Engineering Synthetic Electron Transfer Chains from Metallopeptide Membranes

The energetic and geometric features enabling redox chemistry across the copper cupredoxin fold contain key components of electron transfer chains (ETC), which have been extended here by templating the cross-β bilayer assembly of a synthetic nonapeptide, HHQALVFFA-NH2 (K16A), with copper ions. Similar to ETC cupredoxin plastocyanin, these assemblies contain copper sites with blue-shifted (λmax 573 nm) electronic transitions and strongly oxidizing reduction potentials. Electron spin echo envelope modulation and X-ray absorption spectroscopies define square planar Cu(II) sites containing a single His ligand. Restrained molecular dynamics of the cross-β peptide bilayer architecture support metal ion coordination stabilizing the leaflet interface and indicate that the relatively high reduction potential is not simply the result of distorted coordination geometry (entasis). Cyclic voltammetry (CV) supports a charge-hopping mechanism across multiple copper centers placed 10–12 Å apart within the assembled peptide leaflet interface. This metal-templated scaffold accordingly captures the electron shuttle and cupredoxin functionality in a peptide membrane-localized electron transport chain.

5.6, accompanying another vortex period.Peptide solutions assembled at room temperature for >1 week.
Pre-assembled K16A-Cu(II) fibrils: Mature K16A fibrils were titrated with a small volume of 100 mM CuCl 2 solution to the desired metal equivalent.
Co-assembled Cu(II)-K16A nanoribbons: a 2x peptide stock was prepared as described previously, except with the 100 mM Cu(II) stock addition occurring directly after diluting to 1x with 50 mM MES, along with a third vortex period.Note, the volume of water used to prepare the 2x stock was reduced to accommodate the final addition of Cu(II) stock.Peptide solutions assembled at room temperature for >2 months.

Visible Absorbance Spectroscopy
Scatter-free absorbance spectra of peptide-metal assemblies were measured with an RSM 1000 at Olis, Inc. (Bogart, GA).Samples were diluted 2x with buffer to a final volume of 3.2 mL in a 9 mL DeSa Suspension Presentation Cavity.Spectra were collected with an 850-400 nm grating, 2.25 nm data interval, and a 180 nm/min scan rate.

Cyclic Voltammetry
Cyclic voltammetry was performed at room temperature using a WaveDriver potentiostat (Pine Research Instrumentation, Durham, NC, USA).Experiments used a standard three-electrode arrangement and a microvolume cell.Electrodes used were a 1 M Ag/AgCl reference electrode (0.228 V vs SHE), a Pt wire counter electrode, and a glassy carbon working electrode (CH Instruments Inc. Electrochemical Instrumentation, Austin, TX, USA).Buffered sample (2 mM peptide + 2 mM CuCl 2 ) was taken in 75 uL aliquots and diluted 2x with 25 mM MES, pH 5.6 + 100 mM KCl.The supporting electrolyte was 25 mM MES, pH 5.6 + 50 mM KCl. Cyclic voltammetry was conducted in the potential range of -0.500 to 0.600 V vs. Ag/AgCl at scan rates of 0.025, 0.050, 0.100, 0.250 and 0.500 (V/s).The analyte's open circuit potential was defined as the starting sweep potential, and all runs were cycled starting in the positive potential direction.Before each measurement, the electrolyte was degassed with N 2 for ~1 min.The working electrode was polished before use and in-between runs with an alumina slurry.A buffer measurement between each run at a scan rate of 0.100 V/s verified each experiment was performed with a clean working electrode with no detectable carryover signal/drift.Data was processed in AfterMath.Signal intensities were baseline-corrected in the software using its built-in free-sloped correction function, using a tangent of the preceding wave to define the baseline current.To convert Ag/AgCl-referenced potentials to SHE-referenced reduction potentials at any given pH at 25

Circular Dichroism Spectroscopy (CD)
CD spectra were recorded using the Jasco-810 CD spectropolarimeter at room temperature.Spectra between 260 nm and 180 nm were collected with a 0.1 mm path length cell, with a step size of 0.2 mm and a speed of 50 nm/s.All spectra were recorded in triplicate and averaged.Ellipticity (mdeg) was converted to mean residue molar ellipticity (MRME, [θ], degcm 2 dmol -1 ) by [θ] = θ/(10nCl), where 'n' is the number of backbone amide bonds per peptide, 'C' is the molar concentration (mol/L) and '1' is the cell path in cm.

