Efficient and Selective Electrochemically Driven Enzyme-Catalyzed

Article Cite This: Acc. Chem. Res. XXXX, XXX, XXX−XXX

pubs.acs.org/accounts

Efficient and Selective Electrochemically Driven Enzyme-Catalyzed Reduction of Carbon Dioxide to Formate using Formate Dehydrogenase and an Artificial Cofactor Buddhinie S. Jayathilake,† Supriyo Bhattacharya,‡ Nagarajan Vaidehi,‡ and S. R. Narayanan*,† †

Downloaded via UNIV OF NEW ENGLAND on February 12, 2019 at 07:27:52 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Loker Hydrocarbon Research Institute, Department of Chemistry, University of Southern California, Los Angeles, California 90089, United States ‡ Department of Computational and Quantitative Medicine, Beckman Research Institute of the City of Hope, 1500 E. Duarte Road, Duarte, California 91010, United States S Supporting Information *

CONSPECTUS: Increasing levels of carbon dioxide in the atmosphere and the growing need for energy necessitate a shift toward reliance on renewable energy sources and the utilization of carbon dioxide. Thus, producing carbonaceous fuel by the electrochemical reduction of carbon dioxide has been very appealing. We have focused on addressing the principal challenges of poor selectivity and poor energy efficiency in the electrochemical reduction of carbon dioxide. We have demonstrated here a viable pathway for the efficient and continuous electrochemical reduction of CO2 to formate using the metalindependent enzyme type of formate dehydrogenase (FDH) derived from Candida boidinii yeast. This type of FDH is attractive because it is commercially produced. In natural metabolic processes, this type of metal-independent FDH oxidizes formate to carbon dioxide using NAD+ as a cofactor. We show that FDH can catalyze the reverse process to generate formate when the natural cofactor NADH is replaced with an artificial cofactor, the methyl viologen radical cation. The methyl viologen radical cation is generated in situ, electrochemically. Our approach relies on the special properties of methyl viologen as a “unidirectional” redox cofactor for the conversion of CO2 to formate. Methyl viologen (in the oxidized form) does not catalyze formate oxidation, while the methyl viologen radical cation is an effective cofactor for the reduction of carbon dioxide. Thus, although the thermodynamic driving force is favorable for the oxidized form of methyl viologen to oxidize formate to carbon dioxide, the kinetic factors are not favorable. Only the reverse reaction of carbon dioxide reduction to formate is kinetically viable with the cofactor, methyl viologen radical cation. Binding free energy calculated from atomistic molecular dynamics (MD) simulations consolidate our understanding of these special binding properties of the methyl viologen radical cation and its ability to facilitate the two-electron reduction of carbon dioxide to formate in metal-independent FDH. By carrying out the reactions in a novel three-compartment cell, we have demonstrated the continuous production of formate at high energy efficiency and yield. This cell configuration uses judiciously selected ion-exchange membranes to separate the reaction compartments to preserve the yields of the methyl viologen radical cation and formate. By the electroregeneration of the methyl viologen radical cation at −0.44 V versus the normal hydrogen electrode, we could produce formate at 20 mV negative to the reversible electrode potential for carbon dioxide reduction to formate. Our results are in sharp contrast to the large overpotentials of −800 to −1000 mV required on metal catalysts, vindicating the selectivity and kinetic facility provided by FDH. Formate yields as high as 97% ± 1% could be realized by avoiding the adventitious reoxidation of the methyl viologen radical cation by molecular oxygen. We anticipate that the insights from the electrochemical studies and the MD simulations to be useful in redesigning the metal-independent FDH and alternate artificial cofactors to achieve even higher rates of conversion. generated from solar and wind resources.3 Thus, the production of carbonaceous fuels such as formate (or formic acid), carbon monoxide, and methanol by ERC is a topic of immense scientific and technological interest.2,4−15 Of the various fuels, formic acid or formate salt is the simplest to derive from carbon dioxide.

1. INTRODUCTION Increasing levels of carbon dioxide in the atmosphere1 and the growing global demand for energy necessitates renewable methods of electricity production that can help reduce carbon emissions. Electrical energy from renewable sources can be used to make organic compounds and fuels by the electrochemical reduction of carbon dioxide (ERC).2 Producing fuels via ERC would also enable the storing and distribution of energy © XXXX American Chemical Society

Received: October 31, 2018

A

DOI: 10.1021/acs.accounts.8b00551 Acc. Chem. Res. XXXX, XXX, XXX−XXX

Article

Accounts of Chemical Research Further, formic acid is of commercial value as a chemical feedstock,16 an efficient carrier of hydrogen, and suitable for direct use in fuel cells.15,17 Thus, achieving the efficient conversion of carbon dioxide to formate using renewable energy at a large scale would be a significant step toward the goal of curtailing carbon emissions and recycling carbon dioxide. The major challenge for producing formate by ERC is the poor energy efficiency and selectivity of the process. Large energy losses result from the slow kinetics of electroreduction of CO2 to formate; the energy efficiency for formate production by ERC is in the range of 40−60%.2,3,18,19 The selectivity for formate production is also quite low because of the competitive process of electrochemical hydrogen evolution from the water.20 While the selectivity may be improved by using metal electrocatalysts such as lead, bismuth, mercury, and tin, energy efficiencies have continued to be very low.2,3,21 Enzyme-catalyzed processes are known for their selectivity and also their relatively fast kinetics.22 The use of enzymes in the chemical manufacturing industry has resulted in reduced energy usage and the production of materials of high purity.23,24 Therefore, we have focused on achieving continuous generation of formate from carbon dioxide with 100% selectivity using an enzyme-catalyzed process. In nature, the enzyme formate dehydrogenase (FDH) catalyzes the oxidation of formate to carbon dioxide supporting the metabolic processes in microorganisms and plants.25,26 Our goal was achieving biocatalysis using metal-independent FDH with redox cofactors in preference to using electron-transfer processes using mediators and metal-dependent FDH.27,28 The naturally produced redox cofactor for metal-independent FDH is nicotinamide adenine dinucleotide (NAD+/NADH).25,26 We have found that by selecting other redox cofactors that have favorable binding properties with FDH, we can reverse the process that occurs in nature.29−31 Specifically, we demonstrate the rapid and continuous conversion of CO2 to formate using a commercially available, robust variety of metal-independent type of FDH (derived from the yeast, Candida boidinii), and electrochemically generated methyl viologen radical cation (MV•+) as the redox cofactor.

