Preparative asymmetric reduction of ketones in a biphasic medium with an (S)-alcohol dehydrogenase under in situ-cofactor-recycling with a formate dehydrogenase

Article abstract of DOI:10.1016/j.tet.2003.11.066

The substrate range of a novel recombinant (S)-alcohol dehydrogenase from Rhodococcus erythropolis is described. In addition, an enzyme-compatible biphasic reaction medium for the asymmetric biocatalytic reduction of ketones with in situ-cofactor regeneration has been developed. Thus, reductions of poorly water soluble ketones in the presence of the alcohol dehydrogenase from R. erythropolis and a formate dehydrogenase from Candida boidinii can be carried out at higher substrate concentrations of 10-200 mM. The resulting (S)-alcohols were formed with moderate to good conversion rates, and with up to >99% ee.

Full text of DOI:10.1016/j.tet.2003.11.066

Tetrahedron 60 (2004) 633–640
Preparative asymmetric reduction of ketones in a biphasic
medium with an (S)-alcohol dehydrogenase under
in situ-cofactor-recycling with a formate dehydrogenase
a
Harald Groger, Werner Hummel, Claudia Rollmann, Francoise Chamouleau,
b
a
a
,
,
*
*
¨
Hendrik Hu¨sken,a Helge Werner,a Christine Wunderlich,c Kofi Abokitse,b Karlheinz Drauzd
and Stefan Buchholza
aDegussa AG, Project House Biotechnology, P.O. Box 1345, 63403 Hanau, Germany
b
¨
¨
Institut fur Molekulare Enzymtechnologie der Heinrich-Heine-Universitat, Forschungszentrum Julich, Wilhelm-Johnen-Strabe,
¨
¨
52426 Julich, Germany
cDegussa AG, SU Process Technology, P.O. Box 1345, 63403 Hanau, Germany
dDegussa AG, BU Fine Chemicals, FC-TRM, P.O. Box 1345, 63450 Hanau, Germany
Received 9 September 2003; revised 17 November 2003; accepted 19 November 2003
Abstract—The substrate range of a novel recombinant (S)-alcohol dehydrogenase from Rhodococcus erythropolis is described. In addition,
an enzyme-compatible biphasic reaction medium for the asymmetric biocatalytic reduction of ketones with in situ-cofactor regeneration has
been developed. Thus, reductions of poorly water soluble ketones in the presence of the alcohol dehydrogenase from R. erythropolis and a
formate dehydrogenase from Candida boidinii can be carried out at higher substrate concentrations of 10–200 mM. The resulting (S)-
alcohols were formed with moderate to good conversion rates, and with up to .99% ee.
q 2003 Published by Elsevier Ltd.
1. Introduction
reactions are usually carried out at a substrate concen-
tration in the range of 5–10 mM or below due to the low
solubility of hydrophobic ketones in water. The presence
of an organic solvent could improve the solubility of
poorly water-soluble ketones, but generally causes
significant enzyme deactivation. In particular, this is
known for the formate dehydrogenase from C. boidinii8
which is sensitive to organic solvents.9 The best solution
so far is represented by a continuous process with an
enzyme-membrane reactor.10 Albeit good space–
time yields in the range of 60–104 g/(L d) can be
obtained,10a,b the reaction is limited by the solubility of
the ketone in water, which is often below 5–10 mM.
Thus, the development of alternative reaction concepts
for the asymmetric reduction with isolated enzymes still
is a challenge, as well as the access to NAD-dependent
ADHs, which cover a broad substrate range. They
preferably should be cloned and overexpressed in
Escherichia coli under conditions showing a high
production potential.
The enzymatic asymmetric reduction with an in situ-
NADH-cofactor regeneration using a formate dehydrogen-
ase (FDH) from Candida boidinii (Scheme 1, part (a)) is an
interesting technology for the production of optically active
alcohols which are valuable specialty chemicals.1 A related
process, namely the enzymatic reductive amination by
means of an enzyme-coupled in situ-cofactor regeneration
according to Scheme 1, part (b) has been shown to be
technically feasible, and is a powerful tool for the
preparation of enantiomerically pure a-amino acids. This
process runs on the ton scale in the industrial synthesis of
L-tert-leucine (Scheme 1, part (b)).2 The extension of this
attractive enzymatic concept3,4 towards the large scale
synthesis of optically active alcohols5–7 is highly desirable
but still some drawbacks exist.
