Efficient Nicotinamide Adenine Dinucleotide Phosphate [NADP(H)] Recycling in Closed-Loop Continuous Flow Biocatalysis

Article abstract of DOI:10.1002/adsc.202000058

Biocatalytic redox reactions regularly depend on expensive cofactors that require recycling. For continuous conversions in flow chemistry, this is often an obstacle since the cofactor is washed away. Here, we present a quasi-stationary recycling system for nicotinamide adenine dinucleotide phosphate utilizing an immobilized alcohol dehydrogenase. Four model substrates were reduced with high enantioselectivity as a proof of concept. The two-phase system enables continuous production as well as quick substrate changes. This setup may serve as a general cofactor regeneration module for continuous biocatalytic devices employing (co-)substrates being miscible in organic solvent. The system resulted in space-time yields up to 117 g L^?1 h^?1 and total turnover numbers for nicotinamide adenine dinucleotide phosphate higher than 12,000 mol/mol are possible. (Figure presented.).

Full text of DOI:10.1002/adsc.202000058

Title: Efficient Nicotinamide Adenine Dinucleotide Phosphate
NADP(H)] Recycling in Closed-Loop Continuous Flow
[
Biocatalysis
Authors: Benedikt Baumer, Thomas Classen, Martina Pohl, and Jörg
Pietruszka
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Adv. Synth. Catal. 10.1002/adsc.202000058
Link to VoR: http://dx.doi.org/10.1002/adsc.202000058

Advanced Synthesis & Catalysis
10.1002/adsc.202000058
VERY IMPORTANT PUBLICATION
FULL PAPER
DOI: 10.1002/adsc.201((will be filled in by the editorial staff))
Efficient Nicotinamide Adenine Dinucleotide Phosphate
[NADP(H)] Recycling in Closed-Loop Continuous Flow
Biocatalysis
[
a]
[b]
[b]
[a,b]
Benedikt Baumer , Thomas Classen , Martina Pohl and Jörg Pietruszka*
a
Institut für Bioorganische Chemie der Heinrich-Heine-Universität Düsseldorf im Forschungszentrum Jülich,
Stetternicher Forst, Geb. 15.8, D-52426 Jülich, Germany
E-mail: j.pietruszka@fz-juelich.de
Institut für Bio- und Geowissenschaften (IBG-1: Biotechnologie), Forschungszentrum Jülich GmbH, D-
b
5
2456 Jülich, Germany
Received: ((will be filled in by the editorial staff))
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/adsc.201######.((Please delete if not appropriate))
Abstract. Biocatalytic redox reactions regularly depend on regeneration module for continuous biocatalytic devices
expensive cofactors that require recycling. For continuous employing (co-)substrates being miscible in organic
conversions in flow chemistry, this is often an obstacle since solvent. The system performed space-time yields up to
-
1
-1
the cofactor is washed away. Here, we present a quasi- 117 g L h and total turnover numbers for nicotinamide
stationary recycling system for nicotinamide adenine adenine dinucleotide phosphate higher than 12,000 mol/mol
dinucleotide phosphate utilizing an immobilized alcohol are possible.
dehydrogenase and four model substrates were reduced with
high enantioselectivity as a proof of concept. The two-phase Keywords: cofactor regeneration; enzyme catalysis; flow
system enables continuous production as well as quick chemistry; oxidoreductases; phase separation
substrate changes. This setup may serve as general cofactor
1
.A). Three different conceptual approaches have
[6]
Introduction
been developed in the past, which will be briefly
explained below (Scheme 1.B-D).
Biocatalysis is playing an ever-increasing role in the The first concept (Scheme 1.B) is a simple flow-
synthesis of fine chemicals, pharmaceuticals, and through system. Usually, a mixture of substrate,
natural products.[ Enzymes are already widely used cofactor, buffer etc. is pumped through a reactor filled
in industrial applications as they readily enable access with immobilised enzyme. The substrate is converted
of highly functionalised products due to their high into product and the entire solution leaves the reactor
1]
[2]
regio-, stereo- and enantioselectivity. At the same again including the cofactor. Different cofactor
time, also flow chemistry approaches affected many regeneration methods are used, but since it is not
[
3]
fields of synthesis including biocatalysis:
the recirculated but washed out, the cofactor consumption
catalyst is immobilised within a reactor, the substrates and costs, respectively, will scale proportional to the
[7]
are passed by and the final product should rinse out substrate amount. The second concept towards
ideally without any need of purification.[ However, continuous cofactor regeneration (Scheme 1.C)
while this looks conceptually simple, the utilizes (co-)immobilised or (co-)entrapped cofactor,
4]
implementation is often challenging, especially when allowing a stream of substrate, buffer etc. to pass by
it comes to cofactor regeneration.[5] In the present and perform several reaction cycles.
[3i, 6b, 8]
In case of
work a standalone module for chemoenzymatic immobilisation, just recent work by Scott et al.
reactions in continuous flow with regard to closed- demonstrated the use of PEG-immobilised NAD(H)
loop cofactor regeneration was developed. The proof as well as ATP and made use of the flexibility of a
of principle was demonstrated using a nicotinamide linker to reach the active catalytic centre. Excellent
adenine dinucleotide phosphate [(NADP(H)]- space-time yield (STY) and total turnover number
[6a]
dependent alcohol dehydrogenase (ADH, Scheme (TTN) over 10,000 mol/mol could be achieved.
1
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.202000058
Moreover, in recent years the (co-)entrapment of the system. The cofactor remains inside the aqueous phase,
cofactor in matrices has also moved into focus and led while the product and acetone (co-product) are extracted
to impressive results, such as remarkable TTNs over into the organic layer to leave the column. The protein
4,000 mol/mol.[
7f, 7g]
Unlike the first concept structure was generated from pdb:1zk4.
