Copper(I)-Dioxygen Adducts and Copper Enzyme Mechanisms

Abstract

Primary copper(I)-dioxygen (O₂) adducts, cupric-superoxide complexes, have been proposed intermediates in copper-containing dioxygen-activating monooxygenase and oxidase enzymes. Here, mechanisms of C–H activation by reactive copper-(di)oxygen intermediates are discussed, with an emphasis on cupric-superoxide species. Over the past 25 years, many synthetically derived cupric-superoxide model complexes have been reported. Due to the thermal instability of these intermediates, early studies focused on increasing their stability and obtaining physical characterization. More recently, in an effort to gain insight into the possible substrate oxidation step in some copper monooxygenases, several cupric-superoxide complexes have been used as surrogates to probe substrate scope and reaction mechanisms. These cupric superoxides are capable of oxidizing substrates containing weak O–H and C–H bonds. Mechanistic studies for some enzymes and model systems have supported an initial hydrogen-atom abstraction via the cupric-superoxide complex as the first step of substrate oxidation.

1. Introduction

Copper plays a crucial role in biology, being the third most abundant transition metal in living organisms.^[1] Among the many physiological functions performed by copper-containing enzymes, dioxygen activation is perhaps one of the most important aspects to understand. Oxygenase and oxidase enzymes couple the thermodynamic driving force of O₂ reduction to incorporate oxygen atoms into unactivated bonds or drive endergonic processes such as ATP synthesis.^[2] Dioxygen-utilizing copper-containing proteins are found in mononuclear, homo-multinuclear (coupled and noncoupled), and hetero-multinuclear (including iron or zinc) active sites.^[3,4] The study of dioxygen-processing metallozymes containing binuclear copper active sites, such as hemocyanin, tyrosinase, and particulate methane monooxygenase,^[5] is not only important in elucidating biochemical mechanisms, but it is also a crucial aspect towards deciphering the chemistry of copper and dioxygen. However, the emphasis of this article will be placed on enzymes and synthetic model chemistry involving single copper sites.

Formation of mononuclear copper-dioxygen derived intermediates, including a cupric-superoxide (O₂^•−) species, is invoked (or in some cases observed) in the catalytic cycles of certain enzymes, despite differences in active-site environments and enzymatic functions. These enzymes include the noncoupled binuclear enzymes peptidylglycine α-hydroxylating monooxygenase (PHM),^[6,7] dopamine-β-monooxygenase (DβM),^[6,7] and tyramine-β-monooxygenase (TβM),^[7] and the mononuclear enzymes lytic polysaccharide monooxygenases (LPMOs),^[8,9] galactose oxidase (GO),^[10] and copper amine oxidases (AOs).^[10]

Due to the ephemeral nature of the intermediates formed in enzymatic systems, synthetic models have been used to investigate the specific copper-(di)oxygen species that potentially are responsible for the biologically observed chemistry. The use of small molecule inorganic model complexes allows for precise control and subsequent deduction of those electronic and geometric factors which govern copper-dioxygen adduct stability and reactivity. Careful selection of the model system and the reaction conditions may allow for the generation, and physical and chemical characterization of proposed copper-(di)-oxygen intermediates. Studying the chemistry of these synthetic analogues may reveal complementary information about the reaction pathways occurring in these copper-containing enzymes. This information may thereby implicate a particular intermediate as part of a given enzymatic cycle, or yield new insights for known processes, which may be unobtainable from the study of proteins under physiological conditions.

2. Enzymatic Systems Involving Cu^II(O₂^•−)

2.1 Noncoupled Binuclear Systems

PHM, DβM, and TβM are members of a small family of enzymes responsible for the biogenesis of neurotransmitters. PHM and DβM are found in mammals, while TβM is the homologue of DβM for invertebrates.^[6,11]

PHM along with peptidylglycine α-hydroxyglycine α-amidating lyase (PAL) make up the two domains of the bifunctional enzyme peptidylglycine α-amidating monooxygenase (PAM), which catalyzes the α-amidation of the peptide backbone of a glycine extended prohormone.^[12] PHM (monooxygenase domain) catalyzes the hydroxylation of the C_α of a C-terminal glycine, while PAL (lyase domain) mediates the disproportionation of the α-hydroxyglycine hydroxylated product (Figure 1).^[13]

Figure 1.

A) X-ray structure of the oxy-form of PHM.^[13] B) Catalytic reactions of copper-containing monooxygenases PHM, DβM, and TβM.^[7]

DβM plays an important role in the physiological regulation of neurotransmitters, specifically dopamine and norepinephrine. DβM catalyzes the conversion of dopamine to norepinephrine through oxygen atom insertion.^[6] In the same way, in insects, TβM catalyzes the hydroxylation of tyramine to octopamine (Figure 1).^[11]

PHM, DβM, and TβM share similarities, such as the conservation of six amino-acid residues coordinated to copper centers (five His and one Met), despite their different substrate specificities. Even more remarkably, kinetic and mechanistic studies support the idea that these enzymes possess an identical mechanism for substrate hydroxylation, which includes the use of ascorbate as the physiological reductant.^[7,14]

Crystallographic characterization has only been obtained for PHM (e.g., PDB code: 1PHM),^[12] exhibiting two domains, each binding a copper ion (Cu_{H(or A)} and Cu_{M(or B)}). These two copper centers possess different coordination environments and are separated by 11 Å (Figure 1). Domain I binds the active-site Cu_H with three histidines (H107, H108, and H172), while domain II binds the active-site Cu_M with two histidines and a methionine residue (H242, H244, and M314). There is agreement about the main role of the two copper centers; Cu_H is associated with electron transfer,^[15]] whilst Cu_M is involved in dioxygen binding and substrate hydroxylation. The crystal structure of the oxy-form of PHM shows a cupric-superoxide species bound in an end-on (η¹) fashion at the Cu_M site^[13] (Figure 1).

2.2 LPMOs

Rather recently rediscovered LPMOs have attracted attention due to their remarkable role in biomass conversion through an oxidative mechanism, which involves hydroxylation of a polysaccharide substrate at the C1 or C4 carbon, allowing the subsequent cleavage of the glycosidic bond (Figure 2).^[8,9]

Figure 2.

A) X-ray structure of the LPMO-AA9 from T. aurantiacus.^[20] B) Schematic of the copper active sites observed in LPMOs AA9^[20] and AA10^[21] (Cu(II) with two coordinated water ligands) and AA11^[18] and AA13^[22] (reduced Cu(I)). AR=amino acid residue. C) Glycosidic bond cleavage via hydroxylation at either the C1 or C4 position in cellulose by LPMOs.^[8]

The families of LPMOs that have been reported to date include cellulose-active fungal AA9 (also known as GH61),^[16] cellulose-active and chitin-active bacterial AA10 (also known as CBM33),^[17] chitin-active fungal AA11,^[18] and starch-active fungal AA13.^[19]

X-ray crystal structures of at least one member of each of these LPMO families have been solved to date, highlighting a highly conserved monocopper active site in the four subclasses of these enzymes (AA9,^[20] AA10,^[21] AA11,^[18] and AA13^[22]). Each structure displays a single copper ion that is coordinated to three nitrogen-based ligands. In the reduced form, the copper ion is in a T-shape configuration. This ligand arrangement consists of two histidine residues, one of which acts as a chelating (bidentate) histidine or “histidine-brace”, while the other binds in a monodentate fashion. Comparison of the active sites shows that these enzymes share some common features, but also have subtle differences. For example, all of the subclasses have a tyrosine residue near the copper ion, except AA10, which has a phenylalanine residue (Figure 2B).^[8] Furthermore, LPMOs AA9 and AA13 show an unusual methylation on the τ-nitrogen atom of the “histidine-brace”, whose function remains enigmatic.^[9] Additional ligands, such as water molecules, have been observed in the oxidized form of LPMOs AA9 and AA10. Finally, copper-dioxygen species have been observed in the crystal structures of AA9 enzymes from Neurospora crassa. A cupric-superoxide species bound in an end-on fashion (η¹) was observed in an LPMO that selectively oxidizes the C4 carbon, and a cupric hydroperoxide species was observed in an LPMO that oxidizes both C1 and C4 (PDB codes: 4EIR and 4EIS).^[23] These relevant intermediates will be discussed in the analysis of the catalytic mechanism of copper-containing monooxygenases.

2.3 Galactose Oxidase and Amine Oxidase

GO and AO catalyze the oxidation of substrate alcohol and amine, respectively, using a copper ion and peptide-derived organic cofactor, with concomitant reduction of O₂ to H₂O₂.^[10] These cofactors, a tyrosine covalently cross-linked to a cysteine (Tyr-Cys) in GO and 2,4,5-trihy-droxyphenylalanine quinone (TPQ) in AO, are post-translationally derived.

