A Process Concept for High-Purity Production of Amines by Transaminase-Catalyzed Asymmetric Synthesis: Combining Enzyme Cascade and Membrane-Assisted ISPR

Article abstract of DOI:10.1021/acs.oprd.5b00055

For the amine transaminase (ATA)-catalyzed synthesis of chiral amines, the choice of donor substrate is of high importance for reaction and process design. Alanine was investigated as an amine donor for the reductive amination of a poorly water-soluble ketone (4-phenyl-2-butanone) in a combined in situ product removal (ISPR) approach using liquid-membrane extraction together with an enzyme cascade. This ISPR strategy facilitates very high (>98%) product purity with an integrated enrichment step and eliminates product as well as coproduct inhibition. In the presented proof-of-concept alanine shows the following advantages over the other frequently employed amine donor isopropyl amine: (i) nonextractability of alanine affords high product purity without any additional downstream step and no losses via coextraction, (ii) higher maximum reaction rates, and (iii) broader acceptance among ATAs.

Full text of DOI:10.1021/acs.oprd.5b00055

Concept Article
pubs.acs.org/OPRD
A Process Concept for High-Purity Production of Amines by
Transaminase-Catalyzed Asymmetric Synthesis: Combining Enzyme
Cascade and Membrane-Assisted ISPR
Tim Borner,*,† Gustav Rehn,‡ Carl Grey,† and Patrick Adlercreutz†
̈
†Department of Biotechnology, Lund University, P.O. Box, 221 00 Lund, Sweden
‡Department of Chemical and Biochemical Engineering, Technical University of Denmark, 2800 Kgs. Lyngby, Denmark
S
* Supporting Information
ABSTRACT: For the amine transaminase (ATA)-catalyzed synthesis of chiral amines, the choice of donor substrate is of high
importance for reaction and process design. Alanine was investigated as an amine donor for the reductive amination of a poorly
water-soluble ketone (4-phenyl-2-butanone) in a combined in situ product removal (ISPR) approach using liquid-membrane
extraction together with an enzyme cascade. This ISPR strategy facilitates very high (>98%) product purity with an integrated
enrichment step and eliminates product as well as coproduct inhibition. In the presented proof-of-concept alanine shows the
following advantages over the other frequently employed amine donor isopropyl amine: (i) nonextractability of alanine affords
high product purity without any additional downstream step and no losses via coextraction, (ii) higher maximum reaction rates,
and (iii) broader acceptance among ATAs.
amine donor in excess thereby “pushing” the reaction to high
conversion.19 In situ coproduct removal (IScPR) renders an
attractive concept and can be achieved by evaporation of
acetone (ACE) in IPA reactions20,21 or by enzymatic
transformation of pyruvate (PYR) in ALA reactions. Various
combinations of auxiliary enzymes targeting PYR have been
proven to promote high synthetic yields and are summarized in
recent reviews.22,23 A well-established enzyme cascade reaction
for PYR removal represents the combined use of ATA together
with a lactate dehydrogenase (LDH) and a glucose
dehydrogenase (GDH) for PYR removal and cofactor
generation, respectively.12,24 Alternatively, chemo-enzymatic
synthesis steps may be an option to drive the reaction forward,
but they remain case-specific and thus very limited to certain
types of amine donors.25 Moreover, IScPR methods not only
affect the Keq but can also prevent coproduct inhibition.
Contrary to IScPR, in situ product removal (ISPR) strategies
represent a physical separation approach.16,26 Practically, the
method should realize amine product capturing or isolation.
Ionic exchange resins have been employed in a continuous
process to bind the product amine in situ, but for amine donor/
product selectivity and capacity reasons, they were found to
lack (so far) industrial practicality.27 Similarly, a silica gel-based
“catch-and-release” method (in 0.5 mL scale) was recently
published exploiting the principle of flash-chromatography to
simply separate amines form ketones.28 Hence both amines
(donor and product) are coreleased when flushing the silica
cartridge with the eluent.
1. INTRODUCTION
Optically pure amines stand in high demand for the synthesis of
biologically active compounds, such as agrochemicals and
drugs.1,2,2b The biocatalytic synthesis of chiral amines with
amine transaminases (ATA) affords high enantio- and
regioselectivity and offers a sustainable and more environ-
mentally friendly alternative to the traditional transition metal
catalysis.3−7 Amine transaminases belong to the pyridoxal 5′-
phosphate (PLP) dependent enzymes and catalyze the amine
group transfer from a donor molecule (e.g., amine or amino
acid) to an amine acceptor substrate (e.g., ketone or keto
acid).8,9 Although other emerging enzymatic routes can yield
chiral amines, for example, employing NADH-dependent amine
dehydrogenases10 or imine reductases,11 this work will focus on
ATA-catalyzed amine synthesis by presenting a conceptually
unique process concept affording pure amines.