Among the main limitations are the lacks of isolated
NAD-dependent ADHs commercially available on a
technical scale, and the lack of suitable reaction
media, which guarantee a high ketone solubility. Thus,
In this contribution, we report the investigation of the
synthetic scope of the recently developed (S)-ADH from
Rhodococcus erythropolis overexpressed in E. coli, as well
as the first example of an asymmetric enzymatic reduction
of poorly water soluble ketones including an in situ-
recycling of the cofactor NADH with a FDH, which runs in
Keywords: Ketones; In situ-cofactor regeneration; (S)-Alcohol.
*
Corresponding authors. Tel.: þ49-6181-596401; fax: þ49-6181-532961;
e-mail addresses: harald.groegar@degussa.com;
w.hummel@fz-juelich.de
0040–4020/$ - see front matter q 2003 Published by Elsevier Ltd.
doi:10.1016/j.tet.2003.11.066

¨
H. Groger et al. / Tetrahedron 60 (2004) 633–640
634
R. erythropolis,3k since previous studies12 revealed that
this enzyme possess a potential for general applicability.
This new alcohol dehydrogenase (ADH) isolated from
R. erythropolis shows a tetrameric structure, it has a
molecular weight of 36,206 kDa per subunit, and belongs
to the group of zinc-containing medium-chain ADHs.3k,11a
The developed expression in E. coli is efficient and has a
high production potential. Thus, this E. coli strain should be
suited to make the desired enzyme available on a larger
scale. With this efficient ADH in hand,3k,11a we carried out a
detailed substrate screening of various ketones in order to
evaluate its general utility.
In general, a broad range of substances represent suitable
substrates, and a graphical survey of the photometer-based
investigation of the substrate range is presented in
Tables 1–3.13 Acetophenone was used as a reference
substrate, and its activity was defined as 100%. With respect
to a good preparative applicability, an activity, which is at
least in the range of 50–100% is desirable. To start with the
aromatic ketone substrates bearing one substituent (Table 1),
acetophenones with an interesting variety of substitution
pattern are well accepted. Although the acceptance of any
type of substituent in the para-position is evident (entries
2–9), the type of substituent significantly determine the
degree of activity. Any p-halogenated acetophenone
derivatives (entries 2–4), as well as p-methyl- and
p-methoxysubstituted analogues (entries 5, 8), gave satis-
factory to excellent activities in the range of 194–1333%
compared to acetophenone as a reference substrate.
Notably, the p-bromo- and p-chloro-derivatives led to the
highest activities whereas, the corresponding fluoro-deriva-
tive showed a lower activity. In contrast, very low activities
of only 19 and 17% were found for p-hydroxyacetophenone
and p-aminoacetophenone (entries 7, 9). An activity of
469% was observed for p-(n-butyl)acetophenone (entry 6).
These latter examples might indicate that electronic
properties of the substituent rather than its steric effects
are determining the suitability of a substrate. The þI- or 2I
effect does not appear to have a significant impact.
However, the non-satisfactory results for the p-hydroxy
and p-amino-substituted acetophenone indicate that nucleo-
philicity of the substituent and the ability to form hydrogen
bond interactions might play a significant role.
Scheme 1. Production of chiral compounds with NAD-dependent
dehydrogenase coupled with formate dehydrogenase for NADH regener-
ation. (a) Chiral alcohols with an (S)-specific ADH; (b) L-amino acids via
reductive amination catalyzed by an amino acid dehydrogenase.
the ‘direct’ presence of an organic solvent at high substrate
concentrations.11
2. Results and discussion
2.1. The scope of a new (S)-alcohol dehydrogenase from
R. erythropolis
We recently developed an E. coli expression system for
the NAD-dependent (S)-alcohol dehydrogenase from
Table 1. Enzymatic activities for monosubstituted acetophenones (determined by photometrical assay)
Entrya
R1
R2
R3
Activity (%)
Entrya
R1
R2
Cl
OH
OCH3
NH2
H
R3
Activity (%)
1
2
3
4
5
6
7
8
9
H
F
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
100
194
1198
1333
640
469
19
10
11
12
13
14
15
16
17
H
H
H
H
H
H
H
H
H
H
H
H
2384
47
394
44
85
16
9
Cl
Br
CH3
n-C4H9
OH
Cl
CH3
OH
H
H
H
OCH3
NH2
231
17
OCH3
9
a
The photometrical assay is described in Section 3.

