Ionic Liquids as Bifunctional Cosolvents Enhanced CO2 Conversion Catalysed by NADH-Dependent Formate Dehydrogenase

: Efﬁcient CO 2 conversion by formate dehydrogenase is limited by the low CO 2 concentrations that can be reached in traditional buffers. The use of ionic liquids was proposed as a manner to increase CO 2 concentration in the reaction system. It has been found, however, that the required cofactor (NADH) heavily degraded during the enzymatic reaction and that acidity was the main reason. Acidity, indeed, resulted in reduction of the conversion of CO 2 into formic acid and contributed to overestimate the amount of formic acid produced when the progression of the reaction was followed by a decrease in NADH absorbance (method N ). Stability of NADH and the mechanism of NADH degradation was investigated by UV, NMR and by DFT calculations. It was found that by selecting neutral–basic ionic liquids and by adjusting the concentration of the ionic liquid in the buffer, the concentration of NADH can be maintained in the reaction system with little loss. Conversion of CO 2 to methanol in BmimBF 4 (67.1%) was more than twice as compared with the conversion attained by the enzymatic reaction in phosphate buffer (24.3%).


Introduction
Carbon dioxide (CO 2 ) emissions from combustion of fossil fuels and its greenhouse effect on climate change (i.e., global warming) are considered a current threat [1]. To minimize environmental problems and produce clean energy, efficient utilization of CO 2 and carbon regeneration has been the focus of a tremendous amount of research [2][3][4][5]. Recently, a strategy by which CO 2 may be converted enzymatically into valuable chemicals and fuels such as formate, formaldehyde and methanol has inspired many researchers [6,7]. Obert and Dave were the first to report a cascade reaction involving the three enzymes, formate dehydrogenase (FDH), formaldehyde dehydrogenase (FaldDH) and alcohol dehydrogenase (ADH), where the product of the first reaction serves as a substrate of downstream reaction. Reduced nicotinamide adenine dinucleotide (NADH) was used as the terminal electron donor for the enzymatic reaction [8]. However, enzymatic hydrogenation of CO 2 to formic acid (CH 2 O 2 ), formaldehyde (CH 2 O) and methanol (CH 3 OH) is hampered by the low concentration of CO 2 that is available for the enzyme (formic acid dehydrogenase) in the reaction mixture. Such a low overestimation was much higher when the reaction was performed in buffer (~60 times) as compared to the reaction carried out in a 20/80 (v/v) BmimBF4/buffer mixture (~8 times). The values of CO2 conversion provided by the C method are within a typical range for such an enzymatic reaction under thermodynamic equilibrium, which generally has a low conversion rate (0.002 mM·min −1 ) at the conditions at which the reaction was performed (FDH = 1.3 μg/mL, Tris-buffer, 37 °C) [9,29]. Liu and co-workers reported a higher conversion rate for the same reaction, however the final yield of formate was at the same level (less than 1%) [30]. The higher values of CO2 conversion by the method N suggested a possible degradation of NADH during reaction, which was further investigated.

Stability of NADH in ILs
In order to evaluate the effect of temperature on NADH degradation, aqueous mixtures of NADH with BmimBF4 (5.0 equiv.) were monitored over a temperature range of 10-80 °C ( Figure 1A). NADH degradation curves were obtained by in-situ UV-vis spectroscopy. A remarkable reduction of NADH was observed at temperatures higher than 40 °C, whereas such reduction was alleviated at lower temperature. At 37 °C, the kinetics approximated to a first order, with a degradation energy Ea of 59.5 kJ mol −1 ( Figure S5) and a degradation rate constant of 6.2 × 10 −3 (min −1 ). Degradation decreased exponentially with time, with 90% loss of NADH after 4 h ( Figure 1B), suggesting that the decrease of the NADH concentration in the presence of FDH was not only due to CO2 conversion, but also due to spontaneous degradation under the reaction conditions. Degradation of NADH does not only contribute to overestimate the final CO2 conversion, but also has a direct impact in the actual conversion of CO2 into formic acid, as it also plays a role as a substrate in the CO2 reduction. In order to find likely structure-performance relationships and verify whether NADH degradation occurs in a wide range of ionic liquids, degradation was evaluated in five imidazolium-based ILs, including BmimBF4 (Table 2). Results showed that the stability of NADH was not only dependent on temperature, but also highly dependent on the pH value of the reaction mixture. Under neutral-basic conditions, almost no degradation of NADH was detected (entry 1). In contrast, the concentration of NADH diminished markedly at lower pH values, even under weakly acidic conditions (entry 2). Notably, under strongly acidic conditions, the absorption at 340 nm assigned to NADH disappeared coincidently with the appearance of a blue-shifted peak at 332 nm (entries 3~5 and Figure S8). Finally, when DBULat (known to be a strong basic IL) was employed rather than imidazolium-based ILs (entry 6), only traces of degradation of NADH were observed (non-statistically significant).

