Bio-mimicking galactose oxidase and hemocyanin, two dioxygen-processing copper proteins

The modelling of the active sites of metalloproteins is one of the most challenging tasks in bio-inorganic chemistry.

Copper proteins form part of this stimulating field of research as copper enzymes are mainly involved in oxidation bio-reactions.

Thus, the understanding of the structure–function relationship of their active sites will allow the design of effective and environmental friendly oxidation catalysts.

This perspective illustrates some outstanding structural and functional synthetic models of the active site of copper proteins, with special attention given to models of galactose oxidase and hemocyanin.


The study and modeling of the active site of copper-containing proteins is a field of great interest within the scientific community.1

Indeed, proteins containing copper ions at the active site are mainly involved as redox catalysts in a range of biological processes, such as electron transfer, dioxygen transport and oxidation of various bio-substrates (Fig. 1).2,3

Thus, new and very effective bio-inspired homogeneous catalysts for common reactions under mild conditions may be discovered.4

Copper-containing proteins are widely found in plants, insects, and mammals and have been divided into seven classes according to the structure of their active site,5–7 namely type-1, type-2, type-3, multicopper, CuA, CuB, and CuZ clusters.

The present review primarily deals with the background of the structural and functional modeling of the type-2 and the type-3 active sites of, respectively, galactose oxidase (GOase) and hemocyanin (Hc).

These two metalloproteins have received considerable attention, and some recent outstanding developments have been reported in the literature, updating the future outlook of copper chemistry.8–11

In this perspective, a number of biomimetic coordination compounds are presented together with their potential applications in catalysis, or drug discovery.

Galactose oxidase: A type-2 copper protein


Galactose oxidase (GOase) is an enzyme with a type-2 copper site, first isolated in 1959,12 that selectively catalyzes the oxidation of primary alcohols to the corresponding aldehydes and simultaneous reduction of molecular oxygen to hydrogen peroxide.13

The active site of GOase involves a mononuclear copper(ii) species with a distorted square pyramidal geometry.

As the oxidation reaction requires 2 e and copper is a 1 e redox metal ion, one of the phenol ligands is most likely involved in the oxidative transformation.

Indeed, in the active form of the bio-catalyst, the equatorial plane is formed by two nitrogen atoms from two histidine imidazoles, one oxygen atom of an ortho-S-modified tyrosyl radical and one oxygen atom of a water molecule.

The axial position is occupied by a second tyrosinate ligand (Fig. 2, species A).

The catalytic mechanism of GOase has been extensively studied (Fig. 2).14

The primary alcohol first coordinates to the active species A, leading to the metal–phenoxyl radical complex B.

This species undergoes proton abstraction from the substrate by the axial tyrosinate, followed by a rapid intramolecular electron transfer from the intermediate ketyl radical anion with reduction of CuII to CuI.

The copper(i) species C reacts with dioxygen to form the hydroperoxo copper(ii) complex D with the release of the aldehyde.

Finally, hydrogen peroxide is released to give back the active form of the enzyme.

Over twenty years, a number of structural models have been reported in the literature15,16, but it appears to be more difficult to model the catalytic activity of GOase17.

Structural and functional models of GOase

The first effective biomimetic catalysts were described in 1998 by Stack etal. 18 and by Wieghardt et al. 19


Stack et al.18 described a Cu(ii) species, namely [Cu(ii)BSP] where BSP symbolises a salen-type ligand with a binaphthyl backbone and thioether functions (Fig. 3), able to catalyse the oxidation of benzylic and allylic alcohols under dioxygen at room temperature.

The [Cu(ii)BSP] complex has spectroscopic characteristics similar to those of GOase.

0.01 to 0.06% of this copper catalyst can convert neat benzylic and allylic alcohols to their respective aldehydes or ketones in the presence of a basic co-catalyst (Table 1).

Up to 1000 turnovers (TON) were achieved and no further oxidation of the aldehyde products was observed.

The oxidation reactions can also be performed in acetonitrile solutions, although the catalyst is much less efficient in this solvent (Table 1).


