JBIC Journal of Biological Inorganic Chemistry

, Volume 15, Issue 4, pp 461–483

CuA centers and their biosynthetic models in azurin


  • Masha G. Savelieff
    • Department of ChemistryUniversity of Illinois at Urbana-Champaign
    • Department of ChemistryMichigan State University Dubai
    • Department of ChemistryUniversity of Illinois at Urbana-Champaign

DOI: 10.1007/s00775-010-0625-2

Cite this article as:
Savelieff, M.G. & Lu, Y. J Biol Inorg Chem (2010) 15: 461. doi:10.1007/s00775-010-0625-2


CuA is a binuclear copper center that functions as an electron transfer agent, cycling between a reduced Cu(I)Cu(I) state and an oxidized mixed-valence Cu(+1.5)···Cu(+1.5) state. The copper ions are bridged by two cysteine thiolate ligands and form a copper–copper bond, the first reported of its kind in Nature. Such a “diamond-core” Cu2S(Cys)2 structure allows an unpaired electron to be completely delocalized over the two copper ions and contributes to its highly efficient electron transfer properties. This review provides accounts of how the CuA center was structurally characterized and highlights its salient spectroscopic properties. In the process, it introduces the CuA center in four different systems—native protein systems, soluble protein truncates of native proteins, synthetic models using organic molecules, and biosynthetic models using proteins as ligands—with a greater emphasis on biosynthetic models of CuA, especially on new, deeper insights gained from their studies.


BiosynthesisElectron transferProtein engineeringCytochrome c oxidase



Cytochrome c oxidase


Circular dichroism


Charge transfer


CuA in amicyanin


CuA in azurin


Electron–nuclear double resonance


Electron paramagnetic resonance


Electron transfer


Extended X-ray absorption fine structure


Highest occupied molecular orbital


Magnetic circular dichroism


Molecular orbital


Nitrous oxide reductase


Resonance Raman


Type 1


Type 2


X-ray absorption spectroscopy


Metalloproteins play crucial roles in vital cellular processes, from aerobic [15] and anaerobic [6] respiration to photosynthesis [7, 8]. Proteins are strategically compartmentalized within these multistep processes so that electrons flow down electron transfer (ET) chains to provide reducing equivalents for these biological processes [9]. Many of these ET processes are performed by metal centers within proteins and the electrons are transferred from one center to the next. Numerous types of metal ion based ET centers have been found in biology [10], from the ubiquitous iron–sulfur clusters (2Fe–2S, 4Fe–4S) [11, 12] and hemes [13, 14], to the well-established blue type 1 (T1) copper center [1522], and more recently to binuclear purple CuA centers [2331]. Iron–sulfur clusters, heme groups, and blue T1 copper centers span a broad range of reduction potentials and are employed throughout the electron transport chains. As a new member of this ET family, however, the purple CuA sites have been found to date only at the terminal electron acceptor position of transport chains such as in cytochrome c oxidase (CcO), (the terminal electron acceptor in aerobic respiration) [5, 26, 32], in nitrous oxide reductase (N2OR), (the terminal acceptor in anaerobic respiration) [33, 34], and in the terminal oxidase from Sulfolobus acidocaldarius (SoxH) [35]. A number of reviews on CuA have been published previously [2331].

Before the CuA center was discovered, copper sites in proteins had been classified into only three types on the basis of their spectroscopic properties: mononuclear T1 blue copper sites, which exhibit an intense blue color [1522]; mononuclear type 2 (T2) or normal copper sites [3638], whose spectral features are similar to those of common copper complexes found outside biological systems; and binuclear Cu(II)Cu(II) type 3 copper sites, where both Cu(II) species are antiferromagnetically coupled [39, 40]. The CuA center, however, exhibited very different spectroscopic signatures, such as an intense purple color, and a Cu(+1.5)···Cu(+1.5) resting state with one unpaired electron delocalized over both of its copper ions [2331]. It functions as an ET center like the previously known and well-established T1 copper centers, so researchers were initially puzzled by its existence. Blue T1 copper centers, whose reduction potentials vary depending on their axial ligand [4144], span a broad range of reduction potentials and could fulfill the role of CuA at the terminal position. So why did Nature employ CuA sites? What properties of the site impart advantages that make it suitable for its role at the terminal position of the electron transport chain?

CuA is a binuclear copper center with two cysteine ligands bridging the copper ions to form a Cu2S2 “diamond-core” structure (Figs. 1, 2, 3, 4, insets) [26, 3234]. CuA is unique among biological metal centers as the first example of a metal–metal bond in biology and therefore is of great interest to bioinorganic chemists. The Cu–Cu bond in CuA was debated initially, but was established later mostly from the short distance (at approximately 2.5 Å) observed in extended X-ray absorption fine structure (EXAFS) studies [45, 46] and X-ray crystallography [26, 3234], and the Cu–Cu stretching frequency observed in resonance Raman (RR) spectroscopy [47]. The coordination environment of each copper atom is completed by a histidine ligand to form a trigonal NS2 coordination environment with one additional, but weak and distant axial ligand. The axial ligands are a methionine at one copper and a backbone carbonyl at the other (Fig. 1, inset). The oxidized resting state of the cluster is best described as mixed-valence [48] Cu(+1.5)Cu(+1.5) with the one unpaired electron fully delocalized over both copper ions [4954]. The one-electron-reduced form is the Cu(I)Cu(I) state. CuA therefore acts as a one-electron ET center [29, 31]; the more oxidized Cu(II)Cu(II) state has not been observed to date [55, 56]. Oxidized, resting CuA is intensely purple in color in contrast to the blue T1 copper and the red (cysteine-coordinated) copper proteins. The purple color arises from S(Cys) → Cu charge transfer (CT) bands in the visible region (approximately 480 and 530 nm) and from Cu(+1.5)Cu(+1.5) intervalence CT bands (approximately 760–800 nm) (Fig. 3, inset) [4951, 54]. The reduced Cu(I)Cu(I) form is colorless.
Fig. 1

Ribbon diagram of the cupredoxin chain B of Paracoccus denitrificans cytochrome c oxidase (CcO); the intramembrane helices are evident in the background. Inset CuA site in P. denitrificans CcO [Protein Data Bank (PDB) ID 1AR1]. The CuA ligand set includes two copper bridging cysteine ligands (Cys216 and Cys220), one terminal histidine on each copper (His181 and His224), and one weak axial ligand interaction to each copper (thioether of Met227 and peptide bond oxygen of Glu218). The copper–copper distance is short, approximately 2.5 Å [26]. (Rendered using VMD [211])

Fig. 2

Ribbon diagram of the truncated CuA domain from Thermus thermophilus CcO. Inset CuA site in T. thermophilus CcO (PDB ID 2CUA). The CuA ligand set includes two copper bridging cysteine ligands (Cys149 and Cys153), one terminal histidine on each copper (His114 and His157), and one weak axial ligand interaction to each copper (thioether of Met160 and peptide bond oxygen of Gln151). The copper–copper distance is short, approximately 2.5 Å [70]. (Rendered using VMD [211])

Fig. 3

Ribbon diagram of a biosynthetic model of CuA in azurin (CuA-Az) from Pseudomonas aeruginosa (PDB ID 1CC3) [90]. a CuA site in CuA-Az (inset). The CuA ligand set includes two copper bridging cysteine ligands (Cys112 and Cys116), one terminal histidine on each copper (His46 and His120), and one weak axial ligand interaction to each copper (thioether of Met123 and peptide bond oxygen of Glu114). The copper–copper distance is short, approximately 2.4 Å. b UV–vis spectroscopic properties of CuA from CcO (green line) and biosynthetic CuA-Az (purple line). The UV–vis spectrum of blue T1 copper is shown for comparison (blue line). c X-band electron paramagnetic resonance (EPR) spectrum of CuA from CuA-Az (purple line). The parallel signal shows distinct hyperfine splitting, which arises from a seven-line pattern from delocalization of one electron over two copper nuclei. The EPR spectrum of blue T1 copper is shown for comparison (blue line). Az azurin, CT charge transfer, WT wild type

Fig. 4

Ribbon diagram of the cupredoxin domain from Pseudomonas nautica nitrous oxide reductase (N2OR). Inset CuA site in P. nautica N2OR (PDB ID 1QNI). The CuA ligand set includes two copper bridging cysteine ligands (Cys561 and Cys565), one terminal histidine on each copper (His526 and His569), and one weak axial ligand interaction to each copper (thioether of Met572 and peptide bond oxygen of Trp563). The copper–copper distance is short, approximately 2.55 Å [33]. (Rendered using VMD [211])

The history of CuA

The discovery, full electronic description, and structural determination of CuA have been the result of work spanning several decades [28]. The advances could not have been made without contributions from researchers in multiple disciplinary fields including molecular biologists, biochemists, biophysicists, and inorganic/bioinorganic chemists. Even today, CuA is still a widely studied and active field of research. The history of CuA can be traced through the interweaving stories of two different proteins: CcO [1, 4, 5] and N2OR [6]. The evolution of ideas and experimental progress made was complicated by different factors in each protein, but together they completed the picture of CuA.

The first report of CuA was in CcO, the terminal electron acceptor in aerobic respiration. Although the presence of copper in CcO did not become widely accepted until the 1970s, the debate went back as far as the 1930s, when the presence of copper in CcO was first implicated. The idea was contested and supported by opposite camps over several decades [28]. However, with the advent of modern protein purification techniques and more accurate methods for the determination of copper, it became clear that CcO did indeed contain copper and a copper-to-iron ratio was established [5760]. Earlier, the structure of this new site, termed CuA, was assumed to be mononuclear owing to an underestimation of the copper content. To complicate the situation, the spectroscopic signatures of the site were obscured by overlapping signals from heme centers, which made spectroscopic characterization of the site difficult. The protein was also membrane-bound and therefore challenging to purify and work with.

