Sulfur K-edge X-ray absorption spectroscopy and density functional theory calculations on monooxo Mo(IV) and bisoxo Mo(VI) bis-dithiolenes: insights into the mechanism of oxo transfer in sulfite oxidase and its relation to the mechanism of DMSO reductase.

Sulfur K-edge X-ray absorption spectroscopy (XAS) and density functional theory (DFT) calculations have been used to determine the electronic structures of two complexes [Mo(IV)O(bdt)2](2-) and [Mo(VI)O2(bdt)2](2-) (bdt = benzene-1,2-dithiolate(2-)) that relate to the reduced and oxidized forms of sulfite oxidase (SO). These are compared with those of previously studied dimethyl sulfoxide reductase (DMSOr) models. DFT calculations supported by the data are extended to evaluate the reaction coordinate for oxo transfer to a phosphite ester substrate. Three possible transition states are found with the one at lowest energy, stabilized by a P-S interaction, in good agreement with experimental kinetics data. Comparison of both oxo transfer reactions shows that in DMSOr, where the oxo is transferred from the substrate to the metal ion, the oxo transfer induces electron transfer, while in SO, where the oxo transfer is from the metal site to the substrate, the electron transfer initiates oxo transfer. This difference in reactivity is related to the difference in frontier molecular orbitals (FMO) of the metal-oxo and substrate-oxo bonds. Finally, these experimentally related calculations are extended to oxo transfer by sulfite oxidase. The presence of only one dithiolene at the enzyme active site selectively activates the equatorial oxo for transfer, and allows facile structural reorganization during turnover.

Introduction

Molybdenum is the one seconpan>d row trapan> class="Chemical">nsition metal that has been found in almost all forms of life, including microorganisms, plants and animals.[1] It is part of a multinuclear active site in the FeMoco cofactor in nitrogenase,[2,3] while most other enzymes have a mononuclear center, which is redox-active between the MoIV and MoVI oxidation states and has pyranopterin-dithiolene ligands.[4−7] The MoV oxidation state is not directly involved in the oxo transfer reaction, but is involved as enzymes and models cycle between MoIV (d2) and MoVI (d0) oxidation states.[8] These mononuclear molybdenum enzymes have been divided into three families: dimethyl sulfoxide (DMSO) reductases, sulfite oxidases, and xanthine oxidases.[6] DMSO reductase and sulfite oxidase catalyze oxo-transfer reactions. DMSO reductase reduces DMSO to DMS (dimethyl sulfide), with oxo transfer to the Mo center, which converts from a MoIV desoxo (i.e., no oxo ligand, but a with an alkoxide–Mo bond) to a MoVI monooxo species. Sulfite oxidase (SO) transfers an oxo ligand to substrates, converting the Mo center from a MoVI bisoxo to a MoIV monooxo species.[5]

Model complexpan> class="Chemical">es have been synthesized and characterized that mimic either the structure and/or activity for these enzyme active sites. These have greatly assisted in developing an understanding of the mechanisms of oxo transfer to MoIV and oxo transfer from MoVIO2.[9] A number of experimental and theoretical studies have focused on DMSO reductase reactions to understand the geometric and electronic structural contributions to function.[10−15] Our previous study of [MoIV(OSi)(bdt)2]− and [MoVIO(OSi)(bdt)2]− complexes (OSi = [OSiPh2Bu]−, bdt = benzene-1,2-dithiolate(2−), Figure 1B,D), provided the following insight into the nature of the DMSO oxo transfer reaction:[10] (1) the dithiolene ligands are innocent (i.e., not redox-active), but are strong π-donors that stabilize the singlet ground state of the MoIV in the absence of a strong oxo ligand; and (2) in the oxo transfer from the DMSO to the MoIV center, the S–O bond elongation polarizes the bond, and decreases the energy gap between frontier molecular orbitals (FMOs, the LUMO on the substrate and the HOMO occupied d-orbital on the MoIV), allowing the transfer of the electron pair from Mo to the substrate to complete the oxo transfer.

Figure 1

Crystal and geometry optimized structures of the complexes in this study. Bond distances in angstroms (Å) are listed below the structure. Mo–S* indicates the sulfur atom that is trans-axial to the oxo ligand. In panel D, similar Mo–S bond distances were averaged.

There have also been a number of sulfite oxidase model studipan> class="Chemical">es. Holm and co-workers have synthesized a model that matched the native enzyme structure well, but there was no reported reactivity.[16] The Sarkar group measured the kinetics of oxo transfer from [MoVIO2(mnt)2]2– (mnt = maleonitriledithiolate(2−)) to HSO3– and found Michaelis–Menten kinetics.[17] Other model studies focused on the transfer of an oxo group from a MoVIO2 species to phosphine substrates. Hall and co-workers studied the reaction: [MoVIO2(tBuL-NS)2] + PMe3 → [MoIVO(tBuL-NS)2] + OPMe3 computationally[18] based on the experimental data of Holm an co-workers on this reaction.[19] The tBuL-NS ligand had nonconjugated nitrogen and sulfur groups bound to the Mo. Basu and co-workers[20] studied the reaction: [LiPrMoO2(OPh)] + PMe3 → [LiPrMoO(OPh)-Solvent] + O=PMe3 both experimentally and computationally, where LiPr is a tridentate ligand with nitrogens bound to Mo.[20] Neither study utilized the dithiolene ligation present in the enzymes. There are also studies to evaluate electronic structures of model complexes that do not have oxo transfer reactivity, but contain dithiolene ligation.[21,22] The Holm group has synthesized a series of MoIVO and MoVIO2 bis-dithiolene complexes, characterized their geometric structures and showed that they undergo oxo transfer to phosphite esters with well determined kinetics parameters for the reaction[23]

The focus of the pan> class="Chemical">current study is to use spectroscopic and computational methods to determine the electronic structure of these Mo dithiolene complexes, and use experimentally calibrated calculations to evaluate this oxo transfer reaction. The goal is to understand this and the other oxo transfers to substrate on a molecular level and to compare these to the reaction coordinate obtained for DMSO reductase, which involves oxo transfer from substrate to MoIV.

In our previous study of the DMSO redupan> class="Chemical">ctase reaction, S K-edge X-ray absorption spectroscopy (XAS) was used to experimentally determine the electronic structures of the dithiolene coordinated Mo centers.[10] The transition of a S 1s electron into the low-lying unoccupied valence orbitals, which have some S 3p character mixed into predominantly metal d-orbitals, leads to pre-edge features in XAS. The energies of the pre-edge transitions reflect the energies of these unoccupied Mo d-orbitals, which depend on the effective nuclear charge (Zeff) of the metal and the ligand field strength.[24] Thus, changing from MoIV to MoVI will shift the pre-edge to lower energy, while binding a strong (i.e., oxo) ligand will shift the dσ and dπ orbitals to higher energy. Of particular utility is that the intensity of a pre-edge transition is proportional to the amount of S 3p character mixed into the unoccupied metal d orbitals due to covalent bonding. Thus, the metal–sulfur bond covalencies can be measured experimentally.

