The role of terminal heavy chalcogenide ligands in enabling hydride transfer reactivity of W=E (E = S, Se) models for molybdenum hydroxylases

Abstract

Whereas molybdenum-dependent hydroxylases in the xanthine oxidase family catalyze site-selective C–H hydroxylation reactions utilizing terminal molybdenum sulfide (Mo=S) active sites, certain variants such as nicotinate dehydrogenase (NDH) replace the sulfide ligand by a selenide (Mo=Se) to enable activation of less reactive C–H bonds. To probe the origins of this dependence on terminal chalcogenide identity, we prepared and crystallographically characterized a synthetic model complex of the NDH active site, [(bdt)W(O)₂Se]²⁻, a dioxotungsten(VI) selenide complex supported by benzene-1,2-dithiolate (bdt), for comparison to the previously reported [(bdt)W(O)₃]²⁻ and [(bdt)W(O)₂S]²⁻ analogues. Electronic structure analyses combining experimental UV-Vis spectroscopy and DFT calculations indicate that replacement of terminal oxo ligands by heavier chalcogenides introduces accessible low-energy acceptor orbitals associated with more spatially distributed W=E (E = S, Se) interactions. Treatment of the terminal selenido and sulfido complexes with BH₄⁻ led to reduced tungsten products accompanied by liberation of free hydroselenide (HSe⁻) and hydrosulfide (HS⁻), respectively, whereas the corresponding trioxo analogue was unreactive under analogous conditions. These findings provide direct experimental support for the proposal that terminal heavy chalcogenide ligands play an active electronic role in enabling biological hydride-transfer reactivity in molybdenum hydroxylase enzymes.

Graphical Abstract

INTRODUCTION

The xanthine oxidase (XO) family of metalloenzymes constitutes a widespread class of molybdenum-dependent hydroxylases that mediate C–H bond hydroxylation across diverse biological systems.^1,2 The canonical XO active site structure features a distorted square-pyramidal (PDT)Mo(O)(S)(OH) core (PDT = pyranopterin dithiolate), with a terminal oxo ligand in the apical position and an equatorial plane containing a catalytically essential terminal sulfido ligand and a terminal oxo ligand unmasked by deprotonation during catalysis (Figure 1a).^3,4 In desulfo (inactive) forms, replacement of the sulfido ligand by another oxo unit (i.e., (PDT)Mo(O)₂(OH)) abolishes catalytic activity.⁵ These observations illustrate the critical role of the terminal chalcogenide ligand in modulating enzymatic XO reactivity.

Figure 1.

(a) Representative examples of C–H oxidation reactions catalyzed at the active sites of the xanthine oxidase enzyme family, (b) previously proposed mechanism for C–H activation by H⁻ abstraction at the nicotinate dehydrogenase active site, (c) summary of this work.

In select members of the XO family, the canonical sulfido ligand is substituted for selenide.⁶ For example, the anaerobic bacterium Eubacterium barkeri metabolizes nicotinate to propionate, acetate, CO₂, and ammonia initiated by a nicotinate hydroxylation step catalyzed by nicotinate dehydrogenase (NDH).^7–10 Structural characterization of NDH by Dobbek revealed a terminal Mo=Se group at the active site (Figure 1a).⁵ It has been proposed that catalysis involves initial hydride abstraction from the nicotinate C⁶–H position by the terminal Mo=Se unit (Figure 1b).¹¹ Terminal Mo=Se ligation has likewise been identified in certain bacterial xanthine oxidoreductases and purine hydroxylases.⁶ Based on DFT studies, Dobbek has proposed that the covalent Mo=Se unit possesses a low-energy π*_Mo=Se orbital that stabilizes the transition state for C–H activation relative to the Mo=S congener, resulting in an estimated 200–300-fold rate enhancement.^5,11 On this basis, we hypothesized that a synthetic model complex mimicking the NDH active site would serve as a platform to precisely correlate metal-chalcogenide electronic structure with reactivity relevant to biological C–H oxidation.

