Beyond Blue: Systematic Modulation of Electronic Structure and Redox Properties of Type 1 Copper in Azurin

Abstract

The reduction potentials of metal ions (E°′), crucial for optimizing biological processes like electron transfer and catalysis, are finely tuned by interactions between the primary and secondary coordination spheres (PCS, SCS). While previous successes in tuning E°′ in azurin have provided deeper insights into how the SCS influences electronic structure and associated redox properties of “classic” blue copper proteins, our understanding of E°′ tuning in other subclasses of type 1 Cu (T1Cu) proteins, such as green and red copper proteins, remains rudimentary. To address this issue, we report the design of a green copper center in azurin where an equatorial-to-axial shift in a histidine binding interaction leads to reorientation of the Cu-centered redox active molecular orbital and a +100 mV shift in E°′.In contrast to a 22 mV decrease in E°′ when a hydrophobic interaction is introduced in wild-type azurin through the Met13Phe mutation, this same mutation leads to a 65 mV increase in our designed green Cu azurin. More importantly, using a combination of EPR spectroscopy, protein crystallography, and quantum mechanical calculations, we uncover correlations between E°′, d−s orbital mixing, and the angle between S_Cys−Cu and N_δH46−Cu bonds, ∠(S_Cys−Cu−N_δH46), allowing rationalization of increases in E°′ of green Cu proteins through an entropically driven T-shape distortion. By providing direct connections between geometry, electronic structure, and functional properties such as E°′, this work opens previously unexplored routes to systematically modulating E°′ through the combination of spatial reorientation of the redox active molecular orbital and varying geometric distortion in the primary coordination sphere.

Graphical Abstract

1. INTRODUCTION

The redox and catalytic properties of biological metal centers are intimately tuned not only through the primary coordination sphere (PCS), but through the extended surrounding environment in the secondary coordination sphere (SCS), including H-bonding, electrostatic, and hydrophobic/hydrophilic interactions.¹⁻⁶ These SCS interactions modulate metalloprotein redox and catalytic properties by influencing the electronic structure of metal centers, as imposed by the geometry and metal−ligand bonding strength of the PCS.⁷ Unraveling the interplay between the electronic structure of the metal center and its SCS environment is essential to understanding native metalloenzyme function, as well as for the rational design of finely tuned artificial metalloproteins as molecular redox mediators, and catalysts.

The type 1 Cu (T1Cu) protein azurin (Az) provides an excellent platform to investigate the interplay between the SCS and electronic structure.^2,3,8−10 Wild-type (WT) Az employs a stable cupredoxin fold to minimize the reorganization energy between Cu^II and Cu^I states and allow for efficient electron transfer.¹¹⁻¹⁴ In turn, the stability of the cupredoxin fold has made Az amenable to extensive mutagenesis in both the PCS and SCS surrounding the Cu binding site. Taking advantage of these properties, we and others have previously demonstrated how SCS interactions can modulate the T1Cu^II/I reduction potential (E°′) in Az.^2,3,10,15 These SCS effects can be grouped into two general categories, namely (1) those that modulate the hydrophobicity of the T1Cu site, and (2) interactions that modulate the metal/ligand character. Importantly, the distance and orientation of these SCS interactions relative to the Cu−S bond play a critical role in their ultimate influence on E°′,⁷ and their careful combination has allowed us to access a range of E°′ spanning nearly 1000 mV.³

While much success has been demonstrated in tuning E°′ through the SCS of Az, these effects alone cannot account for the much broader range of electronic and functional properties found among native T1Cu sites. Specifically, our current understanding has largely focused on one subclass of the T1Cu sites, namely the “classic” blue Cu center, which is characterized by its intense blue color due to a strong S_π ligand-to-metal charge transfer (LMCT) absorption at ~600 nm, as well as small hyperfine A_|| in electron paramagnetic resonce (EPR) spectra indicating a highly covalent Cu−S_Cys interaction.¹⁶⁻¹⁸ However, geometric changes in the Cu binding site, such as tetrahedral or tetragonal distortions and the presence of additional coordinating residues, can lead to increasing intensity of an S_σ LMCT ~400−450 nm (Figure S1). As a result, varying ratios of S_π and S_σ LMCTs and changes in Cu−S covalency arising from geometric changes leads to additional subclasses of T1Cu-like proteins, including green/purple T1.5Cu such as nitrite reductase, which exhibit both significant S_π and S_σ LMCTs and intermediate A_|| (~250−350 MHz), and red/yellow T2Cu, which display a dominant S_σ LMCT and A_|| > 400 MHz (such as nitrosocyanin).^9,19−28 Underlying these spectroscopic observations are changes in the electronic structure of the Cu site, which dictate their functional properties such as E°′ and electron transfer.

In contrast to the success in understanding structural features in the SCS that can tune the electronic structure and associated E°′ and electron transfer functions of the “classic” blue Cu Az, our understanding of other subclasses of T1Cu proteins is poor. While we can prepare mutants in the SCS of these other subclasses of T1Cu proteins, it is difficult to obtain a holistic understanding of how the influence of the SCS varies as a function of T1Cu electronic structure due to the myriad of differences in the identity of surrounding SCS residues and variations in the positioning and stability of PCS interactions with the Cu center between different proteins. Instead, redesign of the classic T1Cu in Az into another subclass of T1Cu proteins, such as the green Cu protein, provides a promising route to overcoming this challenge by minimizing the number of changes in protein environment, allowing for more systematic investigation. Based on the previously observed dependence of the electronic structure (S_π vs S_σ) on the positioning of Cu−N/O_σ interactions within the ψ(S_Cys−Cu−N_H46−N_H117) plane,^7,21,22,29 we postulated that these interactions could be leveraged to generate new spatial orientations of the Cu−S bonding interaction, providing a new context to explore the interplay of E°′, T1Cu geometry, and SCS, and—in the process—uncover new routes to tuning E°′.