Fourier Transform Infrared Spectroscopy (FT-IR)
All FT-IR spectra were recorded using a Jasco FT-IR 4100 (Easton, MD, USA).Aliquots (10 μL) of peptide solution were fan-dried onto a Pike GaldiATR (Madison, WI, USA) diamond at room temperature and averaging 1,068 scans with 2 cm -1 resolution.Background spectra were subtracted from each sample.

Fluorimetry
H 2 O 2 production was measured using a commercially available Amplex Red assay kit (Sigma) in stickered 96-well, black-walled, optical glass-bottom plates (Greiner Bio-one).Samples and standards were prepared as described by the kit's instructions with no procedural modifications.Samples were diluted 20x in buffer and reaction mixture prior to measurement.Cu(I)-mediated ROS production was initiated with freshly prepared 0.45 eq.sodium ascorbate (0.9 reducing eq.) prior to mixing with the kit reaction mixture.Sacrificial reductant was intentionally supplied as the limiting reagent as it triggers H 2 O 2 production in the kit's master mix through the activity of the included horseradish peroxidase.All experiments were performed in a Synergy Mx plate reader (BioTek) with 300 μL sample volume maintained across all wells.The scan rate was set to 1 read/10 min with a 7 s shake period prior to every measurement.

Powder X-Ray Diffraction (pXRD)
Assembled K16A fibers and ribbons were collected by centrifugation using an Eppendorf centrifuge 5810 R (Mississauga, Ontario, CAN) at 4000 rpm held at 4 °C to prevent heating and melting the sample and to remove free monomer and small heterogeneities from the assembly mixture.The pelleted sample was decanted, resuspended in metal-free water, and re-pelleted 3x to remove buffer salts as well.The washed pellet was frozen and lyophilized to yield a dry powder for pXRD.The powder spectra were obtained using APEXII diffractometer with a Cu X-ray source, provided through Emory's X-Ray Crystallography Center.

Transmission Electron Microscopy (TEM)
10 μL K16A assembly aliquots were drop cast onto CF200-Cu Carbon Film 200 Mesh Copper grids (Electron Microscopy Sciences, Hatfield, PA).Peptide assemblies were settled on the grid surface for 1 minute.Excess solution was wicked away using a kimwipe.Immediately following, 10 μL of a 1.5% uranyl acetate stain was applied for 20 s and similarly wicked way.TEM grids were vacuum desiccated overnight before imaging.Electron micrographs were imaged with a Hitachi HT-7700 Microscope at 80 kV at the Robert P. Apkarian Integrated Electron Microscopy Core.
Image measurements were performed using ImageJ, which were fit to Gaussian functions in Origin 2018b.

Atomic Force Microscopy
Samples were diluted 100-200x to ensure sample dispersion, 45 μL aliquots were drop cast onto silicon chip for 2 min (Ted Pella Inc., Redding, CA, USA).Excess solution was removed via micropipette.
Tapping mode analysis on a Veeco Dimension 5000 atomic force microscope with a full acoustic enclosure was employed using ultra-sharp non-contact silicon cantilevers with typical frequencies ranging from 150 kHz for the largest tip to 315 kHz for the smallest tip (NSC35/AIBS, MikroMasch, Watsonville, CA, USA).Samples were imaged simultaneously in topography and phase modes, with a filter of 0.2 Hz, a scan clock of 0.1667 ms, and a reference of at 70% of the maximum amplitude.The images were analyzed in Gwyddion. 3