The intensity of a pre-edge peak D0 is given bywhere α2 is the bonpan>d covalency (i.e., sulfur p character mixed into a metal d orbital), N is the total number of sulfurs bound to the metal, h is the total number of d electron holes, and Is is the electric dipole integral, which is dependent on the Zeff of the S ligand and has been experimentally determined in reference.[25] Density functional theory (DFT) calculations supported by the S K-edge XAS data can then provide further insight into the bonding and be used to explore the reactivity.[26]

In this study, S K-edge XAS is used to experimentally determine the electronpan>ipan> class="Chemical">c structures of the dithiolene complexes [MoIVO(bdt)2]2– and [MoVIO2(bdt)2]2– (Figure 1A,C, bdt = benzene-1,2-dithiolate(2−)) that model the oxo atom transfer reaction of SO. These data are correlated to DFT calculations, and the calculations are used to evaluate oxo transfer from MoVIO22– to trimethylphosphite substrate. The nature of this reaction is then compared to our previous results[10] on oxo transfer from DMSO to a MoIV dithiolene complex. This study provides new insight into these oxo transfer reactions, which are found to be fundamentally different. The DMSO reductase reaction involves a late transition state where oxo transfer leads to electron transfer, while the sulfite oxidase reaction has an early transition state with electron transfer inducing the oxo transfer. This study further considers the intrinsic barriers of both sets of structures for both reactions and is extended to consider the active site reactivity of SO.

Experimental Section

Sample Preparation

The [MoIVO(bdt)2]2– and [MoVIO2(bdt)2]2– complexes were prepared as in the literature.[27]

S K-Edge XAS Data Collection and Analysis

All sulfur K-edge XAS data were measured at the Stanpan>ford Synpan> class="Chemical">chrotron Radiation Lightsource (SSRL) using the 54-pole wiggler beamline 6–2 operating in high field mode of 10 kG with a Ni-coated harmonic rejection mirror and a fully tuned Si(111) crystal monochromator. Details of the beamline configuration are given in ref (25). The solid samples were ground into fine powders in an inert atmosphere (N2) dry glovebox where the O2 level was less than 1 ppm and dispersed as thinly as possible on Mylar tape to minimize potential self-absorption. A 6 μm-thick, sulfur-free polypropylene front window was used to prevent sample exposure to air upon mounting in the sample box. The photon energy was calibrated to the maximum of the first pre-edge feature of Na2S2O3·5H2O at 2472.02 eV. A total of 3–5 scans were measured for each sample to ensure reproducibility. Raw data were calibrated and averaged using MAVE in the EXAFSPAK software package.[28] With the use of the PySpline program,[29] the background was removed from all spectra by fitting a second-order polynomial to the pre-edge region and subtracting it from the entire spectrum. Normalization of the data was accomplished by fitting a flat second-order polynomial or straight line to the post-edge region and normalizing the edge jump to 1.0 at 2490.0 eV.[24] Since the post-edge region also contains intensity from the Mo L3-edge, the data were further scaled such that the region between the S K-edge and Mo L3-edge had the same intensity as the corresponding region of a tungsten dithiolene complex.[10] The error from background subtraction and normalization is less than 1%. Intensities of the pre-edge features were obtained by fitting the spectra with peaks having pseudo-Voigt line shape peaks with 1:1 Lorentzian to Gaussian contributions, using the EDG_FIT program.[28] The reported intensity values were based on the average of 10–12 good fits. The intensities of the S 1s → C–S π*/σ* transitions at ∼2473.8 eV were based on previously published values for similar complexes,[10] where these edge features were better resolved, allowing for ±10% variation of the intensities and peak widths. The error from the fitting procedure was less than 10%. The fitted intensities were converted to %S 3p character using eq 1, where D0 is the peak area and the value of Is was set at 14.1,[10] the same as previously published results for similar complexes.

DFT Calculations

DFT calpan> class="Chemical">culations were performed using the Gaussian 09 package[30] with both the pure functional BP86 (Becke GGA exchange with Perdew 1986 nonlocal correlation[31,32]) and the hybrid functional B3LYP (Becke GGA exchange including 20% Hartree–Fock mixing and Lee, Yang, and Parr correlation effects[33]), using the SDD basis set (triple-ζ quality with effective core potential) on the Mo atom,[34] 6-311G(d) basis set on the S, O, N and P atoms and 6-31G(d) on C and H atoms. Both functionals gave similar results, and only the B3LYP results are reported here (BP86 results given in Table S1). The geometry optimizations[35] were performed starting with the published crystal structures. Frequency calculations showed no imaginary frequencies for both reactants and products. All calculations used the polarized continuum model (PCM)[36] in acetonitrile. Mulliken populations and Mayer bond orders were obtained using the QMForge program, which makes extensive use of the cclib library.[37,38] For reaction coordinate studies, intrinsic reaction coordinate (IRC) calculations were run after the frequency calculation at the transition state (only 1 imaginary frequency) to ensure that the pathway connects the transition state to both the reactant and product.

Results and Analysis

Electronic Structures

S K-Edge XAS Data

The sulfur K-edge XAS spepan> class="Chemical">ctra for the monooxo MoIV and bisoxo MoVI bis-dithiolene complexes are shown in Figure 2. The spectra of the [MoIV(OSi)(bdt)2]− and [MoVIO(OSi)(bdt)2]− complexes are also included for reference. The [MoIVO(bdt)2]2– complex (Figure 2A) has a dominant feature at ∼2473.1 eV. There is also a low energy shoulder at ∼2472.4 eV. On the basis of the second derivative of the spectrum (Figure S1A), there is an additional peak at ∼2473.7 eV below the rising edge. Compared to the spectrum of the [MoIV(OSi)(bdt)2]− complex (Figure 2B), which requires a total of four peaks below the edge for a good fit, the three pre-edge features in the [MoIVO(bdt)2]2– complex are shifted to higher energy due to the presence of the strong oxo ligand, which destabilizes the energies of the unoccupied Mo d orbitals.

Figure 2

S K-edge XAS spectra of (A) [MoIVO(bdt)2]2–, (B) [MoIV(OSi)(bdt)2]−, (C) [MoVIO2(bdt)2]2–, and (D) [MoVIO(OSi)(bdt)2]−.

