attributed to the steric encumbrance imposed by the ligand.^30–31 The syntheses of 1 – 4 are shown in Scheme 2. Reaction of the Fe^II precursor³⁰ [Fe^II(Me₃TACN)(CH₃CN)₃](OTf)₂ in acetonitrile with 2-aminothiophenol and triethylamine added as base in THF led to Fe^II(Me₃TACN)(abt)(OTf) (1), which was isolated as colorless crystals in good yield (80%). Synthesis of the CF₃-substituted abt analog (abt^CF3) was accomplished by adding a THF solution of triethylammonium 2-amino-4-(trifluoromethyl)benzenethio-late to [Fe^II(Me₃TACN)(CH₃CN)₃](OTf)₂ in CH₃CN. The same workup and crystallization as described for 1 led to Fe^II(Me₃TACN)(abt^CF3)(OTf) (2) in good yield (75%). The iPr₃TACN complexes, [Fe^II(iPr₃TACN)(abt)](OTf) (3) (50%) and [Fe^II(iPr₃TACN)(abt^CF3)](OTf) (4) (30%), were prepared in a similar fashion to 1 and 2.

Scheme 2.

Synthesis of Complexes 1 – 4

The molecular structures of 1 – 4 were determined by single-crystal X-ray diffraction (XRD) and are shown together in Figure 2. Selected bond distances and bond angles for 1 – 4 are provided in Table 1. The ferrous center in each complex is facially ligated by the TACN ligand through the neutral N donors. The abt ligands coordinate through the anticipated bidentate bonding mode via the neutral N and anionic S donors. The first coordination spheres of 1 and 2 are saturated by triflate, leading to neutral complexes, whereas 3 and 4 remain 5-coordinate and cationic, with a triflate counterion outside the primary coordination sphere but hydrogen-bonded to the NH₂ group of the abt ligand.

Figure 2.

Displacement ellipsoid plots (50% probability level) for 1 – 4 at 110(2) K. The triflate counterions in 2 and 3, and hydrogen atoms in 1 – 4 (except those on N4) have been omitted for clarity.

Table 1.

Selected Bond Distances (Å) and Angles (deg) for Complexes 1 – 4

	1	2	3	4
Bond Distances
Fe1–S1	2.4352(4)	2.4343(4)	2.3573(4)	2.3753(6)
Fe1–N1	2.225(6)	2.220(5)	2.2818(13)	2.243(2)
Fe1–N2	2.274(6)	2.256(5)	2.1642(12)	2.1920(18)
Fe1–N3	2.263(5)	2.249(5)	2.1651(12)	2.1662(19)
Fe1–N4	2.2344(13)	2.2393(12)	2.2555(13)	2.2656(19)
Fe1–O1	2.1497(11)	2.2190(10)	–––	–––
Bond Angles
S1–Fe1–N1	101.60(16)	100.1(3)	100.82(3)	100.33(5)
S1–Fe1–N2	179.59(18)	179.0(2)	136.97(3)	138.78(5)
S1–Fe1–N3	101.43(15)	100.1(2)	138.07(3)	137.49(5)
S1–Fe1–N4	79.29(4)	79.28(3)	81.30(3)	80.61(5)
S1–Fe1–O1	95.91(3)	95.24(3)	–––––	–––––
N4–Fe1–N1	178.45(17)	179.3(3)	175.10(5)	176.81(8)
N4–Fe1–N2	100.34(17)	101.4(2)	100.88(5)	100.67(7)
N4–Fe1–N3	99.11(17)	101.0(2)	92.93(5)	94.70(7)
N4–Fe1–O1	89.59(5)	88.27(4)	–––	–––
N1–Fe1–O1	91.58(17)	91.5(2)	–––	–––
N2–Fe1–O1	84.26(17)	84.1(2)	–––	–––
N3–Fe1–O1	161.75(16)	163.3(3)	–––	–––
N1–Fe1–N2	78.8(2)	79.2(3)	80.64(5)	80.66(7)
N1–Fe1–N3	79.5(2)	79.3(3)	82.53(5)	82.54(8)
N2–Fe1–N3	78.4(2)	80.5(3)	84.95(5)	83.71(7)
Calculated Values
λocta	1.028	1.027	–––	–––
τ5b	–––	–––	0.62	0.63

^aOctahedral quadratic elongation, ${\text{λ}}_{\text{oct}}=\sum {\left(1_{\text{i}}/1_0\right)}^2/6$. l₀ represents the center to vertex distance of an octahedron with O_h symmetry whose volume is equal to that of the distorted octahedron with distances l_i. λ_oct = 1 for an ideal octahedron.³³

^b5-coordinate geometry index, τ₅ = (β – α)/60. β is the largest bond angle observed, and α is the second largest bond angle.³⁴

Complexes 1 and 2 both contain iron in a pseudo-octahedral geometry, with octahedral quadratic elongation parameters λ_oct = 1.028 and 1.027.³³ The λ_oct parameter is a measure of octahedral distortion, with λ_oct = 1 representing an ideal octahedron, and values greater than 1 reflecting distortions from ideality. The λ_oct values for 1 and 2 are slightly greater than that calculated for the six-coordinate Fe^II center bound to the same TACN derivative in [(Me₃TACN)Fe^II-(μ-OH)-Fe^III(MST)]⁺ (λ_oct = 1.021).³⁵ We hypothesize these distortions are due to the electron-donating thiolate donors in 1 and 2. All metal-ligand bond lengths for 1 and 2 are consistent with high-spin ferrous centers.^{30–31, 36} The Fe–S bonds in 1 and 2 are long compared to those typically observed for mononuclear iron thiolate complexes. A search of the Cambridge Structural Database (CSD)³⁷ reveals that there are only three other iron complexes^38–40 with longer terminal Fe–S bond distances, excluding those with partial thione character, and one of these systems is an Fe(abt) complex containing an oxidized ligand radical.³⁹ The Fe–OTf bond length in 2 is 0.07 Å longer than that seen in 1, which seems opposite to what is expected given the electron-deficient thiolate donor in 2. This observation may be explained by the fact that the triflate ligand in both 1 and 2 forms a hydrogen bond with the nearby coordinated NH₂ moiety. The latter group in 2 is expected to be a better H-bond donor because of the electron-withdrawing CF₃ substituent, leading to a stronger TfO---NH₂ interaction and consequently weakening the Fe–OTf bond. Indeed, we observe both shorter –NH₂---O (1: d(N---O) = 3.147(2) Å; 2: d(N---O) = 3.070(2) Å) and longer S–O distances in 2 compared with 1.

The iPr₃TACN complexes 3 and 4 are both 5-coordinate. For each complex, the unbound triflate counterion is hydrogen-bonded to the NH₂ group of the abt ligand (3: TfO---N = 3.055(2) Å; 4: TfO---N = 3.110(3) Å). Complexes 3 and 4 exhibit geometries between square pyramidal (sp) and trigonal bi-pyramidal (tbp) (τ₅ = 0.62 for 3 and τ₅ = 0.63 for 4; where τ₅ =0.0 for sp and τ₅ =1.0 for tbp).³⁴ Overall, the bond metrics for 3 and 4 are similar to each other. The Fe–N1 bond in 3 is slightly elongated (0.04 Å) compared to the Fe–N1 in 4 (Table 1). This elongation can be rationalized by the anticipated stronger trans-effect from the more electron-rich abt ligand in 3. Other bond lengths are nearly the same for 3 and 4.

UV-vis Spectroscopy.

Complexes 1 – 4 are colorless in solution, but exhibit intense absorbance features in the UV region (Figure 3). Similar features in other arylthiolate-ligated ferrous complexes have typically been assigned as S → Fe(II) charge transfer (CT) transitions based on spectroscopic and computational analyses (UV-vis, MCD, DFT).^41–42 The spectrum for 1 is shown in Figure 3 (black line) and exhibits a peak at λ_max = 269 nm, with a shoulder at 304 nm. In comparison, the CF₃-substituted 2 (blue line) exhibits similar bands which are red-shifted by 17 and 5 nm, respectively. The spectra for the 5-coord complexes 3 – 4 give rise to similar charge transfer bands as seen in 1 – 2, and a similar red-shift is also observed when comparing the CF₃-substituted 4 (272 nm) to unsubstituted 3 (259 nm). The red-shift in the LMCT band upon introduction of an electron-withdrawing group was initially unexpected and opposite to what has been seen for a series of Ni-arylthiolate complexes with varied aryl-group substituents⁴³; however, this trend can be rationalized by a significant contribution of ligand-to-ligand charge transfer in the observed transitions. This hypothesis is supported by the UV-vis spectra of the free thiols (Figure S2), which exhibit a red shift in the CF₃-substituted free ligand, and by time-dependent density functional theory (TD-DFT) calculations (vide infra).

