Introduction

Redox-active metalloenzymes possess optimized primary, secondary, and outer-sphere interactions to control catalysis and electron transfer (ET)1,2,3,4,5. The active site minimizes oxidation state-dependent structural changes, which lowers the reorganization energy (λ), facilitating ET and catalysis. The coordination number, geometry, and protein matrix are critical determinants facilitating redox catalysis. Copper (Cu)- -containing metalloproteins are at the heart of numerous biochemical processes, such as respiration, ET, biomass degradation, hydrocarbon oxidation, and the biosynthesis of hormones and cofactors1,6,7. The type-1 Cu centers are involved only in ET, but the type-2 Cu proteins can participate in ET or catalysis1,6. Cu is predominantly found in a 3-5 coordinate geometry with S, N, and O-containing ligands such as Cys, Met, His, terminal amine, Asp/Asn, Tyr, or water. Amine oxidase (AO)8, nitrite reductase (NiR)9, and the CuB site of cytochrome c oxidase (CcO)10 feature a Cu(His)3 ligation (Fig. 1A, B), also found in the coupled binuclear active site of tyrosinase (Ty). The lytic polysaccharide monooxygenases (LPMOs) have a T-shaped Cu(His)2NH2 geometry formed by a His-brace (Fig. 1C)11,12,13,14,15. In addition to O2, LPMOs can utilize H2O2 to oxidatively depolymerize recalcitrant cellulosic substrates, causing them to be designated as peroxygenases 16,17,18,19,20,21.

Fig. 1: Select Cu proteins with CuN3 or CuN4 binding sites.
figure 1

A Amine oxidase (AO), PDB: 2W0Q, B the CuB site of cyt c oxidase (CcO), PDB: 2OCC, and C lytic polysaccharide monooxygenase (LPMO), PDB: 5FJQ feature three protein-derived nitrogen ligands. The CuB site of particulate methane monooxygenase (pMMO), PDB: 7S4H, has a (His)3(NH2) ligation (D) with axial waters. The cartoons are shown in turquoise, His as turquoise/blue sticks, Cu as an orange sphere, and water as red spheres. The figure is made in PyMol.

Full size image

Particulate methane monooxygenases (pMMOs) are membrane-bound enzymes initially thought to feature a bimetallic Cu-Cu active site22,23. However, based on recent structural and EPR studies, pMMOs are now considered to contain only mono-copper active sites24,25,26,27. The membrane-bound CuC or the CuD site is the proposed active site where O2 binding and CH4 oxidation occur25,26,27. Another Cu site, CuB, located in the soluble PmoB subunit, has a unique square pyramidal Cu(His)3(NH2)(OH2) coordination from 3 His, a terminal amine, and axial water (Fig. 1D). Although not the hydrocarbon binding site, the CuB site is important since deleting CuB ligands in homologous hydrocarbon monooxygenase reduces the enzymatic activity by >80%28. Besides being unreactive to exogenous substrates, the CuB site is also difficult to reduce25. Thus, the CuB appears to be a structural site but is still important for function25. The unreactive nature of the CuB is intriguing, given that many synthetic Cu complexes with tripodal tetradentate ligands can activate O2 and H2O229,30,31,32.

How the structural differences translate to function that makes a Cu(His)2NH2 site (LPMO, etc.) active to exogenous substrates but not the Cu(His)3NH2 site of pMMO CuB remains unknown. Given that the active sites are embedded in diverse protein matrices of varying secondary and tertiary structures, it becomes challenging to attain a unified and generalized view of how the coordination number, ligand environment, and remote interactions influence ET, λ, and H2O2/O2 reactivity, ultimately determining functional outcomes. Governed by the primary sequence repeat pattern, the de novo protein design approach enables the creation of active sites within similar structural motifs, such as the α-helix, offering insights into how the immediate coordination environment of metalloenzymes determines their functions3,33,34,35. The knobs-into-holes (KIH) packing of hydrophobes provides a distinctive handle to produce self-assemblies36,37,38 with controlled oligomeric states where an active site of interest can be introduced. Given the modular nature of the design, different types of coordination numbers and geometries can be constructed within de novo proteins, as dictated by self-assembly. Indeed, recent success has been achieved in creating artificial metalloenzymes (ArMs) within de-novo scaffolds 39,40,41,42,43,44,45,46,47,48.

One such example is the artificial Cu proteins (ArCuPs) that function as peroxidases and peroxygenases featuring Cu(His)3 active sites within trimeric (3SCC) self-assemblies45,46. The 3SCC ArCuP activates H2O245 and electrocatalyzes the 2e− and 4e− oxidation and peroxygenation44. To elucidate what factors control the physical and reactivity outcomes of Cu(His)2NH2 vs. Cu(His)3NH2 sites in biology that make the former reactive but the latter unreactive, we have designed a Cu(His)4 site within a de novo designed tetrameric ArCuP (4SCC), as a model of the pMMO CuB site. We demonstrate how the differences in the coordination properties of 4SCC and 3SCC alter their O2 and H2O2 activation and ET properties. The reduction and reoxidation kinetics probed by electronic and EPR spectroscopy suggest that 3SCC is easier to reduce and reoxidize than 4SCC. The 4SCC is also unreactive to the peroxygenation of C-H substrates. To address how coordination differences influence ET properties and the resultant catalytic outcomes, the λ values of these ArCuPs have been experimentally determined. 4SCC has an inherently higher λ than 3SCC, which is significantly dominated by the solvent reorganization energy, mediated by an H-bond involving a primary-sphere His and a second-sphere Glu, which enables the formation of an extended H2O-mediated H-bonding network. Intriguingly, when this H-bond is deleted, the water-mediated H-bonding network is also removed. This lowers the solvent reorganization and renders the inactive 4SCC to become an active peroxidation catalyst.

