Minimal Heterochiral de Novo Designed 4Fe–4S Binding Peptide Capable of Robust Electron Transfer

Abstract

Ambidoxin is a designed, minimal dodecapeptide consisting of alternating L and D amino acids that binds a 4Fe–4S cluster through ligand–metal interactions and an extensive network of second-shell hydrogen bonds. The peptide can withstand hundreds of oxidation–reduction cycles at room temperature. Ambidoxin suggests how simple, prebiotic peptides may have achieved robust redox catalysis on the early Earth.

The 4Fe–4S cubane cluster is a ubiquitous inorganic cofactor mediating a wide variety of electron transfer reactions across the tree of life.¹ The structurally conserved ferredoxin fold binds a 4Fe–4S cubane cluster with four cysteine residues in a common sequence motif (Cys-X₁-X₂-Cys-X₃-X₄-Cys-(X)_n-Cys).^2–4 The ferredoxin α/β topology is one of a small number of folds proposed to have given rise to myriad proteins.⁵ The origin and evolution of the ferredoxin fold family was likely a key evolutionary milestone in origins of biological electron transfer.⁶ Although the structures of ferredoxins are well studied, it has proven challenging to design simpler, artificial peptides that stably bind iron–sulfur complexes and are capable of reversible electron transfer reactions. De novo designs based on intrinsic symmetry of iron–sulfur clusters have produced artificial folds that bind metal and exhibit reversible redox stability for a few dozen cycles.^7–9 Here, we present a minimal dodecapeptide ferredoxin, consisting of alternating L and D amino acid residues, comprised only of cysteine with arginine or lysine. This peptide binds an 4Fe–4S cluster through ligand–metal interactions and an extensive network of second-shell hydrogen bonds. The resulting “ambidoxin” peptide has robust redox stability; the holo-peptide can undergo hundreds of oxidation–reduction cycles at room temperature.

Analyses of backbone dihedral angles of iron–sulfur cluster binding heptapeptide (CXXCXXC) fragments from non-redundant structures in the Protein Data Bank (Figure 1a,b) revealed a five-residue alternating alpha-left and alpha-right structural motif. This motif is similar to the alpha-pleated sheet postulated by Pauling,¹⁰ and nest motifs observed by Watson and Milner-White.^11,12 Extending this pattern to 12 residues produces a unique curved backbone that encircles the iron sulfur cluster (Figure 1c,d, S1). This unusual conformation could be enforced with a sequence of alternating L and D amino acids that stabilize alpha-right and alpha-left backbone conformations respectively (Figure S1g). Molecular dynamics simulations, starting from geometrically and energetically idealized conformations, converged on a unique stable structure that closely aligns with natural ferredoxin (Figure 1e, S1). Because of the equal contributions of L and D amino acids in the design, we refer to this class of molecules as “ambidoxins”.

Figure 1.

Design of ambidoxin. (a) Bacterial ferredoxin has a conserved CXXCXXC motif near iron sulfur cluster structures. (b) Ramachandran plot of the X₁-X₂-C-X₃-X₄ sequence motif surrounding the central cysteine reveals conserved alpha-right conformation (R) in X₁ and X₄ positions (green) and alpha-left conformation (L) in X₂ an X₃ positions (orange). (c) Ferredoxin loop was built by extending the original XXCXX peptide (α_Rα_Lα_Rα_Lα_R conformation) with its mirror image (α_Lα_Rα_Lα_Rα_L conformation), maintaining an alternating right/left conformation throughout. (d) Sequences and computationally determined structural models of ambidoxin peptides. (e) Portion of ambidoxin model and natural ferredoxin structures are super-imposable.

