Arylmalonate Decarboxylase-Catalyzed Asymmetric Synthesis of Both Enantiomers of Optically Pure Flurbiprofen

Article abstract of DOI:10.1002/cctc.201501205

The bacterial decarboxylase (AMDase) catalyzes the enantioselective decarboxylation of prochiral arylmalonates with high enantioselectivity. Although this reaction would provide a highly sustainable synthesis of active pharmaceutical compounds such as flurbiprofen or naproxen, competing spontaneous decarboxylation has so far prevented the catalytic application of AMDase. Here, we report on reaction engineering and an alternate protection group strategy for the synthesis of these compounds that successfully suppresses the side reaction and provides pure arylmalonic acids for subsequent enzymatic conversion. Protein engineering increased the activity of the synthesis of the (S)-and (R)-enantiomers of flurbiprofen. These results demonstrated the importance of synergistic effects in the optimization of this decarboxylase. The asymmetric synthesis of both enantiomers in high optical purity (>99 %) and yield (>90 %) can be easily integrated into existing industrial syntheses of flurbiprofen, thus providing a sustainable method for the production of this important pharmaceutical ingredient. Optically pure flurbiprofen: A novel deprotection strategy for the preparation of the starting material combined with decarboxylase (AMDase) variants optimized by enzyme engineering allowed the asymmetric synthesis of both enantiomers of the non-steroidal anti-inflammatory drug (NSAID) flurbiprofen in excellent yield and optical purity.

Full text of DOI:10.1002/cctc.201501205

DOI: 10.1002/cctc.201501205
Full Papers
Arylmalonate Decarboxylase-Catalyzed Asymmetric
Synthesis of Both Enantiomers of Optically Pure
Flurbiprofen
Sarah Katharina Gaßmeyer,[a] Jasmin Wetzig,[b] Carolin Mügge,[a] Miriam Assmann,[c]
Junichi Enoki,[a] Lutz Hilterhaus,[c] Ralf Zuhse,[b] Kenji Miyamoto,[d] Andreas Liese,[c] and
Robert Kourist*[a]
The bacterial decarboxylase (AMDase) catalyzes the enantiose-
lective decarboxylation of prochiral arylmalonates with high
enantioselectivity. Although this reaction would provide
a highly sustainable synthesis of active pharmaceutical com-
pounds such as flurbiprofen or naproxen, competing sponta-
neous decarboxylation has so far prevented the catalytic appli-
cation of AMDase. Here, we report on reaction engineering
and an alternate protection group strategy for the synthesis of
these compounds that successfully suppresses the side reac-
tion and provides pure arylmalonic acids for subsequent enzy-
matic conversion. Protein engineering increased the activity of
the synthesis of the (S)- and (R)-enantiomers of flurbiprofen.
These results demonstrated the importance of synergistic ef-
fects in the optimization of this decarboxylase. The asymmetric
synthesis of both enantiomers in high optical purity (>99%)
and yield (>90%) can be easily integrated into existing indus-
trial syntheses of flurbiprofen, thus providing a sustainable
method for the production of this important pharmaceutical
ingredient.
Introduction
Enzymatic asymmetric synthesis is considered a sustainable
strategy as it allows the full conversion of a prochiral substrate
into an enantiomerically pure product. The bacterial arylmalo-
nate decarboxylase (AMDase) from Bordetella bronchiseptica
allows the synthesis of optically pure carboxylic acids by enan-
tioselective decarboxylation of prochiral malonates. The
enzyme shows outstanding enantioselectivity and has been
successfully applied for the preparation of optically pure aryl,[1]
alkenyl,[2] aliphatic, and hydroxycarboxylic acids.[3] The high se-
lectivity of the enzyme and the mild reaction conditions offer
significant advantages in terms of sustainability. The arylpropi-
onates flurbiprofen (S)-2a and naproxen (S)-2b are widely
used as non-steroidal anti-inflammatory agents (NSAIDS) and
belong to the most common family of over-the-counter
drugs.[4] Moreover, (R)-2a is under investigation for the treat-
ment of Alzheimer’s disease[5] and has been shown to reduce
proliferation of gastric cancer cells.[6] The distinctive biological
activity of both enantiomers of 2a makes the synthesis of their
optically pure enantiomers a desirable goal.
