The ability of H4pep to coordinate copper ions was verified via different spectroscopic methods. Notably, the CD data demonstrate that H4pep-Cu2+ complexes adopt a beta-sheet conformation, as it is indeed observed in the MFS of laccase. Our data demonstrate that, despite their beta-sheet conformation, H4pep-Cu2+ complexes do not undergo aggregation or formation of fibrils. We also performed preliminary catalytic activity measurements to ascertain if these metallo-peptide complexes exhibit activity based on their parent laccase. Our assessment focuses on indications of positive catalytic activity with respect to O2 reduction, serving as proof-of-concept for the validity of the proposed MeSA tool. To the best of our knowledge, this is the first example of a synthetic β-sheet metallo-peptide complex that is stable in solution and features catalytic activity. These findings demonstrate that the MetalSite-Analyzer tool can provide relevant indications for designing bioinspired catalytically active metallo-peptides.
We have created the MeSA tool with the scope of expanding the MFS concept to include sequence conservation analysis. Starting from an input PDB structure, MeSA allows the selection of user-defined mono-/multinuclear metal sites and, from the extracted MFS, runs a PSI-BLAST search of the metal-binding sequence fragments. This step consists of mapping and aligning the binding fragments to all related sequences contained in Uniprot, the world-leading protein sequence database. The output information allows for the analysis of the conservation of the residues in each specific position of the starting MFS sequence. This provides a basis for further rational design of the desired peptide mimics, by highlighting completely conserved residues, which are presumably strictly necessary for function, moderately variable positions, where one of two/three different amino acids can be selected (e.g., based on considerations of stability or ease of synthesis), and highly variable positions, where almost any amino acid can be introduced. The tool has been implemented as a user-friendly web server, requiring no registration (https://metalsite-analyzer.cerm.unifi.it/).
To assess our methodological approach, we have selected laccase as a model to design a minimal-length peptide ligand for binding copper ions. Laccases belong to the family of multicopper oxidases, which catalyze one-electron oxidation of different organic substrates, coupling it with the four-electron reduction of O to HO. In these enzymes, substrate oxidation occurs at a mononuclear copper site (type 1), followed by electron transfer to a trinuclear copper cluster (type 2/type 3), where oxygen reduction occurs. The latter is the site we have selected to test our bioinformatic tool.
Firstly, we accessed the MetalPDB database to gain information on the binding site(s) of the small laccase from Streptomyces viridosporus (PDB ID: 3tbc). In bacterial small laccases, so-called because they have fewer domains than fungal laccases, the trinuclear Cu sites are located at the interface between monomers, in a three-fold symmetry. As in a typical multicopper oxidase, they consist of eight histidine residues coordinating the metals, in a ligand-nonligand-ligand motif (LXL). In the apo-form of the enzyme, the monomer exposes the coordinating residues to the surrounding environment, making it potentially accessible to solvent and substrate molecules involved in catalysis. This makes the site particularly attractive to inspire the design of artificial biomimetic peptide catalysts. Feeding the pdb structure as input to MeSA and selecting the trinuclear copper site, the tool was able to extract four fragments contributing to the coordination of the Cu ions with two histidine residues each (Fig. 1A). The conservation of the residues in each fragment was analyzed via the PSI-BLAST algorithm. As a result, four alignments of metal-binding motifs were generated. Finally, a sequence profile was obtained for each fragment composing the starting site (Fig. 1B).
In our endeavor to obtain a minimal-length peptide mimicking the laccase site's activity, further rational design steps are needed to implement the information obtained from MeSA. On a closer look at the structure of the site, we have noticed that the binding His residues belong two by two to antiparallel β-sheets, located on adjacent monomers in a C2 symmetry with respect to the axis running across the Cu atoms (Fig. 1C). The four metal-binding fragments identified by our bioinformatic search thus represent the four β-strands composing this structural motif. In particular, two of the four fragments extracted by the bioinformatic search are considerably shorter (7-9 residues for fragment 1 of chain A and B versus 11-13 for fragment 2 of chain A and B, see Fig. 1B), thus more suited for our purpose of designing a minimal-length peptide. Hence, for the final sequence, the choice of the amino acid residues in each position was driven by the combination of the consensus sequences of these two fragments, with further modifications led by the rational analysis of the model site (see "Methods" for details). The final minimal-length peptide ligand is eight residues long and was named H4pep (sequence: HTVHYHGH).
H4pep was synthesized via SPPS and purified via reverse-phase high-pressure liquid chromatography (RP-HPLC). To verify the consistency of our design strategy, we preliminarily carried out UV-visible and NMR experiments to confirm the ability of H4pep to bind copper ions. Such experiments were performed at pH 5.6 to prevent copper oxide formation and N-terminal amine deprotonation (vide infra), still favoring imidazole deprotonation and binding (pKa ~6).
