Ketoreductases: Stereoselective catalysts for the facile synthesis of chiral alcohols

Article abstract of DOI:10.1016/j.tetasy.2005.10.002

The results of a reduction of a wide range of ketones using 31 commercially available isolated ketoreductases (KREDs) are presented. All enzymes accepted a wide substrate range. The stereoselectivity of each enzyme was measured for the reduction of benzoyl-hydroxyacetone and ethyl-3-oxobutanoate, and in each case, enzymes which produce enantiomerically pure (R)- and (S)-alcohols were found. The preparative scale reactions were investigated using two ketones with different hydrophobicities (benzoyl-hydroxyacetone and α-tetralone) and using enzymes with varying specific activities for their reduction. Regardless of the hydrophobicity of the substrate, high titers of ketone (0.75-1.4 M) were reduced in high yield using catalytic amounts of enzyme (1-7% g/g relative to substrate) and cofactor (0.1-0.2% equiv. relative to ketone) within 4-24 h. The cofactor was efficiently regenerated in situ via the oxidation of glucose by glucose dehydrogenase, an enzyme that has also been cloned and over-expressed. These results show that isolated ketoreductases can be quickly and easily screened against target ketones, and the reactions can be scaled to produce preparative amounts of chiral alcohols. Ketoreductase enzymes should become a standard addition to the organic chemists toolbox of asymmetric catalysts for stereoselective ketone reduction.

Full text of DOI:10.1016/j.tetasy.2005.10.002

Tetrahedron:
Asymmetry
Tetrahedron: Asymmetry 16 (2005) 3682–3689
Ketoreductases: stereoselective catalysts for the
facile synthesis of chiral alcohols
Iwona A. Kaluzna, J. David Rozzell and Spiros Kambourakis*
BioCatalytics Inc, 129 N Hill Ave, Suite 103, Pasadena, CA 91106, USA
Received 12 August 2005; revised 3 October 2005; accepted 5 October 2005
Abstract—The results of a reduction of a wide range of ketones using 31 commercially available isolated ketoreductases (KREDs)
are presented. All enzymes accepted a wide substrate range. The stereoselectivity of each enzyme was measured for the reduction of
benzoyl-hydroxyacetone and ethyl-3-oxobutanoate, and in each case, enzymes which produce enantiomerically pure (R)- and
(
(
S)-alcohols were found. The preparative scale reactions were investigated using two ketones with different hydrophobicities
benzoyl-hydroxyacetone and a-tetralone) and using enzymes with varying specific activities for their reduction. Regardless of
the hydrophobicity of the substrate, high titers of ketone (0.75–1.4 M) were reduced in high yield using catalytic amounts of enzyme
(
1–7% g/g relative to substrate) and cofactor (0.1–0.2% equiv. relative to ketone) within 4–24 h. The cofactor was efficiently regen-
erated in situ via the oxidation of glucose by glucose dehydrogenase, an enzyme that has also been cloned and over-expressed. These
results show that isolated ketoreductases can be quickly and easily screened against target ketones, and the reactions can be scaled to
produce preparative amounts of chiral alcohols. Ketoreductase enzymes should become a standard addition to the organic chemists
toolbox of asymmetric catalysts for stereoselective ketone reduction.
ꢀ
2005 Elsevier Ltd. All rights reserved.
1. Introduction
and KREDs that can catalyze the synthesis of either
the (R)- or (S)-alcohol from a starting ketone are usually
3
–5
Stereoselective reduction of ketones to form chiral alco-
hols is one of the most useful reactions in organic syn-
available.
The costof KREDs also compares favor-
ably to that of other chiral catalysts. KRED enzymes
can be produced economically in large quantities scale
using microbial fermentations. Enzymatic reductions
of ketones are thus an important addition to the syn-
thetic chemistÕs toolbox, complementing and expanding
the established stereoselective chemical reductions.
1
,2
thesis.
Chiral alcohols are presentin a vari ey t of
pharmaceutical and other high value compounds and
they are frequently used as intermediates for multi-step
chiral syntheses. It is therefore not surprising that signi-
ficant effort has been devoted to the development of
chemical methods for the reduction of ketones to pro-
2
duce alcohols of high enantiomeric purity. Ketoreduc-
Identification of the best enzyme for a desired applica-
tion is the key step in developing a KRED-catalyzed
route to a chiral alcohol. Although whole cell systems
tases (KREDs) offer important advantages as catalysts
in ketone reduction processes. Recent enzyme discovery
efforts have expanded dramatically upon the number of
4
,6
have been frequently reported, screening of isolated
enzymes is a more rapid and convenientapproach for
chemists. In contrast to whole cell systems, which
require labor-intensive culturing of viable cells, KRED
enzymes can be added directly to the reaction mixture
using standard laboratory equipment. Rate and purity
is another advantage: whole cells typically display low
reaction rates and multiple competing ketoreductase
activities, while isolated KRED enzymes can convert
high titers of substrate quickly using catalytic amounts
of enzyme. Importantly, no competing reactions by
other ketoreductases occur since each isolated enzyme
contains a single, specific ketoreductase activity.
3
,4
KREDs available for use and cost-effective methods
for recycling of the nicotinamide cofactors have been
4
developed. Reactions using ketoreductases typically
require no special conditions (20–45 ꢁC, atmospheric
pressure), and waste disposal problems are minimal
since enzymes are completely biodegradable and reac-
tions can be performed in aqueous media. High yields
and optical purities (>99%) are frequently achieved,
*
Corresponding author. Fax: +1 626 356 3999; e-mail:
skambourakis@biocatalytics.com
0957-4166/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2005.10.002

I. A. Kaluzna et al. / Tetrahedron: Asymmetry 16 (2005) 3682–3689
3683
Herein, we report the results of the screening of 31
cloned ketoreductase enzymes for the stereoselective
ing of 31 NADPH-dependent ketoreductases.5 Every
enzyme was found to act on multiple ketone substrates
with activities ranging from 0.01 to 59 U/mg (Table 1).