Overall, the production of chiral amines using ATAs as
biocatalyst faces several challenges which need to be overcome
in order to yield an attractive industrial process. The ATA-
catalyzed reductive amination reaction is essentially reversible
and prone to both substrate and product inhibition.8 In
addition, although ATAs are enzymes with a rather broad
substrate scope (giving access to a variety of chiral amines),12,13
the substrate specificity may vary depending on the source
organism.14 The most commonly used amine donating
substrates are alanine (ALA) and isopropyl amine (IPA). In
both cases, an unfavorable reaction equilibrium (Keq) hinders
high synthetic yields, which becomes even more pronounced
for pharmaceutically interesting amine products.15 Conse-
quently, strategies have to be developed to displace the Keq
and avoid inhibition affording high yields and high reaction
rates.
Generally, ISPR may have the benefit of an integrated
downstream step compared to IScPR.29 The principle of
selective removal/recovery of the amine product renders
There exist several strategies to counteract the unfavorable
Received: February 23, 2015
K
eq in ATA reactions.16−18 The easiest method is to supply the
© XXXX American Chemical Society
A
DOI: 10.1021/acs.oprd.5b00055
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
Scheme 1. Principle of in Situ Product Removal (ISPR) for Asymmetric Synthesis of Chiral Amines Using a Supported-Liquid
a
Membrane (SLM)
a
Isopropyl amine (IPA) or L-alanine (ALA) can serve as amine donor in the ATA-catalyzed amination of ketone substrates. Amine donor and
coproduct extraction into the stripping phase occurs only when using donor A (IPA, framed blue). Donor B (ALA) and coproduct B (PYR) allow
high product purity (framed red). In both cases (red and blue) ISPR influences the reaction equilibrium towards chiral amine synthesis. For more
explanations see text.
thereby perhaps an even more promising strategy in
asymmetric amine synthesis.30 Rehn et al. presented the
production of α-methylbenzylamine from acetophenone
together with IPA as an amine donor using supported-liquid
membrane (SLM) extraction as ISPR method,31 which affords
high product concentrations (about 100 g/L).32 The principle
of SLM extraction is schematically illustrated in Scheme 1. The
different pH of the reactor and stripping phase as well as the
hydrophobic membrane solvent represent the main driving
forces for selective extraction. Only the deprotonated amine
species extracts into the solvent and is then trapped in the
stripping phase (due to protonation). This resembles a dead-
end extraction and theoretically all amines can so be extracted
into the stripping phase, as long as the stripping phase pH is
kept sufficiently below the pKa of the amine product. Due to
the usually high pKa values of amines, the reactor phase should
feature rather alkaline conditions to establish a large fraction as
possible of extractable (deprotonated) amine species. ATAs are
advantageous for this approach, because they remain catalyti-
cally active and stable also at higher pH.13 If IPA serves as
amine donor, partial coextraction is unavoidable (Scheme 1,
blue frames).31 However, if the amine donor features a
permanently charged character at high pH such as the
zwitter-ionic ALA (Scheme 1, red frames), coextraction
would be avoided, and ALA is thus an ideal candidate in this
respect. Furthermore, as depicted in Scheme 1, the coproduct
(PYR) will not cross the membrane either, and hence the ALA
reaction should facilitate a much higher amine product purity
compared to IPA.
pharmaceutical precursors and challenging for large-scale
applications of ATA-catalyzed reactions.
2. MATERIALS AND METHODS
Materials. All chemicals such as 4-phenyl-2-butanone
(benzylacetone, BA), (S)-(+)-1-methyl-3-phenylpropylamine,
isopropyl amine, acetone, L-alanine, sodium pyruvate, NADH
(β-nicotinamide adenine dinucleotide, reduced disodium salt
hydrate), pyridoxal 5′-phosphate, and all used solvents were
purchased from Sigma-Aldrich Corporation. The amine
transaminase (ATA-50, from metagenomic library) crude
enzyme powder and freeze-dried cells were provided by c-
LEcta GmbH, Leipzig, Germany. Lactate dehydrogenase
(LDH) from hog muscle (Boehringer Mannheim GmbH,
Germany) and from bovine heart type III (Sigma-Aldrich
Corporation) was kindly provided by Prof. Per-Olof Larsson,
Pure & Applied Biochemistry, Lund University, Sweden.
Transamination Reactions. In small scale, 4.5 mL glass
vials (with Teflon septum sealed screw cap) were used
containing 2 mL of reaction solution placed in a thermoshaker
(HLC Biotech, Bovenden, Germany), shaken at 600 rpm, 30
°C. A typical substrate solution contained 450 mM of amine
donor, either IPA or ALA, and 10 mM amine acceptor, BA. A
20 mM sodium phosphate buffer pH 8 including 0.1 mM PLP
was used for kinetic experiments. The pH was adjusted with
HCl or NaOH after all substrates and cofactor had dissolved.
For initial rate experiments (specific activity determination, U/
mg) 0.05−0.5 mg/mL of ATA-50 crude enzyme or cells was
added to the reaction solution. The stopped-assay method
(batch reaction) was used taking aliquots at different time
intervals. The product concentration (of the quenched reaction
in mobile phase of final pH 11) was analyzed by HPLC. Initial
rates were determined by linear regression of product
concentrations over time below 10% conversion (considering
reaction equilibrium position, see Results). To obtain the half-
maximal inhibitory concentration (IC50%), the inhibition data
were fitted to the Michaelis−Menten equation for competitive
inhibition. For conversion profile experiments 2 mg/mL ATA-
50 crude powder or 4 mg/mL cells were added.