Synthetic modeling of NDH and other terminal selenido molybdenum enzymes has been extremely limited. To date, the only reported effort was described by Itoh; however, crystallographic characterization of the proposed O=W=Se complex was not achieved, nor were reactivity studies with hydride donors reported.¹² More broadly, synthetic mimics of XO active sites remains scarce.^13–16 Of relevance to this study is a 2007 precedent from Holm, who successfully characterized a dithiolene-supported cis-O=W^VI=S complex as a structural analogue of XO active sites.¹⁷ Notably, tungsten is frequently employed in synthetic modeling as an isostructural and nearly isometric surrogate for molybdenum.¹⁸ The tungsten analogues exhibit enhanced resistance to metal-centered reduction, enabling isolation of high-valent species that can be inaccessible in the molybdenum series.¹⁹ Herein, we report the complex [(bdt)W^VI(O)₂Se]²⁻ (1, bdt = benzene-1,2-dithiolate) that represents the closest structural analogue of the NDH active site to date. Notably, this complex exhibited reactivity toward the hydride donor, [Bu₄N][BH₄], affording [(bdt)₂W^IV=O]²⁻ and {[(bdt)W^VO]₂(μ-Se)₂}²⁻ along with the release of free SeH⁻ (Figure 1c). This reactivity is consistent with the terminal selenide ligand acting as a hydride acceptor site, as proposed for the catalytic mechanism of NDH. Links between this reactivity and the unique properties of terminal selenide ligation were provided by electronic structure calculations.

RESULTS AND DISCUSSION

Initially, we attempted to access a terminal Mo=Se complex by reaction of [Mo^VIO₃Se][NEt₄]₂ with bdt·2SiMe₃. However, these trials inevitably gave the reduction product, [(bdt)₂Mo^IV=O][NEt₄]₂, and formation of unknown [MoSe_xO_y] species. This behavior parallels previously reported sulfido analogues and suggests the instability of high-valent Mo=Se complexes in the presence of dithiolene ligands.¹³ Analogous reactions using [W^VIO₃Se][NEt₄]₂ similarly proved unsuccessful, giving [(bdt)₂W^VI(O)₂][NEt₄]₂ and unknown Se-containing byproducts.

Use of the less electron rich ^Clbdt·2H (^Clbdt = 3,6-dichlorobenzene-1,2-dithiolate) enabled formation of the desired selenide complex, [^ClbdtW^VI(O)₂Se][NEt₄]₂ (1^Cl). However, the reaction also produced tungsten tetraselenide species, as confirmed by X-ray crystallography, that proved inseparable from the main product (Figure 2a). To overcome these limitations, we adopted a silyl protection/deprotection strategy inspired by Holm’s work (Figure 2b).¹⁹ Starting from the silyl-protected precursor, [(bdt)W^VI(O)₂(OSiⁱPr₃)][NEt₄], treatment with ⁱPr₃SiSeH enabled controlled selenium incorporation to afford [(bdt)W^VI(O)(Se)(OSiⁱPr₃)][NEt₄] (2). Subsequent NEt₄F-mediated desilylation yielded the target terminal selenide complex, [(bdt)W^VIO₂Se][NEt₄]₂ (1), in 62% yield. Crystallographic characterization of 1, 1^Cl, and 2 revealed distorted square pyramidal geometries at W, as expected (Figure 2c). The W=Se bonds in 1 and 1^Cl are elongated (2.3990(12) Å and 2.3759(19) Å, respectively) relative to Mo=Se distances reported for the enzyme active sites (~2.30 Å), likely due to their dianionic molecular charges. Accordingly, monoanionic 2 has a W=Se distance of 2.3071(10) Å, more in line with the native systems.

Figure 2.

(a) Synthesis of 1^Cl, (b) synthesis of 1 via 2, (c) solid-state structures determined by X-ray crystallography (50%-probability thermal ellipsoids; cations, hydrogens, and disordered iso-propyl components omitted).

The UV–vis spectrum of 1 exhibits absorptions at 310 nm (32,258 cm⁻¹) and 420 nm (23,810 cm⁻¹). To assign these transitions, geometry optimization followed by TD-DFT calculations were carried out at the ωB97X-D/def2-TZVPP level of theory (see Computational Details in Supporting Information).^20,21 Natural transition orbital (NTO) analysis²² assigns the low-energy transition (calcd. 335 nm) as a ligand-to-metal charge transfer (LMCT), with the excited particle centered on tungsten and the corresponding hole centered on the terminal selenido ligand (Figure 3). The corresponding sulfido analogue [(bdt)W^VI(O)₂S]²⁻ (3) was reported to have a band at 405 nm (24,691 cm⁻¹).¹⁹ According to our NTO analysis, this excitation (calcd. 323 nm) is also a LMCT, with the excited particle localized on tungsten and the corresponding hole predominantly centered on the terminal sulfido ligand. The literature spectrum of the trioxo analogue, [(bdt)W^VI(O)₃]²⁻ (4),¹³ lacks a corresponding low-energy transition in this region and has a single band at 320 nm (31,250 cm⁻¹; calcd. 277 nm), confirming that the features at 405-420 nm for 1 and 3 are not derived from the W=O or W(bdt) units. The high-energy absorption bands at ~310–320 nm observed for all three complexes are assigned from NTO analysis to intra-ligand charge transfer transition within the bdt framework, with the excited particle localized primarily on the aromatic carbon backbone and the corresponding hole distributed over the bdt sulfur atoms. Thus, the experimental UV–Vis spectra together with TD-DFT and NTO analyses indicate that substitution of the terminal sulfido ligand by selenido stabilizes the metal-chalcogen π* acceptor orbital involved in the LMCT excitation, consistent with predictions for the native molybdenum systems.^5,11

Figure 3.