(ii) Concentration of NADH higher than 0.3 mM is known to inhibit the FDH function shown in eq 1.35 (iii) To sustain a continuous process of formate production, NADH will have to be generated continuously and maintained at high concentrations. However, the continuous generation of NADH in an active form by the electroreduction of NAD+ is not promising because of the isomerization and dimerization of the reduced products.36,37 (iv) Electrochemical generation of NADH is not energy efficient because of the slow kinetics of electroreduction requiring several hundreds of millivolts of overpotential. Therefore, we shifted our focus to using other cofactors that were potentially free of these disadvantages.

3. REPLACING NADH WITH AN ARTIFICIAL REDOX COFACTOR, METHYL VIOLOGEN We overcame the challenges with NADH by using an artificial redox cofactor, methyl viologen (MV). For that, we replaced the use of NADH as a cofactor in the reverse reaction of eq 1 by the use of methyl viologen radical cation as an artificial cofactor to facilitate the enzymatic reaction, as per eq 2. FDH‐catalyzed CO2 + 2MV •+ + H+ XooooooooooooooooooooY 2MV2 + + HCOO− (2)

There are isolated reports of reversing formate oxidation with the methyl viologen radical cation as the reducing cofactor instead of NADH.29,30,38,39 In some cases, MV•+ was used as a mediator (or a reducing agent) to generate NADH.26,40 In the present study, specifically, we noted the following advantages of using MV•+ as a cofactor to carry out the reaction in eq 2: (i) MV•+ provides a favorable thermodynamic driving force for the reduction of carbon dioxide to formate, unlike NADH (SI 1). The standard reduction potential of the MV2+/MV•+ couple is 0.120 V more negative than that of the NAD+/NADH couple.41 Thus, the equilibrium constant for eq 2 is about 10 000 times larger than that with NADH (SI 1). (ii) Our studies have proven that the oxidation of formate to carbon dioxide does not occur readily with MV2+. Thus, the overall yield of formate would be preserved without the loss from reoxidation.39 (iii) MV•+ can be generated efficiently by the electrochemical reduction of MV2+ at a carbon electrode as per eq 3 without the need for any external reducing agent, a distinct advantage for continuous production. The direct electron-transfer process is fast without significant energy losses. Previous reports of using dithionite for the reduction of MV2+ to MV•+ add other byproducts to the reaction mixture that are not desirable for the subsequent enzymatic process; the direct electrochemical production of MV•+ avoids such contamination entirely.30,38,39

2. CARBON DIOXIDE REDUCTION USING METAL-INDEPENDENT FDH AND ITS NATURAL COFACTOR, NADH In the natural environment, FDH converts formate to carbon dioxide as per eq 1 using a naturally produced redox cofactor, NAD+. FDH‐catalyzed NAD+ + HCOO− XooooooooooooooooooooY NADH + CO2

(1)

To shift the equilibrium in eq 1 toward the production of formate, we could use an excess of CO2 and the reduced form of the cofactor (NADH).26,32−34 However, we encountered at least four major technical issues with this approach: (i) The equilibrium constant for the reaction as written in eq 1 is calculated to be 107 (see Supporting Information (SI) 1). Thus, shifting the equilibrium toward formate will require at least 2 orders of magnitude higher NADH concentration compared to the concentration of generated formate. Even if one is successful in producing formate by such an approach, the rapid rate of formate reoxidation to CO2, supported by the normal role of FDH, will reduce the net yield of formate.

MV2 + + e− → MV •+

E° = − 0.44 V vs NHE (3)

Cognizant of these enabling advantages of MV, we set out to address three central questions: (a) Can we use MV effectively to reverse the normal function of FDH and generate formate from carbon dioxide? (b) Can we gain mechanistic insight using molecular dynamics (MD) simulations into the effectiveness of B


Article

Accounts of Chemical Research

Figure 1. Electrochemical formate generation: (A) reactor configuration, (B) experimental setup, and (C) rate of formate generation for various concentrations of MV•+, and the continuous current density for producing MV•+ at an applied potential of −0.44 V vs NHE using 3−40 mM MV2+, 1.2 μM FDH, 15 mL of buffer solution in each chamber, and Tokuyama A901 Anion Exchange Membrane (AEM).

water to oxygen occurred (eq 5), and (iv) a region where the formate was separated from the reaction mixture. These electrochemical compartments were separated by specifically selected cation-exchange membranes (PEMs) or anionexchange membranes (AEMs) to allow for the transport of protons or hydroxide ions and to exclude anions or cations as required.

