Cupric-Superoxo Mediated Inter-Molecular C-H Activation Chemistry

Abstract

A new cupric-superoxo complex [LCu^II(O₂^•−)]⁺, which possesses particularly strong O–O and Cu–O bonding, is capable of intermolecular C-H activation of the NADH analogue 1-benzyl-1,4-dihydronicotinamide (BNAH). Kinetic studies indicate a first-order dependence on both the Cu-complex and BNAH with a deuterium kinetic isotope effect (KIE) of 12.1, similar to that observed for certain copper monooxygenases.

Copper(I) reactions with molecular oxygen play fundamental roles in many chemical and biological processes.^1,2 Copper-dependent proteins perform a diverse array of oxidative and oxygenative reactions. This has inspired considerable efforts in the design of novel ligands and copper coordinated complexes as well as the study of ligand-copper(I) dioxygen adducts to elucidate their structures, electronic characteristics and substrate reactivity.^2–4 Compared to binuclear copper-dioxygen derived species, mononuclear analogues have been synthetically challenging and hence are less understood.^3,5 However, they are fundamentally important and are directly relevant to copper proteins including dopamine-β-monooxygenase (DβM) and peptidlyglycine-α-hydroxylating monooxygenase (PHM).⁶ These enzymes possess a so-called non-coupled binuclear active site,⁷ which comprises two Cu centers separated by ~11Å. Dioxygen binding and substrate hydroxylation occur at one of the copper sites (designated Cu_M). In an important PHM X-ray structure, a dioxygen derived species presumed to be an end-on bound cupric superoxide species (i.e., Cu^II-O-O^•−) resides adjacent to an inhibitory substrate analog.^6c Along with biochemical,6a,b^,8 chemical and computational studies,^5,9 the cupric-superoxo species is thought by many to be the reactive intermediate responsible for initiating oxidation via hydrogen-atom abstraction. However, other species have been considered as important intermediates in enzymatic turnover, either prior to or following substrate attack, including cupric-hydroperoxo (Cu^II-⁻OOH)¹⁰ and high-valent cupryl (Cu^II-O• <—> Cu^III=O) entities.^4b,9c,11

In our own research program, we seek to elucidate the chemical nature of all of these mononuclear species. In this report, we describe the generation and characterization of a new Cu^II(O₂^•−) and an example of substrate C-H activation (i.e., oxidative C–H bond cleavage). To this point, cupric-superoxo complexes have been shown to exhibit phenol O–H bond cleavage reactions,^10a,12 and in one case, Itoh and coworkers¹³ provided evidence for a Cu^II(O₂^•−) mediated intramolecular benzylic C–H oxygenation. Here, for the first time, an intermolecular C–H substrate oxidation reaction has been achieved, with kinetic data clearly implicating the involvement of the Cu^II(O₂^•−) complex in rate-limiting substrate C–H bond cleavage.

The Cu^I complex of a ligand previously employed by Masuda and coworkers,^14a L [bis(pyrid-2-ylmethyl){[6-(pivalamido) pyrid-2-yl]methyl}amine], [LCu^I]⁺ (as the B(C₆F₅)₄⁻ salt,¹⁴ was exposed to O₂ (by bubbling via a syringe needle) in 2-methyltetrahydrofuran (MeTHF) solvent at −125 °C to form the adduct [LCu^II(O₂^•−)]⁺ (1) (Figure 1) with UV-Vis absorptions [λ_max = 410 nm (3700 M⁻¹ cm⁻¹), 585 nm (900 M⁻¹ cm⁻¹), 741 nm (1150 M⁻¹ cm⁻¹)]¹⁵ characteristic of a mononuclear end-bound Cu^II-superoxo complex. This well below −80 °C temperature is necessary in order to observe this Cu/O₂ = 1:1 adduct. Under these conditions, this green EPR-silent species is quite stable, decaying only very slowly (½ life > 4 hrs) with conversion to [{(LCu^II)}₂(O₂²⁻)]²⁺ (2), a μ-1,2-peroxodicopper(II) complex [λ_max = 541 nm (9,900 M⁻¹cm⁻¹)], that is observed when [LCu^I]⁺ is oxygenated at −80 °C.^14b

Figure 1.