Twelve enzymes showed activity on almost every ketone
assayed, while the other 19 enzymes were active on four
or more ketones. High activities, in the range of 10–
59 U/mg, were frequently observed, and more impor-
tantly, for every ketone there was at least one enzyme
identified, which showed good to excellent catalytic
activity. Even with an activity of 0.1 U/mg or less, com-
plete product formation was achieved if the reaction was
allowed to progress for 24–30 h. Details about the titers
and yields of product ht atwere achieved are discussed
later on. The activity range of these enzymes as shown
in Table 1 is especially impressive, given the structural
diversity of the assayed substrates. The ketones are
depicted in Figure 1 and include various substituted b-
keto esters 1–4, one a-keto ester 9, one a-keto acid 8
one aromatic hydroxy ketone 5, benzyl protected
hydroxyacetone 10, a bicyclic ketone 6, and an aliphatic
ketone 7.
5
reduction of a set of 10 structurally different ketones.
The substrate range of these reductive enzymes, the
stereoselectivity thus obtained and the potential scale-
up of such enzymatic processes, are discussed using
some selected examples.
2
. Results and discussion
Until recently, very few ketoreductases were commer-
cially available with the potential for scale-up to com-
7
mercial production. With the recent introduction of
ketoreductases from a variety of sources as individual,
3
–5
isolated enzyme activities,
the rapid screening of
libraries of these enzymes for reduction of a target
ketone has become possible for the first time. A number
3
,4
of such enzymes have been cloned and characterized,
however, using easily accessible, commercially available
ketoreductase screening sets offers a convenient and
rapid method to screen a group of diverse enzymes
against the reduction of any target ketone. In Table 1,
we present an example of the range of substrates
that can be reduced using such a commercial set consist-
In addition to the substrates shown in Table 1, the same
setof commercially available enzymes has recen tl y been
utilized by our laboratory and others for the reduction
À1
Table 1. Specific activities (lmol/min · mg ) on assays performed with 10 substrates (Fig. 1) and 31 ketoreductases
a
KRED
Enzyme activities (U/mg) for substrates 1–10
1
2
3
4
5
6
7
8
9
10
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
123
124
125
126
127
128
129
130
131
132
2.00
4.27
0.960.32
1.26
1.18
1.98
0.084
2.19
0.94
0.18
1.48
0.120
0.076
0.074
0.013
0.014
1.18
0.168
27.0
1.48
4.74
0.18
0.083
0.64
0.94
0.056
0.022
0.148
0.384
0.038
0.074
0.022
0.54
0.022
0.178
0.073
2.17
0.040
0.029
13.8
0.035
0.011
0.021
0.092
0.678
0.44
0.204
0.112
1.03
0.020
0.137
0.028
0.105
0.134
0.834
0.053
0.127
0.046
0.864
0.068
0.023
0.168
1.11
0.072
0.014
2.61
18.0
9.60
0.054
0.29
30.12
0.049
0.582
0.053
0.564
1.19
0.12
0.027
0.051
1.02
0.060
4.26
0.94
0.146
0.138
0.064
3.24
0.074
0.678
0.127
0.168
0.084
4.32
0.270
0.516
0.300
0.492
0.144
0.020
0.210
0.024
0.110
0.008
0.10
0.075
0.336
0.20
7.86
14.46
3.24
2.70
0.288
0.22
1.07
11.88
2.52
3.84
0.23
0.142
0.026
36.9
0.088
0.348
0.216
1.92
0.072
0.122
0.33
0.630
7.44
8.10
26.64
26.58
15.9
0.91
5.82
3.66
8.94
0.061
0.324
0.4860.282
0.030
0.040
0.020
0.021
1.63
0.088
32.1
45.2
0.510
0.708
0.234
0.052
0.055
0.056
0.016
0.024
0.024
0.038
0.038
0.034
0.15
0.034
0.084
22.38
47.04
0.774
0.060
1.38
23.52
1.94
15.60
0.318
4.26
5.16
20.52
0.930
0.058
0.198
58.9
0.089
0.306
0.063
0.049
0.021
0.046
0.474
0.046
0.042
3.60
0.246
0.396
0.954
0.068
0.29
0.124
0.112
0.060
0.342
1.82
0.085
0.173
0.018
0.053
0.607
0.314
1.59
0.028
0.046
0.060
0.022
0.08
0.142
0.156
0.157
1.15
0.055
1.32
1.33
0.046
0.081
0.183
0.058
0.074
1.81
0.23
0.036
0.097
0.013
0.045
0.085
1.11
0.045
0.146
0.100
1.20
0.652
0.60
0.122
2.09
1.04
0.936
0.015
0.042
0.94
0.318
7.20
3.10
2.88
2.20
5.09
Activity was measured in spectrophotometric cuvette assays by the loss of absorbance at 340 nm resulting from the decrease of NADPH concen-
tration in enzyme–substrate mixtures. Details in experimental.
Bold: Activities between 0.1 and 1 U/mg; bold italics: activities > 1 U/mg.
a
Unit definition: 1 U/mg is the conversion of 1 lmol ketone per minute per mg of enzyme.

3684
I. A. Kaluzna et al. / Tetrahedron: Asymmetry 16 (2005) 3682–3689
O
O
O
O
O
5
CO Et
Cl
CO Et
2
CO Et
2
2
CO Et
2
OH
Ph
Ph
Ph
Ph
O
1
2
4
3
O
O
O
O
Ph
O
CO H
Ph
CO Et
2
2
O
6
7
8
9
10
Figure 1. Substrates shown in Table 1.
of a variety of other ketones with high efficiency and
stereoselectivity. Substrates that were utilized for reduc-
ginally different between the two substrates as shown
by the product ratios of KRED110, KRED113 and
KRED125. For example, KRED110 gave close to a
racemic mixture upon reduction of benzyl-hydroxyace-
tone, while 97% of the (S)-enantiomer was produced
from the reduction of ethyl-3-oxobutyrate. Such an
unpredictable preference of reduction using the same
KRED screening sethas previously been observed in
8
tion included a-alkyl substituted 1,3-diketones, a- and
9
b-alkyl substituted b-keto ketoesters, a wide range of
1
0
substituted acetophenones and substituted diethyl 2-
1
1
alkyl ketoglutarates. The latter compounds have been
used in the synthesis of various amino alcohols and
1
1
hydroxyl-amino acids.