UV-Vis absorption spectrum of 1 and natural transition orbital (NTO) surfaces corresponding to the excitation at 420 nm (0.05 e Å⁻³ isosurfaces).

To elucidate the electronic origin of these spectroscopic trends, we analyzed the experimentally determined structures using the formal shortness ratio (FSR) metric, defined as the ratio of the observed bond length to the sum of covalent radii.²³ We also analyzed Wiberg bond indices (WBIs) obtained from the DFT calculations, as these are widely used qualitative indicators of covalent bond order in transition metal-ligand multiple bonds.^24,25 The W=Se and W=S bonds in 1 and 3, respectively have similar FSRs of ~0.85, indicative of multiple bond character in both cases. The calculated Wiberg bond index (WBI) for the W=Se bond in 1 is 1.84, compared to 2.08 for the W=S bond in 3. In 1, the axial and equatorial W=O bonds exhibit WBIs of 2.43 and 2.33, respectively, while in 3 the corresponding values are 2.38 and 2.28. For the trioxo analogue 4, the W=O WBIs are 2.30 for the axial oxo ligand and 2.20-2.21 for the two equatorial oxo ligands. Collectively, these values are consistent with double-bond character for the W=E (E = O, S, Se) interactions. Frontier molecular orbital analysis using Mulliken populations shows that the HOMOs of 1 and 3 are predominantly ligand-centered, with significant contributions from the terminal chalcogenide p orbitals (Se: 28%; S: 28%) and the dithiolene sulfur atoms (~28%), and minimal tungsten character (~2%). In contrast, the corresponding LUMOs are largely metal-centered, containing ~58% tungsten character and only 12–13% chalcogenide contribution, in qualitative agreement with chalcogen-to-metal charge-transfer transitions observed spectroscopically (see above). Indeed, the HOMO–LUMO gap (Figure 4) is smaller for 1 (7.65 eV) than for 3 (7.99 eV), in alignment with the lower-energy LMCT transition observed experimentally for 1. In contrast, the trioxo analogue 4 exhibits a markedly different electronic structure, with a HOMO primarily localized on the bdt sulfur atoms (~56%) and a LUMO centered on the bdt carbon atoms. These results indicate that incorporation of a terminal heavy chalcogenide ligand in place of oxo introduces molecular orbitals associated with metal-chalcogen π-bonding into the frontier manifold, consistent with previous proposals for biological molybdenum hydroxylase active sites.^26–28 The notable dependence of enzymatic reactivity on the identity of the equatorial chalcogenide ligand likely derives from this effect.¹¹

Figure 4.

Frontier orbital analysis for the [(bdt)WO₂E]²⁻ series (E = Se (1), S (3), O (4)) showing trends in calculated HOMO–LUMO energy gap and chalcogenide p-orbital contribution (0.05 e Å⁻³ isosurfaces)

Atoms-in-molecules (AIM)²⁹ analysis provided complementary insight into the electronic structures.^30,31 When examining bond critical point (bcp) descriptors across the W=E series in 1, 3, and 4, the most pronounced changes were observed for the Laplacian of the electron density, ∇²ρ(r_bcp), and normalized kinetic energy density, G(r_bcp)/ρ(r_bcp), terms. The W=Se bond in 1 (Figure S21) exhibits comparatively small values of ∇²ρ(r_bcp) = 0.042 e Å⁻⁵ and G(r_bcp)/ρ(r_bcp) = 0.514, whereas larger values are observed for the W=S bond (Figure S22) in 3 (∇²ρ(r_bcp) = 0.106 and G(r_bcp)/ρ(r_bcp) = 0.665). Each equatorial W=O bond in 4 displays even larger values of ∇²ρ(r_bcp) = 0.894 e Å⁻⁵ and G(r_bcp)/ρ(r_bcp) = 1.389 (Figure S23). Conventionally, the positive ∇²ρ(r_bcp) values describe the extent of charge concentration at the bcp relative to the atomic basins. Relatedly, the G(r_bcp)/ρ(r_bcp) term provides a measure of the spatial confinement energy for charge density at the bcp. Overall, these data reveal a clear progression wherein, going from W=O to W=S to W=Se, the bonding becomes more spatially distributed and atoms less polarized. This conclusion is in agreement with the electronic spectroscopy and frontier orbital analyses discussed above.