MV as a redox cofactor for FDH in the production of formate from carbon dioxide, and can we specifically predict the relative binding energies of the cofactors with FDH? (c) Can the electrochemical regeneration of the artificial redox cofactor be coupled with the enzymatic conversion by FDH to achieve a continuous and efficient process for the production of formate from carbon dioxide? To answer these questions, we have studied the FDHcatalyzed reduction of CO2 in various electrochemical reactor configurations. The studies by Amao et al. focus on developing photocatalytic pathways with sacrificial agents.29,30,38,39 Our study is focused on an electrochemical method for the generation of the cofactor and oxidation of water to oxygen to achieve FDH-catalyzed reduction of carbon dioxide to formate (eq 4).

H 2O → 1 2 O2 + 2H+ + 2e−

E° = 1.23 V vs NHE (5)

Please note that in all our experiments, we used sodium phosphate buffer (0.1 M) at the pH value of 6.6 in which FDH is optimally active, and CO2 is almost entirely in the form of the bicarbonate anion. Therefore, in all the CO2 reduction experiments, we have used bicarbonate (0.1 M) to achieve a high concentration of CO2 (eq 6), although we nominally refer to the process as the electrochemical reduction of CO2.

FDH‐catalyzed with cofactor CO2 + H 2O ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ HCOO− + H+ + 1 2 O2

(4)

H3O+ + HCO3− ⇆ CO2 + 2H 2O

Typically, these reactors had four functional sections: (i) a chamber for the reduction of CO2 to formate by FDH and MV•+ (eq 2), (ii) an electrode where the reduced form of redox cofactor, MV•+, was regenerated electrochemically (eq 3), (iii) an electrode where the conjugate electrochemical oxidation of

(6)

In the configuration shown in Figure 1A,B, the reduced form of methyl viologen, MV•+, was electrogenerated at a porous carbon fiber paper electrode (Toray Inc.) held at −0.44 V vs the normal hydrogen electrode (NHE), while oxygen evolution C


Article


electron donor and NADH being a hydride donor. Further, unlike NADH, two molecules of MV•+ are needed to reduce CO2 to formate, and it is not evident where the second molecule of MV•+ would bind in FDH. Thus, we found it is important to understand the molecular interactions of the two cofactors with FDH to explain the differences in their effectiveness. Therefore, we conducted ligand docking and MD simulations to study the binding of the MV2+/MV•+ and NAD+/NADH redox couples to FDH in the presence of formate/bicarbonate.

occurred on a platinum electrode in chamber A. For every mole of methyl viologen that was reduced, we can expect 0.25 mol of oxygen to be evolved. Electrochemical production of intensely blue-colored MV•+ was monitored with time using a carbon fiber microelectrode (Figure S1).42 MV•+ was allowed to react with FDH and CO2 to produce formate in chamber B (Figure 1B). The rate of formate production as followed by 1H NMR increased with the concentration of MV•+, reaching a maximum value at a concentration of 20 mM. Correspondingly, the steadystate current required to support the electrogeneration of MV•+ also increased and reached a plateau. The rate of formate production at various concentrations of MV•+ followed Michaelis−Menten-type kinetics. These results demonstrated the technical feasibility of producing formate from carbon dioxide using FDH and electrochemically-generated MV•+. Further, we proved that the MV•+ was a viable cofactor to reverse the natural function of FDH and produce formate from carbon dioxide.

4.2. Binding Free Energy Re-Affirms the Thermodynamic Feasibility of the Bicarbonate Conversion by Methyl Viologen and FDH

Our goal in this section is to provide mechanistic insights into the binding energetics of the reactants and products shown in eq 7 (the net reaction of eqs 2 and 6 at neutral pH) and study the effect of the protein dynamics on the stability and flexibility of the reactants and products. FDH‐catalyzed 2MV •+ + H3O+ + HCO3− ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯→ 2MV2 +

4. METHYL VIOLOGEN IS A UNIDIRECTIONAL COFACTOR FOR FDH

+ OH− + HCOO− + H 2O

4.1. Experimental Evidence

(7)

We used ligand docking and MD simulation techniques as described in SI 4. Briefly, starting from the crystal structure of the FDH bound to NAD+ and azide ion (PDB ID 5DN9), MV•+ was docked to the FDH active site (the docking box in Figure S3 in SI 4). We replaced the azide ion with the HCO3− (Figure 3A) and docked H3O+ to the active site with MV•+ and HCO3− present. The best pose for H3O+ was located between MV•+ and HCO3− (Figure 3A). R258, the highly conserved residue in the FDH family, and H311, which was implicated by mutagenesis to be involved in the catalytic activity of FDH, are both present clamping the HCO3− in our structural model.37 MV•+ occupies the binding site FDH and makes similar residue contacts as the nicotinamide moiety of NADH. These consistencies in the docked structures indirectly validate our structural models. We used the same docked positions and converted the reactants to products, MV2+ and HCOO−, in place of MV•+ and HCO3−, respectively. The OH− was docked separately in the presence of the rest of the products. Subsequently, we performed atomistic MD simulations on FDH bound to MV and NADH as cofactors, in the presence of the reactants and products. During MD simulations, the HCO3− mostly stayed sandwiched between H311 and R258, as can be seen from the distance distribution (Figure 3B). However, in the MD simulations of NADH-bound FDH, the bicarbonate ion is distant from both H311 and R258 (distances 10.5 and 8.5 Å) and oriented away from the plane of the nicotine ring (Figure 3C). The HCO3− is also highly dynamic in the presence of NADH compared to that by MV•+ (Figure 3D), suggesting its instability in NADH-bound FDH. Thus, the forward reaction of bicarbonate to formate is disfavored by NADH compared to MV•+. Using the Molecular Mechanics Generalized Born Surface Area method (MMGBSA), we calculated the difference in binding free energy of the products and reactants in FDH, with MV•+ and NADH as cofactors. As shown in Figure 3E, with MV•+ as a cofactor, the binding free energy favors the binding of reactants by 25 kcal/mol compared to the products. The weaker binding free energy of the products in the presence of MV2+ will facilitate the forward reaction of bicarbonate reduction by efficient removal of the reaction products from the active site of FDH. In contrast, the products in the NADH-based reaction (NAD+ and formate) show better binding energy than the