Due to the potential advantages of amine selectivity and
higher acceptance of the natural substrate among ATAs,33 the
aim of this study was to investigate the feasibility of using ALA
as an amine donor in combination with the SLM strategy for
ISPR. Such an integrated process design would have a
significant advantage to the existing methods, because in
principle, it continuously yields a pure and enriched amine
product without any additional downstream step. A model
reaction was chosen, where the ketone (4-phenyl-2-butanone)
features a poor water-solubility, which is often typical for
B
DOI: 10.1021/acs.oprd.5b00055
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
Cascade reactions were conducted in 50 mM borax-HCl
buffer pH 9.0 and 9.5, 0.2 mM PLP, 5 mM NADH, 0.5 M ALA,
10 mM BA, and 0.5 mg/mL (LDH) from hog muscle and
bovine heart type III together with 2 mg/mL ATA-50 at 30 °C
and shaken at 600 rpm.
Product (MPPA, ACP) formation was analyzed by HPLC
(Dionex Ultimate 3000) provided with a reversed phase (C18)
Gemini NX 3u 110 Å column (100 × 2 mm) equipped with a
guard column purchased from Phenomenex (Denmark). The
analytes were eluted in isocratic mode within 2.45 min (MPPA)
and 1.9 min (ACP) using a mobile phase composition of 65%
aqueous phase pH 11 (10 mM NaOH) and 35% acetonitrile.
s-ISPR System and Reaction. The setup of the
simultaneous in situ product and coproduct removal (s-ISPR)
system is given in Figure 5. Two Liqui-Cel MiniModule hollow
fiber (polypropylene) membrane contactors with a surface area
of 100 cm2 was purchased from Membrana (Charlotte, NJ,
USA). The stripping phase contained 100 mM citrate-HCL
buffer (pH < 3). The reactor phase was filled with 50 mL of
reaction solution (50 mM borax-HCl buffer, pH 9.5, 0.5 M
ALA, 10 mM BA, 0.2 mM PLP) and LDH (5 mg) together
with NADH (5 mM) was added initially and two more times
during the process at 40 and 70 h. Immobilized ATA-50 cells
were used in the SLM experiments: 250 mg ATA-50 cells were
entrapped in chitosan and packed into a column together with
Celite (0.2−0.5 mm, 30−80 mesh from BHD Laboratory
Supplies, Poole, England) as a scaffold stabilizer (reducing
compressibility). For detailed immobilization procedure, see
Rehn et al.34 Both phases were circulated with peristaltic pumps
(3 mL/min, Alitea, Stockholm, Sweden) and passing the
supported-liquid membrane contactor.
The isolated amine product was obtained through solvent
extraction. For the single step isolation the pH of the stripping
phase (10 mL) was adjusted to pH 12 (conc. NaOH), and
MPPA plus BA was extracted using 3 × 10 mL ethyl acetate,
pooled, and concentrated (under vacuum). For increased
isolated purity BA was removed first from the stripping phase
(pH < 3) by ethyl acetate (3 × 10 mL) extraction. Then MPPA
was extracted as described above. Products were analyzed by a
Varian gas chromatography (GC 430-GC-FID, Agilent
Technologies Inc., Santa Clara, CA.) equipped with a flame
ionization detector and a SGE analytical capillary column
(BPX35, 15 m, i.d. 0.25 mm).
Figure 1. Kinetic comparison between two amine donors in the amine
transaminase (ATA-50)-catalyzed amination of benzylacetone (BA).
Although different in maximum reaction rates alanine (ALA) and
isopropyl amine (IPA) feature a similar, apparent Km value of 120
8
mM and 128 14 mM. The reaction progress curves (inset) show
that about 90% and 16% of BA was converted into MPPA (1-methyl-
3-phenylpropylamine) with IPA and ALA, respectively. Ketone
concentrations were set to 10 mM at varying amine donor
concentrations using 0.2 mg/mL ATA-50 crude powder.
values commonly reported for donor excess (50−100).36 The
specific activity of ATA-50 at 0.5 M IPA and 10 mM BA was
determined to about 0.04 U/mg for the crude enzyme powder
and 0.02 U/mg freeze-dried E. coli cells containing the same
enzyme overexpressed. Compared to a previous study, which
reported 0.003 U/mg activity for cells (with a different ATA) in
a reaction using 6 mM BA and 1 M IPA,8 the here obtained
activity was almost 10-fold higher.