The [MoVIO2(bdt)2]2– complex (Figure 2C) has a dominant feature at ∼2473.0 eV, and a clear pre-edge feature at ∼2471.2 eV. From the second derivative (Figure S1B), there are two additional peaks required at ∼2472.2 and ∼2473.7 eV. Compared to the spectrum of the [MoVIO(OSi)(bdt)2]− complex (Figure 2D), which requires five peaks below the edge for a good fit, all four pre-edge features are again shifted to higher energy due to the additional oxo ligand in the bisoxo MoVI complex.

In comparing the MoVI to the MoIV complexes, the pre-edge features for the [MoVIO2(bdt)2]2– complex are at lower energy relative to those for the [MoIVO(bdt)2]2– complex. A similar low energy feature is also observed in the comparison of the [MoVIO(OSi)(bdt)2]− and the [MoIV(OSi)(bdt)2]− complexes. The lower pre-edge energy for the MoVI complexes is due to both the presence of an additional unoccupied d-orbital (2 spin holes) for the MoVI relative to the MoIV complexes, which corresponds to the lowest energy peak, and the increased Zeff of MoVI, which shifts the d-manifold down in energy.

The energies and inpan>tensities obtained from the fits are given in Table 1. Peak assignments are presented below.

Table 1

Experimental and Theoretical S Covalency in Mo Bis-dithiolene Complexes

		S K-edge XAS					DFT
complex	orbital	E (eV)	Do	h	S p (%)	RE (eV)a	S p (%)	RE (eV)
[MoIVO(bdt)2]2–	dxz/dyz	2472.4	1.12	4	38	0	31	0
	dxy/dz2	2473.1	1.36	4	59	0.7	59	0.7
[MoIVO2(bdt)2]2–	dyz/dy2(O out of plane π+/π−)	2471.2	0.79	4	33	–1.3	31	–1.2
	dx2−z2(O in plane π+)	2472.2	0.67	2	15	–0.3	10	–0.3
	dx2−y2(O in plane σ+)	2473.0	1.02	2	53	0.6	38	0.6
	dxz(O in plane σ–)	Not shown	NA	2	NA	NA	23	2.5

RE stands for Relative Energy. We use the lowest pre-edge energy in [MoIVO(bdt)2]2– as a reference.

RE stands for Relative Energy. We use the lowest pre-edge energy inpan> [MoIVO(bdt)2]2– as a reference.

DFT Correlations to XAS Spectra

Spin-unrestripan> class="Chemical">cted DFT calculations were used for peak assignments and for quantitative comparisons to the experimental data. The fully optimized geometric structures are consistent with the crystal structures which have bond distances within 0.04 Å, and bond angles within 3° (Figure 1).[39] Molecular orbital diagrams for the [MoIVO(bdt)2]2– and [MoVIO2(bdt)2]2– complexes are presented in Figure 3. The MO contours for the reference compounds in Figure 1B,D are shown in Figure S2 for comparison.

Figure 3

MO diagrams of [MoIVO(bdt)2]2– (left) and [MoVIO2(bdt)2]2– (right). Selected atoms in [MoVIO2(bdt)2]2– have been removed for clarity.

In the five-coordinate, square pyramidal pan> class="Chemical">[MoIVO(bdt)2]2– complex, the Mo d2–2 orbital is predominantly nonbonding and is the HOMO (z-axis along the Mo-oxo bond, and x-axis bisects the dithiolene ligands). There is a large HOMO/LUMO gap of 4.1 eV. The LUMO is close to the nearly degenerate Mo d/ set, which is strongly π-antibonding to both the oxo atom and dithiolene ligands (i.e., contains significant S p character), thus the low energy peak at ∼2472.5 eV in Figure 2A is assigned as the S 1s → Mo d/ transitions. The calculated total S 3p character in these two MOs is 31%, while experimentally, the S pre-edge intensity gives 38%. The Mo d2 and Mo d orbitals are σ antibonding to the ligands and thus at higher energies. On the basis of the calculation shown in Figure 3A, the ∼2473.3 eV peak is assigned as the S 1s → Mo d/d2 transitions. The total S 3p covalency is 59% both experimentally and computationally. In the [MoIV(OSi)(bdt)2]− reference complex, d2 and d are well separated in energy with the d at higher energy by ∼1 eV (see figure S2). In the [MoIVO(bdt)2]2– complex, the d2 is shifted to higher energy, close to d due to the strong axial Mo-oxo bond. From our previous studies on the Ni and Mo bis-dithiolene complexes, the ∼2473.7 eV peak in Figure 2A is assigned as the S 1s → C–S π*/σ* intradithiolene ligand transitions.[40,41]

As shown in Figure 4, for the six-coordinate [MoVIO2(bdt)2]2– complex, the z-axis is defined as bisecting the O–Mo–O angle, the x-axis is perpendicular to z and in the O–Mo–O plane, and the y-axis is perpendicular to this plane. For this cis bisoxo Mo complex, the energies of the Mo d-orbitals are dominated by their σ/π bonding interactions with the two strong oxo donor ligands. The six 2p-orbitals on the 2 oxo atoms form six combinations: the out of plane π+ and π–, in-plane π+ and π–, and σ+, σ–; π indicates perpendicular to the O–Mo–O plane while σ indicates along the bond, and the +/– subscripts indicate the symmetric and antisymmetric combinations of the specific O 2p-orbitals. The in-plane π– is nonbonding. The out-of-plane π+ and π–, and in-plane π+ interact with the dπ orbitals on Mo, with the in-plane π+ interaction being the strongest due to its increased overlap with the Mo (d2–2) (see Figure 3). Thus, the low energy peak at ∼2471.2 eV in Figure 2C is assigned as the combined set of transitions from the S 1s → Mo d (that interacts with the out-of-plane oxo π+) and the S 1s → Mo(d2) (that interacts with the out-of-plane oxo π−). Experiment gives 33% S p character, while the DFT calculation has a total of 31% S 3p character in these two transitions. The peak at 2472.2 eV, which is ∼1 eV to higher energy, is assigned as S 1s → Mo(d2–2) (that interacts with in-plane oxo π+) transition. Experimentally, this transition has 15% S 3p character, while the calculation gives 10%.

Figure 4

Symmetry adapted linear combinations (SALCs) of oxo pπ and pσ orbitals and the d-orbital mixings in the antibonding MOs of bisoxo MoVI bis-dithiolene complex. The coordinate system is given in the upper right panel.