Given the stability of the 4Fe–4S cubane cluster in the absence of oxygen,^13,14 we expected that reconstitution of peptides in the presence of iron and sulfur salts would assemble as a cluster fully coordinated by a single peptide. To test this, we synthesized three ambidoxin variants that included either basic or acidic amino acids at the positions between cysteines, and a tryptophan at the third position to allow quantification by absorption at 280 nm. Upon iron–sulfur cluster assembly, K and R ambidoxin peptides showed a 420 nm charge-transfer absorption band prior to reduction with sodium dithionite (Figure S6),¹⁵ and rhombic EPR signatures consistent with a bound 4Fe–4S cluster in the reduced state after treatment with sodium dithionite (Figure 2a, S8).^16–18 Holo-peptide yields of 80–94% were determined by spin counting experiments. Dipole–dipole interactions between clusters were not observed by EPR, supporting the presence of a monomeric holo-peptide. After reconstitution, holo-peptides were stable for days under anaerobic conditions at room temperature from pH 5 to 8 (Figure S9).

Figure 2.

Structural characterization of ambidoxin. (a) X-band EPR spectra of a dithionite-reduced [4Fe–4S]⁺ cluster reconstituted in K and R peptides in HEPES/pH 5.0 buffer, measured at 9.48 GHz and 10 K (blue traces). The simulated spectra (red dashed traces) used three rhombic g-factor values (g₁ = 2.0445, g₂ = 1.9365, g₃ = 1.903) as marked by vertical lines. (b) Cosine HYSCORE spectrum for [4Fe–4S]⁺ cluster with K peptide in deuterated HEPES buffer at pH 5.0, measured at magnetic field 340 mT (cp. the g₁ field position in panel a) and temperature 12 K. Cross-peaks from three nonexchangeable cysteine protons (H1–H3) were resolved. (c) 600 MHz ¹H NMR spectrum of a dithionite-reduced [4Fe–4S]⁺ cluster reconstituted in K peptide in 95% D₂O HEPES-d₁₈/pH 8.5 buffer measured at 278 K. Hyperfine shifted lines (labeled A–E) are from β-CH₂ protons of four cysteines ligating to paramagnetic [4Fe–4S]⁺ cluster. Weak resonances marked with asterisks are from unknown minor (<10%) species. (d) Temperature dependence of the chemical shifts: Lines A–C with Curie temperature dependence are from four β-CH₂ protons of two cysteine ligands bound to mixed-valence Fe^2.5+−Fe^2.5+ pair, and lines D–E with anti-Curie dependence are from two β-CH₂ protons of two cysteine ligands bound to ferrous Fe²⁺−Fe²⁺ pair.

We did not observe cluster binding to the negatively charged E peptide (Figure S7). Assembled 4Fe–4S–4(Cys) complexes would have a −2 or −3 charge depending on the oxidation state. It is plausible that the significant charge associated with seven surrounding glutamates electrostatically destabilizes the holo-peptide, preventing assembly.

To validate the modeled structure of the paramagnetic peptide-cluster complex, we examined proton hyperfine interactions within 4 Å of the cluster by pulsed-EPR methods. Orientation-selective ESEEM and HYSCORE experiments were performed at several magnetic field positions across the rhombic EPR line-shape (Figures S10, S12–S17).^19,20 Both techniques consistently identified several protons strongly coupled to an unpaired electron spin residing on the 4Fe–4S cluster. Experiments in deuterated buffer revealed that four protons (labeled H1–H4 in Figure 2b, S16–S17) were nonexchangeable and likely belong to β-protons of the four cysteine ligands. From the HYSCORE cross-peak analysis (Figure S18), we were able to extract isotropic and anisotropic hyperfine coupling constants of the four protons, including relative orientations of their anisotropic hyperfine tensors with respect to the g₁ axis of electron g-factor (Table S2). These hyperfine parameters provided both Fe–H distances and angles between Fe–H vectors and the g₁ axis, providing eight constraints that were consistent with models from molecular mechanics simulations of ambidoxin (Figure S19 and Table S2).