So far, all known wildtype AMDases have strict selectivity for
the formation of the (R)-enantiomer.[7] As several arylpropio-
nates, such as the non-steroidal anti-inflammatory drugs ibu-
profen, naproxen and ketoprofen, exert their anti-inflammatory
activity only in the (S)-form, the successful inversion of the en-
zyme’s enantiopreference greatly expanded the synthetic po-
tential of the enzyme (Scheme 1).[8] Several rounds of struc-
ture-inspired saturation mutagenesis increased the activity of
[a] S. K. Gaßmeyer, C. Mügge, J. Enoki, Prof. Dr. R. Kourist
Junior Research Group for Microbial Biotechnology
Ruhr-University Bochum
44780 Bochum (Germany)
E-mail: Robert.Kourist@rub.de
[b] J. Wetzig, Dr. R. Zuhse
Chiracon GmbH
Im Biotechnologiepark, 14943 Luckenwalde (Germany)
[c] M. Assmann, L. Hilterhaus, Prof. Dr. A. Liese
Institute for Technical Biocatalysis
Hamburg University of Technology TUHH
Denickestr. 15, 21071 Hamburg (Germany)
[d] Prof. Dr. K. Miyamoto
Department for Biosciences and Bioinformatics
Keio University
3-14-1 Hiyoshi, Yokohama 223-8522 (Japan)
Supporting Information and ORCID(s) from the author(s) for this article
are available on the WWW under http://dx.doi.org/10.1002/
cctc.201501205.
Scheme 1. Mechanism of the AMDase-catalyzed asymmetric synthesis of op-
tically pure (R)- and (S)-flurbiprofen, (R)- and (S)-2a. The rate-determining
step is the cleavage of the pro-(R)-carboxylate.
ChemCatChem 2016, 8, 916 – 921
916
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
the (S)-selective variant considerably, the best variant with six
substituted amino acids converting the malonate precursors
naproxen and ibuprofen with 0.6 and 1.2-fold relative activity
compared with the (R)-selective wildtype.[9] Although an asym-
metric reaction employing AMDase would be simple and ele-
gant, its broad commercial application has been hampered so
far by the limited stability of several arylmalonic acids. The
saponification of the commonly used diethyl malonates bears
the risk of spontaneous decarboxylation during work-up, re-
sulting in racemic arylpropionic acids. These spontaneously
formed products “dilute” the enantiomeric excess of the other-
wise highly enantiopure product of the enzymatic reaction.
Their removal, however, entails laborious work-up and thus
makes an up-scaling very difficult. Avoiding this side reaction
is particularly challenging for substrates such as 2a, where
electron-withdrawing substituents at the aryl unit increase
their tendency to undergo spontaneous decarboxylation even
at neutral pH. For industrial synthesis of rac-2a, the methyl-
malonic acid is generated by cleavage of the respective diethyl
ester under acidic conditions, followed by decarboxylation of
the free malonic acid at high temperatures.[10] An enzymatic
decarboxylation under mild reaction conditions (neutral pH,
only slightly elevated temperatures, aqueous solvents) would
clearly present an environmentally friendly alternative and di-
rectly lead to the optically pure product in 100% maximal
yield.[11] A previous attempt to integrate the AMDase-catalyzed
decarboxylation step into the industrial synthesis route to-
wards 2a showed the feasibility of the approach. Unfortunate-
ly, the occurrence of spontaneous decarboxylation upon proto-
lytic ester cleavage limited the enantiomeric excess in the
AMDase-catalyzed synthesis of (R)-2a to a maximum of 92%
ee,[12] which is too low for pharmaceutical applications. As the
enzymatic mechanism is strictly stereospecific, suppression of
the side reaction can be expected to increase the optical
purity considerably. In this paper, we report an approach to cir-
cumvent the side reaction by an alternative protection group
strategy and optimization of the reaction conditions. Protein
engineering and immobilization strategies are used to increase
the total turnover number of the enzyme considerably and
thus should lead to appreciable yields of optically pure com-
pounds 2.
Scheme 2. Preparation of arylmethylmalonic acids 1 as the starting material
for asymmetric enzymatic synthesis by deprotection of benzylmalonic esters
4 as key intermediates for asymmetric synthesis. Reaction conditions:
[i] a) 1.5 equiv. benzyl alcohol, cat. toluenesulfonic acid, toluene, reflux,
b) OHÀ, RT; [ii] a) 1.2 equiv. diisopropylamine, 1.2 equiv. n-butyllithium,
1.5 equiv. benzyl chloroformate, THF, À208C to RT, b) H+; [iii] 0.5% Pd/C, H2,
MTBE, 08C.