The oxidized form of both of these enzymes has been characterized by X-ray crystallography (Figure 3).^[24,25] The active-site copper ion in both GO and AO is in a distorted square pyramidal geometry. In GO, the copper(II) ion is coordinated by two His, a Tyr, and a water molecule in the equatorial plane. It is also axially bonded to the Tyr-Cys cross-link. In AO, the copper(II) ion is coordinated by three His ligands and a water molecule in the equatorial plane, with another water molecule in the axial position. A crystal structure of AO with the TPQ cofactor bound to copper has also been published (PDB code: 1AV1).^[25]

Figure 3.

A) X-ray structure of a GO and its enzymatic reaction.^[24] B) X-ray structure of an AO and its enzymatic reaction.^[25]

GO and AO both operate via a ping-pong reaction mechanism, including oxidation of the substrate and reduction of O₂. The oxidative half-reaction includes O₂ reduction to H₂O₂ and concurrent oxidation of the organic cofactor. In GO, O₂ binds and receives an electron from the reduced copper(I) center. This leads to a cupric-superoxide intermediate that is proposed to abstract a hydrogen atom from the nearby Tyr-Cys cofactor, resulting in the formation of a cupric hydroperoxide species plus Tyr^•-Cys moiety, i.e., with a tyrosyl radical. Following hydrogen-atom transfer (HAT), the hydroperoxide ligand is protonated by the nearby Tyr residue to form H₂O₂, producing the oxidized cofactor [Cu(II)-Tyr^•-Cys].^[4]

For AO, there are two proposed oxidative mechanisms, due to both copper(II) aminoquinol (Cu(II)-TPQ_AMQ) and copper(I) semiquinone (Cu(I)-TPQ_SQ) existing in equilibrium in the reduced state of the enzyme. The first mechanism involves inner-sphere reduction of O₂. Binding of O₂ to the reduced copper ion results in a cupric-superoxide complex with TPQ_SQ. This superoxide intermediate is then further reduced (followed by protonation) by the semiquinone, resulting in release of H₂O₂ and formation of an iminoquinone moiety which can then be hydrolyzed to bring the catalyst to its oxidized form, TPQ_OX. The second proposed mechanism involves outer-sphere electron transfer. Starting with the Cu(II)-TPQ_AMQ species, O₂ is reduced to superoxide by TPQ_AMQ. Ligand exchange of water for superoxide would then lead to the cupric-superoxide complex, which can be reduced and protonated to release H₂O₂. These two mechanisms differ by the species capable to carry out the initial reduction of O₂ to superoxide, either copper(I) or the organic cofactor, TPQ_AMQ.

Although a cupric-superoxide species is not directly related to substrate reactivity in GO and AO, this intermediate is proposed to be important for generating the oxidized species of the active site ([Cu(II)-Tyr^•-Cys] or TPQ_OX), by O–H bond HAT giving cupric hydroperoxide complexes. The oxidized cofactors then effect the oxidation of either a substrate alcohol or amine, for GO and AO, respectively.

3. Proposed Pathways in Cu Monooxygenases

Due to the importance of these copper-containing enzymes, there has been great interest in determining their mechanism of action. There have been many mechanistic pathways proposed to occur for PHM, DβM, and LPMOs. Scheme 1 depicts relatively recently proposed mechanisms which differ either by their mode of substrate activation or by the step in which radical rebound (hydroxylation of substrate) occurs. Yet all of these reaction mechanisms proceed through similar intermediates, corresponding to O₂-reduced species occurring in aqueous chemistry, but here bound to a copper ion center.^[26] These include a cupric superoxide (^ES, end-on binding or ^SS, side-on bonding), a copper hydroperoxide (Hp), and a copper oxyl (Cp). The Hp and Cp intermediates will be further discussed in Section 4.

Scheme 1.

Reaction pathways proposed in the catalytic mechanism of copper-containing monooxygenase enzymes (PHM, DβM and LPMOs).

3.1 Pathway 1

In Pathway 1, a cupric-superoxide species (shown with superoxide bound end-on in Scheme 1, ^ES) is responsible for HAT from the substrate. The two proposed variations of Pathway 1 differ in when rebound with the substrate radical occurs and when electron transfer (ET) from Cu_H (in PHM and DβM) occurs. In Pathway 1A, ET from Cu_H occurs directly after HAT, cleaving the O–O bond of the Hp intermediate. Rebound then occurs to a copper-oxyl species. In Pathway 1B, rebound occurs directly after HAT, resulting in product and Cp, which is then reduced by Cu_H.

Klinman and coworkers have conducted many kinetic studies on PHM and DβM to ascertain the mechanism of these enzymes.^[7] One important study with DβM, using substrates that showed three orders of magnitude difference in catalytic efficiency, indicated that substrate oxidation and O₂ consumption are “tightly coupled.”^[27] This shows that the same amount of O₂ is taken up, with respect to hydroxylation, despite differing substrate reactivity. Kinetic isotope effects (KIEs) have also played an important role in proposing a cupric superoxide as the reactive intermediate responsible for substrate HAT.^[28,29] Using both isotopically labeled O₂ (¹⁶O₂ or ¹⁸O₂) and substrate (C–H or C–D), the calculated KIEs showed that the O₂ reduction and C–H activation processes are coupled. This means that the C–H bond must be broken before the O–O bond is cleaved, supporting a cupric superoxide as the reactive intermediate. Support for the rebound step occurring after O–O bond cleavage of the Hp intermediate has been reported. Klinman and coworkers showed that substrates with electron-withdrawing substituents dissociate from the enzyme-substrate complex faster than those without, suggesting that the substrate may be bound to the copper ion after the rebound step (i.e., LCu-OR in Pathway 1A, see Scheme 1).^[14]

Solomon and coworkers have provided computational support for the cupric superoxide performing HAT from the substrate.^[30,31] Modeling Cu_M with one methionine, two histidines, and a water ligand, the authors computed the energy pathways for a cupric hydroperoxide complex and both side-on (η²) and end-on (η¹) cupric-superoxide complexes using the substrate formylglycine (FmG). It was calculated that the homolytic O–O bond cleavage of the hydroperoxide complex to form a copper-oxyl species was energetically unfavorable. Furthermore, calculations showed that HAT by the cupric hydroperoxide was also not favored. However, both cupric-superoxide complexes were shown to be capable of HAT from FmG in an energetically favorable manner. Two different rebound pathways were also investigated computationally. Homolytic cleavage of the O–O bond of the cupric hydroperoxide intermediate, followed by subsequent rebound, was proposed to be unfavorable due to the calculated reaction barrier of at least 20 kcal mol⁻¹.^[30] In contrast, direct rebound with the cupric hydroperoxide complex was calculated to be a downhill process and was thus proposed to be the pathway operating in the enzyme (Pathway 1B).

By analogy to the conclusions made by Klinman for PHM and DβM, Marletta and coworkers have also proposed Pathway 1A to be the working mechanism in LPMOs.^[16,32]] However, a computational study pointing instead to a different mechanism of action for LPMOs has also been reported (vide infra).

3.2 Pathway 2

Pathway 2 was proposed by Parisel, Reinaud, and coworkers in 2008, based on DFT and spin-flip TD-DFT calculations.^[33] Using a simplified model system, Me₁tren (methyltriethyleneamine), the reaction pathways of intramolecular oxidation of the ligand were investigated. Interestingly, it was proposed that cleavage of the C–H bond by a superoxide intermediate resulted in a carbocation intermediate; that is, the superoxide complex performed a hydride abstraction (HA) instead of an HAT reaction. After HA, ET from Cu_H to Cu_M would occur (cleaving the O–O bond of Hp heterolytically), before oxygen-atom transfer to the carbocation. The authors claimed that their proposed mechanism avoids certain problems that the other proposed mechanisms do not address, such as an organic radical in the presence of Cu^II and ET from Cu_H to Cu_M to make Cu^II-O^• in the presence of the organic radical.

3.3 Pathway 3

Here, the reactive intermediate implicated in HAT from the substrate is a copper-oxyl species, with rebound following directly in the next step. Pathways 3A and 3B are differentiated by the number and source of electrons required to complete the reaction. In Pathway 3A, an external electron (with the proton) is enough to facilitate O–O cleavage, while in Pathway 3B, that electron comes from the ligand (i.e., an active-site amino acid); thus, the O–O cleaved product is formulated as L^•+Cu-O^• (L^•+Cp). The exogenously added electron in Pathway 3A results in the final copper oxidation state being reduced relative to that in Pathway 3B (Scheme 1).