Herein, we present the design and characterization of a novel green Cu Az protein, H117A/M121H-Az, a T1Cu protein that displays an effectively rotated Cu−S_p interaction, accomplished by removal of the in-plane His117 ligand together with the addition of an axial His121 ligand (Figure 1). Electrochemical characterization reveals a reduction potential of E°′ = 380 mV, nearly 100 mV higher than WT-Az despite maintaining a T1Cu-like PCS and lacking additional SCS mutations. Furthermore, while the Met13Phe mutation in the SCS decreases E°′ of classic T1Cu WT-Az by 22 mV,¹⁵ the same mutation increased the green Cu H117A/M121H-Az by 65 mV. Using EPR spectroscopy, we find Cu^II−H117A/M121H-Az exhibits a highly rhombic unpaired spin distribution, with an unprecedentedly strong hyperfine interaction at A_x. This strong interaction is rationalized through a reduction in centrosymmetry of the Cu ligation environment, allowing for d−s mixing which in turn produces a significant positive Fermi contact contribution to the hyperfine coupling. Excitingly, a comparison of Cu^II−H117A/M121H-Az and Cu^II-M13F/H117A/M121H-Az with previously reported T1Cu and T1Cu-like proteins, such as plantacyanin, amicyanin, and variants of Az, reveals positive correlations between E°′, the isotropic hyperfine A_iso, and the angle between S_Cys−Cu and N_δH46−Cu bonds, ∠(S_C112−Cu−N_δH46). These results are further supported through DFT and DLPNO−CCSD calculations, which capture the dependence of A_iso on unpaired Cu s-orbital spin density through an increase of ∠(S_Cys−Cu−N_δH46) in a T-shape distortion. Energetic calculations rationalize the observed increase in E°′, revealing that the T-shaped distortion leads to increased stabilization of the Cu^I state in an entropically driven process. Together, the demonstrated dependence of SCS interactions on Cu−S electronic structure and the relationship between PCS geometry and Cu^I stabilization provides design principles for redox potential tuning extending well beyond the classic T1Cu proteins.

Figure 1.

Schematic of the modifications to wild-type Azurin to generate H117A/M121H-Az, effectively rotating the equatorial coordinating His117 to the axial 121 position.

2. RESULTS AND DISCUSSION

2.1. Design and Spectroscopic Characterization of H117A/M121H-Az.

Previous studies have shown that replacing the axial Met121 in Az side chain with His results in an out-of-plane tetrahedral distortion of the otherwise near-planar T1Cu consistent with the spectral features observed for “green Cu” site, with ψ(S−Cu−N_δH46−N_δH117) moving from 170° in WT-Az to ~150° in M121H-Az at pH 3.5.^30,31 To further rotate the N_σ interaction in the T1Cu site with respect to the Cys112 side chain, we mutated His117 to Ala to generate the H117A/M121H-Az variant, which maintains the His₂Cys₁ ligand set of T1Cu but exchanges one of the two coordinating histidines from an equatorial to axial position. Titration of one equivalent of Cu^II into H117A/M121H-Az produces an absorption spectrum with intense peaks at 22,030 cm⁻¹ (454 nm) and 16,530 cm⁻¹ (605 nm) (Figure 2A).^{19,20,26,27,43,44} These features correspond to the LMCT transitions from the S_σ → SOMO and S_π → SOMO, respectively, with significant contributions of both transitions to the absorption spectrum indicating significant mixing of S_σ/ S_π character in the SOMO (singly occupied molecular orbital).¹⁷ The spectra remain unchanged between pH 5−8 (Figure S2).

Figure 2.

Comparison of (A) absorption and (B) CW X-band EPR spectra of Cu^II−WT−Az and Cu^II−H117A/M121H-Az. (C) CW Q-band EPR spectrum of Cu^II−H117A/M121H−Az prepared at pH 7 and measured at 30 K using 0.2 mW power and a 20 G modulation field. EPR fitting parameters are provided in Table 1

Continuous wave (CW) X-band EPR measurements of H117A/M121H-Az reveal an unusual, highly rhombic spectrum that is readily resolved using CW Q-band EPR (Figure 2B,C); pH-dependent X & Q-band spectra are provided in Figures S3 and S4. Whereas typical Cu^II spectra display significantly larger hyperfine at A_z (A_||) than A_y or A_x (A_⊥), deconvolution of these spectra reveals a single species with very small A_z and large splitting at A_x (Table 1), indicating a significant difference in the nature of the SOMO of H117A/M121H-Az. While H117A/M121H-Az presents an extreme case, we note that significant hyperfine at A_x has also been observed in several other green T1Cu proteins with rhombic EPR such as stellacyanin, plantacyanin, and M121H-Az at low pH (Table 1 and Figure S5).^{17,20,27,45−48}

Table 1.