Stability of NADH in ILs
In order to evaluate the effect of temperature on NADH degradation, aqueous mixtures of NADH with BmimBF 4 (5.0 equiv.) were monitored over a temperature range of 10-80 • C ( Figure 1A). NADH degradation curves were obtained by in-situ UV-vis spectroscopy. A remarkable reduction of NADH was observed at temperatures higher than 40 • C, whereas such reduction was alleviated at lower temperature. At 37 • C, the kinetics approximated to a first order, with a degradation energy Ea of 59.5 kJ mol −1 ( Figure S5) and a degradation rate constant of 6.2 × 10 −3 (min −1 ). Degradation decreased exponentially with time, with 90% loss of NADH after 4 h ( Figure 1B), suggesting that the decrease of the NADH concentration in the presence of FDH was not only due to CO 2 conversion, but also due to spontaneous degradation under the reaction conditions. Degradation of NADH does not only contribute to overestimate the final CO 2 conversion, but also has a direct impact in the actual conversion of CO 2 into formic acid, as it also plays a role as a substrate in the CO 2 reduction. In order to find likely structure-performance relationships and verify whether NADH degradation occurs in a wide range of ionic liquids, degradation was evaluated in five imidazolium-based ILs, including BmimBF 4 (Table 2). Results showed that the stability of NADH was not only dependent on temperature, but also highly dependent on the pH value of the reaction mixture. Under neutral-basic conditions, almost no degradation of NADH was detected (entry 1). In contrast, the concentration of NADH diminished markedly at lower pH values, even under weakly acidic conditions (entry 2). Notably, under strongly acidic conditions, the absorption at 340 nm assigned to NADH disappeared coincidently with the appearance of a blue-shifted peak at 332 nm (entries 3~5 and Figure S8). Finally, when DBULat (known to be a strong basic IL) was employed rather than imidazolium-based ILs (entry 6), only traces of degradation of NADH were observed (non-statistically significant). lower temperature. At 37 °C, the kinetics approximated to a first order, with a degradation energy Ea of 59.5 kJ mol −1 ( Figure S5) and a degradation rate constant of 6.2 × 10 −3 (min −1 ). Degradation decreased exponentially with time, with 90% loss of NADH after 4 h ( Figure 1B), suggesting that the decrease of the NADH concentration in the presence of FDH was not only due to CO2 conversion, but also due to spontaneous degradation under the reaction conditions. Degradation of NADH does not only contribute to overestimate the final CO2 conversion, but also has a direct impact in the actual conversion of CO2 into formic acid, as it also plays a role as a substrate in the CO2 reduction. In order to find likely structure-performance relationships and verify whether NADH degradation occurs in a wide range of ionic liquids, degradation was evaluated in five imidazolium-based ILs, including BmimBF4 (Table 2). Results showed that the stability of NADH was not only dependent on temperature, but also highly dependent on the pH value of the reaction mixture. Under neutral-basic conditions, almost no degradation of NADH was detected (entry 1). In contrast, the concentration of NADH diminished markedly at lower pH values, even under weakly acidic conditions (entry 2).
Notably, under strongly acidic conditions, the absorption at 340 nm assigned to NADH disappeared coincidently with the appearance of a blue-shifted peak at 332 nm (entries 3~5 and Figure S8). Finally, when DBULat (known to be a strong basic IL) was employed rather than imidazolium-based ILs (entry 6), only traces of degradation of NADH were observed (non-statistically significant).