In 2006, Yoshizawa et al. published a QM-MM study using a whole-enzyme model of DβM, including 4700 atoms, investigating the reactivity of the three proposed active oxygen species (copper hydroperoxide, oxyl, and superoxide).^[34] The authors stated that the whole-enzyme model is better for these calculations because the small molecule model system has fictitious hydrogen-bonding interactions that change the orientation of the substrate molecule. The activation barriers for HAT were calculated to be 23.1 kcal mol⁻¹ and 5.4 kcal mol⁻¹ for the cupric superoxide and Cu^II-O^• species, respectively. The authors thus concluded that the Cu^II-O^• has a higher oxidizing power and should therefore be the reactive intermediate in the HAT step, followed by oxygen rebound (Pathway 3A).

Also in 2006, Amzel and coworkers reported QM-MM computations on PHM, examining multiple reactive copper-oxygen species towards HAT.^[35] The species Cu^II-O₂^•− and Cu^I-O₂^•− were ruled out as the active oxidants due to their high calculated activation parameters of 25 kcal mol⁻¹. The Cu^II-OOH intermediate was also ruled out, since its activation energy was calculated to be 42.8 kcal mol⁻¹. The activation energy of HAT by two other reactive copper-oxygen species, namely [CuO]²⁺ (formed from protonation and heterolytic O–O bond cleavage liberating water from a Cu^II-OOH species, without reduction by an external electron) and [CuO]⁺, were also computed. The [CuO]²⁺ species was found to undergo HAT in a nearly barrierless process (0.15 kcal mol⁻¹) to spontaneously form the hydroxylated product with no Cu-OH intermediate. The activation energy calculated for the [CuO]⁺ species was found to be 4.1 kcal mol⁻¹, with an activation barrier of 5.4 kcal mol⁻¹ for the oxygen rebound step. These results led the authors to suggest that [CuO]²⁺ is the reactive intermediate in the enzyme, due to the near-zero activation energy of this species to undergo HAT (Pathway 3B). The authors calculated that there is unpaired electron density localized over the three ligands bound to the copper ion, so that [CuO]²⁺ should be thought of as [L^•+Cu^IIIO²⁻]²⁺ or [L^•+Cu^IIO^•−]²⁺. In a protein active-site environment, the ligand radical might be a protein radical, perhaps a nearby tyrosyl radical, and analogy could be made to Compound I in cytochrome P450 chemistry, the oxo-iron(IV) porphyrinate π-cation radical^[36]]

Beckham and coworkers also used computational studies to propose Pathway 3A as the active mechanism in LPMOs.^[37] The authors investigated an active-site model of a fungal LPMO using DFT. Both a cupric superoxide and a copper-oxyl intermediate were examined as the substrate oxidizing species. It was shown that the activation barrier for HAT by a superoxide complex was higher in energy than the combined energy of both the HAT and rebound steps of a copper-oxyl species. Because of this, the authors propose that a copper-oxyl species is the active intermediate.

4. Cu Hydroperoxide or Oxyl Complexes

In this review, cupric-superoxide complexes are emphasized, since it is likely and generally accepted that, at least in the noncoupled binuclear monooxygenases PHM, DβM, and TβM, this species initiates substrate HAT chemistry. However, since cupric hydroperoxide and copper-oxyl complexes must form at some point during monooxygenase reactivity,^[26] we have highlighted here relevant or important aspects of such species, some of which have been synthesized and characterized.

4.1 Cupric Hydroperoxide

During early investigations into the biochemistry and mechanism of action of PHM and DβM, it was hypothesized that a cupric hydroperoxide intermediate was responsible for the hydroxylation of substrate.^[6] In response to this, the synthesis and reactivity of multiple copper model complexes featuring hydroperoxide coordination were described and they have been reviewed.^[38,39] In 1998, Masuda and coworkers isolated a cupric hydroperoxide containing H-bonding moieties in the secondary coordination sphere and analyzed the complex using X-ray diffraction.^[40] Karlin and coworkers have reported the reactivity of several cupric hydroperoxides.^[41–45] However, experimental and computational studies have shown that the cupric hydroperoxide species is not capable of being the reactive intermediate in these enzymes.^{[27,30,31,35]}

4.2 Copper(II) oxyl

While a mononuclear copper-oxyl species has been shown computationally to be a competent reactive intermediate for hydroxylation in the monooxygenase enzymes PHM, DβM,^[34,35] and LPMOs,^[37] such a species has not been directly observed experimentally. However, a copper-oxyl intermediate has been proposed to form and be active in some copper model complex systems.

Karlin and coworkers proposed that a copper-oxyl intermediate effects the intramolecular oxidation of an aliphatic C–H bond on the ligand to copper.^[46] It was suggested that this reactive intermediate was formed upon homolysis of the O–O bond of the cupric hydroperoxide. Similarly, ligand oxidation was observed upon addition of excess H₂O₂ and triethylamine or the well-known O-atom transfer agent iodosobenzene. This evidence indirectly supported the formation of a copper-oxyl intermediate capable of C–H bond oxidation.

Itoh and coworkers have reported on the thermal O–O bond homolysis of a novel copper(II)-cumylperoxide (Cu^II-OOR) complex.^[47] Using the radical trapping agent 5,5-dimethyl-1-pyrroline-N-oxide (DMPO), the putative copper(II) oxyl species formed could be trapped and characterized by ESI-MS and EPR spectroscopy. Without the radical trap, oxidation of N-methyl-9,10-dihydroacridine (AcrH₂) and 1,4-cyclohexadiene to N-methylacridinium and benzene, respectively, occurred. Therefore, it may be possible for a copper(II) oxyl intermediate generated in situ to oxidize exogenous substrates.

Recently, Itoh and coworkers published a more thorough investigation into the reactivity of a Cu^II-OOR complex.^[48] Using results from kinetic isotope effect studies, product analysis, and DFT calculations, homolysis of the O–O bond in the Cu^II-OOR complex and simultaneous H-atom abstraction from the substrate was proposed. The calculations performed pointed to the proximal oxygen (bonded to the copper ion) as the destination of the H-atom. These data supported the conclusion that the Cu^II-OOR itself, and possibly not a copper oxyl, is responsible for substrate oxidation.

5. Cupric-Superoxide Model Complexes

As described, cupric-superoxide complexes have been proposed by some to be the reactive intermediate in certain copper-containing enzymes. At least partial evidence to support this claim was provided by Amzel and coworkers in 2004, when they crystallized the enzyme PHM, finding that dioxygen bound in an η¹ fashion to Cu_M (Figure 1A).^[13] The bonding metrical parameters derived from this crystal structure determination strongly suggested that a cupric superoxide could be generated in the enzyme. As discussed above, experimental and computational studies, both on the enzymes and on model complexes, have offered even more information about the possibility of a cupric-superoxide complex being a possible reactive intermediate for substrate oxidation.

In these kinds of Cu-O₂ complexes, the superoxide ligand can bind in two ways, either through one oxygen (η¹) or through both oxygens (side-on, η²). Crystal structures of cupric-superoxide model complexes showing both of these coordination modes have been reported (Figure 4).^[49,50]

Figure 4.

A) Crystal structure^[50] and UV-Vis spectrum of [Cu^II(TMG₃tren)(O₂^•−)]⁺ (TMG₃tren = tris(tetramethylguanidine)tri-ethyleneamine). B) Crystal structure and UV-Vis spectrum of [Cu^II(HB(3-^tBu-5-ⁱPrpz)₃(O₂^•−)] (HB(3-^tBu-5-ⁱPrpz)₃ = tris(3-tert-butyl-5-iso-propylpyrazolyl)hydroborate.^[49]

Although some cupric-superoxide model complexes had been synthesized and characterized prior to Amzel’s crystal structure of PHM, this seminal paper sparked new interest in investigating the reactivity of new superoxide complexes, especially those that involved superoxide bound in an η¹ fashion (vide infra). Table 1 provides a listing of known mononuclear cupric-superoxide complexes, along with their relevant physical properties, while Figure 5 depicts the chelating polydentate synthetically derived ligands utilized in the complexes of Table 1.

Table 1.

Binding mode and spectroscopic features of cupric-superoxide model complexes.