Figure 3.

Electronic absorption spectra of 1 – 4 in CH₃CN at 23 °C.

TD-DFT.

Complexes 1 – 4 were studied by DFT calculations to help explain the trends in the observed electronic absorbance spectra. The optimized geometries of 1 – 4 were calculated using the BP86/6–311G*/6–31G*(C, H atoms) functional/basis set combination and the conductor-like screening model (COSMO) to model CH₃CN solvation. In all cases, the optimized geometries gave structural parameters that matched well with those determined by X-ray crystallography (Table S6).

Spectral excitations were simulated by the TD-DFT method using several functionals of varying Hartree-Fock (HF) exchange (TPSSh 10%, B3LYP 20%, PBE0 25%, CAM-B3LYP 19–65%). These were all paired with an Ahlrichs def2 family triple-ξ basis set (def2-TZVP).^44–45 With the exception of PBE0, all functionals produced satisfactory results when compared to experimental data (Figure S3). However, in the region of interest, experimental features were best reproduced using the CAM-B3LYP range-separated functional. This result is to be expected as CAM-B3LYP extends the standard B3LYP functional to include long-range corrections and has been shown to more accurately estimate charge-transfer excitation energies.⁴⁶

The theoretical and experimental spectra for complexes 1 – 4 are in good agreement; however, a systematic shifting of the calculated spectra by 20 nm is required to match the experimental spectra. Such a systematic overestimation of absolute transition frequency is a well-established problem inherent to TD-DFT calculations.⁴⁷ The high-energy transitions from 250 – 350 nm, in particular, are well reproduced, and match the experimental trends in 1 – 4. These features are characteristic of Fe–S complexes and have been historically assigned as S → Fe(II) charge transfer (CT) bands.^{41–42, 48}

Representative spectra for 3 – 4 are shown in Figure 4. The computed spectrum for each complex is dominated by a feature between 260 – 285 nm (Figures S4 – S7). For each case, this band contains a number of excitations in both the α and β spin manifolds that are close enough in energy to coalesce into one feature. Analysis of the transition difference density plots reveals that the vertical transitions in the region have very similar character across the series of complexes. At higher energies, an α to α excitation is observed from an admixed molecular orbital of S_p and Fe_d character into the π* manifold of the aromatic ring (Figure 4). Given the dominant sulfur character of the donor MO, we assign these transitions as ligand-to-ligand charge transfer (LLCT). Moving across the dominant band to lower energies, we note β to β transitions comprised mostly of S_p donor and Fe_d acceptor character (Figures S4 – S7). While these appear to be characteristic ligand-to-metal charge transfer (LMCT) transitions, the acceptor MOs also contain reduced, but non-negligible contributions from the aromatic ring of the abt ligand. At higher energies (300 – 325 nm), an absorption shoulder is seen in the experimental and simulated spectra for each complex. This feature arises from S_π → Fe(d_xz)/Fe(d_yz) LMCT transitions and does not exhibit a major energy shift upon –CF₃ substitution of the abt ligand.

Figure 4.

TD-DFT computed absorption spectra for 3 (top) and 4 (bottom) (red dashed line) overlaid with experimental spectra (blue line). The black sticks and letters mark the energies and intensities of specific vertical excitations. Electron density difference maps for representative computed transitions are shown with the blue and red regions indicating gain and loss of electron density, respectively.

Analysis of the calculated absorbance spectra for complexes 1 – 4 explains the observed red-shift between compounds 1 – 2 and 3 – 4. For 3 – 4, addition of an electron-withdrawing CF₃ functional group para to the sulfur results in contraction of both the donor (S_p) and acceptor (ligand π*) MOs involved in the high-energy LLCT excitations (Figures S6 – S7). However, by examining the energies for 1 – 4, we conclude that the acceptor MOs are contracted to a greater degree and the overall orbital energy gap is diminished relative to the corresponding transition in 3. Thus, contributions from LLCT transitions, and their corresponding decrease in energy with –CF₃ functionalization, result in the experimentally observed red-shift between complexes 3 and 4. Identical behavior is observed between 1 and 2 (Figures S4 – S5).

NMR Spectroscopy.

Complexes 1 – 4 exhibit paramagnetically shifted peaks between −10 – 140 ppm. The effective magnetic moments of 1 – 4 in CD₃CN were determined by Evans method^49–50 to be 5.37, 5.50, 5.52, and 5.46 μ_B, respectively, consistent with high-spin (S = 2) ferrous complexes. The ¹H NMR spectra of 1 – 2 in CD₃CN are shown in Figure 5. The spectrum of 2 is similar to 1 (Figure 5) with the exception of one less peak near 12 ppm, which can be assigned to the H atom para to the thiolate donor in 1. Other assignments can be made based on integrations and comparison with the selectively deuterated 1–d₉ complex, which has the methyl groups on TACN replaced by CD₃ groups (Figure S8).

Figure 5.

¹H NMR spectra of 1 (top) and 2 (bottom) in CD₃CN at 24 °C. 1,1,1-trifluorotoluene internal standard designated with an asterisk (*).

The ¹H NMR spectra for the 5-coord iPr₃TACN complexes 3 – 4 in CD₃CN at 24 °C are shown in Figures S9. Peaks corresponding to the abt derivatives are observed, but no prominent peaks in either 3 or 4 could be attributed to the iPr₃TACN ligand. However, paramagnetically shifted peaks (−15 – 180 ppm) appear in the spectra for 3 – 4 upon cooling to −40 °C, which can be assigned to iPr₃TACN (Figures S10 – S11). These observations are consistent with significant conformational fluxionality in the iPr₃TACN ligand at 24 °C, leading to severe line-broadening in the ¹H NMR spectra. Similar observations have been made for Fe(iPr₃TACN)(OTf)₂.³¹

The ¹⁹F NMR spectra of 1 – 4 provide insight into the coordination of triflate in solution. All four complexes exhibit a sharp peak near −78 ppm in CD₃CN at 24 °C, indicative of free OTf^- anion (Figures 6, S12 – S14).³⁰ The presence of free OTf^- for 3 – 4 is expected given the lack of triflate coordination in the crystal structures of these complexes. In contrast, peaks for free OTf^- in 1 – 2 implies dissociation of the anion in solution. Further examination at low temperature shows that there is an equilibrium between bound and unbound triflate for 1 – 2. Variable temperature data for 2 are shown in Figure 6. The peak for the CF₃ group shifts and broadens from −25 to −12 ppm, while the sharp peak at −78 ppm assigned to free OTf^- also shifts and broadens upon lowering the temperature to −40 °C. These changes can be attributed to a fast equilibrium between bound and unbound triflate, as has been reported for another high-spin iron(II)-triflate complex.⁵¹ In contrast, the OTf^- peaks in 3 and 4 show no broadening with decreasing temperature (Figure S13 – S14). Changing the NMR solvent from CH₃CN to butyronitrile allowed for data collection for 2 at temperatures down to −100 °C (Figure 7). A sharp peak is observed at −83 ppm for free OTf-, and two additional peaks are now seen at −18 and +61 ppm. These peaks can be assigned to terminal and bridging triflate ligands, respectively, from previous studies on Fe^II(Me₃TACN) triflate complexes.³⁰ The peak for the bridging OTf^- is consistent with a μ–1,3 bonding mode between two Fe^II centers.

Figure 6.

Variable temperature ¹⁹F NMR spectra of 2 in acetonitrile-d₃. 1,1,1-trifluorotoluene internal standard designated with an asterisk (*).

Figure 7.

¹⁹F NMR spectrum of 2 in PrCN at −100 °C. 1,1,1-trifluorotoluene internal standard designated with an asterisk (*).

Integration of the free, terminal, and bridging OTf^- peaks shows that these species exist in a 1:0.1:0.1 ratio at −100 °C. A variable temperature analysis for 1 also shows broadening for OTf^- at lower temperatures, but no resolution of individual species is seen at −100 °C (Figure S15), indicating that the equilibrium for 1 remains in fast exchange. The difference in the exchange rates for 1 and 2 may arise from the stronger H-bonding interaction between OTf^- and the NH₂ group of the abt ligand seen in the solid-state structure for 2 (vide supra). Overall, the variable temperature ¹⁹F NMR data indicate that the OTf^- ligand in 1 and 2 is labile, leading to the presence of either 5-coord or solvent-bound species together with OTf-bound species in solution.

