[NiFe] hydrogenases - ACS Publications - American Chemical Society

field.76,77 For each simulation, the tleap program in AMBER 16 was used to load the force ..... must be slower than the rates of 0.1-50 x 104 M-1 s-1 ...
1 downloads 0 Views 7MB Size
Subscriber access provided by THOMAS JEFFERSON UNIV

Article

Going beyond structure: Nickel-substituted rubredoxin is a mechanistic model for the [NiFe] hydrogenases Jeffrey W. Slater, Sean C. Marguet, Haleigh A. Monaco, and Hannah S. Shafaat J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b05194 • Publication Date (Web): 18 Jul 2018 Downloaded from http://pubs.acs.org on July 18, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

Going beyond structure: Nickel-substituted rubredoxin is a mechanistic model for the [NiFe] hydrogenases Jeffrey W. Slater‡, Sean C. Marguet‡, Haleigh A. Monaco†, and Hannah S. Shafaat* The Ohio State University, 100 W. 18th Ave., Columbus, Ohio 43210; [email protected] ABSTRACT: Well-defined molecular systems for catalytic hydrogen production that are robust, easily generated, and active under mild aqueous conditions remain underdeveloped. Nickel-substituted rubredoxin (NiRd) is one such system, featuring a tetrathiolate coordination environment around the nickel center that is identical to the native [NiFe] hydrogenases and demonstrating hydrogenase-like proton reduction activity. However, until now, the catalytic mechanism has remained elusive. In this work, we have combined quantitative protein film electrochemistry with optical and vibrational spectroscopy, density functional theory calculations, and molecular dynamics simulations to interrogate the mechanism of H2 evolution by NiRd. Proton-coupled electron transfer is found to be essential for catalysis. The coordinating thiolate ligands serve as the sites of protonation, a role that remains debated in the native [NiFe] hydrogenases, with reduction occurring at the nickel center following protonation. The rate-determining step is suggested to be intramolecular proton transfer via thiol inversion to generate a NiIII-hydride species. NiRd catalysis is found to be completely insensitive to the presence of oxygen, another advantage over the native [NiFe] hydrogenase enzymes, with potential implications for membrane-less fuel cells and aerobic hydrogen evolution. Targeted mutations around the metal center are seen to increase the activity and perturb the rate-determining process, highlighting the importance of the outer coordination sphere. Collectively, these results indicate that NiRd evolves H2 through a similar mechanism as the [NiFe] hydrogenases, suggesting a role for thiolate protonation in the native enzyme and guiding rational optimization of the NiRd system.

INTRODUCTION In recent years, extensive efforts have focused on developing sustainable methods for hydrogen generation from water.1 This energy-dense compound has potential application as a transportation fuel and plays a key role in industrial reactions around the world, including the Haber-Bosch process for ammonia synthesis and the Fischer-Tropsch process for diesel production, highlighting its importance in secondary processes as well. Currently, approximately 95% of the global hydrogen supply derives from steam reforming of methane and the water-gas shift reaction, both of which contribute to rising CO2 levels.2 Nature has instead delicately optimized environmentally benign methods for hydrogen generation through the use of metalloenzymes called hydrogenases (H2ases).3 H2ases exhibit high turnover frequencies, utilize earth-abundant transition metals such as Ni and Fe, are active in aqueous solutions, and show high thermodynamic efficiencies for H2 generation.3,4 However, several factors prevent H2ases from being used on an industrially relevant scale. The most pronounced drawback is sensitivity to oxygen, coupled with difficulties in expression and production of these enzymes in high quantities.5–8 Additionally, the large sizes of native H2ases limit the possible enzyme density for heterogeneous electrocatalysis, which must be high for effective industrial processes3,9 Noting that the [NiFe] hydrogenases are traditionally thought to be less sensitive towards oxygen inactivation4,10 than the [FeFe] systems and are also active for H+ reduction,11,12 with some variants demonstrating strong catalytic biases towards reductive catalysis,13–16 this enzyme has served as a target for diverse approaches to develop catalysts that could be utilized for H2 generation. Our strategy has been centered around development of a protein-based mimic of the [NiFe] hydrogenases using rubredoxin