Supporting Ligand	Binding Mode	UV-Vis:λ,nm(ε, M−1cm−1)	rR: v(O-O), cm−1 (Δ(18O))	rR: v(Cu-O), cm−1 (Δ(18O))	Ref.
TMPA	η1	410 (4000), 747 (1000)[a]	N. R.[b]	N. R.	[51,52]
DMATMPA	η1	418 (4300), 615 (1100), 767 (840)[c]	1121 (−63)	472 (−20)	[53]
DMMTMPA	η1	409 (4250), 587 (1100), 743 (1030)[d]	1121 (−63)	474 (−18)	[54]
DMAN3S	η1	418, 605, 743[e]	1117 (−61)	460 (−20)	[55]
PVTMPA	η1	410 (3700), 585 (900), 741 (1150)[f]	1130 (−63)	482 (−20)	[56]
TMG3tren[g]	η1	448 (3400), 676, 795[h]	1120 (−63)	435(−20)	[57,58]
Me6tren	η1	412 (4800)[b]	1122	N.R.	[59,60]
Me3Bn3tren	η1	416 (5400), 591 (1640), 737 (2340)[d]	1120 (−61)	474 (−20) 459 (−17)	[61]
HIPT3tren	η1	434 (3850), 549 (1350), 675 (1910), 790 (3520)[d]	1096 (−67)		[62]
HB(3-tBu-5- iPrpz)3-[g]	η2	352 (2330), 510 (230), 660 sh[i]	1112 (−52)[j]	554 (−20)[k]	[49]
HB(3-Ad 5-iPrpz)3	η2	383, 452, 699, 980[l]	1043 (−59))	N. R.	[63]
DPH2	η1	524, 650[m]	964 (−55)[j]	N. R.	[64]
TACNPhO	η2	416 (4600), 654 (1800)[n]	1120 (−62)	450 (−8), 422 (−5)	[65]
PDCA	η1	627 (1700)[o]	1104 (−60)	N.R.	[66]
PEDACO-EtPh- OCH3	η1	397 (4550), 570 (865), 705 (1250)[p]	N.R.	N.R.	[67]
PEDACO-EtPh-H	η1	397 (4200), 570 (850), 705 (1150)[p]	1033 (−65) N.R.	457 (−15) N.R.	[67]
PEDACO-EtPh- NO2	η1	397 (4150), 570 (790), 705 (1025)[p]			[67]
PEDACO-iPr	η1	395 (5810), 585 (980), 723 (1240)[q]	N.R.	N.R.	[68]
UN-O	µ-1,2/µ-1,1	404 (5400), 635 (670)[r]	1144/1120 (−61/−65)[s]	478/383 (−22/−26)[t]	[69,70]
Hexpy		370 (11000), 550 (530)[u]	1131 (−63)	368 (−13)[v]	[71]

^[a]Propionitrile @ −90°C.

^[b]N. R. = Not Reported

^[c]THF @ −85°C.

^[d]Acetone @ −90°C.

^[e]MeTHF:TFE 4 : 1 @ −135°C.

^[f]MeTHF @ −125°C.

^[g]Characterized by X-ray crystallography, see Refs. [49,50].

^[h]MeTHF @ −130°C.

^[i]DCM @ −50°C.

^[j]IR data.

^[k]From Ref. [63].

^[l]DCM @ −70°C, all extinction coefficients are below 400.

^[m]MeOH @ −50°C.

^[n]THF @ −20°C.

^[o]DMF/THF 1 : 1 @ −80°C.

^[p]Acetone/MeCN 9 : 1 @ −85°C.

^[q]Acetone @ −85°C.

^[r]DCM @ −80°C.

^[s]The O-O stretches at 1144 and 1120 cm⁻¹ were assigned to the µ-1,2 and µ-1,1 isomer, respectively.

^[t]Both Cu-O stretches were assigned to the µ-1,1 isomer.

^[u]MeCN/DCM 1 : 9 @ −20°C.

^[v]Proposed to be a Cu-(O₂^•−)-Cu stretch.

Figure 5.

The chelating polydentate synthetically derived ligands utilized in the complexes of Table 1.

5.1 Complexes Bearing the TMPA Ligand Framework

5.1.1 TMPA

The isolation of cupric-superoxide complexes in small molecule synthetic systems is inherently difficult due to the thermodynamic stability of 2:1 copper dioxygen adducts. As a result, early examples of structurally characterized copper dioxygen species, generated from the addition of O₂ to mononuclear LCu^I complexes under cryogenic conditions, yielded peroxodicopper(II) complexes in either a trans-µ-1,2 (^TP)[^[72]] or µ-η²:η² side-on binding geometry (^SP).^[73] The formation of these secondary di-oxygen adducts from monomeric Cu^I and O₂ logically implicates the existence of a primary 1 :1 Cu/O₂ species as an initial kinetic intermediate, due to the unlikely ternary reaction to directly yield a 2:1 Cu/O₂ product (Scheme 2).

Scheme 2.

Primary and secondary dioxygen adducts formed from addition of O₂ to ligand-Cu^I mononuclear compounds. Binuclear LCu^II₂(O₂²⁻) complexes may possess trans-µ-1,2-peroxide, cis-µ-1,2-peroxide (with an additional 2^nd bridging ligand) or side-on µ-η²:η²-peroxide structures. The ligand (L) may also favor a form in which the copper ions each provide an additional electron, giving rise to a bis-µ-oxo-dicopper(III) structural form.

Karlin and coworkers were able to confirm these predictions in reporting a short-lived intermediate after initial addition of O₂ to [Cu^I(TMPA)(EtCN)]⁺ (TMPA = tris-[2-pyridyl-methyl]amine) at −90°C in propionitrile (EtCN), which then rapidly transformed to the trans-µ-1,2-peroxide, [{Cu^II(TMPA)}₂(O₂²⁻)]²⁺.^[51,74]] This short-lived intermediate was determined to be the superoxide complex, [Cu^II(TMPA)(O₂^•−)]⁺.

A subsequent study using a “flash and trap” method via photolysis of CO from [Cu^I(TMPA)(CO)]⁺ at −85°C in oxygenated THF showed rate and equilibrium (k₁ and K₁) values of 1.5 × 10⁸ M⁻¹s⁻¹ and 6.5 × 10⁵ M⁻¹, respectively.^[75] Extrapolating k₁ to 25°C, the rate constant of O₂ binding is 1.3 × 10⁹ M⁻¹s⁻¹, which is nearly 10 to 100 times faster than myoglobin and hemoglobin, k₁ = 1.4–25 × 10⁷ M⁻¹s⁻¹ and 2.9–22 × 10⁷ M⁻¹s⁻¹, respectively.^[76]

Despite a large k₁, [Cu^II(TMPA)(O₂^•−)] ⁺ , and synthetic model copper superoxides, in general, are not observed at ambient conditions. In the same aforementioned flash and trap experiment, k₋₁ was calculated as 1.8 × 10⁸M⁻¹ at 25°C, relative to 1.2 × 10¹–2.3 × 10⁴ in myoglobin/hemoglobin. This large k₋₁ value contributes to a significantly smaller K₁ value of 15.4 M⁻¹ for the [Cu^II(TMPA)(O₂^•−)] ⁺ system, as compared with the large values for myoglobin and hemoglobin (K₁ = 0.74–117 × 10⁴ M⁻¹, and 2.87–46.6 ×10⁵ M⁻¹, respectively).[^[76]] The sum of these thermodynamic and kinetic values indicates O₂ binding to Cu^I is facile, but the resulting primary oxygen adduct is unstable and will either revert back to Cu^I and O₂ or rapidly convert to a secondary dioxygen adduct. To generate stable cupric superoxides, one must carefully balance the kinetic and thermodynamic properties of O₂ binding, release, and dimerization to the peroxide (^TP or ^SP) or bis-µ-oxide (O) intermediates (see Scheme 2).

5.1.2 ^DMATMPA

Efforts to stabilize a [LCu^II(O₂^•−)]⁺ complex led to systematic variations of the TMPA ligand to modulate the electronic environment of the copper center. Karlin and coworkers synthesized [Cu^I(^DMATMPA)] ⁺ , a [Cu^I(TMPA)]⁺ derivative bearing a dimethylamino group at the para position of each pyridyl ligand arm. Cyclic voltammetry of the compound revealed a Cu^II/I reduction potential of −700 mV vs. Fc⁺/Fc in MeCN,^[55] 380 mV lower than the parent TMPA complex, measured to be −420 mV under the same conditions (Table 2).^[77] Oxygenation of the carbonyl derivative [Cu^I(^DMATMPA)(CO)]⁺ in THF, at −85°C, yielded a brilliant green solution that was EPR silent.^[53] Resonance Raman spectroscopy confirmed the oxygenated intermediate as [Cu^II(^DMATMPA)(O₂^•−)]⁺ , consistent with an N₄-ligated end-on superoxide complex.^[78]

Table 2.