An absorbance found at 830 nm came to be diagnostic of this CuA site since this absorbance, assigned to CuA by a magnetic circular dichroism (MCD) study [61], was free from any overlap from the strongly absorbing heme signals. A double-resonance MCD experiment further characterized CuA-attributed bands at 475, 525, and 830 nm [62, 63]. Electron paramagnetic resonance (EPR) spectroscopy at various fields also demonstrated a hyperfine splitting structure which could only be assigned to copper [6466]. With the advent of molecular biology came the preparation of soluble fragments of CuA from native proteins [35, 6772] and biosynthetic models [7376] which allowed the beautiful purple color of CuA and its associated spectroscopy to be displayed without being masked by colors and spectra of other metal centers. It was then firmly established that the CuA visible spectrum was dominated by peaks at approximately 480 and 530 nm, with a broad band at approximately 800 nm. The EPR spectrum of CuA could also now be observed free from heme signals.

While the CuA site was being investigated in CcO, studies were also under way in N2OR, the terminal electron acceptor in anaerobic respiration [7781]. Unlike CcO, N2OR was a copper-only metalloprotein with unusual spectroscopic signatures that indicated the presence of two new types of copper–thiolate sites distinct from the previously known blue copper sites. The copper content suggested that they were probably binuclear in nature [81]. The consensus sequence derived for one of the copper sites in N2OR indicated that the coordination ligands included two cysteine residues, two histidine residues, and a single methionine residue and was probably very similar to that of CuA from CcO [8284].

Determination of the N2OR copper sites was complicated by the uncertainty in the copper content and the apparent shortage of cysteine residues. The spectroscopy indicated two types of copper–thiolate clusters [81], yet the lack of cysteine residues to account for them as well as an underreported copper content led to the hypothesis that CuA was the same as the catalytic site in the protein termed CuZ [85]. However, this hypothesis was later discarded as the enzyme was investigated further. EPR spectroscopic studies on [15N]histidine-labeled protein firmly established a clear seven-line pattern, which was attributed to a Cu(+1.5)Cu(+1.5) oxidation state of a binuclear copper site [50, 52]. The binuclear site in N2OR was proposed to be similar to the CuA site from CcO [83].

Various binuclear structures were proposed for CuA on the basis of EXAFS [45, 46, 86] or MCD [87, 88] data, with one group proposing a structure which came very close to the actual structure of CuA [88]. The debate was definitively laid to rest with the publication of three crystal structures containing the CuA center in 1995, independent of each other [26, 32, 89]. They were CuA from native Paracoccus denitrificans CcO [32] (Fig. 1) and native bovine CcO [26] and a restored CuA center in a soluble fragment of cyoA from Escherichia coli (vide infra) [89]. The structure of the site was also reaffirmed later by the publications of crystal structures of a truncated CuA domain of Thermus thermophilusba3-type CcO (Fig. 2) [70], and a biosynthetic model of CuA in azurin (CuA-Az) (Fig. 3) [90], and N2ORs from both Pseudomonas nautica (Fig. 4) [33, 91], and P. denitrificans [34]. The crystal structures of N2OR also established the high similarity between CuA-CcO, CuA-N2OR, and CuA-Az.

Spectroscopic properties of CuA

The binuclear nature of the copper site with its two bridging cysteine ligands imparts many of the unique and very interesting electronic and spectroscopic properties to CuA [10, 4951, 54, 92, 93]. These properties have been elucidated by UV–vis absorption, MCD, RR, EPR, electron–nuclear double resonance (ENDOR), paramagnetically shifted 1H NMR, and EXAFS investigations.

UV–vis studies

The UV–vis spectrum of CuA in native, fully assembled CcO is masked by transitions from heme groups, but the near-IR band at 830 nm is still discernable. CuA’s complete spectrum free of other spectral signals has been obtained from soluble truncates of CcO from various organisms [35, 6772] and from the biosynthetic models CyoA [73, 89], CuA-Az [75, 94], and CuA in amicyanin (CuA-Ami) [74] (vide infra). Their UV–vis spectra establish the presence of two intense transitions at approximately 480 and 530 nm, with weaker bands at approximately 350 and 800 nm (Table 1, Fig. 3b). The UV–vis signatures of the CuA site within N2OR are also similar despite some overlap with signals from other copper sites. The spectra of CuA from N2OR are more discernable in N2OR(V), a mutant of N2OR which only contains only a CuA site [80].
Table 1

UV–vis features for CuA from various native enzymes (including soluble truncates) and biosynthetic models



λmax (nm)

CcO native (aa3) [67]

Paracoccus denitrificans

363, 480, 530, 808

CcO native (ba3) [69]

Thermus thermophilus

365, 480, 530, 790

CcO native (caa3) [68]

Bacillus subtilis

360, 480, 530, 790

N2OR native [34]

Paracoccus denitrificans

480, 540, 850

N2OR native [80]

Pseudomonas stutzeri

350, 480, 540, 780

N2OR native [95]

Achromobacter cycloclastes

350, 481, 534, 780

CuA-CyoA (biosynthetic model) [96]

Escherichia coli

358, 475, 536, 765

CuA-Ami (biosynthetic model) [74]

Thiobacillus versutus

360, 483, 532, 790

CuA-Az (biosynthetic model) [75, 94]

Pseudomonas aeruginosa

350, 485, 530, 765

CcO cytochrome c oxidase, N2OR nitrous oxide reductase, Ami amicyanin, Az azurin

A combined low-temperature UV–vis, circular dichroism (CD), and low-temperature MCD study on Bacillus subtilis truncated CcO, biosynthetic CuA-Az and CuA-Ami, and two small-molecule models has led to a full molecular orbital (MO) description of the CuA site and assignment of all spectral peaks [49]. The CD and MCD spectra reveal ten transitions in the 300–1,100-nm range which overlap in the UV–vis region to give a UV–vis spectrum dominated by transitions at approximately 350, 480, 530, and 800 nm for CuA.

The MO description of CuA involves Cu–S orbital combinations as the lowest in energy. Taking the Cu2S2 diamond core to lie on the xy plane, this leaves the S px and py orbitals available to combine with the Cu-based dxy and \( d_{{x^{2} - y^{2} }} \) orbitals. The S pz orbitals are involved in bonding to the cysteine β-carbons, which are perpendicular to the Cu2S2 plane and lying along the z-axis. Next higher in energy are the d-orbital combinations between the \( d_{{z^{ 2} }} , \)dxz, and dyz orbitals, which combine to form six occupied MOs. The highest in energy are the combination of dxy and \( d_{{x^{2} - y^{2} }} \) orbitals owing to their antibonding interaction with the S ligand orbitals; their combination forms four orbitals: σg, \( \sigma_{u}^{*} ,\pi_{g}^{*}, \) and πu. In the MO scheme for the small-molecule model, based on D2h symmetry, the highest occupied MO (HOMO) ψ* has πu designation. However, the increased Cu–Cu interaction in CuA enhances the energy splitting of the \( \sigma_{g}{-}\sigma_{u}^{*} \) pair, which raises the energy of \( \sigma_{u}^{*} \), making it the HOMO ψ*. The lower symmetry from D2h to Ci in the CuA protein environment relaxes σ/π designations. These high-energy orbitals in CuA demonstrate a significant level of direct Cu–Cu orbital overlap owing to the short (approximately 2.44 Å) Cu–Cu bond distance.

All ten transitions have been assigned on the basis of the MO description and a comparison of the UV–vis spectra for CcO truncate, biosynthetic CuA, and small-molecule models [49]. A concomitant RR excitation profile supports some of the transition assignments. The most intense absorption bands, at approximately 480 and 530 nm, are assigned to S(Cys) → Cu CT, specifically S(px) → Cu ψ*(HOMO) and S(py) → Cu ψ*(HOMO). The more intense, higher-energy transition is assigned to the S(px) → Cu ψ*(HOMO) transition because the S px orbital has greater overlap with the Cu ψ*(HOMO). Enhancement of these two bands in their RR excitation profile supports their assignment as CT bands.

The broad absorbance at approximately 800 nm is attributed to a ψ → ψ* Cu–Cu intervalence transition which is associated with electron delocalization of the mixed-valence site. The RR excitation profile of this band also supports a direct Cu–Cu interaction and its assignment as an intervalence transition. The absorption band is broad because it overlaps with several, lower-intensity dd transitions. The absorption at approximately 350 nm has been assigned to two overlapping N(His) → Cu transitions and not to any sulfur-based transitions owing to its RR profile, which shows resonances attributable to a histidine ring vibration. There are two N(His) → Cu transitions due to the two inequivalent histidine rings.

The high-intensity S(cys) to Cu CT transitions in CuA are redshifted compared with their small-molecule thiolate counterparts. The reason for this phenomenon in purple CuA centers is the same as the reason for the redshift in blue copper proteins compared with small-molecule thiolate complexes: the highly covalent nature of the Cu–S(Cys) bond and the low coordination number resulting from the weakened axial ligand interaction to the copper center [93].