Electron Paramagentic Resonance (EPR)
EPR simulations of spectra obtained for the range of Cu(II)-assembly binding, at [Cu(II)]/[K16A] ratios of 0.1 -0.6, indicate contributions from three dominant components (Fig. 3, Table S2).The [Cu(II)]/[K16A] = 0.1 spectrum cannot be simulated by using a single set (single Cu(II) species) of EPR simulation parameters (electron g and strain-broadening, Cu hyperfine, 14 N superhyperfine), because the amplitude of the simulated 14 N superhyperfine multiplet, associated with the requisite Cu(II) g-and hyperfine parameters, exceeds the experimental amplitude by >5fold.This introduces a critical constraint on the model for the different Cu(II) site contributions.The overall Cu lineshape is preserved, and the 14 N superhyperfine amplitude attenuated, if a second Cu(II) site (Component 2) with enhanced strain-broadening is present, relative to the Cu(II) site (Component 1) with less strain-broadening, that expresses the 14 N superhyperfine features.At [Cu(II)]/[K16A] = 0.1, Components 1 and 2 are present in proportions of 80 and 20%, respectively, and there are no additional significant components (Fig. S7).
Components 1 and 2 are present in the spectrum at [Cu(II)]/[K16A] = 0.2 at altered proportions of 13 and 72%.A small deviation of the sum of Components 1 and 2 from the observed spectrum in the g perp region, that extends into the g || region, is accounted for by a Component 3 (15%), which is characterized by a broad derivative feature and absence of distinct Cu hyperfine features.The decline in 14 N superhyperfine amplitude at [Cu(II)]/[K16A] = 0.4 is accounted for by a relative decrease in the contribution of Component 2 (7%).Component 1 also decreases (53%) and is compensated by a rise in Component 3 (40%).At [Cu(II)]/[K16A] = 0.6, which approaches the saturation of Cu(II) binding sites, Components 1, 2, and 3 reach proportions of 6, 47 and 47%, respectively.These proportions appear to persist at [Cu(II)]/[K16A] ≥ 0.8, in the range where the dominant aqueous Cu(II) spectrum precludes quantification of bound Cu(II) line shape contributions (Fig. 2E).

Solid-State Nuclear Magnetic Resonance
HHQ[1- 13 C]ALV[ 15 N]FFA-NH 2 (enriched K16A) assemblies were pelleted, decanted, and washed with metal-free water 3x at 4°C for 90 min using an Eppendorf centrifuge 5810 R (Mississauga, Ontario, CAN) at 4,000 to remove any unassembled monomer and buffer.The washed pellet was lyophilized to a powder.TEM confirmed assembly persistence following lyophilization.The NMR samples were packed into 4 mm solid-state NMR rotors and centered using boron nitride spacers.Rotational-Echo DOuble-Resonance (REDOR) 4 spectra were collected as described previously 5 .In dephasing the carbonyl carbon of HHQ[1- 13 C]ALV[ 15 N]FFA-NH 2 nanoribbons, the distance of the H-bonded 15 N from the adjacent peptide was set to 4.93 Å (r 1 ) and the distance to the non-H-bonded 15 N was set to 6.0 Å (r 2 ).The angle between the two 13 C- 15 N internuclear vectors was set to 114.7°.The experimental data were fit to a linear combination of 3-spin (one 13 C and two 15 N's) and 13 C{ 15 N}REDOR curve corresponding to the 13 C- 15 N distances sufficient to fit the experimental data points using the Non-Linear Fit routine in Mathematica.

X-ray Absorption Spectroscopy (XAS)
XAS Spectra were collected at Argonne National Laboratory (ANL) on beamline 20-B in fluorescence excitation mode with a 1 x 1 mm X-ray beam 10% defocused at 9.5 KeV (harmonic cutoff = 13.2KeV).An insertion device beamline was not utilized to minimize potential beam damage.
The data were collected at 72 K, which was maintained by a liquid nitrogen cryostat (< 10 mTorr) with a sample-to-edge distance of 25 mm and a detector-cryostat distance of < 45 mm.A 12element Ge detector with 2 us peaking time and 0.55 s integration time was used.
All samples were pelleted and resuspended in buffer to 80% original volume to remove potential peptide monomer-Cu(II) complexes.Immediately before freezing, samples were diluted back to the original volume with glycerol and mixed thoroughly to prevent Bragg scattering by water ice crystallites.Reduced samples were prepared in a portable glovebag (Sigma) that was purged 5x with argon.
Samples were aliquoted into a 3D-printed sample holder supplied by the beamline.The sample holder had two sample chambers stacked vertically, each with a volume of 350 uL and an 18 x 5 mm window area.The window material consisted of kapton film that was secured via an aluminum frame and steel screws.The entire cuvette was covered with a piece of tin sheeting with slits cut in to expose the windows while preventing excitation of the adventitious copper detected in the screws.
To minimize beam damage, sample was aliquoted in both chambers in duplicate with twenty scan sites that alternated across both sample chambers.Each scan collected 467 data points over 13.42 min between 8,830 and 9,953 eV.
All spectra were calibrated to standard copper foil data 6 in Athena (Demeter) 7 and individually inspected for photodamage.Satisfactory replicate scans were aligned, deglitched, and merged.
XAS data was fit in a custom IgorPro routine 8 as previously described in Shearer and Soh, 2007 9 .