While there are some reports in the literature to suggest that the oxidation of formate to carbon dioxide does not occur with FDH and the MV2+,39 it was important for us to verify this property of MV2+. We combined FDH with the MV2+ and formate in various concentrations and analyzed the decrease in concentration of formate. No change in the concentration of formate was detectable even after 30 h (Figure 2). UV−visible spectroscopic

Figure 2. Effect of 20 mM MV2+ on the concentration of sodium formate (11−44 mM) in the presence of 4.3 μM FDH.

studies also showed no decrease in the concentration of the MV2+ nor the appearance of MV•+ (SI 3 Figure S2). Thus, we proved that MV2+ is not a viable cofactor for the oxidation of formate to carbon dioxide, unlike NAD+ that oxidizes formate to carbon dioxide quite efficiently. Therefore, the yield of formate from the reduction of carbon dioxide would be preserved without loss by reoxidation by using MV as the artificial cofactor. The unique property of MV as a redox cofactor that supports only carbon dioxide reduction is crucial for achieving the high efficiency of formate production. This behavior is consistent with the favorable thermodynamic driving force for carbon dioxide reduction over formate oxidation using MV, as determined from the equilibrium constant value (SI 1).43,44 Although the equilibrium constant favored the reduction of CO2 to formate with MV•+, we expect the binding properties of MV to FDH to also play a key role in facilitating the catalytic reaction. We also want to rationalize the differences between MV•+ and NADH as redox cofactors, with MV•+ being just an D


Article


Figure 3. (A, C) Orientation of bicarbonate ion in the representative conformation from the highest populated cluster in the MD ensemble in the presence of hydronium ions and (A) MV•+ (green) or (C) NADH (green); major active site residues shown in magenta; (B, D) probability density of bicarbonate position as a function of distance with R258 and H311 as sampled during the MD simulations of (B) MV•+ bound FDH and (D) NADH bound FDH; for each residue, the minimum distance of bicarbonate oxygens with the side chain polar nitrogen atoms was measured. (E) Difference in binding free energy between the reactants and products of bicarbonate reduction in FDH in the presence of MV•+ and NADH.

Figure 4. Comparison of the reactant and product orientation in the FDH active site in MV facilitated bicarbonate reduction. (A) Orientation of bicarbonate and hydronium ions in the presence of MV•+ (green) and residues making hydrogen bonds with the reactants (blue). (B) Orientation of the products formate and hydroxyl ions in the presence of MV2+ (green) and residues that formed hydrogen bonds with the reactants in the MV•+ bound FDH (blue); these hydrogen bonds are disrupted upon product formation, as shown by the increased distances with the formate.

respectively. The hydronium ion is stabilized between the bicarbonate and the pyridine ring of MV•+ and is 4 Å away from the nitrogen of the pyridine. This binding facilitates the conversion of bicarbonate to formate by stabilizing the reactants (Figure 4A). In contrast, in the simulations of MV-bound FDH (Figure 4B), the products hydroxyl and formate ions leave the binding site after 6 and 50 ns, respectively, of the MD simulations. These simulations indicate that for MV, the reactant binding is favored over that of the products.

reactants, indicating high stability of the reaction products, which hinders the forward reaction. These results validated the thermodynamic nature of the HCO3− reduction with MV•+. We have used these simulation results to provide further insights into the mechanism of action of MV in FDH catalysis. 4.3. Relative Stability and Flexibility of the Reactants and Products in FDH

We clustered the conformations obtained from the MD simulations of the reactants and products bound to FDH independently by their root-mean-square deviation (RMSD) in coordinates (SI 4). We used the most populated conformation cluster for all further analysis presented in this section. We observed that the distances between the HCO3−, H3O+, and MV•+ are optimal for facilitating the proton-coupled electron transfer to the HCO3− ion. During the MD simulations of FDH bound to MV•+, HCO3− and H3O+, the bicarbonate ion is held in place by a hydrogen bond and salt bridge to H311 and R258,

4.4. Involvement of a Second MV•+ in the Catalytic Reaction

The two electrons required for converting bicarbonate to formate should come from two molecules of MV•+ involved in the reaction.38 To identify the binding site of the second molecule of MV•+, we docked a second MV•+ to FDH in the presence of the first MV•+, HCO3−, and H3O+ in the active site and performed MD simulations of the docked conformation. The second MV•+ molecule was bound at the entry of the active E


Article


Figure 5. (A) Cross section of FDH showing the binding site of two bound MV•+ molecules (green), the aromatic residues making contact with the second MV•+ in the secondary pocket (magenta), the major active site residues involved in catalysis (blue), and reactants bicarbonate and hydronium (green). (B) Magnified view of the FDH cross section near the active and secondary sites and (C) possible electron transfer pathway from the second MV•+ molecule.