(Rd), a small, electron transfer protein, as a scaffold. By replacing the native iron center in rubredoxin with nickel, a monomeric, tetrahedral nickel(II) system is generated that reproduces the tetrathiolate primary coordination sphere of the redox-active metal in the [NiFe] H2ases. This approach differs from that pursued in the synthetic community, where homoleptic nickel(II) thiolates have a high propensity towards oligomerization into inactive species. Of those previously reported, only bulky, aryl tetrathiolate nickel(II) compounds have been isolated, and none show reversible redox chemistry, decomposing under reducing or oxidizing conditions.17 Instead, smallmolecule nickel compounds for hydrogen generation typically include less basic phosphine and thioether ligands or redox-active, conjugated dithiolene ligands in a square-planar geometry.18–21 In NiRd, the protein scaffold appears to prevent against dimerization, providing a molecular system that closely mimics the native enzyme.22 Additionally, the use of a structured protein provides a defined secondary coordination sphere, which is known to play an important role in modulating catalytic activity.23–27 Previous work by Moura had shown that native rubredoxin proteins isolated from sulfate-reducing bacteria and substituted with nickel were active for hydrogen evolution and H/D exchange using traditional H2ase assays.28 Our group has expanded this characterization of NiRd activity using solution-phase, light-driven, spectroscopic, and electrochemical measurements.22,29,30 NiRd displays a moderate electrocatalytic overpotential of approximately 550 mV, which is similar to those seen for many first-generation small-molecule and recent proteinderived catalytic systems operating in purely aqueous solution,31–38 and exhibits estimated electrocatalytic rates that are similar to those of the native enzymes.22,30 As such, NiRd shows great potential as an artificial hydrogenase, as it is easily expressed, oxygen tolerant,

ACS Paragon Plus Environment

S Cys Cys S SCys SS SS Cys III Cys Ni Cys NiIII S S Cys S S Ni0 0 Cys Cys Ni Cys Journal of the American Chemical Society Page S S S S H H II S II S S Cys S S Cys Ni Cys Cys Ni S I S Cys I Ni Cys S Cys Ni Cys S H II S H H S H Cys Ni S S Cys Cys S Cys to dramatically perturb catalytic activity. Collectively, these efforts S Cys H Ni-SI Ni-L’ a’ H H 1 reveal that the mechanism of catalysis by NiRd closely models that Cys -S +Cys Cys Cys S Cys Cys Cys Cys Cys Cys Cys Cys Cys S Cys Cys Cys 2 S Cys proposed for the [NiFe] hydrogenase, with many of the same cataSS S S H S SS I Cys S S S lytic 0intermediates S Ni invoked. The correlation between structure and 3NiII S Cys S S 0 S S Cys S S Cys Ni S S NiII Cys Ni Cys Cys Cys Ni0 NiII Cys Cys Cys Ni0 Cys function of NiRd and the native hydrogenases motivates further de4 S S S Cys H S Cys Cys Cys SS Cys H5 S velopment of NiRd as a tunable, robust system for H2 evolution. SS Cys S Cys S Cys S S S Cys Cys Cys NiII NiII S Cys H H H 6 C9 MATERIALS AND METHODS 4 Y s Cys 11 S S -V 3 Cys III S Cys Cys Cys S NiCys Cys C 32 CysCys Cys 7S S C32-C35 S Materials. All materials were obtained from Fisher Scientific, Cys S Cys Cys Cys Cys Cys Cys S Cys Cys S S Cys H S 2 II 8 S S S H H Cys Ni S S Cys Sigma-Aldrich, or VWR unless otherwise stated. Optical elements S S S Cys S Hydride NiI I Cys S NiI II S Cys Ni S S S C6 -C II SS II IIISpurchased S Cys H 9 Cys S H Cys 9 Ni were from Thorlabs or Newport Corporation. All soluNi Ni III Formation Cys S Cys Ni Cys Ni S Cys -V 8 Cys C35 -V3 SH Cys 10 C6 H S H tions were prepared using 18.2 MW/cm deionized water (ELGA H S Cys 7 H S Cys S S Cys Cys S Cys 11 Cys Cys Technologies). HH S H Cys Cys S 12 Protein expression, purification, and metal substitution. DesulS S Cys S S Ni0 Cys 13Cys Cys Cys 0 Cys Cys Ni fovibrio Cys Cys Cysdesulfuricans ATCC 27774 wild-type rubredoxin and muCys Cys Cys S S Cys Cys S S S S S 14 S Cys tants were heterologously expressed, purified, and metallated as preS S S SS Cys S S S S Cys S S III S S NiII III S Cys NiII 22,29,30 III NiII 15 Cys Cys Ni Cys viously described. Mutagenesis DNA primers and conditions Ni NiIII CysCys NiH H H 16 + Cys H S Cys H S for DdRd can be found in Supporting Information. Preparation of S Cys S Cys Cys S S CysCys Cys S Cys S H 17 CysH Cys CoRd was performed similarly to NiRd and FeRd, but with CoCl2 • S Ni-C’ H S S Ni-R’ 18 6H2O added after TCA precipitation. All metal-substituted proteins Cys NiI S S Cys NiII 19 were run on a size-exclusion column (Superdex-75, prep-grade, Figure 1. Proposed catalytic mechanism for hydrogen generation by S Cys H 20 S Cysof full NiIIRd model. The hySigma-Aldrich) and concentrated in a centrifugal filter device to reNiRd. (Inset) DFT-optimized geometry H 21 drogen-bonding network is indicated with dashed lines and labeled. move adventitiously bound metal adducts. Stoichiometric metal in-