Mössbauer spectroscopy.

The zero-field ⁵⁷Fe Mössbauer spectra for crystalline samples of 1 – 4 dispersed in a boron nitride matrix are shown in Figure 8. Each complex exhibits a single, sharp quadrupole doublet that accounts for >95% total Fe. Fitting of the data gives the Mössbauer parameters listed in Table 2. All spectra are consistent with high-spin (S = 2) ferrous centers.⁵² There is a decrease of ~0.15 mm s⁻¹ in the isomer shift (δ) for the five-coordinate complexes as compared to the six-coordinate complexes, consistent with a lowering in coordination number and a shortening of the average iron-ligand bond lengths (Figure S16). This trend is well-established for high-spin iron(II).⁵²

Figure 8.

Zero-field ⁵⁷Fe Mössbauer spectra for 1 – 4 as crystalline solids dispersed in BN matrix at 5 K. Fits shown in red.

Table 2.

⁵⁷Fe Mössbauer Parameters for 1 – 4 in the Solid State^a,b

Complex	δ (Calcd)c	|ΔEQ| (Calcd)c
1	1.07 (1.06)	3.55 (3.35)
2	1.08 (1.06)	3.20 (3.26)
3	0.93 (0.93)	1.97 (2.34)
4	0.95 (0.92)	2.06 (2.33)

^aCollected at 5 K.

^bAll parameters in mm s⁻¹.

^cCalculated by DFT using B3LYP/CP(PPP)(on Fe)/def2-TZVP/COSMO(MeOH).

The quadrupole splitting is different for 1 – 2, whereas it is the same in 3 and 4. The former difference in |ΔE_Q| can be ascribed to slight differences in OTf^- coordination for 1 and 2. The most significant difference seen in Figure 8 is a decrease in |ΔE_Q| upon changing from 6-coord to 5-coord complexes. Trends in quadrupole splitting values as a function of Fe coordination number are, in general, not well-established, and typically only small differences are noted.^53–55 In a recent report on a bis(thioether)amide-ligated iron complex derived from the abt ligand, it was proposed that five- versus six-coordination induced only a slight difference in |ΔE_Q|.⁵⁶ Thus 1 – 4 provide a unique set of Fe^II complexes in which 6-coord (Fe^II(N₄SO)) versus 5-coord (Fe^II(N₄S)) sites are easily identified by their large and small |ΔE_Q| values, respectively.

The Mössbauer spectra for ⁵⁷Fe-labelled 1 – 4 in different solvents provide information regarding the solution state structures of the series. The spectra for 3 in the solid state versus solution are shown in Figure 9. The spectra are nearly identical, indicating that the 5-coord structure for 3 remains unchanged in solution. The same result is seen for 4 (Figure S18). In contrast, Mössbauer spectra of solutions of 1–⁵⁷Fe and 2–⁵⁷Fe are highly solvent dependent and display more than one quadrupole doublet (Figure 10). The spectra for 1 and 2 in CH₃CN (Figure 10a,d) exhibit two quadrupole doublets, and the doublets with the larger |ΔE_Q| values are similar to the solid-state spectra for 1 and 2. These doublets can be assigned to the 6-coord, OTf-bound structures determined by X-ray crystallography. The narrower doublets seen in CH₃CN correspond to a minor component for 1–⁵⁷Fe and a major component for 2–⁵⁷Fe. These data are consistent with the ¹⁹F NMR spectra for each complex, which indicate an equilibrium between free and bound OTf^- species for both 1 and 2. Thus the narrower doublets can be assigned to the complexes without bound OTf-, corresponding to either a 5-coord species, or a solvent-bound form in which CH₃CN replaces the OTf^- ligand. The Mössbauer spectra for 1 and 2 in butyronitrile also exhibit two doublets (Figure 10b,e), and a third, minor component (~10% of total Fe) is observed for 2–⁵⁷Fe. The three components seen for 2 are in line with the low-temperature ¹⁹F NMR spectrum in Figure 7, which shows peaks for free, terminal, and bridging OTf^- species. The relative amounts of the different components obtained by fitting of the Mössbauer spectrum for 2 corroborates the 10:1:1 ratio seen for the different species by ¹⁹F NMR. The Mössbauer spectra for 1–⁵⁷Fe and 2–⁵⁷Fe in CH₃OH are shown in Figure 10c,f, and are distinctly different than those seen for the nitrile-based solvents. Each spectrum displays a broad doublet that is best fit with two closely overlapping subcomponents (Table 3). As in the case for 1–⁵⁷Fe and 2–⁵⁷Fe in nitrile solvents, one doublet is assigned to an OTf-bound structure, and the other doublet could correspond to either a methanol-bound or 5-coord species.

Figure 9.

Zero-field ⁵⁷Fe Mössbauer spectra of 3–⁵⁷Fe dissolved in CH₃CN (top) and of 3 as a crystalline solid dispersed in BN matrix (bottom) at 5 K. Fits shown in red.

Figure 10.

Zero-field ⁵⁷Fe Mössbauer spectra at 80 K of 1–⁵⁷Fe in (a) MeCN, (b) PrCN (c) MeOH and 2–⁵⁷Fe in (d) MeCN (e) PrCN (f) MeOH. Overall fits are shown in red, and each subspecies fit is shown in green, blue, or purple.

Table 3.

Experimental and Calculated ⁵⁷Fe Mössbauer Parameters for Complexes 1 and 2^a,b

1				2
Solvent	Lc	δ, |ΔEQ| (DFT)d	%	Solvent	Lc	δ, |ΔEQ| (DFT)d	%
MeCN	L = OTf	1.05, 3.57 (1.06, 3.35)	65%	MeCN	L = OTf	1.04, 3.63 (1.06, 3.26)	32%
	L = MeCN	1.05, 2.07 (1.06, 2.48)	35%		L = MeCN	1.05, 2.06 (1.06, 2.91)	68%
PrCN	L = OTf	1.05, 3.63 (1.06, 3.35)	24%	PrCN	L = OTf	1.05, 3.62 (1.06, 3.26)	12%
	L = PrCN	1.04, 1.89 (1.05, 2.96)	76%		L = PrCN	1.05, 1.99 (1.05, 2.89)	77%
	–––	–––		MeOH	L = OTf (bridged)	1.09, 2.85	11%
MeOH	L = MeOH	1.04, 3.80 (1.01, 3.48)	51%		L = MeOH	1.04, 3.80 (1.02, 3.35)	39%
	L = OTf	1.05, 3.49 (1.06, 3.35)	49%	Solid	L = OTf	1.05, 3.56 (1.06, 3.26)	61%
Solid	L = OTf	1.06, 3.63 (1.06, 3.35)	–––		L = OTf	1.06, 3.33 (1.06, 3.26)	–––

^aCollected at 80 K.

^bAll parameters in mm s⁻¹.

^cL represents sixth ligand for [Fe^II(Me₃TACN)(abt^x)(L)]^+/0.

^dCalculated using B3LYP/CP(PPP)(on Fe)/def2-TZVP/COSMO(MeOH).

Mössbauer Spectroscopy: Computational Analysis.

Mössbauer parameters were computed by employing the B3LYP functional in conjunction with the optimized geometries for 1 – 4. The δ and |ΔE_Q| values were determined by empirically derived calibration curves composed from a series of known iron complexes with similar coordination environments to 1 – 4 (Figure S19, Table S7).^57–61 As seen in Table 2, the calculated δ and |ΔE_Q| splitting values for all four complexes in the solid state are in good agreement with the experimental values. We next performed DFT geometry optimizations and Mössbauer parameter calculations for the putative 5-coordinate forms of 1 – 2 (1–5coord and 2–5coord), as well as solvent-bound, 6-coordinate forms of 1 – 2 (1–MeCN, 1–PrCN, 1–MeOH, 2– MeCN, 2–PrCN, and 2–MeOH) (Table S8 – S9). The calculated δ values for 1–5coord and 2–5coord are both 0.85 mm s⁻¹ (Table S10), which does not match well with the observed solution state isomer shifts. In contrast, the calculated δ values for the 6-coord, solvent-bound forms of 1 and 2 are close to the observed δ values in each solvent (Table 3). The calculated |ΔE_Q| values for the nitrile-bound solvent complexes 1–MeCN, 1–PrCN, 2–MeCN, and 2–PrCN are slightly larger than the experimental values; however, these values are smaller than the calculated values for 1–OTf and 2–OTf, following the same trend as what is seen experimentally.