Summary of Spin-Hamiltonian Parameters of Components From EPR Spectral Simulations and Comparisons to Other Select Blue and Green Cu Proteins, Together with Reported and Presently Determined Reduction Potentials vs NHE^a

proteinb	color	component	gx, gy, gz	|Ax, Ay, Az| (MHz)	Aiso (MHz)	E°′ (mV)
WT-Az7	blue		2.04, 2.06, 2.26	30, 30, 180	−40	278 ± 315
amicyanin32	blue		2.03, 2.03, 2.25	50, 50, 160	−20	29433
pseudoazurin (A. faecalis)34	blue		2.04, 2.05, 2.24	32, 18, 177	−42	27035
umecyanin36	blue		2.02, 2.07, 2.33	20, 20, 105	−30	28337
stellacyanin20	blue/green		2.03, 2.08, 2.30	150, 20, 120	17	26020
plantacyanin (spinach)38	green		2.03, 2.07, 2.21	185, 17, 165	12	34539
plantacyanin (cucumber)20	green		2.02, 2.06, 2.21	184, 25, 166	14	33340
nitrite reductase41	green	pH 7	2.04, 2.04, 2.21	30, 30, 243	−61	24042
M121H-Az (A. denitrificans)27	green	pH 3.5	2.04, 2.04, 2.30	230, 0, 80	50	35027
	red	pH 7	2.05, 2.05, 2.25	30, 30, 310	−83	20027
H117A/M121H-Az	green		2.01, 2.12, 2.24	315, 160, 70	135	380 ± 5
M13F/H117A/M121H-Az	green		2.01, 2.12, 2.26	325, 170, 20	158	445 ± 4

^aFor several systems (plantacyanin, nitrite reductase, stellacyanin, amicyanin, umecyanin) spectra were adapted from their original sources and resimulated (see Supporting Information). Calculations of A_iso are made assigning a negative sign to A_z, and positive signs for A_x and A_y.

^bCompared cupredoxin proteins were selected from the literature as representatives of blue and green Cu proteins with complete sets of spectroscopic (UV/vis absorption & EPR), crystallographic, and electrochemical (E°′) data from a single organism. M121H-Az was selected for comparison of the influence of the Met121His mutation without an equatorial His117Ala mutation.

2.2. Reduction Potentials & Correlation with A_iso.

The Cu^II/I reduction potential of H117A/M121H-Az was determined to be 380 ± 5 mV vs mined to be 380 ± 5 mV vs NHE via ferrocyanide redox titration (Figure 3A). Notably, this represents a ~ 100 mV increase relative to WT-Az (278 mV) and is comparable to plastocyanin and SCS variants of WT-Az in which the axial Met side chain is substituted with hydrophobic residues such as M121A-Az, M121L-Az, and M121V-Az.^7,10

Figure 3.

A) Redox titrations of H117A/M121HAz (blue) and M13F/H117A/M121H-Az (red), where A represents the absorption at 604 nm. Experimental data are shown as black dots, while fits are provided as dashed colored lines. (B) Correlation of A_iso and reduction potential, E°′, of the systems provided in Table 1. The provided linear line correlation results in a coefficient of determination (R²) of 0.86.

We further explored the influence of hydrophobicity in the SCS environment in the new context of H117A/M121H-Az Recently, we investigated the influence of Phe incorporation on the reduction potential in blue Cu Az, notably at the Met13 and Met44 positions.¹⁵ While the Met13Phe mutation produced ΔE°′ = −22 mV relative to WT-Az, adding this mutation to H117A/M121H-Az to generate M13F/H117A/ M121H-Az produces a significant increase in potential with ΔE°′ = +65 mV (E°′ = 445 ± 4 mV). Further EPR and absorption spectra of M13F/H117A/M121H-Az are provided in Figure S6.

Given the unusually large, positive A_iso of H117A/M121H-Az and M13F/H117A/M121H-Az together with observed increases in E°′ relative to WT-Az, we were interested in whether any broader correlation A_iso and reduction potential existed Combining our observations with those of the T1Cu proteins detailed in Table 1, we find a positive correlation between E°′ and A_iso (Figure 3B). Here, A_iso, also referred to as the Fermi contact term, is proportional to unpaired electron spin density at the Cu nucleus. Classically, the d-orbitals do not permit direct spin density at the nucleus; therefore, A_iso(Cu) arises from the spin polarization of inner s-electron pairs by unpaired d-electrons (also known as core polarization), leading to a net negative electron spin density at the nucleus and producing a negatively signed A_iso. Alternatively, net unpaired electron spin density can also occur at the nucleus by mixing of s-orbital character with the d-orbitals (d−s mixing), providing a positively signed contribution to A_iso.⁴⁹ The magnitude of A_iso can also be impacted by covalency, where unpaired spin delocalization leads to both weaker d−s mixing and core polarization.

Significant d−s mixing requires a sufficient reduction in symmetry at the Cu site.⁵⁰ which should in turn be observable through structural characterization. Therefore, we proceeded to investigate the coordination environments of H117A/M121H-Az and M13F/H117A/M121H-Az by protein crystallography.

2.3. Structural Characterization of H117A/M121H-Az & M13F/H117A/M121H-Az.

To determine whether the Cu site in H117A/M121H-Az has low symmetry, we obtained crystal structures of Cu^II−H117A/M121H-Az and Cu^II− M13F/H117A/M121H-Az, which were resolved at 2.1 Å and 2.6 Å resolutions, respectively (Figure 4). Inspection of the Cu binding sites in either system reveals an exceptionally distorted geometry. Distinct from all other systems listed in Table 1, we observe N_ε—rather than N_δ—coordination of Cu through His121. A dihedral angle ψ(C_β−S_Cys−Cu−N2) ~30° is found, deviating from the ideal 0° for perfect axial positioning, but much closer to an axial position than the ψ(C_β−S_Cys−Cu−N2) of 90−110° found in traditional T1Cu sites such as WT-Az or amicyanin. Furthermore, incorporation of Ala117 produces a change in configuration of the 113−120 loop region, enabling access of H₂O as exhibited by the appearance of two H₂O in the vicinity of Cu (Figure 4B).