[12]
1
(
Scheme 1.B), no continuous addition of cofactor is
The presented continuous cofactor regeneration
systems demonstrated impressively the proof-of-
principle. However, an ideal cofactor regeneration
system serving as a blueprint for future applications
would need to fulfil several requirements:
required. However, these methods depend on
chemical modifications of the cofactor and enzymes
tolerating these modification and entrapment,
respectively. Thus, the method cannot be considered
as a general solution.



Catalytic use of the cofactor: The cofactor should
remain within the oxidoreduction-module or
should be recirculated. There should be no further
addition of cofactor to run the system
continuously. In addition, the cofactor should not
be diluted during the course of reaction.
Modularity: The system should be a standalone
module that can easily be integrated into flow
cascades. Various substrates can be applied to the
very same setup without cross-contaminations.
The cofactor regeneration can be used for various
enzymes rather than single cases.
Easy handling: There should be no need for
chemical modifications of the respective cofactor
to ensure the natural function of the enzyme. The
system should be independent of the kind of
immobilisation of the biocatalyst.
Scheme 1.D shows
a
closed-loop cofactor
regeneration system separating an aqueous layer
including the ionic cofactor (e.g. NADH) from the
organic layer containing the product. This
discrimination in polarity allows recirculation of the
cofactor-layer. Šalić et al. used two micromixers in
series for this purpose, one for product formation and
one for cofactor regeneration. The process remained
stable for three days under continuous conditions
without addition of fresh cofactor.[
9]
Another
innovative system was recently published by Paradisi
and Contente performing functional group
a
interconversion from amines into alcohols. After
phase separation the aqueous layer was refed into the
+
system. A high TTN of 2,000 mol/mol for NADP
was achieved.[
6a, 10]
Nevertheless, it was reported that
a dilution of the substrate feed could not be prevented.
To setup a new modular system, we based our
work on the studies of Döbber et al. using an
immobilised Halo-tagged alcohol dehydrogenase
LbADH from L. brevis. It catalyses the
enantioselective asymmetric reduction of prochiral
ketones to the corresponding secondary alcohols
[7a, 7b]
(
Scheme 1.A).
We propose a standalone closed-
loop cofactor regeneration system using a membrane-
based phase separation technique. It enables the usage
of cofactor-dependent enzymes in continuous flow
chemistry and previously impossible rapid substrate
changes without changing the recirculated cofactor
solution.
Results and Discussion
Simple flow-through system
Initially a simple flow-through system (cf. Scheme
1
.B) with co-substrate (2-propanol)-coupled cofactor
regeneration was established. This has been extended
by a Syrris Asia FLLEX (Flow Liquid Liquid
EXtraction) system (see Figure 1). It allows a simple
phase separation of an aqueous cofactor-containing
layer and an organic product-containing layer. As
organic solvent, ethyl acetate was introduced via
channel 2. The setup was utilized to determine phase
Scheme 1.A) Overview of the LbADH from Lacto- separation efficiency as well as a screening for
bacillus brevis catalysed asymmetric reduction of ketones preliminary
process
parameters.
HaloTag
[
7a, 7b, 7f, 7i, 8c,
including a substrate-based cofactor recycling.
immobilised alcohol dehydrogenase LbADH was
used in a packed-bed reactor to reduce acetophenone
11]
B) Simple flow-through system. A substrate together
with the cofactor NADP /NADPH are pumped through a
reactor filled with immobilised enzyme. C) Flow-through
+
(
(
1a) asymmetrically yielding (R)-phenylethan-1-ol
2a) as the product (Scheme 2).
systems with immobilised enzyme and immobilised or Various reactions were carried out. For this purpose,
entrapped cofactor. D) Closed-loop cofactor regeneration the flow rates (FR) were varied from 30 up to
2
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.202000058
5
00 µL/min and different concentrations of the co- and the co-substrate 2-propanol has been added to the
substrate 2-propanol were used. For a concentration biocatalyst cartridge and the aqueous layer containing
of 50 mM of 1a, 10% (v/v) 2-propanol and a flow rate the cofactor (channel 1) is refed to the cartridge after
of 30 µL/min, a turnover of 92% was achieved (lit. passage and phase separation, respectively. This has
[7a]
95% ). This turnover remained constant even at two advantages: the cofactor can be used in catalytic
higher flow rates of up to 65 µL/min, which means a amounts and multiple turnovers are performed during
doubled productivity and STY with respect to the retention time in the enzyme reactor allowing for
previous results.[ An increase in the 2-propanol full-conversion in dependence of the flow-rate and
concentration to 15% (v/v) resulted in marginal catalyst loading. Once the mixture leaves the enzyme
increase of conversion of approx. 1%. For further reactor an organic solvent (channel 3) is added for
details see supporting information (SI). Furthermore, product extraction. Finally, the layers are separated by
the phase separation was possible and thus was the FLLEX-system, which also prevents any layer
applied in the advanced systems described below. from dilution over time, which is one of the issues
7a]
[10a]
Although many organic substrates are to a certain described in literature.
The organic layer contains
extend water soluble, they preferably solve in organic the product, while the aqueous, NADP(H)-containing
layers whilst the charged cofactor strictly remains layer is again mixed with fresh substrate dissolved in
within aqueous layers. So, it is possible to carry-out the co-substrate 2-propanol. The substrate/co-
the reaction in aqueous phase or even in an substrate is fed and the product is extracted
aqueous/organic-mixture and extract the cofactor continuously, respectively, whilst the cofactor
[10a]
afterwards to feed it again to the enzyme reactor.
remains within the recirculated aqueous layer. Due to
this closed-loop set up NADP(H) is no longer
removed with the product, but efficiently regenerated,
which enables a further decrease to catalytic amounts
(
0.1 mol% relative to the substrate).