The time courses of the reactions, depicted in the inset of
Figure 1, show that for a simple batch reaction the highest
conversion of the amine acceptor (BA) into the amine form
(MPPA) was about 90% and 16% with IPA and ALA,
respectively. Assuming that equilibrium was reached, the
reaction equilibrium constants (Keq) were calculated, and the
IPA reaction featured the more favorable Keq of 0.17 compared
to only 0.0007 for ALA and both are consistent with literature
values.15
When the BA concentration was varied at constant amine
donor concentration (IPA or ALA), a relatively low apparent
Km of about 2.5 mM was determined (Figure 2). Fitting the
data in Figure 2 to the Michaelis−Menten equation, the
predicted maximum reaction rate (V) for 0.45 M donor was
estimated to about 0.05 and 0.07 U/mg for IPA and ALA,
respectively. Since higher BA concentrations above solubility
(ca. 10.8 mM) are required to obtain V, the practical achievable
rate is thus about 80% for both donors. The low Km of the
amine acceptor, however, may circumvent the necessity to
increase the solubility of BA in the reaction phase. Normally,
water miscible or immiscible solvents are employed to enhance
substrate concentrations in ATA reactions.37,33,38,39 In fact,
cosolvent addition studies revealed poor ATA-50 stability (see
SI, Figure S1). Among the tested water-miscible solvents ATA-
50 exhibited the highest tolerance toward dimethyl sulfoxide
(DSMO). The kinetic analysis given in Figure 3, however,
shows that increased ketone solubility in the presence of 25%
(v/v) DMSO had no beneficial effect on the rates. The reaction
rates decreased drastically upon cosolvent addition, and the
3. RESULTS AND DISCUSSION
3.1. Kinetics of Asymmetric MPPA Synthesis. In order
to carry out process development in a rational way a basic
kinetic analysis was performed, comparing amine donors ALA
and IPA. In Figure 1, the kinetics of the ATA-catalyzed
reductive amination reaction of benzylacetone (BA) is shown
using IPA and ALA at different concentrations. In terms of
initial reaction rate (U/mg) ALA resulted in approximately 30%
higher activity than IPA. Although substrate inhibition in ATA-
catalyzed reactions is known to occur,8,35 no obvious substrate
inhibition of either donor was detected within the tested
concentration range (up to 0.8 M). The apparent Michaelis
constant, Km, for IPA and ALA were very similar, about 128 and
120 mM, respectively. Thus, a donor concentration of about 0.5
M may be sufficient for efficient synthesis allowing about 80%
of the maximum reaction rate (V) of 0.05 and 0.07 U/mg for
IPA and ALA, respectively (from fitted data in Figure 1).
Donor concentrations at 0.5 M and 10 mM BA corresponds to
a donor to acceptor ratio of 50, which is at the lower end of the
C
DOI: 10.1021/acs.oprd.5b00055
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
Figure 4. Product inhibition profiles when using alanine (ALA) and
benzylacetone (BA) as a amine donor and acceptor, respectively. A
similar half-maximal inhibitory concentration (IC50%) was determined
for 1-methyl-3-phenylpropylamine (MPPA) and pyruvate (PYR) of
about 2.5 mM and 3.0 mM, respectively. Varying concentrations of
product and coproduct were added to constant amine donor (450
mM) and ketone (10 mM) concentration. The reaction pH was
adjusted to 8 before 0.2 mg/mL ATA-50 crude powder was added.
Figure 2. Influence of ketone substrate (BA) concentrations for the
two different donors: isopropyl amine (IPA) and alanine (ALA). At
the solubility limit of benzylacetone (BA, 10.8 mM) ATA-50 exhibited
80% of its theoretical maximum activity (1-methyl-3-phenylpropyl
amine (MPPA) formation rate). For both donors a similar, apparent
Km value of about 2.5 mM was determined. Amine donor
concentrations were set to 450 mM at varying ketone concentrations
using 0.2 mg/mL ATA-50 crude powder.
inhibition (IC50% of ∼4 mM) was also determined for ATA-50
in whole cells (see SI Figure S2). The inhibition kinetics show
that not only MPPA but also PYR removal is required to avoid
a rapid decrease in reaction rate.
To achieve efficient removal of dual product inhibition, we
attempted a new concept of the simultaneous in situ product
and coproduct removal (s-ISPR) which is a combination of
SLM-extraction of the product (MPPA) and an auxiliary
enzyme for coproduct (PYR) removal.
3.2. Simultaneous in Situ Product and Coproduct
Removal (s-ISPR). The process design of the s-ISPR system is
schematically depicted in Figure 5 (see also SI Figure S7). The
Figure 3. Effect of cosolvent addition on the ATA-50 kinetics.
Increased ketone solubility was achieved upon dimethyl sulfoxide
(DMSO, 25%, v/v) addition but associated with a drastic activity loss
in MPPA (1-methyl-3-phenylpropyl amine) formation compared to no
cosolvent (aqueous). Reaction conditions: 450 mM amine donor
(isopropyl amine, IPA) in 20 mM sodium phosphate buffer, 0.1 mM
PLP, pH 8 at 30 °C.
apparent Km value (ca. 7.4 mM) for BA had more than doubled
compared to no cosolvent (Figure 3). The use of organic
solvents is not only problematic for the enzyme stability and
kinetics but also complicates the downstream process as well as
affecting the “greenness” of the process in total.4 Hence,
maintaining saturation of BA by feeding the ketone
continuously the SLM methodology would allow for high
conversions and high product concentrations despite low
ketone solubility.
As expected, both product and coproduct inhibition were a
serious problem. It was found that 2 mM PYR inhibited the
reaction rate by about 50% (IC50%), and it was about 3 mM for
the amine product (MPPA) (Figure 4). A similarly strong
Figure 5. Simultaneous in situ product and coproduct removal (s-
ISPR) system for the production of 1-methyl-3-phenylpropylamine
(MPPA) from L-alanine (ALA) and benzylacetone (BA). The whole
cell biocatalyst containing the amine transaminase (ATA-50) was
immobilized in a packed-bed reactor (A). The reactor stream (reactor
phase, green) was filtered (B) before entering the SLM extraction unit
(C). MPPA was selectively extracted and enriched in the stripping
phase (red). The coproduct pyruvate (PYR) was reduced in the
reactor phase to lactate (LAT) by a lactate dehydrogenase (LDH)
consuming the cofactor (NADH).