The deprotection of 4a and 4b by hydrogenolysis proceed-
ed smoothly with a conversion of >99% (as judged by TLC
controls) and produced only minute amounts (<0.5%) of side-
product rac-2 (Figure S8 in the Supporting Information). After
one round of recrystallization, 58% of 1a was obtained, which
is a promising starting point for further optimization. Interest-
ingly, the analogous deprotection of 4b did not require further
purification and produced 1b with excellent yield (99%). The
new route using hydrogenolysis thus proved to be a practica-
ble and scalable substitute to provide the starting material for
asymmetric decarboxylation.
Once 1a and 1b were successfully isolated, they could be
used for enzymatic conversion. However, spontaneous decar-
boxylation of 1 competes with the stereoselective enzymatic
reaction once the substrate is dissolved in aqueous medium.
Suppression of this background reaction is essential to achieve
a pharmaceutically relevant optical purity. We qualitatively as-
sessed the rate of decarboxylation of 1a at pH 6 and 9 over
time and found that upon dissolving the powder in buffered
solutions, spontaneous decarboxylation sets in. By increasing
the pH from 6 to 9, we were able to slow down this side reac-
tion considerably (see the Supporting Information). The data
suggests that a biocatalytic reaction at the pH activity maxi-
mum of AMDase (pH 8.5)[7c] with a reaction time below 1 h
should yield compounds 2 in satisfactory enantiopurity, given
sufficient activity of the enzyme. To this aim, the conversion
rates of both (R)- and (S)-selective AMDase variants should be
optimized in the next step.
Results and Discussion
The preparation of a biocatalyst requires a considerable
amount of energy and is thus one of the key factors for sus-
tainability. Increasing the total turnover number is, in addition
to the aforementioned issues, therefore an important objective
for a catalytic process as it allows the use and preparation of
minimized amounts of biocatalyst. This can be achieved by the
complementary methods of enzyme and reaction engineering.
For AMDase, both (R)-selective and (S)-selective variants are
available.[1,2,9,14] Previous work showed that the effect of amino
acid substitutions is highly substrate-specific, and at the same
time synergistic effects of different mutations affect the activity
of the enzyme strongly.[9] The rate-determining step of enzy-
matic decarboxylation by AMDase is the destabilization of the
pro-(R) carboxylate, which depends on the stereoelectronic ef-
We reasoned that an alternative deprotection procedure might
avoid deviations of the pH value and thus suppress the non-
desired side reaction.[13] The benzyl protection group (Bn) was
chosen to enable hydrogenolysis instead of saponification to
isolate arylmethyl malonic acids 1. Dibenzyl arylmethyl malo-
nates 4a/b are thus key intermediates to this alternate syn-
thetic route.
To prove the principle of this approach, compounds
4a/b were generated from the respective profenes 2 via their
benzyl-protected monoesters 3 (Scheme 2). Although 4a is an
oil, compound 4b was obtained as crystals suitable for single
crystal X-ray analysis (Figure S1 in the Supporting Information)
and proves the presence of all the desired structural features.
ChemCatChem 2016, 8, 916 – 921
www.chemcatchem.org
917
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
fects of the substrate and its spatial fit into the enzyme active
site. Although naproxen malonate 1b has a much higher sta-
bility than flurbiprofen malonate 1a, wildtype AMDase decar-
boxylates 1b four times faster than 1a.[12] In contrast, (S)-selec-
tive AMDase variants convert 1a faster than 1b (Figure 1). The
variant AMDase G74C with racemizing activity shows a four-
fold preference for naproxen.[14b] Binding of 1a inside the enzy-
matic active center is expected to decrease the degrees of
freedom for rotation of the second phenyl ring, which might
result in an unfavorable entropy. This example shows that for
an efficient catalysis, a good fit of the substrate in a productive
binding mode inside the active center is a decisive parameter
for conversion. The enzymatic decarboxylation is initiated by
cleavage of the pro-(R) carboxylate group (Scheme 1). Semi-
empiric calculations confirmed that the resulting planar ene-
dionate intermediate is protonated by a cysteine residue,
which decides the stereoselective outcome of the reaction.[8b]
Transferring the catalytic cysteine 188 to the opposite side of
the active center switches the enzyme’s enantiopreference to-
wards the formation of the (S)-enantiomer.[8a,15] The active site
of AMDase consists of a so-called dioxyanion hole to accom-
modate the pro-(S)-carboxylate, a hydrophobic pocket for
cleavage of the pro-(R)-carboxylate and the catalytic C188 resi-
due (Figure 2). In the (S)-selective AMDase, the catalytic residue
is transferred to the opposite side of the substrate. It should
be noted that the stereoconfiguration of the rate-determining
decarboxylation remains the same. This means that the active
sites of the variants with (R)- and (S)-selectivities do not repre-
sent mirror images, and that the binding of substrates might
differ considerably in such variants.[14b] Taking into account the
importance of synergistic effects that multiple mutations have
on AMDase activity, the same amino acid substitution may
affect (R)- and (S)-selective variants very differently.
