Flow Process for Ketone Reduction Using a Superabsorber-Immobilized Alcohol Dehydrogenase from Lactobacillus brevis in a Packed-Bed Reactor

Flow processes and enzyme immobilization have gained much attention over the past few years in the field of biocatalytic process design. Downstream processes and enzyme stability can be immensely simplified and improved. In this work, we report the utilization of polymer network-entrapped enzymes and their applicability in flow processes. We focused on the superabsorber-based immobilization of an alcohol dehydrogenase (ADH) from Lactobacillus brevis and its application for a reduction of acetophenone. The applicability of this immobilization technique for a biotransformation running in a packed bed reactor was then demonstrated. Towards this end, the immobilized system was intensively studied, first in a batch mode, leading to >90% conversion within 24 h under optimized conditions. A subsequent transfer of this method into a flow process was conducted, resulting in very high initial conversions of up to 67% in such a continuously running process.


Introduction
The development of flow processes has gained tremendous interest in recent years in organic chemistry due to numerous advantages of such types of processes over classic batch approaches [1]. In addition to higher efficiency and safety advantages, the ability to ensure constant product quality has turned out to be a major driving force for the development of flow processes. In particular, for active pharmaceutical ingredients (APIs), this aspect is important, as the FDA along with the EPA, as the respective institutions in United States and Europe for approval of medicines, added aspects of continuous manufacturing to their guidelines [2,3], whereas for a long time the focus of flow processes was on "classic" chemical processes with, for example, hazardous and thermally labile substances and chemocatalytic reactions [4][5][6]. Recently, biocatalysis in flow processes has become an emerging field with an increasing number of examples [7][8][9]. An enzymatic reaction of particular interest for an application in flow mode due to its matured technology level is asymmetric ketone reduction, having a proven track record underlined by many successful examples demonstrated in the past for the batch mode [10][11][12][13]. Recently, significant process in the biocatalytic [14][15][16][17] reductive synthesis of alcohols starting from ketones, as well as syntheses by non-enzymatic means, have been reported [18][19][20][21]. In recent years, an increasing number of examples of biocatalysis in flow processes have been reported [22][23][24][25][26][27]. Although only being studied to a minor extent, some examples of ketone reductions with an alcohol dehydrogenase (ADH) in flow processes have been described [28][29][30][31][32][33][34]. For example, Wandrey's and Liese's groups studied in detail the application of enzyme membrane reactors for enzymatic ketone reduction in a continuously running mode [28,29], and packed bed reactor setups Figure 1. Scheme of superabsorber-based enzyme immobilization based on ionic interactions of acetate moieties of the superabsorber polymer particles and amino acid residues on the protein surface (A), hydrogen bonds of, for example, amino acids with carboxylic polymer residues (B), or entrapment of protein in water residues bound in the polymer matrix (C). Immobilized enzyme converts starting material dissolved in surrounding organic solvent to product, diffusing back into the organic solvent. The light blue circle with dashed line illustrates superabsorber particles with polymer network indicated as wavy lines in the surrounding organic solvent. An (R)-enantioselective ADH from Lactobacillus brevis (LB-ADH) has been shown to be a versatile catalyst for the reduction of a variety of ketones, ranging from aliphatic ketones to benzylic or propargylic ketones [46,47]. It can be used as an isolated enzyme or as a whole-cell catalyst. In particular, its tolerance towards organic solvents is outstanding. The stability depends on several factors, such as the nature of the buffer or solvent additives [46,47].
In this contribution, the application of a superabsorber-immobilized ADH from L. brevis for an enzymatic reduction of acetophenone (1) in a flow process is described.
Because the ADHs are cofactor-dependent enzymes and the cofactor NADPH is very expensive, an efficient cofactor regeneration system is required. In the literature, different options are described [48]. An elegant method is the use of isopropanol for an oxidation of the cofactor using the ADH itself. A further commonly used method is an enzyme cascade using an ADH and a glucose dehydrogenase (GDH). Here, the GDH reduces NADP + back to NADPH, consuming one equivalent of glucose, which is oxidized to gluconolactone and irreversibly opens to gluconic acid in an aqueous environment (see Figure 2).

