Mechanism of O–Atom Transfer from Nitrite: Nitric Oxide Release at Copper(II)

Abstract

Nitric oxide (NO) is a key signaling molecule in health and disease. While nitrite acts as a reservoir of NO activity, mechanisms for NO release require further understanding. A series of electronically varied β-diketiminatocopper(II) nitrite complexes [Cu^II](κ²-O₂N) react with a range of electronically tuned triarylphosphines PAr^Z₃ that release NO with the formation of O=PAr^Z₃. Second-order rate constants are largest for electron-poor copper(II) nitrite and electron-rich phosphine pairs. Computational analysis reveals a transition-state structure energetically matched with experimentally determined activation barriers. The production of NO follows a pathway that involves nitrite isomerization at Cu^II from κ²-O₂N to κ¹-NO₂ followed by O-atom transfer (OAT) to form O=PAr^Z₃ and [Cu^I]-NO that releases NO upon PAr^Z₃ binding at Cu^I to form [Cu^I]-PAr^Z₃. These findings illustrate important mechanistic considerations involved in NO formation from nitrite via OAT.

Graphical Abstract

INTRODUCTION

Nitric oxide (NO) serves as a signaling molecule involved in a wide array of physiological functions, such as its role as a vasodilator connected with blood pressure and cardiovascular health.¹ While endothelial nitrogen oxide synthase (NOS) produces NO from arginine, this enzyme loses activity when the oxygen levels become low.² Tightly regulated reduction of nitrite to form NO can compensate for this loss of NOS activity under hypoxia.^3–5 The enzymes deoxymyoglobin, cytochrome c oxidase, and xanthine oxidase are each able to act as nitrite reductases.^6–9

Cu-containing nitrite reductases are common in soil bacteria where they perform a key denitrification step in the nitrogen cycle by converting nitrite to NO. In mammalian biology, Cu has been implicated in the reverse transformation, converting NO to nitrite at the redox-active Cu site of ceruloplasmin.¹⁰ Interestingly, a recent report indicates that the ubiquitous carbonic anhydrase II enzyme may act as a nitrite reductase when complexed with Cu rather than Zn, generating NO.¹¹

While one-electron reduction of nitrite to NO may occur in the presence of proton sources to generate water, O-atom transfer (OAT) pathways also exist.¹² The former has been well characterized in biological^13,14 and biomimetic systems;^15–19 however, OAT to nucleophiles is less studied.^6,12,20–22 Potential O-atom acceptors such as thiols RSH and dialkyl sulfides RSR′ are readily biologically available and may be oxidized to sulfenic acids RS(O)H and sulfoxides RS(O)R′ (Figure 1A).

Figure 1.

(A) OAT reactions of metal complexes with nitrite. (B) Metal scaffolds supporting OAT from nitrite. (C) Calculated OAT to the SMe₂ transition state.

A range of Fe, Co, and Cu biomimetic complexes demonstrate OAT with a range of nucleophiles (Figure 1). The oldest of these models, which dates to 1979, involves a cobalt nitro complex supported by a dianionic tetraazacyle that undergoes reduction by PPh₃ to generate the corresponding cobalt nitrosyl and O=PPh₃.²³ Most model complexes have focused on iron/cobalt porphyrin complexes using a variety of O-atom acceptors,^24–33 although other coordination motifs have also been examined (Figure 1B).^12,21,34,35 These synthetic models typically possess a nitrite ion bound through the N atom. As shown by computational studies with an iron porphyrin system, [M](κ¹-NO₂) structures can enable smooth conversion to a metal nitrosyl [M]-NO upon OAT with SMe₂ with modest barriers, ΔG^⧧(298 K), calculated in either the absence (10.4 kcal mol⁻¹) or presence of an axial ligand L (17.8 kcal mol⁻¹; L = pyridine; Figure 1C).³⁶

Two recent reports document OAT from nitrite at Cu^II that results in NO loss via OAT to PPh₃ with reduction to Cu^I.^12,21 Importantly, these complexes each possess O-bound nitrite with the [Cu^II](κ²-O₂N) binding mode. To better understand OAT from nitrite at Cu, we examine herein the electronic role of the Cu center and incoming O-atom acceptor through kinetic studies supported by computational analysis to develop a reaction pathway for NO release from nitrite at Cu^II.

