Continuous Flow Biocatalytic Reductive Amination by Co‐Entrapping Dehydrogenases with Agarose Gel in a 3D‐Printed Mould Reactor

Abstract Herein, we show how the merge of biocatalysis with flow chemistry aided by 3D‐printing technologies can facilitate organic synthesis. This concept was exemplified for the reductive amination of benzaldehyde catalysed by co‐immobilised amine dehydrogenase and formate dehydrogenase in a continuous flow micro‐reactor. For this purpose, we investigated enzyme co‐immobilisation by covalent binding, or ion‐affinity binding, or entrapment. Entrapment in an agarose hydrogel turned out to be the most promising solution for this biocatalytic reaction. Therefore, we developed a scalable and customisable approach whereby an agarose hydrogel containing the co‐entrapped dehydrogenases was cast in a 3D‐printed mould. The reactor was applied to the reductive amination of benzaldehyde in continuous flow over 120 h and afforded 47 % analytical yield and a space‐time yield of 7.4 g L day−1 using 0.03 mol% biocatalysts loading. This work also exemplifies how rapid prototyping of enzymatic reactions in flow can be achieved through 3D‐printing technology.


Table of Contents
Materials and suppliers are reported in the main manuscript, experimental part.

Co-Immobilization of LE-AmDH-v1 and Cb-FDH from cell free extract on EziG 3 beads and biocatalytic amination of 1 in batch
Note: The initial experiments for the co-immobilization of LE-AmDH-v1 and Cb-FDH were performed using EziG 3 (Fe-Amber) cation-affinity beads. Analytical yields for initial batch reactions for the conversion of benzaldehyde (1) to benzylamine (2) at different substrate loading (10-20 mM) and buffer concentrations (375-750 mM) are reported in Table S1 and Fig. S2. Further experiments with co-immobilized LE-AmDH-v1 and Cb-FDH on the other carrier materials were conducted at 20 mM substrate concentration in 750 mM ammonium/ammonia buffer (section 3.2; Fig. S4).
Co-immobilization of LE-AmDH-v1 and Cb-FDH on EziG metal-ion affinity beads was carried out as follows. Cell free extracts containing a mixture of overexpressed LE-AmDH-v1 and FDH (20 % enzymes loading w w -1 compared with carrier material) were added to a 15 ml falcon tube containing 200 mg of EziG 3 (Fe-Amber) carrier material. The mixture was shaken on an orbital shaker at 120 rpm for 3 h at 4 °C. Small aliquots from the aqueous phase (20 μl) were taken before and after the incubation period; thus, the outcome of the immobilization was checked by analyzing the aliquots through SDS-PAGE. Finally, the immobilized enzyme was left to sediment, the buffer was removed by pipetting, and the immobilized enzyme was used directly in biotransformations.
Reaction conditions for reductive amination of benzaldehyde (1): 1 ml final volume in 2 ml Eppendorf tubes; ammonium formate buffer (375 or 750 mM, pH 8); NAD + (1 mM); 1 (10 or 20 mM) as substrate. Co-immobilized LE-AmDH-v1 and Cb-FDH were prepared as reported above (ca. 25 mg of total mass: enzymes on beads, for each sample). The biotransformations were incubated at 170 rpm and 30 °C in an orbital shaker for 24 h. Next, 500 μl of the reaction mixture was stopped with 70 μl of 10 M KOH, extracted once with 650 μl of EtOAc containing 10 mM toluene as internal standard (1 min vortexing + 5 min centrifugation, 4 °C, 14800 rpm) and dried over MgSO4. Analytical yields (Table S1) were determined by GC-FID (see paragraph Analytics) using toluene as internal standard. Reactions were performed in duplicate.