Figure 2. Configuration of the active site of (S)-selective and (R)-selective
AMDase variants.
P15G was reported to have an improved (R)-activity,[2] a combi-
nation of this double mutant with the (S)-selective variant to
give G74C/M159L/C188G/P14V/P15G resulted in an inactive
mutant. Variation of the hydrophobic pocket proved to be
a more successful approach. The single substitutions V43I and
V156L led to a fairly measurable activity increase. Their combi-
nation into variant G74C/M159L/C188G/V43I/V156l, however,
doubled their specific activity towards (S)-2. An additional re-
placement of A125 by P led to an activity of 55 UmgÀ1, which
is a further 3.7-fold improvement compared with the previous
variant. A125 is situated on a sharp loop between a b-strand
and an a-helix in the vicinity of the catalytic C74. Although all
known natural AMDases possess this alanine, the closely relat-
ed maleate cis–trans isomerases contain a proline at this posi-
To optimize the activity of AMDase in the synthesis of (S)-2,
we chose variant G74C/M159L/C188G[1] as a starting point
(Figure 1). Micklefield et al. showed that the double mutant
AMDase P14V/P15G leads to a 12-fold activity improvement
over the (R)-selective wildtype.[2] Although the mutation P14V/
Figure 1. Activity improvement of (S)-selective and (R)-selective AMDase variants in the enzymatic decarboxylation of substrates 1a and 1b. Mutants given
on the left part of the graph are based on the CLG variant and are (S)-selective, whereas mutants given on the right part are based on the wildtype (WT)
enzyme and are thus (R)-selective.
ChemCatChem 2016, 8, 916 – 921
www.chemcatchem.org
918
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tion. The substitution A125P proved to be beneficial for the
(S)-activity towards 1b[9] and the racemizing variant AMDase
G74C.[14b] The variant G74C/M159L/C188G/V43I/A125P/V156l
(CLG-IPL) was previously found to have a higher activity in the
synthesis of naproxen.[9] In our studies, this variant showed
a four-fold activity increase towards 1a with respect to the
starting variant G74C/M159L/C188G, which nicely confirms its
increased activity towards a-methyl arylmalonic acids. Interest-
ingly, it converts 1a even faster than the (R)-selective wildtype.
In a pH-based quick assay, the (R)-selective AMDase variant
M159V was reported by Micklefield and co-workers to have
a 43-fold increased activity in the decarboxylation of phenyl-
malonate.[2] It was shown in a later publication that the activity
of AMDase M159V towards pro-chiral malonates is actually in-
creased two-fold.[3] We were pleased to find that this also
holds true for the decarboxylation of 1a with a decarboxylation
activity of 59 UmgÀ1. AMDase P14V/P15G did not show any
measurable activity in the decarboxylation of 1a. Interestingly,
simultaneous introduction of the substitutions A125P, V43I,
V156L, and M159L, which were found to be favorable for in-
creasing the (S)-activity, led to a further activity increase to
209 UmgÀ1. AMDase V43I/A125P/V156L/M159L (WT-IPLL) has
a six-fold higher activity than the wildtype, which is the high-
est activity improvement for the (R)-selectivity of AMDase in
the decarboxylation of prochiral malonates.
lished method.[17] Affinity immobilization showed a very high
immobilization yield of 80.7 mggÀ1 carrier and a high activity
yield of about 19.8% towards a solution of purified free
AMDase. After a slight initial activity loss, the carrier could be
reused for ten consecutive batches without a marked decrease
in yield per batch (Figure 3). A half-life of the biocatalyst prepa-
Figure 3. Enzymatic decarboxylation of 1a catalyzed by covalently immobi-
lized AMDase variant CLG-IPL on C2 amino acrylate in repeated batches. Re-
action conditions: 1 mL reaction volume, 308C, 1500 rpm, 20 mm 1a, 10 mm
Tris buffer, 20.3 mg immobilized enzyme (loading: 80.7 mggÀ1 carrier).
ration of 16.5 h was determined in this application, resulting in
an overall total turnover number of 20.103 for the AMDase.
