A Thioether-Ligated Cupric Superoxide Model with Hydrogen Atom Abstraction Reactivity

Abstract

The central role of cupric superoxide intermediates proposed in hormone and neurotransmitter biosynthesis by noncoupled binuclear copper monooxygenases like dopamine-β-monooxygenase has drawn significant attention to the unusual methionine ligation of the Cu_M (“Cu_B”) active site characteristic of this class of enzymes. The copper–sulfur interaction has proven critical for turnover, raising still-unresolved questions concerning Nature’s selection of an oxidizable Met residue to facilitate C–H oxygenation. We describe herein a model for Cu_M, [(^TMGN₃S)Cu^I]⁺ ([1]⁺), and its O₂-bound analog [(^TMGN₃S)Cu^II(O₂^•−)]⁺ ([1·O₂]⁺). The latter is the first reported cupric superoxide with an experimentally proven Cu–S bond which also possesses demonstrated hydrogen atom abstraction (HAA) reactivity. Introduction of O₂ to a precooled solution of the cuprous precursor [1]B(C₆F₅)₄ (−135 °C, 2-methyltetrahydrofuran (2-MeTHF)) reversibly forms [1·O₂]B(C₆F₅)₄ (UV/vis spectroscopy: λ_max 442, 642, 742 nm). Resonance Raman studies (413 nm) using ¹⁶O₂ [¹⁸O₂] corroborated the identity of [1·O₂]⁺ by revealing Cu–O (446 [425] cm⁻¹) and O–O (1105 [1042] cm⁻¹) stretches, and extended X-ray absorption fine structure (EXAFS) spectroscopy showed a Cu–S interatomic distance of 2.55 Å. HAA reactivity between [1·O₂]⁺ and TEMPO–H proceeds rapidly (1.28 × 10⁻¹ M⁻¹ s⁻¹, −135 °C, 2-MeTHF) with a primary kinetic isotope effect of k_H/k_D = 5.4. Comparisons of the O₂-binding behavior and redox activity of [1]⁺ vs [2]⁺, the latter a close analog of [1]⁺ but with all N atom ligation (i.e., N₃S vs N₄), are presented.

The noncoupled binuclear (NCBN) copper monooxygenase family of peptidylglycine-α-hydroxylating monooxygenase (PHM), dopamine-β-monooxygenase (DβM), and tyramine-β-monooxygenase (TβM) is crucial to hormone and neurotransmitter biosynthesis. This family is characterized by two distant (11–14 Å) copper centers Cu_M (Cu_B) and Cu_H (Cu_A) (Figure 1). The former has a MetHis₂ coordination sphere and is thought to play the major role in O₂ activation and substrate-binding, while the latter has His₃ coordination and behaves as an electron relay.^1–4 After years of study,^2,5–10 experimental and computational evidence have mostly converged on an end-on (η¹) superoxide Cu_M^II(O₂^•−) intermediate hypothesized as the active species in substrate hydrogen atom abstraction (HAA), initiating C–H hydroxylation.¹¹ Such an intermediate is distinct from other proposed cupric superoxides^12–14 due to the presence of an oxidizable methionine at Cu_M. In fact, this Met has proven critical to turnover, as mutants lacking sulfur ligation (Met, Cys) have been inactive.¹⁵

Figure 1.

A depiction (PDB: 1SDW)² of the copper centers of PHM, the O₂-binding Cu_M, and the electron relay Cu_H.

The criticality of the methionine remains unexplained^16–19 prompting model system investigations. These have demonstrated the influences of denticity,^4,20–25 electron richness,^26,27 steric modulation,^28,29 coordination geometry,³⁰ temperature, solvent,³¹ and hydrogen bonding^32,33 on electronic structure and reactivity.

Despite these insights, we still lack a Cu^IIO₂^•− model with a clear Cu^II–S interaction and its corresponding HAA reactivity. Although thioether-containing tridentate or tetradentate ligand-copper models have been investigated,^34–41 many synthetic attempts rather form binuclear cupric peroxides,^37,42–44 and none exhibit substrate oxidative reactivity. Our recent report⁴⁵ using a modified TMPA ligand²⁶ succeeded in observation of a mononuclear Cu^IIO₂^•− in the presence of a thioether; however, direct evidence of the Cu–S interaction by (EXAFS) spectroscopy eluded us. Furthermore, the extreme instability of this complex prevented rigorous spectroscopic and reactivity studies, as it rapidly formed the more stable trans-peroxide.