RESULTS AND DISCUSSION

We previously demonstrated that the β-diketiminatocopper(II) nitrite [ⁱPr₂NN_F6]Cu^II (κ²-O₂N) reacts with PPh₃ to release NO with the formation of [ⁱPr₂NN_F6]Cu^I(PPh₃) and O=PPh₃ .¹² To examine the electronic roles of the Cu center and the phosphine that serves as the O-atom acceptor, we sought to employ a range of sterically similar, yet electronically different, β-diketiminatocopper(II) nitrite complexes [Cu^II](κ²-O₂N) and para-substituted triarylphosphines PAr^Z₃. To modulate the electronic environment of the copper(II) nitrite complexes with minimal steric impact, we employed β-diketiminate ligands with backbone substituents X = Me or CF₃ along with N-aryl o-Cl or o-Me substituents. The reaction of copper(I) β-diketiminates [Cu^I] with AgNO₂ serves as a reliable route to the corresponding copper(II) nitrites [Cu^II](κ²-O₂N) (1−4; Scheme 1).

Scheme 1.

Synthesis of Copper(II) Nitrite Complexes 1–4 and X-ray Structure of Complex 2

Cyclic voltammetry experiments of the copper(II) nitrites 1–4 performed in tetrahydrofuran (THF) reveal quasi-reversible waves that correspond to a range of reduction potentials that span 0.33 V (vs NHE) for 1 (backbone CF₃ and N-aryl Cl substituents) to −0.11 V (vs NHE) for 4 (backbone Me and N-aryl Me substituents) (Table 1). Copper(II) nitrite complexes 1–4 are dark green in color, possessing UV–vis spectra in toluene with λ_max values around 600 nm. This can be used to follow NO release upon reaction with PAr^Z₃, which results in reduction of the Cu center to give essentially optically silent [Cu^I]-PAr^Z₃ complexes (Scheme 2). The reaction of 1 with 2 equiv of $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ resulted in a yield of 72% $\left[{\text{Cl}}_2{\text{NN}}_{{\text{F}}_6}\right]\text{Cu}-\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ by ¹⁹F NMR (Figure S19).

Table 1.

Reduction Potentials and OAT Activation Parameters with $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ for Copper(II) Nitrites 1–4

[Cu]	E1/2 (V vs NHE) in THF	ΔH⧧ (kcal mol−1)	ΔS⧧ (cal mol−1 K−1)	ΔG⧧(298 K) (kcal mol−1)
1	0.3345	4.6 ± 0.7	−38.5 ± 3.1	16.0 ± 1.2
2	0.30	5.7 ± 0.5	−34.5 ± 2.4	16.0 ± 0.9
3	−0.02	13.9 ± 0.5	−13.3 ± 1.8	17.9 ± 0.8
4	−0.11	13.1 ± 0.2	−18.5 ± 0.8	18.6 ± 0.3

Scheme 2.