Cys

S

Cys

Cys

Cys S

e, H

H

e, H

22

S 23

Hydrogen atoms have been omitted for clarity. Cys

24S Cys 25 S 26 Cys 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

NiII

Cys Cys Cys Cys chemically S and highly tunable for catalyst optiS and thermally stable,

22 Further, the NiRd model mization. S S system provides an opportunity S S Cys NiIII Cys NiIIIthe specific to assess contributions of the [Fe(CO)(CN)2] fragment 39–42 H to activity in the native [NiFe] hydrogenases. S Cys S Cys

A preliminary mechanism for hydrogen generation by NiRd has been proposed based on the current model for the mechanism of the native [NiFe] H2ase (Figure 1).29 The observed shift in onset potential with pH suggests that protonation is coupled to reduction, which, as is proposed in the [NiFe] H2ases,43 may initially occur at two separate sites to give a Ni-L-like species (Ni-L’) featuring a NiI center with a protonated thiolate. In wild-type (WT) NiRd, the absence of nearby basic groups suggests the thiolate ligands serve as the site for protonation and subsequent intramolecular proton transfer, though this role remains debated in the [NiFe] hydrogenases.27,44 Similar to the native enzyme, at least one nickel-hydride species (NiC’) may be involved in catalysis.44,45 Hydrogen evolution is suggested to occur from a state featuring both a metal-hydride and a protonated thiolate ligand (Ni-R’), regenerating the air-stable NiSIa’ resting state. In this work, an array of experimental and computational methods have been used to determine the mechanism of H2 evolution by NiRd. Protein film electrochemistry (PFE) involves immobilization of a protein monolayer on an electroactive surface to enable direct electron transfer to the catalytic active site, providing a real-time readout of activity.46–48 By applying mixed solutions of FeIIRd and NiIIRd to an electrode, the non-catalytic FeIII/II couple serves as an internal redox standard for determining total electroactive protein coverage. This quantitative PFE (qPFE) approach provides absolute NiRd turnover frequencies as a function of pH, scan rate, isotopic substitution, and across different protein variants, enabling the use of traditional biochemical analyses to study activity. In conjunction with the qPFE measurements and electrochemical modeling, spectroscopy, density functional theory (DFT) calculations, and molecular dynamics (MD) simulations were used to interrogate the NiRd enzymatic mechanism. The rate-determining step for H2 production was identified, and targeted secondary sphere mutations were shown

corporation was verified by inductively-coupled plasma mass spectrometry (ICP-MS, Table ST1). Protein film voltammetry. All electrochemical experiments were carried out in a 1 V/s) voltammetry. A three-electrode system was used containing a homemade edge-plane pyrolytic graphite working electrode (PGE, construction described in Supporting Information), a platinum wire counter electrode, and a Ag/AgCl reference electrode. Reported potentials were converted to the normal hydrogen electrode (NHE) by the addition of +0.198 V. Voltammetric experiments at fast scan rates (>1 V/s) were carried out using a CHI 760E bipotentiostat (CH Instruments). All other voltammetry experiments were performed using a WaveNow potentiostat (Pine Research Instrumentation). Analysis of non-catalytic signals was performed using the SOAS program.49 Laviron analysis of NiRd was carried out using the anodic potential at the termination of catalysis, which becomes an oxidation feature at faster scan rates. Using the following equations: 𝐴𝑛𝑜𝑑𝑖𝑐 𝑆𝑙𝑜𝑝𝑒 = 𝑘1 = (1 − 𝛼)

