Geometric and electronic structure differences between the type 3 copper sites of the multicopper oxidases and hemocyanin/tyrosinase

Abstract

The coupled binuclear “type 3” Cu sites are found in hemocyanin (Hc), tyrosinase (Tyr), and the multicopper oxidases (MCOs), such as laccase (Lc), and play vital roles in O₂ respiration. Although all type 3 Cu sites share the same ground state features, those of Hc/Tyr have very different ligand-binding properties relative to those of the MCOs. In particular, the type 3 Cu site in the MCOs (Lc^T3) is a part of the trinuclear Cu cluster, and if the third (i.e., type 2) Cu is removed, the Lc^T3 site does not react with O₂. Density functional theory calculations indicate that O₂ binding in Hc is ≈9 kcal mol⁻¹ more favorable than for Lc^T3. The difference is mostly found in the total energy difference of the deoxy states (≈7 kcal mol⁻¹), where the stabilization of deoxy Lc^T3 derives from its long equilibrium Cu–Cu distance of ≈5.5–6.5 Å, relative to ≈4.2 Å in deoxy Hc/Tyr. The O₂ binding in Hc is driven by the electrostatic destabilization of the deoxy Hc site, in which the two Cu(I) centers are kept close together by the protein for facile 2-electron reduction of O₂. Alternatively, the lack of O₂ reactivity in Lc^T3 reflects the flexibility of the active site, capable of minimizing the electrostatic repulsion of the 2 Cu(I)s. Thus, the O₂ reactivity of the MCOs is intrinsic to the trinuclear Cu cluster, leading to different O₂ intermediates as required by its function of irreversible reduction of O₂ to H₂O.

A number of important biological systems contain coupled binuclear copper active sites that play important roles in O₂ binding, activation, and reduction to H₂O. The term “coupled” refers to the exchange interaction between two Cu(II) (S = 1/2) centers caused by bridging ligation, leading to antiferromagnetic coupled, diamagnetic, and thus, electron paramagnetic resonance (EPR) silent ground states. They have been historically categorized as “type 3” Cu sites in biology (1) and consistently referred to as one class of related sites. The simplest of the metalloproteins containing type 3 sites are the binuclear Cu proteins, hemocyanins (Hc), catechol oxidase (CatO), and tyrosinase (Tyr) that reversibly bind O₂; CatO and Tyr further activate O₂ for substrate hydroxylation or oxidation (2, 3). Hc/CatO/Tyr have been shown to have structurally equivalent active sites. The type 3 sites are also found in the multicopper oxidases (MCOs), which include tree and fungal laccase (Lc), ceruloplasmin, Fet3p (the MCO found in yeast with ferroxidase activity), and ascorbate oxidase. MCOs use a minimum of four Cu centers: a “blue,” type 1 Cu site and a trinuclear Cu cluster composed of a “normal,” type 2 Cu and a binuclear type 3 Cu site that together catalyze the 4-electron reduction of O₂ to water with concomitant oxidation of substrates (2, 3). Crystal structures indicate that both the type 3 Cu active sites in Hc/CatO/Tyr (4–10) and the MCOs (11–25) are similarly held in the protein by three His ligands on each Cu center. No additional ligands are present in the deoxy forms, whereas oxygen-derived ligands bridge and exchange couple the two Cu(II)s in the oxidized forms.

O₂ binding to the type 3 Cu sites of deoxy Hc/Tyr generates a side-on μ-η²:η² peroxo-bridged Cu₂O₂ oxy structure with Cu–Cu distance of ≈3.6 Å (Fig. 1A) (7, 8, 10). The side-on geometry promotes a large overlap between the O₂²⁻ π* and the two Cu d_x²−y² orbitals and provides a direct pathway for 2-electron reduction of O₂ to peroxide (26). Detailed spectroscopic and theoretical studies have shown that the side-on bridged geometry in oxy Hc/Tyr leads to a large energy splitting between the lowest-unoccupied molecular orbital (LUMO) and the highest-occupied molecular orbital (HOMO), which results in the strong antiferromagnetic coupling between the two Cu(II) centers (2, 3). Evaluation of the O₂-binding reaction coordinate by using density functional theory (DFT) calculations has shown that the spin-forbidden transition in the formation of the antiferromagnetically coupled (i.e., singlet) oxy Hc from triplet O₂ involves a simultaneous 2-electron transfer, where the initial ferromagnetically coupled coordination modes change along the O₂-binding coordinate with increasing metal–ligand overlap to turn on antiferromagnetic coupling to generate the final planar singlet Cu₂O₂ structure (Fig. 1A).