NO Release from Cu^II-Bound Nitrite 1

Demonstrating NO release, copper(II) nitrite 1 reacts with 2 equiv of PPh₃ or $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ in toluene to provide the corresponding [Cu^I]-PAr₃ complex. Capture of NO by the cobalt(II) porphyrin meso-tetra(4-methoxyphenyl)-porphyrincobalt(II) [T(OMe)PP]Co gives the diamagnetic [T(OMe)PP]Co(NO) amenable to ¹H NMR analysis, which shows 64% and 59% yield respectively for PPh₃ and $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ (Scheme S2).³⁷ Consistent with the 1:2 [Cu^II](κ²-O₂N)/PAr₃ stoichiometry, the reaction of ${[\text{Cl}}_2{\text{NN}}_{{\text{F}}_6}\left]\text{Cu}\right({\kappa }^2-{\text{O}}_2\text{N})$ (1) with 1 equiv of $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ reduces the yields of $\left[{\text{Cl}}_2{\text{NN}}_{{\text{F}}_6}\right]\text{Cu}-\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ and NO to 46% and 51%, respectively. Because the NO that is released can react with copper(I) β-diketiminates [Cu^I] to reform copper(II) nitrites such as [Cu^II](O₂N)[Cu^I] and N₂O,¹² this may represent a pathway that competes with the trapping of [Cu^I] by PAr₃.

To establish the rate law, we followed the reaction of $\left[{\text{Me}}_2{\text{NN}}_{{\text{F}}_6}\right]\text{Cu}\left({\kappa }^2-{\text{O}}_2\text{N}\right)$ (2; ca. 2.0 mM) with excess $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ in toluene by UV–vis spectroscopy by monitoring the loss of the band at λ_max = 600 nm for 2. We chose the electron-poor phosphine $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ to slow the reaction enough to enable a study by straightforward UV–vis kinetics at −60 °C in the presence of 25–50 equiv of phosphine. These studies reveal pseudo-first-order decay of 2 with observed rate constants k_obs that vary linearly with $\left[\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3\right]$ (Figures S25 and S26) to give the overall second-order rate law $k\left[\text{Cu}\left({\text{O}}_2\text{N}\right)\right]\left[\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3\right]$, where k(−60 °C) = 0.27(2) M⁻¹ s⁻¹.

Eyring analysis allowed for quantification of the activation parameters for each electronically different copper nitrite complex 1–4 (Figure 2). We obtained second-order rate constants at various temperatures under pseudo-first-order conditions with 20 equiv of $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$(Figure S27). The very different rates of the reaction require Eyring analysis over different temperature spans in order to conveniently follow these reactions by UV–vis spectroscopy. For instance, we employed a temperature range of −60 to −30 °C in the reaction of 1, while we used a range of 15–45 °C for less reactive 4. These studies reveal that the experimental enthalpy and free energies of activation ΔH^⧧ and ΔG^⧧(298 K) generally increase with decreasing reduction potential of the copper(II) nitrites 1–4 (Table 1 and Figures S29 and S30). The β-diketiminato backbone substituent exerts the most prominent effect: ΔH^⧧ increases from ca. 5 kcal mol⁻¹ (X = CF₃) to 13–14 kcal mol⁻¹ (X = Me). We observe a smaller range of free energies of activation ΔG^⧧(298 K) = 16–19 kcal mol⁻¹ that reflect more negative entropies of activation observed with X = CF₃ versus Me (Figure 2).

Figure 2.

Free energies of activation ΔG^⧧ (kcal mol⁻¹) at 298 K for the reaction of 1–4 with excess $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ versus reduction potential of copper(II) nitrite. Bars represent standard errors in ΔG^⧧.

Employing compound 3 as a model along with a set of para-substituted phosphines PAr^Z₃, we monitored the electronic influence on the rate of reaction. As can be seen from the Hammett plot (Figure 3), electron-rich phosphines accelerate the reaction [ρ = −1.5(6)]. Thus, the combination of an electron-poor Cu center with an electron-rich phosphine favors nitrite reduction at Cu^II. Among the phosphines studied, this effect represents a nearly 20-fold increase in the second-order rate constant.

Figure 3.

Hammett analysis of the reaction of copper(II) nitrite 3 with para-substituted triarylphosphines P(Ar^Z)₃.