Wieghardt et al.19 similarly reported the catalytic oxidation of primary and secondary alcohols by a dinuclear Cu(ii)–phenoxyl complex at 20 °C under air.

The reaction of CuICl with 2,2′-thiobis(2,4-di-tert-butylphenol) and triethylamine (1∶1∶2) in dry methanol under air leads to a mononuclear blue copper complex (Fig. 4).

This coordination compound was synthesized to model the active site of GOase.

Its crystal structure shows a non-expected complex where the sulfur atom of the ligand coordinates to the uncommon tetracoordinated CuII ion (Fig. 5b).

In addition, only one triethylamine is coordinated to the metal center while two were required to mimic the N donors of the histidine residues, and no methanol molecule is present as substrate model (Fig. 5).

Furthermore, a diamagnetic copper compound is formed in THF solution which exhibits spectral features characteristic for coordinated phenoxyl groups (λmax ≈ 404 nm, ε = 8.0 × 103 L mol−1 cm−1; Raman: λexc = 458 nm, ν(C–O˙) = 1451 cm−1).

This complex is most likely dinuclear in solution with two bridging phenolate groups and two phenoxyl ligands as depicted in Fig. 6.

0.2% of the [CuICl(2,2′-thiobis(2,4-di-tert-butylphenol))(NEt3)2] catalyst prepared in-situ at 50 °C in THF was used to oxidise several alcohols at room temperature in air and the most important results are summarized in Table 2.

Thus, ethanol was converted to acetaldehyde with a yield of 63% (TON = 315), but the reaction is not chemoselective.

Indeed, several side-products resulting from C–C coupling were detected (Table 2).

The same results were observed for the oxidation of benzyl alcohol with the formation of 60% (TON = 300) benzaldehyde, small amounts of 1,2-glycol and α-hydroxyketone, and 3–4% of α-diketone (Table 2).

Contrary to GOase, this copper catalyst is also able to oxidise secondary alcohols.

Thus, 2-propanol and diphenylcarbinol can be used as substrates but, in both cases, only the 1,2-diol derivatives resulting from an oxidative C–C coupling are formed with a yield of 61 and 68%, respectively (Table 2).

In contrast, the aerobic oxidation of 2-butanol exclusively leads to the formation of butan-2-one (yield = 39%; TON = 195), suggesting two different reaction pathways for the formation of the carbonyl compounds and for the formation of the 1,2-glycols (Fig. 7).19

Ethanol, benzyl alcohol or 2-butanol coordinate to the proposed dinuclear copper complex (Fig. 6) to produce the active species I (Fig. 7).

After H abstraction by one of the two phenoxyl groups, a radical is created which is oxidised to the corresponding carbonyl derivative by the second phenoxyl radical.

In the case of 2-propanol and diphenylcarbinol, two alcohol molecules coordinate to the dinuclear moiety resulting in species II (Fig. 7).

The H abstractions by the two phenoxyl ligands generate two secondary radicals which can undergo C–C coupling to give the 1,2-glycol products.

GOase obviously represents an attractive target for the development of bio-inspired approaches aimed at designing selective and environmentally friendly, green oxidation catalysts. 20


The two pioneering bio-mimicking catalysts described above have inspired the scientific community resulting in a number of oxidation Cu-catalysts being reported over the past five years.22

Nevertheless, none of these catalytic systems are able to selectively convert primary alcohols to the corresponding aldehydes like the natural enzyme.13

Recently, some of us developed a catalytic procedure for the selective aerobic oxidation of primary alcohols to aldehydes based on a CuBr2(Bpy)–TEMPO system (Bpy = 2,2′-bipyridine).21,23

The reactions are carried out under air at room temperature and are catalysed by a [copperII(bipyridine ligand)] complex and 2,2,6,6-tetramethylpiperidinyl-1-oxy (TEMPO) and base as co-catalysts (Fig. 8).

Several primary benzylic, allylic and aliphatic alcohols have been successfully oxidized with excellent conversions (61–100%) and selectivities as no by-products were detected (Table 3).