MCD and RR studies

Early MCD spectra were the key to unambiguously assigning the 830-nm peak in the absorption spectrum of CcO to CuA [61]. Combined MCD–EPR double-resonance experiments on CcO further uncovered absorbances at 475 and 525 nm, as well as the absorption band at 830 nm, associated with CuA [62, 63]. MCD has also been employed to investigate CuA sites in N2OR [51, 85, 97] and some biosynthetic models [49, 94]. Before the X-ray crystal structure of CuA was known, MCD spectra ruled out CuA as a mononuclear site. The MCD features could only be explained by a binuclear site with bridging cysteine ligands [87, 88]. The best proposed structure for CuA before the crystal structure was based on the MCD data [88], and MCD spectra also played a key role in assigning electronic absorption bands and understanding electronic structures of CuA [49]. RR has also been employed to study CuA, including comparison of the excitation modes using isotopic substitution [47, 49, 98101]. The experiments demonstrate that the symmetry of the site is no higher than Ci [47, 49, 100, 101]. Although this is not true of the Cu2S2 core, which has idealized D2h symmetry, the different axial ligands on each copper break the symmetry and reduce it from idealized D2h to Ci.

EPR and ENDOR studies

The EPR spectrum of CuA in CcO at X and Q bands shows a superhyperfine structure from at least two nitrogen atoms and the low g values implicate at least one cysteine ligand [66, 102, 103]. The hyperfine structures from copper are not observed at X band but are easily detected at low field (2–4 GHz); the seven-line hyperfine structures suggested the presence of a binuclear copper center, even before the crystal structure was solved, and played a critical role in determining the binuclear nature of the center [64, 65]. Multifrequency EPR on CcO [104] and N2OR [79, 105, 106] suggested that both contained similar copper sites and were both probably binuclear [52, 53, 104]. The most resolved seven-line pattern was obtained by EPR on 63Cu- or 65Cu-labeled N2OR, or on N2OR doubly labeled with [15N]histidine [50]. This isotopically labeled N2OR exhibited a clear seven-line pattern in a 1:2:3:4:3:2:1 intensity ratio arising from a single electron delocalized over two copper nuclei (I = 3/2, n = 2), which the authors described using a detailed MO description [50]. Iterative extended Hückel (IEHT) and unrestricted Hartree–Fock (UHF-INDO/S) calculations were used to fit the EPR data to provide the best MO description of CuA. The MO scheme which resulted involved four bonding Cu–S MOs, four antibonding Cu–S MOs, and six predominantly metal-based nonbonding MOs. The orbital designations were based on D2h symmetry although Ci is a better representation of CuA’s symmetry. On the basis of their MO scheme, the authors debated between two possible ground-state electronic configurations: 2B2u and 2B3u. ENDOR experimental results demonstrated significant (approximately 3–5%) electron spin density on the imidazole nitrogens [103], so 2B2u was eliminated as the ground state because it has a very weak electron spin density on NHis. The authors therefore proposed 2B3u as the ground-state electronic configuration of CuA. The ground-state HOMO was determined to be highly covalent, with 34% of the total electron density shared between the two copper ions and 60% of it shared by the two sulfur atoms from the bridging cysteine ligands. The doubly labeled 65Cu–[15N]histidine–N2OR protein was also studied by ENDOR to interpret the CuA hyperfine couplings [107].

The seven-line hyperfine pattern has since been observed and verified in various soluble truncates [35, 68, 69, 73, 108, 109], biosynthetic models (Fig. 3c) [74, 75, 94, 96, 109], and a wide range of small-molecule models [110118] and has come to be diagnostic of the fully delocalized mixed-valence state of CuA.

Paramagnetically shifted 1H NMR studies

Copper(II) ions contain an unpaired electron whose relaxation time (τs) is long, on the order of 10−9 s, compared with the relaxation times of other paramagnetic centers. This unpaired electron’s magnetic moment interacts with neighboring protons, resulting in highly broadened, paramagnetically shifted 1H NMR signals, and rendering them quite uninformative in elucidating numerous Cu(II)-binding protein sites [119, 120]. This disadvantage can be overcome by magnetically coupling Cu(II) to other paramagnetic metal ions [such as Co(II) and Ni(II)], thereby sufficiently shortening τs for Cu(II) to obtain useful spectra. This has been extensively applied, with great success, to Cu2Co2 superoxide dismutase, which has native diamagnetic Zn(II) replaced with paramagnetic Co(II) [119, 121, 122]. One of the most exciting developments in the spectroscopic studies of CuA centers is the discovery that the CuA center naturally has an extremely short relaxation time, which makes the site amenable to paramagnetically shifted 1H NMR study [123130]. The paramagnetically shifted peaks are sharp and narrow, producing NMR spectra with resolved peaks which are assignable. The reason for such a short relaxation time is the presence of low-lying excited states for CuA. NMR spectroscopy is a solution-based method that can be performed at room temperature, which offers significant advantages compared with EPR (frozen glass or single crystal) and X-ray crystallography (single crystal). Solution-based methods provide a more biologically relevant picture of the protein as it would exist in a cellular environment. Low-lying excited states are accessible at room temperature, and allow the protein to sample different conformations which may have relevance for dynamic processes such as protein folding, substrate binding, and substrate recognition, and for enzyme catalysis. Paramagnetic 1H NMR investigations have been performed on a number of CuA centers in native proteins [128], soluble truncates of native proteins [123, 124, 126, 127, 129, 130], and biosynthetic CuA-Ami [125]. Moreover, paramagnetic NMR has been used to describe the ground-state electronic structure of CuA in a solution environment at physiologically relevant temperature [130]. Paramagnetically shifted 1H NMR peaks corresponding to both cysteine and histidine ligands of CuA have been observed, although resonances to the distal ligands methionine and backbone carbonyl have not been detected.

Comparison of 1H NMR investigations on unlabeled and 13C- and 15N-labeled T. thermophilus CuA truncate allowed full assignment of all proton peaks [130]. Heteronuclear NMR studies on labeled protein involving 13C NMR and 1H,13C heteronuclear multiple quantum correlation experiments led to a full assignment of all ligand-based carbon resonances using a combination of direct and indirect experiments. Previous methods had demonstrated the nonequivalence of the two CuA histidine ligands [103, 123, 125]; however, the NMR study provided the first evidence for the nonequivalence of the cysteine ligands as well. The technique detected a greater electron spin density on Cys153 (T. thermophilus truncate numbering) than on Cys149. Inspection of the crystal structures shows that Cys149 is involved in a hydrogen bond to the amide backbone from Gly115 [32, 33, 70, 90]; this hydrogen bond is conserved in all reported CuA structures and has been likened to the hydrogen bond in blue copper proteins between the copper Cys ligand and the amide backbone [131133]. This hydrogen bond in blue copper proteins has been shown to be important in modulating the reduction potential of the site [134] and an analogous scenario may be in effect in CuA. Another application of NMR spectroscopy to CuA is deuterium exchange studies to give an indication of the extent of solvent exposure of the CuA site within various models and the extent of copper–ligand participation in hydrogen-bonding with the solvent [125, 127, 128].

Another exciting avenue for paramagnetic NMR spectroscopy has been the exploration of the low-lying excited states and the source of the discrepancy for the energy gap between the ground state and the first excited state as determined by NMR spectroscopy and EPR spectroscopy. NMR spectroscopy has shown the energy difference between the ground-state electronic structure and the lowest-lying excited state to be around 350–600 cm−1 [127, 130]. On the other hand, EPR spectroscopy has shown this energy difference to be 3,000–4,500 cm−1 [49, 50]. A significant difference between EPR spectroscopy and NMR spectroscopy is the temperature at which the technique is performed. EPR spectroscopy on CuA is performed at around 5–15 K, whereas NMR spectroscopy is a room temperature technique. The source of the discrepancy has been debated, which has led to a couple of theories.

One theory proposed by Abriata et al. [130] suggests that the discrepancy may be resolved if one assumes that NMR spectroscopy observes a temperature-averaged population of the ground state and the first excited state [127, 130]. To identify the first excited state, a 1H and 13C NMR temperature dependence study determined the Curie behavior of the resonances and assigned the first excited state as πu [130]. The assignment of the ground-state electronic configuration was based on the short Cu–Cu bond (approximately 2.4 Å), which would enhance the σ contribution to the HOMO ψ* and make \( \sigma_{u}^{*} \) the ground state [49]. Computational work has shown the energy gap between \( \sigma_{u}^{*} \) and πu to be highly sensitive to bond length [49, 54]. This bond length, at room temperature under the conditions of NMR, would be lengthened, lowering the energy gap between \( \sigma_{u}^{*} \) and πu to 350–600 cm−1 as determined by NMR spectroscopy.

Computational work from the Solomon group [135] on a model compound for CuA has suggested a different reason for the difference arising in the energy gap between NMR and EPR. The calculated excited-state potential energy surface for the CuA\( \sigma_{u}^{*} \) ground state is flat, and is in thermal equilibrium with a πu state from an elongated (approximately 2.9 Å) Cu–Cu bond. Since paramagnetic NMR spectra show only one set of resonances for each proton, the exchange between the \( \sigma_{u}^{*} \) ground state and the πu excited state is fast on the NMR time scale. On the other hand, the EPR-measured energy gap is a Frank–Condon excited state measuring approximately 3,200 cm−1.

X-ray absorption spectroscopy and EXAFS studies

EPR and MCD data had indicated that CuA contained two copper ions which were close enough for copper hyperfine splitting or exciton coupling. However, the earliest structural indication that CuA contained a Cu–Cu bond was from EXAFS data where the scattering of signals suggested a direct Cu–Cu bond [45, 46, 136140] particularly in analogy to a small-molecule binuclear copper azacryptand [45]. Although the proposed structures of CuA based on EXAFS data were not exactly the same as in the final structure, the EXAFS-measured Cu–Cu bond lengths for CuA were 2.5 Å in a soluble truncate from B. subtilis [45] and 2.46 Å in bovine heart CcO [46], which were in excellent agreement with the value from the crystal structures. EXAFS data have also been collected on the CuA sites from various soluble truncates [70], biosynthetic models [94, 141], and native N2OR enzymes [97, 142, 143] as well as on a selenomethionine-substituted variant of CuA-CcO truncate [144]. EXAFS has also demonstrated that there are very few changes in the geometry of CuA upon reduction of the oxidized Cu(+1.5)Cu(+1.5) form [138].