Preparation of the Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
The preparation of recombinant LB-ADH overexpressed in Escherichia. coli has been previously described elsewhere, for example, in [46,47]. The recombinant LB-ADH for our study was kindly provided by Prof. Dr. Werner Hummel as whole cells. In order to obtain a crude extract of LB-ADH, the whole cells (4.5 g) were suspended in potassium phosphate buffer (6.8 mL, 0.1 M, pH 7, 1:3 (w/v)) and water (6.8 mL), then the resulting suspension was sonicated (5 × 2 min at 20%). After centrifugation (20,000 rpm, 4 °C, 30 min), the supernatant crude extract was decanted. The obtained crude extract was analyzed via photometer and Bradford assay, and aliquots were stored in the freezer at −20 °C for further use.

Preparation of the Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
The preparation of recombinant LB-ADH overexpressed in Escherichia. coli has been previously described elsewhere, for example, in [46,47]. The recombinant LB-ADH for our study was kindly provided by Prof. Dr. Werner Hummel as whole cells. In order to obtain a crude extract of LB-ADH, the whole cells (4.5 g) were suspended in potassium phosphate buffer (6.8 mL, 0.1 M, pH 7, 1:3 (w/v)) and water (6.8 mL), then the resulting suspension was sonicated (5 × 2 min at 20%). After centrifugation (20,000 rpm, 4 • C, 30 min), the supernatant crude extract was decanted. The obtained crude extract was analyzed via photometer and Bradford assay, and aliquots were stored in the freezer at −20 • C for further use.

Characterization of the Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
To determine the determination of biomass concentration, a standard Bradford assay was carried out as follows. Different dilutions of the samples were prepared (crude extract/buffer; 1:20, 1:30, 1:40). Reference (BSA (protein from Bovines Serumalbumin as standard) 1.4, 0.7, 0.35, 0.18, and 0.09 mg mL −1 ) and sample (5 µL) were pipetted in a microtiter plate, and Bradford reagent (250 µL) (see Table 1) was added. After incubation for 15 min at 25 • C while shaking, the absorption was measured at 550 nm. Three-fold determination was made for all measurements.

Compound Composition
Coomassie Blue G 250 50 mg mL −1 in methanol ortho-phosphoric acid 100 mL water to 1 L For the specific activity of the LB-ADH, a photometer assay was carried out. All photometer assays were carried out using a V-30 UV/VIS spectrophotometer from Jasco (Hachioji, Tokyo). According to Equation (1), the activity (in enzymeunits U mL −1 ) was calculated using the measured change in absorption δA, the total volume V tot (1 mL), the volume of crude extract V crude (10 µL), the molar attenuation coefficient of NADP + at 340 nm ε 340 (6200 L mol −1 cm −1 ), the pathlength d (1 cm), and the dilution f.
In a photometer cuvette, buffer (446 µL), acetophenone solution (500 µL), and NADPH solution (24 µL) were mixed and blank measured. Then, ADH solution (30 µL) was added and the mixture was immediately measured at 20 • C and 340 nm after mixing (see Table 2). As described above, the enzyme activities (in U mL −1 ) were calculated from the cofactor absorption in the first 30-90 s.

Preparation of the Recombinant Glucose Dehydrogenase from Bacillus subtilis
Glucose dehydrogenase (GDH) from Bacillus subtilis was prepared according to the literature [49,50]. Enzymatic activity has shown similar activity, as reported in the literature. Aliquots of the obtained crude extract were stored at −20 • C until usage.
Further experimental data and procedures can be found in Appendix A.