D
DOI: 10.1021/acs.oprd.5b00055
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
Figure 6. Amine production and extraction performance of the s-ISPR system. The composition of the (A) reactor and (B) stripping phase at
different reaction times are given (pie charts). The amine substrate (alanine, ALA) was present in 50-fold excess (98%) compared to the ketone
benzylacetone (BA). The product accumulation and extraction profile of MPPA in the (A) reactor and (B) stripping phase is given in mmol. 5 mg of
LDH was added to the reactor phase initially as well as at 40 and 70 h.
whole cell biocatalyst (ATA-50 cells) was immobilized using
chitosan flocculation.34 This formulation, composed of 90%
cells (10 g cells/g chitosan), retained nearly all of its activity
compared to free cells (see SI Figure S3). Continuous ketone
supply (as second phase) was initially tested but was rejected
because the freely dissolved LDH enzyme (and also ATA-50 to
some extent) appeared to be very sensitive to interfacial
inactivation retaining hardly any activity (data not shown). In
addition the resulting protein precipitates/aggregates caused
clogging of the system. The pH induced inactivation (showing
much less precipitations) was not causing any clogging, and a
filter was sufficient to prevent particle entering the SLM
module. In order to evaluate the feasibility of the s-ISPR system
the reactor phase was initially saturated with BA, and no further
BA was added.
Both enzymes ATA and cascade enzyme must remain
catalytically active at alkaline conditions, because the efficiency
of the SLM-extraction of the amine product MPPA is
dependent on the difference between the pH of the reactor
phase (9.5) and the pKa of MPPA (10.5) (Scheme 1). ATA-50
remained catalytically active at pH 9.5 (data not shown). For
the cascade reaction, however, two lactate dehydrogenases
(LDHs) were chosen and tested under conditions related to the
s-ISPR system. LDH from hog muscle exhibited activity at pH
9.5 and enabled roughly 30% conversion in a batch reaction
together with crude ATA-50 enzyme (see SI Figure S4).
Nonetheless, much higher conversion improvement was
reported using LDH as cascade enzyme at around neutral
pH.40 But as the high pH caused protein precipitation, full
inactivation was detected after a certain reaction time (data not
shown). The little amount of precipitated LDH compared to
the reactor volume could simply be retained through a filter. To
compensate for the poor pH stability of the LDH, a fed-batch
mode for the auxiliary enzyme supply was used in further
experiments. Also, because sufficient for the purpose of
demonstration, simple PYR removal was envisaged without
any other enzyme recycling the cofactor NADH, which has
been reported elsewhere.23
mL), MPPA was concentrated 5-fold during the process. For
comparison, we refer to amounts (mmol) or content (%, ni/
n
total) of substrates and products rather than concentrations. As
can be seen in Figure 6B, during a reaction time of 24 h, the
amine product (MPPA) amount increased linearly in the
stripping phase, suggesting constant MPPA production and
extraction rates. About 50% of the final MPPA (0.14 mmol at
114 h) was produced within the first 24 h of the reaction
(which was about 20% of the total operation time).
Accumulation of MPPA on the reactor side was detected
only at a very early stage of the reaction (Figure 6 A), when the
reaction rate was high due to sufficient BA amount and low
inhibition. After 24 h the extraction rate of MPPA decreased
(Figure 6B). The MPPA amount in the reactor phase was
orders of magnitude lower compared to the stripping phase and
exhibited a further decrease until process termination (Figure
6A). Considering reaction rates, a decrease in MPPA
production rate was probably caused by reduced ketone
substrate (BA) availability and increasing concentration of
PYR. The measured composition of the reactor phase given in
Figure 6 at 24 h revealed that the BA content (25%) was lower
than that of the coproduct (75%). In fact, due the
hydrophobicity of BA a significant amount partitioned into
the membrane solvent as well as into the plastic tubings. Thus,
the membrane solvent acts as a substrate reservoir. Here is
another strong benefit of the SLM strategy; since it utilizes a
very small solvent volume, less BA would be captured
compared to extraction alternatives including a separate organic
phase.41 In addition, back-extraction of BA was observed in the
later stage of the reaction. The measured BA amount in both
phases depleted while the MPPA amount almost doubled from
24 to 114 h (Figure 6B). When the experiment was terminated
(at 114 h), some BA remained in the system. Conclusively, to
achieve a high final purity, the feeding of BA should be stopped,
and the residing ketone in the stripping phase and membrane
phase then diffuses back into the reactor phase where it is
converted.