Figure 4.

(A) Schematic description of nomenclature for the description of T1Cu site geometric parameters summarized in Table 1. (B−D) Cu binding site of Cu^II−H117A/M121H-Az as determined by protein crystallography together with several PCS geometric parameters including distances (B) and angles (C). Residues Gly45, His46, Cys112, and His121 are shown as sticks. Two waters are observed at 3 of 4 Cu sites, referred to as “1” and “2”. (D) Crystallographic depiction of the H117A/M121H-Az Cu binding site. (E) Overlay of Cu-binding sites of subunit 1 of H117A/M121H-Az (PDB ID 9OH6, teal, red H₂O) and M13F/H117A/M121H-Az (PDB ID 9OH7, gray, dark red H₂O). (F) Correlation of A_iso and ∠(S_Cys−Cu−N1) of proteins detailed in Tables 1 and 2. (G) Correlation of E°′ and ∠(S_Cys−Cu−N1) Linear correlations shown as dashed lines in (E,G) produce coefficients of determination (R²) of 0.85 and 0.82, respectively. Atoms in (D,E) are color coded as Cu (orange), N (blue), O (red), and S (yellow).

Several parameters are important in defining the Cu geometry in 3 and 4 coordinate systems, i.e., trigonal planar vs T-shaped vs tetrahedral vs square planar, and in turn, the degree of distortion between different conformations. These include the angles between ligands x,y,z (∠(L_x−M−L_y)) and the dihedral describing planarity,ψ(L_x−M−L_y−L_z). Distorting from a perfectly planar to tetrahedral geometry, L_x−M−L_y−L_z moves from |180°| to |119.66°|, with L_x−M−L_y angles of trigonal and square-planar geometries shifting from 120° and 90° (respectively) to 109.75°.

Comparing the crystal structure of H117A/M121H-Az with WT-Az and several other T1Cu proteins (Table 2), we observe that the dihedral angle describing “planarity” in Cu− H117A/M121H-Az, ψ(S_Cys−Cu−N1−N2), averages to 164°, approximately 14° larger than found in other green Cu proteins and ~−10° smaller than that of WT-Az. Therefore, rotation of the ligand environment does not appear to induce a significant tetrahedral distortion. However, ∠(S_Cys−Cu−N1) significantly increases to 145°. Combining observations of ψ(S_Cys−Cu−N1−N2) and ∠(L_x−M−L_y), Cu^II−H117A/M121H-Az represents a near-perfect distortion between trigonal (120°) and T-shaped (180°) geometries (Figure S7). Furthermore, we observe positive correlations between the angle ∠(S_Cys−Cu−N1), which corresponds to ∠(S_Cys−Cu−N_δΗ46), A_iso, and E°′ across the protein series detailed in Tables 1 and 2, as shown in Figure 5E,F.

Table 2.

Structural Parameter Comparison of Cu¹¹−H117A/M121H-Az Variants with Several Literature Examples of Blue and Green TICu Proteins^a

		(Å)				∠ (°)		Ψ (°)
protein	PDB	Cu−SCys	Cu−N1	Cu−N2	Cu−X	S−Cu−N1	S−Cu−N2	Cβ−S−Cu−X	Cβ−S−Cu−N1	S−Cu−N1−N2
WT-Az	5AZU51	2.26	2.06	2.03	3.13	134	121	−1	−87	−172
amicyanin	1AAC52	2.11	1.95	2.03	2.91	136	113	−4	−109	−147
pseudoazurin	8PAZ53	2.13	2.01	2.01	2.71	133	114	−7	−113	−136
umecyanin	1 X 9R54	2.21	1.94	2.04	2.23	137	116	−10	−106	−144
stellacyanin	1JER55	2.18	1.96	2.04	2.21	133	101	−1	−107	−148
plantacyanin (spinach)	1F5656	2.18	2.05	1.99	2.70	137	112	−7	−108	−138
plantacyanin (cucumber)	2CBP57	2.16	1.93	1.95	2.61	138	110	−6	−109	−135
nitrite reductase	1NID58	2.24	2.02	2.03	2.56	132	104	−21	−125	−118
M121H-Az pH 3.5	1A4C31	2.06	2.08	1.94	2.12	135	122	−15	−111	−152
M121H-Az pH 7	1A4B31	2.15	2.13	2.03	2.26	131	110	−22	−122	−125
		(Å)				∠ (°)		Ψ (°)
protein	PDB	Cu−SCys	Cu−N1	Cu−N2	Cu−X	S−Cu−N1	S−Cu−N2	Cβ−S−Cu−X	Cβ−S−Cu−N1	S−Cu−N1−N2
H117A/M121H-Az	9OH6	2.15(3)	2.05(3)	2.19(6)	2.86(2)b	145(1)	112(4)	35(4)	−128(7)	164(4)
H117A/M121H-Az (QM/MM)		2.18	2.00	2.02	3.29	135	115	31	−117	151
M13F/H117A/M121H-Az	9OH7	2.14(10)	2.03(8)	2.16(15)	2.46c	146(7)	110(7)	35(2)	−120(8)	156(8)

^aThe nomenclature for positions 1, 2 and X for both Az and the TICu proteins summarized in Figure 4. For all multimeric structures parameters are reported as averages over all subunits, apart from M121H—Az pH 3.5 in which parameters are reported for subunit 2 (chain C). Errors are calculated as s based on the deviation in each parameter between each subunit.

^bFound at 3 of 4 Cu sites.

^cFound at 1 of 4 Cu sites.

Figure 5.