Figure 1. Schematic representation of the simple flow-
through system. It consists of two pumps: One for the
reaction mixture (channel 1) consisting of the substrate,
+
NADP , 2-propanol and KPi-buffer and one for the organic
extraction solvent (channel 2). The packed-bed-reactor is
filled with Halo-tagged LbADH and a FLLEX is applied
for phase separation. Colour code: Blue indicates the
aqueous, red the organic and purple the mixed stream.
Acetone and 2-propanol are found in both layers after
extraction.
Figure 2. Schematic overview of our system for
continuous cofactor regeneration. It consists of two pumps
containing four separated channels in total. Channel 1 is
used for the aqueous/cofactor stream and 2 for the substrate
and co-substrate (2-propanol). Channel 3 delivers
Scheme 2. Reduction of acetophenone (1a) to (R)- extraction solvent. 1 and 2 are connected via a y-piece. The
phenylethan-1-ol (2a) by substrate-coupled regeneration of column is filled with immobilised HaloTag-LbADH as
[
7a]
NADPH using an immobilised LbADH variant.
catalyst. The last and important device is the Asia FLLEX
for extraction and phase separation. Colour code: Blue
indicates the aqueous, red the organic and purple the mixed
stream.
Closed-loop cofactor regeneration system
Structure of device: The new advanced closed-loop
cofactor regeneration system is shown in Figure 2. In Comparing simple flow-through vs closed-loop
contrast to the simple flow-through system (Figure 1), system: A comparison of the simple flow-through
a third channel (channel 2) providing the substrate system with the closed-loop has been performed.
3
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.202000058
-1
-1
Therefore, the flow rates were optimized and could be obtained with a STY of 121 g L h (96%).
compared with the previous results (see SI table 2 and Even longer runs were possible without loss on
[7a]
ref. ) of the very same reaction (Scheme 2). Two performance (Figure 3). The TTN could be drastically
+
mixtures were prepared: 1.00 mL of 0.6 mM NADP
in KP for channel 1 (aqueous layer) and 500 mM our knowledge the highest value in literature for
acetophenone (1a) in 2-propanol for channel 2. closed-loop continuous cofactor regeneration
increased to 12,855 mol/mol, which is to the best of
i
[6a, 9-10]
Conversions were determined by gas chromatography systems.
GC) and the flow rates and results are shown in
(
Table 1. The screening results show that the closed-
loop system produces comparable results to the
simple flow-through system. For example, for a
combined flow rate of approx. 30 µL/min, a
conversion of 96% was achieved (simple flow-
[7a]
through system 92% and lit. 95%).
Table 1. Optimisation of flow rates with closed-loop
cofactor regeneration system according to Figure 2. The
given errors are technical standard deviations (SD) from
individual runs (N=3).
flow rate [µL/min]
conversion
channel 1
channel 2
channel 3 [%]±SD
2
3
4
5
5
7.0
6.0
5.0
4.0
9.0
2.7
3.6
4.5
5.4
5.9
29.7
96.6±1.8
92.5±0.4
92.4±0.4
91.4±0.1
93.0±1.4
39.6
49.5
59.4
64.9
Application on different substrates and continuous
run: The reliability of the system has been
demonstrated for the reduction of acetophenone (1a).
Scheme 3. Reduction of different ketones 1a-1d to the
corresponding alcohols 2a-2d using HaloTag-LbADH in
flow chemistry with continuous cofactor regeneration
system (Figure 2). Isolated yields after purification. The
enantiomeric excesses were determined by GC and HPLC,
respectively. The configuration was validated by optical
rotatory power. Space-time yield (STY) was calculated as
shown in the supporting information. The total turnover
[7a, 7c, 7f,
But as LbADH shows a broad substrate range,
1
1a, 11c, 11d, 13]
the closed-loop cofactor regeneration
system was tested for three further substrates as well.
Different ketones 1b-1d were reduced to the
corresponding secondary alcohols 2b-2d. To obtain
high STY, some flow rate optimization was carried
+
number (TTN) refers to NADP consumption.
+
out and for the ketones 1a and 1d, the NADP -
concentration was lowered from 0.6 mM to 0.3 mM.
More details of the studies can be found in the
supporting information and the respective conditions
are shown in Scheme 3.
-
1
-1
Moderate to high space-time yields from 14 g L h
-1
-1
up to 117 g L h and high to very high total turnover
numbers from 128 up to 2023 could be achieved with
excellent enantioselectivities (>99%). Furthermore,
the aqueous phase was extracted and analyzed with
respect to the absence of substrate/product. In fact,
signals neither for substrate nor for product could be
detected in the extracted aqueous layer (NMR, see
SI). Moreover, three different extraction solvents such
as dichloromethane, ethyl acetate, and diisopropyl
ether were tested successfully, which offers a certain
flexibility for follow-up reactions in future studies.