Accumulation of the coproduct PYR was another important
factor influencing the overall conversion rate. PYR was
converted into lactate (LAT) by LDH and NADH added at
the start of the reaction and twice later at 40 and 70 h. The
Using the s-ISPR setup shown in Figure 5, continuous
extraction and enrichment of MPPA was achieved. Due to the
volume difference in reactor (50 mL) and stripping phase (10
E
DOI: 10.1021/acs.oprd.5b00055
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
further additions of LDH clearly enhanced the production of
MPPA, shown as accumulation in the stripping phase (Figure
6B). Evident from Figure 6B, almost 20% of the total MPPA
amount was produced and extracted upon the second LDH
addition (40 h), which demonstrates the importance of PYR
removal. However, it also shows that sequential addition of
small LDH/NADH amounts could be an option. Once the
reaction rate has decreased to a certain threshold, it can quickly
be reestablished by LDH/NADH addition. It is likely that
substantial improvements in the process can be achieved using
auxiliary enzyme(s) with better stability than the LDH used
here. The combination of continuous ketone supply and
effective cascade reaction would maintain the initially (24 h)
observed MPPA production/extraction rate shown in Figure 6B
and thus yielding a much higher productivity.
Although far from an optimized process, the presented
concept achieved very high product purity without any
additional purification steps. The use of ALA in the s-ISPR
approach facilitated a direct purity of ≥98% MPPA in the
stripping phase (Figure 6B). This represents a major advantage
in contrast to other existing ISPR strategies, because no other
method allows such selective amine isolation of the desired
product from the other components. This also includes the
previously reported single ISPR method employing SLM
extraction at a lower reactor pH (9.0) together with IPA and
ACP (which has a 5-fold higher solubility than BA).31 Even
higher purity can be attained if the contaminating BA (here
about 2%) is allowed to back-extracted completely before the
amine product is harvested (SI, Figure S5). Thus, no further
purification step may be needed, and a high isolated yield can
be achieved.
selected to be relatively close to the pKa value of the amine
product. Note, the SLM extraction allows one-way extraction.
The deprotonated amine becomes reprotenated once it diffuses
from the solvent into the stripping phase (low pH), thus
preventing back-extraction. The stripping phase should have a
pH preferable three units below the pKa, but needs to be
maintained, for example, by high buffer concentrations or active
titration (e.g., for continuous process and in large scale).
Hence, the amine product determines the pH gradient from the
reactor to the stripping phase required for sufficient extraction.
Many modern drugs contain pyridines and other nitrogen
containing heterocycles,42 some of which might contain pH
sensitive moieties. Generally, aromatic amines feature a lower
basicity than aliphatic ones, for example, benzimidazole (pKa =
5.5) or 2-aminopyridine (pKa = 6.8) compared to sec-
butylamine (pKa = 10.6). Producing such class of amines
with the s-ISPR strategy would also be possible, because the pH
of the reactor and stripping phase can be selected according to
the product pKa to ensure sufficient extraction and mild enough
conditions for the sensitive amines. A difference in pKa can also
be found among primary amines containing aromatic
structures, such as MBA and MPPA. Due to the more closely
adjacent phenyl ring the pKa of MBA (9.6) is about one unit
lower than that of MPPA (10.6). Previously, Rehn et al. have
synthesized and extracted MBA at pH 9.0, which constituted a
good compromise between reaction and extraction perform-
ance.31 In this study, however, synthesizing MPPA a higher
reaction pH (9.5) was required to achieve sufficient extraction.
4. CONCLUSIONS
The conceptually novel approach, simultaneous in situ product
and coproduct removal (s-ISPR), employing ALA as amine
donor in combination with supported-liquid membrane (SLM)
extraction and a cascade reaction affords the amine product in
high purity (≥98%) without any additional purification step. In
contrast, for reactions using IPA as an amine donor
coextraction causes contamination of the product solution
and to a loss of donor substrate.31 Thus, despite its often more
favorable reaction equilibrium, IPA reactions require further
downstream operations to yield similar purity as achieved with
ALA. Another key point of the s-ISPR is that it allows the
simultaneous removal of product and coproduct, which cannot
be achieved by any other methods at present. Other advantages
of employing ALA as amine donor represents its wide
acceptance among ATAs, which is not the case for IPA,33
and a faster reaction velocity compared to the unnatural amine
donor (IPA). In principle, a single s-ISPR process setup could
be used for the production of different types of chiral amines
with high purity. Noteworthy, the (alkaline) reaction pH can be
adjusted according to the pKa of the amine product and,
similarly, the acid pH of the stripping phase. This, in turn,
permits the application of pH sensitive substrates and/or
products, such as nitrogen containing aromatic heterocycles.
Overall, our study represents a proof-of-concept for the s-
ISPR strategy using ATAs, but it requires certain optimization
for increased performance. Two major bottlenecks were
encountered: first, the pH stability of the auxiliary enzyme(s)
and, second, a more sufficient supply of the ketone substrate.
The development of a whole-cell system furnished with a
transaminase and the entire cofactor regeneration system (e.g.,
LDH/GDH) is currently under investigation. Immobilized cells
would then permit the supply of the poorly water-soluble
ketone via a second phase. This targets both limitations
It is worth mentioning that the stripping phase and so the
purity remained unaffected by operating the s-ISPR with an
excess of amine donor as long as it features a charged character.