-. (01)34

78=> 34

056 78



(1) (2)

the electron transfer coefficient (a) and average interfacial electron transfer rate constant (k0) were estimated for NiRd, where F is Faraday’s constant, R is the ideal gas constant, T is temperature, and n is number of electrons. na is the scan rate from which the linear fit of the anodic peak crosses E˚ of the electrochemical process (Figure S1).50,51 Temperature-dependent experiments were performed in a custom-made, jacketed electrochemical cell with an isothermal reference sidearm.50,52 Data was fit to the Eyring equation to extract thermodynamic parameters using a transmission coefficient (k) of 1.0.53 Oxygen tolerance experiments were carried out on a modulated speed rotator (MSR) assembly combined with a 5 mm diameter edge-plane pyrolytic graphite rotating disk electrode (RDE)

ACS Paragon Plus Environment

2 of 28

Page 3 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

(Pine Instruments). An O2-saturated buffer (2 mM at 4°C) was injected into the electrochemical cell during rotation (w = 1500 RPM) to a final concentration of 4, 40, or 400 µM dissolved O2.54 Electrochemical simulations. Elementary mechanistic steps were adapted from the proposed NiRd mechanism (Figure 1) to generate differential rate laws for each of the possible components and scripted into Mathematica (Wolfram Research, Inc.) for analysis.55 Electron transfer rate constants (kred/kox) were defined in terms of potential and time using Butler-Volmer electron transfer theory,56 as shown in the following equations: CD

𝑘?@A = 𝑘1 ∗ 𝑒 5EF 𝑘MN = 𝑘1 ∗ 𝑒

GH IJK5G L



(OPC)D (GH IJK5G L ) EF

(3) (4)

where k0 is the interfacial electron transfer rate constant, 𝛼 is the charge transfer coefficient (determined to be 0.5, Figure S2), Ei is the initial potential of the linear sweep voltammogram, v is the scan rate, t is time, and E˚ is the equilibrium potential. Similar to simulations of H2ase catalysis on an electrode surface,57 a dispersion of values for k0 was taken into account based on the through-space distances from active site metal to the surface of Rd (4-12 Å) (Figure S3). These distances were applied to the simulations using the following equation: 𝑘1 = 𝑘QRN 𝑒 5SA (5) where kmax is the Fermi contact electron transfer rate, ß is the distance decay parameter (set to 1 Å-1), and d is the distance between the active site and the electrode (Figure S4).57 From the populations of species generated from the calculations, a model linear sweep voltammogram (LSV) was produced. Additional details on the simulation procedures and sample Mathematica scripts can be found in the Supporting Information. Absorbance spectroscopy. UV-visible spectra were measured using a Shimadzu UV-2600 spectrophotometer. To probe the effects of pH, protein samples were diluted to ~100 µM from a stock of ~35 mM in 50 mM Tris, pH 8.0 buffer with either 50 mM phosphate buffer, pH 8.0 or 50 mM acetate buffer, pH 4.0 immediately before spectral acquisition. Extinction coefficients for Dd FeIIIRd, NiIIRd, and CoIIRd were experimentally determined using a Bradford assay to be 7850 ± 120 M-1 cm-1, 3400 ± 70 M-1 cm-1, and 3160 ± 60 M-1 cm-1 at 492 nm, 455 nm, and 355 nm, respectively (Figure S5). Resonance Raman spectroscopy. Resonance Raman (RR) spectroscopy was performed at room temperature with ~ 1 mM sample contained within a flame-sealed sodium borosilicate capillary tube (∼1 mm i.d.). The resonance Raman setup and spectral analysis procedures were described previously.29 ZnRd spectra of the appropriate mutant were subtracted from the NiRd spectra to remove nonresonant protein peaks, and broad baseline features were removed by spline subtraction. DFT calculations. All DFT calculations were performed using the ORCA58–60 computational chemistry package at the Ohio Supercomputer Center.61 The construction of the computational models of NiRd was carried out as previously described.29 Geometry optimizations were performed in the gas phase using the B3LYP functional with the RIJCOSX approximation.62–64 The def2-SVP basis set was used on all atoms except those in the primary coordination sphere, for which the def2-TZVPP basis set was used.65 To retain the fold of the protein, the Cartesian coordinates of the Ca atoms were constrained to the crystallographic positions in the FeRd structure, a