These results show that the oxidation of activated alcohols is faster than aliphatic ones, demonstrating that the hydrogen abstraction from the α-carbon atom by TEMPO (Fig. 9b) is most likely the rate-determining step.

No reaction was observed with 1-phenylethanol, an activated secondary benzylic alcohol, or with octan-2-ol which can be associated with the activity of GOase.

Indeed, the proposed active species for this catalytic oxidation can be considered as a model of the active site of GOase (Fig. 9).

The bipyridine N atoms mimic the two histidine ligands, while TEMPO plays the role of both the phenolate and the phenoxyl groups and is most likely involved as a hydrogen acceptor.21

Competitive experiments were performed in order to confirm this high selectivity towards primary alcohols.

Therefore, mixtures of benzyl alcohol with octan-2-ol or 1-phenylethanol were reacted and no carbonyl products corresponding to the oxidation of the secondary alcohols were observed at all (Table 3).

This lack of reactivity of secondary alcohols may be explained by two factors affecting the catalytic cycle.

(1) Steric effects between the methyl groups of TEMPO and the alcohol can hinder the formation of the active species (Fig. 10a), crucial for the C–H abstraction.

(2) In addition, in the case of primary alcohols, the second β-hydrogen atom can be H-bonded to the oxygen atom of TEMPOH, stabilising the radical intermediate (Fig. 10b).

This bpy/TEMPO-based catalytic system should be regarded as the first synthetic functional model of galactose oxidase as both, the achieved chemoselectivity and the proposed reaction mechanism are close to those of the biological copper enzyme.

Nevertheless, this functional model is not able to compete with the natural enzyme in terms of catalytic efficiency.

Indeed, the rate of turnover is only 0.006 s−1 while it is 800 s−1 for GOase.24

The goal of future research investigations is therefore to improve the proficiency of the catalyst to obtain an economically interesting system for an industrial application.

Hemocyanin: A type-3 copper protein


Hemocyanin (Hc) is a protein with a type-3 copper site, initially studied in the early 70's25 and the crystal structure of its active form was first reported in .199326,27

Hcs are copper-containing respiratory proteins, freely dissolved in the hemolymph of many arthropod and mollusc species.28

This function is related to the capacity of Hc to reversibly bind molecular oxygen at the active site that is a dinuclear copper center.

The protein is thus found in two different forms, i.e. deoxy-Hc, containing a [Cu(i) Cu(i)] pair with a Cu–Cu distance of 4.6 Å, and oxy-Hc.

In the latter form, dioxygen was found to bridge between both copper(ii) ions separated by a distance of 3.6 Å, reflecting an electron transfer process from Cu(i) to dioxygen, where the peroxide dianion is coordinated as a bidentate ligand in a µ-η22 fashion (Fig. 11).

The active site of Hc involves a dinuclear copper species.

In the active form of the biomolecule, the equatorial plane of each CuII ion is formed by two nitrogen atoms from two histidine imidazoles and the two oxygen atoms of a dioxygen molecule.

The axial position is occupied by a third histidine ligand (Fig. 11).

Extensive synthetic efforts have been made to prepare CuII dioxygen complexes using different types of ligands in order to provide a better insight into the binding mode of molecular oxygen in oxyhemocyanin.

In this context, various binding modes of dioxygen have been brought to light and their characteristic spectroscopic features are reported in Table 4.

Structural and functional models of Hc

Monucleating ligands

As early as 1988, Karlin et al. described the first structurally characterized Cu2O2 species using a tripodal N4 ligand, namely tris(2-pyridylmethyl)amine (Fig. 12).29

This dioxygen adduct is intensely purple coloured while oxyhemocyanin is intensely blue.

The complex shows strong absorptions at 440, 525, and 590 nm and an additional d–d band at 1035 nm, and Raman vibrations at 832 cm−1 and 561 cm−1.

These spectroscopic data are not characteristic of the µ-η22 dioxygen binding mode observed in oxy-Hc (see Table 4), and have been assigned to peroxo-to-copper charge transfer transitions, and to an intraperoxide O–O stretch and a copper–oxygen stretch, respectively.