Studies on the Comparison of Enzymatic Acetophenone Reduction in Biphasic Buffer/Methyl tert-Butyl Ether (MTBE) System versus Superabsorber/MTBE System in Batch Mode
At first, the LB-ADH-catalyzed reduction of acetophenone (1) to (R)-1-phenylethanol (2) was investigated in a biphasic system of methyl tert-butyl ether (MTBE) and buffer to set a benchmark for the immobilized enzymes. Initial investigations on influences of different parameters and optimization thereof were carried out in batch mode.
Compared to the reaction in buffer, which provided 86% conversion to alcohol (R)-2 after 1.5 h, the biphasic reaction with a ratio of 1:1 (buffer/organic) provided only 45% (see Figure 3). After 24 h, the conversion to product (R)-2 reached its maximum of 89%. In the case of a higher ratio of 1:5 (buffer/organic), a much faster conversion could be observed. After 1.5 h, 75% conversion to alcohol (R)-2 and after 5 h full conversion could be obtained. Emulsion formation and formation of protein aggregates, sticking on the glass surface, could be observed within a short reaction time in all experiments by utilizing a biphasic reaction system. For an application of the ADH in a continuous process, immobilization of ADH utilizing a superabsorber was carried out. In this system, the enzymes are entrapped in an immobilized aqueous phase (as illustrated in Figure 4) and therein also cofactor and cosubstrate are present. In our studies, we exclusively focused on co-immobilization of both GDH and ADH. As diffusion of cofactor is crucial for an efficient process, immobilization of enzymes in different compartments being separated from each other was expected to be difficult in terms of cofactor availability. By means of a coimmobilization, such an effect could be minimized and the influence of process parameters on the ADH-catalyzed reduction could be studied. In the mobile organic phase, the substrate was dissolved. Thus, this type of immobilization enabled the enzymes to operate in their native aqueous environment, which also might lead to a sufficient stability when being used in such a superabsorber immobilizate. For an application of the ADH in a continuous process, immobilization of ADH utilizing a superabsorber was carried out. In this system, the enzymes are entrapped in an immobilized aqueous phase (as illustrated in Figure 4) and therein also cofactor and cosubstrate are present. In our studies, we exclusively focused on co-immobilization of both GDH and ADH. As diffusion of cofactor is crucial for an efficient process, immobilization of enzymes in different compartments being separated from each other was expected to be difficult in terms of cofactor availability. By means of a co-immobilization, such an effect could be minimized and the influence of process parameters on the ADH-catalyzed reduction could be studied. In the mobile organic phase, the substrate was dissolved. Thus, this type of immobilization enabled the enzymes to operate in their native aqueous environment, which also might lead to a sufficient stability when being used in such a superabsorber immobilizate.
crucial for an efficient process, immobilization of enzymes in different compartments being separated from each other was expected to be difficult in terms of cofactor availability. By means of a coimmobilization, such an effect could be minimized and the influence of process parameters on the ADH-catalyzed reduction could be studied. In the mobile organic phase, the substrate was dissolved. Thus, this type of immobilization enabled the enzymes to operate in their native aqueous environment, which also might lead to a sufficient stability when being used in such a superabsorber immobilizate.  To gain a better understanding of this system, batch experiments screening different parameters were carried out. The conversion was monitored by gas chromatography (GC) analysis and the results after a reaction time of 25 h were compared. Different organic solvents, amounts of superabsorber, enzyme loadings, and buffer-to-organic solvent ratios were investigated. All aqueous solutions (ADH, GDH, NADPH, and glucose each in buffer) were mixed and immobilized by addition of superabsorber. After a short incubation of 5 min, a 100 mM acetophenone (1) solution in MTBE was added. All reactions were carried out at 25 • C. The results were compared to a buffer/MTBE system without immobilization ( Figure 5). To gain a better understanding of this system, batch experiments screening different parameters were carried out. The conversion was monitored by gas chromatography (GC) analysis and the results after a reaction time of 25 h were compared. Different organic solvents, amounts of superabsorber, enzyme loadings, and buffer-to-organic solvent ratios were investigated. All aqueous solutions (ADH, GDH, NADPH, and glucose each in buffer) were mixed and immobilized by addition of superabsorber. After a short incubation of 5 min, a 100 mM acetophenone (1) solution in MTBE was added. All reactions were carried out at 25 °C. The results were compared to a buffer/MTBE system without immobilization ( Figure 5). Compared to a reduction using a biphasic buffer/MTBE system, the reaction in the superabsorber/MTBE system was much slower. In the biphasic system, full conversion to alcohol (R)- Compared to a reduction using a biphasic buffer/MTBE system, the reaction in the superabsorber/MTBE system was much slower. In the biphasic system, full conversion to alcohol (R)-2 could be achieved after 5 h, whereas in the superabsorber system after 5 h only 60% conversion to (R)-2 was observed. A maximum conversion to product (R)-2 of 85% after 23 h was observed ( Figure 5). Thereafter, no further conversion could be observed.