Furthermore, the system including closed-loop
substrate-coupled cofactor regeneration demonstrated
its reliability by continuous runs over 32 h without Figure 3. Performance of closed-loop continuous cofactor
loss in performance. In total, 1.36 g of alcohol 2d regeneration system over 32 h. A 500 mM solution of
4
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.202000058
ketone 1d was pumped through the system at a flow rate the biocatalytic cartridge (steady-state) since no
of90/9/99 µL/min (channel 1/2/3) and converted into dilution takes place. We demonstrated excellent
+
alcohol 2d by using one stock solution of NADP . performance with full conversion for the reduction of
1
Conversion was determined by H-NMR. The collection ketone (1d) in a 32 h run. Four different ketones
was done in time mode and samples were collected for 2 h could be reduced to the corresponding alcohols in
per test tube.
good to very good yields and excellent
enantioselectivities. Different organic solvents were
successfully tested for extraction of the products
Application on serial conversions: The sound phase- using the Asia FLLEX module. The system is a
separation system offers the opportunity for using the closed module, which can be used for serial
very same setup including the very same batch of biotransformations with the very same setup. As we
enzyme and cofactor for several transformations. To have demonstrated, different ketones can be reduced
prevent cross-contaminations it was necessary to have with just one batch of cofactor solution, when
an intermediate flushing giving the benefit of total implementing rinsing steps.
product recuperation from the device (Figure 4). Finally, while the system has its limitations if
Following the setup shown in Figure 2, only a rinsing exclusively water-soluble substrates including
step after each reduction was added. After the cofactor-regenerating substrates such as glucose are
reduction of the first ketone (1d, step 1), a solution of used, it can be anticipated that it represents a generic
a water miscible organic solvent (here 300 mM of module that could also be used with other
acetone) in KP
i
is pumped through the system (step oxidoreductases, either in separate or parallel mode. If
2
). Then, the second ketone (1c) was reduced (step 3), this approach is not possible, one can also implement
followed by the same rinsing step (step 4). In fact, enzyme-coupled cofactor regeneration by including a
this procedure worked well as none of the products as second oxidoreductase for that purpose. In addition,
well as the aqueous phase were contaminated with the system might be combined with other preceding
substrate/product from a previous run. Alcohol 2d or follow-up reaction modules in organic solvent.
could be received in 92% and alcohol 2c in 87%
yield. Furthermore, in another experiment it was
found that the washing step has no influence on the Experimental Section
activity of ADH (see SI, table 8).
General information
The starting materials used in this work were commercially
purchased by Merck KGaA, TCI International and Alpha
Aesar with the exception of ketone 1c and 1d. They were
[
14]
synthesised according to literature procedures.
The
solvents used were in GC grade purity. For flash
chromatography Macherey-Nagel 60M (0.040-0.063 mm,
2
30-400 mesh) was used as silica gel. For analytical TLC
finished films from Macherey-Nagel, type POLYGRAM®
SIL G/UV254 with fluorescence indicator were used. The
plates were developed using a UV lamp at a wavelength of
2
54 nm and/or by staining with an aqueous potassium
Figure 4. Schematic overview of a consecutive experiment
using the setup shown in Figure 2 with one cofactor
permanganate solution. NMR spectra were measured on a
TM
Bruker 600 Ultra Shield with a frequency of 600 MHz
1
13
solution for the whole experiment. Substrate 1 [500 mM in for H- and 151 MHz for C-spectra. Only deuterated
solvents were used. Chemical shifts were determined by
2
-propanol, flow rate: 90/9/99 µL/min (channel 1/2/3)]. is
using solvent or tetramethyl silane as internal standard and
used for the reaction (step 1), followed by a short rinsing
step with 300 mM acetone in KP
are given in ppm. Coupling constants J are given in Hz.
i
(step 2). Then substrate 2 Multiplets are described as s (singlet), d (doublet), t
(
(
200 mM in 2-propanol, FR: 50/5/55 µL/min) is reduced (triplet), q (quartet) and p (pentet). Furthermore, for
1 1 1 13
evaluation DEPT-(135°-pulse)-, H- H-COSY-, H- C-
step 3) and followed by another rinsing step (step 4).
1
13
HSQC- and H- C-HMBC-spectra were taken. IR: Infrared
spectra were recorded on PerkinElmer FT-IR
a
SpectrumTwo spectrometer. Measurements were taken in
-
1
ATR mode. The absorption bands were given in ῦ [cm ].
The rotary power was measured on a KRÜSS P8000-TF.
Conclusion
3
Probes were solved in CHCl and measured in a 5.00 cm
In summary, a well-functioning system for ADHs and cuvette at 589 nm (sodium-D-line). The specific rotation
water-insoluble substrates was developed with the was calculated with the Biot equation. HPLC: HPLC
enantiomeric analysis was done on a Thermo Scientific
help of a phase-separation system (here Asia FLLEX).
Dionex UltiMate 3000. As stationary phase a Phenomenex
As a basis, the already established system of LbADH
Lux Amylose and as mobile phase n-heptane:2-propanol
and 2-propanol as co-substrate was used. Since a (90:10) was used. GC: Conversion and enantiomeric excess
were measured with gas-liquid chromatography. GC-
permanent recycling and regeneration of the cofactor
chromatograms were measured on a Thermo Scientic
was possible, only catalytic quantities were required.
It is a closed-loop setup allowing for continuous runs
®
TRACE GC Ultra. The probes were dissolved in MTBE.
2
Carrier gas was H at 0.6 bar. For the detection, a FID-
without subsequent supplementation of cofactor and detector was used. For exact HPLC and GC methods see
SI.
the concentration of which should be constant within
5
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.202000058
(
3.00 mm x 50.0 mm) column. Different flow rate
Syrris Asia flow devices:
combinations were used (see Table 1 and SI Table 3).
Recirculation of the aqueous layer started right after the
dead volume of the system was overcome. After finishing
the reactions, the conversion was determined by GC. The
Syringe Pump: For the attempts with a flow rate up to
2
00 µL/min, the pumps were equipped with yellow
syringes (50.0 µL/100 µL). For greater flow rates, they
were equipped with blue ones (500 µL/1.00 mL).