8
–11
the reduction of other classes of compounds.
The substrates that are presented in Table 1 include sev-
eral industrially useful ketones. For example (S)-ethyl-4-
chloro 3-hydroxy butanoate (produced by the reduction
of ketone 1) is a key intermediate in the synthesis of
the cholesterol lowering drugs Lipitor and Zocor, while
These results indicate that the activity and stereoselectiv-
ity of the reduction of an unknown ketone often corre-
lates with the activity and stereoselectivity toward
other similar substrates, but predictions cannot be made
with absolute certainty. It is, therefore, advisable to
screen the set of enzymes for activity and stereoselectiv-
ity for each new ketone to ensure that the best enzyme is
not missed; and this can be very easily done using the
methods described herein. It should be noted that in
some cases even when no activity was detected in spec-
trophotometric assays (Table 1), productwas observed
after overnight incubation with the enzyme.
(
R)-ethyl-4-phenyl-2-hydroxy-butanoate (produced by
the reduction of ketone 9) was utilized in the synthesis
of hypertension drugs Lisinopril and Enalaprilat. Aryl
substituted diols (from ketone 5) can be used for the
1
2
synthesis of various pharmaceuticals. Analogues of
the alcohol derived from a-tetralone (ketone 6) have
been used in the synthesis of antagonists of serotonin
5
-HT7 receptors for the treatment of schizophrenia13a
and in the synthesis of inhibitors of b-secretase in the
1
3b
treatment of Alzheimers disease.
All other ketones
In Table 3, preparative scale (5–25 g) reactions of two
substrates with varying hydrophobicities using enzymes
with varying activities towards their reductions are pre-
sented. Although the reactions shown in this table have
not been optimized, the results presented here clearly
show that high titers and yields of product can be
achieved, even when a hydrophobic substrate or an
enzyme with relatively low specific activity for reduction
is utilized.
8
–11
presented in Figure 1 and in the references
resentke ot ne precursors for commercially useful chiral
alcohols.
also rep-
High stereoselectivity of reduction is another advantage
of using isolated ketoreductases as catalysts for reduc-
2
,3,8–11
tion.
This is clearly illustrated in Table 2, where
the stereoselectivity of the thirty one ketoreductases
towards the reduction of two structurally similar
ketones is presented. The enzymes that gave more than
The amount of enzyme that was necessary to complete
every reduction depended on its specific activity for the
substrate, its stability under the reaction conditions over
time, and the concentration of the dissolved substrate in
the reaction media. The effect of the specific activity and
enzymatic stability on the outcome of the enzymatic
reaction is clearly shown by the reduction of benzoyl-
hydroxyacetone using KRED107 and KRED117 (Table
3, entries 2 and 3). The much higher specific activity of
KRED117 towards benzoyl-hydroxyacetone (Table 3,
entry 3) resulted in the need for significantly less enzyme
to complete the reaction in the same time compared to
KRED107 (Table 3, entry 2). Multiple additions of
KRED117 during the course of the reaction were more
effective in carrying the reaction to completion compared
to a single addition. On the other hand, using a highly
stable enzyme such as KRED107, a single addition of
9
5% formation of one enantiomer are indicated in bold.
The enantiomeric ratios were determined by chiral GC
analysis of crude extracts from 5 mL overnight reactions
containing 20 mM substrate and 2–5 mg of each enzyme
(
details in experimental). Two observations can be made
immediately by looking at this table. First, about half of
the enzymes gave reductions with more than 95%
enantioselectivity for both substrates. Secondly, both
enantiomers of each chiral alcohol can be synthesized
in excellent enantiomeric excess once the appropriate
enzyme is identified. Although in most cases, the abso-
lute stereochemistry of the alcohol formed by an enzyme
was the same for both substrates, there are two cases,
namely KRED104 and KRED123, where enzymes gave
the opposite enantiomer as the major product in each
substrate. In a few cases, the stereoselectivity was mar-

I. A. Kaluzna et al. / Tetrahedron: Asymmetry 16 (2005) 3682–3689
Table 2. Stereoselectivity of reduction using the ketoreductases of Table 1
3685
a
OH
OH
OH
OH
KRED
U/mg
O
Ph
O
Ph
U/mg
CO2Et
CO Et
2
R
S
R
S
O
O
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
123
124
125
126
127
128
129
130
131
132
23%
<1%
<1%
7%
8%
<2%
>99%
<1%
12%
3%
33%
11%
<1%
32%
33%
<1%
<1%
<1%
<1%
<1%
65%
62%
<1%
40%
<1%
<1%
<1%
72%
14%
75%
>99%
77%
>99%
>99%
93%
0.49
1.13
33%
1%
<1%
76%
1%
2%
67%
99%
>99%
24%
99%
98%
<1%
98%
84%
55%
71%
58%
69%
61%
71%
>99%
>99%
91%
>99%
>99%
35%
69%
99%
96%
>99%
96%
96%
32%
78%
36%
ND
1.48
4.74
0.18
0.242
ND
ND
ND
92%
0.204
0.112
1.03
>98%
<1%
>99%
88%
97%
67%
89%
>99%
68%
0.142
2.00
3.18
>99%
2%
30.12
0.049
ND
0.012
0.020
0.360
0.822
0.312
0.058
0.016
6.12
16%
45%
29%
42%
31%
39%
29%
<1%
<1%
9%
<1%
<1%
65%
31%
1%
0.075
0.336
0.20
0.142
0.026
36.9
67%
>99%
>99%
>99%
>99%
>99%
35%
38%
>99%
60%
>99%
>99%
>99%
28%
86%
25%
5.34
32.1
45.2
13.14
23.4
1.20
58.9
3.60
0.033
0.030
0.050
ND
0.060
0.342
1.82
4%
<1%
4%
0.055
1.32
1.33
0.649
0.710
0.404
0.014
0.084
0.059
4.44
4%
0.045
ND
68%
22%
64%
>99%
0.015
0.042
0.94
ND
a
The enantiomeric ratio was measured by chiral GC analysis of crude organic extracts from 5 mL reactions containing 20 mM ketone and 2–5 mg of
each enzyme (details in experimental).