The improved AMDase variants were then used in the syn-
thesis of both enantiomers of 2a. Either cell-free extracts or
immobilized proteins could successfully be applied in prepara-
tive-scale procedures (Table 1). In all cases, spontaneous decar-
The results from protein engineering show that steric and
electronic effects play a distinctive role in the different
AMDase variants. CLG-IPL showed higher activity in the synthe-
sis of ibuprofen than the wildtype.[9] Interestingly, this is also
the case with 1a, whereas CLG-IPL converts 1b markedly
slower than the wildtype. Owing to its electron-withdrawing
phenyl substituent, 1a has a stronger tendency to decarboxy-
late than 1b. It is known that for structurally similar substrates,
electronic effects determine the activity of AMDase.[16] The
preference of wildtype AMDase for 1b over 1a indicates that
in the case of malonates with a large aryl substituent, steric ef-
fects dominate the efficiency of substrate decarboxylation. The
fact that the improved (R)-selective WT-IPLL and the (S)-selec-
tive CLG-IPL both favor 1a as a substrate strongly indicates
that the activity increase is based on a reduction of steric hin-
drances as a result of the varied space-fills of the exchanged
amino acids, which in turn lead to an improved binding of the
bulky biaryl system.
Table 1. Synthesis of optically pure 2a under optimized reaction condi-
tions.
AMDase Variant[a]
Product
Yield
[%]
Optical purity[b]
ee [%]
WT-IPLL (CFE)
CLG-IPL (CFE)
CLG-IPL (imm.)
(R)-2a
(S)-2a
(S)-2a
99
96
99
98.2
98.1
99.5
[a] CFE=cell-free extract; imm.=immobilized enzyme. [b] Determined by
chiral gas chromatography, Æ0.3% accuracy.
boxylation could be suppressed efficiently. After complete con-
version, (R)-2a and (S)-2a were isolated in high yield (>96%)
and excellent optically purity (>98%) from cell-free extracts.
The synthetic usefulness of the method was demonstrated by
using 150 mg of 1a, which was converted by immobilized
AMDase to (S)-2a with excellent yield (99%) and enantiopurity
(>99%).
In addition to an activity increase by protein engineering,
optimization of enzyme expression and immobilization to
enable easy recycling are strategies to increase the productivi-
ty as well as to minimize the consumption of the biocatalyst.
In consequence, costs and effort for catalyst preparation are re-
duced, which in turn improves the sustainability of the pro-
cess. By substituting the vector pBAD with pET28a, the expres-
sion yield of AMDase was increased from 100 mgLÀ1 to
160 mgLÀ1 cultivation volume. In immobilization processes,
the loss of catalyst during the immobilization procedure is
compensated by an increased operational stability, the possi-
bility to reuse the enzyme, and, most importantly, by signifi-
cant savings in downstream processing. Covalent binding onto
a C2 amino acrylate carrier was chosen as a robust and estab-
Overall, an integrated procedure for the preparation of pro-
fenes 2, by using a new protection group strategy for the
chemical synthesis of arylmalonates, combined with reaction
engineering, demonstrated that bacterial arylmalonate decar-
boxylase is a highly efficient biocatalyst for the conversion of
arylmalonate substrates of limited stability. The chosen path-
way of dibenzylmalonic ester cleavage by hydrogenolysis
avoided unwanted deviations of the pH value during work-up
ChemCatChem 2016, 8, 916 – 921
www.chemcatchem.org
919
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
the obtained protein was determined by Bradford Assay.[18] The im-
mobilized enzyme was stored at 48C until use.
and efficiently suppressed the non-desired spontaneous decar-
boxylation as a side reaction. The deprotection of dibenzyl
malonates 4 proceeded smoothly and led to full conversion;
the initial yields of pure arylmethyl malonic acids 1 after work-
up are promising and have good potential for optimization.
Protein engineering provided two highly active enzyme var-
iants for the synthesis of both enantiomers of 2a. For the first
time, this has made the AMDase-catalyzed synthesis of optical-
ly pure (R)-2a and (S)-2a possible in high yields and pharma-
ceutically relevant enantiopurity (>98% ee).