1
results are shown in Table 1 and SI Table 3. H NMR (600
MHz, CDCl ): δ 7.44 – 7.32 (m, 4H, ArH), 7.31 – 7.25 (m,
3
3
1
H, 5-H), 4.91 (q, J1,1’=6.5 Hz, 1-H), 1.81 (s, 1H, OH),
3 13
FLLEX: The FLLEX was equipped with original FLLEX 1.51 (d, J =6.5 Hz, 1‘-H). C NMR (151 MHz, CDCl ):
1’,1
3
PTFE or Merck FHLP02500 PTFE membranes. In the last δ 146.15 (C-2), 128.86 (C-3,7/4,6), 127.84 (C-5), 125.73
case, the polyethylene support was removed. The total (C-3,7/4,6), 70.80 (C-1), 25.53 (C-1‘). IR (ATR): 3339,
system pressure was set to 3.00 bar and the cross- 3064, 3029, 2973, 2928, 2877, 1493, 1450, 1368, 1284,
membrane pressure (CMP) was dependent on the solvent 1261, 1203, 1096, 1076, 1029, 1010, 997, 897, 759, 697,
2
0
(100 mbar for ethyl acetate, 50.0 mbar for CH
50.0 mbar for diisopropyl ether).
2
Cl
2
and 606, 539. R = 0.40 (n-pentane:EtOAc 8:2), [α] = +44,6
f
D
(c=1.16 in CHCl )
3
Automated collector: The collector, type: Gilson FC 203B,
was used for the collection of products.
(
R)-3-[(Benzyloxy)methyl]but-3-en-2-ol (2b): Synthesis
with closed-loop continuous regeneration system
Software: The controlling of the devices was done with the
Asia Manager PC Software (1.69 beta) and herein
presented flow rates were defined in the pump settings.
As basis, the setup shown in Figure 2 was used. Channel 1
2 i
was filled with 1.00 mL 0.5 mM NADPNa -solution in KP .
Channel 2 was filled with a 200 mM solution of ketone 1b.
Therefore, 23.2 mg (0.12 mmol) were solved in 583 µL 2-
propanol. As extraction solvent in channel 3 diisopropyl
ether was used. As catalyst an immobilised HaloTag-
Enzyme digestion and immobilization procedure
LbADH
filled
into
an
Omnifit
column
2
8
.00 g cell pellet of HaloTag-LbADH was resuspended in
(
3.00 mm x 50.0 mm) was used. The flow rate combination
i 2
.00 mL KP (100 mM, pH 7, 1.00 mM MgCl ). For cell
was set to 30/3/33 µL/min (aqueous/organic/extraction).
Both the collection of the organic layer and the
recirculation of the aqueous layer started right after the
dead volume of the system was overcome. After the
reservoir of channel 2 was empty, 2-propanol was added,
and the run was continued. After completion, the organic
lysis ultrasonication has been applied and the crude cell
extract has been centrifuged (7084 rcf). Ultrasonication
was carried out using a Bandelin SONOPLUS with a
SONOPLUS KE76 cone tip. The mode used was 2 x 5 min,
5
x 10% cycle at 35% power. The supernatant was used for
immobilisation
without
further
filtration
steps.
1
layer was evaporated and a H-NMR was taken. A
Immobilisation was done in flow mode. Therefore, an
Omnifit column (3 mm x 50 mm) was filled with HaloLink
resin (~400 mg wet weight) and plugged to an Asia syringe
conversion of 80% (20% of substrate) was achieved and
2
1.0 mg of product/substrate mixture was collected.
Purification of the product was done via flash
chromatography (n-pentane:EtOAc 8:2) and the product
was obtained as yellow oil in 72% yield (16.7 mg,
pump. Afterwards, it was washed with KP
i
-buffer
2
100 mM, pH 7, 1.00 mM MgCl ) for 30 min at a flow rate
(
of 250 µL/min. Then, 6.00 mL of supernatant was pumped
through the column at 30 µL/min and collected afterwards.
-
1
-1
ee <99%) with a STY of 14.1 g L h and a TTN of 128
1
[
7a]
mol/mol. H NMR (600 MHz, DMSO-d ): δ 7.38 – 7.31
6
Finally, the washing step was repeated.
The activity
(
m, 4H; ArH), 7.31 – 7.26 (m, 1H; 9-H), 5.11 (s, 1H; 2’-H),
before immobilization was 768 U and afterwards 406 U.
The difference of 362 U is the theoretical maximum
activity of enzyme in the column.
5
.02 (s, 1H; 2’-H), 4.78 (s, 1H, OH), 4.46 (m, 2H, H-5),
3
3
4
.17 (q, J3,4=6.8 Hz, 1H; 3-H), 4.01 (q, J1gem=13.2 Hz,
3
3
J
1,2’=1.2 Hz, 1H; H-1), 1.17 (d, J4,3=6.5 Hz, 3H; 4-H).
13
C-NMR (151 MHz, DMSO-d
6
): δ 150.39 (C-2), 138.47
(
C-6), 128.24 (C-7,11/8,10), 127.41 (C-9), 127.36 (C-9),
(
R)-Phenylethan-1-ol (2a): Screening with simple flow-
109.24 (C-2‘), 71.28 (C-5), 69.83 (C-1), 66.69 (C-3), 22.58
(
C-4). IR (ATR) 3401, 3030, 2976, 2859, 1652, 1496,
through system
1
6
2
2
454, 1366, 1260, 1204, 1070, 1040, 1028, 908, 735, 696,
-
1
+
02, 467 cm . HRMS (ESI): m/z for C12
H
16
O
2
+Na : calcd:
A stock solution of acetophenone (1a) in different
+
+
15.1043 [M+Na] ; found: 215.1044 [M+Na] ; calcd:
16.1076 [M+Na] ; found: 216.1079 [M+Na] . R = 0.33
f
concentrations (10, 30, 50 mM), 10/15% (v/v) 2-propanol,
+
+
KP
i
(100 mM, pH 7, 1.00 mM MgCl
2
) and NADPNa
2
20
(
퐷
3
PE:EtOAc 8:2). [훼] = +16 (c=1.00 in CHCl ).