Co-Immobilization of LE-AmDH-v1 and Cb-FDH on other carrier materials and biocatalytic amination of 1 in batch
Immobilization on EziG 1 beads. 100 mg of EziG 1 (Opal) beads pre-loaded with Fe 3+ were poured in a 15 ml Falcon tube, cooled down in an ice bath and suspended in the immobilization buffer (Tris-HCl, 10 ml, 100 mM, pH 7.8). Purified LE-AmDH-v1 (335 μl from a 572 μM stock solution equal to 192 nmol, ca. 8.5 mg) and Cb-FDH (26 μl from a 1.33 mM stock solution equal to 35 nM, ca. 1.5 mg) were added to the suspension (total enzyme ca. 10 mg, equal to 10% w w -1 , enzyme loading to support material) and the mixture was shaken on an orbital shaker (120 rpm) for 3 h at 4 °C. Small aliquots from the aqueous phase (30 μl) were taken before and after the immobilization procedure; their concentrations were determined using Bradford analysis and visualized via SDS-PAGE. The carrier bearing the immobilized enzyme was left to sediment, the buffer was removed by pipetting, and the immobilized enzyme was used directly in biotransformations. Immobilization yield is reported in Table S2.
Immobilization on Purolite resin. 300 mg of Purolite resin pre-loaded with Co 2+ were washed with KPi buffer (2 x 0.6 ml, 100 mM, pH 7.5). The buffer was removed after 5 minutes of manual shaking. Short pulse centrifugation was used to sediment the resin and the buffer was removed by pipetting. Purified LE-AmDH-v1 (335 μl from a 572 μM stock solution equal to 192 nmol, ca. 8.5 mg) and Cb-FDH (26 μl from a 1.33 mM stock solution equal to 35 nM, ca. 1.5 mg) were diluted with KPi buffer (100 mM, pH 7.5) and added to the resins (1.2 ml final volume). The vials were shaken for 2 h at 4 °C using an orbital shaker (90 rpm, horizontal positioning of vials). Small aliquots from the aqueous phase (30 μl) were taken before and after the immobilization procedure; their concentrations were determined using Bradford and visualized via SDS-PAGE. Immobilized enzyme was washed with KPi buffer (2 x 0.6 ml) and then incubated for 45 min in KPi buffer (1 x 1.2 ml). Finally, another washing step was performed with KPi buffer (2 x 0.6 ml) and the immobilized enzyme was ready for activity testing. Immobilization yield is reported in Table S2.
The mixture was shaken on an orbital shaker at 4 °C for 72 h (90 rpm, mild shaking). After centrifugation (10.0 krpm, 1 min, 4 °C), the buffer solution was discarded and the beads were washed with KPi (100 mM, pH 7.5, 1 ml) for several minutes (600 rpm). After centrifugation, the remaining solution was discarded and the beads were washed with KPi (100 mM, pH 7.5, 1 ml) for 30 minutes (600 rpm). After centrifugation the supernatant was discarded, and beads were directly used in the activity measurements. Small aliquots from the aqueous phase (30 μl) were taken before and after the immobilization procedure and during the washing steps; their concentrations were determined using Bradford and visualized via SDS-PAGE. Immobilization yield is reported in Table S2.
For all immobilization procedures, SDS-PAGE analysis is shown in Fig. S3. Data on process for Immobilization are reported in Table S2. Reaction conditions for the reductive amination of benzaldehyde (1): 1 ml final volume in 2 ml Eppendorf tubes; ammonia/ammonium formate buffer (750 mM, pH 8); NAD + (1 mM); 1 (20 mM) as substrate. Co-immobilized LE-AmDH-v1 and Cb-FDH prepared as reported above (ca. 25 mg of total mass enzyme plus beads for each sample). The biotransformations were incubated at 170 rpm and 30 °C in an orbital shaker for 24 h. Next, an aliquot of 500 μl of the reaction mixture was stopped with 70 μl of 10 M KOH, extracted once with 650 μl of EtOAc containing 10 mM toluene as internal standard (1 min vortexing + 5 min centrifugation, 4 °C, 14800 rpm) and dried over MgSO4. Analytical yields (Table  S3) were determined by GC-FID (see paragraph Analytics) using toluene as internal standard. Reactions were performed in duplicate.
Results of activity tests in batch experiments are reported in Table S3 S4. Activity of LE-AmDH-v1 plus Cb-FDH co-immobilized on different supports at 750 mM ammonium/ammonia species concentration and 20 mM of 1.
A stainless-steel column (50 mm length x 2 mm diameter) was filled with EziG 3 (Fe-Amber) (500 mg) and hydrated with Tris-HCl buffer (50 ml, 100 mM, pH 7.8, flow 0.5 ml min -1 ). The soluble protein fraction obtained after centrifugation of the CFE (ca. 100 mg total enzymes, 20 % w w -1 ), prepared as reported in first paragraph of section 3.1., was loaded onto the column using a peristaltic pump (flow rate = 150 μl min -1 ). After complete loading, the flow was stopped, and the crude cell lysate was left to incubate in the column for 45 min at room temperature. Next, Tris-HCl buffer (50 ml, 100 mM, pH 7.8, flow 0.5 ml min -1 ) was flowed through the column to wash out the other protein impurities. Samples (20 μl) of the loaded CFE and the flow-through obtained during washing of the column were collected and analyzed by SDS-PAGE.
The column containing the co-immobilized enzymes was subsequently mounted on a Dionex P680 HPLC pump unit and conditioned by flowing the reaction buffer (ammonia/ammonium formate 30 ml, 750 mM, pH 8, flow rate 0.3 ml min -1 ). After conditioning, the flow reactor containing the immobilized enzymes was disconnected from the HPLC line. Next, the reaction mixture (5 ml final volume; benzaldehyde (1) (20 mM) in ammonia/ammoniun formate buffer (750 mM, pH 8); NAD + (1 mM)) was injected into the system directly from the HPLC solvent lines at 1 ml min -1 . When all of the reagent's solution was injected, the input of the HPLC line was switched back to the reservoir containing the reaction buffer. At this stage, the flow rate was kept constant at 1 ml min -1 until all the void volume was flashed out (ca. 13 ml). Subsequently, the HPLC line was connected again to the column containing co-immobilized enzymes; the flow rate was set at 0.01 ml min -1 . Column was warmed up at 50 °C using a heated water bath, and the system was maintained at constant pressure through a BPR element (40 psi). The reactor outcome was collected in 2 ml Eppendorf tubes (ca. 1 ml in each tube).

Co-immobilization of LE-AmDH-v1 and Cb-FDH from cell free extract and activity test
A 15 ml Falcon tube containing 100 mg of EziG 1 (Opal) carrier material was cooled down in an ice bath and suspended in the immobilization buffer (Tris-HCl, 10 ml, 100 mM, pH 7.8). Combined CFE, prepared as reported in section 3.1, containing LE-AmDH-v1 and Cb-FDH (total enzymes ca. 10 mg, equal to 10 % w w -1 , enzyme loading to support material) were added to the suspension and the mixture was shaken with an orbital shaker (120 rpm) for 3 h at 4 °C. Small aliquots from the aqueous phase (30 μl) were taken before and after the immobilization procedure; their concentrations were determined using Bradford analysis and visualized via SDS-PAGE. The carrier bearing immobilized enzyme was left to sediment, the buffer was removed by pipetting, and the immobilized enzyme was used directly in biotransformations.
Reaction conditions: 1 ml final volume in 2 ml Eppendorf tubes; ammonia/ammonium formate buffer (750 mM, pH 8); NAD + (1 mM); benzaldehyde (10 mM) as substrate. Co-immobilized LE-AmDH-v1 and Cb-FDH were prepared as reported above (ca. 25 mg total mass of enzymes plus beads, in each sample). The biotransformations were incubated at 170 rpm and 30 °C on an orbital shaker for 24 h. Next, 500 μl of the reaction mixture was stopped with 70 μl of 10 M KOH, extracted once with 650 μl of EtOAc containing 10 mM toluene as internal standard (1 min vortexing + 5 min centrifugation, 4 °C, 14800 rpm) and dried over MgSO4. Analytical yields (Table S6) were determined by GC-FID (see paragraph Analytics) using toluene as internal standard. Reactions were performed in duplicate.

In-flow co-immobilization of LE-AmDH-v1 and Cb-FDH from cell free extract
A stainless-steel column (50 mm length x 2 mm diameter) was filled with EziG 1 Fe-Opal (500 mg) and hydrated with Tris-HCl buffer (40 ml, 100 mM, pH 7.8, flow 0.5 ml min -1 ). The soluble protein fraction of CFE prepared as reported in section 3.1 was loaded onto the column using a peristaltic pump (flow rate = 200 μl min -1 ). After complete loading, the flow was stopped, and the cell lysate was left to incubate in the column for 30 min at room temperature. Then, Tris-HCl buffer (50 ml, 100 mM, pH 7.8, flow 0.5 ml min -1 ) was flowed through the column to wash out any possibly unbound component. Buffer samples (20 μl) of the loading enzyme solution and of the flow-through obtained during washing were taken and immobilization was visualized via SDS-PAGE (Fig. S6).

In-flow co-immobilization and cross-linking of dehydrogenase enzymes.
A stainless-steel column (50 mm length x 2 mm diameter) was filled with EziG 1 (Fe-Opal) (500 mg) and hydrated with Tris-HCl buffer (30 ml, 100 mM, pH 7.8, flow 0.5 ml min -1 ). The soluble protein fraction of CFE prepared as reported in section 3.1 was loaded onto the column using a peristaltic pump (flow rate = 150 μl min -1 ). After complete loading, the flow was stopped, and the cell lysate was left to incubate in the column for 30 min at r.t. Subsequently, a glutaraldehyde solution (5 % in KPi buffer, 10 mM, pH 7.6, final volume 5 ml) was flowed onto the column (0.3 ml min -1 ) and then the column was washed with Tris-HCl buffer (100 mM, pH 8, 20 ml, 0.3 ml min -1 ) in order to stop cross-linking reaction and to wash any unbound component. Buffer samples (20 μl) of the loading enzyme solution and of the flow-through obtained during washing were taken and immobilization was monitored through SDS-PAGE (Fig. S7).

Preparation of agarose-based hydrogel "bricks" for activity tests
An agarose solution (3% w w -1 , + NaCl 10 mM, 2 ml final volume) was prepared in Tris-HCl buffer (100 mM, pH 7.8) and heated up with a microwave (300 W, 10 sec) until a clear solution was obtained. Afterwards, the solution was allowed to cool for up to 2-3 minutes and then purified LE-AmDH-v1 (90 μM as final concentration) and Cb-FDH (16 μM as final concentration) were added. Quickly the resulting mixture was poured into a hole of an Eppendorf tube holder and allowed to cool down and solidify. Finally, the "solid agarose brick" was removed from the mold, split in two equal parts and directly used for biotransformation (Fig. S8).

Preliminary activity test in batch
Agarose-based hydrogel bricks containing entrapped LE-AmDH-v1 and Cb-FDH (90 μM and 16 μM) were prepared as reported in section 4.1.
Reaction conditions: 1 ml final volume in 2 ml Eppendorf tubes; ammonia/ammonium formate buffer (750 mM, pH 8); NAD + (1mM); benzaldehyde (10 mM) as substrate. Entrapped LE-AmDH-v1 and Cb-FDH were prepared as reported above. The biotransformations were incubated at 170 rpm and 40 °C in an orbital shaker for 24 h. Next, an alquot of 500 μl of the reaction mixture was stopped with 70 μl of 10 M KOH, extracted once with 650 μl of EtOAc containing 10 mM toluene as internal standard (1 min vortexing + 5 min centrifugation, 4 °C, 14800 rpm) and dried over MgSO4. Furthermore, the second half of the reaction mixture (500 μl) containing also the agarose-based hydrogel brick were extracted with the same procedure described above. Analytical yields (Table S9) were determined by GC-FID (see paragraph Analytics) using toluene as internal standard. Reactions were performed in duplicate. Reaction mixture 6.9 Reaction mixture + agarose hydrogel (reactor) 9.3
The agarose-based hydrogel brick was then removed from the Eppendorf tube, transferred into a new one and incubated again for 24 h (this procedure was repeated for 7 days). Between the 3 rd and the 4 th cycle, agarose-based hydrogel bricks were left over weekend in the fridge, hydrated with substrate-free reaction buffer. Finally, in the last cycle, also the second half of the reaction mixture (500 μl) containing the agarose-based hydrogel brick was extracted with the same procedure described above. Analytical yields (Table S10) were determined by GC-FID (see paragraph Analytics) using toluene as internal standard. Reactions were performed in duplicate.