Synthesis of (R)-2a and (S)-2a by using cell-free extract of
different AMDase variants
For the synthesis of optically pure 2a, the cell-free extract contain-
ing the appropriate AMDase mutant was used. 1a (0.347 mmol,
100 mg) was dissolved carefully in Tris buffer (50 mm) to a final
concentration of 10 mm. To avoid spontaneous decarboxylation,
the pH was kept at pH 8.5. The reaction mixture was incubated
over 20 min at 308C and then stopped by addition of hydrochloric
acid (2m). Complete conversion was confirmed by thin layer chro-
matography. After extraction with MTBE (methyl-tert-butyl ether)
and drying with MgSO4, the analytical purity of 2a was determined
by HPLC (Knauer HPLC Azura; Nucleodur C18 Pyramid Column, Ma-
cherey–Nagel, isocratic with MeCN/H2O/trifluoroacetic acid (TFA)
Conclusions
The results presented here demonstrate that successful incor-
poration of biocatalysts into synthetic pathways needs to con-
sider all aspects of a synthetic route, including the provision of
the starting material. After this first proof-of-concept, the new
protecting group strategy can be easily integrated into the in-
dustrial synthesis of flurbiprofen[12] or other arylpropionates.
The possibility to produce optically pure 2a without increasing
the number of reaction steps is expected to significantly im-
prove the sustainability of the production of this important
pharmaceutical agent.
1
60:40:0.05), and H NMR spectroscopy.
(S)-2a: Using the (S)-selective AMDase variant G74C/M159L/C188G/
V43I/A125P/V156L, (S)-2a (81 mg, 0.331 mmol, 95% yield) was iso-
lated with an optical purity of 98.1% ee.
1H NMR (400 MHz, Chloroform-d): d=7.58–7.50 (m, 2H), 7.47–7.34
(m, 4H), 7.21–7.13 (m, 2H), 3.80 (q, J=7.2 Hz, 1H), 1.57 ppm (d, J=
7.2 Hz, 3H).
(R)-2a: Using the (R)-selective AMDase variant V43I/A125P/V156L/
M159L, (R)-2a (84 mg, 0.344 mmol, 99% yield) was isolated with an
optical purity of 98.2% ee.
Experimental Section
Compounds 1, 2, 3, and 4 were prepared and characterized as de-
scribed in the Supporting Information.
1H NMR (400 MHz, Chloroform-d): d=7.59–7.51 (m, 2H), 7.49–7.34
(m, 4H), 7.23–7.13 (m, 2H), 3.82 (q, J=7.2 Hz, 1H), 1.59 ppm (d, J=
7.2 Hz, 3H).
Enzyme preparation
Escherichia coli BL21 (DE3) cells bearing a pET28a-Vector with an N-
terminal His-Tag sequence and the desired AMDase mutant were
cultivated in LB-medium (200 mL) containing 30 mgmLÀ1 kanamy-
cin at 378C in a 1 L shake flask. After reaching an OD600 of 0.5, IPTG
(isopropyl-b-d-thiogalactopyranoside, 1 mm) was added and the
cells were cultivated for another 12 h at 308C. Cells were harvested
by centrifugation (15 min, 48C, 5000 g) and washed twice with Tris
buffer (tris(hydroxymethyl)-aminomethane, 20 mL, 50 mm, pH 8.0).
Cells were resuspended in Tris buffer (5 mL) containing 10 mm imi-
dazole and were disrupted by sonication. Cell debris was removed
by centrifugation (15 min, 48C, 8000 g). AMDase variants were, if
applicable, purified by His-tag purification by using a Ni sepharose
spin column (ThermoFischer) according to the instructions of the
manufacturer. The protein was eluted with 250 mm imidazole. The
enzyme solution was prepared by washing the protein in centri-
cons (10 kDa membrane, ThermoFischer) with Tris buffer (50 mm,
pH 8.0).