Table 3. Preparative-scale (5–25 g) synthesis of selected substrates and ketoreductases with varying activities
Entry Substrate, KRED U/mg Substrate Substrate Enzyme DMSO NADPH
concentration (M) titer (g/L) concentration (g/L) % (v/v) concentration (mM) time (h) yield
Rxn
Isolated
Figure 1
c
1
2
3
4
5
6
7
10
10
10
10
10
6
111
107
117
117
117
101
101
0.075
1.03
32.1
32.1
32.1
1.48
1.48
0.140
0.140
0.140
0.750
1.40
25
25
25
134
249
18
1.0
0.5
3.5
3.5
3.5
5.0
5.0
0.20
1.25
0.80
7
4
4
75%
91%
89%
89%
92%
87%
86%
b
a
b
0.25
a
e
b
3.0
1.5
2.0
5
a
e
b
5.0
13
10
24
d
0.125
0.750
1.2
7.5
8.0
10.0
0.34
e
c,d
6
110
1.5
a
Because KRED117 was less stable under these reaction conditions it was added in batches (usually every 1–2 h) giving the combined final
concentration shown in the table.
b
c
>
98% ee was observed in the product of these reactions in agreement with the small-scale reactions (20 mM, Table 2).
The isolated product contained a small amount of starting ketone due to incomplete reaction. The reported yield only refers to the alcohol produced
and isolated.
The major enantiomer was R (92% ee).
d
e
1
mM NADPH was added at the beginning of each reaction, and at later time points NADP+ was added giving the final concentrations of cofactor
shown in the table.
enzyme at the beginning of the reaction was sufficient to
complete the reaction in 4 h. The ketoreductase concen-
trations shown in Table 3 represent the total amount of
enzyme that was added during the course of the reaction.
Even when the specific activity of a KRED towards a
substrate was relatively low, as in the case of KRED111
towards benzyl-hydroxyacetone (Table 3, entry 1), high
yields of conversion to product were achieved by increas-
ing the enzyme concentration and the reaction time. Such
an enzymatic process might even be economical if the
enzyme is active over long reaction times. It can be
produced on large scale by fermentation of recombinant
bacteria, especially if the substrate that it is reducing is a
high value intermediate.

3686
I. A. Kaluzna et al. / Tetrahedron: Asymmetry 16 (2005) 3682–3689
High titers of substrate can be effectively reduced using
catalytic amounts of enzyme and cofactor even when
ketones with low aqueous solubility are utilized as
shown by the results in Table 3. The ketone does not
need to be completely dissolved in the reaction mixture
for the reaction to reach completion. Both reactions
with a-tetralone (Table 3, entries 6 and 7) and the high
titer reactions with benzoyl-hydroxyacetone (Table 3,
entries 4 and 5) gave heterogeneous mixtures after the
addition of each substrate to the enzyme aqueous med-
ium. However, a small increase of the co-solvent added
and the KRED concentration were necessary to com-
plete these high substrate titer reactions in a timely fash-
ion. In the case of benzoyl-hydroxyacetone, 1% (g/g
relative to the substrate) of KRED117 was enough to
complete the reaction containing 0.14 M of substrate
in 4 h (Table 3, entry 3). A small increase of the enzyme
loading to 2% (g/g relative to the substrate) was only
required to complete the same reduction when the sub-
strate concentration increased 5- or 10-fold (Table 3,
entries 4 and 5, respectively) albeit a longer reaction time
was required for the reaction containing 1.4 M of
substrate (Table 3, entry 5). In the case of the more
hydrophobic a-tetralone, the same relative concentra-
tion of KRED101 (6.7% g/g relative to ketone) com-
pleted both the low and high titer reactions (Table 3,
entries 6 and 7) requiring, as before, longer incubation
time in the high titer reaction. The co-solvent concentra-
tion, which is required in every case has to be carefully
examined since it can affect the stability of some
KREDs. A concentration of 10% (v/v) of DMSO is, in
most cases, enough to partially dissolve any ketone
in the aqueous enzymatic media and facilitate its reduc-
tion without significantly affecting the stability and
catalytic activity of most KREDs. Higher concentra-
tions of this (or other water miscible) co-solvent can
be utilized; however, the stability of the desired KRED
in each case mustbe evalua te d.
the purity of the starting material. If larger quantities
of the product are required, the reaction methods
described herein can be directly scaled-up using stirred
tanks with mixing and pH and temperature control.
3. Conclusion
The substrate range and stereoselectivity of 31 cloned,
overexpressed and partially purified ketoreductases were
investigated in this report. With a few exceptions, most
enzymes showed a broad substrate range (Table 1)
9
reducing a- and b-ketoesters 1–4 and 9, an a-ketoacid
8, hydroxy acetophenone 5 as well as various alkyl-
substituted acetophenones 10 aliphatic ketone 7, a
benzyl-substituted hydroxy acetone 10 and a bicyclic
ketone such as a-tetralone 6. Using two structurally
similar ketones, ethyl-3-oxobutyrate and benzoyl-
hydroxyacetone, the stereoselectivity of reduction for
every enzyme was identified. Both the (R)- and (S)-enan-
tiomers were produced in high enantiomeric excess
(>99%) by one or more enzymes.
The potential for its use in preparative-scale (5–25 g)
reactions was also investigated for two substrates.
We showed that large titers of product can be effectively
reduced even when hydrophobic substrates are utilized,
and that it is not necessary for a ketone substrate to be
completely dissolved in order for its reduction to reach
completion (Table 3). In the heterogeneous mixtures
that were produced when high titers of ketone substrate
were prepared (0.75–1.40 M), complete product forma-
tion was achieved by slightly increasing the amount of
DMSO co-solvent and the reaction time when compared
with the reductions on lower substrate concentrations
(0.125–0.14 M). Depending on the reaction, enzymatic
amounts of 1–7% (w/w relative to substrate) were
enough to complete the reactions within 4–24 h. Product
isolation was accomplished in a straightforward fashion
by organic extraction of the aqueous reaction mixture.