The calculated |ΔE_Q| values for the methanol-bound complexes 1–MeOH and 2–MeOH are a better match with experiment as compared to the nitrile-bound complexes. Taken together, the calculated and experimentally determined Mössbauer parameters in the different solvents confirm that the iron(II) centers in 1 and 2 remain six-coordinate, with equilibrium mixtures of solvent- and OTf-ligated species being observed.

Comparison of Mössbauer Data with CDO.

The Mössbauer spectrum of cys-bound CDO is reported to contain two distinct doublets designated as A and A’ (A: δ = 1.03 mm s⁻¹ |ΔE_Q| = 2.28 mm s⁻¹; A’: δ = 1.10 mm s⁻¹ |ΔE_Q| = 3.14 mm s⁻¹), which were proposed to arise from a mixture of a five-coordinate, N₄S-ligated ferrous center and a six-coordinate N₄SO-ligated ferrous center.¹⁰ DFT calculations of the Mössbauer parameters on a truncated model of the enzyme active site predicted A to be the water-bound form of cys-bound CDO and A’ to be the five-coordinate form based on the trends in the calculated ΔE_Q values. However, the calculated Mössbauer parameters did not accurately reproduce the experimental values. The calculations significantly underestimated the isomer shift for both A and A’. These deviations likely stem from the fact that the previously used calibration method was derived from a limited set of iron complexes with a relatively narrow range in isomer shifts.⁶² Given our ability to accurately calculate the isomer shifts in our model complexes with a new calibration method, we recalculated Mössbauer parameters for A and A’ using the previously published coordinates on a truncated model of the enzyme.¹⁰ Using our new calibration method, the calculated Mössbauer parameters for site A (5-coord) are δ = 1.05 mm s⁻¹, |ΔE_Q| = 2.95 mm s⁻¹ and for site A’ (6-coord) are δ = 1.17 mm s⁻¹, |ΔE_Q| = 3.19 mm s⁻¹. These values are much closer to the experimental values for the two subsites in cys-bound CDO and properly match the experimental trends for isomers shift and quadrupole splitting for the two doublets. The trends seen in the model complexes 1 – 4 further support the assignments of the two species observed in cys-bound CDO. The crystal structures and solidstate Mössbauer spectra of 1 – 4 clearly show distinct differences in the isomer shifts and quadrupole splittings for the five-coord and six-coord N₄S-ligated high-spin ferrous complexes. The δ and |ΔE_Q| values are ~0.9 mm s⁻¹ and ~2 mm s⁻¹, respectively, for the five-coord complexes, and ~1 mm s⁻¹ and >3 mm s⁻¹, respectively, for the six-coord complexes in the solid state. These values are close to those observed for A and A’.

Other mononuclear nonheme iron enzymes, including homoprotocatechuate 2,3-dioxygenase,⁶³ and several carotenoid cleavage oxygenases (CCOs),⁶⁴ display two quadrupole doublets in the presence of substrate. For the CCOs, these two species have been proposed to arise from multiple conformations within the active site. This idea is supported by the fact that the isomer shift for these two species are identical, suggesting similar coordination environments. Furthermore, nitric oxide (NO•) binding studies show that in one of the CCOs, two different {FeNO}⁷ species are present in a ratio equal to the different conformers before introduction of NO•, supporting the existence of two distinct conformations of the active site.⁶⁴ In contrast, EPR studies of the {FeNO}⁷ formed upon introduction of NO• to cys-bound ferrous CDO implies the presence of only one major cys-bound species, suggesting that both A and A’ are ultimately converted to the same {FeNO}⁷ species.⁶⁵ We believe that the data for models 1 – 4, and the observation that A and A’ give the same species with NO•, provides strong evidence that A is the five-coordinate form of cys-bound CDO, while A’ is a six-coordinate, water-bound form (Scheme 1).

O₂ Reactivity.

Reaction of 1 with excess O₂ in MeOH at 23 °C results in the formation of 2-aminophenyl disulfide (Scheme 3), as seen by TLC and ¹H NMR spectroscopy (Figure S20). However, the same reaction conducted at −95 °C yields a deep blue solution, which gradually turns brick red upon warming to 23 °C. TLC analysis indicated a new, major product distinct from disulfide, and ¹H NMR spectroscopy of the crude reaction mixture confirmed the production of a new species (Figure S21). Purification of this product on neutral alumina gave a yellow oil with the ¹H NMR spectrum in Figure 11. This spectrum together with ¹³C and ¹H-¹³C HSQC NMR, as well as EI-MS, identified this product as methyl 2-aminobenzenesulfinate (42 % yield; Figures S21 – S26) (Scheme 3).⁶⁶ The use of isotopically labeled ¹⁸O₂ and analysis by EI-MS shows that one of the O atoms in the product originates from O₂ (Figure S26). Attempts to perform catalytic S-oxygenation using Fe^II(Me₃TACN)(OTf)₂ as catalyst³⁰ and excess O₂ and abt yielded no S-oxygenated products.

Scheme 3.

Reaction of 1 with O₂ in MeOH at 23 °C and −95 °C

Figure 11.

¹H NMR spectrum in CD₃CN of the purified reaction product from the oxygenation of 1 in MeOH at −95 °C.

Reaction of the CF₃-substituted complex 2 with excess O₂ at 23 °C resulted in the formation of disulfide⁶⁷ (Figure S27), as seen with 1. In contrast, as observed for 1, the reaction of complex 2 with excess O₂ at −95 °C leads to S-oxygenation. As shown in Scheme 4, reaction of 2 with O₂ at −95 °C in MeOH leads to a µ-oxo diferric complex (5) with bridging sulfinate ligands. This complex was characterized by single crystal XRD, and the molecular structure is shown in Figure 12. The structure is similar to other Fe^III₂(µ-O)(Me₃TACN)₂(L)₂ (e.g., L = RCO₂-, Cl-, CO₃-) complexes,^68–72 with Fe–µ–O distances = 1.8096(7) and Fe–O–Fe angle = 126.41°. The ESI-MS spectrum of crystalline 5 displays a mono-cation peak with m/z = 1066 and an isotope distribution pattern consistent with [(Fe^III₂(Me₃TACN)₂(abt)₂)] + 5O – 1H + 1OTf^-. Use of ¹⁸O₂ results in the two sets of peaks consistent with the incorporation of both 4 and 5 labelled O atoms (Figure S29). These labelling studies indicate that the sulfinate and bridging oxo ligands are derived from O₂; however, the bridging oxo ligand can exchange with adventitious water, as has been observed in other Fe^III (µ-O) complexes.⁷³ In contrast to 1, esterification of the S-oxygenated sulfinate is not observed. This result supports the conclusions that 1 goes through a sulfinate species en route to forming the methyl ester, and both 1 – 2 react with O₂ to give the S-oxygenated sulfinate product, similar to CDO.

Scheme 4.

Reaction of 2 with O₂ in MeOH at −95 °C

Figure 12.

Displacement ellipsoid plot (50% probability level) for 5 at 110(2) K. The triflate counterions and hydrogen atoms (except those on N4 and N4’) have been omitted for clarity.

In contrast to 1 – 2, Complexes 3 and 4 are unstable in MeOH, as seen by the appearance of at least three quadrupole doublets in the Mössbauer spectra for 3–⁵⁷Fe and 4–⁵⁷Fe dissolved in MeOH. The ¹H NMR spectra of 3 and 4 after dissolution in MeOH reveal the presence of protonated [iPr₃TACNH]⁺ (Figure S30),³¹ suggesting that decay of 3 and 4 results from protonation and decomplexation of the iPr₃TACN ligand in MeOH.

A proposed reaction pathway is shown in Scheme 5. Reaction of 1 – 2 with O₂ should lead to the formation of an Fe^III–superoxo species.¹⁹ Attack of the bound O₂ on the sulfur center initiates S-oxygenation,⁷⁴ giving an Fe(sulfinate) species, which can undergo esterification in MeOH to give the methylsulfinate ester in the case of 1. Esterification of sulfinic acids in the presence of Lewis acids is documented.⁷⁵ The oxo-bridged diferric product seen for 2 is typical for the oxidation of Fe^II by O₂.^76–77 At this time we are not able to determine if the formation of the µ-oxo core occurs either before or after S-oxygenation.

Scheme 5.

Proposed Reaction Pathway for Reaction of 1 – 2 with O₂ in MeOH

Enzyme studies: Reaction of C93G CDO with abt and O₂.