Effect of Acidity on the Degradation of NADH
For better identification of the role of acidity on the stability of NADH during enzymatic conversion, the pH of the phosphate buffer was adjusted within the range 1.0-7.0 by the addition of proper amounts of phosphoric acid. In the absence of FDH, NADH degraded steadily with decreasing pH from 4.5 to 7.0 ( Figure S6). At pH 4.5 (Figure 2), complete degradation of NADH was observed after incubation at 37 • C for 3 h. With further reduction of pH to 2.5, a new band appeared at 332 nm, and the intensity of this band gradually increased within the pH range from 2.5 to 4.5 ( Figure 2). Such variation of NADH degradation with increasing acidity is consistent with our observations for the aqueous IL system in Table 2. However, the newly formed species referenced at 332 nm was not stable under extreme acid conditions, as indicated by decrease in absorbance at this wavelength over the pH range from 1.0 to 2.0.  For better identification of the role of acidity on the stability of NADH during enzymatic conversion, the pH of the phosphate buffer was adjusted within the range 1.0-7.0 by the addition of proper amounts of phosphoric acid. In the absence of FDH, NADH degraded steadily with decreasing pH from 4.5 to 7.0 ( Figure S6). At pH 4.5 (Figure 2), complete degradation of NADH was observed after incubation at 37 °C for 3 h. With further reduction of pH to 2.5, a new band appeared at 332 nm, and the intensity of this band gradually increased within the pH range from 2.5 to 4.5 ( Figure 2). Such variation of NADH degradation with increasing acidity is consistent with our observations for the aqueous IL system in Table 2. However, the newly formed species referenced at 332 nm was not stable under extreme acid conditions, as indicated by decrease in absorbance at this wavelength over the pH range from 1.0 to 2.0. Next, a buffer solution (2 mL) of BmimBF4 (VIL/Vbuffer = 20%) and NADH (2 μmol) was prepared separately for in-situ UV-vis detection at 37 °C. Initially, the presence of two typical peaks assignable to NADH at 340 and 260 nm, respectively, were observed. Upon incubation at 37 °C for 3 h, an

Proposed Mechanism of NADH Degradation
Next, a buffer solution (2 mL) of BmimBF 4 (V IL /V buffer = 20%) and NADH (2 µmol) was prepared separately for in-situ UV-vis detection at 37 • C. Initially, the presence of two typical peaks assignable to NADH at 340 and 260 nm, respectively, were observed. Upon incubation at 37 • C for 3 h, an apparent decrease of absorption intensity at 340 nm and a simultaneous increase of absorption at 260 nm were observed ( Figure S7). Such variation was once referred to conversion of NADH to NAD + [31]. However, subsequent 1 H NMR studies indicated that a new compound besides NAD + was formed. Typical signals of the 1.4-dihyropyridine unit in NADH (δ: ppm), such as C(1)H 2 (2.80, 2.89), C(3)H (6.00) and C(4)H (6.95), were fully diminished ( Figure 3, Figure S10, and Table S2). On the other hand, representative C-H signals of pyridinium in NAD + arising from hydride abstraction of NADH were not formed. representative C-H signals of pyridinium in NAD + arising from hydride abstraction of NADH were not formed.  DFT calculations were performed to better unravel the mechanisms of NADH degradation in acid conditions [32]. On the basis of previous literature and from our results, α-O 2′ -3B-cyclotetrahydronicotinamide adenine dinucleotide was most likely to be the final product and two potential pathways for such conversion were evaluated ( Figure S9) [28]. In pathway 1 (denoted by black curve in the Figure 4), the decomposition of NADH started with protonation of N-contained 6membered ring at the β-position, with leading to b. Subsequent configuration isomerization from b to d1 occurred through ring opening of O-contained 5-memered ring. Specifically, nucleophilic addition of H2O to the iminium cation in b at the α-position and C-O bond breaking of the Ocontained 5-membered ring in the presence of acid affords c. Next, dehydrogenation of c at the αposition of N-contained 6-membered ring readily proceeded to generate d1 denoted as an α   DFT calculations were performed to better unravel the mechanisms of NADH degradation in acid conditions [32]. On the basis of previous literature and from our results, α-O 2 -3B-cyclotetrahydronicotinamide adenine dinucleotide was most likely to be the final product and two potential pathways for such conversion were evaluated ( Figure S9) [28]. In pathway 1 (denoted by black curve in the Figure 4), the decomposition of NADH started with protonation of N-contained 6-membered ring at the β-position, with leading to b. Subsequent configuration isomerization from b to d 1 occurred through ring opening of O-contained 5-memered ring. Specifically, nucleophilic addition of H 2 O to the iminium cation in b at the α-position and C-O bond breaking of the O-contained 5-membered ring in the presence of acid affords c. Next, dehydrogenation of c at the α-position of N-contained 6-membered ring readily proceeded to generate d 1 denoted as an α configuration, wherein the N-contained heterocycle and hydroxyl group sit at the same side of O-contained 5-memered ring (the β configuration in b represents two substituents sit at the opposite position). Optimization of the position of H 2 O in the structure of d 1 is found favourable in energy. A ternary hydrogen-bonding interaction system -OH/H 2 O/C=O was found in the relatively stable structure d. Finally, the intra-molecular nucleophilic attack of OH to iminium cation of d affords the product e, coincidently with one equivalent of proton being released. In comparison to pathway 1, configuration isomerization occurred directly in pathway 2 (denoted by a red curve in the Figure 4) through an O-contained ring opening intermediate f. As a result, d 2 with α configuration was formed, which is slightly different from d 1 in pathway 1 imposed by the interaction mode towards H 2 O. Subsequently, d 2 follows the same route as d 1 to eventually provide e. All structures of transitional state for both pathways are also optimized (Table S1). It suggests that the reaction leading to d is more likely following pathway 1 since its energy barrier step (b→c) is 3.1 kJ/mol lower as compared with that (f→d 2 ) in pathway 2.  DFT calculations were performed to better unravel the mechanisms of NADH degradation in acid conditions [32]. On the basis of previous literature and from our results, α-O 2′ -3B-cyclotetrahydronicotinamide adenine dinucleotide was most likely to be the final product and two potential pathways for such conversion were evaluated ( Figure S9) [28]. In pathway 1 (denoted by black curve in the Figure 4), the decomposition of NADH started with protonation of N-contained 6membered ring at the β-position, with leading to b. Subsequent configuration isomerization from b to d1 occurred through ring opening of O-contained 5-memered ring. Specifically, nucleophilic addition of H2O to the iminium cation in b at the α-position and C-O bond breaking of the Ocontained 5-membered ring in the presence of acid affords c. Next, dehydrogenation of c at the αposition of N-contained 6-membered ring readily proceeded to generate d1 denoted as an α configuration, wherein the N-contained heterocycle and hydroxyl group sit at the same side of Ocontained 5-memered ring (the β configuration in b represents two substituents sit at the opposite position). Optimization of the position of H2O in the structure of d1 is found favourable in energy. A ternary hydrogen-bonding interaction system -OH/H2O/C=O was found in the relatively stable structure d. Finally, the intra-molecular nucleophilic attack of OH to iminium cation of d affords the