The required NADPH cofactor was added in catalytic
amounts (0.1–0.2% equiv relative to substrate) and
was efficiently recycled using glucose and glucose
dehydrogenase (GDH). Considering the low cost of
D-glucose and the fact that GDH has been cloned,
over-expressed and can be produced in large quantities,
the additional cost of the cofactor recycling to the over-
all process is low.
In all the reactions presented in Table 3, NADPH was
added in catalytic amounts and was efficiently recycled
using D-glucose and glucose dehydrogenase (GDH), an
enzyme that regenerates NADPH from NADP+ with
the concomitant oxidation of D-glucose. In the high titer
reductions (0.75 M–1.4 M, entries 4, 5 and 7 Table 3),
the concentration of NADPH (added as NADP+ which
was immediately converted to NADPH with the glucose
DH/glucose system) was kept at 1.5–2.0 mM (or 0.14–
0
.2% equiv relative to substrate), which was enough to
give quantitative yields using the NADPH regeneration
system.
The results presented herein clearly show that isolated
cofactor-dependent ketoreductases are important and
versatile catalysts for the production of chiral alcohols.
Besides the ketoreductases that were investigated herein,
many more similar enzymes have been cloned and char-
acterized; this number will increase in the following
years. For example, our group has recently cloned and
characterized another set of more than 30 ketoreducta-
ses with wide substrate and stereoselectivity range. In
this report, we sought to demonstrate the ease of using
and handling ketoreductase enzymes as catalysts for
organic synthesis and to show that any organic chemist
with minimal training in their handling can benefit from
their tremendous potential.
The ease of using KREDs for stereoselective ketone
reduction is another important feature. The reactions
described herein were performed using equipmentavail-
able in any organic chemistry lab. The enzymes were
added in the reaction media as lyophilized powders,
and the substrates were dissolved in DMSO prior to
their addition to the reaction media. Product isola-
tion involved an aqueous organic extraction where all
enzymes, cofactors and buffers remained in the aqueous
layer and only the organic products were extracted. The
productpur iyt was very high and depended only on
3
,4

I. A. Kaluzna et al. / Tetrahedron: Asymmetry 16 (2005) 3682–3689
3687
4
. Experimental
TMSOTf14 prior to injection. A temperature gradient
of 95–130 ꢁC (1 ꢁC/min) was used to separate the two
acetylated enantiomers of ethyl-3-hydroxybutanoate
and a gradientof 145–165 ꢁC (1 ꢁC/min) for the acety-
lated enantiomers 1-benzoyl-2-acetyl-1,2 dihydroxy-
propane. The assignments of each enantiomer for
ethyl-hydroxybutanoate were based on pure enantio-
mers purchased by Aldrich. The (R)- and (S)-assign-
ments for 1-benzoyl-1,2 dihydroxypropane were based
on the literature reports where the (R)-enantiomer was
synthesized using whole cell reactions of Saccharomyces
cerivisiae. We utilized these published methods to syn-
thesize each enantiomer and used it as a standard. The
enantiomeric alcohols of the reduction of a-tetralone
were separated using Chiral HPLC analysis utilized
(S,S)-Whelk-01 column (Regis Technologies) elution
conditions (1 mL/min, hexane/isopropanol, 99:1, v/v).
No derivatization of the alcohol was required. The
absolute configuration was determined using enantio-
merically pure (R)-enantiomer and racemic 1,2,3,4 tetra-
hydro-1-naphthol both purchased by Aldrich.
4
(
.1. General method for the spectrophotometric assays
Table 1 data)
The specific activities in units/mg (1 unit/mg is the con-
version of 1 lmol substrate/mg enzyme · min ) were
À1
calculated by cuvette assays by measuring the rate of
NADPH consumption from the decrease of the absor-
bance at340 nm. In a yt pical assay, 940 lL of a potas-
sium phosphate buffer (Kpi, 50 mM, pH 6.5) was
mixed with 10 lL of an NADPH solution (giving an
absorbance between 1.2 and 1.5 absorbance units),
1
5
3
0 lL DMSO containing every ketone substrate (0.5 M
substrate concentration in DMSO) and 20 lL of every
enzyme solution that was made by dissolving 10 mg of
every enzyme lyophilized powder in 1 mL of potassium
phosphate buffer (50 mM pH 7.0). In many assays, dilu-
tions of 1/10 to 1/100 were required in order to obtain
good absorbance slopes.
Essentially, the same assays can be (and were) per-
formed in 96-well plates by mixing 150 lL of a reaction
buffer (containing 50 mM Kpi, 3.5% v/v DMSO,
All large-scale reaction products were analyzed by GC
H NMR and C NMR:
1
13
2
–10 mM substrate and NADPH) with 5 lL of enzyme
lysate. Simultaneous measurement of all the slopes can
be measured using a UV-plate reader (SpectraMax Plus,
Molecular Devices Inc). Using this method, 96 assays
can be performed at the same time allowing for a rapid
identification of active enzymes towards a substrate as
well as the comparison and the simultaneous calculation
of enzymatic activities.
(i) 1-Benzoyl-1,2-dihydroxy-propane:
1
H NMR (400 MHz, CD Cl ): d = 1.24 (d, J = 6.4 Hz,
2 2
3H), 4.14 (m, 2H), 4.27 (m, 1H), 7.44+7.57+8.02
1
3
(m+m+m, 5H).
C NMR (100 MHz, CD
Cl
Cl ), 66.4, 70.4, 128.8, 129.8,
2
):
2
2
d = 19.5, 53.8 (m, CD
2
130.4, 133.4, 166.8.