¨
H. Groger et al. / Tetrahedron 60 (2004) 633–640
635
Table 2. Enzymatic activities for di- and trisubstituted acetophenones (determined by photometrical assay)
Entrya
R1
R2
R3
R4
R5
Activity (%)
1
2
3
4
5
H
H
OH
OCH3
F
OCH3
OCH3
H
H
H
OCH3
OCH3
OH
OCH3
F
H
OCH3
H
H
H
H
H
H
H
H
48
21
6
7
305
a
The photometrical assay is described in Section 3.
three substituents. Thus, trimethoxyacetophenone gave a
low activity of 21% (entry 2).
Table 3. Enzymatic activities for aliphatic ketones (determined by
photometrical assay)
In addition, several types of 2-alkanones also serve as good
substrates (Table 3). High activities were observed in
particular when using longer-chain alkanones, for example,
2-hexanone, 2-heptanone, and 2-decanone (entries 1–3).
Furthermore, b-keto esters are good substrates, in particular
when bearing an ethoxy group (entry 7). The presence of
such an ethyl ester group gave noticeably higher results
compared to the methyl ester group (entries 6, 7). The study
of ester substrates is ongoing. Further good substrates are
substituted acetone derivatives, in particular phenoxy-
acetone with an activity of 4180% (entry 5). It should be
added at this stage, that in general substrates bearing an
aceto functionality are good substrates.
Entrya
R
Activity (%) Entrya
R
Activity (%)
4180
1
2
3
4
n-C4H9
n-C5H11
n-C8H17
CH2Cl
1096
3328
2521
119
5
6
7
CH2OPh
CH2C(O)OCH3 134
CH2C(O)OC2H5 1020
a
The photometrical assay is described in Section 3.
In addition, the influence of the o-, m-, and p-position on the
activity was studied. In general, acetophenone derivatives
with substituents in o- or m-position represent suitable
substrates, too. However, it appears to be a general tendency
that o-substituted acetophenones gave less satisfactory
results compared with their m- or p-substituted analogues.
For example, the p-chloroacetophenone and the m-analogue
gave high activities of 1198 and 2384%, respectively,
whereas an activity of only 85% was detected for
o-chloroacetophenone (entries 3, 10, 14). A comparable
tendency can be found for other types of substituents, for
example, methyl- and methoxy-substituents (see entries 5,
8, 15, 17).
Thus, this new recombinant (S)-alcohol dehydrogenase
from R. erythropolis shows both a broad substrate
acceptance and high specific activities.14 Subsequently,
the preparative potential of this (S)-alcohol dehydrogenase
was studied. An initial reduction on a preparative scale with
formate and a formate dehydrogenase for cofactor-regener-
ation was carried out using p-chloroacetophenone, 1a, as a
substrate in homogeneous aqueous solution at low substrate
concentration (5 mM). According to these standard reaction
conditions the product (S)-2a was obtained with a conver-
sion rate of 97% after a reaction time of 21 h. The
subsequent chromatographic purification gave the product,
(S)-2a, in 78% yield, and with an enantioselectivity of
.99% ee (Scheme 2).
Acetophenones bearing two substituents can also represent
suitable substrates in dependence on the substitution pattern
(Table 2). For example, 2,4-difluoroacetopheonone gave an
activity of 305% whereas a remarkably decreased activity of
7% only was determined when using a 2,4-dimethoxy-
acetophenone (entries 4, 5). A significant decrease of the
activity was achieved when using acetophenones bearing
2.2. Discovery of a new reaction medium
Albeit this reduction works well in principle, the main
problem of these standard conditions was the low substrate
Scheme 2. Enzymatic reduction in aqueous medium at low substrate concentration (10 mM).