Effect of NADH Degradation in BmimBF 4 and BmimDCA
pH was found to be the most crucial factor leading to NADH degradation. However, variation of pH value cannot be avoided given the intrinsic nature of the reaction. For instance, both the absorption of CO 2 into the aqueous solution and the formation of formic acid will result in a more acidic environment [33]. Interestingly, the results indicated that with CO 2 , NADH degradation is significantly prevented in the presence of contained-BmimBF 4 buffer (Table 1, method N). To clarify and rationally understand this phenomenon, the influence of CO 2 absorption with and without BmimBF 4 on NADH degradation and pH variation were evaluated and compared. Experiments of NADH degradation were conducted in an aqueous solution with a range of volume ratio of BmimBF 4 to buffer in the presence or absence of CO 2 ( Figure 5A). The degradation concentration-course plot (in black) in the case of without CO 2 offers a downward parabola. It demonstrated that the maximum degradation of NADH in a combination of BmimBF 4 and buffer is 22% when a volume ratio of 40% (V IL /V buffer ) is reached. The in-situ pH evaluation of the solution ( Figure 5B) indicated this degradation follows a function of pH value variation, with maximum degradation appearing at the lowest pH value. The pH value variation was attributed to the hydrolysis of BmimBF 4 , which may occur with a different extent under these conditions [34]. On the other hand, a significant degradation of NADH was detected in the buffer as soon as CO 2 (1 bar) was introduced ( Figure 5A, plot in red). Interestingly, such a degradation was gradually relieved with loading BmimBF 4 from 0 to 20% (V IL /V buffer ). The stabilization has been attributed to the formation of ternary aggregates (CO 2 /IL/H 2 O), wherein the amphipathic BmimBF 4 could self-assemble at the interface between the nonpolar CO 2 and the aqueous phase [35]. In this manner, formation of carbonic acid was therefore avoided. Specifically, both hydrolysis of BmimBF 4 and the formation of carbonic acid will be inhibited when a higher ratio of V IL /V buffer was launched, with less NADH degradation.
In contrast, NADH is much more stable in the presence of a weakly basic BmimDCA ( Figure 6A, plot in black), with less than 10% degradation of NADH occurring at maximum. The pH value of the solution progressively increased when the ratio of BmimDCA was increased ( Figure 6B). A similar stabilization of NADH with IL functioning as the surface-active molecule was also observed when CO 2 was introduced ( Figure 6A, plot in red).
of VIL/Vbuffer was launched, with less NADH degradation.
In contrast, NADH is much more stable in the presence of a weakly basic BmimDCA ( Figure 6A, plot in black), with less than 10% degradation of NADH occurring at maximum. The pH value of the solution progressively increased when the ratio of BmimDCA was increased ( Figure 6B). A similar stabilization of NADH with IL functioning as the surface-active molecule was also observed when CO2 was introduced ( Figure 6A, plot in red).