Fig. 1.

O₂ binding in the type 3 Cu sites of Hc/Tyr and MCOs.

The type 3 Cu sites in the MCOs are part of the trinuclear Cu cluster where the 4-electron reduction of O₂ to water occurs. Reaction of the fully reduced form of MCO with O₂ proceeds via two sequential 2-electron steps generating, first, the peroxy intermediate and then, the native intermediate. The spectral features of the peroxy intermediate are very different from those of oxy Hc/Tyr, indicating that the peroxy intermediate acquires a very different geometry (27). In our recent DFT study, we have generated a spectroscopically relevant structure, where the peroxide is bridged internally in a μ₃-1,1,2 geometry with the type 2 and one of the type 3 coppers oxidized (Fig. 1B) (28). Importantly, this structure allows for the irreversible binding of O₂ at the trinuclear Cu site that leads to efficient OO bond cleavage in the subsequent 2-electron reduction step to generate the native intermediate.

The different roles of type 3 Cu sites in Hc/Tyr and MCOs are closely related to their distinctive ligand-binding properties. In particular, by using a derivative of tree Lc where the type 2 Cu is reversibly removed [i.e., type 2-depleted (T2D) Lc], we have demonstrated that the reduced type 3 Cu center alone cannot bind O₂ in the MCOs, (29, 30). However, no detailed account of the origin of this O₂ reactivity difference between the two type 3 Cu sites has been available. In this work, DFT calculations are systematically performed to compare the two classes of type 3 Cu sites in biology. Focus is on evaluating the different O₂-binding properties of Hc/Tyr and the type 3 site in the MCOs and elucidating their origins. This work provides molecular insight into the critical role of the protein environment in directing the O₂ reactivity of these catalytic active sites.

Results and Analysis

Reduced Structures.

As a starting point for the evaluation of O₂ binding to the type 3 sites in Hc/Tyr and MCOs, optimized structures of Hc and Lc [the type 3 Cu center in the absence of the type 2 Cu (Lc^T3)] with reduced (i.e., deoxy) binuclear Cu centers were obtained (Fig. 2 A and B). Copper atoms are labeled as Cu_A and Cu_B for Hc and Cu_α and Cu_β for Lc^T3 for clarity in further descriptions of these sites. Both type 3 sites were modeled with three His ligands on each Cu center; Cu centers were coordinated to εN atoms except for one His ligand of the Cu_α center in Lc^T3, which was coordinated to the δN atom as found in all available crystal structures of MCOs. Each model was geometry optimized with frozen αC positions to reflect the different protein structures at the type 3 sites in Hc and Lc^T3.

Fig. 2.

Optimized structures of deoxy and oxy forms of Hc and Lc^T3. (A) Deoxy Hc. (B) Deoxy Lc^T3. (C) Oxy Hc (M_S = 0). (D) Oxy Lc^T3 (S = 1). (E) Oxy Lc^T3 (M_S = 0). Relevant bond lengths are shown in Å.

The geometry-optimized structures of deoxy Hc and deoxy Lc^T3 are shown in Fig. 2 A and B. In general, the optimized Cu centers have trigonal planar geometry, typical for d¹⁰-ML₃ units (31). However, the Cu atoms in the Cu_A and Cu_B centers in Hc and Cu_β center in Lc are ≈0.2–0.3 Å out of the plane formed by the N₃ ligands, whereas the Cu atom in the Cu_α site in Lc remains in the plane. The planar geometry of the Cu_α site is largely caused by the δN coordination of one of its His ligands because the αC constraint of the δN-His ligand is not along the CuN bond and therefore, does not contribute in orienting the Cu_α atom out of the N₃ plane. The pyramidal distortion in Hc and the in-plane geometry in the Cu_α site of Lc is consistent with Cu K-edge X-ray absorption spectroscopy (XAS) data (29, 30, 32, 33) (see Discussion).