¨
H. Groger et al. / Tetrahedron 60 (2004) 633–640
636
Scheme 3. Concept of an enzymatic reduction in biphasic media.
concentration of the hydrophobic ketones leading to non-
satisfactory volumetric productivities (Scheme 2). Thus, the
next key step was the development of a suitable aqueous-
organic solvent reaction medium according to Scheme 3
which fulfils the following criteria: (i) both enzymes,
namely the recombinant (S)-ADH and FDH from C. boidinii
and mutants thereof, remain stable in the presence of the
organic solvent; (ii) good solubility of poorly water-soluble
ketones which led to high substrate concentrations of up to
200 mM; (iii) simple reaction protocol for lab scale
applications, which allows a flexible and fast preparation
of chiral alcohols; (iv) robust process with potential for
technical scale applications in the future.
the organic solvent component (Scheme 4).9 In addition,
typical solvents for a biphasic medium failed. For example,
MTBE led to a remarkable decrease of the activity after 2
days. A rapid loss of activity was found when ethylacetate
was used. In this case a (nearly) complete deactivation was
observed within 6 h only. Non-polar aromatic hydro-
carbons, for example, toluene, deactivate FDH remarkably,
too. However, in the presence of 10% (v/v) of n-hexane a
long-term stabilization was achieved (Scheme 4). Even after
66–69 h, an activity of 92 and 90% remained for aqueous
solutions containing 10% ( v/v) and 20% (v/v), respectively,
of n-hexane. Notably, other aliphatic hydrocarbons, for
example, n-heptane also showed a positive effect (with 79%
activity after ca. 3 days when using 10% (v/v)), and the
stability was maintained at increased amount of the organic
solvent of up to 60% (v/v). For this study, we have chosen
n-hexane and in particular n-heptane as an organic solvent
component. The high degree of stability independent of the
amount of organic solvent guarantees sufficient flexibility
with respect to the fine-tuning of the reaction later in the
process development step.
As mentioned above, the stability of the formate dehydro-
genase (FDH) from C. boidinii as the most sensitive enzyme
in this system turned out to be in particular a critical issue.
We focused on this enzyme (or mutants thereof) for
cofactor-regeneration due to several advantages such as
favorable economic data, the large scale availability, and a
proven technical applicability.8 A screening among polar
solvents showed a strong deactivation of the FDH (mutant
C23S, C262A)8 even in the presence of only 10% (v/v) of
In addition, the (S)-selective alcohol dehydrogenase from R.
Scheme 4. Stability of the FDH in aqueous-organic media (the activities were measured after ca. 3 days for n-hexane, 2 days for n-heptane, and 2 days or less
for the other solvents).

¨
H. Groger et al. / Tetrahedron 60 (2004) 633–640
637
Table 4. Reduction in a biphasic medium with different substrates
Entry
Product
Conversion (%)
ee (%)a
1
69
.99
2
65
97
Scheme 5. Conversions of p-chloroacetophenone at increased substrate
concentrations.
enantioselectivity of 97% ee was observed when using
p-bromoacetophenone (entry 2). The asymmetric reduction
of phenoxyacetone proceeded quantitatively under for-
mation of intermediate (S)-2c with an enantioselectivity of
.99% ee (entry 3).
3
.95
.99
a
These studies, which proved the applicability of the new
reaction system for preparative conversions were followed
by further investigation of the new reaction media at higher
substrate concentrations in order to reach high volumetric
productivities. As a model reaction, the reduction of
p-chloroacetophenone, 1a, in the presence of the (S)-ADH
from R. erythropolis3k was investigated. Increased substrate
concentrations from 10 mM (69% conversion) up to
100 mM led to comparable conversion rates (Scheme 5).
At 20 and 40 mM, slightly improved conversion rates of 77
and 75%, respectively, were obtained. Even at 100 mM, a
good conversion of 74% was still achieved. An analogous
reaction with a subsequent chromatographical purification
of the crude product gave the desired product (S)-2a with a
yield of 67%. A further increase of the substrate
concentration up to 200 mM gave a still satisfactory
conversion rate of 63%. Thus, higher volumetric
productivities can be obtained with this new reaction
medium compared to the previous ADH/FDH-coupled
reductions in pure aqueous medium.
The enantiomeric excess (ee) was determined by GC chiral
chromatography.
erythropolis turned out to be stable under the same
conditions. In particular, the stability in heptane and
hexane—which turned out to be very compatible with the
FDH—is high. The activity of the (S)-ADH within 48 h
remains nearly unchanged in the presence of 20% (v/v) as
well as 60% (v/v) of n-heptane indicating a good stability in
the biphasic medium consisting of an aqueous phase and
heptane. Thus, a suitable solvent system was found in which
both enzymes remain stable.
With this enzyme-compatible reaction medium in hand,
preparative conversions were carried out as a next step.