(A) Aligned overlap of the crystal structures of WT-Az (gray, PDB 5AZU) and H117A/M121H-Az (teal, 9OH6). (B) DFT-calculated spin densities of WTAz and (C) H117A/M121H-Az abbreviated model 2. SCF spin density surfaces are plotted using an isovalue of 0.005. Arrows indicate orientation of g-tensors as labeled. Noncarbon atoms are color coded as N (blue), S (yellow), orange (Cu).

Refinement of the XRD structure of M13F/H117A/M121H-Az reveals minor perturbations to the Cu binding site from that of H117A/M121H-Az (Figure 4E), including a small increase in the average ∠(S_Cys−Cu−N1) to 146° and almost full exclusion of H₂O; rather than the 1−2 H₂O at each Cu site of the H117A/M121H-Az tetramer, only one H₂O is observed within 4 Å of one Cu site in the M13F/H117A/ M121H-Az tetramer.

2.4. Rotation of the Cu PCS and Interplay of ∠Cu(S_C112−Cu−N_δH46) & A_iso via QM Calculations.

To better understand the connections between electronic and geometric structure in Cu^II− H117A/M121H-Az, we performed a combination of DFT-based QM/MM calculations and angle scans at the DFT and DLPNO−CCSD levels. QM/MM calculations were performed based on models of the measured X-ray crystallographic structure of Cu^II−H117A/M121H-Az as described above. Optimization of a 243 atoms QM cluster displays reasonable agreement with the experimentally determined geometries of H117A/M121H-Az and M13F/H117A/M121H-Az (Table 2), producing a moderate tetrahedral distortion as gauged by ψ(S_Cys−Cu−N1−N2). Inspection of the calculated distribution of unpaired spin density in WT-Az and H117A/M121H-Az reveals a major reorientation from the S_p plane, following the Cu−N_σ interaction (Figure 5). These results support our hypothesis that the positioning of Cu−N/O_σ interactions play a significant role in guiding the orientation of the Cu contribution to the SOMO, and that this dependence can be extended to rotations outside of the S_p plane.

Although the A-tensor of Cu is notoriously challenging to accurately calculate, the calculation of qualitative trends in A_iso(Cu) is much more straightforward. To dissect the geometric dependence of A_iso and d−s mixing, an abbreviated form of the QM/MM optimized model was generated, involving Cu coordinated by Cys112, His46, and His121 side chains truncated at the C^α position together with two H−F to simulate the H-bonding interactions otherwise provided by the amides of the proteinaceous backbone at positions N47 and F114. Rigid scans of ∠(S_Cys−Cu−N1) performed from 120°−180° (effectively moving from trigonal to T-shaped geometries) at both the TPSSh* and DLPNO−CCSD levels. A rigid scan approach was chosen to better correlate changes in electronic structure with an isolated geometric variable. The dependence of A_iso and Cu s-orbital spin density, ρ(s), on ∠(S_Cys−Cu−N1) are shown in Figure 6. A linear correlation is observed between ρ(s) and A_iso, as anticipated based on the formal first-order definition of A_iso(Cu) as the Fermi contact term. Similarly, the Löwdin Cu spin population increases from 0.54 to 0.68 across this range. We note that population analysis (such as Löwdin and Mulliken) requires partitioning of the three-dimensional spin density onto specific atoms and orbitals,⁵⁹ leading to deviations of ρ(s) = 0 from A_iso = 0. Importantly, we find that increase of ∠(S_Cys−Cu−N1) results in a progressive increase in A_iso, moving from negative to positive in sign. Correspondingly, increase of ∠(S_Cys−Cu−N1) induces increases in ρ(s), where the change in sign of ρ(s) from negative to positive around ~135°−140° demonstrates the influence of geometry on the competing action of inner s-shell spin-polarization vs d−s mixing.

Figure 6.

Changes in electronic structure of the H117A/M121H-Az model as a function of ∠(S_Cys−Cu−N1), which corresponds to ∠(S_Cys−Cu−N_δH46); a simplified model showing the change in this angle is provided in (A). (B) Changes in Cu s-orbital spin density (∠(s)) as a function of ∠(S_Cys−Cu−N1). (C) Correlation between ρ(s) with A_iso at indicated angles of ∠(S_Cys−Cu−N1). Scans provided in both (B,C) were performed for rigid (unrelaxed) models to isolate geometric/electronic correlations. (D,E) The change in free energy, △G°, of Cu^II and Cu^I states as a function of ∠(S_Cys−Cu−N1), relative to ∠(S_Cys−Cu−N1) = 120°. Calculations of △G° were performed for relaxed models which were optimized with limited constraints (as described in the Experimental Methods).

2.5. How Does the T-Shape Distortion Increase E°′?

To better understand the interplay of d−s mixing and ∠(S_Cys−Cu−N1) with E°′, additional energy calculations were performed. Changes in E°′ correspond to changes in the relative stability of the Cu^I vs Cu^II states, where stabilization of Cu^I/destabilization of Cu^II results in an increase of E°′.⁶⁰ To estimate the stability of Cu^I vs Cu^II as a function of ∠(S_Cys−Cu−N1), second-order Hessian calculations were performed for a series of geometrically relaxed structures along the ∠(S_Cys−Cu−N1) coordinate from 120°−160° generated by partially restricted DFT optimization (see Experimental Methods for details). We find that as ∠(S_Cys−Cu−N1) is increased the Cu^I state is preferentially stabilized, decreasing ΔG°(Cu^I−Cu^II) and consequentially increasing E°′ based on the Nernst equation, consistent with our experimental observations (Figure 6D).