The most conspicuous difference between the two deoxy structures is found in the equilibrium Cu–Cu distances, in which deoxy Hc is 4.23 Å and deoxy Lc^T3 is 6.48 Å (Fig. 2 A and B). For deoxy Hc, the calculated R(Cu–Cu) is in good agreement with those of known crystallographic data of Hc and Tyr, which range from 3.5 to approximately 4.6 Å (5–7, 10). For Lc^T3, the calculated R(Cu–Cu) is ≈1.2 to ≈1.5 Å larger than those of the known crystal structures of reduced ascorbate oxidase (13), CotA (the MCO found in bacteria Bacillus subtilis) (11) and Fet3p (24), which range from 5.0 to ≈5.3 Å. The deviation in the calculated Cu–Cu distance is reduced when two hydrogen bonds involving His ligands and backbone carbonyl groups^† are included in the calculation [supporting information (SI) Fig. S1]. The resulting Cu–Cu distance is 5.80 Å, which is in reasonable agreement with those of the crystal structures. However, the additional hydrogen bonds have virtually no effect on the overall Cu geometries in deoxy Lc^T3 aside from the Cu–Cu distance. Moreover, the energy difference between deoxy Lc^T3 at R(Cu–Cu) of 6.48 Å and 5.4 Å is only ≈2 kcal mol⁻¹ (≈1.4 kcal mol⁻¹ if a dielectric with ε = 4.0 is included using the polarizable continuum model) caused by the shallow potential energy surface (PES) along R(Cu–Cu), implicating the flexible nature of the Lc^T3 site. Thus, despite the long-calculated Cu–Cu distance, the deoxy Lc^T3 structure shown in Fig. 2B is a reasonable representation of the type 3 site in the MCOs, in particular, in providing energetic descriptions of O₂ binding.

Reaction Coordinate for O₂ Binding.

As a second step in evaluating O₂ binding in the type 3 sites of Hc and Lc^T3, an O₂ molecule was introduced to the deoxy Hc and deoxy Lc^T3 structures. Geometry optimizations of deoxy Hc+O₂ and deoxy Lc^T3+O₂ complexes were performed with frozen αC positions, in both the spin-unrestricted triplet (S = 1) and broken-symmetry (M_S = 0) states; the energies of the pure-singlet (S = 0) states were then obtained from the spin-projection method (34–36) by using the broken-symmetry state energies (see Computational Methods). Calculations were performed both with unconstrained Cu–Cu distances and with these distances systemically varied.