Figure 6. Electrochemical formate generation: (A) Illustration of the two-chamber electrochemical reactor and (B) Results of continuous accumulation of formate from electrochemical reduction of carbon dioxide in the two-chamber reactor at an applied potential of −0.44 V vs NHE using 40 mM MV2+, 2.6 μM FDH, 7.5 mL of buffer solution in each chamber, and Nafion 117 PEM.

site of FDH (Figure 5A), and it stayed stable in this position during the MD simulations. As shown in Figure 5B, several

aromatic and hydrophobic residues in this surface pocket interact with the second MV•+. Residues H232 and Y358 π-stack F


Article


Figure 7. Electrochemical formate generation: (A) Illustration and (B) experimental setup showing diffusion of MV across PEM and (C) Steady-state CO2 reduction in the novel three-chamber reactor at an applied potential of −0.44 V vs NHE using 40 mM MV2+, 2.4 μM FDH, 7.5 mL of buffer solution in each chamber, Tokuyama A901 AEM, and Nafion 117 PEM.

with the pyridine rings of MV•+. The distance between the two MV•+ molecules is about 8.5 Å. We speculate that the aromatic residues could be involved in electron transfer to the HCO3−. Based on the structure of the FDH binding pocket, we can expect that Y358 π-stacks with the second MV•+ to transfer electrons onto the neighboring D96. D96 is involved in a salt bridge network with R258 and bicarbonate as shown in Figure 5C. Although the distance between Y358 and D96 is 5.5 Å, we can expect that the bridging water molecules between these two residues (as observed in the MD simulations) can facilitate the electron transfer.

production of MV•+ (SI 5). The continuous electroregeneration of MV•+ occurred at −0.44 V vs NHE, followed by the FDHcatalyzed carbon dioxide reduction to formate. Since the equilibrium potential value for carbon dioxide reduction to formate at pH 6.6 is −0.42 V vs NHE (SI 1), we can conclude that the conversion to formate using FDH and MV•+ can be achieved at an overpotential of 20 mV. This overpotential loss is very small compared to the several hundreds of millivolts of overpotential encountered during carbon dioxide reduction at metal electrodes. We found that in the configuration presented in Figure 6A with a PEM, MV•+ and MV2+ could crossover from the enzyme chamber to the counter electrode, as the PEM allows for the facile transport of cations. Consequently, the concentration of MV•+ and MV2+ in the enzyme chamber decreased with time. Therefore, we took further steps to prevent the crossover of MV•+ and MV2+ to the counter electrode chamber by introducing an AEM between the oxygen-evolving counter electrode and the MV-reducing electrode (Figure 7). Such a membrane would block the transport of cations, allowing only anions to be transported. Further, in this new configuration (Figure 7A), we also introduced a PEM between the enzyme chamber where formate was produced and chamber B where the MV•+ was produced. This additional separation ensured that formate did not leave the enzyme chamber. Thereby, the MV2+ and the MV•+ could freely diffuse between chambers B and C through the PEM (Figure 7B) and participate in the conversion of carbon dioxide to formate. We verified that

5. CONTINUOUS REDUCTION OF CARBON DIOXIDE TO FORMATE IN TWO-CHAMBER AND THREE-CHAMBER ELECTROLYZER CONFIGURATIONS 5.1. Formate Accumulation under Continuous Steady-State Carbon Dioxide Reduction

The experimental results in Figure 1C and Figure 2 and the MD simulations confirmed the feasibility of the continuous accumulation of formate by the reduction of carbon dioxide using FDH and the artificial cofactor MV•+. In a two-chamber electrochemical reactor with a PEM (Figure 6), we could demonstrate the continuous production and accumulation of formate in the enzymatic chamber. The concentration of formate in chamber B (Figure 6A) increased linearly (Figure 6B) with time, consistent with the constant current used for the G


Article

Accounts of Chemical Research no MV2+ or MV•+ crossed over to chamber A, even after 21 h of electrolysis (Figure 7B). While the rate of transport of MV•+ into chamber C was dependent on the membrane thickness and the concentration of MV•+ in chamber B, the steady-state generation of formate in chamber C was governed by the rate of the carbon dioxide reduction reaction with FDH and the MV•+. Under steady-state conditions, the current required to maintain the production of MV•+ in chamber B is equal to both the mass flux of MV•+ into chamber C and the rate of consumption of MV•+ in chamber C for the conversion of carbon dioxide to formate. Thus, upon the addition of FDH and carbon dioxide to chamber C, we could observe an immediate jump in the value of the reduction current, indicating a sudden increase in the rate of consumption of MV•+ resulting from the reduction of carbon dioxide (Figure 7C). In 2 h after introducing FDH and CO2 into chamber C, the formate concentration rose to 0.72 ± 0.01 mM. The rate of carbon dioxide reduction per unit concentration of the enzyme was calculated to be 75 ± 3 h−1 (SI 6), and this value compares well with the kcat value of FDH that has been reported by Ikeyama et al. (61 h−1, based on the Michaelis−Menten equation) in the photochemical reduction of carbon dioxide using FDH and MV•+.38 In all the studies discussed here, the regeneration of the MV•+ was carried out at just a few millivolts away from the reversible potential for the electrochemical reduction of CO2 to formate. Thus, the conversion to formate could be achieved at high energy efficiency. Energy losses that result from the oxygen evolution reaction are small with catalytic electrodes based on iron-doped nickel hydroxide, well-studied by us and others.45,46

CO2 reduction experiment was recalculated to be 97% ± 1%. By introducing a glass frit that restricted the oxygen transport, we could increase the CE of MV•+ generation to 96% ± 1% (Table 2). In the future, we intend to implement a glass frit in conjunction with an AEM to separate chambers A and B (Figure 7). In yet another modification to realize 100% CE, we completely avoided the oxygen evolution reaction by using a magnesium metal strip as the counter electrode. Thus, instead of oxygen evolution, magnesium dissolution to magnesium ions was the reaction at the counter electrode, while MV was reduced at the working electrode. Under these conditions, the CE of MV reduction was 98% ± 1%. While the use of a magnesium strip proved the ability to reach high Coulombic efficiencies, magnesium is a sacrificial electrode and is not recommended because of the effects of dissolved magnesium ions on the electrolyte pH and deactivation of the enzyme.