Good conversion rates accompanied by excellent enantio-
selectivities were obtained with a variety of aromatic ketone
substrates. Selected examples are shown in Table 4. In the
presence of the (S)-ADH from R. erythropolis, p-chloro-
acetophenone was converted into the optically active (S)-
enantiomer, (S)-2a, with .99% ee with a conversion of 69%
(entry 1). A conversion of 65%, and a slightly lower
The reaction conditions were further optimized, for
Scheme 6. Quantitative conversion in enzymatic reduction of p-chloroacetophenone in a biphasic medium.

¨
H. Groger et al. / Tetrahedron 60 (2004) 633–640
638
example, with respect to the preferred enzyme/substrate
ratio for a high conversion of .90%. Using improved
reaction conditions (pH 7.0, 30 8C, (S)-ADH and FDH
amount of 60 U/mmol of substrate), a conversion of 95%
was observed at a substrate concentration of 150 mM
(Scheme 6).
0.1 mmol of NADH (70.2 mg) in 100 mL of a phosphate
buffer (50 mM; pH 7.0). After stirring the reaction mixture
for 21 h, the aqueous phase was extracted with 3£100 mL of
methyl tert-butyl ether. The collected organic phases were
analyzed with respect to the conversion rate (97% according
to HPLC), dried over magnesium sulfate, and evaporated in
vacuo. After chromatographical purification (eluent: ethyl
acetate/n-hexane 25:75) of the resulting crude product, the
desired isolated product (S)-2a was obtained in 78% yield
(64.0 mg), and with an enantioselectivity of .99% ee
(determined by HPLC).
In conclusion, the substrate range of a novel recombinant
(S)-alcohol dehydrogenase from R. erythropolis was
studied, and a practical and efficient reaction medium for
the reduction of ketones via enzymatic in situ-cofactor-
regeneration was found. The conversions proceed at high
substrate concentrations in the ‘direct’ presence of an
organic solvent. The general applicability of this reaction
system as well as the simple access to optically active
alcohols on a lab scale are further advantages. Currently,
further process development is in progress.
3.3. Measurement of the enzymatic activity of the
formate dehydrogenase (FDH) from C. boidinii (mutant:
C23S, C262A; overexpressed in E. coli; see also Ref. 8)
after storage in aqueous organic solvent mixtures
At first, 100 mL of a formate dehydrogenase solution
(formate dehydrogenase (FDH) from C. boidinii (mutant:
C23S, C262A; overexpressed in E. coli; see also Ref. 8)
were dissolved in 10 mL of a solvent mixture containing
10–20% (v/v) of an organic solvent (an overview about
the investigated solvents is given in Scheme 4). The
samples were stored for a duration of up to 3 days, and
regularly samples of 100 mL are taken from the solution.
These samples were analyzed with respect to the enzymatic
activity by spectrophotometrically measuring the reduction
of NADþ to NADH at 340 nm (1340¼6.22 mM21 cm21)
with sodium formate (spectrophotometer: JASCO V-530
UV/VIS). The activity was measured at 30 8C in a cuvette
(1 mL) containing 500 mL of a sodium formate solution
(0.8 M; pH 8.2; 0.1 M phosphate buffer), and 500 mL of a
NADþ-solution (0.4 mM). The reaction was started by the
addition of the enzyme solution (10 mL of the 100 mL
solution). One unit of FDH activity was defined as the
amount of enzyme that converted 1 mmol of NADþ per
minute. The resulting enzymatic activities of the FDH in
dependency of the storage time in the presence of organic
solvents are described in Scheme 4.
3. Experimental
The measurements of the enzymatic activity were carried
out using the spectrophotometer JASCO V-530 UV/VIS.
The solution of the (S)-alcohol dehydrogenase from
R. erythropolis (overexpressed in E. coli) was prepared
according to Refs. 3k and 11a. The solution of the formate
dehydrogenase from C. boidinii (mutant: C23S, C262A;
overexpressed in E. coli) was prepared according to Ref. 8.
The ketones, cofactors, sodium formate and the solvents
are commercially available and were used in the activity
tests and preparative conversions without further
purification.
3.1. Measurement of enzyme activity (typical protocol;
according to Tables 1–3)
The NAD-dependent ADH activity was determined spectro-
photometrically measuring the oxidation of NADH at
340 nm (1340¼6.22 mM21 cm21) in the presence of the
ketone. The activity was measured at 30 8C in a cuvette
(1 mL) containing 1.45 mM of the ketone in 100 mM
sodium phosphate buffer (pH 6.0), and 0.25 mM of NADH.
The reaction was started by the addition of the recombinant
(S)-alcohol dehydrogenase from R. erythropolis (10 mL of
the (S)-ADH solution described below as a crude extract or
partially purified by ion exchange chromatography, which
was diluted by a factor of 100). One unit of ADH activity
was defined as the amount of enzyme that converted 1 mmol
of NADH per minute. The activity tests described in entries,
1–5, 8, 10, 14 in Table 1, and in all entries in Table 3, are
based on the use of an (S)-ADH solution with an activity of
13.4 U/mL (for acetophenone); the activity tests described
in entries, 6, 7, 9, 11–13, 15–17 in Table 1, and in all entries
in Table 2, are based on the use of an (S)-ADH solution with
an activity of 43.3 U/mL (for acetophenone).
3.4. Enzymatic reduction with different types of
substrates (according to Table 4)
A general procedure is as follows: At a reaction temperature
of 30 8C, 10 U of the (S)-alcohol dehydrogenase from
recombinant R. erythropolis (see also Ref. 3k), and 10 U of
the formate dehydrogenase from C. boidinii (mutant: C23S,
C262A; expressed in E. coli; see also Ref. 8) are added to a
solution of 0.5 mmol of the ketone component (83.9 mg),
2.5 mmol sodium formate (171.6 mg), and 0.1 mmol of
NADH (70.2 mg) in a solvent mixture, consisting of 10 mL
of n-heptane, and 40 mL of a phosphate buffer (50 mM;
pH 7.0). After stirring the reaction mixture for 21 h, the
organic phase was separated and the aqueous phase is
extracted with 3£50 mL of methyl tert-butyl ether. The
collected organic phases are dried over magnesium sulfate,
and after filtration and evaporation of the volatile com-
ponents in vacuo the resulting oily crude product was
analyzed with respect to the conversion rate (via NMR,
HPLC). The ee was determined by chiral GC chromato-
graphy using a chiral column ‘CP-Chirasil-DEX CB’
(length: 25 cm; diameter: 25 mm; 1.3 mL/min; gas: He)
from the company Chrompack.
3.2. Enzymatic reduction in aqueous media (according to
Scheme 2)
At a reaction temperature of 30 8C, 10 U of the (S)-alcohol
dehydrogenase, and 10 U of the formate dehydrogenase was
added to a solution of 0.5 mmol p-chloroacetophenones
(83.9 mg), 2.5 mmol sodium formate (171.6 mg), and

¨
H. Groger et al. / Tetrahedron 60 (2004) 633–640
639
J. Mol. Biol. 1998, 278, 967–981. (f) Holt, P. J.; Williams,
R. E.; Jordan, K. N.; Lowe, C. R.; Bruce, N. C. FEMS
Microbiol. Lett. 2000, 190, 57–62. (g) Matsuda, T.; Harada,
T.; Nakamura, K. Chem. Commun. 2000, 1367–1368.
(h) Riebel, B.; Hummel, W. Biotechnol. Lett. 2001, 23,
231–234. (i) Schubert, T.; Hummel, W.; Kula, M.-R.; Mu¨ller,
M. Eur. J. Org. Chem. 2001, 4181–4187. (j) Stampfer, W.;
Kosjek, B.; Moitzi, C.; Kroutil, W.; Faber, K. Angew. Chem.
2002, 114, 1056–1059. (k) Abokitse, K.; Hummel, W. Appl.
Microbiol. Biotechnol. 2003, 62, 380–386.
3.5. Enzymatic reduction at higher substrate
concentrations (according to Scheme 5) exemplified for
the reaction with a substrate concentration of 100 mM
At a reaction temperature of 30 8C, 10 U of the recombinant
(S)-alcohol dehydrogenase from R. erythropolis (see also
Ref. 3k), and 10 U of the formate dehydrogenase from
C. boidinii (mutant: C23S, C262A; expression in E. coli; see
also Ref. 8) were added to a solution of 0.5 mmol of
p-chloroacetophenone (83.9 mg), 2.5 mmol sodium formate
(171.6 mg), and 0.1 mmol of NADH (70.2 mg) in a solvent
mixture, consisting of 1 mL of n-heptane, and 4 mL of a
phosphate buffer (50 mM; pH 7.0). After stirring the
reaction mixture for 21 h, the organic phase was separated
and the aqueous phase was extracted with 3£5 mL of
methyl tert-butyl ether. The collected organic phases were
dried over magnesium sulfate, and after filtration and
evaporation of the volatile components in vacuo the
resulting oily crude product was analyzed with respect to
the conversion rate (via NMR, HPLC).
4. For reviews about biocatalytic reductions in general, see:
(a) Hummel, W. Adv. Biochem. Engng Biotechnol. 1997, 58,
146–184. (b) Faber, K. Biotransformations in organic
chemistry; 4th ed. Springer: Berlin, 2000; Chapter 2.2.3,
pp 192–194.
5. For selected contributions of an alternative approach via
asymmetric metal-catalyzed hydrogenation of ketones, see:
(a) Burk, M. J.; Hems, W.; Herzberg, D.; Malan, C.; Zanotti-
Gerosa, A. Org. Lett. 2000, 2, 4173–4176. (b) Ohkuma, T.;
Koizumi, M.; Yoshida, M.; Noyori, R. Org. Lett. 2000, 2,
1749–1751. (c) Ohkuma, T.; Takeno, H.; Honda, Y.; Noyori,
R. Adv. Synth. Catal. 2001, 343, 369–375. Review: (d) Noyori,
R.; Okhuma, T. Angew. Chem., Int. Ed. 2001, 40, 40–73.
6. For selected contributions of an alternative approach via
asymmetric whole-cell-biocatalytic reduction of ketones, see:
(a) Yasohara, Y.; Kizaki, N.; Hasegawa, J.; Wada, M.;
Kataoka, M.; Shimizu, S. Tetrahedron: Asymmetry 2001, 12,
1713–1718. (b) Korkhin, Y.; Kalb, A. J.; Peretz, M.; Bogin,
O.; Burstein, Y.; Frolow, F. J. Mol. Biol. 1998, 278, 967–981.
(c) Riebel, B.; Hummel, W. Biotechnol. Lett. 2001, 23,
231–234. (d) Stampfer, W.; Kosjek, B.; Faber, K.; Kroutil, W.
J. Org. Chem. 2003, 68, 402–406. (e) Haberland, J.; Hummel,
W.; Daubmann, T.; Liese, A. Org. Proc. Res. Dev. 2002, 6,
458–462.
In addition, for the asymmetric reduction at a substrate
concentration of 100 mM the product (S)-2a has been
purified from the crude product and isolated via TLC-
chromatography (eluent: ethylacetate/n-hexane (25:75);
thin-layer-chromatography plate: Merck TLC plate
20£20 cm, silica gel 60 F254; Rf-value: 0.21; yield: 67%).
Acknowledgements
This work was supported by the Bundesministerium fu¨r
Bildung und Forschung (Biotechnologie 2000—
Nachhaltige BioProduktion; Project: ‘Entwicklung eines
biokatalytischen und nachhaltigen Verfahrens zur
industriellen Herstellung enantiomerenreiner Amine und
Alkohole unter besonderer Beru¨cksichtigung der
7. For selected contributions of an alternative approach via
asymmetric reduction of ketones based on substrate-coupled
cofactor-regeneration, see: (a) Wolberg, M.; Ji, A. G.;
Hummel, W.; Mu¨ller, M. Synthesis 2001, 937–942.
(b) Schubert, T.; Hummel, W.; Mu¨ller, M. Angew. Chem.
¨
Atomokonomie’).
¨
2002, 114, 656–659. (c) Stillger, T.; Bonitz, M.; Villela, F.;
Liese, A. Chem. Ing. Techn. 2002, 74, 1035–1039.
8. The properties of the cofactor-regenerating enzyme FDH from
C. boidinii have been improved remarkably by protein
engineering in the Kula group. Thus, a stable and efficient
formate dehydrogenase is now available on large scale for
NADH-regeneration, see: Slusarczyk, H.; Felber, S.; Kula,
M.-R.; Pohl, M. Eur. J. Biochem. 2000, 267, 1280–1289.