Reduction potentials of Cu^II complexes in MeCN vs. Fc⁺/Fc.^[55,77]

Supporting Ligand	DMATMPA	DMMTMPA	PVTMPA	DMAN3S	TMPA
E1/2 (mV)	−700	−570	−520	−470	−420

Remarkably, the generated superoxide species was stable at −85°C for several hours (t_1/2≥4hrs), without formation of the bridging dicopperperoxide species. This surprising stability allowed for exploration of [Cu^II(^DMATMPA)(O₂^•−)]⁺ reactivity with exogenous substrates. The addition of p-OMe-2,6-di-tert-butylphenol to [Cu^II(^DMATMPA)(O₂^•−)]⁺ at −85°C yielded the corresponding phenoxyl radical characterized by UV-Vis (λ_max = 407nm, sharp peak) and EPR (g~2) spectroscopies. ¹⁸O₂ labeling experiments revealed incorporation of labeled oxygen into the 2,6-di-tert-butyl-1,4-benzoquinone and 2-hydroperoxyl-4-OMe-6-tert-butylphenol products, confirmed by GC-MS. Oxidation of 2,4,6-tri-tert-butylphenol was also tested and yielded the 2,6-di-tert-butyl-1,4-benzoquinone exclusively, where the inserted oxygen atom coming from O₂ was confirmed by GC-MS (Scheme 3).

Scheme 3.

Substrate oxidation performed by synthetic cupric-superoxide model complexes.

5.1.3 ^DMMTMPA

To further investigate how cupric-superoxide complexes oxidize exogenous substrates, the modified TMPA derivative, ^DMMTMPA, bearing methyl groups at the 3- and 5-positions and a methoxy group at the 4-position of the pyridyl ligand arm, and the corresponding Cu^I complex were synthesized. The Cu^II/I reduction potential of this complex was found to be in between [Cu^II(^DMATMPA)]²⁺ and [Cu^II(TMPA)]²⁺ at −570 mV vs. Fc⁺/Fc in MeCN (Table 2).^[55] It was expected that the less electronically stabilized Cu^II center in [Cu^II(^DMMTMPA)]⁺ , relative to [Cu^II(^DMATMPA)] ⁺ , imparts the corresponding superoxide species with greater electrophilic character, capable of reacting with substrates possessing a larger range of bond dissociation energies (BDEs).

Addition of O₂ to [Cu^I(^DMMTMPA)(CO)]⁺ at −90°C in either acetone or 2-methyltetrahydrofuran (MeTHF) yielded a green solution that was consistent with an end-on superoxide species [Cu^II(^DMMTMPA)(O₂^•−)]⁺, as deduced using UV-Vis and resonance Raman spectroscopies.^[54] [Cu^II(^DMMTMPA)(O₂^•−)]⁺ (t_1/2 = 3hrs, −90°C) was found to be less stable than [Cu^II(^DMATMPA)(O₂^•−)] ⁺ , but more stable than [Cu^II(TMPA)(O₂^•−)] ⁺ .

Reactions of [Cu^II(^DMMTMPA)(O₂^•−)]⁺ were performed with various para-substituted 2,6-di-tert-butylphenols (p-X-2,6-DTBP) (Scheme 3). Adding p-X-2,6-DTBP to [Cu^II(^DMMTMPA)(O₂^•−)]⁺ yielded the corresponding phenoxyl radical, identified by the distinctive UV-Vis features near 400 nm and by EPR spectroscopy displaying a sharp peak at g~2. Final products of the low temperature reactions were evaluated by GC-MS, and in all cases, the major product was the 2,6-di-tert-butyl-1,4-benzoquinone. KIE studies using p-OCD₃ and p-CD₃ 2,6-di-tert-butylphenols were employed yielding values of 11 and 4, respectively, suggesting an overall HAT pathway. To provide evidence that HAT was the rate-determining step, cumylperoxyl radical (Cm-OO^•), a pure hydrogen atom acceptor with respect to phenols, was used as an oxidant with the p-X-2,6-DTBP substrates. Comparison of the second-order rate constants of phenol oxidation with Cm-OO^• and [Cu^II(^DMMTMPA)(O₂^•−)]⁺ yielded a linear trend (slope = 4.5), indicating HAT as the likely rate-determining step. Further evidence for HAT was obtained via KIE studies of p-OCD₃ 2,6-DTBP oxidation using Cm-OO^•, which yielded a value of 9, similar to that found for [Cu^II(^DMMTMPA)(O₂^•−)]⁺ . These results strongly suggest an initial HAT step yielding LCu^II-OOH and phenoxyl radical. KIE studies and control reactions with the pure HAT acceptor Cm-OO^• agree that [Cu^II(^DMMTMPA)(O₂^•−)]⁺ reacts through a HAT process as opposed to ET/PT, PT/ET, or PCET pathways, in line with what would occur with copper monooxygenase mechanistic Pathway 1.

5.1.4 ^DMAN₃S

Modeling of synthetic mononuclear copper systems bearing an N₂S or N₃S ligation is of great interest due to the structural similarities to the 2-His-1-Met coordination motif found in the biological copper-containing monooxygenases PHM, DβM, and TβM (vide supra).^[7] Previously reported models containing N₂S geometries generally yielded secondary dioxygen adducts as final low temperature products,^[79,80]] indeterminate Cu/O₂ adducts,^[81–83]] or exhibited no O₂ reactivity.^[84]] Recent efforts by Karlin and coworkers have shown that selective steric modulation of the sulfur donors in ^DMMN₃S systems leads to preferential formation of the trans-µ-1,2 peroxide or bis-µ-oxide copper complexes (Figure 6).^[85]

Figure 6.

Secondary dioxygen adducts formed with N₂S and ^DMMN₃S ligands.^[79,80,85]

Very recently, by combining knowledge gleaned from the ^DMATMPA, ^DMMTMPA, and ^DMMN₃S systems, Karlin and coworkers were able to report the first example of a thioether-ligated cupric-superoxide complex, [Cu^II(^DMAN₃S)(O₂^•−)]⁺ , a [Cu^II(^DMATMPA)(O₂^•−)]⁺ derivative where one pyridyl arm is replaced with a 2-methyl-benzyl-thioether group.^[55] Oxygenation of [Cu^I(^DMAN₃S)]⁺ at −135°C in MeTHF led to formation of [Cu^II(^DMAN₃S)(O₂^•−)] ⁺ , which rapidly converted to [{Cu^II(^DMAN₃S)} (O₂²⁻)]²⁺ over 50s Fortunately, [Cu^II(^DMAN₃S)(O₂^•−)]⁺ was found to be selectively stabilized by oxygenating [Cu^I(^DMAN₃S)]⁺ in MeTHF solutions containing the polar, and potentially hydrogen-bonding solvent trifluoroethanol (TFE) in a 4 :1 mixture. This new superoxide complex was confirmed via UV-Vis and resonance Raman spectroscopies, containing features similar to previously characterized tetradentate end-on cupric superoxides. ^{[53,54,56,78]}

Probing the reactivity of [Cu^II(^DMAN₃S)(O₂^•−)]⁺ with p-OMe-DTBP and AcrH₂ remarkably showed reactivity at −135 °C (Scheme 3), whereas both [Cu^II(^DMATMPA)(O₂^•−)]⁺ and [Cu^II(^DMMTMPA)(O₂^•−)] ⁺ were unreactive under the same conditions. The increased reactivity of [Cu^II(^DMAN₃S)(O₂^•−)]⁺ was attributed to the weaker donating ability of a thioether relative to pyridine, yielding a more electrophilic cupric superoxide. Indeed, this hypothesis is supported by cyclic voltammetry results that show the reduction potential of [Cu^II(^DMAN₃S)]²⁺ as −470 mV vs. Fc⁺/Fc in MeCN (Table 2).

Previous attempts at creating N_xS thioether-ligated cupric superoxides were unfruitful due to either weak Cu–S bonds, most likely the result of steric constraints of the thioether ligand, and/or insufficient donor ability of the nitrogen donors. The ^DMAN₃S system demonstrates careful and deliberate balancing of the steric and electronic contributions towards thioether-coordinated superoxide stability, based on years of discovering the fundamental principles that govern Cu^I/O₂ chemistry.

5.1.5 ^PVTMPA

Recent work in synthetic bioinorganic chemistry has illustrated that insertion of hydrogen bond donor groups in the secondary coordination sphere leads to stabilization of a series Mn and Fe oxygen intermediates.^[86,87] Earlier, Masuda and coworkers observed similar stabilization effects in a copper system leading to the first crystal structure of a mononuclear cupric hydroperoxide complex.^[40] Taking inspiration from these seminal examples of (di)-oxygen stabilization, Karlin and coworkers synthesized [Cu^I(^PVTMPA)]⁺ (^PVTMPA = {6-pivalamido-[2-pyridyl-methyl]}-bis[2-pyridylmethyl]amine), a TMPA derivative where one pyridyl ligand contains a pivalamide H-bond donor in the secondary coordination sphere, a ligand previously made by Masuda and coworkers (Figure 5).