UV-visible spectra of H4pep display the characteristic peak of tyrosine at 276 nm (Fig. 2A). Upon gradual copper(II) addition, the characteristic band of Cu d-d transitions arises in the region between 500 and 800 nm. This feature is in good agreement with reported λ values for histidine coordination in peptide-copper complexes. Notably, with copper concentration up to 1 equivalent, this band could be fitted with two Gaussian functions, suggesting the contribution of multiple species to the absorbance profile (see "Stoichiometry of H4pep-Cu complexes" section for details). In excess of copper, the absorption peak of unbound copper(II), with maxima at 786 nm, becomes noticeable and increasingly contributes to the observed spectrum.
Binding of copper(I) was also tested to verify the ability of H4pep to bind copper also in the reduced oxidation state. The affinity of H4pep to Cu ions was determined via competitive titration with bicinchoninic acid (BCA, Fig. S1). The absorption changes at 562 nm, corresponding to the formation of [Cu(BCA)], could be fitted satisfactorily assuming a single class of binding sites with K = 2.4*10M. This affinity is comparable to the values found for a series of nitrite reductase mimics featuring a Cu(His) site bound to a triple-stranded α-helical coiled-coil.
To further investigate the coordination of metal ions to H4pep, the peptide's structure was analyzed by NMR spectroscopy under different experimental conditions. In aqueous solution, the apo-peptide showed only four very broad amide signals of the nine expected (seven backbone and two C-terminal amide protons) due to fast exchange with the solvent, indicating the absence of stable hydrogen bonds and secondary structure elements (Figs. S2A and S3A). In methanol, while amide resonances became observable, their chemical shift dispersion remained limited, suggesting lack of defined tertiary structure in the apo-form (Figs. S3B and S4A). The use of Zn as a diamagnetic probe in NMR studies is a strategy used to gain preliminary structural insights when the paramagnetic nature of Cu precludes direct NMR analysis, including in the particularly demanding pentacoordinated geometry found in LPMOs. In such cases, Zn allows for the identification of potential binding sites and ligand interactions without the complications associated with paramagnetic broadening. The addition of Zn in a 1:1 ratio induced selective changes in the histidine resonances, with downfield shifts (∼0.1 ppm) of the proton signals belonging to the Cδ and Cε of the imidazole ring (Fig. S2B). The broadening of these resonances suggests a chemical exchange process, widely observed in NMR analysis, which primarily involves the histidine residues. This spectral behavior suggests a dynamic equilibrium between multiple zinc-bound species, where either one or two Zn²⁺ ions are alternately coordinated to the two different available binding sites in the peptide scaffold. More dramatic spectral changes were observed upon addition of Cu (at H4pep:Cu ratios of 1:1 and 2:1) under anaerobic conditions, both in aqueous buffer (Figs. S2C and S5) and methanol (Figs. S4B and S6). All resonances significantly broadened, suggesting that Cu binding affects the overall peptide conformation more extensively than Zn, leading to multiple conformational states in exchange. The spectral broadening persisted in the presence of sodium dithionite, excluding contributions from paramagnetic Cu species. From the above results, we can conclude that H4pep binds both Zn and Cu ions in solution; additionally, binding of Cu was demonstrated by the UV-visible spectra (Fig. 2A). Furthermore, the NMR data of H4pep indicate that Zn and Cu binding occurs via the histidine side chains.
As anticipated by NMR, CD spectroscopy of H4pep in the far-UV range (190-250 nm) reveals only a deep band at 198 nm indicative of a random coil for the apo-peptide in buffer solution (Fig. 2B). Based on our peptide design strategy, we can reasonably anticipate that binding of Cu to H4pep will induce the formation of a β-sheet structural motif. Thus, we expect to observe spectral changes characteristic of β-sheet conformation when H4pep is titrated with Cu. In line with our design and the previous NMR analysis, we expect metal binding to occur via histidine side chain coordination rather than backbone amide coordination, previously reported for 3-residue-Cu systems. These modes of coordination can be differentiated by their pH dependence: histidine coordination is favored at lower pH due to the side chain imidazole group's pKa (~6), while backbone amide coordination typically predominates at higher pH (pKa ~7-8). Therefore, CD spectra recorded at different pH values allowed us to test Cu-binding ability under varying conditions. H4pep was dissolved in acetate buffer solutions at pH 4.4, 4.8, 5.2, and 5.6, respectively (Fig. S7). Higher pH values are excluded to avoid the formation of copper oxide species, limiting Cu availability to the peptide ligand, and to mitigate terminal amine binding. No significant changes were observed in the CD spectra at pH 4.4 and 4.8 in presence of Cu, indicating unfavorable binding conditions due to the protonated state of the histidine ligands. In contrast, at pH 5.2 and 5.6, a clear change in conformation is visible. Therefore, pH 5.6 buffer conditions were selected to record the titration spectra in Fig. 2B. Upon Cu addition, the CD spectrum displays a negative band at 227 nm and a positive one at 210 nm. This suggests that Cu binding to H4pep induces a β-sheet conformation, consistent with our initial design. These features saturate upon the addition of 0.5 Cu equivalents and remain essentially unchanged until a 1:1 H4pep:Cu ratio. As additional copper equivalents are introduced, the spectrum undergoes further changes, with the positive feature increasing and shifting to 205 nm. This suggests the formation of another species, favored at high copper concentrations.