The distinct electronic structures observed for the terminal W=E bonds (E = O, S, Se) revealed from electronic spectroscopy, frontier orbital analysis, and charge density analysis provides some insight into the contrasting reactivity patterns of the corresponding Mo=E motifs in biological systems. Whereas incorporation of a terminal sulfido or selenido ligand introduces accessible low-energy frontier orbitals that could be capable of mediating H⁻ abstraction reactivity, the corresponding oxo complexes lack analogous acceptor orbitals, thus providing an electronic rationale for the absence of hydroxylase chemistry in naturally occurring trioxo cofactors.^5,11

Next, we investigated the reactivity of the model compounds with a range of hydride donors to probe whether the terminal W^VI=E units can, indeed, serve as H⁻ acceptor sites. We have recently studied a bioinspired dioxomolybdenum(VI) complex capable of C–H hydroxylation of benzimidazolines via H⁻ abstraction.³² When the same substrates were reacted with 1 and 3, they produced complex mixtures of unidentified products. On the other hand, treatment of 1 with [NBu₄][BH₄] at 45 °C in acetonitrile solvent (Scheme 1) resulted in conversion to reduced complexes [(bdt)₂W^IV(O)₂]²⁻ and [bdtW^V(O)(μ₂-Se)₂W^V(O)bdt]²⁻ in ~ 1:1 ratio , the identities of which were proposed by comparison of ¹H NMR data to literature reports combined with X-ray crystallography.^17,33 This reaction presumably also forms WO_x species that were not characterized. Notably, the ¹H and ⁷⁷Se NMR spectra (CD₃CN) of the product mixture of showed resonances at −6.57 ppm (¹H) and −406 ppm (⁷⁷Se), both consistent with literature values for free hydroselenide (HSe⁻).^12,34 The generation of HSe⁻ was further confirmed by trapping experiments using excess benzyl bromide (BnBr), which afforded dibenzyl selenide (BnSeBn) as verified by ¹H NMR spectroscopy.³⁵

Scheme 1.

Experimental confirmation of H⁻ delivery to the terminal W=Se unit.

Similarly, treatment of the sulfido analogue 3 with [NBu₄][BH₄] produced a ¹H NMR resonance at −3.84 ppm consistent with free hydrosulfide (HS⁻).³⁶ Trapping of HS⁻ with excess benzyl bromide likewise afforded dibenzyl sulfide (BnSBn). In contrast, the trioxo analogue [(bdt)W^VI(O)₃]²⁻ was unreactive toward [NBu₄][BH₄] under identical conditions. Although the initial hydride-transfer step closely parallels the proposed enzymatic chemistry, the subsequent loss of the terminal chalcogenide ligand is not thought to occur in the enzyme systems.^5,10 This difference likely reflects the greater propensity of reduced W^V/W^IV synthetic complexes to undergo ligand reorganization under non-biological conditions.^37,38 Collectively, these findings establish that replacement of terminal oxo ligands by heavier chalcogenides fundamentally reorganizes the frontier orbital structure of high-valent tungsten centers, introducing frontier metal-chalcogen molecular orbitals amenable to hydride transfer. The observed hydride transfer to terminal W=Se and W=S units, therefore, provides experimental validation of long-standing mechanistic proposals that terminal sulfido and selenido ligands play an active electronic role in the catalytic hydride-transfer chemistry of molybdenum hydroxylase enzymes.^4,5

CONCLUSIONS

In summary, a benzene dithiolate-supported W^VI(O)₂Se complex (1) has been prepared as the first structural model for enzyme active sites containing terminal Mo^VI=Se groups. Combined spectroscopic, computational, and reactivity studies demonstrate that substitution of terminal oxo ligands by heavier chalcogenides introduces frontier molecular orbitals associated with metal-chalcogen π-bonding evidenced by ligand-to-metal charge-transfer excitations and available for hydride-accepting reactivity. Consistent with this electronic structure proposal, the terminal selenido and sulfido complexes undergo hydride transfer from BH₄⁻ with concomitant liberation of HSe⁻ and HS⁻, whereas the corresponding trioxo analogue remains unreactive under identical conditions. These results provide direct experimental validation of the proposed catalytic role of terminal heavy chalcogenide ligands in molybdenum hydroxylases and establish a biomimetic platform for directly correlating terminal chalcogenide electronic structure with hydride-transfer reactivity.

Supplementary Material

The following files are available free of charge.

Experimental details, materials and methods, and spectral data (PDF)

Computational output (XYZ)