Left: Representation of the cationic portion of the X-ray structure of [LCu^I](B(C₆F₅)₄⁻), revealing the N₄O_(amide) coordination. (Cu-O = 2.190 Å) Right: Calculated structure (see Supporting Information) of [LCu^II(O₂^•−)]⁺ (1), indicating the H-bonding interaction between the ligand and the superoxo β-oxygen atom. See text for further discussion.

A resonance Raman (rR) spectrum (excitation at 413 nm, 77 K) of 1 in MeTHF frozen solution is shown in Figure 2. The O-O stretch is observed at 1130 cm⁻¹ as a single peak when the complex was formed with ¹⁶O₂ (Figure 2B, red); but upon ¹⁸O₂ isotopic substitution, two features are observed (Figure 2B, blue). This behavior is consistent with a Fermi resonance of the ¹⁸O₂-vibration with a non-enhanced mode at similar energy. Using the energy and intensity of the two observed mixed-modal features, the pre-interaction energy of the resonantly enhanced ¹⁸O-¹⁸O stretch was calculated to be 1067 cm⁻¹. Thus the O-O stretch shifts down in energy by 63 cm⁻¹ upon ¹⁸O₂-substitution, consistent with a bound superoxo species.¹⁶ For the lower energy region, both the ¹⁶O₂ and ¹⁸O₂ data show two peaks with an intensity distribution which changes with isotope. Therefore these are both mixed due to Fermi resonance. Analysis¹⁵ of the lower energy region (Figure 2A) leads to a Cu-¹⁶O stretch at 482 cm⁻¹, which shifts to 462 cm⁻¹ with ¹⁸O.¹⁷

Figure 2.

Solvent-subtracted rR spectra of MeTHF solutions of 1 (λ_ex=413 nm). (A) v(Cu-O) region. (B) v(O-O) region. Red, ¹⁶O₂; blue, ¹⁸O₂.

Thus the rR spectroscopic data confirm the formulation of [LCu^II(O₂^•−)]⁺ as an end-on superoxo-containing complex, with O-O and Cu-O stretches of 1130 cm⁻¹ and 482 cm⁻¹, respectively.¹⁵ Notably, these values are higher than those found for all cupric superoxo complexes previously described: ν(O-O) for the one structurally defined side-on bound cupric superoxo complex is 1043 cm⁻¹, while those for the end-on bound species range up to 1122 cm⁻¹; ν(Cu-O) varies from 422 to 474 cm⁻¹.¹⁵ Using DFT calculations, we have evaluated the Cu-O and O-O vibrational frequencies of 1, those for this structure with the pivalamido group at the para instead of the ortho position, and for a structure with Cu-N bonds constrained but no pivalamido group.¹⁵ The comparison between para and ortho substitution eliminates an inductive effect from the pivalamido group as the origin of the higher frequencies. Consistent with the ν(O-O) of 1130 cm⁻¹, 1 is calculated to be an end-on bound Cu(II)-superoxo species with a triplet ground state (the singlet/triplet splitting is calculated to be 1581 cm⁻¹,¹⁵ after correction for spin contamination). For L and its close analogs, Masuda and coworkers^14a have found that the pivalamido pyridyl substituent forms intramolecular H-bonds to the alpha oxygen atoms of a peroxide and/or hydroperoxide moiety ligated to the copper(II),^14a,18 or to the alpha N-atom for an azide coordinated to copper(II).^14a Our calculations¹⁵ suggest that the superoxo moiety in [LCu^II(O₂^•−)]⁺ (1) forms an intramolecular H-bond, however with either the alpha- or beta-oxygen. This would contribute to the relative stability of 1, and can account for the higher O-O and Cu-O frequencies. rR data on the analogs [Cu^II(TMPA)(O₂^•−)]⁺ (TMPA = tris(2-pyridylmethyl)amine)) and [Cu^II(NMe₂-TMPA)(O₂^•−)]⁺ show a ν(O-O) of 1120 cm⁻¹ and a ν(Cu-O) of 472 cm⁻¹;¹² thus ν(O-O) and ν(Cu-O) are both about 10 cm⁻¹ higher in 1. Optimized geometries of [Cu^IITMPA(O₂^•−)]⁺, [Cu^II(NMe₂-TMPA)(O₂^•−)]⁺ have also been obtained through DFT calculations and compared to 1. These calculations give an increase in both ν(Cu-O) and ν(O-O) when H bonding is included in 1, particularly to the beta oxygen of the superoxo ligand.^15,19 The calculated structure shows that Cu-O bond length decreases by 0.005 Å indicating that a slightly stronger Cu-O bond is associated with the higher ν(Cu-O). From the DFT calculation of the structure with Cu-N bond lengths constrained at the values of the optimized structure of 1 but with no pivalamido group, this appears to reflect a distortion of the N_equatorial ligand system which decreases its donor interaction with the Cu. The donor interaction of the superoxo with the Cu thus increases and this leads to a stronger Cu-O bond. Alternatively, the calculations show that the O-O bond actually increases by 0.005 Å in 1, indicating that the increase in ν(O-O) relative to [Cu^IITMPA(O₂⁻)]⁺ does not reflect a stronger O-O bond but derives from structural coupling of vibrations within the ligand system due to the H-bond. This H-bonding stabilizes the superoxo species (relative to the binuclear peroxo species) and allows its reactivity to be studied.