Oxygenation of [Cu^I(^PVTMPA)]⁺ in MeTHF at −80 °C yields [{Cu^II(^PVTMPA)}₂(O₂²⁻)]²⁺[^[56] with UV-Vis features similar to previously observed trans-µ-1,2 peroxide dicopper complexes.^[72] When [Cu^I(^PVTMPA)]⁺ was oxygenated in MeTHF at −125°C, a green, EPR silent, species was observed, and further confirmed by UV-Vis and resonance Raman spectroscopies to be an end-on cupric-superoxide species.^[56]] Addition of the pivalamido group led to pronounced changes in the resonance Raman spectrum. As shown in Table 1, [Cu^II(^PVTMPA)(O₂^•−)]⁺ exhibits the strongest v_O-O and v_Cu-O stretches of any reported N₄-ligated cupric superoxide (Table 1). Geometry optimization studies showed that the pivalamido N–H group forms an intramolecular hydrogen bond to either the proximal or distal oxygen of [Cu^II(^PVTMPA)(O₂^•−)]⁺. Computational models for hydrogen bonding to the proximal or distal oxygen both show an increase in the v_O-O and v_Cu-O vibrations relative to having no hydrogen bonding, where hydrogen bonding to the distal oxygen displayed the largest increase.

[Cu^II(^PVTMPA)(O₂^•−)]⁺ was found to be unreactive towards several common C–H substrates, including dihydroanthracene, xanthene, and AcrH₂. Addition of 1-benzyl-1,4-dihydronicotinamide (BNAH), an NADH mimic and a strong H-atom and H donor, leads to decay of the superoxide species to the trans-µ-1,2-peroxide di-copper complex. KIE studies employing 1-benzyl-1,4-di-hydro[4,4’–²H₂]nicotinamide (BNAD), displayed a pronounced primary KIE of 12.1. Comparative reactions with 1,3-dimethyl-2,3-dihydrobenzimidazole (BzImH), a preferential H⁻ donor (heterolytic hydride affinity = 49.5kcal mol⁻¹; homolytic BDE = 73.4 kcal mol⁻¹) vs. BNAH (heterolytic hydride affinity = 64.2 kcal mol⁻¹; homolytic BDE = 70.7 kcal mol⁻¹ showed a 2.4 rate increase of superoxide reactivity with BNAH over BzImH, suggesting rate-determining HAT, over hydride transfer, as the preferred C–H transfer cleavage pathway, thus agreeing with Pathway 1 and not Pathway 2 with respect to discussions surrounding Scheme 1.

5.2 Complexes Bearing the Tren Ligand Framework

5.2.1 TMG₃tren

Stabilizing cupric superoxides requires careful balancing of electronic and steric demands of the copper center and ligand scaffold. Schindler and coworkers incorporated such principles when designing TMG₃tren, a triaminoethylene amine (tren) modified with three tetramethylguanidine moieties. Oxygenation of [Cu^I(TMG₃tren)]⁺ at −70°C in acetone led to formation of [Cu^II(TMG₃tren)(O₂^•−)]⁺ as a bright green solution, which, upon warming in a closed cell, reverted back to Cu^I and O₂.^[78] The superoxide complex was confirmed by resonance Raman spectroscopy and X-ray crystal structure determination. This remains the only crystallographically characterized copper superoxide complex possessing a tetradentate ligand.^[50] Variable temperature NMR spectroscopy, ¹⁸O equilibrium isotope effect (EIE) measurement, and variable temperature variable field (VTVH) MCD spectroscopies, as well as DFT calculations conclude [Cu^II(TMG₃tren)(O₂^•−)]⁺ is a ground state triplet, i.e., S = 1.^[58,88]

Karlin and coworkers have reported on the chemistry of [Cu^II(TMG₃tren)(O₂^•−)]⁺ with various exogenous substrates. Reaction of [Cu^II(TMG₃tren)(O₂^•−)]⁺ with p-OMe-2,6-DTBP has been tested and found to yield a mixture of corresponding benzoquinone, phenoxyl radical and 2-hydroperoxyl-4-OMe-6-tert-butylphenol (Scheme 4).^[46] Surprisingly, crystallographic analysis of the product revealed one of the methyl groups on the guanidinyl arms had been hydroxylated.

Scheme 4.

Reactivity of [Cu^II(TMG₃tren)(O₂^•−)]⁺ with phenol and acid/reductant.

To complement the reactivity of [Cu^II(TMG₃tren)(O₂^•−)]⁺ with exogenous H-atom donors, reactivity with exogenous proton and electron sources was explored. Addition of reducing agent decamethylferrocene (Me₁₀Fc; E_red=−0.52 V vs. Fc⁺/Fc in MeTHF), or stronger reducing agent cobaltocene (CoCp₂; E_red= −1.345 V vs. Fc⁺/Fc in MeTHF), exhibited no reactivity with [Cu^II(TMG₃tren)(O₂^•−)]⁺ at −130°C.^[57] Upon addition of 10 equiv. of trifluoroacetic acid (TFA) with no reductant present, the superoxide complex changed to a new intermediate with UV-Vis features at λ_max (ε, M⁻¹ cm⁻¹): 330 (5200), 549 (370), 670 (170), 985 (430) nm. A Job’s plot indicated maximum formation of [Cu^II(TMG₃tren)(O₂^•−)(TFA)]+ at X_Cu = 0.5 suggesting a 1:1 copper-TFA adduct. ¹⁹F NMR spectroscopy also confirmed the formation of a TFA adduct, yielding a broad, thermally sensitive, chemically distinct peak, generated at 80% yield relative to copper. Resonance Raman spectroscopy showed a dramatic shift of the v_O-O stretch from 1117 cm⁻¹ to 1149 cm⁻¹ upon TFA addition to [Cu^II(TMG₃tren)(O₂^•−)]⁺ . Addition of d¹-TFA yields an indistinguishable v_O-O stretch relative to [Cu^II(TMG₃tren)(O₂^•−)(TFA)]⁺ , suggesting formal proton transfer from TFA to [Cu^II(TMG₃tren)(O₂^•−)]⁺ does not occur due to the lack of an isotope effect on the O–O stretch. X-ray absorption spectroscopy of [Cu^II(TMG₃tren)(O₂^•−)(TFA)]+ is consistent with no change in oxidation state or Cu^II geometry after addition of TFA, ruling out protonation of the ligand by TFA. DFT studies suggest the most likely structural isomer of [Cu^II(TMG₃tren)(O₂^•−)(TFA)]⁺ , consistent with spectroscopic observations, contains a TFA moiety hydrogen bonded to the distal oxygen of [Cu^II(TMG₃tren)(O₂^•−)] ⁺ .

Addition of 1 equivalent of either Me₁₀Fc or Me₈Fc (octamethylferrocene) to [Cu^II(TMG₃tren)(O₂^•−)(TFA)]+ showed reduction to the hydroperoxide complex, [Cu^II(TMG₃tren)(OOH)]⁺ , as shown in Scheme 4, which quickly becomes protonated to release H₂O₂. Formation of [Cu^II(TMG₃tren)(OAc^F)]⁺ , the trifluoroacetate adduct, was confirmed by comparison with the authentic compound by EPR spectroscopy. The reduction of [Cu^II(TMG₃tren)(O₂^•−)(TFA)]+ was found to be first order in reductant, suggesting a simple one electron reduction to the hydroperoxide species. Further studies using Me₂Fc (dimethylferrocene) with [Cu^II(TMG₃tren)(O₂^•−)(TFA)]⁺ showed no reactivity, suggesting the reduction potential of the TFA adduct is in between Me₈Fc (E_red = −0.43 V vs. Fc⁺/Fc, in MeTHF) and Me₂Fc (E_red = −0.115 V vs. Fc⁺/Fc, in MeTHF). This stepwise hydrogen-bonding reduction pathway is suggestive of a reaction which might occur in copper monooxygenase reaction Pathway 3 (Scheme 1), where pronation and reduction of the initial cupric superoxide yields a hydroperoxide intermediate.

5.2.3 Other tren Derivatives

Schindler and coworkers reported the synthesis of [Cu^I(Me₆tren)]⁺ as an aliphatic analogue to TMPA-based tripodal tetradentate ligands. Oxygenation of [Cu^I(Me₆tren)]⁺ at −60°C in EtCN led to the formation of a transiently stable intermediate that decayed to [{Cu^II(Me₆tren)}₂(O₂²⁻)]²⁺.^[59] The dicopper peroxide product was verified by resonance Raman spectroscopy (v_O-O = 880 cm⁻¹). Oxygenation of the same Cu^I compound at −90°C in EtCN led to stable generation of the formally transient intermediate. Resonance Raman spectroscopic inquiries later concluded this intermediate as [Cu^II(Me₆tren)(O₂^•−)]⁺.^[60] Interestingly it was observed that oxygenation of [Cu^I(Me₆tren)]⁺ in acetone led to faster and more complete formation of the cupric superoxide, which the authors attributed to a decrease in solvent competition with O₂ for Cu^I binding.