Symmetry adapted linear combinpan>ations (SALCs) of oxo pπ and pσ orbitals and the d-orbital mixings in the antibonding MOs of bisoxo MoVI bis-dithiolene complex. The coordinate system is given in the upper right panel.

The MOs with oxo σ pan> class="Chemical">character are strongly antibonding and thus at higher energy. The ∼2473.0 eV peak is assigned as the S 1s → Mo(d2–d2) (antibonding with oxo atoms σ+). The S covalency is 53% from calculation vs 38% from experiment. The S 1s → Mo(d) (antibonding with oxo atoms σ−) transition is calculated to be ∼3.8 eV above the LUMO, overlapping the rising edge and thus could not be experimentally resolved. Computationally, this transition has 23% S p character. Finally, on the basis of our previous studies[5], the ∼2473.6 eV peak is assigned as S 1s → C–S π*/σ* intra-dithiolene ligand transitions.

Comparisopan> class="Chemical">ns of the DFT calculations with S K-edge XAS experimental data show that the calculations reproduce the pre-edge transition energy splittings, but somewhat underestimate the S covalency (Table 1). However, these DFT calculations did well in reproducing the experimental energies and intensities in the lower energy region. As the low-energy unoccupied MOs are the frontier molecular orbitals (FMO) in the oxo transfer reaction, these DFT calculations will be used to further evaluate the reaction coordinate for oxo transfer to phosphite ester (Section 3.3).

In comparing the pan> class="Chemical">[MoIVO(bdt)2]2– to the [MoVIO2(bdt)2]2– complexes, the total S experimental covalency changes from 194% to 248%, an increase of 54%, (computationally 180% to 204%, an increase of 24%). This increase is much smaller than the increase observed for the tris-dithiolene MoIV and MoVI complexes[40] (118%), which undergo a ligand-based redox process (i.e., noninnocent behavior). Thus, for the monooxo MoIV and bisoxo MoVI complexes, the presence of oxo ligands leads to innocent behavior of the dithiolenes where the covalent donor bonding to the Mo simply increases upon oxidation of the metal center.

Oxo Bond Strengths

The DFT calpan> class="Chemical">culations were further used to estimate the MoVI-oxo bond strengths in the bisoxo and monooxo complexes and to compare to experimental values where available. These values are necessary for consideration of the reaction coordinate results in the next section.

For the oxo transfer reaction: X + 1/2O2 → XO, the X=O bond strength is the difference between one-half the O2 bond strength and the ΔH of the reaction. The ΔHrxn values were obtained from the DFT calculations, using 119 kcal/mol for the value of the O2 bond strength.[42−44] These oxo bond strengths are given in Table 2. The computational results are within 7 kcal/mol of the experimental values, where available.[42−45]

Table 2

Comparative Experimental and Computational Oxo Bond Energies

compound	oxo bond	experimental bond strength (kcal/mol)	computational bond strength (kcal/mol)a
O2	O=O	119	–
DMSO	S=O	87	91
O=P(Me)3	P=O	139	144
O=P(Pr)3	P=O	138	144
O=P(Bu)3	P=O	137	144
O=P(OMe)3	P=O		151
O=P(OEt)3	P=O	150	151
SO42–	S=O		146
HSO4–	S=O	123	122
H2SO4	S=O		107
[MoVIO2(bdt)2]2–	Mo=O		104
[MoVIO2(mdt)2]2–	Mo=O		106
[MoVIO2(mnt)2]2–	Mo=O		102
[MoVIO(OMe)(bdt)2]–	Mo=O		116
[MoVIO(OSi)(bdt)2]–	Mo=O		118
[MoVIO2(mnt)(SMe)]−	Mo=O		126

DFT calculations were carried out in Gaussian 09, using B3LYP functional, SDD basis set for Mo, 6-311G(d) for S, O, N and P, and 6-311G(d) for other atoms.

The Mo–oxo bonpan>d strength in pan> class="Chemical">[MoVIO2(bdt)2]2– is ∼104 kcal/mol, while that in [MoVIO(OSi)(bdt)2]− is ∼118 kcal/mol. Thus, the monooxo compound has a 14 kcal/mol stronger Mo–oxo bond, out of ∼118 kcal/mol. From the energy diagram in Figure 3, the MoIV–oxo bond order in the monooxo complex is 3, which will remain the same in the MoVI monooxo complex (i.e., 2 electrons are lost from the nonbonding d orbital), while the Mo–oxo bond order in the MoVI bisoxo complex is 2.5 per bond. On breaking the Mo–oxo bond in the monooxo complex, the bond order of 3 is lost. Alternatively, breaking one Mo–oxo bond in the bisoxo complex generates a monooxo product which has a Mo–oxo bond order of 3; thus, the bond order only decreases by 2. This reflects the spectrator oxo effect described by Rappe and Goddard.[46]

This ∼10% decrease in bonpan>d strength pan> class="Chemical">can be considered quantitatively in terms of the Mayer bond orders (MBO)[47] acquired from the DFT calculations. In the [MoVIO2(bdt)2]2– compound, the MBO for each Mo–oxo bond is 1.81, and in the [MoIVO(bdt)2]2– compound, the Mo–oxo MBO is 1.99. Loss of one oxo ligand from the [MoVIO2(bdt)2]2– complex to produce the [MoIVO(bdt)2]2– complex eliminates a MBO contribution of 1.81, while the remaining spectator Mo–oxo MBO increases from 1.81 to 1.99; thus, the total Mo–oxo MBO lost is 1.63 (1.99–1.81 × 2). In the [MoVIO(OSi)(bdt)2]− compound, the Mo-oxo MBO is 1.86 and in going to [MoIV(OSi)(bdt)2]−, this MBO is lost. The ∼13% decrease in Mo–oxo MBO lost in going from the bis- to monooxo, relative to mono- to desoxo conversion (both involving changing from a six-coordinate MoVI to a five-coordinate MoIV), is consistent with the ∼10% Mo–oxo bond strength decrease, and reflects the effect of the spectator oxo in the bisoxo complex and its change in going to the monooxo MoIV complex.

Reaction Coordinates for Oxo Transfer

Transfer to Phosphite Ester

In this sectionpan>, we evaluate the pan> class="Chemical">oxo transfer process computationally using the Mo complexes with mnt ligands instead of bdt, because detailed kinetic data are available for oxo transfer from the [MoVIO2(mnt)2]2– complex to P(OR)3.[23] As described in Table S2, the geometric and electronic structures of the [MoVIO2(mnt)2]2– and [MoIVO(mnt)2]2– complexes are very similar to those of the analogous bdt ligand complexes studied above.

