Formate dehydrogenase from Candida boidinii (FDH, EC.1.2.1.2, homo-dimer, 76 kDa), formaldehyde dehydrogenase from Pseudomonas sp. (FaldDH, EC.1.2.1.46, homo-dimer, 150 kDa), yeast alcohol dehydrogenase (ADH, EC 1.1.1.1, 141 kDa), reduced and oxidized nicotinamide adenine dinucleotide (NADH/NAD⁺, 98 wt%), L-Histidine (99%), glycerol (99%), L-glutamic acid (99%), L-serine (99%), L-arginine (99%), dopamine hydrochloride (DA), poly (ethyleneimine) (PEI), 2,2′-bipyridyl-5,5′-dicarboxylic acid, and dichloro-(penta-methylcyclopentadienyl)rhodium (III) dimer ([Cp*RhCl₂]₂) were purchased from Sigma-Aldrich (St Louis, MO, United States). CO₂ gas (>99.5%) in a cylinder was purchased from Linde Gas (Sweden).

NADESs were synthesized with the method referring to the literature (Ren et al., 2018). Taking SerGly as an example, SerGly was synthesized by mixing L-serine with glycerol at a molar ratio of 1:6, and the mixture was heated at 70°C until the liquid changed from colorless to colored. The unreacted amino acid was removed by centrifugation, and the obtained product was dried under vacuum at 60°C for 2 h. Similarly, other NADESs were synthesized following the same procedure, and the obtained NADESs were stored in a glassware dryer at ambient temperature. Finally, the yields of GluGly, SerGly, ArgGly, and HisGly were 92, 95, 100, and 89%, respectively.

The FDH activity was determined by monitoring the absorbance changes at 340 nm during the redox reactions catalyzed by the FDH at 25°C. The oxidation of formate was conducted in the conventional buffer and NADES-contained solution, respectively. The assay solution (2 ml) contains 20 μg of FDH, 100 mM sodium formate, and 1 mM NAD⁺. One unit of oxidation activity was defined as the amount of enzyme required to produce 1 μmol of NADH per minute under standard conditions.

Electrochemical regeneration of NADH was carried out using the CHI-760e electrochemical workstation in a conventional three-electrode H-cell. The Rh complex-grafted carbon felt fabricated according to the previous work (Zhang et al., 2021c) served as the working electrode, and the platinum wire and Ag/AgCl/KCl (3 M) were used as the counter and reference electrodes, respectively. The anodic and cathodic chambers were filled with 10 ml Tris-HCl (50 mM) buffer and degassed using N₂ to prevent the oxidation of NADH. 1 mM NAD⁺ was added to the solution and then reduced to NADH by using the Rh-grafted electrode. The progress of NAD⁺ reduction was monitored with a spectrophotometer (Shimadzu) by measuring the absorbance at 340 nm. Cyclic voltammetry (CV) measurements were recorded at a scan rate of 50 mV⋅s⁻¹ and the potential range is from −1 to 0 V. The electrochemical impedance spectroscopic (EIS) was carried out at open circuit potential (OCP) within the frequency extent of 100–100,000 Hz and at an amplitude of 5 mV.

CO₂ absorption was determined by weighting the samples (Yan et al., 2020). Typically, 5 g of NADES-based solution was added into an absorption glass tube with an inner diameter of 2.0 cm. Then, the absorption tube was partially immersed in a water bath at the desired temperature, and the standard uncertainty of temperature was ±0.1°C. After that, CO₂ was passed into the absorption tube at a flow rate of 100 ml/min. The weight of the captured CO₂ was obtained by the electronic balance with an accuracy of ±0.0001 g until CO₂ absorption in the DESs reached an equilibrium.