Guided by the experimental rate law and activation parameters in the generation of NO from β-diketiminatocopper(II) nitrites 1–4, we sought to uncover further details of the reaction pathway through density functional theory (DFT) computational analysis. We chose to focus on the synthetic model 4 because its experimental parameters had the lowest estimated errors. In accordance with previous DFT complexes on copper β-diketiminato complexes, we carried out calculations at the BP86+GD3BJ/6–311+ +G(d,p)/SMD-toluene//BP86/6–311+G(d)/gas level of theory.

Direct attack of the phosphine $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ on the copper(II) nitrite 4 with a κ²-O₂N nitrite binding mode led to a transition state much higher in energy [ΔG^⧧(298 K)_calc = 31.1 kcal mol⁻¹] than experimentally determined [ΔG^⧧(298 K)_exp =18.6(3) kcal mol⁻¹] (Figure S37). By starting with a κ²-O₂N bonding mode, OAT to phosphine most directly results in a metastable [Cu]-ON isonitrosyl complex,^38,39 calculated to be 20.6 kcal mol⁻¹ higher in free energy than the corresponding three-coordinate nitrosyl [Cu^I]-NO (Scheme S3).

There are several crystallographically determined binding modes for nitrite at Cu centers, which include κ²-O₂N, κ¹-ONO, and κ¹-NO₂ (Figure 4).^12,13,40,41 At copper(II) complexes of lower coordination number, nitrite generally prefers the κ²-O₂N binding mode in both synthetic complexes^21,42 as well as Cu-containing nitrite reductases.²² Tetradentate supporting ligands, however, can lead to the κ¹-ONO binding mode.^43,44 On the other hand, copper(I) complexes typically display the κ¹-NO₂ binding mode^12,41 because of the availability of back-bonding into the NO₂ π* orbitals (Figure S75). While the β-diketiminatocopper(II) nitrites in this study exhibit κ²-O₂N binding modes in the solid state,^12,45 a recent computational study indicates that both κ¹-ONO and κ¹-NO₂ binding modes are energetically accessible at Cu^II in the presence of hydrogen bonding.²⁰ Guided by examples in iron porphyrin chemistry for which the [Fe](κ¹-NO₂) binding mode can enable direct transformation to the corresponding [Fe]-NO nitrosyl complex upon OAT,³⁶ we were eager to consider this κ¹-NO₂ binding mode as a possible intermediate in OAT from nitrite at Cu^II.

Figure 4.

Nitrite bonding modes at Cu.

Computationally examining nitrite isomerization in the copper(II) complex 4, we find that a distorted κ¹-NO₂ binding mode is only 0.9 kcal mol⁻¹ higher in free energy than the κ^2-O₂N ground state with a barrier of ΔG^⧧(298 K) = 8.1 kcal mol⁻¹. This transition state resembles the κ¹-ONO binding mode, which in our model did not result in an optimized stable point (Figure S46). Curiously, this κ¹-NO₂ binding mode is not symmetrical as found in the crystallographically characterized ${\left\{\left[{}^i{\mathrm{Pr}}_2{\text{NN}}_{{\text{F}}_6}\right]\text{Cu}\left({\kappa }^1-{\text{NO}}_2\right)\right\}}^{-}$ in which the nitrite ONO plane is orthogonal to the β-diketiminato backbone.¹² Rather, this κ¹-NO₂ binding mode has a close Cu···O contact of 2.440 Å, which results in a highly distorted square-planar coordination. This distortion unequally polarizes a modest amount of unpaired electron density present at nitrite toward the proximal O atom (0.14 e⁻) at the expense of the distal O atom (0.05 e⁻) (Figure S74).