ALA was confirmed as excellent amine donor, because it
remained on the reactor side (see SI, Figure S6). Since the
maximum soluble amine acceptor (BA) concentration was
about 10 mM (0.5 mmol, which is ∼2% of total substrate), the
aqueous reactor phase mainly contained ALA (∼98%, 0.5 M,
see Figure 6) throughout the entire course of reaction.
Normally, a large excess of amine donor imposes separation
and impurity problems using other processes, such as 2-phase
extraction or ionic exchange resins,41,27 which do not facilitate
selective in situ amine product recovery strategies. With the s-
ISPR concept, in fact, the donor excess can be minimized,
because both the kinetic (inhibition) and the thermodynamic
(equilibrium) barrier are controllable. Yet, the amine donor
needs to be available in sufficient amounts, supplied either
batch-wise or continuously, in order to maintain the reaction
velocity.
3.3. s-ISPR Product Scope. This work is focused on the
asymmetric MPPA synthesis in high purity from BA using the
simultaneous in situ product removal (s-ISPR) approach. ALA
represents the key to high purities since it is not extracted
compared to using IPA as an amine donor.31 In the following
we give a brief description of the general applicability of the s-
ISPR to a broad range of amine products and its extraction
principle depicted in Scheme 1. In principle, any amine which is
soluble in organic solvents could be extracted, purified, and
enriched. In the s-ISPR approach both the reactor pH and the
membrane solvent influence the extraction of the amines. Only
the deprotonated form extracts from the reactor phase into the
thin solvent film (Scheme 1). To achieve this, the reactor pH is
F
DOI: 10.1021/acs.oprd.5b00055
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Concept Article
(13) Schatzle, S.; Steffen-Munsberg, F.; Thontowi, A.; Hohne, M.;
Robins, K.; Bornscheuer, U. T. Adv. Synth. Catal. 2011, 353, 2439−
2445.
̈
̈
mentioned above, and a direct comparison between IPA and
ALA including further s-ISPR optimization will then be
reasonable.
(14) Sayer, C.; Isupov, M. N.; Westlake, A.; Littlechild, J. A. Acta
Crystallogr., D 2013, 69, 564−576.
ASSOCIATED CONTENT
* Supporting Information
■
(15) Tufvesson, P.; Jensen, J. S.; Kroutil, W.; Woodley, J. M.
Biotechnol. Bioeng. 2012, 109, 2159−2162.
S
Additional experimental information on solvent stability and
immobilization of ATA, screening of LDH in batch, as well as a
detailed setup of the s-ISPR system with illustrations and
pictures are provided. This material is available free of charge
via the Internet at http://pubs.acs.org/.The Supporting
Information is available free of charge on the ACS Publications
website at DOI: 10.1021/acs.oprd.5b00055.
(16) Koszelewski, D.; Tauber, K.; Faber, K.; Kroutil, W. Trends
Biotechnol. 2010, 28, 324−332.
(17) Kroutil, W.; Fischereder, E.-M.; Fuchs, C. S.; Lechner, H.;
Mutti, F. G.; Pressnitz, D.; Rajagopalan, A.; Sattler, J. H.; Simon, R. C.;
Siirola, E. Org. Process Res. Dev. 2013, 17, 751−759.
(18) Mathew, S.; Yun, H. ACS Catal. 2012, 2, 993−1001.
(19) Yun, H.; Cho, B.-K.; Kim, B.-G. Biotechnol. Bioeng. 2004, 87,
772−778.
(20) Park, E.-S.; Dong, J.-Y.; Shin, J.-S. Org. Biomol. Chem. 2013, 11,
6929.
AUTHOR INFORMATION
Corresponding Author
*E-mail: tim.borner@biotek.lu.se.
■
(21) Tufvesson, P.; Bach, C.; Woodley, J. M. Biotechnol. Bioeng. 2014,
111, 309−319.
(22) Ricca, E.; Brucher, B.; Schrittwieser, J. H. Adv. Synth. Catal.
2011, 353, 2239−2262.
Notes
The authors declare no competing financial interest.
(23) Simon, R. C.; Richter, N.; Busto, E.; Kroutil, W. ACS Catal.
2013, 4, 129−143.
ACKNOWLEDGMENTS
(24) Pressnitz, D.; Fuchs, C. S.; Sattler, J. H.; Knaus, T.; Macheroux,
P.; Mutti, F. G.; Kroutil, W. ACS Catal. 2013, 3, 555−559.
(25) Wang, B.; Land, H.; Berglund, P. Chem. Commun. 2013, 49,
161−163.
■
Financial support by the European Union FP7 Project
BIOINTENSE−Mastering Bioprocess integration and intensi-
fication across scales (Grant Agreement Number 312148) is
gratefully acknowledged.
(26) Tufvesson, P.; Lima-Ramos, J.; Jensen, J. S.; Al-Haque, N.; Neto,
W.; Woodley, J. M. Biotechnol. Bioeng. 2011, 108, 1479−1493.
(27) Truppo, M. D.; Rozzell, J. D.; Turner, N. J. Org. Process Res. Dev.
2010, 14, 234−237.
ABBREVIATIONS
■
(28) Andrade, L. H.; Kroutil, W.; Jamison, T. F. Org. Lett. 2014, 16,
6092−6095.