practice often employed in DFT calculations on bioinorganic systems.29,66–68 The zeroth-order relativistic approximation (ZORA) was applied to all atoms to correct for scalar relativistic effects, though this correction was found to have only a minimal effect on calculated structures and energies (Figure S6, Tables ST2-ST3).69 Dispersion interactions were treated with the DFT-D3 atom-pairwise correction.70 Vibrational frequencies were calculated for each model using two-sided numerical differentiation to obtain the Gibbs free energies. All possible intermediates were investigated. The charges, multiplicities, and electronic energies of the optimized structures for each model are given in the Supporting Information (Tables ST4ST5). Energetics were calculated using the isodesmic approximation with the NiIII/IIRd couple used as a reference for reduction potential calculations and the deprotonation of N-acetyl-cysteineamide (NACA) as a reference for the pKa calculations. Methodology and equations are given in Supporting Information. Time-dependent density functional theory (TD-DFT) calculations were used to obtain absorption spectra of the optimized structures of relevant NiRd catalytic intermediates. The Tamm-Dancoff approximation with the B3LYP functional was performed for the first 80 roots.71–73 Vertical excitation energies and molar extinction coefficients for TD-DFT transitions were calculated using the ORCA_MAPSPC module. MD simulations. Full protein models of Dd NiRd were constructed from the X-ray crystal structure of oxidized Dd iron rubredoxin (PDB code 6RXN), with the iron substituted by nickel.68 Double conformations of residues, N-terminal formylation, and crystallographic waters were removed from the structure to avoid errors. Bond stretching and bending force field parameters for the active site of NiRd were obtained using the Visual Force Field Derivation Toolkit (VFFDT) applied to the vibrational frequency Hessian for the given active-site model (Table ST6-ST8).74,75 All MD simulations were performed with AMBER 16 using the ff14SB force field.76,77 For each simulation, the tleap program in AMBER 16 was used to load the force field parameters for the active site, neutralize the system by adding Na+ ions, and solvate the protein using a TIP3P water box with a buffer region of 12 Å in all directions.76,78 Energy minimization was performed for 5000 steps. An initial 80 ps MD simulation was carried out to heat the system to 298 K. Three, 9 ns production MD simulations at 298 K were performed on each system under NPT ensemble with periodic boundary conditions implemented. Temperature and pressure were controlled using the Langevin thermostat and the Berendsen barostat, respectively.79,80 All bonds containing hydrogen atoms were constrained to their equilibrium bond length using the SHAKE algorithm, allowing for a timestep of 2.0 fs.81 Particle mesh Ewald (PME) was employed to calculate long-range electrostatic interactions, with a non-bonded cut-off distance of 12 Å.82 During the production runs, coordinates were saved every 1 ps. RESULTS AND DISCUSSION Use of an internal redox standard in PFE provides total electroactive protein coverage One of the primary limitations of PFE for the study of hydrogen-evolving systems under aqueous conditions is that a non-catalytic redox couple is often absent, instead giving way to catalytic voltammograms. Considering that the variation in protein coverage on a given electrode can vary by up to an order of magnitude depending on the electrode, polishing technique, and drying method, it is gen-

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

erally not possible to obtain the absolute electroactive enzyme coverage, which is necessary to convert measured current values into total turnover frequencies. To circumvent these limitations, a non-catalytic version of the same protein may be used as an internal redox standard to report on the electroactive protein coverage on the surface. Rubredoxin is ideally suited to explore application of this approach because of its versatile metal-binding capabilities. The global structure of Rd is maintained when the active site is loaded with FeII, CoII, NiII, and ZnII ions.83 To verify that metal identity does not impact the orientation preference of the protein on the surface of the electrode, electrostatics, or total electroactive coverage, PFE was performed on adsorbed solutions containing both CoIIRd and FeIIRd in varying ratios. Unlike many square planar CoII compounds,32,84,85 CoIIRd is not electrocatalytically active for H+ reduction (Figure S7), consistent with the stable tetrahedral geometry around the metal center.86 Instead, cyclic voltammetry on electrodes containing both CoIIRd and FeIIRd show a reversible CoIII/IIRd couple at +425 mV vs. NHE along with the FeIII/IIRd couple at +10 mV vs. NHE (Figure 2A). The relative integrated areas seen for the FeIII/IIRd and CoIII/IIRd couples follow the ratios of FeIIRd/CoIIRd in solution (Figure 2B), consistent with metal-independent protein adsorption of Rd on PGE electrodes. This one-to-one correspondence is only seen when the two metals have the same oxidation state. Mixed solutions of FeIIIRd with CoIIRd show relative enrichment in