The covalency of the Cu–S bond in CuA was determined by sulfur K-edge and copper L-edge X-ray absorption spectroscopy (XAS) [141, 145]. The results from sulfur K-edge XAS demonstrate that the two cysteinate sulfurs contribute 46 ± 2% to the HOMO covalency. Copper L-edge experiments demonstrated a 44 ± 2% HOMO contribution, leaving around 10% combined contribution from the two histidine ligands. These results are in agreement with the findings using other spectroscopic techniques which demonstrate that the HOMO has predominantly copper and sulfur character [49, 50]. XAS data on blue T1 copper plastocyanin were also obtained for a comparison with CuA [141]. The amount of covalency measured for plastocyanin was 38%. Although this value is less than for CuA, it is over a single Cu–S bond as compared with 46% for CuA over two Cu–S bonds. This demonstrates that the contribution from anisotropic covalency in CuA is as significant as in blue copper proteins for their role in ET.

Comparison of the CuA in proteins with that in a mixed-valence small-molecule model demonstrates a similar level of Cu–S covalency in both molecules and therefore similar contributions to the electronic Cu–Cu coupling (2HCu–Cu) [141]. However, UV–vis absorption spectroscopy and MCD experiments demonstrated that 2HCu–Cu is approximately 5,600 cm−1 for the mixed-valence small molecule model, but approximately 13,400 cm−1 for CuA in native proteins [49]. The difference in energy of 2HCu–Cu for the two molecules must arise from direct Cu–Cu overlap in CuA, which has a significantly shorter Cu–Cu bond of approximately 2.4 Å compared with approximately 2.9 Å for the mixed-valence small molecule.

A measure of the bond covalencies of CuA has an important relevance for its ET function [49]. Electron entry into CuA is via intermolecular transfer from CcO’s redox partner cytochrome c. Intramolecular ET out of CuA is via heme a of subunit I followed by transfer to heme a3 (in the CuB catalytic site of CcO). The electron exit pathway has been proposed via a His–Cu bond [2, 146] to heme a owing to a lower number of intervening bonds. However, the high degree of anisotropic covalency in CuA makes the highly covalent Cu–S bond a viable electron exit route despite the greater number of intervening bonds to heme a owing to stronger acceptor–donor coupling [141, 145]. A similar scenario for the intermolecular ET from cytochrome c demonstrates that the Cu–S bond may also be an important factor in enhancing the rate of ET from cytochrome c [141]. In both instances, the CuA site enables extremely rapid ET rates compared with other ET sites, but with significantly smaller driving forces. The electron delocalization and highly covalent nature of the Cu–S bonds could be important contributors [141] as well as the low reorganization energy [141, 147].

Reduction potentials

The reduction potential of CuA has been measured in native enzymes, soluble truncates from various organisms, and biosynthetic models of CuA (Table 2) employing a variety of techniques, such as cyclic voltammetery, redox titration, and spectroelectrochemistry. The values are mostly in close agreement with each other, with a spread of approximately 80 mV from 216 mV at one extreme to 297 mV at the other. All the reduction potentials reported are for the Cu(+1.5)Cu(+1.5) → Cu(I)Cu(I) couple; the Cu(II)Cu(II) → Cu(+1.5)Cu(+1.5) couple has never been reported and the Cu(II)Cu(II) state has not been confirmed yet [56].
Table 2

Reduction potentials measured for CuA from various native enzymes, soluble truncates, and biosynthetic models



Potential versus NHE (mV)

Solution pH

CcO native [148]

Bos taurus



CcO native [149]

Paracoccus denitrificans



N2OR native [150]

Paracoccus denitrificans



SoxH native [35]

Sulfolobus acidocaldarius



CcO subunit II truncate [69]

Thermus thermophilus



CcO subunit II truncate [56]

Thermus thermophilus



CcO subunit II truncate [151]

Thermus thermophilus



CcO subunit II truncate [152]

Thermus thermophilus



CcO subunit II truncate [72]

Synechocystis PCC 6803



CuA-Az biosynthetic model [153]

Pseudomonas aeruginosa



NHE normal hydrogen electrode

Functional properties

The main features that make CuA a unique cluster are its fully delocalized mixed-valence binuclear core, a short Cu–Cu bond, bridging thiolate ligation, and low coordination number with an NS2 ligand set. A low reorganization energy and reversible ET between an oxidized Cu(+1.5)Cu(+1.5) and a reduced Cu(I)Cu(I) state are also essential for its function as an ET site. These features have been investigated in a number of ways, in the native enzymes, in soluble truncates, and in bottom-up approaches in small-molecule and biosynthetic models. Some of the studies lend valuable insights into the functioning of CuA, and are discussed later in this review.

Different CuA systems

Although it is satisfying to finally elucidate the three-dimensional structure of the CuA center, the structure in itself does not answer many questions about its other properties, such as electronic structure, spin and magnetic properties, and functional properties such as redox and ET. To elucidate these properties fully, a number of different CuA systems have been employed, each of which has its own advantages and challenges. The systems include CuA in native enzymes, soluble fragments, synthetic models, and biosynthetic models.

CuA in native proteins

The earliest studies were performed on native CuA in fully assembled CcOs and N2ORs. The CuA site has only been established in three proteins to date, all of them terminal electron acceptors: CcO, N2OR, and SoxH. Crystal structures for CuA have been solved in CcO, N2OR, soluble truncate and biosynthetic models of CuA (vide infra). Important cornerstone structures for CcO include the first P. denitrificans (IAR1) [32] (Fig. 1) and bovine heart (1OCC) [26, 146] structures. For N2OR, crystal structures for P. nautica (1QNI) (Fig. 4) [33], P. denitrificans (1FWX) [34], and Achromobacter cycloclastes (2IWF) [154] have been reported. No crystal structure for SoxH appeared in the literature, and the presence of a CuA site is based on spectroscopic characterization mainly by UV–vis and EPR spectroscopies [35].

Mutagenic studies of the CuA binding ligands in native, fully assembled P. denitrificans CcO and native Pseudomonas stutzeri N2OR have shown the cysteine ligands to be particularly important to the functioning of the enzyme and the spectroscopic signatures of CuA. Numerous studies had been performed prior to the structural elucidation of CuA, and most were performed to determine CuA’s ligand set. The results demonstrate the sensitivity of the native system to mutagenesis of CuA ligands and the critical role it plays as an ET site within the protein. For example, Speno et al. [155] reported on mutations of His161, His204, Cys196, Cys200, and Glu198 of CcO from yeast Saccharomyces cerevisiae, including a number of ligand swaps. On the basis of the evaluation of the expression level and proper folding of expressed yeast CcO of these mutants, the authors proposed that these cysteine and histidine ligands directly bound copper as their mutations almost completely abolished proper functioning and assembly of the enzyme. Mutagenic studies of the vital cysteine ligands were also performed on intact, fully assembled CcO from P. denitrificans [149]. Mutation of Cys216 to serine (Cys216Ser) resulted in a protein which exhibited a four-line EPR consistent with a blue T1 copper site, but which retained only 1% of wild-type activity. Mutation of the second cysteine residue to serine (Cys220Ser) resulted in a greater retention of copper binding and activity, but the spectroscopic properties were not elaborated. These types of investigations have not been limited to CcO, with similar observations being reported for mutagenesis of CuA ligands in N2OR. Mutation of Cys618 to aspartate resulted in almost complete loss of copper-binding ability and activity, but the mutation Cys622Asp retained some copper-binding ability and activity, although the spectroscopic signatures of CuA were abolished [143].

The weaker axial methionine ligand has also been investigated by mutagenesis within native fully assembled P. denitrificans [156]. The Met227Ile replacement resulted in a protein which retained two copper ions within the CuA site; however, the EPR showed a collapse of the characteristic wild-type CuA seven-line pattern to a four-line pattern in Met227Ile, indicative of valence trapping of the delocalized electron in wild-type CuA to a trapped Cu(II)Cu(I) configuration in the Met227Ile mutant. The near-IR spectrum showed a shift in the band at 831 nm in the wild-type CuA to 823 nm in the mutant with a concomitant decrease in the intensity. The ET properties of the mutant were also altered; the kinetics of ET from cytochrome c to CuA were not affected, but the rate of ET to heme a was significantly diminished in the mutant protein compared with the WT protein. The Cu(II)Cu(I) has an altered reduction potential which results in reduced ET properties and subsequently lowered enzyme activity.

The weak axial ligand has also been mutated in native CcO from Rhodobacter sphaeroides with a similar effect [157, 158]. The mutant (Met263Leu) was also shown to retain two copper ions which were still mixed-valence-delocalized as determined by EPR spectroscopy. The mutant was still capable of proton pumping but had a diminished enzyme activity which was only 10% of the native enzyme value. This was attributed to an increased reduction potential of 120 mV relative to the native enzyme. The same publication also reported on mutation of His260 to asparagine, which resulted in a significantly greater perturbation and disruption of enzyme activity. The results suggest that the CuA site is much less tolerant to mutation of its primary coordination sphere such as histidine than mutation of the more distant axial methionine, and that ET is significantly dependent on the primary coordination sphere, but less so on the weak axial methionine [157, 158].