For the oxo transfer reaction:the calculated ΔHrxn is −49 kcal/mol (Figure 5A, R → P), which reflects the difference in the X=O bond strengths of [MoVIO2(mnt)2]2– and O=P(OMe)3 (151–102 kcal/mol, Table 2).

Figure 5

(A) Geometric structures and enthalpies at different stages of the reaction coordinate of (1). R for Reactant, TS for Transition State, P′ for the state on the path way that the oxo has been transferred while the product is still bound, P for Product. (B) HOMO (left) and LUMO (right) around TS3, showing the bonding/antibonding interaction between the phosphorus lone pair and dπ oxo as well as one of the dithiolene sulfurs.

Three possible trapan> class="Chemical">nsition states were found for reaction 1 (Figure 5A). Each transition state structure in Figure 5A has only one imaginary frequency, and has an intrinsic reaction coordinate (IRC, a one-dimensional path along the potential energy curve of a chemical reaction which describes the path of least resistance between reactants and products) coupling it to both the reactant and product. TS1 has the phosphite ester P bonding to one of the terminal oxo atoms. It is very similar to that described elsewhere.[48] TS2 has the phosphorus atom bound to both oxo atoms to form a bis(μ-oxo) structure. TS3 is similar to TS1, but with a different P–O orientation, such that the phosphorus atom also interacts with one of the sulfurs of a dithiolene ligand. The P–O orientation in TS3 is similar to a known structure,[18] but the compound in that study does not have delocalized dithiolene ligation. TS3 has the lowest energy barrier (12 kcal/mol; TS1 has 21 and TS2 has 20 kcal/mol), which is very similar to the ΔH⧧ that was measured experimentally (10 kcal/mol).[23]

There are two major points to note about the geometric anpan>d elepan> class="Chemical">ctronic structures of these transition states. First, TS3 has a P–S interaction involving the phosphorus lone pair donating electron density into the Mo-based unoccupied d-orbital that is delocalized onto the oxo and the S p orbital that is perpendicular to the dithiolene plane (Figure 5B). The P–S distance is ∼3.3 Å and the P–S MBO is 0.15. This P–S interaction stabilizes the transition state by ∼10 kcal/mol relative to TS1, which has the P lone pair interact only with the oxo pπ orbital. Second, one of the O(Me)–P–O(Mo) angles in TS3 (and TS1) is almost linear (∼165°). This was first noted by Hall and colleagues,[18] and attributed to a nucleophilic attack of the occupied oxo p orbital on the σ* orbital of the phosphite ester. From a fragment analysis, at the transition state, there is about ∼1.5% phosphite ester σ* character mixed into the occupied valence orbitals. This mixing is only observed around the transition state, and can be eliminated by decreasing the O(Me)–P–O(Mo) angle. This results in an ∼4 kcal/mol energy penalty. Alternatively, there is ∼17% of the phosphite ester lone pair orbital mixed into the unoccupied valence d orbital at the transition state (Figure 5B LUMO). Thus, the electron transfer accompanying the oxo transfer process is mainly the phosphorus lone pair into the empty Mo d π orbital through the oxo bridge.

At TS3 in Figure 5A, the Mo–O distance pan> class="Species">is 1.83 Å and the P–O distance is 1.99 Å, while the Mo–O distance for the [MoVIO2(mnt)2]2– reactant is 1.72 Å, and the P–O distance for the O=P(OMe)3 product is 1.48 Å. Thus, the Pauling bond fraction[49] for the Mo–O bond is 1.43 (decreased from 2.11), and that of the P–O bond is 0.33. Figure 6A gives this in terms of Mayer bond orders at the transition state: ∼1/3 of the Mo–O bond order is lost (a decrease in MBO from 1.84 to 1.17) and ∼1/4 of the P–O bond is formed (an increase from 0.00 to 0.43; the MBO for the P=O bond in the product is 1.80). Figure 6B shows that about 1/3 of the charge has been transferred from the substrate to the Mo complex at this transition state. The above results indicate that this oxo transfer reaction has an early transition state, with only 0.1 Å of Mo–O bond elongation.

Figure 6

Mayer bond order (MBO) (A) and Mulliken charge change (B) for reaction 1 along the reaction coordinate in terms of Mo–O distance. Spec Mo=O indicating the spectator oxo in [MoVIO2(mnt)2]2–, which is the oxo atom that is not transferred. The dashed line indicates the distance in TS3.

Mayer bond order (MBO) (A) and Mulliken charge pan> class="Chemical">change (B) for reaction 1 along the reaction coordinate in terms of Mo–O distance. Spec Mo=O indicating the spectator oxo in [MoVIO2(mnt)2]2–, which is the oxo atom that is not transferred. The dashed line indicates the distance in TS3.

The calpan> class="Chemical">culations show that there is not much structural rearrangement of the dithiolenes and the spectator oxo atom in this oxo transfer reaction. When this oxo transfer is complete, but with the phosphate ester product still bound (P′ in Figure 5A, which was obtained from a point on the IRC toward the product with a Mo–O(phosphite ester) bond distance of 2.24 Å), the two dithiolene planes remain perpendicular to each other, as in R and TS3. However, P′ is not a local minimum, and the product is lost along the reaction coordinate to give the monooxo MoIV complex. As the phosphate ester comes off, the two dithiolenes rotate into the same plane to form P (Figure 5A), and the enthalpy is lowered by 25 kcal/mol from P′ to P. This is in contrast to the reaction coordinate in refs (18) and (20), where the product remained bound. This difference reflects the fact that the final product in reaction 1 is a five-coordinate [MoIVO(mnt)2]2– complex that has its two dithiolenes coplanar. The four S pπ orbitals that are perpendicular to the dithiolene planes are parallel with each other, and this stabilizes the square pyramidal five-coordinate Mo monooxo complex.

Transfer to Sulfite

Sarkar et al. found that the [MoVIO2(mnt)2]2– complex can also transfer an oxo atom to bisulfite,[17] which is the native substrate of sulfite oxidase, to form bisulfate. They showed that the reactionexhibits Michaelis–Menten kinetics. However, the proposed transition state has an oxo atom of the substrate bound to the Mo to form a seven-coordinate structure, and is too high in energy (36 kcal/mol) for reaction 2 to occur based on kinetic data.[50] It is also not on the IRC. A lower energy transition state was thus found, and determined to be on the IRC to both the reactant and the product. This transition state is very similar to TS3 in Figure 5A with the sulfur atom of HSO3– bound to one of the terminal oxo atoms (Figure S3).