[LCu^II(O₂^•−)]⁺ (1) is unreactive toward a number of commonly employed C–H substrates, such as dihydroanthracene, xanthene and 10-methyl-9,10-dihydroacridine – substrates possessing C-H bonds significantly weaker than those found for DBM and PHM substrates (dopamine, 85 kcal/mol; hippuric acid, 87 kcal/mol).^6b However, the addition of an excess of 1-benzyl-1,4-dihydronicotinamide (BNAH) – an NADH analogue, which is both a strong H-atom (H^•) and hydride (H⁻) donor,¹⁹ to solutions of [LCu^II(O₂^•−)]⁺ leads to the latter’s decay, as observed by UV-Vis spectroscopy (Figure 3a). Kinetic interrogation of this reaction (−125 °C) showed pseudo first-order decay behavior for [LCu^II(O₂^•−)]⁺ (1) (for 10–40 eq. BNAH). The decay was also first order in [BNAH]. Thus, the cupric superoxo is responsible for promoting the substrate oxidation (vide infra). A second-order rate constant of 0.19 M⁻¹s⁻¹ was obtained (Figure 3b); when the substrate was deuterated in the 4,4’ position, i.e., 1-benzyl-1,4-dihydro[4,4’-²H₂]nicotinamide (BNAD), a significant slowing of the reaction occurred, k = 0.016 M⁻¹s⁻¹ (Figure 3b). This gives a kinetic isotope effect (KIE) of 12.1. This KIE value is comparable to that (KIE = 10) reported for C-H bond cleavage of BNAH by a trans-dioxomanganese(V) porphyrin.²⁰ Product analysis of the [LCu^II(O₂^•−)]⁺/BNAH reaction (following quenching with HCl at −130 °C)¹⁵ confirms that BNAH has undergone oxidation by 1. The substrate’s 4’ C-H bond has been oxidatively cleaved to form 1-benzylnicotinamidium ion (BNA⁺) in 42% yield (¹H-NMR), based on the initial copper concentration (Scheme 1). Additionally, upon acidification, liberated hydrogen peroxide is also detected in ~30% yield, approximately corresponding to the amount of peroxo-dicopper(II) complex [{(LCu^II)}₂(O₂²⁻)]²⁺ (2) that is also a reaction product, identified by its characteristic UV-Vis absorption bands (vide supra).

Figure 3.