CuA in soluble truncated variants of native proteins

Although studying the CuA center in its native proteins provides the most complete picture of its structure and the most relevant information about its function, it has been difficult to purify these CuA-containing proteins to homogeneity with full copper content in their active sites, particularly for the membrane-bound CcO. Furthermore, purification of N2OR requires strict anaerobic conditions to maintain an active protein with full copper content. Finally, studying the CuA site’s spectroscopy is complicated by masking from other spectral signals such as those from low-spin heme a and heme–copper in CcO and CuZ in N2OR. To overcome these problems, the native CcO proteins have been truncated to a small domain fragments containing the CuA center which are soluble and easily expressed and purified [35, 6772, 88, 159161]. For CuA in CcO, these soluble truncates have been prepared from subunit II of CcO, which contains the CuA site. The sequence for subunit II is modified at the genetic level to express only the soluble cupredoxin domain without the intramembrane helices of subunit II, which typically anchor the protein subunit to the mitochondrial membrane. These successful truncates are soluble and amenable to more spectroscopic techniques than the fully assembled native CcO systems [35, 6772, 88, 159161]. The truncates are very similar to one another in their spectroscopic properties (UV–vis, EPR, EXAFS) and in their reduction potentials. The only soluble truncate for which a crystal structure has been solved to date is from T. thermophilus (Fig. 2) [70].

CuA within native N2OR has also been investigated as a platform to understanding the site within a native protein [34, 150], and although it is a soluble periplasmic protein and does not need to be truncated, it is challenging to purify because it requires strict anaerobic conditions [80]. Therefore, a mutant form, N2OR(V), which is devoid of copper sites other than CuA, is often investigated to study the CuA site within N2OR [80].

Soluble CcO truncates are quite amenable to mutagenesis, and a number of systems have been investigated. A complete mutagenic study of the CuA ligands was performed on the P. denitrificans truncate by Farrar et al. [88] prior to the crystallographic elucidation of the CuA structure. They created single mutations of all the proposed CuA ligands and monitored changes in the UV–vis, MCD, and EPR spectral properties as indications that the ligand was vital to copper binding. In particular, they found that mutation of either of the cysteine ligands, now known to be the bridging ligands of the diamond-core structure, was essential to obtain the spectroscopic signature of CuA, with the mixed-valence state abolished, as shown by EPR spectroscopy. Copper-binding ability was diminished for Cys244Ser, and slightly lowered for Cys248Ser. Both mutants had a yellow color, with UV–vis properties which were pH-independent.

Stable mutations to the weakly axial ligands of CuA in CcO truncates remained elusive until relatively recently. Mutants of the axial methionine were prepared by mutagenesis in the T. thermophilus CuA truncate [162]. The study investigated two mutations: Met160Gln and Met160Glu. Both mutants retain essentially a wild-type CuA UV–vis spectrum but with slightly redshifted or blueshifted bands. The EPR spectra are more informative and demonstrate an increased amount of T2 copper overlapping with the predominant CuA signal in Met160Glu, which suggests a population of trapped Cu(II)Cu(I) sites within this mutant. The EPR spectrum is also more rhombic, indicative of a stronger axial ligand interaction for Glu(O)–Cu compared with the native Met(S)–Cu. The results from the investigation suggest that although the axial methionine is a weak ligand, the axial ligand can indeed modulate the electronic structure, and therefore the ET properties of the CuA site. The mutations were studied in more depth in an additional publication which reported the examination of the mutants by pulsed EPR/ENDOR to probe their detailed electronic structure [163].

A seleno-substitution of the sulfur atom of the axial methionine was also performed to obtain a Se-Met-CuAT. thermophilus truncate [144]. The substitution has very little effect on the optical properties of CuA, with identical UV–vis spectra for S-Met and Se-Met CuA, and almost identical EPR spectra. The reduction potential of both proteins was 240 mV as determined by cyclic voltammetry. Combined Se and Cu K-edge EXAFS demonstrates very little change in the CuA geometry in changing between the oxidized and the reduced form.

A recent paper demonstrated a dependence of the reduction potential on the axial ligand of the CuA site [152]. The mutations were performed on the soluble truncate from T. thermophilus and the mutations examined spanned a variety of mutations: Met160Leu, Met160Ser, Met160Tyr, Met160His. UV–vis and EPR data collected on the mutants demonstrated largely conserved wild-type CuA spectroscopic characteristics with some shifting of the UV–vis bands and some changes in g|| values, indicative of some valence trapped species. The reduction potentials were found to vary with the axial ligand from a low value of 148 mV (vs. the normal hydrogen electrode) for Met160His to a high of 348 mV for the Met160Tyr mutant compared with a native protein value of 293 mV. The axial mutations were therefore able to fine-tune the reduction potential of the site over a range of 200 mV.

Synthetic models of CuA

Because of the truncation, several soluble fragments often lack the interface portion that stabilizes the whole intact CuA domain in CcO, making them relatively unstable and prone to the formation of inclusion bodies. Furthermore, to completely understand the structure and function of the CuA center, it is often not enough to study CuA in its native proteins or truncates. Structural features important for electronic and functional properties of CuA could remain hidden in this “top-down” approach because it can elucidate only certain features that are necessary for function, but not all the features sufficient for the function. Therefore, the “bottom-up” approach of synthesizing models of CuA from scratch could complement the “top-down” approach, presenting the ultimate test of our knowledge about the CuA center [164167].

The traditional “bottom-up” approach has been the preparation of small inorganic molecules using organic ligands to model various protein metal sites such as blue T1 copper and purple CuA [168]. A good model of CuA would capture several important features of CuA, such as the short Cu–Cu bond with the bridging thiolates, the highly delocalized mixed-valence nature of the site, and a highly reversible Cu(+1.5)Cu(+1.5)/Cu(I)Cu(I) couple. To this end, a number of binuclear, bridged copper complexes have been prepared using azacryptands [110112, 169172], sulfido bridges [113, 173186], amido bridges [118], and carboxylato bridges [114117, 187]. Mixed-valence dicopper complexes have been around for a while [188, 189] but were not necessarily studied in the context of CuA such as in some of the earlier works cited above.

The investigation of small-molecule models revealed several interesting points. The amido-bridged molecule [([SNS]Cu)2][B(3,5-(CF3)2C6H3)4] [SNS is bis(2-tert-butylsulfanylphenyl)amido)] [118] (Fig. 5a), the carboxylate-bridged molecule [Cu2(BBAN)(μ-O2CCPh3)(OTf)](OTf) [BBAN is 2,7-bis((dibenzylamino)methyl)-1,8-naphthyridine, Tf is CF3SO2] [117] (Fig. 5b), and the amido-bridged molecules [Cu(t-BuNHC(O)NHCH2CH2NMe2)]2+ and [C(t-BuNHC(O)NHCH2CH2)3N]2+ [190] (Fig. 5d) are mixed-valence Cu(+1.5)Cu(+1.5) but with a ligand set that lacks the bridging thiolates. Therefore, these studies have shown that bridging sulfur ligands are not necessary for a fully delocalized mixed-valence Cu(+1.5)Cu(+1.5). In fact, azacryptands have also been reported with fully delocalized mixed-valence Cu(+1.5)Cu(+1.5) and extremely short Cu–Cu bonds, but they do not have any bridging ligands at all (Fig. 5c) [110112, 169172]. They raise interesting questions. For example, can the ON2 ligand set in [Cu2(BBAN)(μ-O2CCPh3)(OTf)](OTf) be adopted in a protein environment? Work on a double-cysteine mutant of biosynthetic CuA-Az examined whether the bridging cysteine ligands can be replaced with bridging oxygen donors, such as serine or aspartate, and still maintain the mixed-valence delocalized Cu(+1.5)Cu(+1.5) state (vide infra) [191]. In addition, [Cu2(BBAN)(μ-O2CCPh3)(OTf)](OTf) (Fig. 5b) is not symmetrical owing to the presence of a trifluoromethanesulfonate ion on only one of the copper atoms. The CuA site is also not fully symmetrical by virtue of the different axial ligands, methionine to one copper atom and backbone carbonyl to the other. Small molecules can also maintain a mixed-valence delocalized Cu(+1.5)Cu(+1.5) state in a slightly asymmetrical site. So the question raised is how unsymmetrical can a site become before the copper atoms are no longer of fully delocalized mixed-valence type and CuA function is impaired?
Fig. 5

Synthetic models of CuA using small organic molecules as ligands: a [{[SNS][Cu]}2][B(3,5-(CF3)2C6H3)4] [SNS is bis(2-tert-butylsulfanylphenyl)amido)] [118], b [Cu2(BBAN)(μ-O2CCPh3)(OTf)](OTf) [BBAN is 2,7-bis((dibenzylamino)methyl)-1,8-naphthyridine, Tf is CF3SO2] [117], c azacryptand binuclear model complexes [110112, 169172], d [Cu(t-BuNHC(O)NHCH2CH2NMe2)]2+ and [Cu(t-BuNHC(O)NHCH2CH2)3N]2+ [190], e (LiPrdacosCu)2(O3SCF3) (LiPrdacos is 1-isopropyl-5-ethylthiolato-1,5-diazocyclooctane) [113]

The above synthetic models do not contain the bridging thiolates present in CuA. Introduction of bridging thiolates is more challenging owing to the tendency for the two thiolates to form a disulfide bond in the presence of copper(II) ions. The thiolate-containing ligand (LiPrdacosCu)2(O3SCF3) (LiPrdacos is 1-isopropyl-5-ethylthiolato-1,5-diazocyclooctane) (Fig. 5e) has been successfully synthesized and shown to bind two coppers with a seven-line EPR pattern [Cu(+1.5)Cu(+1.5) resting state] [113]. This compound represents the closest synthetic model to the CuA reported to date. However, the Cu–Cu bond distance at 2.93 Å is significantly longer than it is in CuA (at 2.48 Å in N2OR from P. nautica) [33].