Enzymatic Reaction in BmimBF4 and BmimDCA
Enzymatic conversion of CO2 to formate in phosphate buffer systems (IL/buffer) consisting of two alternative ILs at different ratios was subsequently conducted ( Table 3). As indicated by experiments of NADH degradation ( Figure 5A), it will lead to not only the formation of a ternary aggregates (CO2/IL/H2O) but also hydrolysis of BmimBF4 when the concentration of BmimBF4 was increased within a range (VIL/Vbuffer < 50%), which actually has an opposite effect towards the degradation of NADH. Therefore, taking these premises into consideration, to find a proper concentration of BmimBF4 is required to attain an optimized conversion with least NADH being degraded (entry 3). In contrast to the acidic BmimBF4, NADH was more stable in the presence of basic BmimDCA. And the conversion of NADH monotonously decreased when the concentration of BmimDCA was increased (entries 6~9), but the yield of formate increased with the concentration of BmimDCA from 10% to 40%, and dropped to the lowest at 60% BmimDCA.
It is well known that water with small amounts of salts is considered to be the best media for proteins. However, high concentration of ILs (salts) can cause conformational changes of peptide chains that can result in enzyme denaturation [36]. Crystallisation and aggregation behaviour of proteins change dramatically upon increasing concentrations of ILs. Indeed, in pure ILs, enzymes can hardly be dissolved in homogenous phase without denaturation. In this enzymatic reaction, increasing concentration of IL to absorb CO2 is expected to influence negatively the activity of the enzyme. This can explain that CO2 conversion achieved the highest value when only a 20% of BmimBF4 was used.

Enzymatic Reaction in BmimBF 4 and BmimDCA
Enzymatic conversion of CO 2 to formate in phosphate buffer systems (IL/buffer) consisting of two alternative ILs at different ratios was subsequently conducted ( Table 3). As indicated by experiments of NADH degradation ( Figure 5A), it will lead to not only the formation of a ternary aggregates (CO 2 /IL/H 2 O) but also hydrolysis of BmimBF 4 when the concentration of BmimBF 4 was increased within a range (V IL /V buffer < 50%), which actually has an opposite effect towards the degradation of NADH. Therefore, taking these premises into consideration, to find a proper concentration of BmimBF 4 is required to attain an optimized conversion with least NADH being degraded (entry 3). In contrast to the acidic BmimBF 4 , NADH was more stable in the presence of basic BmimDCA. And the conversion of NADH monotonously decreased when the concentration of BmimDCA was increased (entries 6~9), but the yield of formate increased with the concentration of BmimDCA from 10% to 40%, and dropped to the lowest at 60% BmimDCA.
It is well known that water with small amounts of salts is considered to be the best media for proteins. However, high concentration of ILs (salts) can cause conformational changes of peptide chains that can result in enzyme denaturation [36]. Crystallisation and aggregation behaviour of proteins change dramatically upon increasing concentrations of ILs. Indeed, in pure ILs, enzymes can hardly be dissolved in homogenous phase without denaturation. In this enzymatic reaction, increasing concentration of IL to absorb CO 2 is expected to influence negatively the activity of the enzyme. This can explain that CO 2 conversion achieved the highest value when only a 20% of BmimBF 4 was used.  4 :NaH 2 PO 4 = 39:61 (mol/mol)), 2 mL); 37 • C, 3 h. a Enzymatic reaction was repeated three times and standard deviations (SD) were calculated that presented in the Table 3. pH value of reaction mixture was added before and after saturating CO 2 .