The 832 cm−1 stretching vibration clearly indicates the oxidation state of the dioxygen moiety to be O22−.

X-ray data of the thermally and moisture sensitive material were obtained at –90 °C.

It contains an end-on trans µ-1,2 peroxo group (viz. µ-1,2-O22−) bridging the two copper(ii) ions (Fig. 12; Table 4) (Cu⋯Cu = 4.36 Å).

This peroxo compound is only stable below −70 °C.30

A very important step forward in type-3 copper protein biomimetics was reported in 1989 when Kitajima et al.31 reported a copper–dioxygen complex obtained from an anionic tripodal N3 ligand, i.e. tris(3,5-diisopropylpyrazolyl)borate ligand (Fig. 13).

This deeply purple coloured compound is a stable solid at room temperature or even in non-coordinating solvent (e.g. CHCl3, CH2Cl2) below −10 °C.32

This good stability of the CuII–dioxygen adduct is probably due to the isopropyl groups of the ligand which shield the coordinated molecular oxygen (Fig. 13).

The most remarkable feature observed in its X-ray structure appeared to be the side-on µ-η22-peroxo coordination that holds the two copper units together.

Moreover, the physical properties reported are qualitatively and quantitatively comparable to those of oxy-Hc (See Table 4).

This complex indeed shows UV-Vis absorptions at 349 nm (ε = 21,000) and 551 nm (ε = 800), while for oxyhemocyanin the values are: 340 nm (ε = 20,000) and 580 nm (ε = 1,000).

The low value of the O–O stretch of 741 cm−1 (resonance Raman) also matches the value reported for oxyhemocyanin (744–752 cm−1).

Dinucleating ligands

The assumed advantage of using dinucleating ligands to prepare dioxo complexes is their ability to overcome the entropic barrier by keeping two N3 or N4 donor sets close to each other, at a chosen distance defined by the spacer length.

The spacers used to connect two coordinating units can be alkyl chains,33 phenoxo bridges,34 xylyl bridges35 or cyclohexyl bridges.36

Thus, Karlin et al. prepared a dinucleating ligand by linking two mononucleating tris(2-pyridylmethyl)amine (TPA) ligands through an ethylene bridge.37,38

A similar trans µ-1,2-peroxo-dicopper(ii) species could be obtained with this ligand (Figs. 12 and 14), which shows the characteristic UV-Vis absorptions (420 nm, 540 nm, and 600 nm) and Raman resonance (O–O stretch at 830 cm−1) for this type of dioxygen binding mode (Table 4).

However, this bis-TPA peroxo complex has a half-life time of 20 s at room temperature, while the TPA-peroxo species is only stable below −70 °C.

This significant enhancement of the stability clearly reflects the benefit achieved through the bridging of two TPA moieties.39

Very recently, Kodera et al. 40 took advantage of all the research investigations performed in this field to synthesize a ligand featuring almost every beneficial characteristic which are: (1) a N3 donor set in order to obtain the biomimetic side-on µ-η22 dioxygen binding mode of oxy-Hc; (2) a bridge between two monucleating units to increase the stability of the peroxo complex prepared; (3) and sterically hindered alkyl groups close to the N donor atoms to further enhance the stability of the µ-1,2-O22−-dicopper(ii) adduct.

Thus, Kodera et al. prepared the ligand 1,2-bis[2-(bis(6-methyl-2-pyridyl)methyl)-6-pyridyl]ethane which possesses all these essential features.

Indeed, a µ-η22 peroxo complex could be synthesized from this ligand which was found stable, at room temperature, for 3.6 h in dichloromethane solution.40

As expected, the peroxo complex exhibits the typical spectroscopic data of the binding mode encountered in oxy-Hc (Table 4) that are UV-Vis absorption bands at 366 nm and 537 nm, and a Raman O–O stretch at 765 cm−1.

In addition, the dioxygen binding is reversible which confers to this biomimetic compound, the status of best synthetic model of hemocyanin reported so far.

Macrocyclic ligands

One step further than dinucleating ligands is the macrocyclic system.

The believed benefit of using macrocyclic ligands in coordination chemistry is the high predictability of their metal complexes.