One of the most useful features of synthetic models is to elucidate structural elements responsible for electronic structures of the native proteins through comparison of spectroscopic features of synthetic models with those of native proteins. For example, the crystal structures of dicopper azacryptands have been known for some time [111], and have been shown to contain a short Cu–Cu bond of 2.36 Å. Before the crystallographic structure of CuA was reported, a short Cu–Cu bond was implicated by comparison of CuA’s RR [169] and EXAFS [45] data with the dicopper azacryptand’s data. Small-molecule models are also more amenable to density functional calculations and have been used to determine CuA’s reorganization energy in a model compound compared with that of blue T1 copper protein models [54].

Biosynthetic models of CuA

The above synthetic models have captured either the short Cu–Cu distance or the thiolates bridging to the binuclear copper. An ideal model would incorporate both structural features and display functional properties. To achieve this goal, ligand design is absolutely critical; in addition to a correct primary coordination sphere, noncovalent interactions in the secondary coordination sphere are important, such as the weak interactions in CuA between the axial thioether methionine with copper and the protein backbone carbonyl oxygen with the second copper (Figs. 1, 2, 3, 4). Small organic ligands have long been the method of choice in making synthetic models of metalloproteins [168] owing to the ease of synthesizing them and the high yield obtained in comparison with synthesis of proteins. However, recent advances in molecular biology and protein biochemistry have made it possible to synthesize a number of proteins as easily as, and sometimes even more easily than, some of the organic molecules, especially when elaborate secondary coordination spheres are required. Hence, biosynthetic models of CuA have been constructed using small, stable, easy-to-make and well-characterized proteins, such as azurin and amicyanin, as “ligands” [164167]. Biosynthetic models of CuA combine the advantages of ease of synthesis and characterization in small-molecule synthetic models with an environment similar to that of the native proteins, making it possible to make models closer to the native targets. In addition to sharing an ease of characterization with the soluble truncates, biosynthetic protein models make it possible to reveal features which are necessary for the structure and function of CuA, as they are built from the “bottom-up” approach using proteins that do not contain the structural elements for the formation of a CuA center [164167].

Biosynthetic CuA models have been prepared in quinol oxidase, a copper-free oxidase homologous to CcO and were named purple CyoA [73, 89, 192], as well as in the T1 copper proteins amicyanin (CuA-Ami) [74, 193, 194] and azurin (CuA-Az), through loop-directed mutagenesis (Fig. 6) [75, 94]. The designs are based on the versatility of the cupredoxin fold, which is an eight-stranded Greek key β-barrel motif. CuA resides in a cupredoxin fold in native CcO and N2OR. CyoA is a quinol ob3 oxidase, which uses quinol instead of CuA for ET. The CyoA loop analogous to the CuA binding loop of CcO has “lost” ligands to bind copper, but designing the CyoA loop by mutagenesis with the consensus sequence for CuA results in a purple protein termed purple CyoA (purple CyoA is a truncate as well but has been grouped with the biosynthetic proteins because the CuA loop was designed). In addition to complementing the studies of CuA centers in native proteins, soluble truncates, and synthetic models, these biosynthetic models have provided many new and sometimes unique insights. Such examples from the authors’ group are given in the next section.
Fig. 6

Preparation of biosynthetic model CuA-Az using loop directed mutagenesis (LDM). The blue T1 copper binding loop in azurin is replaced at the genetic level with the CuA binding loop from CcO [75]. Blue T1 copper and CuA both reside naturally in a cupredoxin fold, which is the basis of the LDM approach. Inset Intensely blue wild-type azurin before LDM and intensely purple CuA-Az after LDM. (Rendered using VMD [211])

Construction, characterization, and insights gained from study of biosynthetic models of CuA in Azurin (CuA-Az)

Construction and characterization of CuA-Az

CuA-Az is a biosynthetic model of CuA within the blue T1 cupredoxin protein azurin. The model was constructed on the basis of the structural homology between the protein fold of native CuA in CcO and N2OR to the cupredoxin fold of azurin. The major difference is the copper binding loop; whereas the ligand loop in azurin contains an amino acid sequence for the formation of the well-known mononuclear blue T1 copper, the corresponding ligand loop in CcO is responsible for CuA formation. Instead of site-directed mutagenesis, loop-directed mutagenesis was performed by replacing the blue T1 copper ligand loop with the corresponding CuA ligand loop from CcO (Fig. 6) [75]. The new protein, CuA-Az, binds a binuclear copper site with spectroscopic features very similar to those of CuA from native CcOs and N2ORs as determined by a host of techniques such as UV–vis absorption and EPR (Fig. 3, inset) [75, 94, 195], ENDOR [109, 196], XAS/EXAFS [94, 141, 195], and MCD and RR [47, 49, 195]. The biosynthetic protein CuA-Az has also been crystallized and the three-dimensional structure has been solved at 1.64-Å resolution [90], which shows that CuA-Az is almost identical to CuA-CcO (Fig. 7) in both the primary coordination sphere and the secondary coordination sphere, making it the closest model to the native CuA reported to date. Moreover, substitution of the blue copper binding loop of azurin with the purple copper binding loop of CuA maintains the overall structure of the azurin cupredoxin fold (Fig. 11a).
Fig. 7

Comparison of CuA from biosynthetic CuA-Az with CuA from native Paracoccus denitrificans (Pd) CcO. Top Top view of the Cu2S2 plane. Bottom Side view of the Cu2S2 plane

Insights gained from study of CuA-Az

Making a biosynthetic model whose structure is almost identical to that of native CuA has allowed us to determine the structural features that are sufficient to form the center from a “bottom-up” approach. However, that is not the only goal of the research. An even more important goal is to use such a small, stable, and simple model to gain insights that may otherwise be difficult to obtain from studying native CuA or other systems. Toward this goal, we have investigated the kinetics of copper incorporation, metal ion/charge selectivity, and the roles of cysteine, methionine, and histidine ligands in the CuA center. Comparison of CuA within CuA-Az with T1 copper in azurin has also demonstrated the utility of the biosynthetic method to gain a deeper understanding of the native CuA system, for properties such as its reorganization energy and its role in ET.

Kinetics of copper incorporation into CuA-Az

One interesting aspect of the CuA center is the facile formation of the mixed-valence Cu(+1.5)···Cu(+1.5) site even though only Cu2+ ions are added to the metal-free apoprotein without any exogenous reductants. To elucidate the mechanism of CuA formation, the kinetics of copper ion incorporation into the apo form of CuA-Az was monitored using stopped-flow UV–vis spectroscopy [197]. A “yellow” intermediate, characteristic of a copper(II)–thiolate center in a tetragonal geometry, is formed first and then converts to the final mixed-valence [Cu(+1.5)···Cu(+1.5)] CuA form (Fig. 8, top). Addition of an external reductant ascorbate results in less intermediate accumulation and more product formation (Fig. 8, bottom). A mechanism is proposed whereby Cu2+ is first incorporated into the protein to produce a tetragonal copper(II)–thiolate center more typical of inorganic copper complexes. The final CuA center is then formed after incorporation of Cu+, which is generated by reduction of Cu2+ either by the cysteine residues from apo-CuA-Az proteins or by exogenous reductants such as ascorbate [197]. The observation that ascorbate increases the amount of final product strongly suggests that reductants play an important role in copper ion incorporation into the CuA centers of CcO and N2OR in vivo.
Fig. 8

Proposed mechanism for incorporation of copper into apo-CuA-Az. Top In the presence of exogenous copper(II) only, a yellow type 2 (T2) intermediate forms, which is then converted to a mixed-valence Cu(+1.5)Cu(+1.5) CuA with the concomitant reduction of sacrificial apoprotein molecules [197]. Bottom In the presence of both exogenous copper(II) and ascorbate reductant; the ascorbate reduces copper(II) to copper(I) such that apoprotein molecules do not need to be sacrificed. Therefore, the final amount of holo-CuA-Az is greater

Metal-binding properties of CuA: preference for a +3 total charge

The facile formation of the mixed-valence Cu(+1.5)···Cu(+1.5) raised an interesting question about the metal ion specificity of the CuA center. The biosynthetic CuA-Az model can address this question quite simply, because the protein is generally purified in apo form, with exogenous copper typically added after purification to form the holo protein. Therefore, different metal ions can be added to the apo form in lieu of the customary copper. The various metal-reconstituted holoproteins may then be subjected to syringe-pump electrospray ionization mass spectrometry, wherein the protein is pumped into the electrospray ionization instrument in a buffer. Detected ions are therefore in native buffer-like conditions, and metal binding can be detected. Addition of Hg(II), Cd(II), Ag(I), or Au(I) to either apo-CuA-Az or holo-CuA-Az always resulted in M(II)M(I) derivatives such as Hg(II)Ag(I)–CuA-Az and Hg(II)Cu(I)–CuA-Az, demonstrating a strong preference for the site to adopt a +3 total charge within (Fig. 9) [198, 199]. In fact, the Cu(II)Cu(II) state has not been attained in the native, truncated or biosynthetic systems, and a reduction potential has not been determined for the Cu(II)Cu(II)/Cu(+1.5)Cu(+1.5) couple [56]. Currently, CuA has been seen to function solely as a one-electron ET center.
Fig. 9

Addition of various metal ions to holo-CuA-Az and apo-CuA-Az [198, 199]. a Addition of Hg(II), M(II) and M(I) to holo-CuA-Az, b addition of Hg(II), M(II), and M(I) to apo-CuA-Az, and c electrospray ionization mass spectrometry data to demonstrate the formation of the Hg(II)Ag(I) metal ion combination within the CuA-Az site