Results and discussion

Design of modular peptide assemblies

A hallmark of de novo protein design is the ability to construct self-assemblies with different oligomeric states by changing the primary sequence. This is achieved by controlling the KIH packing of a and d site residues in the heptad (abcdefg)n. For example, introducing an Ile at the a and d site of the heptad (IAAIKQE)n produces the trimeric 3SCC ArCuP45. To design the tetrameric 4SCC assembly, the (LAAIKQE)n motif was chosen, having a Leu at the a site and an Ile at the d site49. A five-heptad sequence was chosen to optimize packing and ensure tetramer formation with the desired metal coordination number. To introduce the His site in 4SCC, we chose the 9th position of the heptad, similar to the 3SCC, to allow a direct assessment of how the immediate coordination environment alters their physical and reactivity properties. The Ala residues favor the α-helix formation, while the Glu and Lys at e and g positions control the parallel orientation of the helices by ion pairs. An acetylated and amidated Gly residue caps the N- and C- termini, respectively. A Trp chromophore is introduced at an f site on the last heptad to aid in quantification. The final sequence of 4SCC is shown in Table 1.

Table 1 Peptide sequences used in this study
Full size table

Structure and electronic properties of 4SCC

4SCC was crystallized in the Cu-bound form and diffracted to 1.36 Å (statistics in Table S1). The structure reveals that 4SCC has a tetrameric quarternary structure produced by parallel assembly of the monomers (Fig. 2A), as designed. All four His residues point toward the helical axis and bind Cu exclusively via the Nε atoms. A water molecule occupies an axial position at 2.6 Å, forming an approximate square pyramidal Cu(His)4OH2 coordination (Fig. 2B). The His residues are slightly distorted from the Cu plane with Cu-Nε distances of 2.0–2.2 Å (Fig. 2C). In the CuB site of pMMO, the Cu-NHis distances range from 2.1–2.3 Å, and Cu-NH2 distance is 2.5 Å24. Two opposing Nε atoms in 4SCC are located below Cu with an ∠Nε-Cu-Nε angle of 167° (Fig. 2D). The remaining two Nε atoms lie above the Cu with an ∠Nε-Cu-Nε angle of 189°. The ∠Nε-Cu-Nδ and ∠Nδ-Cu-NH2 angles in the pMMO CuB site are 119° and 161°, respectively (Fig. S1). In 4SCC, the orientation of His Cε atoms is such that two opposing His residues are tilted away from each other at a greater extent than the others (140° vs. 163°). The electronic spectrum of 4SCC has a d-d band (Fig. 2E) at ~600 nm. The EPR spectrum of 4SCC (Fig. 2F) is characteristic of a tetragonally elongated type-2 Cu center having an axial g tensor, with gx,y = 2.048, gz = 2.253 and resolved Cu hyperfine splittings of Ax,y = 71 MHz and Az = 543 MHz (Table S2). Additional superhyperfine splittings from the 14N nuclei of four His ligands are observed with AN(x,y) = 51, 41, 44, 41 MHz, and AN(z) = 55, 41, 44, 41 MHz. The observation of well-resolved Cu-14N(His) superhyperfine splittings suggests a high degree of structural homogeneity in the Cu coordination sphere of 4SCC. Finally, the axial symmetry (gz > gx = gy) indicates that the unpaired electron resides in the dx2-y2 orbital, pointing toward the four N ligands on the tetragonal plane.

Fig. 2: X-ray structure, UV-vis, and EPR analysis of 4SCC.
figure 2

A Cartoon representation of a 1.36 Å X-ray structure of 4SCC showing the parallel tetrameric assembly. B The square pyramidal Cu site, viewed perpendicular to the helical axis. C Views from the N-termini showing Cu binding to His Nε atoms. D Orientations of the Nε atoms relative to Cu. One set of Nε-Cu-Nε atoms from opposing His residues creates an obtuse angle, while the other set forms a reflex angle. The Cε-Cu-Cε orientations are wider in one set of His than in the other set. The axial water is omitted for clarity from D. The cartoons are shown in magenta, His as yellow/blue sticks, Cu as an orange sphere, and the axial water as a red sphere. Figures are generated in PyMol (PDB 8VHS). E d-d band of 4SCC. F The X-band EPR spectrum of 4SCC (blue) and the simulated spectrum (dark gray). Insets show expanded AN splittings. EPR data was collected at 77 K.

Full size image

Reactivity differences between 3SCC and 4SCC

3SCC catalyzes H2O2 reduction electrochemically45. However, 4SCC does not, as no significant catalytic current corresponding to H2O2 reduction is observed with the latter (Fig. S2). 3SCC also electrocatalyzes the oxidation and peroxygenation of benzylic C-H substrates with catalytic proficiencies (1/kTS = kcat/KM/kuncat) in the 300–1200 M−1 range. The catalytic proficiencies can be enhanced by outer-sphere steric modifications that affect catalysis by facilitating substrate access, influencing the nature of Cu-oxygen species, and improving stabilities46. On the contrary, 4SCC does not peroxygenate or oxidize these substrates. This is evidenced by the small catalytic current and charge (Fig. S3) and a lack of product after electrolysis. These results indicate that differences in metal coordination in our de novo ArCuPs are critical determinants of their reactivity. This is consistent with the fact that Cu active sites with three protein-derived N ligands (e.g., LPMO, AO, CuB of CcO, NiR, etc.) are reactive, while the CuB site of pMMO, which has four protein-derived ligands, is not.

3SCC is easier to reduce than 4SCC

The reactive CuI form is necessary for generating the Cu-oxygen intermediates6,16. Given the coordination differences between 3SCC and 4SCC, one would expect the inherent reducibility to be different, which could influence their reactivity. To test this, we assessed the reduction kinetics using Trp fluorescence as a reporter of the Cu oxidation state. In the original designs, a Trp was present at the end of the peptide, which is too far from the Cu site to show any oxidation-state-dependent emission change. Thus, the Trp was moved to the 10th position (A10W variants), one residue away from the Cu site (Fig. 3A–B). These constructs also produce the type-2 Cu sites (Fig. S4). Kinetics of CuII → CuI reduction with ascorbate (Asc) shows that A10W-3SCC is reduced faster than A10W-4SCC (Fig. 3C, S5, S6). The rates vary linearly with [Asc] (Fig. 3D), yielding second-order rate constants of 1.9 × 10−6 M−1s−1 and 7.6 × 10−8 M−1s−1 for A10W-3SCC and A10W-4SCC, respectively (Table S3). The two orders of magnitude rate difference validate the notion that the CuI state is easily accommodated within a Cu(His)3 motif rather than Cu(His)4, a conclusion consistent with the experimentally determined λ of these ArCuPs (vide infra). In LPMO, the rate of reduction is fast and not rate-limiting17. In contrast, the CuB site of pMMO is reluctant to reduce25. Our results with 4SCC mirror this lack of reducibility of the pMMO CuB site.