Molecular Modeling and Molecular Dynamics (MD)
Peptide assembly models were generated using Maestro 11.4 (Schrodinger Suite). 10K16A peptides were extended as antiparallel β sheets and arranged into a bilayered lattice of 8 β sheet chains consisting of 8 peptides each for a total of 128 peptides.Peptides were spaced according to the hydrogen-bonding and lamination distances reported by pXRD, and the peptide registry according to the 13C{15N} REDOR fit.
Unrestrained MD was performed on the peptide lattice using GROMACS v2018.3 11in a centered cubic box of TIP4P 12 water with periodic boundary conditions and a 1.0 nm protein-box distance using the OPLS-AA 13 force field.Short-range Van Der Waals and electrostatics were cutoff at 1.0 nm each using the Verlet scheme 14 .The system was neutralized with Na + and Cl -.The system was energy minimized for 1,000 steps using a steepest descent algorithm then equilibrated via 100 ps NVT and 100 ps NPT ensembles.Average temperature was kept at 300 K using a Berendsen thermostat 15 with a 1.0 ps time constant, and average pressure was kept at 1 bar using a Parrinello-Rahman barostat 16 with a 2.0 ps time constant.MD was performed until the aligned protein's RMSD converged at approximately 1.2 ns.Bond distances were constrained with the LINCS algorithm 17 with a 2-fs integration time step.
The final frame's coordinates of the apo-peptide lattice were simulated with a dummy copper (DCU) to estimate the coordination sphere geometry.The DCU was constructed from a centroid placed within the bilayer and between the 4 th and 5 th β sheets to minimize the effects of bulk solvent exposure and peripheral β sheet peeling.The DCU was repositioned into ligation proximity using Chimera.To minimize experimenter bias, only the Cα-Cβ and Cβ-Cγ bond torsions of the Nterminal HISE were adjusted.The DCU-containing models were simulated identically to the apolattice, until the RMSD of an aligned, water-excluded 4 Å radius of DCU visibly plateaued.
The exported coordinates and the OPLS-AA force field was modified to include the experimental parameters from XAFS and ESEEM and simulated for 2 ns.RMSD typically converged in > 1 ns.MD trajectory analysis (trajectory RMSD, bond length measurements, etc.) was performed using VMD. 18Trajectory visualization/rendering and model raytracing was performed using PyMOL. 19.pdb of the final frame was used to simulate a pXRD pattern using CRYSOL (ATSAS suite).20 Default simulation parameters were used except for the order of harmonics and the order of the Fibonacci grid, which were maximized.The simulated pattern was normalized in Origin 2018b.

Electron spin-echo envelope modulation (ESEEM) spectroscopy.
ESEEM data were collected at 6 K on a home-built pulsed-EPR spectrometer 21 by using the threepulse (π/2--π/2-T-π/2--echo) microwave pulse sequence with pulse-swapping. 22The  value was selected from a scan of the  values, corresponding to the relation, , where n = 1, 2, 3, … and is the free precession frequency of the 1 H nucleus at the prevailing external magnetic field.

 1 H
These  values correspond to -suppression 23 of the free 1 H contribution to the ESEEM.The  values of 160 ns (n = 2) and 310 ns (n = 4) were chosen to emphasize the low-frequency 14 N nuclear quadrupole features and their combination lines, and the 14 N double quantum line ( dq ) and its corresponding combination feature (2 dq ), respectively.ESEEM was cosine Fourier transformed to generate the ESEEM frequency spectra.Modulation in the dead time portion of the ESEEM waveform was reconstructed by inverse-Fourier transformation of prominent features in the initial Fourier transform. 24All data processing was performed by using laboratory-written MATLAB (Mathworks, Natick, MA) programs running on local computers.