9. In spite of its high technical potential and applicability, the
FDH from C. boidinii was found to be sensitive towards the
presence of organic solvents. The general problem of rapid
deactivation of enzymes other than hydrolases in the presence
of organic solvents, has been previously described in several
coontributions, e.g. in: Kruse, W.; Kragl, U.; Wandrey, C. DE
4436149, 1996.
References and notes
1. Reviews: (a) Kula, M. R.; Kragl, U. In Stereoselective
biocatalysis; Patel, R. N., Ed.; Marcel Dekker: New York,
2000; pp 839–866 Chapter 28. (b) Hummel, W. TIBTECH
1999, 17, 487–492.
2. (a) Hummel, W.; Schu¨tte, H.; Schmidt, E.; Wandrey, C.; Kula,
M.-R. Appl. Microbiol. Biotechnol. 1987, 26, 409–416.
(b) Bommarius, A. S.; Drauz, K.; Hummel, W.; Kula, M.-R.;
Wandrey, C. Biocatalysis 1994, 10, 37–47. (c) Krix, G.;
Bommarius, A. S.; Drauz, K.; Kottenhahn, M.; Schwarm, M.;
Kula, M.-R. J. Biotechnol. 1997, 53, 29–39.
3. Numerous efficient (S)- and (R)-specific alcohol dehydrogen-
ases have been already found. For selected contributions, see:
(a) Bradshaw, C. W.; Hummel, W.; Wong, C.-H. J. Org.
Chem. 1992, 57, 1532. (b) Ammendola, S.; Raia, C. A.;
Caruso, C.; Camardella, L.; D’Auria, S.; De Rosa, M.; Rossi,
M. Biochemistry 1992, 31, 12514–12523. (c) Cannio, R.;
Fiorentino, G.; Carpinelli, P.; Rossi, M.; Bartolucci, S.
J. Bacteriol. 1996, 178, 301–305. (d) Bogin, O.; Peretz, M.;
Burstein, Y. Protein Sci. 1997, 6, 450–458. (e) Korkhin, Y.;
Kalb, A. J.; Peretz, M.; Bogin, O.; Burstein, Y.; Frolow, F.
10. This efficient ‘three-loops’-concept is based on an enzymatic
reaction in pure aqueous medium, a separation of the aqueous
phase from the enzyme via ultrafiltration, and a subsequent
continuous extraction of the aqueous phase with an organic
solvent. Organic and aqueous phases are separated by a
hydrophobic membrane. This method is described in: (a) Liese,
A.; Seelbach, K.; Wandrey, C. Industrial biotransformations;
Wiley-VCH: Weinheim, 2000. (b) Kruse, W.; Hummel, W.;
Kragl, U. Recl. Trav. Chim. Pays-Bas 1996, 115, 239–243.

¨
H. Groger et al. / Tetrahedron 60 (2004) 633–640
640
11. For preliminary communications, see: (a) Hummel, W.;
¨
Abokitse, K.; Drauz, K.; Rollmann, C.; Groger, H. Adv.
activity was not observed in the experimental investigation.
The specific activities of this (S)-ADH from R. erythropolis
expressed in E. coli is in the range of 100–1000 U/mg of
purified protein depending of the structure to be converted, see
also Ref. 3k. For comparison, horse liver alcohol dehydro-
genase as one of the most widely applied (S)-alcohol
dehydrogenase so far shows an activity in the range of
1–3 U/mg only, see Ref. 1a.
¨
Synth. Catal. 2003, 345, 153–159. (b) Groger, H.; Hummel,
W.; Buchholz, S.; Drauz, K.; Nguyen, T. V.; Rollmann, C.;
Hu¨sken, H.; Abokitse, K. Org. Lett. 2003, 5, 173–176.
12. (a) Hummel, W.; Gottwald, C. DOS 4209022, 1993; Chem.
Abstr. 1994, 120, 29495.. (b) Zelinski, T.; Peters, J.; Kula,
M.-R. J. Biotechnol. 1994, 33, 283–292.
13. The determination of enzyme activities were done using a cell-
free crude extract of E. coli with the recombinant (S)-ADH
enzyme from R. erythropolis. In addition, activity studies were
also done with a (partially) purified form of the enzyme. The
presence of other dehydrogenases causing a background
14. It is to be noticed that several differences in substrate range
were observed in comparison with the corresponding wild-
type alcohol dehydrogenases (from crude extracts), see also
Ref. 11a.