Fig. 3: CuII→I reduction kinetics of A10W mutants monitored using Trp emission.
figure 3

Cartoon showing the A10W mutation in 3SCC (A) and 4SCC (B) to bring the Trp closer to Cu. C Reduction of CuII-A10W-3SCC (red) and CuII-A10W-4SCC (blue) by Asc. D The kobs depend linearly on [Asc] with a significantly higher rate in 3SCC than in 4SCC. Here, 0.125 mM CuII-ArCuPs were anaerobically mixed with varying [Asc] in 10 mM MES buffer pH 6.5, and time-dependent emission was monitored (C shows the data with 10-fold Asc). The error bars in D represent SD from three independent measurements. λex = 280 nm, λem = 340 nm for 3SCC and λem = 345 nm for 4SCC.

Full size image

H2O2 and O2-dependent reoxidation

LPMOs can utilize either O2 or H2O2 as an oxidant16,17,18,19,20,21. To establish the relative differences in H2O2 vs. O2 activation, reoxidation studies of CuI-ArCuPs with each oxidant were performed. In both cases, the final species is in the CuII oxidation state (vide infra EPR). In contrast to CuII-3SCC, which produces the CuII-hydroperoxo species with H2O245, the CuI reactivity is more complex (Fig. 4A).

Fig. 4: H2O2 and O2 activation by the ArCuPs monitored by UV-vis and Trp emission.
figure 4

A H2O2 vs. O2 activation pathways delineating the production of various CuII-oxygen species. B–E Reoxidation of 3SCC and 4SCC with H2O2 and O2 monitored by UV-vis. The time traces represent spectra collected at 1 min (gray) and 100 min (red, violet, blue, green) after initiating the reactions. F The H2O2-dependent (black = 10, wine = 20, blue = 25, and green = 50-fold) reoxidation of A10W-3SCC monitored by Trp emission, and G the kobs vs. [H2O2] plots for A10W-3SCC (red), and A10W-4SCC (blue). Error bars in G represent SD from three experiments. [ArCuP] = 0.125 mM, pH = 6.5. For oxygenic reoxidation, O2-saturated solutions were used.

Full size image

In the peroxide pathway of LPMO, the CuI--H2O2 association complex undergoes homolytic O-O bond cleavage to produce CuII-OH and •OH by Fenton-type chemistry17,21. H-atom abstraction (HAA) by the •OH produces the oxyl CuII-O• and H2O. In the O2-dependent pathway, cupric superoxo CuII-O2•− is produced initially, which, upon H+/e− transfer, is converted to CuII-OOH. The distal O protonation and ET produce the oxyl and H2O. The corresponding proximal O protonation produces the CuI-H2O2 association complex, which enters the H2O2-dependent pathway. Thus, the reoxidation of CuI-LPMO with H2O2 can produce various Cu-oxygen species, such as CuII–OH---OH•, CuII–O•, or CuII–OH17. By analogy, we expect similar species to be produced during reoxidation of the ArCuPs.

When reacted with H2O2, CuI-3SCC produces a prominent band at ∼332 nm (Fig. 4B). A definitive assignment of this species is challenging due to many possibilities, as discussed above. A CuII-hydroperoxo species appears in several model complexes in the 325-350 nm range50,51,52. Although CuII-OOH can be prepared with CuII and H2O2 or from CuI + O2 in the presence of an electron and proton source, in one example, a CuII-hydroperoxo species has been produced from CuI and H2O253. In this case, after the initial formation of CuII-OH and •OH from the homolytic cleavage, the •OH reacts with CuI to produce more CuII-OH, which then reacts with H2O2 to produce CuII-OOH and H2O (Fig. S7). In the absence of substrate, when CuI-3SCC is reacted with H2O2, the active site His is oxidized45, as observed in LPMOs as well16,20. The reaction product of CuI-3SCC and H2O2 is best assigned as a CuII-oxygen species with a partially oxidized His.

The reaction of CuI-3SCC with O2 is ~11-fold slower than with H2O2 as monitored by UV-Vis (Fig. 4C, S8), producing a major band at ~336 nm plus a weaker feature at ∼422 nm and another broad absorption around 540-570 nm. These features are typically observed for Cu-oxygen species, namely, CuII-hydroperoxo or CuII-alkylperoxo 54,55.

CuI-4SCC with both H2O2 and O2 produces a similar species, with weak features at ~430 nm and 545 nm and a broad shoulder around 360 nm (Fig. 4D, E). 4SCC has poor reactivity, as no noticeable change in absorption features is observed at longer time points.

The reoxidation of CuI-3SCC-A10W with H2O2 and O2 occurs at a rate comparable to that of CuI-3SCC, and it also peroxidizes benzyl alcohol (Fig. S8). The reoxidation was monitored by Trp quenching as well. Increasing the molar ratio of H2O2 (10 to 50-fold) quenched the Trp signal (Fig. 4F, G) for A10W-3SCC, with a second-order rate (kreox) of 1.6 × 10−4 M−1s−1 (Table S3). At higher H2O2 concentrations, saturation is observed in A10W-4SCC (Fig. 4G) with a kreox of 2.8 × 10−5 M−1s−1. Analogous experiments with O2 saturated solutions show that CuI-A10W-3SCC reoxidation is faster than CuI-A10W-4SCC (Fig. S9), with a pseudo-first-order rate of 0.0019 s−1 for the former. No kinetic information was extracted for A10W-4SCC due to its poor reactivity with O2.

Time-dependent EPR is further used to examine the reoxidation of CuI-ArCuPs with saturated O2 solutions or with 100-fold excess H2O2 anaerobically. With O2, a gradual reoxidation is observed (Fig. 5A, C), producing a major species ( ~ 70%) with gx,y of 2.059, gz of 2.276, and Az of 525 MHz (Table S2), and a minor species with g values of 2.054, 2.309, and Az of 489 MHz. With H2O2, the signal saturates within ~30 min (Fig. 5B, C), further validating the faster reoxidation process than O2. The major species (80%) produced with H2O2 have gx,y of 2.057 and gz of 2.265, and Az of 534 MHz (Table S2), while the remaining 20% of the species have g values of 2.063, 2.299, and Az of 489 MHz. The reoxidation is much slower for the CuN4 site (Fig. S10, S11), as anticipated.

Fig. 5: H2O2 and O2-dependent reoxidation of CuI-3SCC monitored by EPR.
figure 5

Time-dependent X-band EPR data for the reoxidation of 3SCC with O2 (A) and H2O2 (B) at 5 min (blue), 10 min (violet), 30 min (green), and 60 min (red). CuI-3SCC in B is shown as a gray trace. C The corresponding EPR intensity plot vs. time from double integrations (H2O2 = circles; O2 = stars). A 100-fold H2O2 was used in these experiments, and samples were frozen at various intervals. For O2, CuI-peptide was mixed with an equal volume of O2-saturated buffer, stirred, and then frozen at different intervals. The simulated spectra for 60-min samples are shown as insets in A and B. Final [ArCuP] = 0.5 mM. Data were collected at 77 K. pH = 6.5.

Full size image

To summarize, 3SCC undergoes faster reoxidation, having higher rates with H2O2 than O2, like LPMOs16. The low reactivity of 4SCC resonates with the inertness of the CuB site of pMMO 25.

Redox properties of the ArCuPs

Protein-film voltammetry was used to determine the redox properties of the ArCuPs adsorbed on a pyrolytic graphite electrode (PGE). Repeated cyclic voltammetry (CV) scans show that the ArCuPs remain stable on the electrode surface (Fig. S12)45. The cathodic and anodic peak currents vary linearly with the scan rate, suggesting a surface-controlled redox process. Both the ArCuPs have E°’ in a narrow range, with a difference of −16 mV, 256 mV for 3SCC45, and 240 mV for 4SCC (Fig. 6A) at pH 6.5. These E°’s are similar to the range of other His-coordinated type-2 Cu proteins irrespective of their role as catalytic or ET sites1,6. The CV for 4SCC is broader than 3SCC, which suggests a slower ET and that dispersion effects contribute to the redox process 56.

Fig. 6: Experimental determination of reorganization energies of the ArCuPs using electrochemistry.
figure 6

A CVs for 3SCC (red), 4SCC (blue), and the blank electrode (black) in 80 mM mixed buffer pH 6.5. Scan rate = 50 mV/s. The inset shows background corrected redox peaks. B RDE data for 0.25 mM 3SCC solutions at 295 K in 80 mM mixed buffer, 100 mM KCl at pH 6.5. Rotation rates = 250 rpm (black), 512 rpm (red), 850 rpm (blue), 1300 rpm (magenta), and 1850 rpm (olive). The inset shows the Koutecký-Levich (K-L) plots in the mass transport region used to obtain kET. The data are shown at potentials 0.017 V (black), 0.039 V (red), 0.062 V (blue), 0.084 V (magenta), and 0.108 V (olive). C Arrhenius plots were constructed using the kET obtained from the K-L analysis. D λT at different potentials obtained from the slopes of the Arrhenius plots. The dashed lines show the average λT. E ln(i) vs. t plots obtained from CA experiments. The slopes yield kapp (s-1) at a fixed η (data shown for η = 100 mV; see the SI for the remaining). F λCu-P measured from CA experiments at different η values. The dashed lines show the average λCu-P. The error bars represent SD from 2-3 independent sets of experiments. For CA experiments, 10 µL of 0.25 mM ArCuPs were dropped on PGE and dried under N2, and data were collected in 80 mM mixed buffer at pH 6.5. In C, D, and F, 3SCC is red, 4SCC is blue, and K6E/E8K-4SCC is orange.

Full size image

4SCC has a higher λ than 3SCC, with a significant solvent contribution that is negligible in 3SCC

We have probed the ET properties to investigate how factors beyond the primary coordination influence the reactivity differences between the ArCuPs.

As the Marcus Eq. (i) describes, ET depends on λ 57,58.

$${k}_{{ET}}=\frac{4{\Pi }^{2}{H}_{{AB}}^{2}}{h\sqrt{4\Pi \lambda {k}_{B}T}}\exp \left[-{\left(\Delta {G}^{0}+\lambda \right)}^{2}/4\lambda {k}_{B}T\right]$$
(i)

The total λ (λT) for metalloproteins is contributed by the inner-sphere (λCu) and the outer-sphere (λOS) reorganization, which comprises the surrounding protein matrix (λP) and the solvent (λH2O) reorganization4,59,60. The λCu reports any geometrical changes associated with the redox event at the Cu site. Traditionally, λ for Cu proteins such as azurin (Az) has been determined by photoexcitation followed by flash-quench of Ru/Os modified Az60 or pulse radiolysis61. Electrochemistry is an alternative method to derive λ from Arrhenius’s analysis of temperature-dependent ET rates62,63,64,65. We have employed two electrochemical methods to determine λ: rotating disk electrochemistry (RDE) and chronoamperometry (CA). In RDE, the ArCuPs are dissolved in solution, and linear sweep voltammograms (LSVs) at different electrode rotation rates are measured as they diffuse to the electrode surface (Fig. 6B, S13−S16). Since the protein is in solution, λ measured from RDE has both the inner-sphere (λCu) and outer-sphere components (λP + λH2O). The ET rates (kET) thus determined (see Methods for details) are subjected to Arrhenius analysis (Fig. 6C) to obtain λT. The λT obtained using this method is 0.34 eV for 3SCC and 0.89 eV for 4SCC (Fig. 6D, Table 2).

Table 2 Selected λ for Cu proteins and model complexes
Full size table

Contrary to the RDE method, the λ measured by CA are obtained from ArCuPs that are directly immobilized on the electrode surface. Therefore, contributions from the solvent reorganization are expected to be minimal compared to situations where the protein is freely diffusing in solution62,63. Thus, the λ determined using CA has contributions from the Cu site and any changes to the polypeptide matrix during ET. We denote this sum as λCu-P. Naturally, the difference in λT measured by RDE and λCu-P measured from CA will yield the solvent contribution, λH2O (λH2O = λT – λCu-P). From this analysis (Fig. 6E, S17−S22), the average λCu-P of 0.36 eV is obtained for 4SCC and 0.27 eV for 3SCC (Fig. 6F, Table 2), which yields λH2O of 0.07 eV for 3SCC and 0.53 eV for 4SCC. These results imply that not only does the 4SCC have a higher λT than 3SCC, but a significant solvent contribution, λH2O of 0.53 eV, is present in 4SCC but not in 3SCC.

λ comparison between ArCuPs and Cu proteins

In classical Cu complexes, λ is high, ~1.88-2.4 eV66. However, the protein fold significantly lowers λ, as seen in various type-1 Cu sites (Table 2)60,67,68,69,70,71,72,73. A flexible coordination produces a higher λ than a constrained site in Az mutants, as exemplified by C112D-Az (type-2) or C112D/M121L-Az (type-0) Cu sites74,75. An exception is Cu-Aβ16, which has a low λ of 0.3 eV76, similar to 3SCC, despite being unstructured. The low λ has been ascribed to a preorganized ET process that minimizes structural changes. In select cases of synthetic Cu complexes, λ is ~0.9577 – 1.1 eV78, with λCu of 0.4 eV in one example78, comparable to 4SCC.

Correlating structural differences to ET and catalysis

A less saturated Cu(His)3 coordination of 3SCC is expected to lower λCu-P than a more saturated 4SCC, which is observed here. However, to rationalize the presence of a significant solvent reorganization in 4SCC, one must look beyond the primary-sphere. Indeed, differences exist between the second and outer coordination-sphere interactions. Specifically, the His9 residues in 4SCC participate in second-sphere H-bonding with Glu8 from a neighboring helix via the available Nδ atoms (Fig. 7A). No such interaction is observed in 3SCC (Fig. 7B), even though both peptides have the -GluHis- motif in their heptads. Twelve H2O molecules constitute an extended outer-sphere H-bonding network in 4SCC involving Glu8 (Fig. 7A). These structural differences alter the physical properties of the ArCuPs, which are translated to their function.

Fig. 7: Structural analyses of the ArCuPs and pMMO CuB site showing the Cu site, 2nd-sphere, and outer-sphere interactions.
figure 7

X-ray structures of 4SCC (A: PDB - 8VHS) and 3SCC (B: PDB - 7L33) show that in 4SCC, the non-coordinating Nδ atoms of His9 make second-sphere H-bond with Glu8 from a neighboring helix, which is involved in extended H2O-mediated H-bonding. The corresponding His residues in 3SCC do not make this H-bond despite having the EH sequence in the heptad. C X-ray structure of K6E/E8K-4SCC (PDB - 9BQR) shows that when Lys and Glu at the 6th and 8th positions are interchanged in the first heptad, the second-sphere H-bond between the His9 and Glu6 plus the H2O-mediated H-bonds are no longer present, which lowers the λH2O. D The CuB site of pMMO (PDB - 7S4H) and its surrounding environment, showing the H-bonding network involving amino acids and solvent, are present. Figures are generated in CCP4mg. The Cu is shown as an orange sphere, and its electron density in A–C is drawn as a red mesh, with the remaining atom density being blue. In these ball-and-stick representations, carbon is green, oxygen is red, and nitrogen is blue. Waters are shown as small red spheres.

Full size image

First, the second-sphere H-bonding will increase the electron density on the imidazole ring, making the CuII state more stable than the CuI state. Second, as seen here, the extended H2O-mediated H-bonding network will increase the outer-sphere contribution to λT. Therefore, the combined primary, secondary, and outer-sphere H-bonding involving H2O molecules add to the higher λT in 4SCC, with the major contribution coming from the solvent reorganization. This observation is coherent with the hybrid quantum mechanics/molecular mechanics (QM/MM) simulation on Az, where the solvent reorganization around His117 is proposed to contribute ~80% of the total λ79. The coordination environment of 3SCC and the absence of second-sphere and outer-sphere water-mediated H-bond lowers λT, with nearly all the contribution coming from the active site alone. In 3SCC, a single Glu8-H2O H-bond is present (PDB 7L33), but since Glu8 does not make an H-bond with primary-sphere ligand His9, the effect of outer-sphere solvent reorganization is expected to be negligible, as seen here. The slower reaction kinetics of 4SCC with O2 or H2O2 and its lack of reactivity for C-H oxidation can be attributed to a significant solvent reorganization barrier compared to 3SCC.

One possible factor for the unreactivity of the pMMO CuB site, even to polar solutes such as nitrite, could be its high reorganization energy. Interestingly, several second-sphere and outer-sphere H-bonds involving amino acid residues and solvent molecules are present around the CuB site (Fig. 7D)27. Although λ determination for the CuB site will be nearly impossible, given the presence of multiple mononuclear Cu sites in pMMO, these extended outer-sphere interactions are expected to increase the reorganization energy barrier for this site, as we see in 4SCC. The poor reactivity of Cu(His)3NH2 or Cu(His)4 sites in a protein or polypeptide framework is intriguing, as several synthetic CuN4 complexes with tripodal tetradentate ligands activate O2, H2O2, and C-H bonds29,30,31,32. One crucial factor governing this difference in reactivity is likely higher outer-sphere reorganization energy in biomolecular systems than in synthetic systems, which lack ligand and solvent-mediated extended H-bonding networks.

Deleting H-bonding interactions modulates λH2O and enables C-H oxidation

An important question in metalloprotein design is, “Is it possible to selectively alter reorganization energy by changing second- and outer-sphere H-bonding but keeping the primary coordination intact?” To address this, we designed a mutant to delete the Glu-His H-bond of 4SCC by replacing the Glu8 (g site heptad) with Lys. Consequently, the Lys6 (e site heptad) was changed to Glu to maintain the charge neutrality and e-g ion pairs. The E°’ for the mutant (Fig. S23) is similar to 4SCC and other type-2 Cu sites. From RDE and CA (Fig. S24–26), the λT and λCu-P determined for K6E/E8K-4SCC are 0.69 eV and 0.36 eV, respectively (Fig. 6D, F; Table 2). This result shows that although the λCu-P is identical to the parent 4SCC, the λH2O is significantly reduced to 0.33 eV compared to 0.53 eV in 4SCC.

To investigate if the experimental λ values correlate to structure, we determined the X-ray structure of the double mutant (Fig. 7C). The Cu site is nearly identical to 4SCC (Fig. S27), although some disorder in the N-terminal region of chain D is present in the mutant. The EPR spectrum of the mutant lacks the Cu-14N(His) superhyperfine splittings in both the perpendicular and parallel regions (Fig. S28) due to more flexibility caused by the deletion of the H-bond, agreeing with this disorder of Chain D. The relative orientations of the His residues are slightly different than 4SCC, causing minor variations in the ∠Nε-Cu-Nε and His Cε angles (Fig. S1). As designed, the second-sphere His-Glu H-bond is eliminated. This, in turn, has removed the H2O-mediated outer-sphere H-bonds involving the Glu residue as it is now 3-residue away from the His. The only H2O-mediated H-bond is seen between the axial Cu-bound H2O and a second H2O nearby (Fig. 7C). Since the Cu site in 4SCC is located within the hydrophobic core, the deletion of second-sphere H-bonding does not alter the Cu(His)4 site, leading to a similar λCu-P for both 4SCC and the mutant. However, deletion of the second-sphere H-bond removes the outer-sphere H2O-mediated H-bonding besides the axial H2O---H2O H-bond, leading to a 0.2 eV lower λH2O in the mutant than 4SCC. The axial H2O-mediated H-bond also contributes to a higher λH2O in the mutant than in 3SCC, which lacks any H2O-mediated H-bond. These findings demonstrate that protein design makes it possible to modulate λH2O via selective modification of second- and outer-sphere H-bonding.

The absorption features of the mutant produced with H2O2 and O2 (Fig. S28) are more pronounced than 4SCC, suggesting a more reactive nature of the former. In addition, there are also differences in band positions in the mutant, attesting to their reactivity differences. Next, we probed whether deleting the extended hydrogen-bonding interactions would make the double mutant active toward C-H oxidation. Unlike 4SCC, K6E/E8K-4SCC produces a significant catalytic current for benzyl alcohol peroxidation (Fig. S3) with H2O2. As a result, an appreciable amount of charge is observed in the controlled potential electrolysis (CPE) experiment, leading to the detection of benzaldehyde in the GC (Fig. 8). The kcat, KM, and kuncat for K6E/E8K-4SCC were determined to be 1.31 s-1, 37.3 mM, and 0.07 s-1, respectively, producing a catalytic proficiency of 500 M−1. This value is still lower than that of 3SCC (1200 M−1), suggesting that differences in the primary coordination sphere remain a major contributor to activity. However, with the same CuN4 primary coordination, K6E/E8K-4SCC is catalytically active, while 4SCC is not. We attribute this to a high solvent barrier caused by the extended H-bonding network of water molecules around the protein, mediated by the His—Glu H-bond. When this H-bond is removed in K6E/E8K-4SCC, the surrounding water-mediated H-bonding network is disrupted, lowering its λH2O and restoring its peroxidation activity.

Fig. 8: Benzyl alcohol to benzaldehyde peroxidation reactivity of the ArCuPs.
figure 8

GC traces for benzyl alcohol oxidation to benzaldehyde after 1 h of CPE experiment. Benzaldehyde was observed with 3SCC (red trace) and in the double mutant K6E/E8K-4SCC (orange trace), but no quantifiable product was observed with 4SCC (blue trace). After CPE, the product was extracted in DCM, dried over Na2SO4, and analyzed by GC.

Full size image

These results demonstrate that protein design allows for controlled tuning of outer-sphere solvent reorganization energies while maintaining the primary coordination and enables switching on or off the activity of a metalloprotein by modulating H-bonding interactions.

In summary, an Ile/Ile to Leu/Ile switch of the hydrophobic layers enables the modulation of the oligomerization behavior in designing the ArCuPs, and structural and spectroscopic evidence supports the design. The 3SCC is easier to reduce and reoxidize than the 4SCC. Additionally, the 3SCC activates peroxide and oxidizes/peroxygenates C-H substrates, whereas the 4SCC does not. These chemical properties of 4SCC resemble the unreactivity of the pMMO CuB site. CuI-3SCC is more rapidly reoxidized with H2O2 than with O2, similar to the peroxygenase reactivity of LPMOs in the absence of a substrate. The reaction profile of H2O2 and CuI-3SCC is complex, starting with homolytic O-O bond cleavage, leading to the CuII-oxygen species. Conversely, the O2 reaction produces features consistent with a Cu-hydroperoxo species via an outer-sphere ET process.

The reactivity differences between 3SCC and 4SCC can be explained by inner-sphere and outer-sphere reorganization energies coupled with structural differences. A less-coordinatively saturated 3SCC accounts for its lower λCu-P than 4SCC. However, 4SCC has a significantly higher solvent contribution due to an extended H2O-mediated H-bond network around the second-sphere Glu8, which is H-bonded to the primary-sphere His. These interactions increase the structural rigidity of 4SCC, leading to high reorganization energy and making it less reactive than 3SCC. Removing the second-sphere His9-Glu8 H-bond eliminates the extended H2O-mediated H-bonding network, reducing the outer-sphere solvent contribution to the reorganization energy while maintaining the primary coordination. Consequently, the lower solvent reorganization barrier converts an inactive metalloprotein 4SCC into an active peroxidation catalyst. This activity switching by protein design to control outer-sphere solvent rearrangement is an exciting discovery. Our findings offer a new dimension to modulating the function, which we believe will be broadly applicable in metalloenzyme design.

Methods

General

All chemicals and reagents were used as received. Buffers were prepared in Milli-Q (Millipore) water, treated with Chelex-100 overnight, pH adjusted as necessary, and filtered. Tris(hydroxymethyl) aminomethane (Tris), 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 2-(N-morpholino) ethanesulfonic acid (MES), sodium acetate, NaCl, and HCl were purchased from Fisher Scientific. The mixed buffers contained 20 mM each of MES, sodium acetate, HEPES, and Tris at the desired pH. CuCl2 (Alfa Aesar), NaOH (VWR), and H2O2 (VWR) stock solutions were made in water, and ascorbic acid (VWR) stock was made in the buffer.

Peptide synthesis and purification

Peptides were synthesized at 0.25 mmol scale on a solid phase peptide synthesizer (CEM) using standard Fmoc chemistry with acetylated N-terminal and amidated C-terminal45. Cleavage from the resin and side chain deprotection was performed using the cocktail consisting of 92.5% trifluoroacetic acid (TFA: Advanced Chemtech), 0.25% triisopropylsilane (Sigma), 0.25% ethanedithiol (Sigma–Aldrich), and 0.25% water for 2 h. After filtration, TFA was evaporated, and peptides were precipitated using cold ether. Purification was done by reverse-phase HPLC (1260 Infinity II, Agilent) using a C18 semiprep column with water and acetonitrile solvent system supplemented with 0.1% TFA. Peptide mass was confirmed by MALDI mass spectrometry using the sinapinic acid matrix (Sigma).

X-ray crystallography

Solutions for CuII-4SCC were prepared by adding one equiv of CuCl2 to 45 mg/mL apo peptide in basic water. K6E/E8K-4SCC was dissolved at 35 mg/mL in 20 mM Tris pH 8.5, followed by metallation. Crystals were grown using the hanging drop method at ambient temperature. Two μL of CuII-4SCC solution was mixed with an equal volume of the well solution containing 0.1 M Tris pH 8.5 and 2 M ammonium sulfate. For CuII-K6E/E8K-4SCC, two μL of this solution were mixed with two μL of the well solution containing 0.1 M Tris pH 8.5, 0.2 M MgCl2.6H2O, and 28% PEG 4000. The well volume was 1 mL. Crystals appeared within 7-20 days. The crystals were harvested and soaked with 30% PEG 400 as a cryoprotectant before flash-freezing in liquid N2.

X-ray diffraction data were collected at the Life Sciences Collaborative Access Team (LS-CAT) of the Argonne National Laboratory, the Stanford Synchrotron Radiation Lightsource, and the FMX beamline at NSLS-II. Diffraction images were indexed and integrated in XDS80, and scaled in Aimless81. Data quality assessment and twinning analysis were performed in phenix.xtriage82. The 4SCC structure was solved using Phaser83 in CCP4i84 by molecular replacement using 3R4A85 as the search model, followed by automatic model building in Buccaneer86. Refinement was performed using phenix.refine87 with iterative model building and density fitting analysis in Coot88. The K6E/E8K-4SCC was solved similarly, using 4SCC as the search model, and refined in Refmac589. Structure validation was performed using the wwPDB OneDep System. The structure factors and coordinate files have been deposited with PDB code 8VHS for 4SCC and 9BQR for K6E/E8K-4SCC. Data collection statistics are in Table S1. 2mFo-DFc and mFo-DFc maps are in Fig. S29−30. Electron density maps shown in Fig. 7 were drawn in CCP4mg 90.

UV-vis spectroscopy

The spectra were recorded on an Agilent 8454 or Cary 5000 spectrophotometer. All samples were prepared in 10 mM MES buffer at pH 6.5. All solutions were degassed individually before mixing and starting the experiments. For the kinetics experiments, 0.125 mM ArCuP solutions and 12.5 mM H2O2 were mixed in a septa-capped anaerobic cuvette (Starna Cells) followed by monitoring the kinetics for 120 min with 1 min intervals in-between scans. CuI-peptides were prepared by adding one equivalent of Asc inside a glove box (MBraun) and incubating for 15 min. UV-vis was checked to ensure complete reduction. For kinetics with O2, 0.25 mM Cu-peptides were mixed with an equal volume of an O2-saturated buffer, and the kinetics started immediately. An O2-filled balloon was used to maintain an O2-saturated environment. Initial rates were calculated from the first 3 min of the kinetic run.

Fluorescence spectroscopy

Kinetics of CuII reduction with Asc and CuI reoxidation experiments with H2O2 were performed anaerobically by monitoring the Trp emission of A10W variants at pH 6.5 (10 mM MES) in septa-capped fluorescence cuvettes using a PTI fluorimeter. For reduction, 0.125 mM CuII-peptides were mixed with different equiv of Asc followed by continuous spectral monitoring. Reoxidation experiments were performed with 0.125 mM CuI-peptides (prepared by adding one equiv Asc to CuII-peptides) and an increasing concentration of H2O2. Each run was monitored for 120–180 min or until saturation was obtained, which depended on H2O2 concentration. In these experiments, the Trp excitation wavelength of 280 nm was used. An emission wavelength of 340 nm for A10W-3SCC and 345 nm for A10W-4SCC was selected based on the highest emission intensity. The final sample volume in each experiment was 800 μL.

Analysis of reduction and reoxidation kinetics

For reduction kinetics, the normalized fluorescence vs. time plots were fitted with a single-exponential function to extract the first-order rate constants at varying Asc concentrations. These rate constants were plotted against Asc concentration and fitted to a linear equation. The slopes of the linear fits represent the second-order rate constants with respect to Asc. A similar procedure was used for the reoxidation kinetics, except that the second-order rate constants were derived against varying H2O2 concentrations. This analysis was not performed for O2 because, in O2-saturated solutions, the O2 concentration was assumed to be constant.

EPR spectroscopy

EPR data were collected at 77 K on a Bruker X-band (9.4168 GHz) EMXPlus Continuous Wave spectrometer at the School of Chemical Sciences EPR lab at the University of Illinois. Samples for EPR contained 0.5 mM ArCuPs in 100 mM MES pH 6.5 plus 25% glycerol as a glassing agent. A power of 10.02 mW was used with an attenuation of 13 dB, receiver gain of 30 dB, a modulation amplitude of 10 G, and a modulation frequency of 100 kHz. All instrument parameters were kept constant for each set of experiments. For reoxidation, a 100-fold H2O2 was added, and at different time intervals, samples were aliquoted to glycerol-containing EPR tubes and flash-frozen in liquid N2 outside the glove box. For reoxidation experiments with O2, one mM CuI-peptides was mixed with an equal volume of O2-saturated buffer, and the reaction was allowed to proceed under an O2 atmosphere outside the box with stirring. At different time intervals, samples were flash-frozen. Baseline correction of the raw data was done in Origin 2023, version 10.0.0.154. The spin concentration of peptide samples was calculated from double spectral integration. Due to extensive superhyperfine features, reliable spin integrations were not obtained for 4SCC.

EPR simulations were done in EasySpin 6.0.091 using the MATLAB (R2024a) interface. An axial g tensor was used for Cu with the appropriate number of 14N nuclei (3 for trimer and 4 for tetramer). The hyperfine tensor Aframe was invoked to transfer the molecular frame to the A tensor. A Lorentzian line broadening was used. All simulations were done using the method suitable for frozen solutions (pepper).

Electrochemistry

All electrochemistry experiments were performed using a Wave Driver 20 bi-potentiostat (Pine Research) with PGE as the working electrode, a coiled platinum wire as the counter electrode, and Ag/AgCl (in saturated KCl) as the reference electrode. The potential values were referenced against Ag/AgCl and converted to SHE. Temperature-dependent experiments were done in an RRDE cell with a water jacket by controlling the temperature from a circulating water bath.

i. λT determination

The RDE experiments were used to determine the total reorganization energy, λT. All RDE experiments were performed with 0.25 mM ArCuPs dissolved in 80 mM mixed buffer, pH 6.5, and 100 mM KCl as a supporting electrolyte under anaerobic conditions. RDE experiments (Fig. 6B, S15, S16, S24) show an increase in the current with an increase in rotation rates following the Koutecký-Levich (K-L) relationship, Eq. (ii)92, with iL and iK representing the Levich current and kinetic currents, respectively. Here, n is the number of electrons transferred during the redox process, F is Faraday’s constant 96485 C mol-1, A = area of the electrode (0.19625 cm2), C0 is the concentration of ArCuP (0.25 mM), and Γ is the surface coverage of the electrode (88 ± 8 ×10−12 mol cm−2). kET is the ET rate constant (kET in M−1s−1). D0 is the diffusion coefficient of the analyte, ω is the angular velocity of the electrode, and ν is the kinematic viscosity of the solution.

$$\frac{1}{i}=\frac{1}{{i}_{L}}+\frac{1}{{i}_{K}}=\frac{1}{0.62{nF}{{AC}}_{0}{({D}_{0})}^{2/3}{{{\rm{\upsilon }}}}^{-1/6}{{{\rm{\omega }}}}^{1/2}}+\frac{1}{{nF}{A{C}_{0}k}_{{ET}} \Gamma }$$
(ii)

From the intercepts of the linear i−1 vs. ω−1/2 plots, the second-order rate constant for ET (kET) was determined given that in the absence of other competing chemical/redox phenomena, the only process occurring is the ET. The kET was obtained at five different potentials in the mass transport-limited region between −0.05 V and 0.4 V. The experiments were performed at three different temperatures, and for each temperature, kET was determined (Table S4, S5, S10). λT was obtained from the slopes of ln(kETT −1/2) vs. T−1 plots (Fig. 6C, S15, S16, S24) according to the Arrhenius Eq. (iii).

$$\,\lambda=-4.03\,\, d \, {{ln}}\left({{kT}}^{-\frac{1}{2}}\right)/d[{T}^{-1})$$
(iii)

ii. λCu-P determination

The CV and CA experiments were performed with ten μL of 0.25 mM ArCuPs drop-cast on PGE and six μL of 0.1 M polymyxin B to facilitate electrostatic interaction between the ArCuPs and the electrode. The protein solution was then dried under N2, and data were recorded in N2-saturated 20 mM mixed buffers. Each CA experiment was performed for a total duration of 0.3 s (Fig. S18). Temperature-dependent CV and CA experiments were performed at seven different temperatures between 280 K and 310 K with 5 K increments. CVs were recorded to extract E°’s (Fig. S17, S23; Table S6, S9) and determine the overpotentials (η) at each temperature. Six different η were chosen, three on the cathodic (−150 mV, −100 mV, −50 mV) and three on the anodic (50 mV, 100 mV, 150 mV) side of Eo’s at each temperature. At a given η, the current follows an exponential relationship with time according to Eq. (iv)

$$i=\,{k}_{{app}}\,Q \, {{\mathrm{exp}}}(-{k}_{{app}}t)$$
(iv)

where kapp is the apparent ET rate constant in s−1, Q is the total charge passed during CA, and t is the time. The decay current consists of the charging current and the Faradaic current. The capacitive current decayed within 1 ms for the blank electrode at pH 6.5. To avoid the charging current decaying completely from the Faradaic response, ln(i) vs. t plots (Fig. 6E, S19, S20, S25) were constructed beyond the decay time for the blank electrode. From the slope of these semilogarithmic graphs, the heterogeneous ET rates kapp (s−1) (Table S7, S8, S11) were determined at each temperature and η, which was used to calculate λ. Arrhenius plots of ln(kappT−1/2) vs. T−1 were made (Fig. S21, S22, S26) using kapp from above93. The λCu-P at each η was calculated from the slopes of these plots.