Electron spin-echo envelope modulation (ESEEM) simulations.
Numerical simulations of the ESEEM were performed by using the OPTESIM ESEEM simulation and analysis software suite. 25Simulation of ESEEM that arises from the coupling of the remote 14 N of imidazole ligands to the unpaired electron spin on Cu(II) has been described in detail. 26riefly, the coupled electron-single 14 N system is parameterized by using the following eight adjustable parameters: A xx , A yy , A zz (electron-14 N superhyperfine coupling constants), e 2 qQ/h and  ( 14 N nuclear quadrupole coupling constant and asymmetry parameter), and angles that relate the principal axes of the superhyperfine and nuclear quadrupole tensors).For simulation of the Cu(II)-bis-imidazole complexes, these eight parameters are assumed to be common to the two imidazole 14 N.For the bis-imidazole complex,  A ,  A , and  A (Euler angles that relate the two 14 N superhyperfine dipolar tensor axes) are additional adjustable parameters.The simultaneous confidence region for each adjustable parameter, at a specific confidence level (here, 99%), was calculated, as described. 25Table S4 presents the values of the simulation parameters.

Figure S1 .
Figure S1.Periodic table-view screening of attempted metal-K16A assemblies to date, displaying several morphologies such as fibers (e.g.Ni(II)), nanotubes (e.g.Zn(II)), helical ribbons (e.g.Zn(II), too), twisted ribbons (e.g.Cu(II)), belts (e.g.Pt(IV)), aggregate (e.g.Fe(III)).(A).Assemblies prepared with green-labeled elements readily yield characterizable K16A assemblies by TEM.Elements labeled yellow produce mixtures of assembly and amorphous aggregate.Red elements form less structured associations with K16A.Representative TEMs of metal-K16A assemblies are shown in panels B-O.All assemblies were prepared as 1:1 co-assemblies in 25 mM MES, pH 5.6.Chloride salts were used as the metal ion source, except for Al(III) and Ag(I), which were introduced as nitrate salts.The periodic table was adapted from freestock.ca.

Figure S6 .
Figure S6.H 2 O 2 production reported as resorufin fluorescence via an Amplex Red Assay across co-, pre-, and particulate (unassembled) Cu(II)-K16A preparations (A).Cumulative H 2 O 2 evolution across measured samples by the final timepoint (B).All samples were diluted to a final concentration of 0.1 mM peptide and 0.08 mM Cu(II) via sample preparation.

Figure S14 .
Figure S14.Cartoon highlighting d-spacing of N-terminal (H) and C-terminal (A) residues and nearest H-H distance with antiparallel orientation.

Figure S17 .
Figure S17.EPR of concentrated Cu(II)-K16A compared with a 1 mM peptide preparation as shown in Fig. 2.

Figure S18 .
Figure S18.Trajectory overview and dative bond-length statistics for Cu(II)-K16A, mono-His models cis (A-B) and trans (C-D).RMSD of simulated the copper center restricted to a 4 Å radius about the DCU, with the frames representing the pre-stabilized and stabilized structure denoted via a gray box.The average bond length values plotted across trajectory measurements in B and D are represented with a dashed red line.Note: RMSD spike within ~650-1,850 ps for A is attributed to a vicinal but non-interacting His ring flip.

Figure S19 .
Figure S19.Trajectory overview and dative bond-length statistics for Cu(II)-K16A, bis-cis-His.Simulated pXRD pattern of Cu(II)-K16A's coordinates exported from its trajectory's final frame (A).RMSD of simulated Cu(II)-K16A lattice, restricted to a 4 Å radius about the DCU, with the frames representing the pre-stabilized and stabilized structure denoted via a gray box (B).Ligating histidines' bite (C) and dihedral angles (D) as a function of time.Both His-Cu(II) bond lengths as a function of time (E-F).The average value plotted across trajectory measurements in C-F are represented with a dashed red line.

Figure S20 .
Figure S20.Simulated bis-cis-His Cu(II)-K16A assembly with ESEEM and XAFS restraints as a single site featuring a dummy copper atom (yellow sphere).Non-histidine sidechains are hidden for clarity (A).Time-averaged coordination sphere isolated as a 4 A selection radius consisting of chelating His sidechains only visualized on orthogonal axes (B-C).

Table S3 .
Statistically valid and chemically reasonable models to the EXAFS data.
a Number of scatterers restrained to the nearest whole number