(a) Spectral changes of 0.4 mM [LCu^II(O₂^•−)]⁺ in the presence of 8 mM BNAH at −125°C in MeTHF. The first spectrum recorded (green) is that immediately following O₂-bubbling to a solution of [LCu^I]⁺ + BNAH. Inset: pseudo first order kinetics fit from 741 nm data. (b) Plots of k_obs as a function of BNAH, BNAD, and BzIMH concentration used to determine second order rate constants.

Scheme 1.

Cupric superoxo-promoted cleavage of the C-H bond likely follows one of two possible pathways (Scheme 1). One is initial H-atom transfer (HAT), in which the C-H bond cleaves homolytically (with the thus-formed BNA-radical rapidly losing a second electron); the other, hydride transfer resulting from heterolytic cleavage. To provide further mechanistic insight into the mode of C-H activation by 1, a second substrate – 1,3-dimethyl-2,3-dihydrobenzimidazole (BzImH)²¹ – was studied for reactivity. Like BNAH, BzImH possesses a weak C-H bond, but with markedly different bond strengths compared to BNAH (homolytic, BDE = 73.4 kcal/mol, compared to 70.7 kcal/mol for BNAH) and heterolytic (hydride affinity of 49.5 kcal/mol, compared to 64.2 kcal/mol for BNAH)).²¹ Kinetic studies reveal that the oxidation of BzImH occurs ~2.4 times slower than that of BNAH, with a second order rate constant of 0.078 M⁻¹s⁻¹ (Figure 3b). The slower rate of C-H oxidation of BzImH (stronger H⁻ donor) vs BNAH (stronger H• donor) thus suggests that at least for these substrates, the preferred mode of C-H activation by [LCu^II(O₂^•−)]⁺ is via rate-limiting homolytic C-H bond cleavage; that is, H-atom transfer (HAT).

Based on extensive studies of the enzymes PHM/DβM, the most accepted mechanistic proposal is that these enzymes initially proceed by (non-classical)^6b,22 hydrogen-atom transfer (HAT) chemistry, promoted by a cupric superoxo generated at the Cu_M site.^8a,9a Interestingly, enzyme kinetic studies carried out on DβM and PHM reveal KIE’s of 10–14.^6,23 (depending on conditions), thus very similar to that found here for our system. Our studies reported here with a “model” cupric superoxide complex suggest that this mechanism is followed, at least for the relatively weak C-H substrates investigated. In one case, theoretical calculations inspired by results on a model system led to the suggestion that an initial hydride abstraction may occur for PHM/DβM.²⁴ Our findings here suggest this is not likely to be a favorable pathway.

Following C-H activation, the overall mechanism of substrate hydroxylation by the enzymes DβM/PHM is poorly understood, with nearly as many proposals for the subsequent steps as there are researchers in the field.¹⁵ The steps to products following the transition state are also difficult to surmise in our chemical systems.²⁶

Despite these uncertainties, the importance of the present studies lies in the observation of intermolecular C-H activation by a cupric superoxo complex. To summarize, a new-ligand supported cupric superoxo complex with reactivity behavior of relevance to the DβM and PHM enzymes has been described. Significant advances are: (a) [LCu^II(O₂^•−)]⁺ (1) is the first cupric superoxo complex which incorporates a hydrogen-bonding ligand feature; (b) this results in a compound with stronger Cu-O and O-O bonds compared to previously known examples; (c) not necessarily related to the latter properties, the reactions of 1 with BNAH and BzImH provide the first examples of inter-molecular C-H bond activation by a cupric superoxo complex. Note that it is not unexpected that relatively weak C–H bonds are cleaved for an inter-molecular situation; one does not have the advantage of a proximate substrate, as found for the known single example of intramolecular Cu^II-O₂^•− mediated C–H oxidation,¹³ and finally, (d) the relative rates for C-H oxidation of these two substrates support rate-limiting homolytic C-H bond activation. Further corroborating studies with other C-H mechanistic probes will be investigated. The clean kinetic behavior, striking deuterium kinetic isotope effect (KIE), and determination of rate-limiting HAT demonstrate that the reaction of [LCu^II(O₂^•−)]⁺ (1) with C-H substrates may possess biological significance in direct comparison to the DβM/PHM enzymes reaction mechanism.