Role of the bridging cysteine ligands

Modeling studies have suggested that the cysteine ligand orbitals mix significantly with the copper orbitals, resulting in strong delocalization and the formation of superexchange pathways [49]. They are proposed to be responsible to a significant extent for the mixed-valence nature of the Cu–Cu interaction. Mutagenesis of CuA-Az has supported this proposal (Fig. 10). For example, Cys112Ser within CuA-Az still maintained a binuclear copper site, but with an altered UV–vis spectrum exhibiting a transition in the yellow region (approximately 400 nm) with a broad dd transition at approximately 750 nm [200]. The EPR was consistent with two T2 coppers in distinct but similar environments. On the other hand, Cys116Ser resulted in a more dramatic perturbation of the site, resulting in the collapse of the binuclear CuA site into a mononuclear T1 copper, with blue transitions in its visible spectrum and EPR spectra which were pH-dependent. At lower pH a spectrum typical of T2 copper was observed, but at higher pH the spectrum was more typical of a T1 copper. These pH-dependent changes were corroborated by low-temperature UV–vis scans performed across a range of pH values. Therefore, the cysteine residues are important to maintain the structural and functional integrity of the CuA site [200].
Fig. 10

Various mutations to the primary and secondary coordination environment of the CuA site. Mutation of the bridging cysteine ligands results in significant perturbation to the CuA site: Cys112Ser results in two T2 Cu sites, Cys116Ser results in one T1 copper site [200], and Cys112SerCys116Ser at low pH results in one T2 Cu site and at high pH results in two T2 Cu sites [191]. Mutation of His120X preserves the CuA site but with slightly diminished UV–vis intensity and altered EPR characteristics [201, 202]. Mutation of the axial methionine also preserves the CuA site; the UV–vis intensity and EPR characteristics are slightly altered from the mild perturbation [203]. Wt wild type

To investigate symmetry as a factor contributing to the mixed-valence nature of the binuclear CuA site, double cysteine to serine mutations of CuA-Az were also performed (Fig. 10) [191]. These mutations eliminated the strong σ-donor-type cysteine residues, altering the electronic structure of the site but maintaining a symmetrical environment about the copper active site. The copper-binding properties of the double mutants were investigated at high and low pH. At low pH a single T2 copper site was observed as shown by UV–vis spectroscopy and syringe-pump electrospray ionization mass spectrometry, so mutation of both cysteine residues clearly abolishes the CuA site. Moreover, the absence of cysteine ligands resulted in a pale T2 copper site devoid of any intensely colored S → Cu CT bands. Copper binding studies were performed at higher pH to enhance the copper-binding affinity of the serine ligands. This goal was indeed achieved since the protein binds two T2 copper sites at pH 8.5 instead of just one at pH 5.1. However, the mixed-valence delocalized nature of CuA was still not restored. The results demonstrate that symmetry is not a sufficient condition to build a CuA site without the bridging thiolates in protein models [191].

Role of the histidine ligands

The histidine ligands contribute only 3% to the HOMO on the basis of density functional theory calculations [49]. This is borne out by experimental results which demonstrate that His120 in CuA-Az is not required for the formation of the purple CuA site. A series of His120X mutations (where X is Gly, Ala, Asn, or Asp) were constructed (Fig. 10), and the resultant His120 mutants were shown to have spectroscopic properties very similar to those of the CuA site, including the purple color, although with a somewhat diminished stability [201]. This histidine ligand is therefore not essential for the formation of a native CuA site at relatively neutral pH and demonstrates the remarkable stability of the Cu2S2 diamond-core structure. Titrations with various exogenous ligands suggest that the His120 replacement ligand is not exchangeable and that an internal amino acid ligand must replace the mutated His120 residue [202]. An examination of the secondary coordination environment suggests Asn119 as a candidate, but the double-mutant constructs His120Gly/Asn119X (X is Asp, Ser, or Ala) retain the CuA spectroscopic signatures. The replacement ligand has not been identified to date, yet the formation of a native-like CuA site within the His120 mutants demonstrates the stability and rigidity of the CuA bridged-core structure.

Despite the His120X series having properties very similar to those of wild-type CuA, there are some properties that are altered as determined by careful multifrequency EPR and ENDOR studies on His120Asn and His120Gly [196]. The Q-band ENDOR perturbations of the cysteine β-protons and the remaining His46 protons do not reveal any discernible differences between His120Asn and His120Gly from wild-type CuA-Az and suggest a similar electronic structure with a mixed-valence binuclear core in all cases. Multifrequency EPR spectroscopic studies at X, C, and S bands indicate an alternative scenario where the characteristic seven-line CuA pattern collapses to a four-line pattern for His120Asn and His120Gly. The results suggested by EPR and ENDOR may be reconciled if valence localization to one copper atom occurs, yet the spin densities on the cysteine sulfur atoms and His46 are unaltered. Another possibility is that the site remains of mixed-valence type in the His120 mutants but with a significantly diminished hyperfine coupling for one of the copper atoms, such that the overall spin distribution remains unaltered.

Role of the methionine ligand

The axial ligand in blue copper proteins has been shown to significantly influence the reduction potential of the blue copper proteins by approximately 200 mV [41, 134]. Isostructural replacement studies of methionine with unnatural amino acids in azurin have shown the correlation to arise from the effect of the hydrophobicity of the axial methionine position on the reduction potential [42, 44]. Mutation of the axial methionine in the soluble truncate from T. thermophilus oxidase has resulted in a similar 200 mV redox potential change, from as low as 148 mV in Met160His to as high as 348 mV in Met160Tyr [152].

Since the axial methionines in both blue T1 copper and purple CuA play important roles in tuning the reduction potentials of the copper center, a natural question is whether the methionine at a similar distance from copper centers can exert a similar degree of influence on either mononuclear blue copper or dinuclear purple CuA. By placement of both blue copper and purple CuA in the same scaffold of azurin, the biosynthetic CuA model in azurin has provided a unique system to answer this question, because the Met–Cu distance in CuA-Az (3.07 Å) is similar to that in blue copper azurin (3.15 Å). Interestingly, although the methionine mutations in blue copper azurin resulted in an approximately 170 mV change in the blue copper reduction potential, the same mutation in purple CuA-Az caused potential changes of less than 25 mV under the same condition [203], suggesting that the same methionine exerts much less influence on the reduction potential of CuA than the same residue in blue copper at a comparable distance. A possible explanation may arise from the diamond-core structure of CuA, where the bridging geometry of the site is more resistant to influences from the axial positions.

The ability of CuA to resist changes in the reduction potential makes sense in the context of its biological role: whereas blue copper proteins are involved in a variety of ET functions that requires a wide range of reduction potentials to match those of redox partners, CuA has been observed to date only at the terminal position of electron transfer chains, where the driving forces are small. Therefore, a CuA site resistant to changes in the reduction potential is advantageous to prevent loss of ET caused by mutant sites. The presence of two different copper sites within the same protein scaffold (T1 copper in azurin and CuA in biosynthetic CuA-Az) is valuable in eliminating other factors that may influence the reduction potential, such as distance, solvent exposure, dielectric medium, and the secondary coordination environment of the protein.

ET properties and reorganization energy

The blue T1 copper centers are common and very efficient ET agents found in many biological systems [17, 2022, 93]. With the recent discovery of the CuA center as an ET agent that shares structural similarities with blue T1 copper centers, it was a natural step to compare its ET properties with those of the well-established blue T1 copper centers. Although ET kinetics have been measured in each of the two classes of proteins, it is still difficult to compare the inherent efficiency of ET of the two sites, because each metal-binding site resides in a different protein scaffold, and the protein contribution to the ET rate occurs through different protein pathways or media. A direct comparison of both copper sites within the same protein scaffold would minimize or eliminate differences in ET rate arising from protein contributions (Fig. 11, left). The biosynthetic approach places both blue copper and purple CuA in the same azurin scaffold, and thus provides an exceptional opportunity to compare ET properties of the two centers directly, a feature that is not possible to accomplish by studying native enzymes, their soluble truncates, or synthetic CuA models.
Fig. 11

A overlay of wild-type blue T1 azurin (PDB ID 4AZU) and biosynthetic CuA-Az (PDB ID 1CC3) demonstrates that there is very little change in the cupredoxin fold outside the ligand binding loop [90]. B line drawing of wild-type azurin overlaid on CuA-Az showing the possible electron pathways from the radiolytically reduced disulfide bond between Cys3 and Cys26 to the oxidized CuA or blue T1 copper center in electron transfer studies [147, 204]

The experiments examined the long-range ET rate from a radiolytically reduced disulfide bond within the CuA-Az protein to the oxidized Cu(+1.5)Cu(+1.5) site in comparison with the same ET to a blue copper site in azurin (Fig. 11, right) [147]. The results demonstrate that CuA is a more efficient ET site despite a lower driving force for transfer between the reduced disulfide and the oxidized site. This result may be explained by examining the reorganization energy of the two types of copper sites. The calculated reorganization energy of the CuA center is half that of the T1 copper center. This lower reorganization energy is attributed to the rigid diamond-core structure of CuA and the bridging nature of the site, because changes in reorganization are averaged over two bridging cysteine ligands and two copper ions in CuA compared with a single cysteine ligand and one copper ion T1 copper sites [147]. More recent results have also shown that the mixed-valence nature of the CuA site is only a secondary contributor to its lower reorganization energy [204]. The advantages of a binuclear site in this regard have been reported previously and verified experimentally by a comparison of EXAFS data of oxidized and reduced CuA which shows very little difference [138]. The low reorganization energy of CuA has also been verified in the soluble CcO truncate from T. thermophilus [56].

pH-dependent effects on CuA: a possible role of CuA as a gate for proton pumping

The CuA center is known to be the electron entry site in CcO, whose function is to couple the ET to O2 reduction and proton pumping. For CcO to function well without causing problems to biological systems, a “gate” is most likely required to trigger the ET, O2 reduction, and proton pumping when there is such a need. More importantly, if too many protons are pumped, the “gate” needs to be closed to stop ET, O2 reduction, and proton pumping. For a long time, the heme–copper center in CcO, the site of O2 reduction, was considered as a candidate for the “gate.” However, experimental evidence for such a proposal is not very strong. A recent pH-dependence study on the biosynthetic model CuA-Az and comparison with similar structural features in native CcO suggest that CuA may fulfill such a role [205].