We now report that the ^TMGN₃S ligand³⁹ supports a cupric superoxide, [(^TMGN₃S)Cu^IIO₂^•−]⁺ ([1·O₂]⁺), with a demonstrable Cu^II–S interaction (Scheme 1) and HAA reactivity. Resonance Raman (rR) studies confirmed the identity of ([1·O₂]⁺) as a superoxide complex, and EXAFS studies demonstrate the first irrefutable Cu–S bond in a CuO₂ complex. Beyond the relevance of [1·O₂]⁺ as an NCBN structural model with Cu–S_thioether ligation, it is the first functional mimic of the Cu_M active site.

Scheme 1.

Compound [1]B(C₆F₅)₄ Reversibly Binds O₂ upon Cooling to −135 °C in 2-MeTHF, Forming [1·O₂]B(C6F5)4

The O₂-chemistry of [1]OTf was previously interrogated, but no superoxide formation was observed in acetone at −76 °C.³⁹ However, the success of the similar TMG₃tren ligand in supporting a stabilized cupric superoxide [2·O₂]^+28,46–48 encouraged our exploration of broader experimental conditions for the −CH₂CH₂SEt ligand analog, [1·O₂]⁺ (Scheme 1). We sought lower temperatures by a switch from acetone to 2-MeTHF, also requiring a switch to tetrakis(pentafluorophenyl)borate, B(C₆F₅)₄⁻, as the counteranion. Combination of ^TMGN₃S with [Cu^I(CH₃CN)₄](C₆F₅)₄ in 2-MeTHF provided [1]B(C₆F₅)₄ (87% yield) following workup and recrystallization. Our present single crystal X-ray diffraction (XRD) study (see Section 2 in the Supporting Information (SI)) revealed a monomeric species with a normal Cu–S distance of 2.2709(5) Å (Figure 2a).^{37,39,42–44} [1]B(C₆F₅)₄ possesses unremarkable Cu–N distances of 2.0362(15), 2.0436(15), and 2.1987(14) Å. This four-coordinate, distorted trigonal pyramidal geometry presents an open fifth coordination site for further ligand binding.

Figure 2.

(a) Displacement ellipsoid plot (50% probability level) of [1]B(C₆F₅)₄ at 110(2) K with a Cu–S distance of 2.2709(5) Å. Hydrogen atoms and counterions have been omitted for clarity. (b) Structure of [1·O₂]⁺ from density functional theory (DFT) calculations. Selected interatomic distances [Å] and angles [deg]: Cu–S 2.50, Cu–O 1.99, O–O 1.29; Cu–O–O 113.0.

Exposure of a precooled solution of [1]B(C₆F₅)₄ (0.39 mM, 2-MeTHF, −135 °C) to excess O₂ evinced the brilliant green color of [1·O₂]B(C₆F₅)₄. UV/vis spectroscopic monitoring (Figures S1 and S2) revealed growth of features at λ_max 442 nm (ε 3700 M⁻¹ cm⁻¹), 642 nm (ε 1525 M⁻¹ cm⁻¹), and 742 nm (ε 1675 M⁻¹ cm⁻¹),⁴⁹ matching well with those of [2·O₂]⁺.⁴⁸ Quantification was achieved by variable temperature O₂ binding studies (Figures 3a and 3b), which revealed that the earlier report of [1]⁺ likely missed superoxide formation because [1·O₂]⁺ only appreciably forms below −90 °C. Interestingly, the difference in temperature profiles of O₂ binding between [1]⁺ and the N₄-analog [2]⁺ was stark (Figures 3b and S3) and could be fit to show dioxygen binding by [1]⁺ was more endergonic by ΔΔH_[1]–[2] = +3.2 kcal/mol and ΔΔS_[1]–[2] = +3.7 cal/mol·K (see Section 4 in the SI). Cyclic voltammetry showed similar results. Solutions of both [1]B(C₆F₅)₄ and [2]B(C₆F₅)₄ (0.1 mM, 17:3 2-MeTHF/acetone, 0.1 M TBAPF₆, 25 °C) exhibited quasi-reversible features and had half-wave potentials at 150 and −126 mV, respectively (Figure 3c; see Figure S4 for CV data in CH₃CN). This corresponds to a 6 kcal/mol difference between Cu^II/Cu^I reduction potentials, in qualitative agreement with the O₂ binding behavior.