A scan of the approach of the phosphine to the O atom with greater spin density led to the optimization of a transition state with ΔG^⧧(298 K) = 19.1 kcal mol⁻¹, extremely close to the experimental value of 18.6(3) kcal mol⁻¹. In this transition state, which features κ¹-NO₂ coordination, the nitrite has twisted to become orthogonal to the β-diketiminato plane. This primes an O atom (O1) for abstraction by the incoming phosphine with N–O1 and P–O1 distances of 1.44 and 1.86 Å, respectively, in the transition state. Developing Cu-NO character is apparent through a shortening of the Cu–N (1.85Å) and N–O2 (1.23 Å) bonds that lead to copper(I) nitrosyl 7 with Cu–N (1.78 Å) and N–O2 (1.19 Å) distances, which are slightly bent orthogonal to the β-diketiminato backbone with a Cu–N–O2 angle of 157.9°. The conversion of copper(II) nitrite 4 and $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ reactants to copper nitrosyl 7 and phosphine oxide $\text{O}=\text{P}\left({\text{Ar}}^{{\text{CF}}_3}\right)$ is significantly exergonic at −27.2 kcal mol⁻¹. Moreover, the displacement of NO at Cu^I by phosphine to give the copper(I) phosphine adduct $\left[{\text{Cu}}^{\text{I}}\right]-\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ is further downhill by another 6.7 kcal mol⁻¹ in free energy (Figure 5). This is congruent with our experimental observation of [Cu^I]-PPh₃ in the reaction of [Cu^II](κ²-O₂N) complexes with 2 equiv of PPh₃ (Scheme 2).¹²

Figure 5.

Reaction coordinate diagram of OAT from [Me₂NN]Cu(κ²-O₂N) (4) to $\text{P}{\left({\text{Ar}}^{{\text{CF}}_3}\right)}_3$ via [Me₂NN]Cu-NO (7). Free energies (bold) are in kilocalories per mole at 298 K.

The identification of a DFT transition state structure for OAT from the Cu^II-bound nitrite to a phosphine that matches the experimental activation energies encourages the consideration of biologically relevant O-atom acceptors. For instance, methionine residues are often found in the vicinity of Cu active sites.^46–49 Employing SMe₂ as a simple model, experimentally we find that it does not undergo OAT with β-diketiminatocopper(II) nitrites at room temperature or even with modest heating. DFT analysis that involves a κ²-O₂N-to-κ¹-NO₂ isomerization of copper(II) nitrites 1–4 prior to OAT to SMe₂ reveals calculated free energies of activation that range from 33.0 to 42.1 kcal mol⁻¹ for the four copper nitrite complexes (Scheme 3 and Figure S38). Thus, the ease of oxidation of the incoming O-atom acceptor is a crucial feature of OAT from nitrite at Cu^II.

Scheme 3. Considering OAT from Copper(II) Nitrites 1–4 to SMe₂^a.

aCalculated free energies are in kilocalories per mole at 298 K.

CONCLUSIONS

Cu-containing nitrite reductases have reduction potentials in the range of 0.17–0.28 V (vs NHE) at the type 2 Cu site,^50–54 similar to both copper(II) models 1 and 2, which exhibit reduction potentials of 0.33 and 0.31 V (vs NHE) in THF. This study reveals that increasing the reduction potential of a copper(II) nitrite facilitates OAT from nitrite, a feature that controls the rate of reaction at roughly isosteric models. Moreover, the combination of an electron-poor Cu center with an electron-rich O-atom acceptor proves optimal for OAT. DFT studies benchmarked on the experimental free energy of activation reveal that OAT requires isomerization of the nitrite to a κ¹-NO₂ binding mode, which enables efficient transfer of a nitrite O atom from N to P to form O=PAr₃ along with the copper nitrosyl [Cu]-NO. In contrast to the strong binding of NO at iron(II) porphyrins, the more labile Cu–NO interaction¹⁶ results in NO release, with an additional equivalent of the incoming nucleophile that binds to the Cu^I center.

These studies reveal higher thermodynamic and kinetic barriers for OAT from copper(II) nitrites to more modest O-atom acceptors such as dialkyl sulfides. Nonetheless, turning on the OAT pathways from nitrite to dialkyl sulfides such as methionine commonly found within the coordination sphere of copper enzymes could represent a pathway to connect nitrite and NO with oxidative methionine signaling, a post-translational means to control protein activity.^55–57