Multi-Enzymatic Reaction of Converting CO 2 to Methanol in BmimBF 4
Sequential reduction of CO 2 to formic acid, formaldehyde and finally methanol is limited by the low reaction rate of the first reaction in the sequence (CO 2 → formic acid), which is much lower than that of the reverse reaction (formic acid → CO 2 ) [9]. Indeed formic acid oxidation was 30 times faster than CO 2 reduction catalyzed by FDH [29]. For the second enzyme, FaldDH, the reaction (formic acid → formaldehyde) was also found to be less efficient than the reverse reaction (formaldehyde → formic acid). However, for the third enzyme, ADH, the forward reaction (formaldehyde → methanol) is much more favorable than the reverse reaction (methanol → formaldehyde). Therefore, to make this multi-enzymatic reaction efficient, formic acid and formaldehyde have to be consumed by the next sequential reaction, in order to shift the equilibrium of the cascade reaction and eventually achieving high conversion of CO 2 to methanol ( Table 4, entry 1). a Reaction conditions: FDH (3 μL, 0.075 U/mL), NADH (2 μmol, 1 mM), CO2 (1 bar), buffer (100 mM (Na2HPO4:NaH2PO4 = 39:61 (mol/mol)), 2 mL); 37 °C, 3 h. a Enzymatic reaction was repeated three times and standard deviations (SD) were calculated that presented in the Table 3. pH value of reaction mixture was added before and after saturating CO2.

Multi-Enzymatic Reaction of Converting CO2 to Methanol in BmimBF4
Sequential reduction of CO2 to formic acid, formaldehyde and finally methanol is limited by the low reaction rate of the first reaction in the sequence (CO2 → formic acid), which is much lower than that of the reverse reaction (formic acid → CO2) [9]. Indeed formic acid oxidation was 30 times faster than CO2 reduction catalyzed by FDH [29]. For the second enzyme, FaldDH, the reaction (formic acid → formaldehyde) was also found to be less efficient than the reverse reaction (formaldehyde → formic acid). However, for the third enzyme, ADH, the forward reaction (formaldehyde → methanol) is much more favorable than the reverse reaction (methanol → formaldehyde). Therefore, to make this multi-enzymatic reaction efficient, formic acid and formaldehyde have to be consumed by the next sequential reaction, in order to shift the equilibrium of the cascade reaction and eventually achieving high conversion of CO2 to methanol (Table 4, entry 1). Another strategy to further increase conversion of CO2 to methanol is to enhance CO2 solubility, as low CO2 solubility in water would also result in low formic acid concentration. As a consequence, production of formaldehyde is prevented, as the second sequential reaction requires a threshold concentration of formic acid to be activated [24]. Both factors (unfavourable equilibrium rates and need of a minimum concentration threshold) make the first reaction of the sequence (the one studied here) to play a decisive role on conversion of CO2 to methanol. Addition of BmimBF4 helps enhance the conversion in the first reaction by both stabilizing NADH and also increasing CO2 solubility. Indeed, the yield of methanol when the reaction was conducted in 20% BmimBF4 was two times higher than in the phosphate buffer.
Other authors followed other strategies to increase conversion. For example, Ober and Dave immobilized the three enzymes in silica sol-gel matrixes, confining and reducing the volume of the enzymes, in such a manner that the local concentration of reactants was enhanced [8]. Jiang followed a similar strategy, and was able to increase the yield of methanol up to 71.6% [24]. These studies also confirmed that breaking the unfavourable equilibrium by switching it to the right is a good manner to enhance the overall conversion. Another strategy to further increase conversion of CO 2 to methanol is to enhance CO 2 solubility, as low CO 2 solubility in water would also result in low formic acid concentration. As a consequence, production of formaldehyde is prevented, as the second sequential reaction requires a threshold concentration of formic acid to be activated [24]. Both factors (unfavourable equilibrium rates and need of a minimum concentration threshold) make the first reaction of the sequence (the one studied here) to play a decisive role on conversion of CO 2 to methanol. Addition of BmimBF 4 helps enhance the conversion in the first reaction by both stabilizing NADH and also increasing CO 2 solubility. Indeed, the yield of methanol when the reaction was conducted in 20% BmimBF 4 was two times higher than in the phosphate buffer.