The C93G variant of Rattus norvegicus CDO was prepared and purified as described previously.²⁶Addition of the non-native abt substrate was examined under aerobic conditions in MOPS buffer (pH 7.1) at 25 °C. Analysis by ¹H NMR spectroscopy (Figure 13) shows that 2-aminophenyl disulfide is the major product of abt oxidation by C93G CDO in the presence of O₂. The same reaction was examined in the presence of alcohol (MeOH, EtOH) (20% v/v), reproducing the conditions employed by Pierce,¹⁸ but the NMR data indicate formation of the same product (Figure 13d, e). The ¹H NMR spectrum for abt + H₂O₂ (Figure 13b), which should give disulfide in situ, matches the spectrum from the abt reaction. Extraction of the organic product into CDCl₃ followed by NMR also led to the same disulfide spectrum. Analysis by ESI-MS (positive ion mode) confirms the presence of disulfide [M-H]⁺ 249.0516 m/z. The ESIMS of the reactions containing ROH did show weak signals for the corresponding benzothiazole at [M-H]⁺ 136.0216 m/z (MeOH derivative) and 150.0372 m/z (EtOH derivative) in addition to disulfide, but the intensities of these signals suggest these products form in trace amounts compared to the disulfide.

Figure 13.

¹H NMR spectra measured in MOPS buffer pH 7.1 of (a) abt; (b) abt after reaction with 2 equiv H₂O₂; (c) C93G CDO (10 µM) reacted with abt (5 mM); (d) C93G CDO (10 µM) reacted with abt (5 mM) with 20% MeOH; (e) C93G CDO (10 µM) reacted with abt (5 mM) with 20% EtOH.

C93G CDO Kinetics with abt and O₂.

The steady state rate of C93G-mediated abt oxidation was measured by observing independently the rates of abt depletion by Ellman’s assay⁷⁸ and 2-aminophenyl disulfide formation by detection at 335 nm using a 96-well microplate reader. In both cases, non-enzymatic oxidation was considered. Values of k_obs versus abt concentration are plotted in Figure 14 and show comparable rates to those observed by Pierce et al.,¹⁸ and approximately half the rate with the correct substrate cysteine.²⁶ Linear regression of these data points provides second order rate constants of 47 ± 3 M⁻¹ s⁻¹ for thiol depletion and 24 ± 1 M⁻¹ s⁻¹ for disulfide formation. The fact that depletion is double the rate of product formation further confirms that disulfide is the major product. Both best-fit lines do not intercept zero, but rather approach zero rate at ~0.5 mM. This intercept is unusual but reproducible. We suggest that this “threshold” concentration, where there appears to be little reaction until ~0.5 mM substrate has been added, may indicate a radical type mechanism. Similar reaction profiles with abt^CF3 as substrate were observed, although significantly slower than with abt (8 ± 4 M⁻¹ s⁻¹ for thiol depletion).

Figure 14.

Variation of k_obs with initial abt concentration for the C93G CDO-mediated turnover of abt in MOPS buffer (pH 7.1) monitoring either the depletion of abt (red) or the formation of disulfide (blue). Depletion of abt^CF3 is given in green.

We have observed disulfide formation when presenting cysteine to Pa3MDO from Pseudomona aeruginosa.^{2, 26} Like the present study, this work also yielded kinetics that are first order in both enzyme and thiol. Previously, we indicated that this type of kinetics behavior is similar to that seen for glutathione peroxidases, which also show such a rate equation where formation of the sulfenate is rate determining.⁷⁹ Weak binding of the substrate would promote such reactivity by prematurely releasing substrate following addition of the first O atom and before rearrangement to allow addition of the second O atom.⁷⁴ Indeed, Pa3MDO shows a K_d for cysteine of ~1.3 mM.² It was therefore proposed to investigate substrate binding by Mössbauer spectroscopy.

Binding Studies of C93G CDO with abt.

Mössbauer spectroscopy was used to investigate abt interactions with C93G CDO over a range of abt concentrations (0 – 9.3 mM) under anaerobic conditions, as Mössbauer is very sensitive to the first coordination sphere of iron. Representative Mössbauer spectra collected are presented in Figure 15 and fitted parameters are given in Table S11. Addition of abt leads to a slight decrease in the isomer shift (~0.05 mm s⁻¹), and slight line broadening as compared to the spectrum for the enzyme alone. Changes in the spectra are minor compared to those observed when cysteine is bound and thus rules out S/N coordination by abt. Overall the Mössbauer spectra suggest that abt does not coordinate directly to the iron but may bind weakly within the CDO active site. Incorrect binding of the substrate agrees well with formation of disulfide as the major oxidation product, rather than sulfination.

Figure 15.

Mössbauer spectra of C93G CDO (middle) in the presence of 1 equiv cysteine (top) and 8.8 equiv abt (bottom). Fits are shown as red, blue, or black lines.

Conclusions

This works describes a combined study including both model complexes (Fe^II(N₄S(thiolate)) and the relevant thiol dioxygenase enzyme cysteine dioxygenase. The same non-native substrate, aminobenzenethiolate (abt), was examined for both models and protein. The synthesized series of structurally well-defined ferrous complexes provided spectroscopic data that informed on their solution state versus solid state structures, and in the case of Mössbauer spectroscopy, allowed for direct comparison with data on the protein. A clear pattern in the Mössbauer data was seen for 5-coord versus 6-coord complexes, which provided important information for comparison with overlapping Mössbauer signals observed for substrate-bound CDO. Previous single-turnover investigations on cys-bound CDO revealed two sub-spectra pointing to a mixture of 5- and 6-coord enzyme,¹⁰ but calculations failed to satisfactorily support these assignments. The Mössbauer parameters for the two species seen in cys-bound CDO and C93G CDO (δ = 1.03, |ΔE_Q| = 2.28 and δ = 1.10, |ΔE_Q| = 3.14 mm s⁻¹) agree remarkably well with the parameters for the 5- and 6-coord model complexes. The models allow us to conclude that the 5-coord/6-coord assignments for the enzyme are indeed correct.

The 6-coord ferrous complexes 1 and 2 both react with dioxygen at low temperature to S-oxygenate the non-native abt substrate in a manner similar to the native thiol dioxygenase reaction. The low temperature conditions likely enhance binding of O₂, and prevent autooxidation pathways that could lead to disulfide formation in the models, as opposed to the enzymatic system which has evolved to readily bind O₂ under ambient conditions. Only a few examples of dioxygenation to give a sulfinate product from an Fe^II-thiolate complex have been reported, and complex 5 provides a rare example of a structurally characterized, sulfinate-bound iron product.⁸⁰ In contrast, the same abt derivatives are not S-oxygenated by C93G CDO, but rather give disulfide as the major product. Examination of the addition of abt substrate to C93G CDO reveals that only minor changes occur in the Mössbauer spectrum, consistent with no binding of the substrate directly to the metal center. The inability to directly coordinate is likely due to the differences between abt and the native cys substrate, including the planar phenyl ring in abt, and the lack of a carboxylate substituent seen in cys, which forms a salt bridge with Arg60. These differences likely prevent correct chelation of abt to the iron center, in contrast to the model complexes. Assuming that the reactivity of the models at low temperature is comparable to the enzyme at room temperature, we can conclude that proper S/N chelation in thiol dioxygenases is key for successful S-oxygenation.

Our conclusion that there are both 5-coord, and water-bound 6-coord ferrous sites in cys-bound CDO suggest that the prevailing paradigm for nonheme iron dioxygenases may need revision. Both 5-coord and 6-coord sites rapidly disappear upon reaction with dioxygen,¹⁰ and within 40 ms approximately 75% of the bound cys is converted to cysteine sulfinic acid. This rapid oxidation of substrate could be explained by very fast conversion of the 6-coord sites into the 5-coord sites, thereby allowing all O₂ reactivity to occur through the 5-coord center. However, it is also possible that the H₂O molecule coordinated in the sixth site is relatively weakly bound and can be displaced by dioxygen, allowing for S-oxygenation to occur via both 5-coord and 6-coord centers. The fact that the model complexes 1 – 2 are 6-coord in both the solid-state and in solution, and readily react with O₂ to give sulfinate products, suggests the latter pathway may be a viable mechanism for CDO.

Experimental Section

General Considerations.