Looking closer at the enthalpic and entropic contributions to G° along the ∠(S_Cys−Cu−N1) coordinate, as shown in Figure 6E, we find both Cu^II and Cu^I states are most enthalpically favored at 140−145°, with a greater degree of stabilization observed for Cu^II (relative to 120°), i.e. positive ΔH°(Cu^I−Cu^II). Above 140°−145°, both Cu^II and Cu^I state are enthalpically destabilized, with Cu^II becoming more destabilized than Cu^I ~ 153°, producing a negative ΔH°(Cu^I−Cu^II). Meanwhile, increase of ∠(S_Cys−Cu−N1) from 120°−145° is entropically favored for both Cu^I and Cu^II states, with Cu^I displaying the greater (more favorable) increase. Interestingly, little change in S° is observed above 145° for either Cu^II or Cu^I. Bringing these results together, we find that increase of ∠(S_Cys−Cu−N1) from 120°−145° is enthalpically and entropically favorable for both Cu^II and Cu^I states; however, the entropic gains in the Cu^I state outweigh the enthalpic gains of the Cu^II state, leading to an increasingly negative ΔG°(Cu^I−Cu^II) and therefore increased E°′ (Figure 6E). Beyond ∠(S_Cys−Cu−N1) ~ 145°, this balance switches from being entropically to enthalpically driven, with the destabilization of Cu^II outpacing that of Cu^I to continually drive to lower ΔG°(Cu^I−Cu^II) (Figure 6D).

From Table 2 and Figure 6E, proteins from WT-Az to M13F/H117A/M121H-Az display ∠(S_Cys−Cu−N1) in the enthalpically favored range of ~130−145°. Across this region, we see the continuous decrease of ΔG°(Cu^I−Cu^II), corresponding to an increase in E°′, driven by an entropic—rather than enthalpic—factor, despite little to no change in tetragonal distortion. This entropic factor may arise through the decreased centrosymmetry of the metal coordination site,⁶¹ borne out through ∠(S_Cys−Cu−N1) in a T-shape distortion (Figure 7), resulting in increased orbital mixing as evidenced by changes in A_iso.

Figure 7.

Schematic of the tetrahedral and tetragonal distortions of the approximately C_3v-symmetric T1Cu site contrasted with the T-shape distortion observed in the present study of H117A/M121H-Az. The term T-shape distortion describes the distortion of equatorial ligands from a trigonal planar toward a T-shaped geometry. Colored spheres represent approximate Cu site color.

2.6. Contextualizing the T-Shape Distortion.

T1Cu proteins have been proposed to function under the entatic principle, where the constrained coordination environment of Cu reinforces a Cu^I geometry on the Cu^II-bound state to minimize the reorganization energy in cycling between the Cu^II and Cu^I states.¹⁷ The degree of reinforcement of this state in turn provides a mechanism for tuning the reduction potential, where distortions toward Cu^II-preferred tetrahedral or tetragonal geometries are “less entatic”.^14,62−64 This has been supported through the extensive study of varying axial interactions in T1Cu sites in both native and redesigned cupredoxins, where coordinating residues (Glu, Gln, His) significantly decrease reduction potential, and conversely noncoordinating (Ala, Ile, Leu) display significantly increases.^2,9,20,65,66

Building on the entatic state principle, the coupled distortion model (CDM) has been proposed to account for both redox potential tuning and S_π vs S_σ bonding based on comparisons of electronic and geometric structures of the blue T1Cu proteins plastocyanin and plantacyanin with the green T1Cu protein nitrite reductase.^67,68 In brief, the CDM describes the weakening of the Cu−S_Cys bond with increasing axial interaction strength, which is coupled to a tetragonal distortion (Figure 7) resulting in mixing of S_σ character into the SOMO. Under the CDM, increased S_σ character is correlated with increased axial ligand strength and consequently lowered reduction potential.

The CDM has proven valuable in understanding the impact of tetragonal distortion on electronic structure and reduction potential, particularly in T1Cu sites with axial Met. However, the properties of a considerable number of T1Cu sites are not well-predicted by the CDM, including the unusually long axial ligation of Met in azurins or increased redox potential in green T1Cu sites like plantacyanin, AcoP, and the H117A/M121H-Az variants of the present study.^39,40,69,70 While the CDM provides a rationale for how T1Cu distortion to a “less entatic” state produces lower redox potentials and green T1Cu, the T-shape distortion model provides a functional rationale for the formation of green T1Cu sites with increased redox potentials without necessitating axial ligation. Practically, these principles are not exclusive, and we hypothesize that distortions through both ∠(S_Cys−Cu−N1) and the CDM function together with varying Cu−S_Cys covalency and secondary coordination sphere effects to intimately tune T1Cu redox potentials.

3. CONCLUSIONS

In summary, we have designed a new T1Cu variant of Az which demonstrates that the N/O_σ bonding interactions play a significant role guiding redox-active molecular orbital orientation in T1Cu centers, as accomplished by exchanging an equatorial N_σ for an axial N_σ results in the effective rotation of the Cu(d) contribution to the SOMO out of the S_π-containing plane This effective rotation not only produces a significant 100 mV increase in E°′ from 280 mV in WT-Az to 380 mV, but recontextualizes surrounding interactions between the SCS and the redox active molecular orbital, as exemplified by the Met13Phe mutation, which increases E°′ by 65 mV compared to H117A/M121H-Az, as opposed to a 22 mV decrease when incorporated in WT-Az. The significant increase in E°′ of H117A/M121H-Az relative to WT-Az is accompanied by a large positive increase in A_iso. Through a combination of spectroscopy, crystallography, QM calculations, and literature support, we have rationalized this increase in A_iso as an increase in d−s mixing as modulated through a T-shape distortion, manifested through the angle ∠(S_Cys−Cu−N1). Through systematic calculations at the DLPNO−CCSD and TPSSh* levels, we also find that the increase of ∠(S_Cys−Cu−N1) stabilizes the Cu^I-bound state to a greater extent than Cu^II in a T-shape distortion, consistent with experimentally observed increases in E°′ along this coordinate. This difference in stabilization arises from a greater entropic penalty for Cu^II at ∠(S_Cys−Cu−N1) < 145°, and a greater enthalpic penalty for Cu^II at ∠(S_Cys−Cu−N1) > 145°. These results provide direct connections between geometry, electronic structure, and functional properties, and, most excitingly, open previously unexplored routes to systematically modulating metal redox potential through the combination of spatial reorientation of the redox-active molecular orbitals and varying geometric distortion in the primary coordination sphere.

4. EXPERIMENTAL METHODS

4.1. Site-Directed Mutagenesis and Protein Expression.

Plasmids containing WT-Az, H117A/M121H (H117A/M121H-Az) and M13F/H117A/M121H mutations were constructed, expressed, and purified similar to previously reported protocols.^22,71 Specifically, the site-directed mutagenesis of wild-type azurin from Pseudomonas aeruginosa in a pET9a vector was carried out using the quick-change polymerase chain reaction procedure. The mutations were confirmed by DNA sequencing. The proteins were expressed in BL21(DE3) E. coli (New England Biolabs, Ipswich, MA). Cultures were grown in LB media for 8 h at 37 °C and used to inoculate 4 L flasks holding 1.5 L of 2x YT media containing 50 mg/L Kanamycin which were further grown at 25 °C while shaking at 190 rpm for approximately 14 h. Induction using IPTG (final concentration 75 mg/L) was performed at 30 °C when cultures reached an O.D. of ~0.8 and harvested by centrifugation at 8000 rpm (10 min, 4 °C) approximately 4.5 h after induction. Protein purification was performed following our previously reported protocol.^22,72

4.2. Preparation of pH-dependent Sample Series for Absorption and EPR Spectroscopies.

The preparation of all measured samples of H117A/M121H-Az were prepared by combining monomer protein fractions obtained by size-exclusion chromatography and concentrating to generate ~4 mM protein stock solutions in 25 mM Tris, pH 7.8. All buffers were prepared with CHELEX to remove extraneous metal ions from solution and further passed through a 0.22 μm filter. Samples of each protein were diluted to 100 mM based on ε₂₈₀ = 8000 M⁻¹ cm⁻¹ in 25 mM Tris (pH 7.8) and titrated using 0.1 eq. aliquots of Cu^IISO₄ to ensure proteins were capable of binding one (full) equivalent of Cu^II. For pH-dependent absorption and EPR measurements, aliquots of protein stock were used to generate ~1.4 mM protein solutions in 200 mM NaOAc (pH 5), 50 mM MES (pH 6), 50 mM MOPS (pH 7), and 50 mM Tris (pH 8). These solutions were metalated with 0.7 eq CuSO₄ using a 20 mM CuSO₄/H₂O stock solution to generate ~1 mM samples. Each protein solution was split to prepare respective samples for absorption, X-band EPR, and Q-band EPR spectroscopies. Final Cu-protein complex concentrations and molar extinction coefficients were determined by spin-integration of X-band spectra and comparison to a 2.054 mM CuSO₄ spin standard.

UV/vis absorption measurements were performed at room temperature using a Cary 60 spectrometer using a 1 cm⁻¹ QSIL quartz cell Aliquots of the same solution used to prepare X-band and Q-band EPR samples were diluted by a factor of 4× to ensure all spectra did not exceed absorption >1 across the 300−1100 nm range.

4.3. Redox Titrations.

All reduction potential measurement experiments were conducted under an N₂ atmosphere. Solutions containing 150 mM Cu^II−H117A/M121H-Az, 5 mM ferricyanide, 50 mM NaCl, and 50 mM MOPS buffer at pH 7 were prepared, along with a 50 mM stock of sodium ascorbate as reductant. Aliquots of ascorbate stock were titrated into the protein/ferricyanide solution while monitoring the absorption spectrum. The relative peak intensities at 420 nm (for ferricyanide) and 604 nm (for Cu^II−H117A/M121H-Az) were monitored as a function of reducing equivalents and used together with the Nernst equation to calculate the Cu^II/I potential with reference to 436 mV vs SHE for ferricyanide/ ferrocyanide.⁷³

4.4. Electron Paramagnetic Resonance Spectroscopy.

Continuous wave (CW) electron paramagnetic resonance (EPR) measurements were performed at X-band and Q-band frequencies (~9.65 and 34.5 GHz, respectively) using a Bruker EMXplus spectrometer. X-band measurements were performed using a Bruker high-sensitivity cavity (ER4119HS) and a nitrogen-flow cryostat (ESR900). A high-sensitivity Bruker Premium-X bridge with an integrated microwave frequency counter was employed, along with a magnetic field controller (ER083CS) that was calibrated using a Bruker NMR field probe (ER035M). All X-band measurements utilized a 4 G modulation amplitude with a 100 kHz modulation frequency, an 80.92 ms time constant, and 1 mW microwave power at ~85 K. Q-band measurements were performed using a Bruker ER5106 QT resonator with a liquid He flow cryostat (CF935). All Q-band measurements utilized a 20 G modulation amplitude with a 100 kHz modulation frequency, an 80.92 ms time constant, and 0.1 mW microwave power at 40 K. All spectra were recorded on frozen solutions. All spectra were processed and analyzed using the software package EasySpin (version 6.0.0),⁷⁴ as implemented in Matlab (version R2022b).