Suzuki and coworkers utilized the tren framework and found systematic alkylation or arylation of the amino groups led to preferential formation of either a trans-µ 1,2-peroxide dicopper or cupric-superoxide complex. Oxygenation of [Cu^I(Me₃H₃tren)]⁺ at −80°C yields [{Cu^II(Me₃H₃tren)}₂(O₂²⁻)]²⁺ as a deep violet solution confirmed as the peroxide species by resonance Raman spectroscopy.^[61] Oxygenation of the arylated derivative [Cu^I(Me₃Bn₃tren)]⁺ at −90°C in acetone yields a stable green species identified by resonance Raman spectroscopy as [Cu^II(Me₃Bn₃tren)(O₂^•−)]⁺. Upon bubbling of argon into a solution of [Cu^II(Me₃Bn₃tren)(O₂^•−)]⁺ , this species transforms into the binuclear peroxide complex analogue [{Cu^II(Me₃Bn₃tren)}₂(O₂²⁻)]²⁺, as also confirmed by resonance Raman spectroscopy.

Itoh and coworkers designed a tren-based ligand featuring hexaisopropylterphenyl (HIPT) groups to impose extreme steric demands on the tren-based ligand to preclude formation of secondary dioxygen adducts in an effort to stabilize the cupric-superoxide form (Scheme 2). Indeed oxygenation of [Cu^I(HIPT₃tren)]⁺ at −90°C in acetone yielded a dark green species confirmed as [Cu^II(HIPT₃tren)(O₂^•−)]⁺ by resonance Raman spectroscopy.^[62] A silent EPR spectrum and apparent lack of paramagnetically shifted signals in the complex’s ¹H-NMR spectrum led the authors to suggest [Cu^II(HIPT₃tren)(O₂^•−)]⁺ had an S=0 ground state. However, a later study by Solomon and Karlin showed, using VTVH-MCD and ²H-NMR spectroscopies, that [Cu^II(HIPT₃tren)(O₂^•−)]⁺ was a ground state triplet,^[89] a result more consistent with other cupric superoxides possessing tetradentate ligation.[^[58,88]

5.3 Anionic Tetradentate Ligand

In 2003, Tolman and coworkers reported the first cupric-superoxide complex supported by a tetradentate ligand with a phenolate moiety to mimic the active site of galactose oxidase.[^[65]] This superoxide complex, [Cu^II(TACN^PhO)(O₂^•−)], was stable at −50°C in tetrahydrofuran. This species was characterized by EPR, UV-Vis, and resonance Raman spectroscopies. Mixed isotope labeling (¹⁶O-¹⁸O) suggested that the superoxide ligand binding mode was side-on, making this the first example of a side-on cupric-superoxide species supported by a tetradentate ligand. The designation of side-on binding was also supported by a later computational study published by Cramer and coworkers.[^[90]]

5.4 Complexes Bearing a Tridentate Ligand Framework

5.4.1 Anionic Ligands

The first crystallographically characterized cupric-superoxide complex was reported by Fujisawa, Kitajima, and coworkers in 1994 using a sterically hindered hydrotris-(pyrazolyl)borate ligand (HB(3-^tBu-5-ⁱPrpz)₃, with tert-butyl groups to prevent dimerization to secondary dioxygen intermediates.[^[49]] [Cu^II(HB(3-^tBu-5-ⁱPrpz)₃)(O₂^•−)] was synthesized by the addition of O₂ to a dichloromethane solution of [Cu^I(DMF)(HB(3-tBu-5-ⁱPrpz)₃)] at −50°C. This new species was characterized by UV-Vis, IR, and resonance Raman spectroscopies. [Cu^II(HB(3-tBu-5-ⁱPrpz)₃)(O₂^•−)] was shown to be diamagnetic as it exhibited sharp signals seen in its ¹H-NMR spectrum in CD₂Cl₂ at −40 °C. Variable temperature magnetic susceptibility measured by SQUID also agreed with the S = 0 assignment. The X-ray crystal structure of this intermediate showed that the superoxide ligand was bound in a side-on fashion, with Cu-O bond distances of 1.84 Å and an O-O bond distance of 1.22 Å

Almost a decade later, Solomon, Fujisawa, and coworkers published an article detailing the electronic structure and spectroscopy of [Cu^II(HB(3-^tBu-5-ⁱPrpz)₃)(O₂^•−)], as well as a new derivatized complex, with adamantyl groups replacing the tert-butyl groups.[^[63]] This study showed that the copper ion and superoxide ligand have a highly covalent interaction, which leads to the diamagnetic singlet ground state, due to magnetic coupling of the unpaired spins formally derived from the separate Cu^II and superoxide radical anion moieties. Interestingly, the UV-Vis spectrum of the superoxide with the adamantyl-ligand derivative did not show any intense absorption bands like its parent analogue (Figure 4), although four weak transitions were observed.

In 1999, the first end-on cupric superoxide supported by an anionic tridentate ligand was published by Wieghardt and coworkers.[^[64]] Addition of O₂ to a mixture of Cu^ICl, DPH₂ (N,N-bis(2-hydroxy-3,5-di-tert-butylpheny-l)amide anion), and triethylamine in methanol at temperatures between −50 and −70°C resulted in deep red crystals. This new complex, [Cu^II(DPH₂)(O₂^•−)], was characterized by UV-Vis and IR spectroscopies. The end-on coordination was postulated due to the IR spectrum observed when mixed isotope labeling was used.

More recently, Tolman and coworkers reported the synthesis of a cupric superoxide from addition of potassium superoxide to a Cu^II complex supported by the dianionic ligand PDCA (N,N-bis(2,6-diisopropylphenyl)-2,6-pyridi-nedicarboxamide).^[66] This superoxide complex, [Cu^II(PDCA)(O₂^•−)]⁻, was shown to be stable in a DMF/ THF (1:1) solution at −80 °C, but decomposed when warmed above −60°C. [Cu^II(PDCA)(O₂^•−)]⁻ was characterized by UV-Vis, resonance Raman, EPR, and ¹H-NMR spectroscopies. These data supported the identification of a cupric superoxide that is paramagnetic. DFT calculations were also performed to better understand the electronic structure. Addition of [Cu^I(TMPA)(CH₃CN)]⁺ resulted in the formation of a trans-µ-1,2-peroxide complex characterized by UV-Vis and resonance Raman spectroscopies. Interestingly, [Cu^II(PDCA)(O₂^•−)]⁻ was shown to not undergo electrophilic reactivity with phenols as other cupric-superoxide complexes have been shown to accomplish (vide supra). Instead, the authors showed that the complex could react as a base, and proposed that this reactivity comes from the anionic nature of the complex.

McDonald and coworkers further probed the nucleophilic reactivity of [Cu^II(PDCA)(O₂^•−)]⁻ in 2014.^[91] This complex was shown to react with acyl chlorides to afford the carboxylic acid and with certain aldehydes to undergo deformylation (Scheme 5). Mechanisms involving a Cu(III)-OOR intermediate were proposed. This constituted the first time that a cupric-superoxide complex was shown to react as a nucleophile.

Scheme 5.

Nucleophilic reactivity of [Cu^II(PDCA)(O₂^•−)]⁻.^[91]

5.4.2 Neutral Ligands

Itoh and coworkers have reported the synthesis of the only series of N₃ (rather than N₄ as for TMPA and its analogues) cupric-superoxide species supported by neutral ligands known to date, which have been characterized by UV-Vis, resonance Raman, and EPR spectroscopies. Each complex is hypothesized to have a four-coordinate tetrahedral geometry (without any solvent ligation) and has been shown to hydroxylate an intramolecular aliphatic C–H bond. Such ligand geometry and substrate functionalization strongly mimics the structure and reactivity of enzymes like PHM and DβM.[^[67,68,92]] This could especially be seen in the similar Hammett ρ values obtained in reactivity studies with DβM and these cupric-superoxide complexes (Scheme 6).

Scheme 6.

Comparison of Hammett reactivity studies done for: A) DβM;^[94] and B) PEDACO-EtPh-R^[68] cupric-superoxide complexes.

The supporting ligands of these complexes, unlike the other neutral ligands that have been used for stabilizing cupric-superoxide species, do not contain strongly electron-donating groups,^[53] sterically bulky groups,^[62] or substituents that can participate in hydrogen bonding interactions.^[56] Instead, Itoh and coworkers chose 1,5-diazacy-clooctane with a 2-(pyridyn-2-yl)ethyl donor substituent (PEDACO). The relevance of the eight-membered cyclic diamine moiety on the stabilization and reactivity of cupric-superoxide species has been highlighted. It was observed that any change in the size or composition of the diamine ring, including opening of the cycle or changing one nitrogen atom for a sulfur atom, yields Cu^I complexes that have a completely different reactivity (or are even unreactive) towards dioxygen.^[84,93] Thus, the superoxide complex is stabilized through the precise geometric configuration of the macrocyclic ligand.