All syntheses and manipulations were conducted in an N₂-filled drybox (Vacuum Atmospheres, O₂ < 0.2 ppm, H₂O < 0.5 ppm) or using standard Schlenk techniques under an atmosphere of Ar unless otherwise noted. Me₃TACN was purchased from Matrix Scientific, degassed by three freeze-pump-thaw cycles, and stored over 3 Å molecular sieves. iPr₃TACN was synthesized according to a reported procedure.⁸¹ 2-aminothiophenol was purchased from Alfa Aesar, degassed by three freeze-pump-thaw cycles, and stored over 3 Å molecular sieves. Fe(OTf)₂•2MeCN and ⁵⁷Fe(OTf)₂•2MeCN were prepared according to a literature procedure.^82,57Fe metal (95.93%) was purchased from Cambridge Isotope Laboratories. Formaldehyde–d₂, formic acid–d₂ in D₂O (21% w/w), and ¹⁸O₂ (98 atom %) were purchased from ICON Isotopes (Summit, N.J.). All other reagents were purchased from commercial vendors and used without further purification. Acetonitrile, acetonitrile-d₃, methanol, methanol-d₄, and hexamethyldisiloxane were distilled from CaH₂. Tetrahydrofuran was dried over Na/benzophenone and subsequently distilled. Butyronitrile was distilled from Na₂CO₃/KMnO₄ according to a reported procedure.⁸³ Diethyl ether was obtained from a PureSolv solvent purification system (SPS). All solvents were degassed by a minimum of three freeze-pump-thaw cycles and stored over freshly activated 3 Å molecular sieves in the drybox following distillation.

Instrumentation.

The ¹H and ¹⁹F NMR spectra were measured on a Bruker 300 MHz or a Bruker 400 MHz spectrometer. ²H NMR spectra were recorded with a broad-band coil on a 300 MHz instrument with ²H resonance at 46 MHz. Solution magnetic susceptibilities were determined by Evans’ method.^49–50 Chemical shifts were referenced to reported solvent resonances.⁸⁴ UV–vis experiments were carried out on Agilent 5453 diode-array spectrophotometer using a 1 cm cuvette. Midwest Microlab (Indianapolis, IN) conducted elemental analyses on samples prepared and shipped in ampules sealed under vacuum. Infrared spectra were recorded on a ThermoNicolet Nexus 670 FTIR Spectrometer with an ATR diamond crystal stage using the OMNIC 6.0a software package. Infrared spectra were collected on crystalline solids that were crushed into a fine powder. High resolution EI mass spectra were obtained using a VG70S double-focusing magnetic sector mass spectrometer (VG Analytical, Manchester, UK, now Micromass/Waters) equipped with an MSS data acquisition system (MasCom, Bremen, Germany). ESI mass spectra were acquired using a Finnigan LCQ Duo ion-trap mass spectrometer equipped with an electrospray ionization source (Thermo Finnigan, San Jose, CA. Mössbauer spectra were recorded on a spectrometer from SEE Co. (Edina, MN) operating in the constant acceleration mode in a transmission geometry. The sample was kept in an SVT - 400 cryostat from Janis (Wilmington, MA), using liquid He as a cryogen for 5 K data collection and liquid N₂ as a cryogen for 80 K measurements. Isomer shifts were determined relative to the centroid of the spectrum of a metallic foil of α–Fe collected at room temperature. Data analysis was performed using version F of the program WMOSS (www.wmoss.org) and quadrupole doublets were fit to Lorentzian lineshapes.⁸⁵

DFT Computational Studies.

All geometry optimizations and Mossbauer calculations were performed in the ORCA-3.0.2 program package.⁸⁶ Initial geometries were obtained from X-ray crystallographic models. Optimized geometries were calculated using the BP86^87–88 or the TPSSh functional.⁸⁹ The 6–311g* basis set^90–92 was used for all Fe, N, O, F, P, Cl and S atoms and the 6–31g* basis set^93–94 was used for all C and H atoms. Solvent effects in these calculations were accounted for by using a continuum solvation model (COSMO)⁹⁵ in all cases. Due to SCF convergence difficulties in some cases, damping parameters were altered using the Slowconv function in ORCA. Frequency calculations at the same level of theory confirmed that all optimizations had converged to true minima on the potential energy surface (i.e., no imaginary frequencies). The optimized structures were used for Mössbauer calculations. Mössbauer parameters were computed using the B3LYP^96–98 functional and a combination of CP(PPP)⁶² for Fe and def2-TZVP^{44, 99} for all other atoms. The angular integration grid was set to Grid4 (No-FinalGrid), with increased radial accuracy for the Fe atom (In-tAcc 7). To simulate solid state effects, a continuum solvation model was included (COSMO) with a solvent of intermediate dielectric (methanol), which has been shown to lead to accurate predictions of Mössbauer parameters.¹⁰⁰ The isomer shift was obtained from the electron density at the Fe nucleus, using a linear fit function δ_calc = α(ρ(0) − C) + β. For the methodology described here, α = −0.44024 mm s⁻¹ a.u.³, β = 2.1042 mm s⁻¹, and C = 11813 a.u.³, which was derived by plotting ρ(0) versus the experimental isomer shift of a series of Fe complexes. The calibrated quadrupole splitting was obtained from a linear fit function: |ΔEQ|_calibrated = η(|ΔEQ|_calc) – B₀ with η = 0.84003 B₀ = −0.0019275 mm s^–1, which was derived by plotting |ΔEQ|_calc versus |ΔEQ|_exp.

All TD-DFT calculations were performed using the quantum computing suite ORCA 4.0.¹⁰¹ The electronic absorption spectra were calculated using TD-DFT and a triple-ξ basis set (def2-TZVP)⁹⁹ in combination with functionals of varying Hartree-Fock (HF) exchange (TPSSh 10%,⁸⁹ B3LYP 20%,⁹⁶ PBE0 25%,¹⁰² CAM-B3LYP 19–65%¹⁰³). All calculations employed the resolution of identity approach (RI-J)¹⁰⁴ and chain-of-spheres (COSX)¹⁰⁵ approximation for the Coulomb integral calculations and numerical integration of HF exchange, respectively. The zeroth-order regular approximation (ZORA) was used to model relativistic effects.¹⁰⁶ Solvent effects in these calculations were accounted for by using the conductor-like polarizable continuum model (CPCM) and specifying the dielectric constant (ε) for acetonitrile.¹⁰⁷ At least 50 vertical transitions were calculated for each compound. Simulated absorption spectra were generated using the ORCA spectral creation tool (orca_mapspc) using a constant fixed-width half-maximum line broadening of 2500 cm⁻¹. Each simulated spectrum was scaled by a factor of 0.8 to allow for better comparison with experimental data. Computed excitations were individually analyzed by generating and visualizing transition difference density plots using the ORCA orbital plotting tool (orca_plot) and the UCSF Chimera extensible molecular modeling program, respectively.

Synthesis of Me₃TACN–d₉.

Synthesis of Me₃TACN–d₉ was performed according to a modified version of a previously reported procedure.¹⁰⁸ H₃TACN (1.050 g, 8.127 mmol) was dissolved in D₂O (2 mL). Excess formaldehyde–d₂ (7 g, 150 mmol) and formic acid–d₂ in D₂O (21% w/w) (10 g, 70 mmol) were added to the H₃TACN solution resulting in a color change to orange. The solution was refluxed for 25 h. The orange solution was cooled to 0 °C and the pH was slowly raised to >10 by dropwise addition of aqueous NaOH (15 M). The resulting slurry was then extracted with CHCl₃ (~100 mL) three times and all the organics were combined and dried over MgSO₄, filtered, and dried under vacuum, yielding a pale yellow oil (1.107 g, 6.13 mmol, 75%), which was then degassed and stored over molecular sieves (3 Å) prior to use.

[Fe^II(Me₃TACN)(abt)(OTf)] (1).

To a solution of Me₃TACN (155 mg, 0.903 mmol) in CH₃CN (~5 mL) was added Fe(OTf)₂•2 CH₃CN (394 mg, 0.903 mmol) to afford the previously reported purple species, [Fe^II(Me₃TACN)(CH₃CN)₃](OTf)₂.³⁰ In a separate vial, trimethylamine (126 µL, 0.903 mmol) was added dropwise to a solution of 2-aminothiophenol (97 µL, 0.90 mmol) in THF (~5 mL) and stirred for 5 min. The resulting pale yellow solution was added dropwise to the solution containing [Fe^II(Me₃TACN)(CH₃CN)₃](OTf)₂, resulting in formation of a yellow slurry. The reaction was stirred for 1 h before filtering through a bed of celite and removing the solvent under reduced pressure. To the resulting pale yellow residue was added THF (~3 mL), followed by minimal CH₃CN to dissolve the solid. Vapor diffusion of Et₂O afforded 1 as colorless crystals after 12 h (360 mg, 80%). UV-vis (CH₃CN): λ_max = 268 nm (ε = 11000 M⁻¹ cm⁻¹). Selected IR bands, ν (cm⁻¹): 3232, 3122, 2992, 2876, 1600, 1467, 1301, 1210 (s, ν_OTf), 1011(s, ν_OTf), 747, 637 (s, ν_OTf). ¹H NMR: (CD₃CN, 400 MHz): δ 133.65, 56.83, 39.08, 37.21, 17.87, 14.30, 12.06 ppm. ¹⁹F NMR: (CD₃CN, 300 MHz): δ −77.18 ppm. Evans method (CD₃CN, 400 MHz): μ_eff = 5.37 μ_B. Anal. Calcd for C₁₆H₂₇N₄F3O₃S₂Fe: C, 38.41; H, 5.44; N, 11.20. Found: C, 38.44; H, 5.51; N, 11.24. [Fe^II(Me₃TACN–d₉)(abt)(OTf)] (1–d₉) was synthesized using the same procedure as that for 1 using Me₃TACN–d₉.

