ChemicalsBuffer salts and HPLC grade solvents were purchased from Sigma-Aldrich. Deuterated solvents were purchased from Thermo Scientific (CDCl3, 99.8% D; MeOH-d4, ≥99.8% D), Sigma-Aldrich (D2O, 99.9% D), and VWR Chemicals (DMSO-d6, 99.8% D).Nitrobenzene (99%) was purchased from Alfa Aesar; N-phenylhydroxylamine (≥95%), nitrosobenzene (≥97%), 4-nitrotoluene (99%), 4-nitrophenol (≥99%), 1,2-dinitrobenzene (97%), 1,3-dinitrobenzene (97%), 1,4-dinitrobenzene (98%), 1-fluoro−4-nitrobenzene (99%), 1-chloro-4-nitrobenzene (99%), 1-nitronaphtalene (99%), benzyl viologen dichloride (97%), N,N-diethylethylenediamine (≥99%), triethylamine (≥99%), 3-chloroperbenzoic acid (≤77%), 1-nitrohexane (98%), hexylamine (99%), and 4-nitrobenzoyl chloride (98%) were purchased from Sigma-Aldrich; 2-nitrotoluene (99%), 3-nitrotoluene (95%), 2-nitrophenol (95%), 3-nitrophenol (99%), 1-bromo-2-nitrobenzene (98%), 1-bromo-3-nitrobenzene (99%), 1-bromo-4-nitrobenzene (95%), 1-iodo-4-nitrobenzene (95%), 4-nitrobenzyl alcohol (99%), 4-nitrobenzaldehyde (98%), 4-nitrobenzoic acid (98%), ethyl 4-nitrobenzoate (95%), 4’-nitroacetophenone (95%), 4-nitrobenzotrifluoride (98%), 1-tert-butyl-2-nitrobenzene (95%), 1-tert-butyl-nitrobenzene (95%), 4-nitrothiophenol (90%), 4-nitrobenzonitrile (97%), 1-ethynyl-4-nitrobenzene (95%), 5-nitrosalicylic acid (95%) were purchased from Fluorochem; 4-nitrostyrene (98%) was purchased from Thermo Scientific. Product standards 4-aminostyrene (96%) and 4-aminothiophenol (97%) were purchased from Fluorochem. Carbon black (carbon, C) Black Pearls® 2000 was purchased from Cabot.All the chemicals and solvents were used as received without prior purification, except for 4-aminothiophenol, which was purified by column chromatography (silica gel, hexane/EtOAc = 5:1, Rf = 0.31). All aqueous solutions were prepared with deoxygenated MilliQ water (Millipore, 18 MΩcm).Methods of analysis1H-NMR spectra were recorded at 400 MHz on a Bruker Avance III HD NanoBay NMR spectrometer equipped with a 9.4 T magnet. Chemical shifts of 1H-NMR spectra (measured at 298 K) are given in ppm by using residual solvent signals as references (CDCl3: 7.26 ppm; D2O: 4.79 ppm; DMSO-d6: 2.50 ppm; methanol-d4: 4.78 ppm and 3.31 ppm). Coupling constants (J) are reported in Hertz (Hz). Standard abbreviations indicating multiplicity were used as follows: s (singlet), brs (broad singlet), d (doublet), dd (doublet of doublets), t (triplet), q (quadruplet), m (multiplet). 1H-NMR signals for samples in 10% D2O in sodium phosphate buffer (PB, concentration and pH are specified in each case) were measured with the water suppression method. NMR-based yields reported by analysis of 1H-NMR signals measured in 10% D2O in PB with a quantitative water suppression method in the presence of internal standard (4-nitrophenol).Analytical thin-layer chromatography (TLC) was performed on MERCK silica gel 60 F254 aluminium plates, which were analysed by fluorescence detection with UV-light (λ = 254 nm, [UV]) or after exposure to standard staining reagents and subsequent heat treatment. The following staining solution was used: potassium permanganate [KMnO4] (1.5 g potassium permanganate, 10 g potassium carbonate, 1.25 mL sodium hydroxide (10% aqueous solution) in 200 mL water). Purification by column chromatography was performed using Geduran® Si 60 (40–63 µm) purchased from Sigma-Aldrich.Production and purification of enzymesThe Hydrogenase-1 (Hyd-1) was produced according to the reported protocol and purified using a Ni-affinity column28. The yield of Hyd-1 was ~0.5 mg of protein per liter of culture. The activity of the overexpressed Hyd-1 was verified using a spectrophotometric assay measuring the hydrogenase-mediated reduction of benzyl viologen dichloride in an H2-saturated solution. The specific activity of (32 ± 1) nmol·min−1·mg−1 was measured on three independent hydrogenase-mediated benzyl viologen dichloride reduction assays. Another batch of Hyd-1 (1.9 mg of protein per liter of culture, similar activity to the previous batch) was prepared according to the