In-flow reductive amination using entrapped LE-AmDH-v1 and Cb-FDH using an agarose-based hydrogel flow reactor
An agarose-based flow reactor (Type II-v1) containing entrapped LE-AmDH-v1 and Cb-FDH (90 μM and 16 μM) were prepared as reported in the general procedure (main manuscript, experimental part). This led us to obtain a longer reactor 1 cm x 6 cm (diameter x length) with several channels inside. After removing the solidified reactor from the mold, this was entered inside an empty HPLC-column (1 cm x 10 cm, diameter x length), and the assembled reactor (Fig S9) was manually filled with reaction buffer (ammonia/ammonium formate, pH 8, 750 mM).
The assembled flow reactor was subsequently mounted on a KD Scientific KD 100 syringe pump unit equipped with a Terumo plastic syringe (10 ml) containing the reaction mixture (10 ml). The flow rate was set at 0.02 ml min -1 and the column was heated up (50 °C) by using a warm water bath. Reactor outcome was collected in 15 ml Falcon tubes. Afterwards, 500 μl of the outflow were mixed with 70 μl of 10 M KOH, extracted once with 650 μl of EtOAc containing 10 mM toluene as internal standard (1 min vortexing + 5 min centrifugation, 4 °C, 14800 rpm) and dried over MgSO4. Analytical yields were determined by GC-FID (see paragraph Analytics) using toluene as internal standard.

Continuous-flow reductive amination using entrapped LE-AmDH-v1 and Cb-FDH in an agarose-based hydrogel flow reactor
An agarose-based flow reactor (Type II-v2) containing entrapped LE-AmDH-v1 and Cb-FDH (90 μM and 16 μM) were prepared as reported in the general procedure (main manuscript, experimental part). After removing the solidified reactor from the mold, it was inserted into an empty HPLC-column (1 cm x 10 cm, diameter x length), and the assembled reactor was manually filled with reaction buffer (ammonia/ammonium formate, pH 8, 750 mM).
The reaction was performed according to the procedure described in section 3.6. Reactor outcome was collected in small round-bottom flasks for 24 h. 2× 500 μl aliquots of the outflow were mixed with 140 μl of 10 M KOH, extracted once with 650 μl of EtOAc containing 10 mM toluene as internal standard (1 min vortexing + 5 min centrifugation, 4 °C, 14800 rpm) and dried over MgSO4. Analytical yields (Table S11) were determined by GC-FID (see paragraph Analytics) using toluene as internal standard.

Determination of process parameters from optimized experiment
To characterize the flow process, we have used several parameters such as yield per gram of catalyst and space-time-yield. All parameters (including the isolated yields and the parameters determined earlier) are listed in Table 2, main manuscript. Additionally, the equations that were used for the calculation of the parameters (if not listed previously) are listed below. Analytical yield was used as a mass of product. Molar mass of LE-AmDH-v1 and Cb-FDH is 44569 and 42429 Da, respectively. [1] The volume of the reactor was calculated in section 4.2 and the reaction time when the flowing of the reaction mixture was stopped is listed in Table S12 (cycle 5). = 0.2205 0.006 * 120 = 0.31 ,2 ℎ ,2 S14 4.8. Oligomerization of  Oligomerization status of LE-AmDH-v1 was determined via size exclusion chromatography using AKTA (GE Healthcare) chromatography system equipped with Superdex 200 26/600 column using 20 mM Tris-HCl pH 8 buffer supplemented with 150 mM NaCl. The molecular weight was estimated using the set of molecular weight standards (29 kDa -669 kDa, 25 mg ml -1 , Figure S12). Table S14 contains the outcome of the size exclusion analysis.

Analytics
GC-FID was performed on an Agilent 7890B chromatograph using H2 as carrier gas.