To resolve the structure and electronic state of the cluster at ambient temperature, we performed paramagnetic NMR experiments²¹ from 5 to 32 °C on holo-K peptide prepared to near-stoichiometric yield. At all temperatures, large chemical shifts were observed, with Curie and anti-Curie temperature dependencies indicative of strong hyperfine coupling between cysteine β-protons and mixed- or ferrous-valence iron, similar to those observed in natural protein environments²² (Figure 2c,d, Figure S20). Combining the HYSCORE-based analysis, the computational ambidoxin model, and the temperature dependencies of NMR derived chemical shifts, we assigned three contact-shifted ¹H resonances to D-Cys2 and L-Cys5 H^β atoms, ligating a mixed-valence Fe^2.5+ pair, and two other ¹H resonances (Figure 2c) to D-Cys8 and L-Cys11 H^β atoms, ligating a ferrous Fe²⁺ pair (Figures S19 and S20). The temperature-dependent shifts were fully reversible and persistent at both pH 5 and pH 8.5 (Figure S20). These results demonstrate a stable peptide-cluster coordination structure around the bound paramagnetic 4Fe–4S center.

To test redox turnover stability, we performed cyclic voltammetry on a pyrolytic graphite edge electrode under anoxic conditions.²³ The midpoint potential of 4Fe–4S bound to K and R peptides at p H 8.5 (Figure 3a) was approximately −450 mV, which is within the range for extant bacterial ferredoxins.²⁴ One would expect the extensive hydrogen bonding and solvent exposure to reduce the Fe–S bond covalency and produce a species with a less reductive potential than bacterial ferredoxin,²⁵ but electrostatic contributions from K and R ambidoxin side chains appear to compensate, leading to a more reductive midpoint potential. Measurement of peak currents at different scan rates confirmed that current was not diffusion limited, and was due to direct electron transfer between the working electrode a metallo-peptide film²⁶ (Figure 3b). The holo-peptide endures hundreds of redox turnovers at room temperature, decaying with a rate constant of around 0.2% per cycle (Figure 3c). This robust redox stability exceeds that of many previously designed artificial metallo-peptides by ten to a hundred-fold. As in natural iron–sulfur proteins such as ferredoxin and rubredoxin, ambidoxin redox stability likely originates from the extensive network of second-shell hydrogen bonds from backbone amides to cysteine thiols and cluster μ-sulfurs.^27,28

Figure 3.

Electrochemical characterization of ambidoxin. (a) Cyclic voltammogram of R peptide (blue trace) and K peptide (red trace) at room temperature, pH 8.5. (b) Peak faradaic current of R peptide measured at different scan rates. (c) Measured faradaic charge from oxidation (red) and reduction (blue) of R peptide repeated over 1000 reduction–oxidation cycles. The rate decay constant estimated from exponential fits (dotted lines) is approximately 0.002 per cycle.

On the basis of analysis of a protein sequence in 1966, Eck and Dayhoff proposed that the ancestral ferredoxin might have had a repeating four-amino-acid pattern.²⁹ Since then, there have been many attempts to make minimal ferredoxin-like peptides based on similar sequence repeats.^30–33 The ambidoxin design, based on repeating structural features rather than sequence, produced a simple, artificial ferredoxin that closely mimics redox behavior of its biologically evolved counterparts. Such structural and sequence simplicity greatly increases the likelihood of prebiotic evolution of catalytic redox-active peptides on early Earth.³⁴ Molecules much like the ambidoxins may have served as soluble electron carriers prior to the emergence of homochirality required for protein translation and folding.³⁵ In this scenario, ambidoxin would have assembled from racemic sources of amino acids present in the environment, with peptide bonds formed abiologically, near hydrothermal vents³⁶ or on mineral surfaces.³⁷

Wächtershäuser and others proposed short heterochiral peptides with catalytic metal centers were critical intermediates in the earliest stages of chemolithotrophic origins of life.^38,39 The robust redox stability of ambidoxins is a clear demonstration in the laboratory that early peptide catalysts could have been functionally on par or even superior to modern proteins in some respects. Without chaperones, metal cluster biogenesis and homeostasis mechanisms present in a cellular milieu, early peptides had to specifically and stably coordinate metals in order to be effective catalysts.⁴⁰ This is in contrast to the view of early enzymes as poorly functioning ancestors of modern catalysts.^41,42 Geometric calculations indicate multiple solutions for short heterochiral molecules that form rings of different lengths (Supporting Information). Ambidoxins may be just one case in a larger family of metalchelating peptides that could have served as prebiotic molecular ancestors for redox catalysts that sparked the emergence of modern oxidoreductases.