Entry
Other authors followed other strategies to increase conversion. For example, Ober and Dave immobilized the three enzymes in silica sol-gel matrixes, confining and reducing the volume of the enzymes, in such a manner that the local concentration of reactants was enhanced [8]. Jiang followed a similar strategy, and was able to increase the yield of methanol up to 71.6% [24]. These studies also confirmed that breaking the unfavourable equilibrium by switching it to the right is a good manner to enhance the overall conversion.

General Procedure for Enzymatic Conversion of CO 2 to Formate
The enzymatic reaction was conducted in a stainless steel reactor (25 mL) equipped with two valves and a pressure gauge. In the reactor, a buffer solution of FDH (3 µL) and NADH (1 mM) was bubbled with CO 2 for 5 min to remove the residual air, after which the pressure was adjusted to 1 bar. The reactor was positioned in water bath at 37 • C for 3 h. Samples were then taken from the reactor for analysis. Buffer (100 mM (Na 2 HPO 4 :NaH 2 PO 4 = 39:61 (mol/mol)), pH = 7, 2 mL); Enzymatic reaction was repeated three times; standard deviations (SD) were calculated and presented in the Table 1.

Kinetic Degradation of NADH in BmimBF 4
Solution of NADH and BmimBF 4 was prepared in deionized water (2 mL) at concentration of 1 mM (1.4 mg, 2 µmol) and 5 mM (2 µL, 5.0 equiv.), respectively. The reaction was performed over a temperature range of 10-80 • C during which sample was measured every ten degree centigrade and kept for ten minutes. Additionally, another experiment was carried out at 37 • C for 9 h and the preparation of sample was same as above during which samples were measured every ten minutes ( Figure 1B). Temperature (37 • C) in such experiment was referring to enzymatic reaction and the variation of NADH concentration with time was monitored by method N.

Stability of NADH in Aqueous ILs
In the reactor, a solution (2 mL) of 20 vol % ILs and 80 vol % deionized water was prepared to dissolve 2 µmol NADH (1 mM). Then, the reactor was positioned in water bath at 37 • C for 3 h. The sample was taken from the reactor and detected at 340 nm by an in-situ UV-vis spectroscopy ( Table 2).

Effect of Acidity on the Degradation of NADH
Phosphate buffers in the pH range of 1.0-7.0 were separately prepared, namely, pH 7, 6.5, 5.0, 4.5, 4.0, 3.5, 3.0, 2.5, 2.0, 1.5, and 1.0, with addition of varied amounts of purity phosphoric acid in phosphate solution (pH = 7). NADH (1 mM) in these buffers was separately incubated at 37 • C for 3 h. Residue NADH is measured at 340 nm by an in-situ UV-vis spectroscopy (Figures 2 and S6).

NMR for NADH Degradation in BmimBF 4
Equimolar of NADH and BmimBF 4 was mixed in deuteroxide. The mixture was heated at 50 • C for 3 h and then the product was purified through extraction of methylene dichloride. After removing methylene dichloride in a vacuum oven at 50 • C for 3 h, solid powder was obtained and detected by NMR.

Effect of NADH Degradation in BmimBF 4 and BmimDCA
Solutions (2 mL) of BmimBF 4 with phosphate buffer in different concentration were prepared, namely, 0%, 5%, 10%, 15%, 20%, 30%, 40%, 60%, and 80% (V IL /V buffer ). In the reactor, NADH (1 mM) in the prepared solution was bubbled with CO 2 for 5 min to remove the residual air, after which the pressure was adjusted to 1 bar. The reactor was positioned in water bath at 37 • C for 3 h. Without CO 2 , it is no need to press CO 2 in the reactor. In the same way, solutions (2 mL) of BmimDCA with phosphate buffer in different concentrations were prepared, namely, 0%, 10%, 20%, 30%, 40%, and 60% (V IL /V buffer ). In the reactor, NADH (1 mM) in the solution was bubbled with CO 2 for 5 min to remove the residual air, after which the pressure was adjusted to 1 bar. The reactor was positioned in water bath at 37 • C for 3 h. Furthermore, pH value of each solution was measured by pH meter (298.5 K, Mettler Toledo FE20, Mettler Toledo, Zurich, Zurich State, Switzerland) before starting reaction.