6. CONCLUSIONS We have demonstrated a novel approach to the continuous reduction of carbon dioxide to formate by reversing the normal function of metal-independent FDH in the wild type, using an electrochemically-generated artificial cofactor, the MV•+. This method of electrochemical reduction of carbon dioxide relies on MV being an effective redox cofactor for the reduction of carbon dioxide but not for the oxidation of formate. Calculation of binding free energies from atomistic MD simulations showed that the wild-type FDH binds to the MV•+, bicarbonate, and the hydronium ion in preference to the products, namely, MV2+ and formate. Additionally, we observed that the cofactor NADH does not favor binding of the reactant, bicarbonate. The MD simulation results also provided insights into how a second MV•+ could bind at the entrance to the active site of FDH and provide the second electron required for the reaction to be completed. We have demonstrated the continuous production of formate at high energy efficiency and yield by carrying out the reactions in a novel three-compartment cell configuration. By the careful selection of ion-exchange membranes to separate these compartments, we could preserve the yields of MV•+ and formate. We have shown that yields as high as 97% ± 1% can be realized by avoiding adventitious reoxidation of MV•+ by molecular oxygen. By the electroregeneration of MV•+ at −0.44 V vs NHE, we could produce formate at just 20 mV negative to the reversible electrode potential for carbon dioxide reduction to formate. Our results are in sharp contrast to the large overpotentials of −800 to −1000 mV required on metal catalysts. We anticipate the insights from the electrochemical studies and the MD simulations to be useful in redesigning FDH and the artificial cofactor to achieve even higher rates of conversion.

5.2. Formate Yield

By comparing the rate of production of formate with the total charge passed for the reduction of MV, the yield of formate was determined to be 61% ± 1% (Table 1). As no product other than Table 1. Yield of Formate in the Three-Compartment Cell Shown in Figure 7 expected formate concentration, mM

detected formate concentration, mM

formate yield %

1.180 ± 0.005

0.72 ± 0.01

61 ± 1

formate is generated, we attributed the reduced yield to the diffusion of oxygen from chamber A to chamber B that results in the parasitic oxidation of the MV•+. We estimated the oxygen crossover rate to be 0.43 ± 0.01 μmol/cm2/h. For attaining 100% Coulombic efficiency (CE) in this reactor, we must avoid the diffusion of oxygen completely from chamber A to chamber B in addition to the loss of formate by crossover. Independent studies of the CE for the production of MV•+ (without any CO2 reduction) using different electrolyte membranes (Table 2) verified that the reaction of the MV•+ with in situ generated molecular oxygen is the most significant parasitic loss leading to the reduction of CE of CO2 reduction. After accounting for the loss of MV•+ by reaction with oxygen, the formate yield in the

■

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.accounts.8b00551.

Table 2. Coulombic Efficiency of Methyl Viologen Reduction counter electrode reaction

separator

coulombic efficiency (±1%)

oxygen evolution oxygen evolution magnesium dissolution

AEM 901 glass frit AEM 901

63 96 98

ASSOCIATED CONTENT

Thermodynamics of FDH catalyzed CO2 reduction with MV•+ or NADH, detection of MV2+, UV/vis detection of MV•+ generation during formate oxidation, crystal structure of cbFDH bound to NAD+ and azide, docking of MV, bicarbonate, and hydronium ions to FDH, MD simulation of MV and NADH bound to FDH, clustering H