Determination of enzyme activity for immobilized AMDase
CLG-IPL in repeated batch experiments
For activity measurements, the concentration of substrate 1a was
set to 20 mm in 10 mm Tris buffer and the pH was adjusted to 8.5
with 1m NaOH. Experiments were carried out in 2 mL round
bottom microreaction tubes on a Biometra thermo shaker at
1500 rpm and 308C. Immobilized purified enzyme (20.3 mg) or im-
mobilized cell lysate (122.7 mg) was mixed with 1 mL of 20 mm 1a
and the reaction progress monitored by reverse-phase HPLC. Sam-
ples were taken in duplicate over the course of the reaction (12 mL
sample to 6 mL MeCN, ten-fold dilution with water/MeCN (1:1)) and
analyzed by using a C18 column (Nucleodur C18 pyramid 250/4.6,
Macherey–Nagel) with
a
mobile phase of MeCN/H2O/TFA
(59.025:39.025:0.05) at isocratic conditions and a flow rate of
0.8 mLminÀ1 for 9 min. Compounds 1a and 2a eluted at 4.9 and
7.0 min, respectively, as observed by a diode array detector at
245 nm. Between the batches, the enzyme-loaded carrier was care-
fully washed with water and stored at 48C until next usage.
Immobilization on amino C2 acrylate resin
Amino C2 acrylate resin (Iris Biotech) was equilibrated in water for
20 min on a shaker and filtered. The resin was preactivated by in-
cubation with 2% (211.75 mm) glutaraldehyde solution in a ratio of
1:4 (w/w resin/2% glutaraldehyde) for 1 h. After preactivation, the
resin was carefully washed with water and filtered. The preactivat-
ed resin (350 mg) was added to the purified enzyme solution
(8 mL, 4 mgmLÀ1) or cell lysate (8 mL) and shaken (8 rpm, over-
head shaker) for 16 h at 208C. Afterwards, the resin was washed
with 0.5m aqueous NaCl solution and water. The concentration of
Preparation of compound 1a by chemocatalytic hydrogenol-
ysis of benzyl esters
Dibenzyl 2-(2-fluoro-[1,1’-biphenyl]-4-yl)-2-methylmalonate 3a
(63.4 g, 135.3 mmol) was dissolved in methyl-tert-butyl ether
(MTBE, 510 mL) and cooled to 08C. After adding palladium (10%
on activated carbon, 7.2 g, 6.8 mmol), the mixture was stirred
under an atmosphere of hydrogen at 08C for 5 h. The crude mate-
rial was filtered through Celite, washed with MTBE, and the solvent
ChemCatChem 2016, 8, 916 – 921
www.chemcatchem.org
920
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
removed by cold evaporation. Pure 1a (22.7 g 58%) was obtained
by crystallization from n-hexane/ethyl acetate as white solid.
Rhine-Westphalia (grant number PtJ-TRI/1411ng006) are grateful-
ly acknowledged for financial support. We thank Ulf-Peter Apfel
(Bochum) for recording and solving the molecular structure of
compound 4b.
1H NMR (250 MHz, [D6]DMSO): d=13.17 (s, b, 1H, COOH), 7.76–7.07
(m, 8H, Har), 1.76 ppm (s, 3H, CH3).
Keywords: asymmetric
immobilization · lyase · protein engineering
synthesis
·
biocatalysis
·
Conversion of compound 1a with immobilized AMDase
CLG-IPL on a preparative scale
Compound 1a (152.2 mg, 526.9 mmol, 26.35 mm) was dissolved in
20 mL water containing 10 mm Tris buffer and the pH was adjusted
to 8.5 by using 1m NaOH. Immobilized cell lysate (250.9 mg) was
added and the suspension mixed in a water-tempered stirred tank
reactor (308C, 600 rpm). Samples were taken to follow the reaction
course by reverse-phase HPLC as described above. After complete
conversion (22 h), the reaction solution was separated from the im-
mobilized cell lysate by centrifugation. The supernatant was isolat-
ed, acidified with 6m HCl (5 mL), extracted with ethyl acetate (3”
15 mL), and dried by addition of MgSO4. The combined organic
solutions were evaporated to dryness to afford (S)-flurbiprofen (S)-
2a (127.9 mg, 523.6 mmol, 99.1% isolated yield, >99% ee) as an
off-white powder.
[1] Y. Miyauchi, R. Kourist, D. Uemura, K. Miyamoto, Chem. Commun. 2011,
47, 7503–7505.
[2] K. Okrasa, C. Levy, M. Wilding, M. Goodall, N. Baudendistel, B. Hauer, D.
Leys, J. Micklefield, Angew. Chem. Int. Ed. 2009, 48, 7691–7694; Angew.
Chem. 2009, 121, 7827–7830.