Figure 3.

(a) UV/vis spectra show [1]B(C₆F₅)₄ solutions (0.39 mM, 2-MeTHF) under an O₂ atmosphere to reversibly form [1·O₂]B(C₆F₅)₄. (b) O₂ binding curves (UV/vis, 740 nm) and fits (see the SI). (c) CVs of [1]B(C₆F₅)₄ and [2]B(C₆F₅)₄ (~1 mM, 17:3 2-MeTHF/Acetone, 0.1 M TBAPF₆). (d) Resonance Raman spectra (413 nm, 20 mW) of the ¹⁶O and ¹⁸O samples, along with a difference spectrum (gray). (e) Cu K-edge XAS spectra of [1]B(C₆F₅)₄ (green) and [1·O₂]B(C₆F₅)₄ (orange) with the expanded pre-edge region (inset). (f) Cu K-edge EXAFS data (inset) and nonphase-shift-corrected FT for [1·O₂]B(C₆F₅)₄ (data, solid orange; fits, dotted black).

The presence of a cupric superoxide moiety within [1·O₂]B(C₆F₅)₄ was confirmed by resonance Raman (rR) spectroscopic studies on a frozen 2-MeTHF sample (Figure 3d). Laser excitation at 413 nm revealed resonantly enhanced bands at 446 and 1105 cm⁻¹ that shifted upon ¹⁸O₂ labeling (425, 1042 cm⁻¹, respectively). The 446 cm⁻¹ band can be assigned as the Cu–O stretch and the 1105 cm⁻¹ band as the superoxide O–O stretch in line with other η¹-end-on-bound superoxides, including [2·O₂]+.^{26,32,33,45,50–52}

Importantly, this assignment left unresolved the question of a Cu^II–S interaction in [1·O₂]⁺. To address this, and as previously successfully applied,^41,42 Cu K-edge X-ray absorption spectroscopy (XAS) studies were undertaken on frozen 2-MeTHF solutions of both [1]B(C₆F₅)₄ and [1·O₂]B(C₆F₅)₄ (Figure 3e). An electric dipole-forbidden quadrupole-allowed 1s–3d transition pre-edge feature was observed for [1·O₂]⁺ at ~8979 eV, indicative of significant Cu^II character.⁵³ In contrast, [1]⁺ shows an intense feature at ~8984 eV from the electric dipole-allowed 1s–4p transition expected of Cu^I complexes.⁵⁴ The absence of this feature in [1·O₂]⁺ demonstrates little-to-no residual [1]⁺.

The Cu K-edge EXAFS data and corresponding Fourier transforms (FT) provided geometrical insight (Figure 3f). Compared to [1]⁺, [1·O₂]⁺ exhibits shifts in the EXAFS beat pattern to higher k and in the first-shell FT peak to lower R (SI Figure S5b), suggesting a general contraction in the first coordination sphere of [1·O₂]⁺. Careful fitting revealed this contraction was not uniform (SI Table S2): [1]⁺ had 3 Cu–N/O at 2.01 Å and 1 Cu–S at 2.26 Å, while [1·O₂]⁺ had 4 Cu–N/O at 1.98 Å and 1 Cu–S at 2.55 Å. Thus, the Cu–S bond lengthened upon O₂ binding. The distances agree with XRD of [1]B(C₆F₅)₄ and related copper–dialkylthioether distances^39,55,56 in the Cambridge Structural Database (Cu^I⋯S 2.24–2.43 Å, Cu^II⋯S 2.30–2.74 Å).⁵⁷ The relatively high σ² for the Cu–S path of [1·O₂]⁺ (SI Table S2) reflects higher disorder, reasonable considering the fairly long interatomic distance. The Cu–S contributions were essential to achieve good fits; fits for [1·O₂]⁺ replacing the Cu–S interaction with Cu⋯C or Cu⋯O scatterers were poorer, most noticeably in the ~2 Å FT feature and the larger errors (Figure S5e–f and Table S3).