Article


■

(11) Li, Y. C.; Yan, Z.; Hitt, J.; Wycisk, R.; Pintauro, P. N.; Mallouk, T. E. Bipolar Membranes Inhibit Product Crossover in CO2 Electrolysis Cells. Adv. Sustain. Sys. 2018, 2, 1700187. (12) Kuhl, K. P.; Cave, E. R.; Abram, D. N.; Jaramillo, T. F. New insights into the electrochemical reduction of carbon dioxide on metallic copper surfaces. Energy Environ. Sci. 2012, 5, 7050−7059. (13) Zhang, W.; Hu, Y.; Ma, L.; Zhu, G.; Wang, Y.; Xue, X.; Chen, R.; Yang, S.; Jin, Z. Progress and perspective of electrocatalytic CO2 reduction for renewable carbonaceous fuels and chemicals. Adv. Sci. 2018, 5, 1700275. (14) Apaydin, D. H.; Schlager, S.; Portenkirchner, E.; Sariciftci, N. S. Organic, Organometallic and Bioorganic Catalysts for Electrochemical Reduction of CO2. ChemPhysChem 2017, 18, 3094−3116. (15) Olah, G. A.; Goeppert, A.; Prakash, G. S. Beyond oil and gas: the methanol economy; John Wiley & Sons, 2011. (16) Reutemann, W.; Kieczka, H. Ullmann’s Encyclopedia of Industrial Chemistry; Wiley, 2011; 13−33. (17) Rees, N. V.; Compton, R. G. Sustainable energy: a review of formic acid electrochemical fuel cells. J. Solid State Electrochem. 2011, 15, 2095−2100. (18) Lee, W.; Kim, Y. E.; Youn, M. H.; Jeong, S. K.; Park, K. T. Catholyte-Free Electrocatalytic CO2 Reduction to Formate. Angew. Chem., Int. Ed. 2018, 57, 6883−6887. (19) Del Castillo, A.; Alvarez-Guerra, M.; Solla-Gullón, J.; Sáez, A.; Montiel, V.; Irabien, A. Sn nanoparticles on gas diffusion electrodes: Synthesis, characterization and use for continuous CO2 electroreduction to formate. J. CO2 Util. 2017, 18, 222−228. (20) Feaster, J. T.; Shi, C.; Cave, E. R.; Hatsukade, T.; Abram, D. N.; Kuhl, K. P.; Hahn, C.; Nørskov, J. K.; Jaramillo, T. F. Understanding selectivity for the electrochemical reduction of carbon dioxide to formic acid and carbon monoxide on metal electrodes. ACS Catal. 2017, 7, 4822−4827. (21) Narayanan, S.; Haines, B.; Soler, J.; Valdez, T. Electrochemical conversion of carbon dioxide to formate in alkaline polymer electrolyte membrane cells. J. Electrochem. Soc. 2011, 158, A167−A173. (22) Blumenfeld, L. A.; Tikhonov, A. N. Principles of Enzyme Catalysis. Biophysical Thermodynamics of Intracellular Processes; Springer, 1994; pp 86−111. (23) Schmid, A.; Dordick, J. S.; Hauer, B.; Kiener, A.; Wubbolts, M.; Witholt, B. Industrial biocatalysis today and tomorrow. Nature 2001, 409, 258. (24) Bhosale, S. H.; Rao, M. B.; Deshpande, V. V. Molecular and industrial aspects of glucose isomerase. Microbiol rev 1996, 60, 280− 300. (25) Davison, D. C. Studies on plant formic dehydrogenase. Biochem. J. 1951, 49, 520. (26) Reda, T.; Plugge, C. M.; Abram, N. J.; Hirst, J. Reversible interconversion of carbon dioxide and formate by an electroactive enzyme. Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 10654−10658. (27) Sakai, K.; Kitazumi, Y.; Shirai, O.; Kano, K. Bioelectrocatalytic formate oxidation and carbon dioxide reduction at high current density and low overpotential with tungsten-containing formate dehydrogenase and mediators. Electrochem. Commun. 2016, 65, 31−34. (28) Sakai, K.; Kitazumi, Y.; Shirai, O.; Takagi, K.; Kano, K. Efficient bioelectrocatalytic CO2 reduction on gas-diffusion-type biocathode with tungsten-containing formate dehydrogenase. Electrochem. Commun. 2016, 73, 85−88. (29) Ikeyama, S.; Amao, Y. The effect of the functional ionic group of the viologen derivative on visible-light driven CO2 reduction to formic acid with the system consisting of water-soluble zinc porphyrin and formate dehydrogenase. Photoch. Photobio. Sci. 2018, 17, 60. (30) Ikeyama, S.; Amao, Y. An Artificial Co-enzyme Based on the Viologen Skeleton for Highly Efficient CO2 Reduction to Formic Acid with Formate Dehydrogenase. ChemCatChem 2017, 9, 833−838. (31) Ikeyama, S.; Amao, Y. Novel electron carrier molecule based on viologen derivative for visible light-driven CO2 reduction to formic acid with the system of zinc porphyrin and formate dehydrogenase. Sustain Energ Fuel 2017, 1, 1730−1733.

of trajectories and calculation of binding free energy, and steady-state reduction of CO2 (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Nagarajan Vaidehi: 0000-0001-8100-8132 S. R. Narayanan: 0000-0002-7259-3728 Notes

The authors declare no competing financial interest. Biographies Buddhinie S. Jayathilake is a Ph.D. candidate focusing on electrochemical energy conversion and energy storage systems. Supriyo Bhattacharya’s research interest focuses on protein dynamics and allostery. Nagarajan Vaidehi’s focus includes developing constrained molecular dynamics methods and application to protein structural dynamics. Sri R. Narayanan focuses on sustainable pathways to energy conversion and storage using electrochemical methods.

■

ACKNOWLEDGMENTS Authors B.S.J. and S.R.N. acknowledge the Loker Hydrocarbon Research Institute and USC for the financial support of this work. S.B. and N.V. acknowledge the support by the National Cancer Institute under award number P30CA033572.