ACE, acetone; ADH, amine dehydrogenase; ALA, L-alanine;
ATA, amine transaminase (omega-transaminase); BA, benzyla-
cetone (4-phenyl-2-butanone); DMSO, dimethyl sulfoxide;
GDH, glucose dehydrogenase; IPA, isopropyl amine; ISPR, in
situ product removal; IScPR, in situ coproduct removal; s-ISPR,
simultaneous in situ product and coproduct removal; LDH,
lactate dehydrogenase; MPPA, 1-methyl-3-phenypropyllamine;
NADH, β-nicotinamide adenine dinucleotide, reduced diso-
dium salt hydrate; PLP, pyridoxal 5′-phosphate; PYR, sodium
pyruvate
(29) Shin, J.-S.; Kim, B.-G.; Liese, A.; Wandrey, C. Biotechnol. Bioeng.
2001, 73, 179−187.
(30) Yun, H.; Kim, B.-G. Biosci. Biotechnol. Biochem. 2008, 72, 3030−
3033.
(31) Rehn, G.; Adlercreutz, P.; Grey, C. J. Biotechnol. 2014, 179, 50−
55.
(32) Rehn, G. ω-Transaminase catalyzed synthesis of chiral amines.
Ph.D. Dissertation, Lund University, Sweden, 2013.
(33) Mutti, F. G.; Kroutil, W. Adv. Synth. Catal. 2012, 354, 3409−
3413.
(34) Rehn, G.; Grey, C.; Branneby, C.; Adlercreutz, P. J. Biotechnol.
REFERENCES
■
2013, 165, 138−144.
(35) Shin, J.-S.; Kim, B.-G. Biosci. Biotechnol. Biochem. 2001, 65,
1782−1788.
(36) Turner, N. J.; Truppo, M. D. Biocatalytic Routes to Nonracemic
Chiral Amines. In Chiral Amine Synthesis; Wiley-VCH Verlag GmbH &
Co. KGaA: Weinheim, 2010; pp 431−459.
(37) Truppo, M. D.; Strotman, H.; Hughes, G. ChemCatChem 2012,
4, 1071−1074.
(38) Fuchs, C. S.; Simon, R. C.; Riethorst, W.; Zepeck, F.; Kroutil,
W. Bioorg. Med. Chem. 2014, 22, 5558−5562.
(39) Simon, R. C.; Fuchs, C. S.; Lechner, H.; Zepeck, F.; Kroutil, W.
Eur. J. Org. Chem. 2013, 2013, 3397−3402.
(40) Mutti, F. G.; Fuchs, C. S.; Pressnitz, D.; Sattler, J. H.; Kroutil,
W. Adv. Synth. Catal. 2011, 353, 3227−3233.
(41) Shin, J.-S.; Kim, B.-G. Biotechnol. Bioeng. 1997, 55, 348−358.
(42) Baumann, M.; Baxendale, I. R. Beilstein J. Org. Chem. 2013, 9,
2265−2319.
(1) Turner, C. Biocatalysis in the Pharmaceutical and Biotechnology
Industries; CES Press LLC: Boca Raton, FL, 2007.
(2) (a) Patel, R. N. ACS Catal. 2011, 1, 1056−1074. (b) Ghislieri, D.;
Turner, N. Top Catal. 2014, 57, 284−300.
(3) Shin, J.-S.; Kim, B.-G. Biotechnol. Bioeng. 1999, 65, 206−211.
(4) Kohls, H.; Steffen-Munsberg, F.; Hohne, M. Curr. Opin. Chem.
̈
Biol. 2014, 19, 180−192.
(5) Savile, C. K.; Janey, J. M.; Mundorff, E. C.; Moore, J. C.; Tam, S.;
Jarvis, W. R.; Colbeck, J. C.; Krebber, A.; Fleitz, F. J.; Brands, J.;
Devine, P. N.; Huisman, G. W.; Hughes, G. J. Science 2010, 329, 305−
309.
(6) Nestl, B. M.; Hammer, S. C.; Nebel, B. A.; Hauer, B. Angew.
Chem., Int. Ed. 2014, 53, 3070−3095.
(7) Nugent, T. C.; El-Shazly, M. Adv. Synth. Catal. 2010, 352, 753−
819.
(8) Shin, J.-S.; Kim, B.-G. Biotechnol. Bioeng. 2002, 77, 832−837.
(9) Eliot, A.; Kirsch, J. Annu. Rev. Biochem. 2004, 73, 383−415.
́
(10) Abrahamson, M. J.; Vazquez-Figueroa, E.; Woodall, N. B.;
Moore, J. C.; Bommarius, A. S. Angew. Chem., Int. Ed. 2012, 51, 3969−
3972.
(11) Scheller, P. N.; Fademrecht, S.; Hofelzer, S.; Pleiss, J.; Leipold,
F.; Turner, N. J.; Nestl, B. M.; Hauer, B. ChemBioChem. 2014, 15,
2201−2204.
(12) Koszelewski, D.; Lavandera, I.; Clay, D.; Rozzell, D.; Kroutil, W.
Adv. Synth. Catal. 2008, 350, 2761−2766.
G
DOI: 10.1021/acs.oprd.5b00055
Org. Process Res. Dev. XXXX, XXX, XXX−XXX