The O₂-binding mechanism in Hc has been studied in detail, and the structure of Hc+O₂ (i.e., oxy Hc) is well established (26). Therefore, the optimized oxy Hc structure, in the broken-symmetry (M_S = 0) state, is first obtained to validate our method and model systems (Fig. 2C). Our calculations indicate that O₂ binding in Hc is exothermic by −2.9 kcal mol⁻¹, which is in reasonable agreement with experiment (ΔH = −11.5 to −6.0 kcal mol⁻¹ and ΔG = −5.0 kcal mol⁻¹ at 25 °C) (37–41). Key structural features are consistent with experimental data, yielding the side-on μ-η²:η² O₂²⁻ core structure with R(Cu–Cu) of 3.55 Å (experimental ≈3.6 Å). The Cu₂O₂ structure is slightly butterflied, and the μ-η²:η² O₂²⁻ bridge provides a strong superexchange pathway for a large antiferromagnetic coupling of the Cu(II) centers with a calculated −2J (Ĥ = −2J Ŝ₁·Ŝ₂) of ≈1,280 cm⁻¹, also consistent with experiment (>600 cm⁻¹). A reaction coordinate was further generated by varying the Cu–Cu distance from the starting R(Cu–Cu) of 4.2 Å (i.e., that of reduced Hc) (Fig. 3Lower Left) and optimizing the rest of the structure with frozen αC positions (note that the S = 0 energies shown in Fig. 3 are obtained from spin-projection method by using the broken-symmetry state energies and geometries). As a result, a spin-state transition from S = 1 to S = 0 occurs with decreasing Cu–Cu distance (at ≈3.9 Å). Associated with this spin-state transition is a structural change from a μ-η¹:η² peroxy bridge at large distances (Fig. S2) to the final μ-η²:η² Cu₂O₂ structure. These results are consistent with our previous study on O₂ binding to deoxy Hc, where a reaction coordinate was generated by systematically varying the distance of the peroxide above the molecular plane (26). It was shown that the charges on both copper centers change at the same rate even in the asymmetric bridged structure, indicating that O₂ binding involves the simultaneous transfer of 2 electrons. Likewise, it is found here that 2 electrons from both Cu centers have already been transferred to O₂ to form peroxide at an R(Cu–Cu) of ≈4.0 Å, as indicated by spin-density distribution among Cu and O atoms in the broken-symmetry wave functions: At R(Cu–Cu) = 4.0 Å, spin densities on Cu_A, Cu_B, O₁, O₂ atoms are +0.46, −0.46, +0.09, and −0.11, whereas at the equilibrium R(Cu–Cu) = 3.55 Å, these are +0.48, −0.48, −0.03, and +0.04, respectively. Thus, the Cu–Cu distance in deoxy Hc is prepositioned for facile 2-electron transfer upon O₂ binding.

Fig. 3.

Reaction coordinates for O₂ binding in Hc and Lc^T3 along Cu–Cu. The O₂-binding energies were obtained by subtracting the sum of the energies of triplet O₂ molecule and the most stable deoxy forms (i.e., deoxy forms in Fig. 2) from that of the oxy forms at each R(Cu–Cu). The S = 0 energies are obtained from the spin-projection method by using the broken-symmetry state energies and geometries.

In contrast to oxy Hc, the most stable structure for oxy Lc^T3 is found in the triplet (S = 1) state. As shown in Fig. 2D, the triplet oxy Lc^T3 structure has R(Cu–Cu) of 6.30 Å. Consequently, O₂ is unable to bridge the two Cu centers but is only very weakly bound to the Cu_β center with an end-on geometry and a low O₂-binding energy of −0.7 kcal mol⁻¹ (note that we were not able to generate a structure with O₂ bound to the Cu_α atom). Bond lengths, R(Cu_βO) = 2.96 Å and R(OO) = 1.23 Å, are also indicative of a very weak Cu_βO₂ bond that lacks any electron transfer between Cu and O₂ (i.e., no superoxide or peroxide formation). At shorter Cu–Cu distances, the side-on μ-η²:η² Cu₂O₂ structure in the broken-symmetry spin state is also obtained (Fig. 2E). The singlet oxy Lc^T3 structure has R(Cu–Cu) of 3.66 Å and R(O–O) of 1.42 Å, indicative of a 2-electron reduced peroxo-species. However, the singlet oxy Lc^T3 is found to have O₂-binding energy of +5.6 kcal mol⁻¹ (6.3 kcal mol⁻¹ higher than the triplet oxy Lc^T3), making its formation thermodynamically unfavorable.