Enzymatic Reaction in BmimBF 4 and BmimDCA
Solutions (2 mL) of BmimBF 4 with phosphate buffer in different concentrations were prepared, namely, 0%, 10%, 40%, and 60% (V IL /V buffer ). Solutions (2 mL) of BmimBF 4 with phosphate buffer were prepared in the same way. Procedure of enzymatic reaction is as same as described above (General procedure).

Analytical Method
Two methods, NADH (named as N) and colorimetric detection (named as C), were evaluated to determine the yield of formate product. The details of these methods are as follows:

Method N for Determination Concentration of NADH
Method N is an indirect method where the yield of formate is determined through quantifying reduction of NADH. It can be easily monitored by UV-vis spectroscopy at 340 nm referenced to NADH [23]. The assumption is that formation of formate during the enzyme-catalysed process would consume an equal-molar amount of NADH. The concentration of NADH was quantified by absorbance measurements, which was monitored at 340 nm by an in-situ UV-vis spectroscopy. After the enzymatic conversion to formate was completed, 0.5 mL of sample was taken and diluted with 1.5 mL of water. After dilution of NADH solution, absorbance at 340 nm is below 2 that is more precise. Then 2 mL of the solution was pipetted into a 10 mm quartz cell. The photometric measurement was monitored at 340 nm using a UV-vis spectrophotometer (Shimadzu UV, 2550 spectrophotometer Shimadzu Corporation, Tokyo, Tokyo metropolitan, Japan). The standard calibration curve is presented in Figure S1.

Method C for Determination Concentration of Formate
Method C was originally used for detecting formate in products from fermentation [37].
The assumption is that a complex compound, recorded at 515 nm using UV-vis spectroscopy, would be formed by the reaction of formate with a mixture of citric acid, isopropanol, acetic anhydride, sodium acetate in an appropriate ratio [38]. The yield of formate is then directly determined. First, 3.5 mL of 100% acetic anhydride, 50 µL of 30% (w/v) sodium acetate, and 1 mL of 2-propanol solution containing 0.5% (w/v) citric acid and 10% (w/v) acetamide were added in a 10 mL vial. Then 0.5 mL of sample was added to this assay solution and incubated for 90 min at 25 • C. Finally, the absorbance was determined at 515 nm using a UV-vis spectrophotometer. The formate standard calibration curve was prepared against the appropriate concentrations of formate ( Figures S2-S4). 500 µL of sample was used for the colorimetric assay.

GC for Determination Concentration of Methanol
A Hewlett Packard HP6890 gas chromatograph (Hewlett-Packard Company, Palo Alto, CA, USA) (GC) equipped with a FID (250 • C) and a Restek XTI-5 column (30 m × 0.25 mm i.d., film thickness 0.25 mm) was used for methanol concentration by using ethyl acetate as an internal standard. The carrier gas was N 2 with a flow rate of 0.4 mL min −1 . The injector temperature was 150 • C and the injection volume was 1 mL. Methanol GC chromatograms were calibrated with 0.01-1 mM methanol solution in 0.1 M pH 7.0 phosphate buffer.

1 H Nuclear Magnetic Resonance Measurements
Spectra were recorded at 298 K on a Bruker av-600 MHz spectrometer (Bruker, Bilka, MA, USA) operating at 600 MHz. Deuterium oxide was as solvent.

Conclusions
In summary, we demonstrated that NADH degradation is unavoidable due to the presence of acid gas during enzymatic CO 2 conversion, which impedes the CO 2 conversion and overestimates the conversion when the absorbance of NADH is used as a method of detection. The mechanism of NADH degradation was investigated by UV, NMR, and DFT calculation and all methods suggested that the occurrence of acid in the reaction mixture is the main contributing factor to the degradation observed. By selecting neutral-basic ionic liquids and adjusting the concentration of ionic liquids in the buffer, stabilization of the cofactor (NADH) can be achieved, along with enabling a higher CO 2 concentration in the buffer. Finally, CO 2 conversion was more than twice as compared with the conversion reached by the enzymatic reaction in a phosphate buffer (traditional buffer). This study is a significant contribution to the use of enzymes like formate dehydrogenase in ionic liquids and paves the way for improving biocatalysts using ionic liquids.
Author Contributions: Z.Z., B.-h.X. and J.L. conceived and designed the experiment, performed the experiments, and analysed the data. M.P., N.V.S. and S.Z. wrote the paper and reviewed drafts of the paper. H.H. and Y.Z. conceived and designed the simulation experiment, analysed the data and reviewed drafts of the paper.