Note that anpan> pan> class="Chemical">ES precursor complex is obtained from the IRC (Figure S3) that has the HSO3– anion bound to the [MoVIO2(mnt)2]2– complex through an H-bond to one of the dithiolene sulfurs. This is consistent with the observation that reaction 2 with a monoprotonated sulfite exhibits Michaelis–Menten kinetics, while reaction 1 with phosphite ester is second-order.

The calpan> class="Chemical">culated ΔH⧧ for reaction 2 is 23 kcal/mol, and the intrinsic barrier (obtained by using the Marcus equation[51,52] to eliminate the effect of the thermodynamic driving force) is 32 kcal/mol. This is similar to the intrinsic barrier of reaction 1 (28 kcal/mol); thus, the difference in barrier heights between reactions 1 and 2 simply reflects the difference between the X=O bond strengths in the product (151 kcal/mol for phosphate ester vs 122 kcal/mol for bisulfate).

A seconpan>d-order rate constant for reaction 2 (at low substrate concentration, before saturation) of ∼60 M–1 s–1 has been reported.[17] This value is 3 orders of magnitude faster than that observed for reaction 1 at the same temperature (1.7 × 10–2 M–1 s–1), which is not consistent with the difference in the product bond strengths. Reaction 2 was performed at pH ∼ 5 (the Mo compound is not stable under basic conditions). At pH 5, the reactant is monoprotonated HSO3– (pKa = 7.2); however, the product should be deprotonated SO42– (pKa = 1.9). Thus, the reaction measured experimentally is

Thermodynamically, the S=O bonpan>d in SO42– is ∼20 kpan> class="Chemical">cal/mol stronger than that in HSO4–; thus, the free energy for the formation of SO42– is larger. Computationally, the ΔG⧧ for reaction 2 is 27 kcal/mol, and the ΔG⧧ for the reaction [MoVIO2(mnt)2]2– + SO32– → [MoIVO(mnt)2]2– + SO42– is 3 kcal/mol. Experimentally, the ΔG⧧ for reaction 3 is ∼13 kcal/mol based on the kinetics data (1.7 × 10–2 M–1 s–1 at 298 K at low concentration).

The conpan>jugate base of propionic acid was used to model the protonation behavior of bisulfite and sulfate in solution, such that bisulfite remain monoprotonated and sulfate is totally deprotonated. This was included next to the Mo complex as a base and the reaction coordinate with HSO3– was calculated as given in Figure 7.

Figure 7

Reaction coordinate of reaction 3. The conjugate base of propionic acid was used to model the protonation behavior of bisulfite and sulfate in solution.

Reactionpan> coordinate of reaction 3. The conjugate base of propionic acid was used to model the protonation behavior of bisulfite and sulfate in solution.

The substrate HSO3– is almost deprotonpan>ated at the trapan> class="Chemical">nsition state, with an O–H MBO of 0.16 to the substrate, and 0.59 to the base. The calculated ΔG⧧ for this reaction is 19 kcal/mol, which is reasonably consistent with experimental data. Thus, loss of the proton from the substrate occurs at an early stage and facilitates the oxo transfer reaction to sulfite.

Discussion

Nature of the Oxo Transfer Reaction to Phosphite Ester

The S K-edge XAS and DFT results show that the transformation from [MoVIO2(bdt)2]2– to [MoIVO(bdt)2]2– involves a metal-based redox process; thus, the dithiolenes act as innocent covalent ligands in the oxo transfer reaction. It is initiated by the phosphorus lone pair attack on an unoccupied metal-based dπ-orbital that contains oxygen pπ character, which leads to the formation of the P–O bond and loss of the Mo–oxo bond. The almost linear R–P–O bond angle[18] and a P–S interaction between the phosphite and a dithiolene ligand (see Figure 5) help stabilize the TS structure. On the basis of Mulliken charges and Mayer bond orders (Figure 6), about 1/3 of the oxo bond from the Mo and the electron pair from the phosphite ester are transferred at the TS, while the Mo–O bond has only been elongated by 0.1 Å. The two dithiolene planes remain almost perpendicular throughout the oxo transfer process, and subsequently, there is an additional ∼25 kcal/mol driving force for the two dithiolenes to reorient into the five-coordinate square pyramidal monooxo MoIV product where the dithiolenes are now coplanar and the product is released (Figure 5A). The ∼25 kcal/mol driving force is consistent with the fact that the phosphate ester product does not remain bound to the MoIVO bis-dithiolene complex, in contrast to nondithiolene Mo systems where the product remains bound after oxo transfer.

Comparison of Oxo Transfer from a Metal-Oxo to Substrate versus Oxo Transfer from a Substrate Oxo to Metal Ion

From a comparisonpan> to our previous study,[53] there is a major differenpan> class="Chemical">ce in the TS for oxo transfer from substrates to desoxo MoIV complexes, relative to oxo transfer from bisoxo MoVI complexes to substrates. In the former, the electron transfers from the occupied d-orbital of MoIV to the substrate X–O σ* orbital. The energy gap between the two FMOs in the reactants (the Mo d HOMO and the O=S σ* substrate LUMO) is ∼6 eV (Figure 8 top). The substrate S=O bond must greatly elongate to bring down the energy of the substrate σ* orbital to interact with the Mo d HOMO, and polarize the S=O bond to allow this electron transfer at a closer energy gap and with better overlap of the FMOs (Figure 8 top). Thus, the TS is late in S–O bond elongation, and for the DMSO reductase reaction, it is oxo atom transfer from the substrate that initiates the electron transfer from the Mo center.

Figure 8

FMOs for DMSO oxo transfer to desoxo MoIV (top), and bisoxo MoVI oxo transfer to phosphine (bottom).

For the reactionpan> coordinate involving oxo transfer from bisoxo MoVI to a phosphite substrate, in analogy to the sulfate oxidase enzyme reaction, electrons are transferred from the phosphorus lone pair into the low energy unoccupied Mo d-orbital. This LUMO is a dπ orbital with oxo pπ character, with a calculated energy gap between the FMOs in the reactants of ∼4 eV (Figure 8 bottom). This low energy splitting allows orbital mixing and thus electron transfer, from the substrate HOMO into the LUMO of the oxo Mo complex. Also, the oxo pπ character of the Mo d-acceptor is large enough to overlap with the phosphorus lone pair with little additional polarization. Thus, the TS is early in Mo–O coordinate, and in this case, it is the transfer of an electron pair from the phosphite ester that induces cleavage of the Mo–O bond.

In general, when an oxo atom is transferred from a metal center to a substrate, an unoccupied d-orbital is involved which is close in energy to the substrate lone pair, allowing electron transfer. Relatively, when an organic oxo atom is transferred, the electron transfer is to the substrate-oxo σ* orbital. The σ* orbital is in general high in energy and has little oxo character due to the high electronegativity of O atom. Thus, the substrate oxo bond needs significant elongation to lower its energy and polarize to initiate the electron transfer.