We have also generated a reaction coordinate for the formation of singlet oxy Lc^T3 by varying R(Cu–Cu), in both the triplet (S = 1) and singlet (S = 0) states (Fig. 3). The PES for the triplet state is uphill from R(Cu–Cu) of 6.30 Å with decreasing Cu–Cu distance, whereas that for the singlet state is downhill until the equilibrium R(Cu–Cu) of 3.66 Å is reached. At R(Cu–Cu) > 4.5 Å, the broken-symmetry optimized structures (representing singlet states) are characterized by the side-on η² superoxo structures formed between O₂ and the oxidized Cu_β center via internal 1-electron transfer from Cu_β to O₂ (see Fig. S3), in contrast to the end-on structure of the triplet state (Fig. 2D). When R(Cu–Cu) decreases below ≈4.5 Å in the broken-symmetry state, a bridged μ-η¹-η² geometry is initially formed, then the final μ-η²-η² peroxo structure, with transfer of the second electron from the Cu_α center to the bridging O₂ moiety (spin density at the Cu_α atom changes from 0.0 to −0.33 to −0.45 with change in R(Cu–Cu) from 6.13 Å to 4.20 Å to 3.66 Å in the broken-symmetry state). Notably, the spin state transition from S = 1 to S = 0 occurs at R(Cu–Cu) ≈4.0 Å, where an energy barrier is found to be ≈9 kcal mol⁻¹ relative to that of the triplet oxy Lc^T3. The presence of this energy barrier makes the O₂ binding to deoxy Lc^T3 kinetically difficult, in contrast to Hc where the O₂ binding to the deoxy Hc is uniformly downhill (Fig. 3 Lower Left). Thus, our calculations indicate the O₂ binding in Lc^T3 is inhibited thermodynamically, as well as kinetically, and therefore the long Cu-Cu distance in deoxy Lc^T3 creates a large energy barrier for the singlet oxy Lc^T3 formation.

At this point, we evaluate the origin of different O₂ reactivities of Hc and Lc^T3. As mentioned above, the O₂-binding energies of Hc and Lc^T3 are different by ≈8.5 kcal mol⁻¹ (−2.9 kcal mol⁻¹ for Hc and +5.6 kcal mol⁻¹ for Lc). However, the total energy of the oxy Lc^T3 structure itself is only ≈1.6 kcal mol⁻¹ higher than that of the oxy Hc structure, which does not account for the difference in O₂-binding energies. Rather, the remaining 6.9 kcal mol⁻¹ difference is found in the energy difference of the deoxy structures of Hc and Lc^T3. As shown below, the relative destabilization of the deoxy Hc structure leads to the thermodynamically favorable O₂ binding in Hc, whereas the relative stabilization in the deoxy Lc^T3 structure results in unfavorable O₂ binding in Lc^T3.

Potential Energy Surfaces of Deoxy Sites.

To gain insight into the origin of the difference in deoxy Hc and deoxy Lc^T3, PESs along Cu–Cu were generated, where geometry optimizations of the deoxy Hc and deoxy Lc^T3 were performed with frozen αC positions (Fig. 4A, filled symbols), as well as without the αC constraints (Fig. 4A, open symbols, Hc* and Lc^T3*). As described above, the deoxy Lc^T3 (with αC constraints) energy minimum is 6.9 kcal mol⁻¹ lower in energy than that for deoxy Hc and has an equilibrium Cu–Cu distance of 6.48 Å, in contrast to the 4.23 Å optimized distance of deoxy Hc. Alternatively, unconstrained PESs for deoxy Hc and deoxy Lc^T3 (Hc* and Lc^T3* in Fig. 4A) are both effectively linear, and the slopes are very similar (approximately −2.5 kcal mol⁻¹Å⁻¹), indicating that the Cu(I)–Cu(I) electrostatic repulsion of the two deoxy forms are the same (note that the difference between the two unconstrained PESs are from the small additional contribution caused by a decrease in steric interactions of the three His ligands between Cu centers, which are eclipsed in Lc and staggered in Hc). Notably, the energy difference of the unconstrained deoxy Lc^T3 at R(Cu–Cu) = 6.5 Å and 4.2 Å is ≈5.8 kcal mol⁻¹ (i.e., the electrostatic energy difference), which accounts for most of the 6.9 kcal mol⁻¹ energy difference between deoxy Hc and deoxy Lc^T3. Thus, the decrease in energy of the deoxy Lc^T3 site reflects the decrease in electrostatic repulsion of the two Cu(I) centers in the low dielectric of the protein. The lack of O₂ reactivity of the type 3 site in Lc^T3 reflects its electrostatic and structural stabilization at its long-calculated equilibrium Cu–Cu distance of 6.48 Å.