5.4 Complexes Bearing a Binucleating Ligand Framework

While there have been multiple reports of mononuclear copper superoxide complexes over the last twenty five years, only two binuclear copper superoxide complexes have been published. In 1992, Karlin and coworkers published the synthesis of [Cu^II₂(UN-O-)(O₂^•−)]²⁺ via two different pathways.^[69] Oxygenation of a mixed-valent Cu^ICu^II species produced a bright green complex characterized by UV-Vis and EPR spectroscopies. Likewise, oxidation of the previously published peroxide dicopper(II) complex produced the same superoxide complex.

Very recently,^[70] a more detailed study involving [Cu^II₂(UN-O-)(O₂^•−)]²⁺ was reported. Using resonance Raman spectroscopy and DFT analysis, it was determined that there are two isomers present in solution (Table 1) that interconvert, with the superoxide anion binding in a µ-1,2 or µ-1,1 fashion. Interconversion of the superoxide µ-1,2 isomer and the peroxide complex could be accomplished using various ferrocene or ferrocenium derivatives. The standard reduction potential of the superoxide/peroxide complexes couple was determined to be E°= + 0.13 V (vs. SCE, in CH₂Cl₂ solvent), in the range of some biologically relevant redox species.

In 1998, Kitagawa and coworkers reported the synthesis of [Cu^II₂(OH)(O₂^•−)(hexpy)]²⁺ via the addition of hydrogen peroxide to the bis-µ-hydroxide dicopper(II) starting material.^[71] The verification of the superoxide complex was supported by UV-Vis, EPR, and resonance Raman spectroscopies, as well as ESI-MS. A µ-1,1-terminal bridging mode of the superoxide ligand was proposed.

6. Conclusion

With the plethora of experimental data collected from both enzymatic and model systems, and aided by computational studies, it is possible to say that the most likely mechanism of hydroxylation in the monooxygenases PHM, DβM, and TβM is Pathway 1 (Scheme 1). Multiple cupric-superoxide model complexes have been synthesized over the past 25 years, and more recently, their reactivity has been reported. With the exception of [Cu^II(PDCA)(O₂^•−)]⁻, which has been shown to have nucleophilic character to its reactivity, probably due to its dianionic ligand PDCA, cupric-superoxide model complexes have been reported to undergo electrophilic HAT with C–H and O-H containing substrates. It was shown that [Cu^II(^PVTMPA)(O₂^•−)]⁺ undergoes HAT with BNAH, and does not carry out hydride abstraction. Therefore, this superoxide complex does not participate in Pathway 2 reaction chemistry. The mechanistic studies with [Cu^II(^DMMTMPA)(O₂^•−)]⁺ showed that this superoxide complex is capable of oxidizing strong O-H bonds. HAT was proposed to be the mechanistic pathway due to KIE studies, product analysis, and comparison of reactivity to the well-known organic hydrogen atom acceptor, the cumylperoxyl radical.

Comparing the reactivity thus far carried out for these small molecule model systems with enzymatic experimental work shows remarkable similarities. KIEs calculated for HAT from C–H and O-H using cupric superoxides have been calculated to be 12.1 and 12 for [Cu^II(^PVTMPA)(O₂^•−)]⁺ and [Cu^II(^DMMTMPA)(O₂^•−]⁺, ^[54,56] respectively. These values are similar to the values reported for PHM reacting with hippuric acid and DβM reacting with dopamine (10.6 and 10.9, respectively),^[7] although the synthetic studies have been carried out at low temperatures in organic solvents. Hammett analysis of C–H activation of para-substituted phenethylamines by DβM and the PEDACO-EtPh-R series showed similar ρ values of −0.4 and −0.63, respectively. Thus, if there is a relevance of the chemistry determined for these synthetic compounds, whose study has revealed previously unknown basic and fundamental chemistry of copper(I) dioxygen interactions, it is that support emanates for a cupric superoxide as the reactive intermediate for PHM, DβM, and TβM (i.e. Pathway 1). However, due to the lack of more experimental evidence for LPMOs, it is not possible to claim the mechanism of action in these monooxygenases.

While many years of research originally went into stabilizing cupric superoxides to study their electronic and physical properties, recent efforts have been aimed towards synthesizing more oxidizing intermediates. To date, cupric-superoxide model complexes can oxidize exogenous substrates with BDEs of O-H bonds ranging from 69.4 to 82.3 kcal mol⁻¹[^[46,54,95]] and of C–H bonds ranging from 70.7 to 73.7 kcal mol⁻¹,^[55,56,96]] although the aliphatic C–H bond in PEDACO-EtPh-R is presumably much stronger (BDE of ethylbenzene is 85.4 kcal mol⁻¹).^[96] While these are impressive oxidation reactions at cryogenic temperatures, most model systems have not been able to achieve the same reactivity of the monooxygenase enzymes. The substrates commonly used in hydroxylation with PHM and DβM (hippuric acid and dopamine) have C–H bond strengths of 87 and 85k cal mol⁻¹, respectively.^[14] As can be seen, the reactivity of cupric-superoxide model complexes with exogenous substrates so far falls short of that known for the native enzymes.

Future studies of cupric-superoxide model complexes should take cues from previous reports to find ways to generate more reactive species. The role of thioether ligation in the enzyme and in the model complex [Cu^I(^DMAN₃S)]⁺ is still largely not understood and merits further investigation. To date, only one cupric-superoxide complex has been published featuring hydrogen bonding in the secondary coordination sphere. How hydrogen bonding affects the stability and reactivity of cupric-superoxide complexes remains to be seen and future studies should expand our knowledge. Finally, as seen with the series of TMPA derivatives, Cu^II/Cu^I reduction potential seems to play a role in, or is an indication of, the reactivity of cupric-superoxide complexes. Complexes with higher (more positive) reduction potentials make cupric superoxides that are more reactive. The interplay of these three factors (thioether ligation, hydrogen bonding, and reduction potential) needs further exploration and the conclusions drawn will enrich our understanding of fundamental copper-oxygen chemistry and its role in biology.

Biographies

Jeffrey J. Liu obtained his Bachelor’s degree in Chemistry from Boston University in 2013. Currently he is working as a Ph.D. student in the Karlin Lab at Johns Hopkins University. His research interests include studying the interaction of O₂ with mononuclear copper(I) complexes, and/or nitrogen monoxide (NO), which includes investigation of peroxynitrite-copper species which may form.

Daniel E. Diaz was born in Santiago, Chile and received his Bachelor’s degree in Chemistry from Universidad de Santiago de Chile. He came to the USA with a Fulbright fellowship, and joined Johns Hopkins University, where he is currently working towards his Ph.D. in chemistry under the direction of Kenneth D. Karlin. His research focuses on the biomimetic study of primary copper-(di)oxygen adducts present in the proposed catalytic mechanisms of Cu-containing monooxygenases, such as PHM and LPMOs.

David A. Quist received his B.S. degree in chemistry from the University of Michigan in 2013. He is currently a Ph.D. student working in the lab of Dr. Karlin at Johns Hopkins University. His work is focused on binuclear copper model complexes, their O₂-ad-ducts and their redox interconversions, and their further reactivity with nitric oxide.

Kenneth D. Karlin is the Ira Remsen Professor of Chemistry and the current department Chair at Johns Hopkins University in Baltimore, Maryland, USA. Educated at Stanford University (B.S. 1970) and at Columbia University, New York (Ph.D. 1975), he was a N.A.T.O. postdoctoral fellow at Cambridge University in England before being appointed Assistant Professor of Chemistry at SUNY Albany (Albany, New York, USA) in 1977. He moved to the Johns Hopkins University as professor in 1990. Dr. Karlin is Editor-in-Chief for Progress in Inorganic Chemistry (John Wiley & Sons) and holds or has held advisory or administrative positions with the Society for Biological Inorganic Chemistry (SBIC), the Petroleum Research Fund (PRF) (of the American Chemical Society (ACS)) and the Division of Inorganic Chemistry (DIC) of the ACS, most recently as 2013 DIC Chair (elected). He is also a fellow of the American Association for the Advancement of Science, and winner of a 2009 ACS National Award, the F. Albert Cotton Award in Synthetic Inorganic Chemistry. He has been organizer/chair of a number of international meetings on copper and/or bioinorganic chemistry, the 1998 Metals in Biology Gordon Research Conference and the 1989 International Conference on Bioinorganic Chemistry (ICBIC-4). Dr. Karlin’s bioinorganic research focuses on coordination chemistry relevant to biological and environmental processes, involving copper and/or heme (porphyrin-iron) complexes and their chemistry with molecular oxygen, its reduced derivatives, and nitrogen oxide compounds.