Directionality

From thermodynamic considerations, the direction of these reactions is in part determined by the difference in oxo bond strengths (Table 2). Here, we use mdt (mdt =1,2-dimethylethene-1,2-dithiolate(2−)) ligation to be consistent with our previously published results for DMSO reductase reactions. It is shown in Table S2 that Mo mdt complexes have geometric and electronic structures similar to those of the Mo bdt complexes. For the phosphite oxidation reaction, the P=O bond strength is ∼150 kcal/mol, while the Mo–oxo bond in the MoVI bisoxo complex is 106 kcal/mol (Table 2). This large difference in oxo bond strength drives the oxo transfer to the phosphite. In the DMSO reductase reaction, the S=O bond is 91 kcal/mol, while formation of mono MoVI–oxo bond gives 116 kcal/mol; thus, it is energetically favorable to transfer the oxo from DMSO to the Mo center.

However, there are also kinetic considerations in the directionality of oxo transfer. In comparing the bisoxo MoVI and the monooxo MoVI complexes in the reaction involving oxo transfer to phosphite, the difference in the reaction enthalpy is 9 kcal/mol, which reflects the bond strength difference (section 3.2). Their difference in the activation enthalpy is 3 kcal/mol (Figure S4), which gives an intrinsic barrier difference of only 1 kcal/mol. Both reactions are initiated by the phosphorus lone pair attacking the well-exposed oxo atom in the six-coordinate MoVI complexes, and the structures of their TS’s are very similar. The distortion of the Mo complexes into five-coordinate products occurs after the TS. Electronically, both complexes use similar dπ* FMOs with similar energy gaps (∼4 eV in the reactant) and have similar phosphorus lone pair mixing into the Mo d LUMO (∼30%) at the TS. Thus, oxo transfers from both bisoxo and monooxo MoVI to phosphite have very similar reaction coordinates with a rate difference reflecting their difference in reaction energy.

Alternatively, it is also thermodynamically favorable to transfer an oxo atom from DMSO to both the des- and monooxo MoIV complexes to form monooxo and bisoxo MoVI complexes, respectively, with the same 9 kcal/mol difference. However, the difference in enthalpic barrier height, ΔH⧧, is 14 kcal/mol (Figure 9A, black vs green). This gives an intrinsic barrier for the monooxo to bisoxo reaction that is 9 kcal/mol higher than that for the des-to-mono oxo reaction.

Figure 9

(A) Reaction coordinate for oxo transfer from DMSO to MoIV complexes. Enthalpic barriers and intrinsic barriers (in parentheses) are indicated. T.B. stands for trigonal bipyramidal. The TS in red has a similar geometry to the TS in black. The starting point is where the Mo complex and the DMSO are ∼4 Å away from each other. (B) Geometry of the DMSO bound intermediate structure for the des-to-mono oxo (left) and mono-to-bis oxo (right) coordinate. Mo–O(DMSO) distance is constrained at 2.24 Å.

(A) Reactionpan> coordinate for oxo transfer from DMSO to MoIV complexes. Enthalpic barriers and intrinsic barriers (in parentheses) are indicated. T.B. stands for trigonal bipyramidal. The TS in red has a similar geometry to the TS in black. The starting point is where the Mo complex and the DMSO are ∼4 Å away from each other. (B) Geometry of the DMSO bound intermediate structure for the des-to-mono oxo (left) and mono-to-bis oxo (right) coordinate. Mo–O(DMSO) distance is constrained at 2.24 Å.

In the des-to-monpan>o reapan> class="Chemical">ction, the reactant is a 5C MoIV complex. The methoxyl group distorts to open a site for DMSO to bind to form a 6C intermediate state (Figure 9B left). In this intermediate, DMSO-bound complex, the DMSO distortion energy is 3 kcal/mol, and the distortion energy of the MoIV desoxo complex is 18 kcal/mol. There is 12 kcal/mol interaction energy between the two fragments, due to 14% donor and 2% backbonding interaction. Thus, the total energy of this DMSO bound state is 9 kcal/mol higher than that of the reactant, and there is an additional 7 kcal/mol required to reach the TS. In the mono-to-bis reaction, the five-coordinate MoIV complex does not distort as described above, because of the higher energy required to distort the strong oxo bond (see trigonal bipyramidal (TB) TS in Figure 9A red). Instead, the dithiolene planes rotate to leave an open position for DMSO binding (Figure 9B right). The DMSO distortion energy is similar (2 kcal/mol), but the bis-dithiolene-MoIV-oxo distortion energy is larger than that for DMSO binding to the desoxo MoIV (23 kcal/mol). There is also a decreased interaction energy (7 kcal/mol) reflecting decreased donor bonding of the DMSO to MoIV (8%) due to the strong oxo–MoIV bond. This leads to a higher total energy (18 kcal/mol) to bind DMSO to the MoIV=O complex, and another 12 kcal/mol is required to reach the transition state.

In summary, for oxo transfer from six-coordinate MoVI complexes to phosphite, electron transfer leads to oxo transfer. The electron density in the phosphorus lone pair orbital readily interacts with the oxo p-character in the low energy unoccupied metal d-orbital; thus, both the bis-to-mono or mono-to-des reactions are accessible. Alternately, for oxo transfer from DMSO to the five-coordinate MoIV complexes, it is the oxo transfer that leads to electron transfer. This requires DMSO binding to the MoIV complex, which is more difficult in the presence of a strong oxo donor ligand.

Correlations to the Enzymes

With the results of the model studipan> class="Chemical">es, calculations can be used to consider the oxo transfer mechanism for native sulfite oxidase. Note that the substrate for the native enzyme is the fully deprotonated sulfite,[54] while the model complex in section 3.3.2 was not stable at the pH ∼8 where SO32– dominates. By systematically varying the substrate and the ligation of the Mo site, the results from the [MoVIO2(mnt)2]2– complex can be extended to the enzyme active site (reaction numbers below correspond to those in Table 3).