In principle, macrocyclic ligands can be designed to hold the proper donor groups in a preorganized rigid three-dimensional environment.41,42

Therefore, the macrocyclic ligand usually has a lower degree of conformational freedom in solution which may prevent the formation of unexpected or unwanted compounds upon coordination to the metallic centres.

Furthermore, highly reactive copper-peroxo species trapped in a macrocycle, might be stabilised.

Recently, the use of the macrocyclic ligand MePy22Pz (Fig. 15) by some of us also produced a room-temperature stable Cu2O2 species.43

Single crystals suitable for X-ray diffraction could not be obtained, but the spectroscopic data (viz. UV-Vis and resonance Raman) are comparable to those observed by Karlin (Section 3.2.1).

Thus, a trans-µ-1,2-peroxo-dicopper(ii) structure has been assigned, which is room temperature stable for 0.5 h in dichloromethane and air-stable for months in the solid-state (Fig. 16)44.

New applications

Synthetic models of tyrosinase45 or catechol oxidase46 have found many applications in catalysis.47

An important and rising field of application is the design of copper-based anti-cancer drugs.48

This perspective will highlight just a few of these last ones.

One way to eradicate cancer cells is to cleave their DNA,49 like the cleavage of the DNA strands through the oxidation of the sugar moieties (Fig. 17).50

This type of cytotoxic activity is shown by some copper complexes which act as synthetic chemical nucleases.51

One goal in the field of anti-cancer drug discovery is to synthesize metal complexes able to cleave both strands of a DNA molecule.

Indeed, a single-strand break (SSB, Fig. 18a) leads to damaged DNA which can often be easily repaired by the cellular enzymatic machinery.

On the other hand, if both DNA strands are broken (DSB, Fig. 18b), the restoration of the initial double-stranded supramolecule becomes almost impossible, the two DNA pieces being far apart.

Consequently, a drug able to perform DSBs will be very effective, as are the known natural bleomycin52 and neocarzinostatin.53

The relaxation of supercoiled circular DNA (form I) into relaxed (form II) and linear (form III) forms is commonly used to quantify the relative cleavage efficiencies of different complexes (Fig. 19).

A single-strand break (SSB) leads to form II while more than one SSB result in form III which is practically not repairable for the cellular enzymes.

A very effective cytotoxic drug would be a synthetic nuclease capable of performing a direct DSB (Figs. 18b and 19).

Indeed, more the one SSB on the same DNA molecule is statistically barely probable.

Recently, Karlin et al. 54–56 tested the peroxo complexes described in Sections 3.2.1 and 3.2.2 (See Figs. 12 and 14), as DNA cleaving agents.

The results are depicted in Fig. 20.

Both coordination compounds are active, but the mononuclear complex obtained from the tripodal ligand tris(2-pyridylmethyl)amine (TPA) is not effective in linearizing supercoiled DNA, as only the circular form II is detected on the agarose gel (Fig. 20a).54

In contrast, the dinuclear copper complex obtained from the ethylene-bridged TPA ligand (see Fig. 14) is capable of mediating the conversion of supercoiled DNA (Form I) to its linear form III (Fig. 20b).54

The results suggest that the dinuclear complex is able to perform more than one SSB (possibly a DSB) on the same DNA molecule while the mononuclear complex obviously cannot (See Fig. 19).

In addition, this significant difference in activity is consistent with their relative abilities to activate molecular oxygen (See Sections 3.2.1 and 3.2.2)54.

Concluding remarks

The study and modeling of the active site of dioxygen-processing copper proteins such as galactose oxidase or hemocyanin was started some fifteen years ago with the pioneering work of Karlin29 and Kitajima.31

However, the first outstanding applications ensuing from these research investigations have only been reported some ten years later, in 1998.

Nevertheless, over the past six years, an increasing amount of applied synthetic models of the type-257 and type-358 active site of copper proteins have been described in the literature almost weekly.

Thus, a considerable development can be expected in the coming years in this still incipient field, demonstrating the importance of bio-mimicking nature.

Copper oxidation chemistry is assured a great future.