same procedure to run a large-scale experiment.The Hydrogenase-2 (Hyd-2) enzyme was produced according to the reported protocol and purified using a Ni-affinity column36. The yield of Hyd-2 was ~0.2 mg of protein per liter of culture. The activity of the overexpressed Hyd-2 was verified using a spectrophotometric assay measuring the hydrogenase-mediated reduction of benzyl viologen dichloride in an H2-saturated solution. The specific activity of (4.18 ± 0.2) µmol·min−1·mg−1 was measured with three independent hydrogenase-mediated benzyl viologen dichloride reduction assays.Preparation of Hyd-1/C catalystCatalyst preparation was carried out in a glove box (Glove Box Technology Ltd.) under a protective N2 atmosphere (O2 < 3 ppm). A 20 mg/mL carbon black suspension in PB (50 mM, pH 6.0 unless stated otherwise) was sonicated for 1 hour. For the preparation of the catalyst for one 2 mL scale reaction with 10 mM concentration of substrate, 52.8 µL of this suspension was transferred to an Eppendorf tube, 15.4 µL of Hyd-1 solution (1.71 mg/mL) was added (C:Hyd-1 = 40:1 mass ratio), the mixture was gently mixed and left in the fridge (4 °C) for 1 hour. After that, the suspension of the catalyst was centrifuged (3 min, 14,100 × g), the supernatant was removed by pipetting, and the catalyst was resuspended in 100 µL of PB (50 mM, pH 6.0 unless stated otherwise). Resuspension-centrifugation-pipetting steps were repeated 3 times, and then the catalyst was resuspended in 100 µL of PB and then directly used for the reaction or frozen in liquid N2 and stored at −80 °C.Small-scale hydrogenation reactions with Hyd-1/C catalystReaction set-up was carried out in a glove box (Glove Box Technology Ltd.) under a protective N2 atmosphere (O2 < 3 ppm). Reactions were run on a 2 mL scale with a 10 mM concentration of substrate in PB (50 mM, pH 6.0 unless stated otherwise) or with 10% v/v of MeCN at room temperature under a gentle H2 flow in an Asynt Octo Mini reactor, which allows running eight reactions in parallel (Supplementary Fig. 5). A stock solution of substrate in buffer or MeCN was transferred to a reaction vessel, 100 µL of catalyst was added, and the volume was adjusted with the corresponding buffer to a total volume of 2 mL. Substrates 8, 12, 14, 17–20, 23, 24, 26, and 30 required double catalyst loading; for substrate 27 quadruple catalyst loading was used; substrate 28 required double catalyst loading and pH 8.0 (PB, 50 mM). Stock solutions of substrates 4, 8–13, 15–18, 20–24, 26–30 were prepared in MeCN so that the total amount of MeCN in the reaction mixture was 10% v/v. The reactor was closed and removed from the glove box. The hydrogen line was connected, and reactions were run at a 30-40 mL/min flow of H2. Time points were taken at 24, 48, and 72 hours and analysed by 1H-NMR spectroscopy. To prepare a sample for the 1H-NMR analysis, an aliquot of 480 µL of the reaction mixture was centrifuged (3 min, 14,100 × g), 450 µL of supernatant was placed in the NMR tube, and 50 µL of D2O was added.Gram-scale synthesis of procainamide using Hyd-1/C catalystCatalyst preparation was carried out in a glove box (Glove Box Technology Ltd.) under a protective N2 atmosphere (O2 < 3 ppm). A suspension of 264 mg of carbon black in 13.2 mL of PB (50 mM, pH 6.0) was sonicated in a 15 mL FalconTM tube for 1 hour. After that, 1.02 mL of Hyd-1 solution (6.5 mg/mL) was added (C:Hyd-1 = 40:1 mass ratio), the mixture was gently mixed and left in the fridge (4 °C) for 1 hour. Next, the suspension of the catalyst was centrifuged (10 min, 11,309 × g), the supernatant was decanted, and the catalyst was resuspended in 25 mL of PB (50 mM, pH 6.0). Resuspension-centrifugation-decanting steps were repeated 5 times, and then the catalyst was resuspended in 25 mL of PB (50 mM, pH 6.0). Then, 50 mL of stock solution of substrate 31 in MeCN (100 mM) was transferred to a three-neck round-bottom flask (1 L) equipped with a stirring bar, the catalyst suspension was added, and the volume was adjusted to a total volume of 500 mL with the same buffer. The flask was closed with three Suba-Seal® septa and removed from the glove box. The hydrogen line was connected, and the reaction was run at a 30 mL/min flow of H2 at 25 °C in a temperature-controlled oil bath. Reaction completion was confirmed after 24 hours by 1H-NMR spectroscopy. The reaction mixture was then transferred into a centrifuge bottle and centrifuged for 20 min at 11,309 × g. The supernatant was transferred into a separation funnel and washed with EtOAc (2 × 50 mL). The aqueous layer was collected into an Erlenmeyer flask and pH was adjusted to ~10-11 by slow addition of saturated aqueous K2CO3 solution followed by addition of brine (100 mL). The resulting solution was extracted with EtOAc (10 × 50 mL), combined organic layers were washed with brine (100 mL), dried with Na2SO4, filtered, and concentrated in vacuo to yield procainamide (31a) as a yellow oil.Recycling of Hyd-1/C catalystReaction set-up was carried out in a glove box (Glove Box Technology Ltd.) under a protective N2 atmosphere (O2 < 3 ppm). Reactions were run on a 2 mL scale with a 10 mM concentration of 1 in PB (50 mM, pH 6.0) at room temperature under a gentle H2 flow in an Asynt Octo Mini reactor. A stock solution of substrate in buffer was transferred to a reaction vessel, 100 µL of catalyst suspension in PB (50 mM, pH 6.0) was added, and the volume was adjusted to a total of 2 mL with the corresponding buffer. The reactor was closed and removed from the glove box. The hydrogen line was connected, and the reaction was run at a 30 mL/min flow of H2. After 24 hours of reaction time, the reaction mixture was transferred to an Eppendorf tube® and centrifuged (3 min, 14,100 × g), the supernatant was removed and analysed by 1H-NMR spectroscopy. The remaining catalyst was resuspended in 100 µL PB (50 mM, pH 6.0) and subjected to the next reaction cycle. The same procedure was repeated until a drop of conversion of the starting material was observed (Supplementary Fig. 43).Electrochemical proceduresAll electrochemistry was conducted using a 3-electrode cell consisting of a pyrolytic graphite edge (PGE) working electrode (WE, 0.031 cm2), a saturated calomel reference electrode (SCE, Palmsens BV), a coiled platinum wire counter electrode (CE), and aqueous electrolyte (PB 50 mM, pH 6.0 unless otherwise specified). Cell temperature was kept constant at 25 °C by a water jacket. All experiments were conducted in a N2 atmosphere (O2 < 3 ppm) glovebox (Glove Box Technology Ltd.).Voltammetry of Hyd-1 and nitrobenzeneCyclic voltammograms at stationary electrodes of nitrobenzene (1), N-phenylhydroxylamine (1b) and nitrosobenzene (1c) were recorded using a PalmSens4 potentiostat (Palmsens BV) and the associated PSTrace 5.8 software. Cyclic voltammograms (CVs) were recorded with n = 2 replicates on independent electrodes and 3 sequential CVs recorded within each replicate. Each measurement included a 20 s poise to equilibrate the electrode before the scan. Aqueous electrolyte for all measurements was PB (50 mM, pH 6.0) containing 1 mM substrate (note: where solubility is <1 mM, this is a saturated solution). Each repeat included a control measurement in PB (50 mM, pH 6.0) without substrate to ensure a clean electrode surface before the start of each measurement. Before use, each WE was polished through 600, 1200 and 4000 grit sandpapers to achieve a high shine, with three 5-second pulses in a sonicator in MilliQ-water and rinsing with ethanol between each grade of sandpaper.Enzyme film voltammogramsTo determine the onset potential of H2 oxidation by Hyd-1 and Hyd-2 voltammograms were recorded using an Autolab PGStat30 potentiostat (EcoChemie) and the associated Nova 1.1 software. The cell headspace was purged with a constant flow of 1000 scc/min H2 gas before and during measurements. Aqueous electrolyte for all measurements was PB (50 mM, pH 6.0). Voltammograms were recorded with n = 2 replicates on independent electrodes. Each experiment included a control measurement with no enzyme film. Before use, each WE was polished with 400 grit sandpaper and rinsed with MilliQ-water, and was then modified with an enzyme film by dropping 2 µL of enzyme solution (Hyd-1 = 1.71 mg/mL, Hyd-2 = 6.3 mg/mL) onto the electrode which was allowed to incubate for 5–10 min before rinsing again with MilliQ water. The modified electrode was mounted onto a rotator (Metrohm) and rotated at 3000 rpm for the duration of the measurements to negate mass transport effects.Whole catalyst voltammogramsWhole catalyst voltammograms were recorded by the same method as the enzyme film voltammograms with the exception that 3 µL of freshly prepared catalyst suspension (Hyd-1/C) was drop-cast on to the polished electrode in place of the enzyme.Determination of substrate reduction onset potentialsA broader study of the reduction onset potentials in aqueous electrolyte of nitro-containing compounds was conducted to assess whether these thermodynamic predictions of catalyst reactivity were accurate. Since these are not currently available in literature under aqueous conditions, we used voltammetry of the substrates under reaction conditions to determine these.Substrate reduction voltammograms were recorded using a PalmSens4 (Palmsens BV) potentiostat and the associated PSTrace 5.8 software. Cyclic voltammograms (CVs) were recorded with n = 2 replicates on independent electrodes and 3 sequential CVs recorded within each replicate following a 20 s poise to equilibrate the electrode. The first sweep of each (equivalent to a linear sweep voltammogram) was taken for analysis. Electrolyte for all measurements was PB (50 mM, pH 6.0) containing 1 mM substrate (note: where solubility is <1 mM, this is a saturated solution). Each repeat included a control measurement in PB (50 mM, pH 6.0) without substrate. Before use, each WE was polished through 600, 1200 and 4000 grit sandpapers to achieve high shine, with three 5 s pulses in MilliQ-water in a sonicator and rinsing with ethanol between each grade of sandpaper. Working electrodes were then mounted onto an AutoLab (Metrohm) rotator and were rotated at 3000 rpm, unless otherwise specified, during data acquisition to negate mass transport effects, therefore providing a study of substrate behaviour at the carbon surface.Onset potentials were then calculated by analysis of the first voltammogram for each substrate once the electrode was placed into solution. Checking for consistency with the control voltammogram taken before each measurement, a linear baseline correction was extrapolated from a region of no activity. The potential at which the difference between the measured current and the baseline began to increase exponentially, exceeding a threshold value of 10 nA (instrument resolution ~0.5 nA), was taken to be the onset potential. Each voltammogram and calculation was repeated with two independent electrodes to ensure true repeats on newly prepared carbon surface.Small-scale hydrogenation reactions with Hyd-2/C catalystCatalyst preparation was carried out in a glove box (Glove Box Technology Ltd.) under a protective N2 atmosphere (O2 < 3 ppm). A 20 mg/mL carbon black (BP2000, Cabot) suspension in PB (50 mM, pH 6.0) was sonicated for 1 hour. For the preparation of the catalyst loading for one 1 mL scale reaction with 10 mM concentration of substrate, 7.5 µL of this suspension was transferred to an Eppendorf tube, 2.13 µL of Hyd-2 solution (6.3 mg/mL) was added (C:Hyd-2 = 5.6:1 mass ratio), the mixture was gently mixed and left on ice for 1 hour. After that, the suspension of the catalyst was centrifuged (3 min, 14,100 × g), the supernatant was removed, and the catalyst was resuspended in 50 µL of PB (50 mM, pH 6.0). The catalyst was then directly added to the reaction vial. Reactions were run on a 0.5 mL scale with 10 mM of substrate in PB (50 mM, pH 6.0) in a pressure vessel reactor, charged to 2 bar H2 and placed on a rocker to facilitate mixing. For 1H-NMR analysis, the reaction mixture was centrifuged to collect the catalyst particles, 450 µL of supernatant and 50 µL of D2O were added to the NMR tube. 1H-NMR spectra were recorded with water signal suppression method.