Table 3

Computational Comparison of Oxo Transfer from MoVIO Dithiolene Complex to Two Different Substrates

rxn no.	MoVI site	substrate	ΔHrxn (kcal/mol)	ΔH⧧ (kcal/mol)a	Mo–O distance at TS (Å)	Mo–O MBO at TS	energy gap (eV)b	mixc
(1)	[MoO2(mnt)2]2–	P(OMe)3	–48	11(30)	1.83	1.17	4.2	21%
(4)	[MoO2(mnt)2]2–	SO32–	–39	1(13)	1.78	1.54	1.2	35%
(5)	[MoO2(mnt)(SCH3) ]−	P(OMe)3	–25	14(25)	1.83	1.17	4.0	23%
(6)	[MoO2(bdt)(SCH3) ]−	P(OMe)3	–20	16(25)	1.85	1.13	4.0	22%
(7)d	[MoO2(bdt)(SCH3) ]−	SO32–	–23	3(12)	1.82	1.41	1.4	41%
(8)e	[MoO2(bdt)(SCH3) ]−	SO32–	–12	32(38)	1.82	1.39	1.1	39%

Intrinsic barriers show in parentheses.

Energy gap between the substrate lone pair and the LUMO of the Mo site in reactant.

Amount of substrate lone pair mixed into the unoccupied orbitals at the transition state.

In this reaction, the product sulfate stayed bound to the Mo site, which remains a five-coordinate site.

The reaction was carried out with a dielectric constant of 4. There is no stable ES or EP complex, and the resultant four-coordinate MoIV site rearranged to a trigonal pyramidal geometry.

Energy gap between the substrate lone pair and the LUMO of the Mo site inpan> reactant.

In this reactionpan>, the product sulfate stayed bound to the Mo site, which remains a five-coordinate site.

The reactionpan> was carried out with a dielectric constant of 4. There is no stable ES or EP complex, and the resultant four-coordinate MoIV site rearranged to a trigonal pyramidal geometry.

Comparisonpan> of reapan> class="Chemical">ctions (1) and (4) shows that for the same six-coordinate [MoVIO2(mnt)2]2– complex, oxo transfer to SO32– has a much lower barrier than to phosphite. The dianionic SO32– has more charge density, thus its lone pair donor HOMO is closer in energy to the LUMO of the Mo bisoxo complex. Thus, even less distortion is required for the overlap of the FMOs for the electron transfer to induce oxo transfer to sulfite and there is an even earlier transition state (Table 3, shorter Mo–O distance and higher Mo–O bond order).

In reactionpan> (1), the Mo pan> class="Chemical">site goes from six-coordinate to five-coordinate, while in reaction (5) one of the dithiolenes is replaced with one thiolate, and the oxo transfer to phosphite corresponds to a five-coordinate to four-coordinate conversion. Reaction (5) has less driving force than (1) due to the difference in Mo–oxo bond strengths (102 kcal/mol for the six-coordinate and 126 kcal/mol for the five-coordinate MoVI–bisoxo complex); however, the intrinsic barrier for reaction (5) is lower (Table 3). For the six-coordinate [MoVIO2(mnt)2]2–, the two oxo atoms are equivalent (9% of each in the LUMO, Figure 10A), and need to polarize toward the oxo that is transferred (Figure 8B). In the five-coordinate [MoVIO2(mnt)(SCH3)]− complex, the two oxo atoms are nonequivalent with the equatorial oxo contributing 22% to the LUMO (1% for the axial oxo, Figure 10B). This activates the equatorial oxo for transfer, consistent with the selectivity of the enzyme.

Figure 10

(A) LUMO of six-coordinate MoVIO2 site. The two oxo atoms are identical in the reactant. (B) LUMO of the five-coordinate MoVI site. The equatorial oxo atom has a much larger contribution than the axial oxo atom.

Another advantage of the five-coordinate pan> class="Chemical">MoVI bisoxo complex is that for a six-coordinate dithiolene complex, the two dithiolene planes have to rotate by about 90° for each reaction cycle. From section 3.3.1, this rotation costs ∼25 kcal/mol. The energy would be even higher in the enzyme where the pyranopterin extends into the protein. Less dithiolene rearrangement is required in the five-coordinate MoVI bisoxo site as the major change is simply equatorial oxo transfer.

In going from reactionpan> (5) to (6), the mnt is pan> class="Chemical">changed into the bdt, without a change in the intrinsic barrier, and reaction (7) simulates the active site of the native enzyme with the native substrate. Relative to reaction (6) with phosphite, sulfite lowers the reaction barrier and leads to an earlier transition state, as discussed above. The product SO42– remains bound to the MoIV monooxo product complex. The calculated ΔG⧧ is 4 kcal/mol, while based on the rate constant (7.0 × 106 M–1 s–1 at 298 K),[54] the experimental ΔG⧧ for chicken liver sulfite oxidase is ∼7 kcal/mol. All the calculations above were done in acetonitrile, which has a dielectric constant of 36. However, in proteins, the dielectric constant is decreased.[55] Reaction (8) repeated reaction (7) with a dielectric constant of 4. This resulted in a large increase in reaction barrier with no stable ES complex, due to the increase in Coulomb repulsion. The Mo site has a total charge of −1, while the sulfite has a charge of −2. This suggests that in SO, the protein environment contributes to the substrate binding (a positively charged pocket with conserved tyrosine, histidine, and arginine residues) to form the ES complex to lower the barrier for oxo transfer,[56,57] while the pyranopterin cofactor functions as a pathway to reoxidize the Mo site subsequent to the oxo transfer.[22,58] [Note that the Oax–Mo–Sthiolate–C dihedral angle in the optimized structure is different from that in the crystal structure (148° vs 78°). However, this dihedral angle distortion only costs a few kcal/mol of energy, consistent with ref (55).]

In conpan>trast, the experimental ΔG⧧ for pan> class="Species">Escherichia coli DMSO reductase is ∼9 kcal/mol (based on the rate constant 4.3 × 105 M–1 s–1 at 298 K),[59] much lower than the value calculated for the model complex (30 kcal/mol).[10] This indicates that the geometry of the enzyme active site is distorted toward the transition state by the protein environment to assist the oxo transfer reaction in DMSOr as suggested by Hall and co-workers[15] and Kirk and co-workers.[13] Also, in this class of oxo transfer enzymes, the two pyranopterins are found to be structurally different with the protein tuning one to be an electron conduit, to effectively reduce the oxidized state.[58]

Conclusions

This study provides ipan> class="Chemical">nsight into the oxo transfer reaction in sulfite oxidase, based on experimental and computational results on its model complexes, and its relation to our previous studies on related DMSO reductase models. In DMSOr, it is the oxo transfer that leads to electron transfer, while in SO, oxo transfer is initiated by the electron transfer. This difference reflects the large energy gap between the LUMO of DMSO and dπ HOMO in the MoIV desoxo complex relative to the small energy gap between the sulfite lone pair HOMO and the dπ orbital-based LUMO of the MoVI bisoxo complex. The five-coordinate MoVI bisoxo active site of SO activates the equatorial oxo, and minimizes the reorganization energy over the reaction cycle.