The conversion of CO₂ to methanol was performed in the setup of the H-cell with three-electrodes. A mixture (10 ml) of 1 mg FDH, 1 mg FaldDH, 1 mg ADH, and 1 mM NADH (7.09 mg) in the cathode was prepared and used in all enzymatic reactions. Before reaction, CO₂ gas was bubbled through 10 ml buffer solution (50 mM Tris-HCl buffer, pH 7.4) for 30 min to achieve CO₂ saturation in the solution. Applying for a reduction peak potential, the electro-enzymatic reaction was initiated, and the methanol concentration was determined by GC using the internal standard method.

An Agilent 7890B gas chromatograph (GC) equipped with a flame ionization detector (250°C) and an HP-5 column (30 m × 0.25 mm, film thickness 0.25 μm) was used for analyzing the products in the liquid phase and determining the methanol concentration by using ethyl acetate as the internal standard. The NADH concentration was measured by the UV-vis spectrophotometer (UV-1280, Shimadzu). Thermogravimetric analyses (TGA) were performed using a thermal gravimetric analyzer (Netzsch, STA449 F5, Germany) and the samples were heated from 30 to 400°C at 20.0°C/min under the nitrogen atmosphere. The freezing point of each DES was determined by a differential scanning calorimeter (DSC) (Netzsch DSC 200F3, Germany), and the samples were heated from −78 to 50°C at 10.0°C/min. The density and viscosity were measured by the Anton Paar DMA 5000 density meter with an uncertainty of ±0.0005 g cm⁻³ and Anton Paar AMVn with ±0.5% precision.

In the synthesis of NADESs, it typically contains hydrogen bonding acceptor (HBA) and hydrogen bond donor (HBD), wherein the amino acid plays as hydrogen bonding acceptor (HBA), and glycerol serves as hydrogen bond donor (HBD). Taking glutamic acid as an example, the reaction between HBA and HBD was proposed in Supplementary Figure S1. To confirm the NADES formation, the synthesis process of NADESs was monitored by a UV-vis spectrometer in the wavelength ranging from 250 to 650 nm. In the beginning, the raw materials, amino acid and glycerol, do not have any absorption from 250 to 650 nm. As the reaction progressed, due to the formation of NADES, the absorption peak was observed for each NADES as shown in Figure 1A. Besides, the freezing point of NADESs was further detected by DSC. According to Figure 1B, the freezing points of GluGly (Supplementary Figure S2), SerGly, ArgGly, and HisGly were estimated to be −61, −72, −60, and −75°C, respectively, which are lower than those of amino acids (Glu: 199, Ser: 225, Arg: 260, and His: 277°C) and glycerol (17.8°C) (Do et al., 2020), also confirming the formation of NADESs successfully. The thermal stabilities of NADESs were analyzed by TG, as shown in Figure 1C. There is no obvious weight loss below 150°C for all the NADESs, indicating their high thermal stability. Also, these NADESs presented different thermal stabilities. Taking the mass loss of 50% as an example, the corresponding temperatures of GluGly, SerGly, ArgGly, and HisGly were respectively 178, 212, 214, and 228°C, and thus their thermal stabilities follow the order of ArgGly > SerGly > HisGly > GluGly. Furthermore, the obtained NADESs present super hydrophilic, which can be dissolved in water at any proportion.

The density and viscosity are key properties of the reaction media. The temperature-dependent densities and viscosities were measured as shown in Figure 2 and listed in Supplementary Table S1. The results showed that the density decreases linearly with the increase of temperature for all NADESs, and follows the order of ArgGly > HisGly > SerGly > GluGly. The viscosity decreases non-linearly (sharply) with the increase of temperature for all NADESs, and follows the order of HisGly > SerGly > GluGly > ArgGly.