[Fe^II(Me₃TACN)(abt^CF3)(OTf)] (2).

To a solution of Me₃TACN (132 mg, 0.767 mmol) in CH₃CN (~5 mL) was added Fe(OTf)₂•2CH₃CN (335 mg, 0.767 mmol) to afford [Fe^II(Me₃TACN)(CH₃CN)₃](OTf)₂.³⁰ In a separate vial, triethylamine (214 µL, 1.54 mmol) was added dropwise to a solution to 2-amino-4-(trifluoromethyl)benzenethiol hydrochloride (176 mg, 0.767 mmol) in THF (~5 mL) and stirred for five min. The yellow slurry was filtered through celite to produce a pale yellow solution, which was then added to the solution containing [Fe^II(Me₃TACN)(CH₃CN)₃](OTf)₂ to afford a yellow solution. The reaction was stirred for 1 h before filtering through a bed of celite and removing the solvent under reduced pressure. To the resulting pale yellow residue was added THF (~3 mL) followed by minimal CH₃CN to dissolve the solid. Vapor diffusion of Et₂O afforded colorless crystals after 12 h (326 mg, 75%). UV-vis (CH₃CN): λ_max = 286 nm (ε = 10400 M⁻¹ cm⁻¹). Selected IR bands, ν (cm⁻¹): 3317, 3223, 2891, 1610, 1462, 1327, 1282, 1232, 1157, 1221, 1107, 1080 (s, ν_OTf), 1012(s, ν_OTf), 890, 831, 635 (s, ν_OTf). ¹H NMR: (CD₃CN, 400 MHz): δ 133.34, 55.46, 41.74, 40.21, 17.02, 14.74 ppm. ¹⁹F NMR: (CD₃CN, 300 MHz): δ −75.03, −23.97 ppm. Evans method (CD₃CN, 400 MHz): μ_eff = 5.50 μ_B. Anal. Calcd for C₁₇H₂₆N₄F₆O₃S₂Fe: C, 35.92; H, 4.61; N, 9.86. Found: C, 35.91; H, 4.62; N, 9.70.

[Fe^II(iPr₃TACN)(abt)](OTf) (3).

A solution of iPr₃TACN (151 mg; 0.589 mmol) in THF (~1 mL) was added to a solution of Fe(OTf)₂•2CH₃CN (257 mg, 0.589 mmol) dissolved in CH₃CN (~3 mL) and stirred for 1 h to afford the previously reported Fe^II(iPr₃TACN)(OTf)₂.³¹ In a separate vial, 2-aminobenzenethiol (74 mg, 0.59 mmol) was stirred with Et3N (82.0 μL, 0.589 mmol) in THF (~2 mL) for 30 min. The resulting solution of 2-aminobenzenethiolate was added to Fe^II(iPr₃TACN)(OTf)₂, and an immediate color change from peach to dark brown was observed along with visible precipitate formation. The reaction was stirred for 12 h and the resulting solution was filtered through celite before removing the volatiles under reduced pressure. The dark brown oily residue was redissolved in a mixture of 1:1 CH₃CN/THF (~2 mL). Vapor diffusion of Et₂O afforded 3 as colorless blocks (177 mg, 51%). UV-vis (CH₃CN): λ_max = 259 nm (ε = 11000 M⁻¹ cm⁻¹). Selected IR bands, ν (cm⁻¹). 3288, 3244, 2970, 1599, 1470, 1382, 1258, 1257, 1144, 1221, 1072 (s, ν_OTf), 1026 (s, ν_OTf), 759, 720, 621 (s, ν_OTf) ¹H NMR: (CD₃CN, 400 MHz): δ 40.69, 30.37, −0.61 ppm. ¹⁹F NMR: (CD₃CN, 300 MHz): δ −79.47 ppm. Evans method (CD₃CN, 400 MHz): μ_eff =5.46 μ_B. Anal. Calcd for C₂₂H₃₉N₄O₃S₂F₃Fe: C, 45.21; H, 6.73; N, 9.58. Found: C, 45.52; H, 6.78; N, 9.57.

[Fe^II(iPr₃TACN)(abt^CF3)](OTf) (4).

A solution of iPr₃TACN (138 mg, 0.540 mmol) in THF (~1 mL) was added to a solution of Fe(OTf)₂•2CH₃CN (191 mg, 0.540 mmol) dissolved in CH₃CN (~3 mL) and stirred for 1 h to afford the previously reported species Fe^II(iPr₃TACN)(OTf)₂.³¹ In a separate vial, triethylamine (150.6 µL, 1.080 mmol) was added dropwise to a solution of 2-amino-4-(trifluoromethyl)benzenethiol hydrochloride in THF (~3 mL) and stirred for 30 min. The slurry was filtered through celite to remove the precipitate. The resulting pale yellow solution of 2-amino-4-(trifluoromethyl)benzenethiolate was added to the solution containing [Fe^II(iPr₃TACN)(OTf)₂] and an immediate color change from pale yellow to red was observed. The reaction mixture was stirred for 12 h and the resulting solution was filtered through celite before removing the volatiles under reduced pressure. The residue was redissolved in a mixture of 1:1 CH₃CN/THF (~2 mL). Vapor diffusion of Et₂O afforded colorless blocks of 4 (103 mg, 30%). UV-vis (CH₃CN): λ_max = 272 nm (ε = 9500 M⁻¹ cm⁻¹). Selected IR bands, ν (cm⁻¹). 3261, 3138, 2985, 1606, 1495, 1380, 1324, 1273, 1248, 1159, 1075, 1068 (s, ν_OTf), 1026 (s, ν_OTf), 964, 894, 721, 636 (s, ν_OTf). ¹H NMR: (CD₃CN, 400 MHz): δ 137.75 (br), 45.24, 28.47, 17.97, −28.21(br) ppm. ¹⁹F NMR: (CD₃CN, 300 MHz): δ −78.87, −27.89 ppm. Evans method (CD₃CN, 400 MHz): μ_eff = 5.52 μ_B. Anal. Calcd for C₂₃H₃₈N₄O₃S₂F₆Fe: C, 42.34; H, 5.87; N, 8.59. Found: C, 42.05; H, 6.04; N, 8.64.

Reaction of 1 with excess O₂ at 23 °C.

An excess amount of O₂ gas was bubbled vigorously through a solution of 1 (19 mg, 38 μmol) in MeOH (10 mL) for ~1 min. The colorless solution rapidly turned blue then yellow/brown within 5 s. The solvent was removed and the crude product was dissolved in CD₃CN for analysis by ¹H NMR spectroscopy. The ¹H NMR spectrum exhibited peaks corresponding to 2-aminophenyl disulfide at δ = 7.14 (td, 1H), 7.05 (dd, 1H), 6.75 (dd, 2H), 6.51 (td, 1H), and 4.72 (s, br, 2H) ppm.

Reaction of 1 with excess O₂ at −95 °C.

A solution of 1 (59 mg, 0.12 mmol) in MeOH (10 mL) was cooled to −95 °C, and O₂ gas was added by vigorous bubbling through the solution. A rapid color change from colorless to dark blue was noted. Further bubbling of O₂ was continued for 15 min at −95 °C, followed by gradual warming to 23 °C. A color change from blue to red/brown occurred during warming. The solvent was removed under vacuum to give a red oil which was dissolved in CD₃CN for analysis by ¹H NMR spectroscopy. Peaks observed at δ = 7.38 (d, 1H), 7.29 (t, 1H), 6.78 (t, 2H), 5.09 (s, br, 2H), and 3.53 (s, 3H) ppm corresponded to methyl 2-aminobenzenesulfinate. The organic product was purified on neutral alumina (100% EtOAc). The solvent was removed yielding pure methyl 2-aminobenzenesulfinate as a yellow oil. Quantification by comparison of ¹H NMR integrations with a standard (1,1,1-trifluorotoluene) indicated that methyl 2-aminobenzenesulfinate was formed in 42% yield. ¹H NMR: (CD₃CN, 400 MHz): δ = 7.38 (d, 1H), 7.29 (t, 1H), 6.78 (t, 2H), 5.09 (s, br, 2H), and 3.53 (s, 3H) ppm. ¹H NMR: (CDCl₃, 400 MHz): δ = 7.39 (d, 1H), 7.28 (t, 1H), 6.81 (t, 1H), 6.68 (d, 1H) 4.92 (s, br, 2H), and 3.61 (s, 3H) ppm. EI-MS (m/z): 171 (M⁺), 140 {(M – OCH₃)}⁺.