Fig. 4.

PESs of deoxy Hc and deoxy Lc^T3 along Cu–Cu. (A) PESs of deoxy Hc and deoxy Lc^T3 with αC constraints are shown as Hc and Lc^T3 in filled symbols, whereas PESs of deoxy Hc and deoxy Lc^T3 without αC constraints are shown as Hc* and Lc^T3* in open symbols. (B) PESs of deoxy Hc and deoxy Lc^T3, where the energies of Hc* and Lc^T3* in A are subtracted from those of Hc and Lc^T3 in A. Symbols are the results of the subtraction, whereas the solid lines are the fit curves using a second-order parabolic function. For comparison of the two PES components, the minima are set as zero in relative energies.

In Fig. 4B, the components of the PESs in Fig. 4A where the electrostatic contributions (i.e., Hc* and Lc^T3* in Fig. 4A) are removed are given. The resultant PESs provide important insight into the elasticity of the two type 3 sites imposed by the protein constraints (through the αC atoms in our models) in the absence of electrostatic contributions. The PES components (Fig. 4B, filled symbols) were fitted to second-order parabolic functions (Fig. 4B, solid lines) to obtain spring force constants, k. It is found that k for deoxy Lc^T3 is 2.3 kcal mol⁻¹Å⁻², whereas that for deoxy Hc is 14.1 kcal mol⁻¹Å⁻², which is ≈7 times larger than that of deoxy Lc^T3. Notably, in Fig. 4B, the minimum in the deoxy Lc^T3 PES component is ≈1 Å shorter than that of the original PES in Fig. 4A (from 6.48 Å to 5.51 Å), reflecting the low value of k and thus, the flexible nature of the Cu geometry in deoxy Lc^T3. In contrast, the minimum in the deoxy Hc PES is shifted by only 0.08 Å (from 4.23 Å to 4.15 Å). Thus, the large k in deoxy Hc demonstrates that the Cu–Cu distance is kept short by the protein environment, whereas it is not in deoxy Lc^T3.

Discussion

Generally, the binuclear Cu sites in Hc/Tyr and MCOs (i.e., Lc^T3) have been classified into the same type 3 group because both have antiferromagnetically coupled diamagnetic ground states and thus, no EPR feature. Despite the same active-site components of [Cu(His)₃]₂, however, the two type 3 sites have very different ligand-binding properties. In particular, the reduced type 3 Cu site in T2D Lc is stable to O₂ binding, in strong contrast to Hc/Tyr, which reversibly bind O₂ to form oxy Hc/Tyr (29, 30). Moreover, exogenous ligands (carbon monoxide and azide) were shown to bind to the type 3 site of T2D Lc, demonstrating that small molecules are not blocked from having access to this active site. However, these ligands bind only to one Cu center in T2D Lc, further demonstrating the different ligand-binding properties compared with Hc/Tyr, in which exogenous ligands bridge the two-Cu center (42).

We have calculated that O₂ binding to Hc is exothermic by 2.9 kcal mol⁻¹. This is in contrast to O₂ binding to the deoxy Lc^T3 in the formation of singlet oxy Lc^T3, which is found to be endothermic by 5.6 kcal mol⁻¹ (Fig. 3), consistent with the fact that T2D Lc does not bind O₂.^‡ The origin of this 8.5 kcal mol⁻¹ destabilization of O₂ binding in Lc^T3 relative to Hc reflects the relative stabilization of the deoxy Lc^T3 structure in the protein environment of the MCOs. The 6.9 kcal mol⁻¹ stabilization of deoxy Lc^T3 relative to deoxy Hc derives from the decrease in electrostatic repulsion of the two Cu(I)s at the long Cu–Cu distance in deoxy Lc^T3 in the low dielectric of the protein. This increase in Cu–Cu distance is allowed by the combined effects of the protein constraints and the flexibility in the active-site environment.