Density functional theory (DFT) computations supported these structural features. Geometry optimizations of [1]⁺ and [1·O₂]⁺ (Figure 2b) with ORCA 4⁵⁸ at the B3LYP-D3(BJ)/Def2-TZVP level of theory^59–65 and a tetrahydrofuran conductor-like polarizable continuum model⁶⁶ predicted an end-on CuO₂^•− triplet with a CuOO angle of 113° and similar lengthening of the Cu–S bond upon O₂ binding (2.32 to 2.50 Å). The calculated thermodynamics of O₂ binding also agreed reasonably with experiment (vide supra), showing ΔΔH_[1]–[2] = +1.3 kcal/mol and ΔΔS_[1]–[2] = +1.7 cal/mol·K. The calculated vibrations (Cu–O 378 cm⁻¹, O–O 1173 cm⁻¹) also comport with our Raman assignments.

Beyond its demonstrable Cu^II–S interaction, [1·O₂]B-(C₆F₅)₄ was remarkable for its stability. Monitoring its formation by UV/vis spectroscopy showed no decay or dimerization over 4 h,^32,49 unlike the previously reported cases with sulfur-substituted ligand frameworks. Cycling a solution from −135 °C to +25 °C and back again gave ~95% recovery of [1·O₂]⁺, illustrating clean and reversible dioxygen binding (see Figure S2).

Such stability enabled further reactivity studies. Hydrogen atom abstraction (HAA) by [1·O₂]⁺ was probed using 2,2,6,6-tetramethylpiperidin-1-ol (TEMPO–H; Scheme 2).⁶⁷ Monitoring the 742 nm band by UV/vis spectroscopy (2-MeTHF, −135 °C) in the presence of excess TEMPO-H⁶⁸ (25–200 equiv) revealed [1·O₂]⁺ to decay isosbestically by pseudo-first-order kinetics (Figure 4), allowing extraction of a second-order rate constant k₂ = 1.28 × 10⁻¹ M⁻¹ s⁻¹ (see SI Figures S6–S7 and Table S4), to the putative cupric hydroperoxide [1·OOH]⁺ (330 nm, 400 nm) in >94% yields (see Section 10 in the SI). Deuteration (100–750 equiv of TEMPO–D⁶⁹) slowed the decay of [1·O₂]⁺ with k_D = 0.24 × 10⁻¹ M⁻¹ s⁻¹, revealing a primary kinetic isotope effect (KIE) of k_H/k_D = 5.4 (see Figures S8–S9), indicating a rate-limiting HAA process. To probe the effect of S_thioether substitution on the HAA reactivity with the O–H bond of TEMPO–H, the rates of the reaction of [2·O₂]⁺ were reevaluated in 2-MeTHF at −135 °C and contrasted with that of [1·O₂]⁺.⁴⁸ The obtained pseudofirst-order rate constants (k_obs) were plotted (see Figures S10–S11 and Table S5) to yield a second-order rate constant k₂ = 0.82 × 10⁻¹ M⁻¹ s⁻¹ indicating a slightly slower HAA reactivity by [2·O₂]⁺ (Figure S12). Quantitation of the products of the HAA reaction between [1·O₂]⁺ and TEMPO–H, i.e., [1·OOH]⁺ and TEMPO^•, was carried out by showing that the hydroperoxide [1·OOH]⁺ protonates to give H₂O₂ (see Figures S13–S14) and that electron paramagnetic resonance (EPR) spectroscopy reveals the production of Cu^II in [1·OOH]⁺ along with the TEMPO^• radical in an ~1:1 ratio (see Figure 4 above). These results illustrate that [1·O₂]⁺ performs HAA both cleanly and competently. Thus, [1·O₂]⁺ is the first thioether Cu_M model to demonstrate Cu_M reactivity.

Scheme 2.

HAA from TEMPO–H by [1·O₂]⁺

Figure 4.

(a) UV/vis (2-MeTHF, −135 °C) and (b) EPR spectroscopic monitoring of the reaction between [1·O₂]⁺ and TEMPO–H to generate [1·OOH]⁺ and TEMPO^•. See Figure S15 for EPR simulation and further discussion.

With this report, we have outlined the stability and reactivity of [1·O₂]⁺, the first Cu_M model complex to both unambiguously possess a Cu^II–S interaction and perform Cu_M-like HAA. Careful selection of solvent and temperature was crucial to successfully access this superoxide complex. Ongoing research includes studies to obtain deeper insights into the electronic structure of [1·O₂]⁺ and their correlation to [2·O₂]⁺. This system will be a valuable point of comparison for future work targeting the influence of a sulfur-based ligand in Cu^I/O₂ oxidative reactivity including the methionine residue at the Cu_M active site of noncoupled binuclear copper monooxygenases.