Page 4 of 28

the FeIIIRd protein (Figure S8), consistent with a dominantly electrostatic adsorption mechanism. These control experiments support the use of this methodology to probe the catalytic activity of NiRd. Application of a deoxygenated solution containing both NiIIRd and FeIIRd to an edge-plane pyrolytic graphite electrode (PGE) using standard protein film deposition techniques results in a cyclic voltammogram (CV) with two distinct electrochemical signals (Figure 2C). The non-catalytic FeIII/II redox couple at +10 mV vs. NHE can be integrated to determine the amount of FeRd on the surface. In the same CV, a large catalytic wave appears at approximately -800 mV vs. NHE, corresponding to proton reduction by NiRd. Both FeRd and NiRd signals are identical to those seen previously in protein films made from pure NiIIRd and FeIIRd,30 suggesting that doping solutions of NiRd with FeRd does not affect the interaction of either protein with the electrode surface. As with the CoRd controls, a linear dependence between the NiRd current and the relative concentration of NiRd in solution indicates that both proteins adsorb on the surface independently (Figure 2C, inset), unaffected by the presence of the other metallated species. This approach is also insensitive to the presence of redox-inactive species, as demonstrated by experiments in which ZnIIRd was added to solutions prior to adsorption (Figure S9). Because FeIIRd and NiIIRd are the same size, with similar isoelectric points (Figure S10), the two variants are expected to adsorb to the electrode surface with comparable affinity. In support of this assumption, optical spectra demonstrate that upon desorption from the electrode surface, NiRd and FeRd are found in the same ratios as in the protein solution that was initially applied to the electrode (Figure S11). This holds true after running electrochemical experiments, indicating the stability of the two proteins on the electrode is comparable over the timescale of the experiment (Figure S12). Unfortunately, common surface analysis techniques such as Xray photoelectron spectroscopy were not sensitive enough to detect the metal ions in the ≤100 pmol/cm2 of Rd on the electrode surface (Figure S13). pH and isotope effects suggest intramolecular proton transfer is the rate-determining step for NiRd catalysis

Figure 2. (A) Cyclic voltammogram of surface-adsorbed Rd from a 50:50 FeIIRd/CoIIRd mixture (n=100 mV/s). Sections highlighted represent FeIII/II (red) and CoIII/II (teal) redox couples. (B) Relative electroactive coverages of FeIIRd/CoIIRd as a function of solution ratios. Lines drawn show a 1:1 correspondence. (C) Cyclic voltammogram of surface-adsorbed Rd from a 76:24 FeIIRd/Ni IIRd mixture (n=100 mV/s). Sections highlighted represent the catalytic wave of NiRd (green) and the FeIII/II redox couple (red). (Inset) Catalytic NiRd currents at Ean normalized to 10 pmol total Rd from NiRd and FeRd mixtures in varying ratios, measured in H 2O and D2O buffers (pL=4.5).

By considering the coverage (GFe) quantified from the area of the FeIII/II couple and extrapolating from the known ratios of FeIIRd and NiIIRd in the applied solution, the absolute coverage of NiRd (GNi) on each electrode can be calculated. As the NiRd catalytic signal appears nearly linear with potential and does not reach a limiting current value due to non-ideal adsorption on the electrode and background H+ reduction by the electrode at low potentials, analysis of the NiRd catalytic activity cannot be performed with traditional solution-phase electrochemical metrics. Instead, the onset potential for catalysis is defined as the inflection point of the first derivative of the CV (Figure S14) for consistency across all protein variants. Using this metric, the onset potential (Eonset) at pH 4.0 of WT NiRd is measured to be -800 mV vs NHE and shifts by -59 mV/pH unit, suggestive of a proton-coupled electron transfer (PCET) process with a catalytic overpotential of 560 mV (Figure 3A). The absolute activities are extracted using current values measured at the analysis potential (Ean) that is 100 mV more negative than the onset potential (Ean = Eonset – 100 mV, Figure S15).22 With the electroactive coverage quantified, the overall apparent turnover frequency (TOFapp, mol H2/s per mol NiRd) of NiRd at Ean is obtained using Equation 6

ACS Paragon Plus Environment

𝑇𝑂𝐹RXX =

YZ[\ ]^H 7_

(6)

Page 5 of 28 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of the American Chemical Society

where 𝑖`[\ is the current at Ean, ΓNi is the moles of NiRd on the electrode after correcting for the ratio of NiRd to FeRd in the protein film, and n = 2. This analysis reveals an absolute TOFapp,H2O for H2 production by WT NiRd of 49.5 ± 6.5 s-1 at pH 4.5 (n = 36), a more