According to the kinetics of the multi-enzymatic reaction (Luo et al., 2015), the first two reactions (CO₂→HCOOH→HCHO) were rate-limiting steps, and the second reaction (HCOOH→HCHO) was regulated by the product concentration of the first reaction. Therefore, the first reaction (CO₂→HCOOH) plays a key role in the whole reaction, as confirmed by a previous study (Zhang et al., 2021c). Besides, biocatalyst FDH plays a key role in driving the first reaction, and thus the activity of FDH was evaluated in four NADESs (20% v/v) as well as a conventional buffer as a blank for comparison. As shown in Figure 3, the activity of FDH was enhanced by SerGly and GluGly, and SerGly presented the better enhancement with 384.6 U/mg, while the activities of FDH in ArgGly and HisGly were seriously decreased to 69.4 and 78.4 U/mg, respectively. To figure out the reason, the pH values of different NADES-based solutions were examined, and those for SerGly and GluGly were close to the optimal pH of FDH (7.0), while the pHs of ArgGly-based and HisGly-based solution were 9.9 and 5.6, respectively. Therefore, the pH values that are far from the optimal value of FDH are probably the main reason for the activity decline. For the case that the pH values of NADESs are close to the optimal pH of FDH, the enhancement can be explained as follows. Previous results revealed that the usage of ILs will increase the rigidity and stability of FDH, and also the substrate (CO₂) residence time in the activity site of FDH (Zhang et al., 2018). The longer residence time will give more chances to adjust the substrate in the right position and thus make it more productive. If the enzyme is in an unfavorable media solution, the enzyme will not perform well with their unnormally folding conformation, leading to the decrease of the activity or even lose activity. Within the studied NADESs, SerGly with desirable pH exhibits the best enhancement on the enzyme activity.

To further identify the optimal NADESs concentration, the concentrations of SerGly and GluGly were set in the range from 10 to 60% (V_NADES/V_buffer) and evaluated their impact. As shown in Figure 3, the enzyme activity was enhanced and kept well in the concentration of NADESs from 10 to 20%, because the rigidity and stability of FDH were increased with increasing concentration from 1 to 20%, as described above. Aqueous solutions (water-rich IL mixtures) have also been confirmed to be the best media for proteins, being consistent with experimental results in this study (Kohno and Ohno, 2012). With further increasing the NADESs concentration from 20%, the enzyme activity was sharply decreased, which was probably attributed to the break of electrostatic balance in proteins by the charge of DES when subjected to high salt (DES) concentrations (Weingartner et al., 2012). Once the electrostatic balance in proteins is broken, the hydrophobic interaction plays a main role, resulting in aggregation of enzyme and thus decreasing the active sites of enzyme exposure out. Therefore, a high concentration of NADES will decrease the enzyme activity. Above all, SerGly is the best among the developed NADESs and 20% is optimal concentration. Unless specific statement, 20% NADES-based aqueous solution was used in the following study.

The high performance of reducing NAD⁺ to NADH is crucial to achieving enzymatic conversion of CO₂ in a sustainable way. Electrochemical regeneration of NADH was used in this study. In this part, the effect of buffers, i.e., DES-contained buffer (SerGly) and the conventional buffer (Tris-HCl), were investigated in the electrocatalysis and explore how the DES could boost the electrochemical reduction of NAD⁺ to NADH. First, the electrocatalytic activities of electrocatalyst (Rh complex) in SerGly-contained solution and Tris-HCl solution were separately characterized by cyclic voltammetry (CV). As shown in Figure 4A, within the cathodic potential from −1 to 0 V, a reduction peak, was observed at −0.62 V, indicating the reduction of Rh^III to Rh^I, where the formed Rh^I can efficiently reduce NAD⁺ to NADH (Chen et al., 2019). At this reduction peak potential, the corresponding current density of SerGly-contained solution (−0.225 mA/cm²) is higher than that of conventional solution of −0.166 mA/cm², confirming a higher electron transfer and NAD⁺ reduction efficiency in the SerGly-contained solution.

Next, the Tafel plot was separately examined in both solutions, as an indicator to quantify how easy/difficult to achieve a reaction, and the lower the overpotential, the easier the reaction (Kang et al., 2020). As presented in Figure 4B, the Tafel slope in SerGly-contained solution (200 dec⁻¹) was lower than that in Tris-HCl buffer, indicating a lower overpotential under the given current density. This indicates a lower energy barrier in the SerGly-contained solution with a higher driving force. As reported by Faggion et al. report (Faggion et al., 2019), ILs could significantly reduce the overpotential in the electrochemical reduction of CO₂ and improve the electrochemical efficiency. This observation is consistent with our findings here.