Studies on the Influence of Organic Solvents on the Reduction of Acetophenone with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
To investigate the influence of different solvents, the organic phase was changed while other parameters were kept constant. A range of standard solvents with different properties were investigated. As dichloromethane (DCM) is usually not suitable for enzyme catalysis, it would be interesting to learn about the performance of the investigated system towards this solvent in comparison to other organic "standard solvents". In addition, methyl tert-butyl ether (MTBE) and diethyl carbonate were chosen as greener solvent options.
As shown in Figure 6, the highest conversion to product (R)-2 after 24 h was obtained for a system using MTBE as an organic solvent component (85%). A maximum conversion to the desired product of 55% for toluene could be observed after 26 h. Thereafter, no further product formation was observed. The result for diethyl carbonate was comparable to that of toluene. With DCM as solvent, the conversion of acetophenone to product (R)-2 was with 10% even after 48 h reaction time very low. By far the best tested solvent was MTBE, which was therefore used for further experiments.
Bioengineering 2019, 6, x FOR PEER REVIEW 7 of 24 solvent, the conversion of acetophenone to product (R)-2 was with 10% even after 48 h reaction time very low. By far the best tested solvent was MTBE, which was therefore used for further experiments.

Studies on the Influence of Incubation Time on the Acetophenone Reduction with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
A further investigation of the storage stability of the absorbed enzymes was carried out. For this purpose, the immobilized aqueous phase was incubated for 24 h at room temperature without organic solvent.
It was found that the immobilized enzymes were not stable over a longer period. After incubation of the immobilized aqueous phase for 24 h, the maximum conversion to alcohol (R)-2 after

Studies on the Influence of Incubation Time on the Acetophenone Reduction with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
A further investigation of the storage stability of the absorbed enzymes was carried out. For this purpose, the immobilized aqueous phase was incubated for 24 h at room temperature without organic solvent.
It was found that the immobilized enzymes were not stable over a longer period. After incubation of the immobilized aqueous phase for 24 h, the maximum conversion to alcohol (R)-2 after 23 h reaction time dropped from 85% to below 10%, as shown in Figure 7. Therefore, the storage stability of the immobilized enzymes was not very high.

Studies on the Influence of Incubation Time on the Acetophenone Reduction with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
A further investigation of the storage stability of the absorbed enzymes was carried out. For this purpose, the immobilized aqueous phase was incubated for 24 h at room temperature without organic solvent.
It was found that the immobilized enzymes were not stable over a longer period. After incubation of the immobilized aqueous phase for 24 h, the maximum conversion to alcohol (R)-2 after 23 h reaction time dropped from 85% to below 10%, as shown in Figure 7. Therefore, the storage stability of the immobilized enzymes was not very high.

Studies on the Influence of the Ratio of Buffer to Solvent on the Acetophenone Reduction with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
The influence of the ratio of the immobilized aqueous phase and organic phase was investigated. Change of phase interface area, distribution of compounds, as well as the enzyme environment can influence the performance of the system.
By changing the aqueous to organic phase ratio from 1:5 to 1:2.5, an increase of conversion to product (R)-2 after 23 h from 85% to 90% could be observed ( Figure 8). Higher enzyme stability due to less direct contact of enzyme and solvent, and more favorable compound distribution might be the reasons for higher product formation.
influence the performance of the system. By changing the aqueous to organic phase ratio from 1:5 to 1:2.5, an increase of conversion to product (R)-2 after 23 h from 85% to 90% could be observed (Figure 8). Higher enzyme stability due to less direct contact of enzyme and solvent, and more favorable compound distribution might be the reasons for higher product formation.

Studies on the Influence of the Buffer Concentration on the Acetophenone Reduction with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
Also, the influence of potassium phosphate buffer (PPB) buffer concentration was investigated. Because the superabsorber itself acts as a buffer system and because stoichiometric amounts of acid were formed due to the GDH-based cofactor regeneration system, this system was difficult to quantify.
By increasing the PPB buffer concentration from 100 to 200 mM, the conversion to the desired product after 24 h increased from 85% to 94% (Figure 9). The results have been confirmed in a second experiment (92%).

Studies on the Influence of the Buffer Concentration on the Acetophenone Reduction with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
Also, the influence of potassium phosphate buffer (PPB) buffer concentration was investigated. Because the superabsorber itself acts as a buffer system and because stoichiometric amounts of acid were formed due to the GDH-based cofactor regeneration system, this system was difficult to quantify.
By increasing the PPB buffer concentration from 100 to 200 mM, the conversion to the desired product after 24 h increased from 85% to 94% (Figure 9). The results have been confirmed in a second experiment (92%).

Studies on the Influence of the Superabsorber Particle Size on the Acetophenone Reduction with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
Another process parameter of potential relevance was the interface area of organic and aqueous phase. By powdering the superabsorber granulate, the interface area, and therefore the mass transfer phenomena, should be significantly increased. For the preparation of superabsorber powder (<100 µ m), the superabsorber granulate was powdered by grinding in a mortar and sieving with a steel mesh.
As shown in Figure 10, a smaller particle size led to lower conversion. After 25 h, 6% less product (R)-2 was formed compared to larger particle size after 23 h. Considering that the surface of the superabsorber particles was increased, phase transfer was not a limiting factor for this system. Due to smaller aqueous superabsorber particles, the solvent played a major role in deactivation of enzymes, as enzymes might be more often near to the surface. Another process parameter of potential relevance was the interface area of organic and aqueous phase. By powdering the superabsorber granulate, the interface area, and therefore the mass transfer phenomena, should be significantly increased. For the preparation of superabsorber powder (<100 µm), the superabsorber granulate was powdered by grinding in a mortar and sieving with a steel mesh.
As shown in Figure 10, a smaller particle size led to lower conversion. After 25 h, 6% less product (R)-2 was formed compared to larger particle size after 23 h. Considering that the surface of the superabsorber particles was increased, phase transfer was not a limiting factor for this system. Due to smaller aqueous superabsorber particles, the solvent played a major role in deactivation of enzymes, as enzymes might be more often near to the surface.

Studies on the Influence of the Biocatalyst Loading on the Acetophenone Reduction with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
Finally, the ADH loading was increased form 2 U mL −1 to 4 U mL −1 . The amount of cofactor regeneration system was adjusted accordingly.
As expected, the reaction using more catalyst performed better (85% vs. 92% after 23 h, Figure  10). This result also supported the assumption that mass transfer was not a limiting factor in the reaction system, as the conversion can be increased using more catalysts. Enzyme deactivation, however, was likely to have played a major role in this reaction because at lower catalyst loading (2 U mL −1 ) prolonging the reaction time after 8 h, no significant further increase of conversion could be observed ( Figure 11).

Studies on the Influence of the Biocatalyst Loading on the Acetophenone Reduction with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
Finally, the ADH loading was increased form 2 U mL −1 to 4 U mL −1 . The amount of cofactor regeneration system was adjusted accordingly.
As expected, the reaction using more catalyst performed better (85% vs. 92% after 23 h, Figure 10). This result also supported the assumption that mass transfer was not a limiting factor in the reaction system, as the conversion can be increased using more catalysts. Enzyme deactivation, however, was likely to have played a major role in this reaction because at lower catalyst loading (2 U mL−1) prolonging the reaction time after 8 h, no significant further increase of conversion could be observed ( Figure 11).
As for the tested solvents, it could be shown that MTBE was the best performing one. As expected, solvents that have previously shown to be not tolerated by enzymes, such as DCM, turned out not to be suitable for the investigated reaction system. MTBE and diethyl carbonate have both shown high conversions, indicating deactivation of the enzyme to a lesser extent. Besides stability issues, according to the solvent-dependent partition coefficient, substrate distribution in the aqueous and organic phases might also have played a role here. In the case of an unfavored partition coefficient (caused by a very high solubility of the substrate in the organic phase), the concentration of substrate in the aqueous superabsorber system might have been below the K M value, thus slowing down the overall process.  Figure 11. Influence of catalyst loading (in U mL −1 ) on the conversion to phenylethanol (R)-2 in an enzymatic reduction of acetophenone (1) using ADH in superabsorber-immobilized aqueous phase.
As for the tested solvents, it could be shown that MTBE was the best performing one. As expected, solvents that have previously shown to be not tolerated by enzymes, such as DCM, turned out not to be suitable for the investigated reaction system. MTBE and diethyl carbonate have both shown high conversions, indicating deactivation of the enzyme to a lesser extent. Besides stability issues, according to the solvent-dependent partition coefficient, substrate distribution in the aqueous and organic phases might also have played a role here. In the case of an unfavored partition coefficient (caused by a very high solubility of the substrate in the organic phase), the concentration of substrate in the aqueous superabsorber system might have been below the KM value, thus slowing down the overall process. Figure 12 shows a schematic flow process for an enzymatic reduction of acetophenone (1) to phenylethanol (R)-2, coupled with a GDH-based cofactor regeneration system. The packed bed reactor was loaded with the immobilized aqueous phase, containing the enzymes, cofactor, and glucose. A total amount of 2 U ADH and a substrate concentration of 100 mM in MTBE were used for the experiments. For the setup, an in-house custom build glass tube reactor (ID 5 mm) was used ( Figure 13). Figure 11. Influence of catalyst loading (in U mL −1 ) on the conversion to phenylethanol (R)-2 in an enzymatic reduction of acetophenone (1) using ADH in superabsorber-immobilized aqueous phase. Figure 12 shows a schematic flow process for an enzymatic reduction of acetophenone (1) to phenylethanol (R)-2, coupled with a GDH-based cofactor regeneration system. The packed bed reactor was loaded with the immobilized aqueous phase, containing the enzymes, cofactor, and glucose. A total amount of 2 U ADH and a substrate concentration of 100 mM in MTBE were used for the experiments. For the setup, an in-house custom build glass tube reactor (ID 5 mm) was used ( Figure 13).  Figure 11. Influence of catalyst loading (in U mL −1 ) on the conversion to phenylethanol (R)-2 in an enzymatic reduction of acetophenone (1) using ADH in superabsorber-immobilized aqueous phase.

Set-Up of a Packed Bed Reactor
As for the tested solvents, it could be shown that MTBE was the best performing one. As expected, solvents that have previously shown to be not tolerated by enzymes, such as DCM, turned out not to be suitable for the investigated reaction system. MTBE and diethyl carbonate have both shown high conversions, indicating deactivation of the enzyme to a lesser extent. Besides stability issues, according to the solvent-dependent partition coefficient, substrate distribution in the aqueous and organic phases might also have played a role here. In the case of an unfavored partition coefficient (caused by a very high solubility of the substrate in the organic phase), the concentration of substrate in the aqueous superabsorber system might have been below the KM value, thus slowing down the overall process. Figure 12 shows a schematic flow process for an enzymatic reduction of acetophenone (1) to phenylethanol (R)-2, coupled with a GDH-based cofactor regeneration system. The packed bed reactor was loaded with the immobilized aqueous phase, containing the enzymes, cofactor, and glucose. A total amount of 2 U ADH and a substrate concentration of 100 mM in MTBE were used for the experiments. For the setup, an in-house custom build glass tube reactor (ID 5 mm) was used ( Figure 13).  The reactor was fed with a solution of ketone 1 in MTBE using a syringe pump. A residence time of 1 hour was set and collected fractions of the resulting product solution were analyzed by GC. Because the catalyst was immobilized, no separation or quenching of the reaction mixture was required. Figure 13. Developed glass reactor packed with superabsorber gel.

Studies on the Acetophenone Reduction in a Flow Process with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
The reactor was charged with different volumes of immobilized aqueous phase. The amounts of superabsorber and catalytic system (ADH, GDH, glucose, NADPH) were kept constant, whereas different volumes of buffer were used to adjust the overall reactor bed volume (RV). After charging of the reactor, the end pieces of the reactor were adjusted to the catalyst bed. The average conversions of collected fractions are shown in Figure 14.
As for the larger reactor volume, the initial conversion to alcohol (R)-2 in the first hour was very high (67%). However, the conversion decreased very quickly to only 24% after 5 h. In the case of the smaller reactor bed volume, the conversion to alcohol (R)-2 for the first fractions collected from the start to the first hour was about 61%. However, the conversion was not constant over a longer period, and dropped to 33% after 7 h of reactor operating time (or seven residence times). Compared to the larger reactor volume, the system was more stable. The smaller system was probably more stable due to lower flow rates and therefore less organic solvent pumped through the reactor bed. In accordance with the batch experiments, the stability of the catalyst was a major issue.  The reactor was fed with a solution of ketone 1 in MTBE using a syringe pump. A residence time of 1 hour was set and collected fractions of the resulting product solution were analyzed by GC. Because the catalyst was immobilized, no separation or quenching of the reaction mixture was required.

Studies on the Acetophenone Reduction in a Flow Process with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
The reactor was charged with different volumes of immobilized aqueous phase. The amounts of superabsorber and catalytic system (ADH, GDH, glucose, NADPH) were kept constant, whereas different volumes of buffer were used to adjust the overall reactor bed volume (R V ). After charging of the reactor, the end pieces of the reactor were adjusted to the catalyst bed. The average conversions of collected fractions are shown in Figure 14.

Studies on the Acetophenone Reduction in a Flow Process with a Superabsorber-Immobilized Recombinant Alcohol Dehydrogenase from Lactobacillus brevis
The reactor was charged with different volumes of immobilized aqueous phase. The amounts of superabsorber and catalytic system (ADH, GDH, glucose, NADPH) were kept constant, whereas different volumes of buffer were used to adjust the overall reactor bed volume (RV). After charging of the reactor, the end pieces of the reactor were adjusted to the catalyst bed. The average conversions of collected fractions are shown in Figure 14.
As for the larger reactor volume, the initial conversion to alcohol (R)-2 in the first hour was very high (67%). However, the conversion decreased very quickly to only 24% after 5 h. In the case of the smaller reactor bed volume, the conversion to alcohol (R)-2 for the first fractions collected from the start to the first hour was about 61%. However, the conversion was not constant over a longer period, and dropped to 33% after 7 h of reactor operating time (or seven residence times). Compared to the larger reactor volume, the system was more stable. The smaller system was probably more stable due to lower flow rates and therefore less organic solvent pumped through the reactor bed. In accordance with the batch experiments, the stability of the catalyst was a major issue.  As for the larger reactor volume, the initial conversion to alcohol (R)-2 in the first hour was very high (67%). However, the conversion decreased very quickly to only 24% after 5 h. In the case of the smaller reactor bed volume, the conversion to alcohol (R)-2 for the first fractions collected from the start to the first hour was about 61%. However, the conversion was not constant over a longer period, and dropped to 33% after 7 h of reactor operating time (or seven residence times). Compared to the larger reactor volume, the system was more stable. The smaller system was probably more stable due to lower flow rates and therefore less organic solvent pumped through the reactor bed. In accordance with the batch experiments, the stability of the catalyst was a major issue.
Compared to batch processes, a much higher initial conversion within 1 h reaction time could be achieved. Due to the nature of a packed bed reactor, the catalyst could be used more efficiently, as the catalyst concentration in the reaction was much higher (a summary can be found in Table 3). It was successfully shown that a superabsorber-immobilisate can be applied in a packed bed reactor, thus offering an alternative to the standard approach using carrier-immobilized biocatalysts.

Conclusions
In conclusion, different factors have shown to be more or less important for the investigated reduction of acetophenone (1) in the presence of the LB-ADH (summarized in Table 4). The choice of solvent turned out to represent a crucial issue for this process. Although MTBE provided high conversions, DCM, for instance, led to deactivation of the catalytic system under the same reaction conditions. Mass transfer seemed to not be a limiting factor for this system under the investigated conditions. Table 4. Influence of different parameters on the reduction of acetophenone (1) in a biphasic superabsorber/MTBE system, considering conversions to alcohol (R)-2 after approximately 24 h in comparison to an initial standard experiment (Section 3.1). Legend: − negative impact, + positive impact, parentheses indicate small impact.

Parameter
Influence Relative Conversion solvent (Section 3.2) -up to −75% longer incubation time (Section 3.3) -−75% higher superabsorber-solvent ratio (Section 3.4) (+) +7% higher buffer concentration (Section 3.5) (+) +8% higher catalyst loading (Section 3.7) (+) +7% smaller particle size (Section 3.6) (−) −6% Furthermore, the application of superabsorber-immobilized LB-ADH with a cofactor regeneration system in a packed bed reactor, thus running in a flow mode, was demonstrated. However, the stability of the immobilized enzymes under the investigated conditions was shown to be insufficient. Nevertheless, very high initial conversions for the PBR setup were observed. Further process development and extension of this PBR methodology with superabsorber-based redox enzymes towards other applications in biocatalytic ketone reduction and related redox biotransformations are currently in progress. In addition, in general (and thus, also for other types of enzymes) this type of PBR methodology represents an alternative to the use of standard carrier-immobilized biocatalysts, and this system also offers the perspective for applications beyond isolated enzymes and, for example, for whole cell catalysts [51].     Appendix A.5 Standard Procedure 3 (SP3): Synthesis of 1-Phenylethanol (R)-2 Using Superabsorber-Immobilized Enzymes A glass reactor was filled with a layer of cotton (5 mm), then superabsorber (35 mg, Favor SXM 9155) was added. A solution of LB-ADH (crude extract in potassium phosphate buffer, 9.4 µL, 320 U mL −1 ), GDH (crude extract in potassium phosphate buffer, 31.9 µL, 188 U mL −1 ), glucose (300 µL of 1 M solution in potassium phosphate buffer (100 mM, pH 7)), and NADPH (20 µL of 10 mM solution in water) in potassium phosphate buffer (1.5 mL, 100 mM, pH 7) was pumped into the reactor to be immobilized at room temperature for 30 min. The reactor was filled with an additional cotton layer (5 mm) and sealed. A solution of acetophenone (35.1 µL, 0.3 mmol) in MTBE (3 mL) was transferred into a syringe (5 mL, from S.G.E. analytics, gas tight, 10.3 mm ID), which was attached to a syringe pump. The pump was connected to the reactor and set to a flow rate of 0.2 mL min −1 corresponding to a residence time of 1 h. Fractions were collected, and the conversion was determined via GC analysis.