(
ii) 1,2,3,4-Tetrahydro-1-naphthol (reduction product
4.2. Small-scale reductions of benzoyl-hydroxyacetone
and ethyl-3-oxobutanoate (Table 2 data)
of a-tetralone):
1
H NMR (400 MHz, CD Cl ): 1.75+1.86–2.0 (m+m,
2
2
In test tubes carrying a stirring bar, 5 mL of phosphate
buffer (200 mM Kpi pH 6.9) 2.5 mM NADPH and
4
7
H), 2.72+2.80 (m+m, 2H), 4.76 (t, broad 1H),
.1+7.2+7.42 (m+m+m, 4H). C NMR (100 MHz,
13
5
0 mM D-glucose were added, followed by the addition
CD Cl ): 18.8, 29.2, 23.3, 53.8 (q, CD Cl ), 68.1,
1
2
2
2
2
of DMSO-substrate solution giving a final concentration
of 4% v/v DMSO and 20 mM ketone substrate. In this
mixture, 2 mg glucose dehydrogenase (GDH) and
26.1, 127.5, 128.6, 129.0, 130.1, 138.8.
2
–5 mg ketoreductase were dissolved and the reactions
4.3.1. Synthesis of (R)-(1-benzyl)-1,2-dihydroxy propane
using KRED107
were left si tr ring at34
was isolated by extraction of the crude reaction mixtures
with EtOAc (3 mL). After drying with Na SO and
evaporation of the organic solvent under reduced pres-
sure, the oily product was dissolved in CH Cl /Ac O/
TMSOTf solution in order to acetylate the alcohol.
The crude mixture was then analyzed by chiral GC chro-
matography as described in the following paragraphs.
ꢁC for 12–16 h. The product
4.3.1.1. KRED107 25 g (0.14 M) substrate, 1.25 mM
NADPH (entry 2 Table 3). In 950 mL of a potassium
phosphate buffer (Kpi, 200 mM, pH 6.9), a mixture of
DMSO-substrate (35 mL DMSO and 25 g substrate)
was added giving a final concentration of 3.5% v/v
DMSO and 140 mM substrate. In this solution, 45 g
glucose (250 mM), 1.10 g NADPH (1.25 mM), 0.25 g
glucose dehydrogenase (GDH 102) and 0.5 g KRED107
were dissolved. The reaction was stirred at 34 ꢁC using a
pH-stat that was keeping the pH at 6.85–6.90 with the
addition of NaOH (3 M solution, ꢀ50 mL added during
the reaction).
2
4
2
2
2
1
4
4
.3. Large-scale reduction of benzoyl-hydroxyacetone and
a-tetralone and product identification (Table 3 data)
Every reduction described in Table 3 is discussed in
detail in the following paragraphs. Product analysis
was based on chiral GC analysis for ethyl-3-hydroxy-
butanoate and 1-benzoyl-1,2 dihydroxypropane, and
chiral HPLC analysis for the reduction products of
a-tetralone. Chiral GC utilized Chirasil Dex CB column
The reaction was followed by TLC and GC analysis.
After 2 h, GC analysis showed 80% conversion and after
4 h traces of starting material were detected. At this point,
the reaction mixture was extracted with EtOAc (200 mL,
3·). The combined organic extracts were back-extracted
with brine, dried over Na SO , and evaporated to dryness
(
P7502, Varian Inc.). Both alcohols were first derivi-
tized to the corresponding acetyl-esters using Ac O/
2
2
4

3688
I. A. Kaluzna et al. / Tetrahedron: Asymmetry 16 (2005) 3682–3689
giving 24 g of a cream-colored viscous liquid that was
crystallized after overnight incubation at 4 ꢁC.
50 mg KRED117 were added to the medium. The heter-
ogeneous mixture was stirred at 34 ꢁC using a pH-stat
that was used to keep the pH at 6.9 by the addition of
3 M NaOH. The reaction progress was monitored by
GC analysis of crude samples every one hour. Product
formation stopped after the first hour since a yield of
40% was measured at the 1 h and the 2 h time points
due to enzyme inactivation as shown earlier. From this
pointon, KRED117 was added every hour: 30 mg at
2 h (40% yield), 20 mg at3 h (70% yield) and 20 mg at
GC analysis of this oil revealed 96% of alcohol present
in the extracted product along with traces of the (S)-
enantiomer and 4% of another unidentified product.
No starting ketone was detected.
Final isolated yield of pure alcohol: 91%; Enantiomeric
excess: >98% ee.
4
h (87% yield) bringing the total amount of KR117
4
.3.2. Synthesis of (S)-(1-benzyl)-1,2-dihydroxy propane
using KRED117 and KRED111
.3.2.1. KRED117 25 g (0.14 M) substrate 0.8 mM
added to 0.12 g. At the 3 h time point, another 20 mg
of GDH and 20 mg NADP+ (0.5 mM) were added to
the reaction. These additions kept the reaction rate at
an almost constant rate giving a 98% yield after 5 h of
reaction time.
4
NADPH (Table 3, entry 3). A reaction similar to the
one described earlier is prepared as follows:
In 950 mL Kpi 200 mM pH 6.9 add DMSO-substrate:
At this point, the reaction mixture was extracted with
EtOAc (50 mL, 2·), the combined organic extracts were
back extracted with brine, dried using sodium sulfate
and evaporated to dryness giving 4.8 g of a clear oil con-
taining >98% of the alcohol and traces of starting
ketone. The total isolated yield was therefore calculated
to be 89%.
3
5 mL–25 g (3.5% v/v DMSO, 140 mM substrate), glu-
cose: 45 g (250 mM), NADPH: 0.7 g (0.8 mM), GDH:
.2 g and KRED117: 0.15 g. The reaction was stirred
0
at34 ꢁC with a pH-stat that kept the pH at 6.8–6.9 by
the addition of 3 M NaOH.
After 3 h, TLC analysis of a reaction aliquot showed
4.3.2.2.2. 1.4 M benzoyl hydroxyacetone.The reaction
mixture was prepared exactly the same way as described
earlier. The only difference was the amount of benzoyl-
hydroxyacetone where 10 g were dissolved in 2 mL
DMSO prior to the addition in the reaction mixture
(38 mL) containing GDH (50 mg), KRED117 (50 mg),
NADPH (35 mg, 1 mM) and D-glucose (14.4 g, 2 M).