4.5. Protein Crystallography.

Cu^II-bound H117A/M121H-Az and M13F/H117A/M121H-Az crystals were prepared using the hanging drop method at 4 °C, where drops were placed on silica slides and equilibrated over wells containing 500 μL of crystallization buffer. Protein stocks were prepared at 1.5 mM in 25 mM Tris at pH 7.8 and were metalated by titration with 1.2 eq of Cu^IISO₄ in aqueous solution. Crystallization buffers contained 50 mM Tris at varying pH, PEG 4000 at varying concentrations, and 100 mM LiNO₃. Crystals of H117A/M121H-Az were collected and measured from drops prepared by mixing 1 μL protein stock with 3 μL crystallization buffer with pH 7.7 Tris and 27% PEG 4000. Crystals of M13F/H117A/M121H-Az were collected and measured from drops prepared by mixing 1.5 μL protein stock with 1.5 μL crystallization buffer with pH 7.4 Tris and 30% PEG 4000. Light green plate-like crystals appeared within 1 day and were harvested at 7 days. The crystals were dipped in cryo-protecting buffer containing 75% well buffer and 25% glycerol (by mass) before mounting and were frozen in liquid N₂ for diffraction analysis.

X-ray diffraction measurements were performed at beamlines 8.2.2. (H117A/M121H-Az) and 5.0.3 (M13F/H117A/M121H-Az) of the Advanced Lightsource (ALS) at Lawrence Berkeley National Laboratory (LBNL). Diffraction images were processed and reduced using Xia2-DIALS software^75,76 in the CCP4 software suite.^77,78 Molecular replacement solutions were identified using M121H/H46EAz (PDB: 4KWX)⁷⁹ as an initial search model for the Phaser module of the Phenix software suite.⁸⁰ Model building and model refinement steps were performed using the Phenix.refine-WinCoot interface,^81,82 and final models were evaluated using ProCheck and MolProbity.^83,84 Refinement statistics are provided in Table S1.

4.6. QM/MM and Ab Initio Calculations.

Quantum mechanics/molecular mechanics (QM/MM) modeling of H117A/M121H-Az was performed based on the crystal structure obtained in the current study. All QM/MM calculations were performed using a previously presented protocol^22,85 in ASH,⁸⁶ which provides an interface for the QM program ORCA,^87,88 version 5.0.3, with the MM program, OpenMM. MM calculations used the CHARMM36⁸⁹ force field. Protons were added for pH 7, and standard protonation states were assumed for all residues, including Ne protonation of His46 and N_d protonation for His121, deprotonation of Cys3 and Cys26 to form the cystine bond, and deprotonation of Cys112 to allow for Cu^II coordination. TIP3P water molecules were added to the protein using a 60 × 60 × 60 ų solvation box, which included 6667 water molecules in total. Additionally, a total ionic strength of 0.1 was set using Na⁺ as positive ions, and Cl⁻ as negative ions. Hydrogen positions were minimized using a 10000-step conjugate gradient algorithm, and water positions were relaxed using a 10000-step NVT simulation of 290 K with a periodic boundary condition. A QM region was selected to include Cu^II, Met13, Gln14, Met44-Asn47, Cys112-Gly123, and the two neighboring crystallographic water molecules, totaling 241 atoms. The QM/MM coupling was described using electrostatic embedding together with H link atoms and a charge-shifting scheme.⁹⁰

DFT-based QM calculations employed the TPSSh density functional^91,92 with a 20% Hartree−Fock exchange (referred to as TPSSh*) with a D3BJ dispersion correction.^93,94 The choice to increase HF exchange was based on improved experimental agreement of spin density distribution in the T1Cu site. Based on previous QM/MM studies of Az,^85,95 the DKH2 Hamiltonian was selected to account for relativistic effects in DLPNO−CCSD calculations, and the zeroth order relativistic approximation (ZORA) was selected for DFT calculations.⁹⁶⁻¹⁰⁰ Recontracted DKH-def2-TZVP(-f) basis sets were used for all atoms and all calculations,^101,102 together with custom basis sets for Cu and S which have previously been shown to reasonably treat hyperfine parameters,^85,95 along with decontracted def2/J basis sets for the RI approximation to the Coulomb intergals.¹⁰³ These basis sets are provided in the Supporting Information. The chain of spheres approximation (COSX) was used for the exchange.¹⁰⁴

Calculations of ΔG° were performed based on models shown in Figure 5 for relaxed Cu^I and Cu^II geometries, where optimizations were subject to limited constraints intended to imitate the steric restrictions of the T1Cu site These restraints include: (1) fixing of spatial coordinates of the three terminal alkyl carbons representing C^α of respective amino acids, (2) fixing dihedral angles C^α−C^β−S_Cys−Cu, C^β−S_Cys−Cu−N1, and C^β−S_Cys−Cu−N2.

Supplementary Material

Supporting Information

The Supporting Information is available free of charge at 10.1021/jacs.5c07009.

Additional Figures, Tables, molecular coordinates and custom basis sets from calculations, and crystallographic information (PDF) calculations_xyzs (ZIP)

ABBREVIATIONS

Az

azurin

CW

continuous wave

DFT

density functional theory

DLPNO−CCSD

domain-based local pairwise natural orbital coupled cluster theory singles doubles

EPR

electron paramagnetic resonance

PCS

primary coordination sphere

QM/MM

quantum mechanics/molecular mechanics

SCS

secondary coordination sphere

SOMO

singly occupied molecular orbital

T1Cu

type 1 copper

WT

wild-type

WT-Az

wild-type azurin