(
2.00 mM) was prepared and filled in channel 1 (see Figure
). As extraction solvent ethyl acetate was used. An
1
Omnifit packed-bed reactor (3.00 mm x 50.0 mm) filled
with Halo-tagged LbADH was used as catalyst and Consecutive experiment
reactions were performed in a twofold manner with flow
rates ranging from 30.0 to 500 µL/min. The results are
presented in Table 2 (SI). The determination of conversion
was done by GC.
As basis, the setups shown in Figure 2 and Figure 4 were
used. A stock solution of NADPNa
Therefore 0.5 mg (0.64 µmol) NADPNa
i 2
.00 mL KP (100 mM, pH 7, 1.00 mM MgCl ) and filled
2
was prepared.
was dissolved in
2
1
into channel 1. With this batch, two different ketones were
converted to the corresponding alcohols. Diisopropyl ether
in channel 3 was used as extraction solvent. An Omnifit
column (3.00 mm x 50.0 mm) filled with immobilised
HaloTag-LbADH was used as biocatalytic cartridge. For
(
R)-Phenylethan-1-ol (2a): Screening with closed-loop
continuous regeneration system
Three different channels were needed to perform the step 1, channel 2 was filled with a 500 mM solution of
screening with closed-loop continuous regeneration system ketone 1d. Therefore, 22.9 mg (0.14 mmol) were solved in
(
cf Figure 2). Channel 1 (aqueous cofactor containing 260 µL 2-propanol. The flow rate combination was set to
layer) was filled with 1.00 mL 0.5 mM NADPNa
in KP . Channel 2 (substrate and co-substrate containing collection of the organic layer and the recirculation of the
layer) was filled with 500 mM solution of acetophenone aqueous layer started right after the dead volume of the
2
-solution 90/9/99 µL/min (aqueous/organic/extraction). Both the
i
(
1a) in 2-propanol. Channel 3 (organic extraction solvent) system was overcome. After the reservoir of channel 2 was
was filled with ethyl acetate. As catalyst, an immobilised empty, 2-propanol was added and the run was continued.
Halo-tagged LbADH was used and filled into an Omnifit After completion, an intermediate flushing with 300 mM
6
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.202000058
2
acetone in KPi (100 mM, pH 7, 1.00 mM MgCl ) was
5
433; e) M. Baumann, I. R. Baxendale, S. V. Ley,
performed for 10 min (step 2). Then, step 1 was repeated
for ketone 1c (step 3). This time, 21.0 mg (0.11 mmol)
were solved in 512 µL 2-propanol resulting in a 200 mM
solution. The remaining steps were carried out as described
above, but the flow rate combination was set to
Mol. Diversity 2011, 15, 613-630; f) F. Lévesque,
P. H. Seeberger, Angew. Chem. Int. Ed. 2012, 51,
1706-1709; g) D. Kopetzki, F. Lévesque, P. H.
Seeberger, Chem. Eur. J. 2013, 19, 5450-5456.
a) R. A. Sheldon, J. M. Woodley, Chem. Rev.
2018, 118, 801-838; b) G. Qu, A. Li, Z. Sun, C. G.
Acevedo-Rocha, M. T. Reetz, Angew. Chem. Int.
Ed. 2019, 10.1002/anie.201901491; c) B. M.
Nestl, B. A. Nebel, B. Hauer, Curr. Opin. Chem.
Biol. 2011, 15, 187-193; d) A. Weckbecker, H.
Gröger, W. Hummel, in Biosystems Engineering
I: Creating Superior Biocatalysts (Eds.: C.
Wittmann, R. Krull), Springer Berlin Heidelberg,
Berlin, Heidelberg, 2010, pp. 195-242; e) J.
Britton, C. L. Raston, Chem. Soc. Rev. 2017, 46,
1250-1271.
a) A. Kirschning, W. Solodenko, K. Mennecke,
Chem. Eur. J. 2006, 12, 5972-5990; b) S. V. Ley,
Chem. Rec. 2012, 12, 378-390; c) R. A. Sheldon,
D. Brady, ChemSusChem 2019, 12, 2859-2881; d)
M. B. Plutschack, B. Pieber, K. Gilmore, P. H.
Seeberger, Chem. Rev. 2017, 117, 11796-11893;
e) F. M. Akwi, P. Watts, Chem. Commun. 2018,
54, 13894-13928; f) D. E. Fitzpatrick, C.
5
0/5/55 µL/min and the intermediate flushing was
[
2]
performed for 15 min. The collected organic layers were
1
evaporated and a H-NMR was taken. A full conversion
was achieved and product 2d yielded in 92% as yellow oil
-
1
-1
(
ee <99%, 21.0 mg, STY 114 g L h , after step 1 and 2).