Additional KRED117 was added at increment amounts
as follows. 1 h: 20 mg (20% yield), 2 h: 20 mg (32%
yield), 3 h: 20 mg (38% yield), 4 h: 20 mg (47% yield),
5 h: 15 mg, 6 h:15 mg (61% yield), 8 h: 20 mg (72%
yield), 10 h: 20 mg. The reaction stopped at 13 h where
>99% productyield was de et c et d by GC analysis of a
crude reaction extract. In addition to KRED117 glucose
dehydrogenase (GDH) and NADP+ were added at3 h
(20 mg GDH, 0.5 mM NADP+) and at6 h (15 mg
GDH and 0.5 mM NADP+), bringing the total concen-
trations of enzymes and cofactor to 0.2 g for KRED117
ꢀ
50% product formation. However, at this point no
decrease of the pH was observed over time (ꢀ10–
5 min) and no more productforma it on was observed
1
by GC analysis. This meant that the reaction had
stopped, probably because of KRED117 inactivation
or GDH inactivation or NADPH decomposition.
GDH and NADPH should be stable over at least 4 h
since no problem was observed in the reduction using
KRED107. As a result, another 0.1 g of KRED117 were
added to the reaction (bringing the total to 0.25 g)
resulting in an immediate decrease of pH from 6.9 to
6
.8 within 5 min. After 1 h from the second KRED117
addition (4 h from the beginning of the reaction) analy-
sis of an aliquot showed complete conversion to the
alcohol. It is clear therefore that KRED117 is not stable
for more than 3 h under the reaction conditions and
should be added at incremental amounts.
(or 5 g/L), 0.085 g for GDH (or 2.12 g/L) and 2 mM for
At this point, the reaction mixture was extracted with
EtOAc and the product was isolated using the method
described earlier for the isolation of the (R)-enantiomer.
A cream viscous liquid (23 g) was isolated. GC analysis
showed 98% of the (S)-enantiomer, traces of the (R)-
enantiomer and 2% of an unidentified product. No start-
ing ketone was detected.
the cofactor (1 mM NADPH and 1 mM NADP+). The
final product (9.2 g) was isolated by EtOAc extraction as
described earlier giving a clear oil containing only the
alcohol productat92% isola te d yield.
4.3.2.3. KRED111 10 g (0.14 M) substrate 0.20 mM
NADPH (Table 1, entry 1). The reaction mixture
was prepared as described earlier for the reduction of
10 g with KRED107. The only difference was that
0.5 g of enzyme KRED111 and 0.2 mM NADPH were
used to reduce 10 g of benzyl-hydroxy acetone
(140 mM). The reaction mixture was stirred at 34 ꢁC
and the progress was followed using GC analysis of
crude mixtures: 2 h—31% yield; 3 h—47% yield; 4 h—
55% yield; 7 h—94 %.
Final isolated yield of pure alcohol: 89% Optical purity:
>98% ee.
4
.3.2.2. KRED117 5.3 g (0.75 M) and 10 g (1.4 M)
substrate 1.5 mM and 2.0 mM NADPH, respectively
Table 3, entries 4 and 5)
.3.2.2.1. 0.75 M benzoyl-hydroxyacetone. The con-
(
4
centrated reactions were prepared as follows. In 38 mL
of potassium phosphate buffer (Kpi, 0.25 M), 7.2 g D-
glucose (1 M), 5.3 g (0.03 mol, 0.75 M) of benzoyl-
hydroxyacetone that was dissolved in 2 mL of DMSO
After 7 h of reaction, the reaction mixture was extracted
twice with EtOAc, combined organic extracts washed
with brine, and evaporated to dryness giving 8.0 g of a
clear oily product that contained 6% of the starting
(
5% v/v), 35 mg NADPH (1 mM), 50 mg GDH and

I. A. Kaluzna et al. / Tetrahedron: Asymmetry 16 (2005) 3682–3689
3689
ketone and 94% of the alcohol product. Final yield of
alcohol: 75%.
Catalysis in Organic Synthesis, 1sted.; John Wiley &
Sons: New York, 1994.
. (a) Noyori, R.; Kitamura, M.; Ohkuma, T. Proc. Natl.
Acad. Sci. U.S.A. 2004, 101, 5356; (b) Noyori, R. Science
2
1
990, 248, 1194; (c) Sandoval, C. A.; Ohkuma, T.; Muniz,
4
.3.3. Large-scale reduction of a-tetralone
.3.3.1. KRED101 9.1 g (0.125 M) substrate 0.34 M
K.; Noyori, R. J. Am. Chem. Soc. 2003, 125, 13490; (d)
Hoffman, R. V.; Maslouh, N.; Cervantes-Lee, F. J. Org.
Chem. 2002, 67, 1045; (e) Corey, E. J.; Link, J. O.
Tetrahedron Lett. 1989, 30, 6275.
4
NADPH (Table 3, entry 6). In 460 mL of a potassium
phosphate buffered solution (200 mM, pH 6.9), 9.1 g of
a-tetralone dissolved in 40 mL of DMSO were added
giving final concentrations of 0.125 M for tetralone
and 8.0% (v/v) for DMSO. In this mixture, 22.5 g of glu-
cose (0.25 M final concentration), 100 mg NADPH
3
. (a) M u¨ ller, M.; Wolberg, M.; Schubert, T.; Hummel, W.
Adv. Biochem. Eng. Biotechnol. 2005, 92, 261; (b) Stewart,
J. D.; Rodriquez, S.; Kayser, M. M. In Enzyme Techno-
logy forPha rm aceutical and Biotechnological Applications ;
Kirst, H. A., Yeh, W.-K., Zmijewski, M. J., Eds.; Marcel
Dekker: New York, 2001; pp 175–207; (c) Kaluzna, I. A.;
Andrew, A. A.; Bonnilla, M.; Martzen, M. R.; Stewart, J.
J. Mol. Catal B: Enzymatic 2002, 17, 101; (d) Katz, M.;
Hahn-Hagerdal, B.; Gorwa-Grauslund, M. F. Enz. Mic-
rob. Technol. 2003, 33, 163; (e) Kaluzna, I. A.; Feske, B.
D.; Wittayanan, W.; Ghiviriga, I.; Stewart, J. D. J. Org.