The same trend was observed in the visible range of the CD spectra, in which changes in the coordination environment around the metal ion(s) can be monitored (Fig. S8A). Titration of up to 1 equivalent of Cu results in the gradual increase of a positive peak centered at 719 nm. At higher Cu equivalents, the peak position shifts to the red, suggesting the formation of a second H4pep-Cu species.
Titration of Cu to H4pep results in similar changes in the UV region of the CD spectrum (Fig. S8B). There are no observable features in the visible region of the same CD spectrum, consistent with the lack of d-d transitions in Cu complexes. This confirms that the observed changes in the UV region of the CD spectrum are associated with the coordination of Cu ions to H4pep to form H4pep-Cu complexes.
EPR spectroscopy was employed to further analyze copper(II) binding and to elucidate the presence of different H4pep-Cu species. Titration of H4pep into a Cu solution revealed the presence of at least two distinct EPR active species (Fig. 2C). Under a large excess of peptide (e.g., H4pep:Cu ratio 1:0.3), a single axial signal is observed, with values of g, g, and g, respectively, at 2.0419, 2.0737, and 2.2447 (Fig. S9). At higher Cu equivalents (between H4pep:Cu ratios 1:0.4 to 1:1) an additional axial signal characterized by lower g values could be detected. Presumably, the coordination of an additional Cu ion into the peptide can cause a small distortion to the local environment of the first Cu, resulting in a shifted EPR spectrum, which could be isolated by spectral deconvolution (Fig. S10). In this configuration, the two Cu binding sites are essentially identical, as only one axial Cu signal is observed. The presence of two distinct EPR signals at different H4pep:Cu ratios is consistent with CD spectroscopy results, indicating that two H4pep-Cu species are formed. Furthermore, above H4pep:Cu ratio of 1:1, an EPR signal associated with unbound Cu becomes observable.
The CD and EPR spectra obtained from titrations between Cu and H4pep were deconvoluted by fitting each spectrum with its respective isolated EPR signals or CD spectra and presented in Fig. 2D. The observation of unbound Cu species in the EPR spectra above H4pep:Cu ratio of 1:1, suggests that each peptide can bind a maximum of 1 Cu equivalent. The formation of a β-sheet conformation upon Cu binding, as evident from the CD spectra, indicates that at least two peptide units are required to form a Cu(H4pep) complex (where y ≥ 2). It is unlikely that the minimum number of peptides to form a Cu(H4pep) is three, since we would expect most of the H4pep to exist in a β-sheet conformation at H4pep:Cu ratio of 1:0.33. On the contrary, our experimental observation led to a fitted molar fraction of peptide in random coil conformation of around 0.5 at 1:0.33 H4pep:Cu ratio (Fig. 2D). Therefore, our hypothesis is that a Cu(Hpep) complex is formed upon addition of Cu ions to H4pep. In this scenario, there are two possible species corresponding to Cu(Hpep) and Cu(Hpep), designated as 1Cu2Pep and 2Cu2Pep, respectively. Deconvolution of UV-visible d-d band spectra could be obtained accordingly (Fig. S11). Fitting of the spectra up to 1 Cu equivalents was possible considering two Gaussian components centered at 636 and 742 nm, likely due to the formation of Cu(Hpep) complexes. At higher Cu concentrations above 1 equivalent, a third component, centered at 786 nm and arising from free Cu in solution, needed to be included for accurate fitting.
Observing the resulting trends, it is clear how the isolation of a single Cu(H4pep) species is far from trivial. The 1Cu2Pep species seems to be favored at low Cu concentrations, up to 0.5 Cu equivalents, while both 1Cu2Pep and 2Cu2Pep species are present in similar quantities up to 1 Cu equivalent. The 2Cu2Pep species is favored at higher Cu concentrations, but an increasing amount of free Cu ions is detected in these conditions. Notably, even in large excess of Cu, dynamic light scattering (DLS) experiments revealed the absence of large aggregates, confirming the molecular nature of the complexes (Fig. S12).