For product 2c, a product/substrate mixture of 97/3 was
collected, which was further purified by flash
chromatography (n-pentane:EtOAc 8:2). Afterwards, the
product was collected as yellow oil in 87% yield (ee <99%,
-
1
-1
1
8.2 mg, STY 30.9 g L h , after step 3 and 4). Product 2c:
1
H NMR (600 MHz, DMSO-d
6
): δ 5.82 – 5.76 (m, 1H; 8-
3
H), 5.12 (d,
J9-trans,8=17.2 Hz, 1H; 9trans-H), 4.97 (d,
3
3
J
9cis,8=10.5 Hz, 1H; 9cis-H), 4.65 (d, JOH,7=4.80 Hz, 1H;
3
OH), 4.04 (q, J1’,2’=7.10 Hz, 2H; 1’-H), 3.89 (m, 1H; 7-h),
3
3
2
.26 (t, J2,3=7.38 Hz, 2H; 2-H), 1.50 (p, J3,4=7.36 Hz,
3
J
3,2=7.38 Hz, 2H; 3-H), 1.38 – 1.33 (m, 2H; 6-H), 1.33 –
3
1
.23 (m, 4h; 4-H, 5-H), 1.17 (t, J2’,1=7.10 Hz, 3H; 2’-H). [3]
13
6
C NMR (151 MHz, DMSO-d ): δ 172.70 (C-1), 142.53
(
C-8), 112.78 (C-9), 70.77 (C-7), 59.46 (C-1‘), 36.60 (C-6),
3.32 (C-2), 28.30 (C-4), 24.46 (C-3), 24.33 (C-5), 13.98
C-2‘). IR (ATR): 3443, 2987, 2934, 2860, 1733, 1463,
3
(
1
6
423, 1373, 1266, 1180, 1119, 1095, 1029, 991, 919, 737,
-1
20
f
퐷
69, 591 cm . R = 0.26 (PE:EtOAc 8:2). [훼] = 5.32
1
(
c=1.08 in CHCl ). Product 2d: H NMR (600 MHz,
3
3
CDCl
3
): δ 4.28-4.23 (m, 1H; 3-H), 4.19 (q, J1’,2’=7.2 Hz,
2
H; 1’-H), 3.64 – 3.58 (m, 2H; 4-H), 3.12 (d, 1H; OH),
3
2
.67 – 2.59 (m, 2H; 2-H), 1.28 (t, J2’,1’=7.2 Hz, 3H; 2’-H).
Battilocchio, S. V. Ley, ACS Cent. Sci. 2016, 2,
1
2
F. Paradisi, F. Molinari, Trends Biotechnol. 2018,
36, 73-88; i) R. Yuryev, S. Strompen, A. Liese,
Beilstein J. Org. Chem. 2011, 7, 1449-1467; j) D.
T. McQuade, P. H. Seeberger, J. Org. Chem 2013,
13
C NMR (151 MHz, CDCl
3
): δ 172.15 (C-1), 68.32 (C-3),
31-138; g) A. M. Foley, A. R. Maguire, 2019,
019, 3713-3734; h) L. Tamborini, P. Fernandes,
6
1.39 (C-1’), 48.48 (C-4), 38.78 (C-2), 14.50 (C-2’). IR
(
ATR): 3452, 2983, 1722, 1406, 1374, 1304, 1259, 1188,
1
4
151, 1086, 1058, 1030, 951, 894, 850, 803, 756, 709, 557,
-
1
25
73 cm . R
f
퐷
= 0.30 (PE:EtOAc 8:2). [훼] = -26.6 (c=1.25
in CHCl
3
).
7
8, 6384-6389.
Calculation of space-time yield (STY)
[
4]
a) J. Britton, S. Majumdar, G. A. Weiss, Chem.
Soc. Rev. 2018, 47, 5891-5918; b) R. Porta, M.
Benaglia, A. Puglisi, Org. Process Res. Dev.
The STY was calculated according to the following
equation:
2
016, 20, 2-25.
M. L. Contente, F. Molinari, Nat. Catal. 2019, 2,
51-952.
Yield [g]
STY =
[5]
[6]
9
column volume [L] ∗ time [h]
a) C. J. Hartley, C. C. Williams, J. A. Scoble, Q. I.
Churches, A. North, N. G. French, T. Nebl, G.
Coia, A. C. Warden, G. Simpson, A. R. Frazer, C.
N. Jensen, N. J. Turner, C. Scott, Nat. Catal.
-
4
column volume= 3.53*10 L
2
019, 2, 1006-1015; b) S. Velasco-Lozano, A. I.
Acknowledgements
Benítez-Mateos, F. López-Gallego, Angew. Chem.
Int. Ed. 2017, 56, 771-775; c) M. Heidlindemann,
G. Rulli, A. Berkessel, W. Hummel, H. Gröger,
ACS Catal. 2014, 4, 1099-1103; d) A. Fassouane,
J.-M. Laval, J. Moiroux, C. Bourdillon,
Biotechnol. Bioeng. 1990, 35, 935-939; e) S.
Kochius, J. B. Park, C. Ley, P. Könst, F.
Hollmann, J. Schrader, D. Holtmann, J. Mol.
Catal. B: Enzym. 2014, 103, 94-99; f) R.
The authors thank the Heinrich-Heine University Düsseldorf and
the Forschungszentrum Jülich GmbH for their support. The
European Regional Development Fund (ERDF) funded this
project. Moreover, we are grateful to Johannes Döbber for
providing the plasmid, Beatrix Paschold for the preparation of
preparative amounts of the HaloTag-LbADH, and to Birgit
Henßen as well as Patrick Ullrich for establishing the analytical
methods (GC/HPLC).
Ruinatscha, K. Buehler, A. Schmid, J. Mol. Catal.
B: Enzym. 2014, 103, 100-105; g) S. K. Yoon, E.
R. Choban, C. Kane, T. Tzedakis, P. J. A. Kenis,
J. Am. Chem. Soc. 2005, 127, 10466-10467.
a) J. Döbber, M. Pohl, S. V. Ley, B. Musio, React.
Chem. Eng. 2018, 3, 8-12; b) J. Döbber, T.