Figure 3. (A) Catalytic onset potentials and (B) turnover frequencies of WT NiRd as a function of pH with experimental and simulated values shown as closed and open symbols, respectively.

precise value than was previously attainable. The activity is independent from pH 3 to pH 5 (Figure 3B), which spans the PFEaccessible pH range for NiRd. The pH-independent rates suggest that the rate-determining step for catalysis does not rely on proton transfer from water or buffer, as either would be expected to increase with decreasing pH. To further investigate the nature of the rate-determining step, solvent isotope effects were measured. The qPFE approach enables global kinetic isotope effects on the rates of H2 production to be precisely measured. Interestingly, mixed solutions of FeIIRd and NiIIRd exchanged into deuterated buffers show approximately three-fold greater total protein coverage on the electrode (Figure S16). While this empirical observation is not easily rationalized, without the use of an internal redox standard, the large apparent currents would have masked any inherent decrease in activity. However, by normalizing the catalytic current values to total protein coverage, a TOFapp,D2O of 13.0 ± 2.9 s-1 at pD 4.5 is obtained. The ratio between the two TOFsapp gives an overall solvent kinetic isotope effect (KIE) of 3.81 ± 0.99 for proton reduction by NiRd. The isotope dependence implicates proton transfer in the rate-determining step, though the pH-independent rates suggest this must be an intramolecular process. The combination of these two observations has important mechanistic implications, as discussed below. The “normal” KIE observed for H+ reduction by NiRd is typical for a semiclassical, primary isotope effect, consistent with that expected for a reaction in which protons themselves are the substrate. This value also compares well to the reported solvent KIE of 2.6 for solution-phase H2 production by the P. furiosus [NiFe] H2ase87 and the electrochemical solvent KIE of 2.7 for H+ reduction by nitrogenase.88 In addition to overall solvent isotope effects, a proton inventory analysis can be employed to determine the number of protons transferred in the rate-determining process by adjusting the mole fraction of deuterated solvent (cD2O). This type of study has been instrumental in elucidating proton transfer in cytochrome c oxidase, Photosystem II, among other systems,89–91 and was recently applied to study electrocatalytic proton reduction by the MoFe nitrogenase.88 In the case of WT NiRd, a decreasing TOFapp is observed as cD2O increases, consistent with the overall KIE. However, the fit expressions for a single, double, or multiproton transfer mechanism fall within the error of the qPFE method (Figure S17). Thus, this methodology can-

not yet discern the number of protons involved in the rate-determining step, though ongoing efforts are aimed at answering this question. Variable temperature qPFE experiments reveal a large, negative entropy of activation Because protein film adsorption efficiencies and desorption rates are known to depend on temperature, extracting thermodynamic parameters for electrocatalysis has also previously presented experimental challenges. However, the qPFE method, applied across multiple electrodes and independently adsorbed films, can now be used to measure the activity of NiRd as a function of temperature. As expected, an exponential dependence of TOFapp on temperature was observed (Figure 4). An Eyring analysis provides an activation enthalpy (DH‡) of 9.5 ± 0.5 kcal/mol and an activation entropy (DS‡) of -17.9 ± 1.0 cal/(mol•K) (n = 6-12). From this, a free energy of activation (DG‡) at 293 K of 14.7 ± 0.6 kcal/mol was calculated. These values are similar to experimentally determined activation parameters for H2 oxidation by the [NiFe] H2ase and H2 production by the [FeFe] H2ases, both of which also exhibit large negative entropies of activation and only modest enthalpies of activation.92,93 The negative entropy of activation for NiRd catalysis indicates an ordered transition state, which is consistent with the intramolecular proton transfer step suggested by the pH dependence and KIE ex-

Figure 4. (A) Linear sweep voltammograms of surface-adsorbed mixtures of NiIIRd and FeIIRd at varying temperatures. Currents normalized to a coverage of 10 pmol NiRd. (Inset) Apparent turnover frequencies as a function of temperature. (B) Eyring plot of NiRd activity shown with linear fit.

periments, and differs from typical values associated with small molecule catalysts for H+ reduction.94,95 No significant effects on catalytic wave shape were seen across the range of temperatures (Figure S18), indicating potential-independent thermodynamic parameters like those seen for native hydrogenases.92,96 Quantitative simulations of the NiRd cyclic voltammograms provide elementary catalytic rate constants The qPFE voltammograms were numerically simulated using a CECEC mechanism (Figure 5, inset), with initial protonation and reduction occurring on similar timescales. By matching the shapes of the voltammograms and first derivatives, correlating current values to absolute enzyme concentration, and ensuring the simulated onset potentials and apparent TOFs agree with those observed experimentally (overlaid in Figure 3), elementary rate constants can be extracted (Figure 5). The observed -59 mV/pH unit shift in onset potential from pH 3-5 requires a pKa value between 2 and 3; pKa values lower than 2 result in calculated shifts of -90 – -120 mV/pH unit, while pKa values higher than 3 showed no shift in onset potential with pH (Figure S19). This low pKa is consistent with unsuccessful