To reveal the difficulty of kinetic reaction, the interface between electrode and electrolyte was investigated by electrochemical impedance spectroscopy (EIS). Figure 4C exhibits typical results of impedance spectra (Nyquist plots), where Z’ and Z” are the real variable and the negatively imaginary variable of impedance, respectively. A semi-circular part at high frequencies in Figure 4C corresponds to the electron-transfer limited process (the electrode/solution interface), and a linear part at low frequencies corresponds to the diffusion process. Charge transfer resistance (R_ct) in the SerGly-contained solution was estimated to be 48 Ω, which is remarkably lower than that in the Tris-HCl buffer of 82 Ω. The results with lower R_ct in the SerGly-contained solution suggested the faster interfacial charge transfer rate between the electrolyte (SerGly) and the working electrode, which is superior to the Tris-HCl buffer in electrocatalysis principally.

To verify the enhancement of NAD⁺ reduction by NADESs, the electrochemical reduction of NAD⁺ to NADH was separately conducted in the SerGly-contained solution and Tris-HCl solution. As shown in Figure 4D, the yield of NADH in the SerGly-contained solution was 60% in 1 h, which was approximately two times higher than that in Tris-HCl buffer. The higher reduction rate of NAD⁺ to NADH in SerGly-contained solution was attributed to higher electron transfer by SerGly, which has been demonstrated by the electrochemical characterization above. Therefore, besides serving as a co-solvent to enhance the enzyme activity, SerGly can also boost electrochemical conversion and is a desirable co-solvent for the enzymatic reaction.

To establish a sustainable process to produce methanol and enhance the process by NADES, the enzymatic reaction coupled with NADH electro-regeneration in the SerGly-contained buffer and Tris-HCl buffer was implemented, and the control reaction (the enzymatic reaction without NADH regeneration) was also studied for comparison. Therefore, three systems were designed and evaluated, as shown in Figure 5. The enzymatic reaction without NADH electro-regeneration quickly reached equilibrium with low methanol concentration (0.03 mM), since such multi-enzymatic reaction is reversible and CO₂ conversion was inhibited by the accumulation of byproduct NAD⁺ over time (Luo et al., 2015). Applying with electro-reduction of NAD⁺ to NADH, the methanol concentration was 0.22 mM, significantly increased up to 7 times higher than the control reaction. It indicated that the byproduct NAD⁺ was efficiently reduced to cofactor NADH, achieving the removal of NAD⁺ and the increased NADH concentration. To further boost the electro-enzymatic conversion of CO₂, SerGly was used to enhance the whole process, i.e., remarkably increase the CO₂ conversion. As a result, the rate of methanol generation reached 0.48 mM/h, which is two times higher than that in the Tris-HCl buffer (0.22 mM/h) and 16 times higher than the control reaction. Being consistent with the above investigation, SerGly enhanced not only enzyme activity but also NADH electro-regeneration rate, which was probably the main reason to improve the CO₂ conversion. Besides, SerGly increased the CO₂ concentration in the solution. As shown in Figure 5, the CO₂ solubility in the SerGly-contained solution was 17.3 mg-CO₂/g-DES at room temperature, which is 11 times higher than that in the Tris-HCl solution (1.5 mg-CO₂/g-water). Zhang et al. (2021c) demonstrated that increasing the CO₂ concentration could significantly drive the reversible reaction forward and thus increase the CO₂ conversion in the enzymatic reaction. Therefore, SerGly plays a role in enhancing enzyme activity and cofactor electro-regeneration as well as in the enrichment of the substrate CO₂, and finally enhances the process of the electro-enzymatic conversion of CO₂.