Gerlach, H. Offermann, D. Rother, M. Pohl,
Green Chem. 2018, 20, 544-552; c) T. Peschke, P.
Bitterwolf, K. S. Rabe, C. M. Niemeyer, Chem.
References
[
1]
a) D. J. Pollard, J. M. Woodley, Trends
Biotechnol. 2007, 25, 66-73; b) M. P. Thompson,
I. Peñafiel, S. C. Cosgrove, N. J. Turner, Org.
Process Res. Dev. 2019, 23, 9-18; c) J. C. Pastre,
D. L. Browne, S. V. Ley, Chem. Soc. Rev. 2013,
[
7]
4
2, 8849-8869; d) F. B. Mortzfeld, J. Pietruszka, I.
R. Baxendale, Eur. J. Org. Chem. 2019, 5424-
7
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.202000058
Eng. Technol. 2019, 42, 2009-2017; d) M. L.
Contente, F. Dall'Oglio, L. Tamborini, F.
Molinari, F. Paradisi, 2017, 9, 3843-3848; e) A.
Abdul Halim, N. Szita, F. Baganz, J. Biotechnol.
2
013, 168, 567-575; f) T. Peschke, P. Bitterwolf,
S. Gallus, Y. Hu, C. Oelschlaeger, N.
Willenbacher, K. S. Rabe, C. M. Niemeyer,
Angew. Chem. Int. Ed. 2018, 57, 17028-17032; g)
T. Peschke, P. Bitterwolf, S. Hansen, J. Gasmi, K.
S. Rabe, C. M. Niemeyer, Catalysts 2019, 9, 164;
h) P. Bitterwolf, S. Gallus, T. Peschke, E.
Mittmann, C. Oelschlaeger, N. Willenbacher, K.
S. Rabe, C. M. Niemeyer, Chem. Sci. 2019, 10,
9
2
752-9757; i) N. Adebar, H. Gröger, Bioeng.
019, 6, 99.
[
8]
a) L. H. Andrade, W. Kroutil, T. F. Jamison, Org.
Lett. 2014, 16, 6092-6095; b) H. Li, J. Moncecchi,
M. D. Truppo, Org. Process Res. Dev. 2015, 19,
6
95-700; c) W. Hummel, H. Schütte, M.-R. J. A.
M. Kula, Biotechnology, 1988, 28, 433-439; d) R.
Wichmann, C. Wandrey, A. F. Bückmann, M.-R.
Kula, 1981, 23, 2789-2802; e) A. I. Benítez-
Mateos, M. L. Contente, S. Velasco-Lozano, F.
Paradisi, F. López-Gallego, ACS Sustainable
Chem. Eng. 2018, 6, 13151-13159.
[
[
9]
A. Šalić, B. Zelić, RSC Adv. 2014, 4, 41714-
4
1721.
a) M. L. Contente, F. Paradisi, Nat. Catal. 2018, 1,
52-459; b) M. Romero-Fernández, F. Paradisi,
10]
4
Curr. Opin. Chem. Biol. 2020, 55, 1-8.
[
11]
a) J. Robertson, K. Clarke, R. L. Veech,
US9034613B2, 2015; b) T. Peschke, M. Skoupi,
T. Burgahn, S. Gallus, I. Ahmed, K. S. Rabe, C.
M. Niemeyer, ACS Catal. 2017, 7, 7866-7872; c)
E. Mittmann, Y. Hu, T. Peschke, K. S. Rabe, C.
M. Niemeyer, S. Bräse, ChemCatChem 2019, 11,
5
519-5523; d) S. Leuchs, L. Greiner, Chem.
Biochem. Eng. Q. 2011, 25, 267-281; e) F.
Hildebrand, S. Lütz, Tetrahedron: Asymmetry
2
006, 17, 3219-3225.
[
[
12]
13]
N. H. Schlieben, K. Niefind, J. Müller, B. Riebel,
W. Hummel, D. Schomburg, J. Mol. Biol. 2005,
3
49, 801-813.
a) T. Classen, M. Korpak, M. Schölzel, J.
Pietruszka, ACS Catal. 2014, 4, 1321-1331; b) C.
Kumru, T. Classen, J. Pietruszka, ChemCatChem
2
018, 10, 4917-4926; c) B. Seisser, I. Lavandera,
K. Faber, J. H. L. Spelberg, W. Kroutil, 2007,
49, 1399-1404; d) K. Goldberg, K. Schroer, S.
Lütz, A. Liese, Appl. Microbiol. Biotechnol. 2007,
6, 237; e) T. Fischer, J. Pietruszka, in Natural
3
7
Products Via Enzymatic Reactions, Vol. 297 (Ed.:
J. Piel), Springer, Berlin, Heidelberg, 2010, pp. 1-
4
3.
[
14]
a) M. Mantel, M. Guder, J. Pietruszka,
Tetrahedron 2018, 74, 5442-5450; b) S. Huo, Org.
Lett. 2003, 5, 423-425.
8
This article is protected by copyright. All rights reserved.

Advanced Synthesis & Catalysis
10.1002/adsc.202000058
FULL PAPER
Efficient Nicotinamide Adenine Dinucleotide
Phosphate [NADP(H)] Recycling in Closed-
Loop Continuous Flow Biocatalysis
Flow & Recycle! Phase separation is the key to
cross the line between continuous flow biocatalysis
and cofactor regeneration. An efficient system is
shown, which addresses both and enables efficient
closed-loop regeneration of the expensive cofactor
NADP(H).
Adv. Synth. Catal. Year, Volume, Page – Page
Benedikt Baumer, Thomas Classen, Martina Pohl,
and Jörg Pietruszka*
9
This article is protected by copyright. All rights reserved.