■

REFERENCES

(1) Trends in Atmospheric Carbon Dioxide. www.esrl.noaa.gov/ gmd/ccgg/trends/?utm_source=www.uoota.com. (2) Jones, J. P.; Prakash, G.; Olah, G. A. Electrochemical CO2 reduction: recent advances and current trends. Isr. J. Chem. 2014, 54, 1451−1466. (3) Jhong, H.-R.; Kenis, P. J.; Ma, S. Electrochemical conversion of CO2 to useful chemicals: current status, remaining challenges, and future opportunities. Curr. Opin. Chem. Eng. 2013, 2, 191−199. (4) Drennan, C. L.; Heo, J.; Sintchak, M. D.; Schreiter, E.; Ludden, P. W. Life on carbon monoxide: X-ray structure of Rhodospirillum rubrum Ni-Fe-S carbon monoxide dehydrogenase. Proc. Natl. Acad. Sci. U. S. A. 2001, 98, 11973−11978. (5) Dobbek, H.; Svetlitchnyi, V.; Gremer, L.; Huber, R.; Meyer, O. Crystal structure of a carbon monoxide dehydrogenase reveals a [Ni4Fe-5S] cluster. Science 2001, 293, 1281−1285. (6) Kuk, S. K.; Singh, R. K.; Nam, D. H.; Singh, R.; Lee, J.-K.; Park, C. B. Photoelectrochemical Reduction of Carbon Dioxide to Methanol through a Highly Efficient Enzyme Cascade. Angew. Chem., Int. Ed. 2017, 56, 3827−3832. (7) Kuwabata, S.; Tsuda, R.; Yoneyama, H. Electrochemical conversion of carbon dioxide to methanol with the assistance of formate dehydrogenase and methanol dehydrogenase as biocatalysts. J. Am. Chem. Soc. 1994, 116, 5437−5443. (8) Aresta, M.; Dibenedetto, A.; Baran, T.; Angelini, A.; Łabuz, P.; Macyk, W. An integrated photocatalytic/enzymatic system for the reduction of CO2 to methanol in bioglycerol−water. Beilstein J. Org. Chem. 2014, 10, 2556−2565. (9) Rusching, U.; Müller, U.; Willnow, P.; Höpner, T. CO2 reduction to formate by NADH catalysed by formate dehydrogenase from Pseudomonas oxalaticus. Eur. J. Biochem. 1976, 70, 325−330. (10) Miyatani, R.; Amao, Y. Bio-CO2 fixation with formate dehydrogenase from Saccharomyces cerevisiae and water-soluble zinc porphyrin by visible light. Biotechnol. Lett. 2002, 24, 1931−1934. I


Article

Accounts of Chemical Research (32) Hwang, H.; Yeon, Y. J.; Lee, S.; Choe, H.; Jang, M. G.; Cho, D. H.; Park, S.; Kim, Y. H. Electro-biocatalytic production of formate from carbon dioxide using an oxygen-stable whole cell biocatalyst. Bioresour. Technol. 2015, 185, 35−39. (33) Peck, H. D.; Gest, H. Formic dehydrogenase and the hydrogenlyase enzyme complex in coli-aerogenes bacteria. J. Bacteriol. 1957, 73, 706−721. (34) Choe, H.; Ha, J. M.; Joo, J. C.; Kim, H.; Yoon, H.-J.; Kim, S.; Son, S. H.; Gengan, R. M.; Jeon, S. T.; Chang, R.; et al. Structural insights into the efficient CO2-reducing activity of an NAD-dependent formate dehydrogenase from Thiobacillus sp. KNK65MA. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2015, 71, 313−323. (35) Kim, S.; Kim, M. K.; Lee, S. H.; Yoon, S.; Jung, K.-D. Conversion of CO2 to formate in an electroenzymatic cell using Candida boidinii formate dehydrogenase. J. Mol. Catal. B: Enzym. 2014, 102, 9−15. (36) Jensen, M. A.; Elving, P. J. Nicotinamide adenine dinucleotide (NAD+). Formal potential of the NAD+/NAD· couple and NAD· dimerization rate. Biochim. Biophys. Acta, Bioenerg. 1984, 764, 310−315. (37) Srikanth, S.; Maesen, M.; Dominguez-Benetton, X.; Vanbroekhoven, K.; Pant, D. Enzymatic electrosynthesis of formate through CO2 sequestration/reduction in a bioelectrochemical system (BES). Bioresour. Technol. 2014, 165, 350−354. (38) Ikeyama, S.; Amao, Y. Novel Artificial Coenzyme Based on the Viologen Derivative for CO 2 Reduction Biocatalyst Formate Dehydrogenase. Chem. Lett. 2016, 45, 1259−1261. (39) Amao, Y.; Ikeyama, S. Discovery of the Reduced Form of Methylviologen Activating Formate Dehydrogenase in the Catalytic Conversion of Carbon Dioxide to Formic Acid. Chem. Lett. 2015, 44, 1182−1184. (40) Parkinson, B. A.; Weaver, P. F. Photoelectrochemical pumping of enzymatic CO2 reduction. Nature 1984, 309, 148−149. (41) Karyakin, A. A.; Ivanova, Y. N.; Karyakina, E. E. Equilibrium (NAD+/NADH) potential on poly(Neutral Red) modified electrode. Electrochem. Commun. 2003, 5, 677−680. (42) Jayathilake, B. S.; Narayanan, S. R. Bio-Electrocatalytic CO2 Reduction into Formate Using Metal-Independent Formate Dehydrogenase from Candida Boidinii (Yeast). Meeting Abstracts 2017, 45, 1995−1995. (43) Berg, J.; Tymoczko, J.; Stryer, L.; Stryer, L. Biochemistry, 5th ed.; WH Freeman, New York, 2002; pp 507−507. (44) Ito, M.; Kuwana, T. Spectroelectrochemical study of indirect reduction of triphosphopyridine nucleotide: I. Methyl viologen, ferredoxin-TPN-reductase and TPN. J. Electroanal. Chem. Interfacial Electrochem. 1971, 32, 415−425. (45) Mitra, D.; Narayanan, S. R. A Stable and Electrocatalytic Iron Electrode for Oxygen Evolution in Alkaline Water Electrolysis. Top. Catal. 2018, 61, 591−600. (46) Mitra, D.; Trinh, P.; Malkhandi, S.; Mecklenburg, M.; Heald, S.; Balasubramanian, M.; Narayanan, S. R. An Efficient and Robust Surface-Modified Iron Electrode for Oxygen Evolution in Alkaline Water Electrolysis. J. Electrochem. Soc. 2018, 165, F392−F400.

J


Efficient and Selective Electrochemically Driven Enzyme-Catalyzed

Recommend Documents