From Fig. 5, the large structural differences between the coupled binuclear site in Hc and the type 3 site in MCOs relate to the very different structural constraints in the two protein environments. The binuclear Cu(I) site of deoxy Hc is kept at its 4.2 Å Cu–Cu distance because of the constraints of the protein. The most conspicuous constraint is associated with two His ligands, one on each Cu, for which the αC–αC distance is only ≈5 Å compared with >9 Å for other αC–αC distances between His ligands on the two Cu centers. The origin of the close αC–αC distance is attributed to an Arg(Lys)–Asp salt bridge between two helices that is conserved in Hc/CatO/Tyr (see sequence alignment in Fig. S4). Alternatively, in the Lc^T3 Cu site, the two Cu(I)s are kept at an electrostatically stable distance of ≈6.5 Å (calculated) by two sets of two His ligands (αC–αC distances of the two pairs of His ligands are ≈7 Å), where each set is from an H–X–H motif that is on a loop extending from a β-sheet, leaving the Cu(I)s relatively unconstrained. Importantly, this shows that the different ligand architecture in Hc and Lc^T3 account for the large difference in equilibrium Cu–Cu distances in deoxy Hc and deoxy Lc^T3. This is a critical factor in the O₂-binding properties of Hc and Lc^T3, where the close Cu–Cu distance in deoxy Hc leads to simultaneous 2-electron transfer from both Cus to O₂, whereas the long Cu–Cu distance in deoxy Lc^T3 leads to stabilization of this state and energetically unfavorable O₂ binding.

Fig. 5.

Features of Cu active sites Hc (A) and Lc^T3 (B) that affect the equilibrium Cu–Cu distances. Structures are adapted from crystal structure database 1JS8 (oxy Hc from O. dofleini Hc) (8) and 1GYC (resting oxidized fungal Lc from T. versicolor) (19).

In addition, the protein constraints further contribute to the very different PESs for deoxy Hc and Lc^T3 along Cu–Cu (Fig. 4B), where the binuclear force constant k in deoxy Lc^T3 is ≈7 times smaller than that of deoxy Hc. This large force constant difference results in the shift in the equilibrium Cu–Cu distance by ≈0.1 Å in deoxy Hc and ≈1 Å in deoxy Lc^T3 because of the linear distorting force from the electrostatic repulsion (Fig. 4). Note that the ≈1 Å shift in deoxy Lc^T3 results in an energy stabilization of ≈1.4 kcal mol⁻¹, whereas the energy of deoxy Hc is lowered by only ≈0.2 kcal mol⁻¹ by the ≈0.1 Å shift.^§ Thus, the difference in force constants between deoxy Lc^T3 and deoxy Hc contributes ≈20% of the total energy difference (i.e., ≈7 kcal mol⁻¹) between these sites. Importantly, the large force constant in deoxy Hc indicates that the active site is tightly controlled to keep the Cu–Cu distance short for facile reversible O₂ binding. Alternatively in deoxy Lc^T3, the low force constant allows a large change in Cu–Cu distance with little change in energy, which is required in the reaction coordinate for OO bond cleavage at the trinuclear Cu cluster in the MCOs (43).

Insight into the origin of the difference in force constants in deoxy Hc and deoxy Lc^T3 can be gained from Cu K-edge XAS data. As shown in Fig. 6, the preedge features of deoxy Hc and deoxy T2D Lc (i.e., deoxy Lc^T3) are very different (29, 30, 32, 33). In deoxy T2D Lc, the sharp intense peak at ≈8,983 eV is the Cu 1s → Cu 4p_z transition in a trigonal planar geometry, with the Cu 1s → Cu 4p_x,y transitions shifted up in energy with lower intensities because of the equatorial ligands. In deoxy Hc, the Cu 1s → Cu 4p_z transition has lower intensity whereas the Cu 1s → Cu 4p_x,y transitions shift to lower energy and higher intensity, reflecting the pyramidal distortion of the Cu centers out of the ligand N₃ plane. These Cu K-edge XAS data are consistent with the calculated structures of deoxy Hc and deoxy Lc^T3 (Fig. 2 A and B); in deoxy Hc, both Cu centers show a pyramidal distortion (Cu centers ≈0.2–0.3 Å out of the N₃ plane) whereas in deoxy Lc^T3, only one Cu is distorted, and the other, Cu_α in Fig. 2B, retains a trigonal planar geometry, where the planarity derives largely from the δN ligation of one of its His ligands. These different Cu geometries are closely related to the different force constants; in the trigonal planar geometry of the Cu_α center in deoxy Lc^T3, the vibrational mode normal to the N₃ plane involves only bends, whereas for the trigonal pyramidal geometries of the Cu centers in deoxy Hc (and Cu_β in deoxy Lc^T3), both bends and bond stretches are involved. Thus, the larger force constant along Cu–Cu in deoxy Hc largely reflects the pyramidal distortion of the Cu center that is imposed by the protein environment, whereas the low force constant in deoxy Lc^T3 derives from its relatively unconstrained environment, resulting in the trigonal planar structure typical of Cu(I) d¹⁰–ML₃ complexes.