Gradual pH titrations to holo-CuA-Az from pH 4 to pH 7 showed a blueshifting of the near-IR band from approximately 800 nm to approximately 750 nm, with a concomitant slight decrease and broadening of the 485–530-nm peaks [205]. Observation of isosbestic points indicated a transition between two species with a pKa of 4.8 ± 0.1 for a one-site protonation. In contrast to these small changes to the UV–vis spectrum, the typical seven-line EPR pattern of CuA collapses to a four-line pattern upon dropping the pH from 7 to 4. Further studies by MCD and RR spectroscopy and combined with density functional theory calculations showed that the unpaired electron is delocalized in the low-pH form and that the four-line hyperfine pattern results from the large EPR spectral effects of an approximately 1% 4s orbital contribution of one copper to the ground-state spin wave function upon protonative loss of its histidine ligand [195].

Examination of the residues in the vicinity of the CuA site has identified several residues which may be responsible for the ionizable pKa of 4.8, including Glu114 and His120. Therefore, Glu114Gln and His120Ala mutants were prepared to test the pH dependence of their UV–vis and EPR spectra. The pH dependence is maintained for Glu114Gln; the mutant retains a seven-line EPR pattern at high pH, which collapses to a four-line EPR pattern at low pH, suggesting that Glu114 is not the protonation site. In contrast, the pH dependence is abolished for the His120Ala variant, which shows a “low-pH” four-line pattern at all pH values, suggesting that His120 is the ionizable residue. More importantly, cyclic voltammetry of the proteins demonstrated that lowering the pH results in a dramatically increased reduction potential of the CuA center (from 160 mV at pH 7 to 340 mV at pH 4) for wild-type CuA-Az [205].

A close inspection of the structures of CuA-Az and native CcO indicates that His120 is equivalent to His204 in bovine CcO and His260 in R. sphaeroides CcO [157, 158], both of which are close to the interface between the membrane and the intermembrane space and can be subjected to protonation. Since CcO is the terminal oxidase, and the driving force for the ET reactions in CcO is quite small (approximately 50 mV), the above-mentioned finding of the protonation and the resulting dramatic increase of the reduction potential suggests that the equivalent of His120 in CcOs may act as a gated residue [205]. At low protein concentration, the reduction potential is low and His120 (or its equivalent) is in the normal deprotonated state, and ET from CuA to heme a proceeds and CcO pumps protons across the mitochondrial membrane (Fig. 12, left). However, as protons accumulate within the mitochondrial membrane, the histidine makes a transition to a protonated form which can raised reduction potential of CuA. This results in lowered ET since the driving force becomes significantly diminished, with concomitant diminished proton pumping (Fig. 12, right). Overall the hypothesis puts forward a mechanism for CuA to gate ET as suggested by studies in biosynthetic CuA-Az. pH-dependent changes on a soluble truncate of a native CuA have been observed previously [35]. Previous work has indeed shown that ET is controlled by proton uptake [206]. His204 is involved in extended hydrogen-bonding to residues along the ET pathways of bovine CcO, whereas mutagenesis of His260 in R. sphaeroides CcO to asparagine (His260Asn) results in a raised reduction potential and altered ET kinetics compared with wild-type protein [157, 158]. Therefore, this hypothesis put forward from the study of a biosynthetic model in CuA-Az is consistent with observations made in native CcO and provides a strong basis for possible gated proton-coupled ET in CcO.
Fig. 12

Proposed proton-gated pumping mechanism by CuA in CcO [205]. Left low protein concentration, the CuA histidine ligand is deprotonated and CuA has a relatively higher reduction potential such that \( E_{{\text{Cu}}_{\text{A}}}^{ \circ } < E_{\text{heme}_{a}}^{ \circ }\) and electrons transfer from CuA to heme a to heme a3 with concomitant proton pumping across the mitochondrial membrane. Right At high proton concentration, the CuA histidine ligand is protonated and CuA has a relatively lower reduction potential such that \( E_{\text{CuA}}^{ \circ } > E_{{{\text{heme}}{\kern 1pt} {\text{a}}}}^{ \circ } \). Electron flow from CuA to heme a is halted along with proton pumping across the mitochondrial membrane. Cyt c cytochrome c

Evolution of CuA

Early phylogenetic analysis of cupredoxin proteins suggested a common ancestry between blue T1 copper proteins and purple CuA proteins despite less than 10% sequence homology [40, 207, 208]. Nevertheless, sequence alignment and consensus sequences in these highly divergent proteins point to a common ancestry. Recent pH-dependence studies on the reconstitution of apo-CuA from native P. denitrificans N2OR with exogenous Cu2+ demonstrated the formation of three types of copper centers within the same, native ligand loop of CuA [209]. The results at high pH (approximately 8), based on UV–vis and EPR spectroscopies, demonstrated the initial formation of blue T1 copper [17, 20, 22] and red T2 copper [3638] sites before slow conversion to the native purple CuA site. At lower pH (pH ~ 6), the formation of T1 copper and T2 copper intermediates was less, and the conversion to purple CuA was significantly faster. This was the first report on the observation of all three types of copper, T1 copper, T2 copper, and CuA, within the same ligand loop, and was the first experimental evidence for an ancestral link between the various copper sites, as also suggested by phylogenetic analysis [40, 207, 208]. The pH-dependent incorporation of copper also has physiological implications for chaperones. Chaperones are responsible for shuttling metals to the apoprotein, but the results reported herein clearly underline their role in preventing misincorporation of metal sites, particularly at nonideal pH conditions which may otherwise result in inhibited protein forms [210].

Why does Nature employ CuA sites?

Earlier in the review a couple of questions were asked: Why did Nature employ CuA sites? What properties of the CuA site impart advantages that make it suitable for its role in the terminal position of the electron transport chains? The combined results from the last few sections can offer some answer to these questions. The proposal that CuA evolved from blue copper sites (see “Evolution of CuA”) suggests that there must have been an evolutionary advantage in terms of protein function for Nature to evolve the CuA site. The observation of CuA in only the terminal position to date implies an advantage to ET function in that terminal position.

The results from “Role of the methionine ligand” demonstrated that CuA could be described as having a rigid diamond-core structure with a reduction potential relatively resistant to mutations of the axial methionine at a distance similar to that in blue T1 copper proteins. The ET studies described in “ET properties and reorganization energy” have shown that the reorganization energy of CuA is significantly smaller than the reorganization energy of blue copper sites, and that they are more efficient ET sites despite the driving force being smaller (a smaller potential difference).

In an electron transport chain, there will be a certain potential difference between the initial electron entry site and the final electron acceptor. As electrons move down the transport chain, the remaining driving force diminishes and the potential of subsequent ET sites needs to fall between a narrower range of values. Therefore, an ET site that functions efficiently with a lower reorganization energy than that of the well-known blue copper, even with a small driving force, would be optimally suited in the terminal position, where the driving force may be small at that point. Moreover, if the site was relatively resistant to mutations in its secondary coordination sphere, then its reduction potential could not be changed sufficiently to change the ET direction to the final catalytic site. The CuA site has both of these advantages over blue copper sites, which may be the reason Nature evolved them in the terminal electron acceptor position.


CuA is a fascinating metal-binding site that has captured the interest of many inorganic chemists, biochemists, and biophysicists. In a complementary approach to studying CuA in native enzymes, soluble truncates, and synthetic models, biosynthetic modelling approach combine the advantages of simple, homogeneous, and interfering-chromophore-free synthetic models with the natural environment of native proteins, particularly the presence of noncovalent secondary coordination spheres. Furthermore, the bottom-up approach of biosynthetic models makes it possible to reveal hidden structural features that make them the only structural and functional models of native CuA, with almost identical structure and function, including the secondary coordination sphere environment. More importantly, the biosynthetic models make it possible to offer deeper insights such as the mechanism of copper incorporation, evolution of cupredoxins including blue T1 copper and CuA, metal-binding specificity, roles of cysteine, methionine, and histidine ligands in spin coupling and redox potential tuning, and proton-coupled ET. Although the above-mentioned insights can be obtained from other CuA systems, the placement of two ET centers, T1 Cu and CuA within the same protein framework, demonstrates that biosynthetic models can provide unique insights that are otherwise difficult if not impossible to obtain in other cases, such as that the axial methionine in CuA-Az exerts much less influence on the copper reduction potentials than that in T1 copper at a comparable distance and that CuA is a more efficient ET center than T1 copper owing to its lower reorganization energy. Therefore, CuA-Az is not just a good structural model of CuA, but is also a valuable resource for gaining functional insight into CuA. This biosynthetic approach can be applied to modeling more complex metalloproteins.


We wish to thank the US National Science Foundation and its Special Creativity Extension Award (award no. CHE 05-52008) for continued financial support. We would also like to thank other Lu group members for their intellectual contributions to work described in this review, Kyle D. Miner for help in modifying the figures, and Dewain K. Garner and Tiffany D. Wilson for their help in reference compilation and proof-reading.

Copyright information

© SBIC 2010