ACS Paragon Plus Environment

Journal of the American Chemical Society 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

experimental attempts to detect significant populations of a protonated thiolate ligand using optical, Raman, and infrared spectroscopies at pH ≥ 3 (vide infra). Both ET steps were modeled using Butler-Volmer kinetics with the same reduction potential and a distribution of rate constants, as previously implemented for [NiFe] H2ases.57 An average ET rate (k0) of 75 s-1 was estimated using a Laviron analysis of the small anodic signal recovered at very fast scan rates (Figure S1). The appearance of an anodic signal is also predicted to occur in the electrochemical simulations, with a similar waveshape to that seen experimentally (Figure S1). Unfortunately, this oxidation peak is positioned on top of residual proton reduction both by NiRd and the electrode at fast scan rates, preventing use of the non-catalytic signal to extract accurate electroactive coverage values. To obtain apparent TOF values at Ean consistent with those seen experimentally for WT NiRd, rate constants of k1=130 s-1 for the intramolecular, rate-limiting step and k2=109 M-1 s-1 for the final protonation step were used, though varying k2 from 108–1010 M-1 s-1 had little effect on the generated traces (Figure S20). Other catalytic schemes were considered, including the simpler, four-step ECEC and ECCE mechanisms (Figure S21). The four-step mechanisms involving an initial chemical step (e.g., CEEC, CECE, and CCEE) or two sequential reduction steps (e.g., EECC) were excluded because these schemes either invoke the generation of a NiIV-hydride or a Ni0Rd species, both of which were considered highly unlikely. The ten other permutations of chemically reasonable, five-step mechanisms were also ruled out (Figure S22). Simulations of these mechanisms either demonstrated an isolated NiII/I couple prior to catalysis, a logarithmic dependence of turnover frequency on pH, or onset potentials that did not display a -59 mV/pH

Figure 5. (A) Averaged currents from baseline-corrected linear sweep voltammograms of NiRd from pH 3 - pH 5 (solid lines, n=6-9) shown with standard deviation (shaded areas) and electrochemical simulations overlaid (dashed lines). Currents normalized to a coverage of 10 pmol NiRd. (Inset) Catalytic scheme used in simulations. (B) Derivatives of experimental (solid lines) and simulated (dashed) voltammetric signals.

Page 6 of 28

unit shift (Figures S23-S25). More complicated mechanisms, such as the six-step, CECCEC process, were also considered (Figure S26). However, little improvement to the fit quality was seen compared to the CECEC scheme, as the voltammograms are dominated by the kinetics of the rate-determining step. These more complex mechanisms cannot be excluded but introduce additional, poorly constrained variables, motivating our selection of the five-step CECEC mechanism. The quantitative agreement between experiment and simulation is not found for any of the other four- or fivestep mechanisms, providing strong support for the mechanism shown in Figure 5.97 NiRd catalyzes H2 evolution in the presence of oxygen Oxygen is known to be a potent inhibitor of the standard [NiFe] hydrogenases for both electrocatalytic proton reduction and hydrogen oxidation, resulting in decreased catalytic currents and some degree of inactivated protein upon O2 exposure.4,10,13,14,98,99 In contrast, addition of O2-saturated buffer to the electrocatalytic cell containing NiRd-adsorbed electrodes does not impact activity (Figure 6A). While a catalytic response for O2 reduction by the electrode becomes evident at -500 mV (Figure S27), there are no effects on the magnitude of electrocatalytic current observed for H+ reduction (Figure S28). This suggests that oxygen does not interact with the reduced NiRd active site on the timescale of the electrochemical process. Because outer-sphere electron transfer from FeIIRd to O2 is readily observed, and ET from a reduced nickel state to O2 at potentials of less than -800 mV would be energetically favorable, the NiRd resistance to oxygen inactivation is attributed to kinetic rather than thermodynamic effects. The apparent oxygen tolerance can be reproduced in the electrochemical simulations by including a competitive, non-productive step for O2 binding to the NiIRd-SH state with kO2