Chem. 2005, 70, 342; (f) Kaluzna, I. A.; Matsuda, T.;
Sewall, A. K.; Stewart, J. D. J. Am. Chem. Soc. 2004, 126,
(
0.11 mM), 0.25 g of glucose dehydrogenase (GDH)
and 0.35 g of KRED101 were added. The reaction was
stirred at 34 ꢁC using a pH-stat that kept the pH at
6
.9 with the addition of 3 M NaOH solution. After 4 h
and while TLC analysis showed about50% conversion,
another 250 mg of KRED101, 50 mg of GDH and
5
0 mg of NADPH were added to the reaction mixture.
The reaction progress was followed by TLC and after
0 h only product was detected by TLC analysis. At this
1
1
2827–12832; (g) Wolberg, M.; Kaluzna, I. A.; M u¨ ller,
point, the reaction mixture was extracted twice with
M.; Stewart, J. D. Tetrahedron: Asymmetry 2004, 15,
2825–2828; (h) Habrych, M.; Rodriguez, S.; Stewart, J. D.
Biotechnol. Prog. 2002, 18, 257–261; (i) Hanson, R. L.;
Goldberg, S.; Goswami, A.; Tylly, T. P.; Patel, R. N. Adv.
Synth. Catal. 2005, 347, 1073.
. (a) Johanson, T.; Katz, M.; Gorwa-Grauslund, M. FEMS
Yeast Res. 2005, 5, 513; (b) Kataoka, M.; Kita, K.; Wada,
M.; Yasohara, Y.; Hasegawa, J.; Shimizu, S. Appl.
Microbiol. Biotechnol. 2003, 62, 437.
EtOAc (180 mL each time), dried over MgSO and the
4
solvent was evaporated to dryness giving 7.9 g of a light
orange oil in 87% isolated yield. NMR analysis con-
firmed the product structure and purity along with
HPLC analysis. Chiral HPLC using (S,S)-Whelk-01 col-
umn separated the two enantiomers giving the (R)-iso-
mer as the major product (92% ee).
4
5
. KRED 30000 (BioCatalytics Inc., Pasadena CA.
www.biocatalytics.com).
4
.3.3.2. KRED101 4.4 g (0.75 M) substrate 1.5 M
NADPH (Table 3, entry 7). The reaction mixture
40 mL total volume, 0.25 M Kpi) was prepared as
6
. For recentreview see: (a) Garcia-Urdiales, E.; Alfonso, I.;
Gotoer, V. Chem. Rev. 2005, 105, 313; (b) Homann, M. J.;
Vail, R. B.; Previte, E.; Tamarez, M.; Morgan, B.; Dodds,
D. R.; Zaks, A. Tetrahedron 2004, 60, 789; (c) Wolfgang,
K.; Mang, H.; Edegger, K.; Faber, K. Adv. Synth. Catal.
2004, 346, 125; (d) Nakamura, K.; Yamanaka, R.;
Matsuda, T.; Harada, T. Tetrahedron: Asymmetry 2003,
(
described earlier for the synthesis of benzoyl-hydroxyac-
etone under the same substrate concentration. The only
difference was the initial concentration of KRED101,
which was 0.2 g, while the rest of the reactants were
1
4, 2659; (e) Ishihara, K.; Yamaguchi, H.; Nakajima, N. J.
0
.75 M or 4.4 g for a-tetralone dissolved in 4 mL
Mol Catal. B: Enzym. 2003, 23, 171.
DMSO, 1 M or 7.2 g for D-glucose, 1 mM or 35 mg
for NADPH and 50 mg for GDH. The reaction mixture
was stirred at 34 ꢁC using a pH-stat to keep the pH at
7
8
. Peters, J. In Biotechnology; Rehm, H.-J., Reed, G., Puhler,
A., Stadler, P., Eds.; Biotransformations I; Kelly, D. R.,
Ed.; Wiley-VCH GmbH: Weinheim, 1998; Vol. 8a, pp
6
.9 and after 8 h of reaction an additional 25 mg of
3
91–474.
NADP+, 0.1 g KERD101 and 25 mg GDH were added
to the heterogeneous mixture. The reaction was almost
complete after 24 h (>95% yield). At this point, the mix-
ture was extracted with EtOAc (50 mL 2·), the com-
bined organic layers were back extracted with brine,
dried using Na SO and evaporated to dryness giving
. Kalaitzakis, D.; Rozzell, J. D.; Kambourakis, S.; Smonou,
I. Org. Lett., in press.
9. Zhu, D.; Mukherjee, C.; Rozzell, J. D.; Kambourakis, S.;
Hua, L., Tetrahedron, in press.
10. (a) Zhu, D.; Rios, B. E.; Rozzell, J. D.; Hua, L.
Tetrahedron: Asymmetry 2005, 16, 1541; (b) Zhu, D.;
Mukherjee, C.; Hua, L. Tetrahedron: Asymmetry 2005, 16,
2
4
3
.97 g as an oily compound. TLC and HPLC analysis
3
275.
1. (a) Kambourakis, S.; Rozzell, J. D. Adv. Synth. Catal.
003, 345, 699; (b) Kambourakis, S.; Rozzell, J. D.
revealed a small amountof residual ke to ne (<5%) giving
a final isolated yield of 89% for the alcohol. Chiral
HPLC analysis showed formation of the same enantio-
meric mixture as before.
1
2
Tetrahedron 2004, 60, 663.
1
1
2. Wei, Z.-L.; Lin, G.-Q.; Li, Z.-Y. Bioorg. Med. Chem. 2000,
8
, 1129.
3. (a) PTC Patent WO 2004/087124 A1.; (b) PTC Patent WO
004/094384 A2.
2
References
14. Procopiou, P. A.; Baugh, S. P. D.; Flack, S. S.; Inglis, G.
G. A. Chem. Commun. 1996, 2625.
1
. (a) Wills, M.; Hannedouche, J. Curr. Opin. Drug Discuss.
Devel. 2002, 5, 881–891; (b) Noyori, R. Asymmetric
15. Kometani, T.; Hisae, T.; Daikaiji, Y.; Michimasa, G. J.
Biosci. Bioeng. 2001, 91, 525.