Reaction of 2 with excess O₂ at −95 °C.

A solution of 2 (63 mg, 0.11 mmol) in MeOH (5 mL) was cooled to −95 °C, and O₂ gas was added by vigorous bubbling through the solution. A rapid color change from colorless to dark blue/purple was noted. Further bubbling of O₂ was continued for 30 min at −95 °C, followed by gradual warming to 23 °C. A color change from blue/purple to red/brown occurred during warming. The solvent was removed under vacuum and the resulting red oil was washed with CH₂Cl₂ to remove organic byproducts. The red solid was dissolved in CH₃CN and layered with diethyl ether affording red crystals of 5 over 24 h. (24 mg, 36%). ¹H NMR: (CD₃CN, 400 MHz): δ = 26.0 (br), 18.2 (br), 15.1 (br), 9.4 (s), and 8.4 (s) ppm. ¹⁹F NMR: (CD₃CN, 300 MHz): δ −77.40, −60.73 ppm.

Protein purification and sample preparation.

The C93G variant of CDO was expressed and purified using the Strep-tag® affinity purification system as described previously.¹⁷ C93G purity was measured to be >95% as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. C93G contained 5% endogenously bound iron as determined by a ferrozine assay.¹⁰⁹ Purified protein was concentrated and used as appropriate.

Product distribution of C93G CDO catalyzed reaction.

The products produced through reaction of abt with enzyme were assessed qualitatively by NMR spectroscopy and mass spectrometry. ¹H NMR experiments were performed using a 600 MHz Bruker Avance III spectrometer using excitation sculpting for water suppression (zgespg) rather than presaturation as used previously.¹¹⁰ C93G (10µM) in the absence and presence of ethanol and methanol (20% v/v) in MOPS buffer (100 mM, pH 7.1) were mixed with abt (5 mM) to give a total volume of 1 mL. The reaction mixture was kept at 25 °C and stirred constantly for 3 hours. Aliquots (700 μL) of each reaction were transferred into NMR tubes with a D₂O capillary for ¹H NMR analysis at 25 °C. Following analysis, the final reaction mixture (200 μL) was added to ice-cold acetonitrile (800 μL) to precipitate the protein and the supernatant injected into a Thermo Exactive Plus Orbitrap mass spectrometer equipped with high voltage electrospray ionization source (H-ESI) for high resolution and accurate mass analysis. The spray voltage, temperature of ion transfer tube and S-lens of the mass spectrometer were set at 3.8 kV, 250 °C and 50%, respectively. The full MS scans were acquired at a resolving power of 140,000 at m/z 200, an auto gain control (AGC) target value of 3.0 × 10⁶ and a maximum injection time of 200 ms.

abt; ¹H NMR (H₂O, 600 MHz): δ = 7.27 (d, J = 8.1 Hz, 1H), 6.80 (t, J = 7.6 Hz, 1H), 6.71 (d, J = 7.8 Hz, 1H), 6.58 (t, J = 7.9 Hz, 1H) ppm. abt reacted with 2 equivalents H₂O₂ in MOPS (100mM, pH 7.1); ¹H NMR (H₂O, 600 MHz): δ = 7.22 (t, J = 8.2 Hz, 2H), 7.04 (d, J = 8.2 Hz, 2H), 6.87 (d, J = 8.6 Hz, 2H), 6.63 (t, J = 7.9 Hz, 2H) ppm. Sample extracted into CDCl₃; ¹H NMR (CDCl₃, 600 MHz): δ = 7.15 (ddd, J = 4.7, 3.8, 1.4 Hz, 4H), 6.71 (dd, J = 7.8, 0.7 Hz, 2H), 6.58 (td, J = 7.5, 1.3 Hz, 2H) ppm. abt reacted with C93G; ¹H NMR (H₂O, 600 MHz): δ = 7.20 (t, J = 8.0 Hz, 1H), 7.03 (d, J = 8.2 Hz, 1H), 6.85 (d, J = 8.3 Hz, 1H), 6.62 (t, J = 7.9 Hz, 1H) ppm. MS: major product: m/z 249.0514 [M-H]⁺ abt reacted with C93G in 20 v/v methanol; ¹H NMR (H₂O, 600 MHz): δ = 7.20 (t, J = 8.2 Hz, 1H), 7.03 (d, J = 8.3 Hz, 1H), 6.85 (d, J = 8.5 Hz, 1H), 6.62 (t, J = 7.9 Hz, 1H) ppm. MS- major product: 2-aminophenyl disulfide m/z 249.0514 [M-H]⁺; minor product: benzothiazole m/z 136.0216 [M-H]⁺. abt reacted with C93G in 20 v/v ethanol; ¹H NMR (H₂O, 600 MHz): δ = 7.20 (t, J = 8.0 Hz, 1H), 7.02 (dd, J = 12.9, 8.9 Hz, 1H), 6.84 (dd, J = 12.6, 8.4 Hz, 1H), 6.65 – 6.58 (m, 1H) ppm. MS- major product: 2-aminophenyl disulfide m/z 249.0514 [M-H]⁺; minor product: benzothiazole m/z 150.0372 [M-H]⁺.

Enzyme kinetics.

Kinetics of abt turnover were followed either via an Ellman’s based assay⁷⁸ to follow thiol disappearance or direct UV-vis to follow product formation. UV-vis measurements were performed using a Clariostar Monochromator Microplate Reader (BMG LABTECH), a Varian Cary 50 Bio spectrophotometer (Agilent Technologies) or a stopped-flow apparatus (SX-20MV, Applied Photophysics).

The extinction coefficient for abt was measured under anaerobic conditions as abt oxidizes readily at higher pH. To overcome this, various concentrations of abt (200–2000 μM) were rapidly mixed with MOPS buffer in a stopped-flow apparatus to give a final pH of 7.1. Initial spectra were averaged and used to calculate the extinction coefficients between 250 and 500 nm (Figure S31). 2-aminophenyl disulfide was produced by reaction of varying concentrations of abt with two equivalents of hydrogen peroxide. UV-vis spectra of the resulting solutions allowed equivalent extinction coefficients between 250 and 500 nm to be calculated (Figure S31).

The rate of 2-aminophenyl disulfide formation was measured at 335 nm, where the largest difference in absorbance between product and reactant was observed. It was found that abt was sensitive to photolysis and so measurements were carried out in a 96-well plate to decrease the pathlength and render photolysis negligible. Abt was also sensitive to non-enzymatic oxidation at the pHs under study and so control experiments in the absence of enzyme allowed the enzymatic rate to be calculated. A pH profile was performed to determine the optimal pH for detection of C93G activity (Figure S36), which was found to be pH 7.1, the maximum activity of C93G with cysteine.²⁶ Typical experiments reacted C93G (2–10 μM) with abt (0–8.5 mM) in MOPS buffer (100 mM, pH 7.1). Absorbance-time profiles were converted to concentration-time profiles and the steadystate rates determined through linear regression of the initial linear section of the curve (approximately first 8 min). Division by the enzyme concentration allowed k_obs (s⁻¹) to be calculated.

Mössbauer spectroscopy of C93G CDO with abt.

Mössbauer spectroscopy of ⁵⁷Fe bound C93G (~1.2 mM) was carried out under anaerobic conditions as previously described. Cysteine and a range of 2-aminothiophenol solutions were made and added anaerobically to purified C93G in Tris buffer (100 mM, pH 8.1) and frozen anaerobically in liquid nitrogen. Spectra were measured on a Mössbauer spectrometer (Science Engineering & Education Co., MN) equipped with a closed-cycle refrigerator system from Janis Research Co. and SHI (Sumi-tomo Heavy Industries Ltd.) and a temperature controller from Lakeshore Cryotronics, Inc. Data were collected in constant acceleration mode in transmission geometry with an applied field of 47 mT parallel to the γ-rays.