[3] R. Lewin, M. Goodall, M. L. Thompson, J. Leigh, M. Breuer, K. Baldenius,
J. Micklefield, Chem. Eur. J. 2015, 21, 6557–6563.
[4] T. A. Hylton, J. A. Walker, EU Patent 0032620, 1981.
[5] H. Geerts, IDrugs 2007, 10, 121–133.
[6] a) H. Jin, Z. Wang, L. Liu, L. Gao, L. Sun, X. Li, H. Zhao, Y. Pan, H. Shi, N.
Liu, L. Hong, J. Liang, Q. Wu, Z. Yang, K. Wu, D. Fan, Mol. Pharm. 2010,
7, 156–168; b) G. Lajos, D. C. Chan, R. S. Hodges, P. A. Bunn, WO 2011/
106721A1, 2011; c) J. K. Liu, S. K. Patel, D. L. Gillespie, K. Whang, W. T.
Couldwell, J. Neuro-Oncol. 2012, 106, 561–569.
[7] a) Y. Yatake, K. Miyamoto, H. Ohta, Appl. Microbiol. Biotechnol. 2008, 78,
793–799; b) K. Miyamoto, H. Ohta, Eur. J. Biochem. 1992, 210, 475–481.
[8] a) Y. Terao, K. Miyamoto, H. Ohta, Chem. Commun. 2006, 3600–3602;
b) M. E. Lind, F. Himo, ACS Catal. 2014, 4, 4153–4160.
[9] S. Yoshida, J. Enoki, R. Kourist, K. Miyamoto, Biosci. Biotechnol. Biochem.
2015, 79, 1965–1971.
[10] J. A. Walker, US Patent 4266069, 1979.
[11] a) G.-S. Shin, K.-W. Lee, T.-K. Kim, H.-D. Shin, Y.-H. Lee, J. Mol. Catal. B
2005, 33, 93–98; b) R. Kourist, P. Domínguez de María, K. Miyamoto,
Green Chem. 2011, 13, 2607–2618.
[12] Y. Terao, Y. Ijima, H. Kakidani, H. Ohta, Bull. Chem. Soc. Jpn. 2003, 76,
2395–2397.
Determination of the optical purity of compounds 2a and
2b
Isolated product powders (ca. 1 mg) were dissolved in MTBE
(600 mL). Methanol (100 mL) and trimethylsilyl diazomethane (25 mL)
were added and the mixture incubated at room temperature for
30 min. Acetic acid (5 mL) was added, the solvents evaporated to
dryness, and the residue re-dissolved in ethyl acetate (200 mL).
Samples were then analyzed by chiral gas chromatography on
a Shimadzu GC Plus 2010 device; using a FS-Hydrodex-b-6TBDM
column (Macherey–Nagel), with a method operating isothermically
at 1708C with an injection split of 1:20. The elution order was
(1) (S)-2a, (2) (R)-2a with baseline separation and a peak difference
of 0.5 min. The ee values were determined from the relative peak
areas according to standard procedures. Errors of ee value determi-
nation were extrapolated from the area difference obtained from
the respective racemic authentic standard. From this, the precision
of the ee values was found to be Æ0.3%.
[13] R. Zuhse, J. Wetzig, R. Kourist, S. Gaßmeyer, DE Patent 10 2014 006 224
A1, 2015.
[14] a) R. Kourist, Y. Miyauchi, D. Uemura, K. Miyamoto, Chem. Eur. J. 2011,
17, 557–563; b) S. K. Gassmeyer, H. Yoshikawa, J. Enoki, N. Hulsemann,
R. Stoll, K. Miyamoto, R. Kourist, ChemBioChem 2015, 16, 1943–1949.
[15] Y. Terao, Y. Ijima, K. Miyamoto, H. Ohta, J. Mol. Catal. B 2007, 45, 15–20.
[16] K. Miyamoto, Ohta Hiromichi, J. Am. Chem. Soc. 1990, 112, 4077–4078.
[17] A. Basso, L. Froment, M. Hesseler, S. Serban, Eur. J. Lipid Sci. Technol.
2013, 115, 468–472.
Acknowledgements
[18] T. Zor, Z. Selinger, Anal. Biochem. 1996, 236, 302–308.
The Deutsche Bundesstiftung Umwelt (AZ30818-32) and the Fed-
eral Ministry for Innovation, Science and Research of North
Received: November 2, 2015
Published online on January 28, 2016
ChemCatChem 2016, 8, 916 – 921
www.chemcatchem.org
921
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim