Continuous Flow Bioamination of Ketones in Organic Solvents at Controlled Water Activity using Immobilized ω‐Transaminases

Abstract Compared with biocatalysis in aqueous media, the use of enzymes in neat organic solvents enables increased solubility of hydrophobic substrates and can lead to more favorable thermodynamic equilibria, avoidance of possible hydrolytic side reactions and easier product recovery. ω‐Transaminases from Arthrobacter sp. (AsR−ωTA) and Chromobacterium violaceum (Cv−ωTA) were immobilized on controlled porosity glass metal‐ion affinity beads (EziG) and applied in neat organic solvents for the amination of 1‐phenoxypropan‐2‐one with 2‐propylamine. The reaction system was investigated in terms of type of carrier material, organic solvents and reaction temperature. Optimal conditions were found with more hydrophobic carrier materials and toluene as reaction solvent. The system's water activity (aw) was controlled via salt hydrate pairs during both the biocatalyst immobilization step and the progress of the reaction in different non‐polar solvents. Notably, the two immobilized ωTAs displayed different optimal values of aw, namely 0.7 for EziG3−AsR−ωTA and 0.2 for EziG3−Cv−ωTA. In general, high catalytic activity was observed in various organic solvents even when a high substrate concentration (450–550 mM) and only one equivalent of 2‐propylamine were applied. Under batch conditions, a chemical turnover (TTN) above 13000 was obtained over four subsequent reaction cycles with the same batch of EziG‐immobilized ωTA. Finally, the applicability of the immobilized biocatalyst in neat organic solvents was further demonstrated in a continuous flow packed‐bed reactor. The flow reactor showed excellent performance without observable loss of enzymatic catalytic activity over several days of operation. In general, ca. 70% conversion was obtained in 72 hours using a 1.82 mL flow reactor and toluene as flow solvent, thus affording a space‐time yield of 1.99 g L−1 h−1. Conversion reached above 90% when the reaction was run up to 120 hours.


General information 2.1 Equipment
For the immobilization of enzymes on carrier material, a C-star orbital shaker no. 12846016 (Thermo Fisher Scientific, UK) was used. Biorad protein assay dye reagent concentrate was purchased from Carl Roth (Karlsruhe, Germany). Biotransformations were performed in an Eppendorf Thermomixer compact 5350 (Germany). Continuous flow experiments were performed with a Dionex P680 HPLC pump unit (Thermo Fischer Scientific, UK).

Bradford assay
Concentrated Biorad protein assay dye reagent was diluted 5-fold with MilliQ water and filtered over a paper filter. The stock solution was freshly prepared before use and kept in the dark at 4 °C. Albumin calibration was performed in the standard range of 200-1000 µg mL -1 protein. For lower protein concentration (<25 µg mL -1 ), the low-concentration assay of 1-20 µg mL -1 was used. Samples were prepared by mixing 980 µL stock solution and 20 µL protein sample (for low-concentration assay: 800 µL stock and 200 µL protein sample) followed by incubation for 5-10 minutes at RT. Absorption at the wavelength of 595 nm was measured and plotted against the protein concentration. Diluted enzyme samples were then measured in the same fashion in order to determine their concentration.

Analytics
Conversions were determined by GC using a 7890A GC system (Agilent Technologies), equipped with a FID detector, using H2 as carrier gas, and a DB1701 column from Agilent (30 m, 250 μm, 0.25 μm). The enantiomeric excess of the derivatized amines was measured using a ChiraSil DEX-CB column from Agilent (25 m, 320 μm, 0.25 μm).

Immobilization yield
In order to determine how much of the enzyme is immobilized during the process, a Bradford assay (UV absorption at 595 nm, section 2.2) was performed before (A595 initial) and after the immobilization process (A595 final) for calculating the amount of enzyme bound to the beads, i.e., the immobilization yield (Equation 1).

Water activity
For obtaining reaction solvents with a controlled αw, organic solvents were stirred for 1 hour in presence of sodium dibasic phosphate hydrate salt pairs (1:1 w w -1 , ratio of hydrates). Previous studies indicate this time to be sufficient for αw to reach an equilibrium between the organic and the solid phase.

Calculation of amine conversion
The reactions with immobilized ωTAs in organic solvents were analyzed by GC (see section 2.3 for details on analytical equipment and methods). Retention times of observed compounds are listed in Table S2. Apart from the substrate 1a and product 1b, corresponding imines (imine-1a/2b and imine-1b/2a, see Table  S1) were observed. It is important to remark that imine formation is due to the spontaneous and reversible addition of amines to ketones that are present in the same reaction mixture. However, the amine product could always be quantitatively isolated upon proper work-up of the organic phase.
The GC response of the observed imines was determined and compared to those of 1a and 1b. We performed GC calibration using 20 mM, 50 mM, or 100 mM 1a (or 1b) by measuring the GC peak area of the standard samples. In a second set of standard samples, we dissolved 1a (20 mM, 50 mM, or 100 mM) in neat 2b as well as 1b (20 mM, 50 mM, or 100 mM) in neat 2a in order to form the imines in solution in nearly quantitative amounts (>95%). The GC peak area of each standard sample was plotted against the concentration of analyte ( Figure S1). Notably, no significant difference in response was observed among 1a, 1b and imines 1a/2b and 1b/2a. Following this observation we calculated the conversion to 1b in the reaction mixture containing imine-1a/2b and/or imine-1b/2a by adding up GC areas of 1a with those of imine-1a/2b and GC areas of 1b with those of imine-1b/2a in order to obtain the actual reaction conversion: imine-1b/2a Figure S1. GC area response of substrates and products. The GC area response of 20 mM, 50 mM, or 100 mM of 1a (or 1b) was measured. Imine formation was generated by dissolving 1a in neat 2b (imine-1a/2b) or 1b in neat 2a (imine-1b/2a).

Reaction rate in flow reactors
In flow reactors, several parameters relate to the reaction rate. An important parameter is space velocity (SV, in units of reciprocal time), which is defined by the volumetric flow rate of the reactant stream (Vo, specified at the inlet conditions of temperature and pressure with zero conversion), and the catalyst volume (Vc). Often catalyst volume (Vc) is equally related to the reactor volume (Vr), which depends on the packing density of the catalyst particles.

Equation 4:
Space time (τ, in units of time) is the inverse of space velocity and it gives the time required to process one reactor volume: The space time yield (STY) refers to the quantity of product produced per quantity of catalyst per unit time.
If the catalyst is well-packed in the full reactor, then the catalyst volume (Vc) can be equated to the reactor volume (Vr).

Equation 6:
Calculation of space-time yield for the flow process described in the manuscript:

Results
Preparation of EziG-immobilized ω-transaminases was optimized for application in neat organic solvents in terms of type of immobilization buffer, buffer pH, water-equilibrated solvent, reaction solvent and αw. AsR-ωTA was chosen as a model enzyme. As indicated in the experiments in this section, either the immobilization conditions or the reaction solvents were changed. Unless stated otherwise, the reaction conditions for the reductive amination assay were as depicted in Scheme 1 (main manuscript).

Effect of the pH of the immobilization buffer on the activity of immobilized ωTA in organic solvent as reaction medium
AsR-ωTA was immobilized (analytical scale) in KPi buffer (100 mM) at 4 °C. AsR-ωTA (2 mg, 54 nmol) was diluted in 1 mL buffer containing PLP (0.1 mM) and EziG carrier material (20 mg) was added. The suspension was shaken (120 rpm) for 3 hours after which aliquots (20 µL) of the buffer solution were taken and the protein concentration was determined by Bradford assay (section 2.2). The immobilized ωTA was water-equilibrated with EtOAc in the presence of Na2HPO3•5H2O/Na2HPO4•7H2O (ca. 25 mg, ratio 1:1 w w -1 , see experimental section ─ main manuscript). The reaction was performed in toluene (αw = 0.7) with 50 mM 1a and 150 mM 2b at 25 °C (see experimental section ─ main manuscript). Conversions were determined by GC (see section 2.3).

Amination in organic solvents catalyzed by ω-transaminases immobilized on three types of EziG carrier materials
AsR-ωTA was immobilized (analytical scale) under standard conditions (see experimental section ─ main manuscript). The immobilization was performed in KPi buffer (100 mM, pH 8) at 4 °C. AsR-ωTA (2 mg, 54 nmol) was diluted in 1 mL buffer containing PLP (0.1 mM) and EziG carrier material (20 mg) was added. The suspension was shaken (120 rpm) for 1 hour in the case of samples supplied with EziG 1 Fe Opal, and 3 hours in the case of samples supplied EziG 2 Fe Coral or EziG 3 Fe Amber. Aliquots (20 µL) of the buffer solution were taken and the remaining protein concentration was determined by Bradford assay (section 2.2). The immobilized ωTA was water-equilibrated with EtOAc in the presence of Na2HPO3•5H2O/Na2HPO4•7H2O (ca. 25 mg, ratio 1:1 w w -1 , see experimental section ─ main manuscript).

Testing different reaction solvents with EziG-immobilized transaminase
AsR-ωTA was immobilized (analytical scale) under standard conditions (see experimental section ─ main manuscript). The immobilized ωTA was water-equilibrated with EtOAc in the presence of hydrate salts (ca. 25 mg, ratio 1:1 w w -1 , see experimental section ─ main manuscript and Table S6). The reaction was performed in different reaction solvents (Table S6)

Activity of EziG-immobilized Cv-ωTA at controlled α w
Cv-ωTA was immobilized (analytical scale) under standard conditions (see experimental section ─ main manuscript). The immobilized ωTA was water-equilibrated with EtOAc in the presence of hydrate salts (ca. 25 mg, ratio 1:1 w w -1 , see experimental section ─ main manuscript and Table S7). The reaction was performed in toluene (αw as specified, Table S7) with 50 mM 1a and 150 mM 2b at 25 °C. Conversions were determined by GC (see section 2.3).

Activity of EziG-immobilized AsR-ωTA at different temperatures
AsR-ωTA was immobilized (analytical scale) under standard conditions (see experimental section ─ main manuscript). The immobilized ωTA was water-equilibrated with EtOAc in the presence of hydrate salts (ca. 25 mg, ratio 1:1 w w -1 , see experimental section ─ main manuscript and Table S8). The reaction was performed in toluene (αw as specified, Table S8) with 50 mM 1a and 150 mM 2b at 40 °C or 50 °C. Conversions were determined by GC (see section 2.3).

Decrease of the amine donor concentration for the amination catalyzed by EziGimmobilized AsR-ωTA and in organic solvent as reaction medium
AsR-ωTA was immobilized (analytical scale) under standard conditions (see experimental section ─ main manuscript). The immobilized ωTA was water-equilibrated with EtOAc in the presence of hydrate salts (ca. 25 mg, ratio 1:1 w w -1 , see experimental section ─ main manuscript and Table S9). The reaction was performed in toluene (αw as specified, Table S9) with 50 mM 1a and 50 mM, 100 mM, or 150 mM 2b at 25 °C. Conversions were determined by GC (see section 2.3).

Recyclability of EziG-immobilized AsR-ωTA in toluene
AsR-ωTA was immobilized (analytical scale) under standard conditions (see experimental section ─ main manuscript). The immobilized ωTA was water-equilibrated with EtOAc in the presence of hydrate salts (ca. 25 mg, ratio 1:1 w w -1 , Na2HPO3•5H2O/Na2HPO4•7H2O). The hydrate salt pair was added in the first reaction cycle; subsequent cycles were run with the same aliquot of hydrate salt pair unless agglomeration of the particles was observed. In that case, further addition of 10 mg of hydrate salt pair was sufficient to obtain an active catalyst. The reaction was performed in toluene (αw = 0.7, Table S12) with 50-400 mM 1a and 1 equiv. of 2b at 25 °C). Conversions were determined by GC (see section 2.3). It is worth to mention that the quality of the Eppendorf reaction vessel was affected by toluene over time and it was necessary to transfer the biocatalyst to a new reaction vessel between cycle #2 and cycle #3. For that reason, no further recycling of the immobilized ωTA was attempted after the fourth reaction cycle.

Continuous flow experiments
EziG 3 -AsR was applied in a continuous flow set-up (Figure 3, main manuscript) for the transamination of 1a with 2b. The immobilized enzyme flow reactor was prepared as depicted in the experimental section of the main manuscript. The reaction mixture was prepared in a round bottom flask (50 mL) under nitrogen atmosphere. All reaction solvents were degassed before use and water-equilibrated by stirring them with hydrate salt pairs for 1 hour at RT. The reaction solvent (50 mL) containing 1a (20 or 50 mM) and 2b (3 or 5 equiv.) was pumped (flow rate; 0.2 mL min -1 ) through a pre-column containing hydrate salt pairs (4 g, Na2HPO3•5H2O/ Na2HPO4•7H2O) before entering the flow reactor. The outlet of the flow reactor was connected to the reactor vessel creating a closed-loop system. The flow reactor showed acceptable reproducibility for few days of operation when applying 50 mM 1a and 3 equivalents (150 mM) 2b in toluene (αw = 0.7, Figure 4 -main manuscript, Table S13, entries 1-18).
[3] Flow reactor was stored for one day before re-use. [4] Flow reactor was stored for 28 days before re-use.