To examine if 1Cu2Pep and 2Cu2Pep exhibit catalytic O reduction activity, as in the original laccase, a preliminary electrochemical characterization of these two complexes by cyclic voltammetry was performed (Figs. 3 and S13A). Cyclic voltammograms of 2Cu2Pep under N atmosphere showed a reduction peak at -0.11 V against a Ag/AgCl reference electrode and a corresponding re-oxidation wave at 0.52 V. The large peak-to-peak separation suggests a slow heterogeneous electron transfer rate between the glassy carbon working electrode and the complex. Notably, during the anodic scan, a sharp reoxidation peak at 0.08 V is observed, characteristic of oxidation of metallic Cu deposited on the working electrode. This suggests that during the reduction of 2Cu2Pep, the copper binding affinity of at least one of the binding sites becomes weaker, resulting in the loss of copper from the complex. Since metallic Cu is also known to participate in electrocatalytic O reduction, we will not examine the O reduction activity of 2Cu2Pep as it is not trivial to verify its catalytic O reduction activity in presence of metallic Cu.
On the other hand, cyclic voltammograms of 1Cu2Pep show a reduction peak at -0.16 V and the corresponding reoxidation peak at 0.36 V under N atmosphere. Importantly, features associated with the deposition of metallic Cu and its subsequent reoxidation are not observable. This is further verified by rinse test experiments performed after cyclic voltammetry of 1Cu2Pep (Fig. S13B). Under O atmosphere, a current enhancement is observable at around -0.20 V, suggesting catalytic O reduction activity (Fig. 3A). To verify this hypothesis, bulk electrolysis experiments were performed in a custom-designed cell that incorporates a Clark electrode for sensing of O concentration. Chronoamperometry experiments at -0.20 V result in a steady-state current of approximately -15 µA with a concomitant decrease in O concentration (Fig. 3B), confirming that the catalytic current is due to O reduction. At present, it is difficult to estimate a meaningful faradaic efficiency parameter, as the current data cannot distinguish between a two-electron reduction of O to HO or a four-electron reduction to water. As such, we present two scenarios to estimate faradaic efficiencies corresponding to two- or four-electron reduction of O in Fig. 3B. In the four-electron reduction scenario, the faradaic efficiency approaches unity, while it is halved for the two-electron reduction scenario. We note that additional scenarios where mixtures of HO and HO products are also highly likely.
Cyclic voltammetry of the 1Cu2Pep solution after bulk electrolysis does not show significant deviation from that recorded prior to bulk electrolysis, with no indication of Cu plated on the electrode surface (Fig. S13C). Furthermore, additional rinse test experiments show a reductive current comparable to the background during chronoamperometry (Fig. S13D). To further rule out the participation of unbound copper species during catalytic O reduction, negative control experiments were performed with dilute solutions of unbound Cu (0.03 mM, 7% of Cu content in 1Cu2Pep). Chronoamperometry of 1Cu2Pep shows a higher current compared to 7% Cu (Fig. S14A), indicating that free Cu ions only contribute minimally, if at all, to the catalytic activity observed for 1Cu2Pep. Moreover, rinse test cyclic voltammogram of the unbound Cu solution shows significant reoxidation of metallic copper species at 0.09 V, not observed for the 1Cu2Pep species (Fig. S14B). Therefore, the O reduction activity in the conditions above is likely to originate from 1Cu2Pep.
We have then tested the performances of 1Cu2Pep at a more reducing potential (-0.30 V). In this case, an interesting behavior of the chronoamperometry signal was observed: while the recorded current decreases in the first 120 s, it then reaches a plateau and starts increasing after 150 s (Fig. S15A). No signals of unbound Cu could be detected in the cyclic voltammogram and rinse test performed after the measurement (Fig. S15B). This result seems to indicate that a more active species of the H4pep-Cu assembly can be formed in these conditions.
For preliminary assessment of O reduction kinetics of 1Cu2Pep and 2Cu2Pep, we monitored the oxidation rate of p-phenylenediamine (PPD), commonly used in laccase activity assay, by spectrophotometry. In a 1 mM PPD and 10 µM 1:1 Cu:H4pep complex reaction mixture, the apparent reaction rate is 82.8 ± 36.6 nM min. In the discussion below, we note that kinetic parameters derived from the use of typical laccase activity may not be representative of intrinsic reactivity.