Fig. 6.

Normalized Cu K-edge XAS spectra of deoxy Hc and T2D Lc. The two Hc spectra were obtained from Hcs from an arthropod Panulirus interruptus (spiny lobster) (32) and a mollusk Busycon canaliculatum (sea snail) (32), and the T2D Lc spectrum was obtained from Lc from Rhus vernicifera (lacquer tree) (29, 30). The features of the type 1 Cu (T1 Cu) was subtracted from the T2D Lc spectrum by using the XAS spectrum of the blue Cu protein plastocyanin.

In summary, deoxy Hc is electrostatically destabilized to react with O₂ to form oxy Hc, and this can be cooperatively regulated by changing the Cu(I)–Cu(I) distance (26) in the protein quaternary structure (i.e., tense and relaxed conformations) (44). Alternatively, the deoxy Lc^T3 Cu site in the multicopper oxidases is electrostatically stable and does not react with O₂ when the T2 Cu is not present. When the type 2 Cu is also present, the trinuclear Cu cluster does react with O₂ to form a 2-electron reduced peroxide, bridging between the type 2 and type 3 Cu centers (27, 28). The trinuclear Cu cluster further promotes cleavage of the OO bond by stabilizing the products of peroxide reduction (oxo and hydroxo moieties) through multiple bridging interactions requiring a large change in Cu–Cu distance.

Computational Methods

Spin-unrestricted DFT calculations were performed on Gaussian 03 (45). All geometry optimizations were performed with B3LYP hybrid functional (46) by using TZVP basis sets for Cu and coordinated N/O atoms and TZV basis sets for the rest (47, 48). The initial geometries of the type 3 sites of Hc and Lc were adapted from the crystal structures of the oxy form of Octopus dofleini Hc [Protein Data Bank (PDB) ID code 1JS8] (8) and the resting oxidized form of Trametes versicolor fungal Lc (PDB ID code 1GYC) (19), where the bridging O atoms were removed. Three His ligands on each Cu center were replaced with methyl-imidazolyl ligands, where Cu centers were coordinated to εN atoms, except for one imidazolyl ligand of the Cu_α center in Lc, which was coordinated to δN atom, as shown consistently in all available crystal structures of MCOs. All geometry optimizations were performed with frozen methyl C atoms (i.e., αC atoms; depicted by gray spheres in Fig. 2, except for the deoxy Hc and deoxy Lc^T3 structures in generating the Hc* and Lc* potential energy surfaces shown in Fig. 4A, in which all of the αC constraints were removed). For oxy Hc and oxy Lc^T3 structures, geometry optimizations were performed in both the triplet (S = 1) and the broken-symmetry (M_S = 0) states. The energies of the pure singlet (S = 0) states were then computed by using the spin-projection method (34–36):

where the triplet energy, E_{S = 1}, is computed for the single-determining high-spin configuration at the broken-symmetry geometry and 〈S_BS²〉 is the spin-expectation value of the broken-symmetry calculation. The O₂-binding energies were obtained by subtracting the sum of the energies of triplet O₂ molecule and the most stable deoxy forms from that of the oxy forms.