Purification and cloning of a ketoreductase used for the preparation of chiral alcohols

Article abstract of DOI:10.1002/adsc.200505045

The synthesis of the leading candidate compound in an anticancer program required (S)-2-chloro-1-(3-chlorophenyl)-ethanol as an intermediate. Other possible candidate compounds used analogues of the S-alcohol. Of 119 microbial cultures screened for reduction of the corresponding ketone to the S-alcohol, Hansenula polymorpha ATCC 58401 (73.8% ee) and Rhodococcus globerulus ATCC 21505 (71.8% ee) had the highest enantioselectivity for producing the desired alcohol. A ketoreductase from Hansenula polymorpha, after purification to homogeneity, gave the S-alcohol with 100% ee. Amino acid sequences from the purified enzyme were used to design PCR primers for cloning the ketoreductase. The cloned ketoreductase required NADP(H), had a subunit molecular weight of 29,220 and a native molecular weight of 88,000. The cloned ketoreductase was expressed in E. coli together with a cloned glucose 6-phosphate dehydrogenase from Saccharomyces cerevisiae to allow regeneration of the NADPH required by the ketoreductase. An extract of E. coli containing the two recombinant enzymes was used to reduce 2-chloro-1-(3-chloro-4-fluorophenyl)-ethanone and two related ketones to the corresponding S-alcohols. Intact E. coli cells provided with glucose were used to prepare (S)-2-chloro-1-(3-chloro-4-fluorophenyl)-ethanol in 89% yield with 100% ee.

Full text of DOI:10.1002/adsc.200505045

FULL PAPERS
Purification and Cloning of a Ketoreductase used for the
Preparation of Chiral Alcohols
Ronald L. Hanson,* Steven Goldberg, Animesh Goswami, Thomas P. Tully,
Ramesh N. Patel
Process Research and Development, Bristol-Myers Squibb, One Squibb Drive, New Brunswick, NJ 08903, USA
Fax: (þ1)-732-227-3994, e-mail: ronald.hanson@bms.com
Received: January 24, 2005; Accepted: March 30, 2005
Abstract: The synthesis of the leading candidate com- lecular weight of 88,000. The cloned ketoreductase
pound in an anticancer program required (S)-2- was expressed in E. coli together with a cloned glu-
chloro-1-(3-chlorophenyl)-ethanol as an intermediate. cose 6-phosphate dehydrogenase from Saccharomyces
Other possible candidate compounds used analogues cerevisiae to allow regeneration of the NADPH re-
of the S-alcohol. Of 119 microbial cultures screened quired by the ketoreductase. An extract of E. coli
for reduction of the corresponding ketone to the S-al- containing the two recombinant enzymes was used
cohol, Hansenula polymorpha ATCC 58401 (73.8% to reduce 2-chloro-1-(3-chloro-4-fluorophenyl)-etha-
ee) and Rhodococcus globerulus ATCC 21505 none and two related ketones to the corresponding
(71.8% ee) had the highest enantioselectivity for pro- S-alcohols. Intact E. coli cells provided with glucose
ducing the desired alcohol. A ketoreductase from were used to prepare (S)-2-chloro-1-(3-chloro-4-fluo-
Hansenula polymorpha, after purification to homoge- rophenyl)-ethanol in 89% yield with 100% ee.
neity, gave the S-alcohol with 100% ee. Amino acid
sequences from the purified enzyme were used to de-
sign PCR primers for cloning the ketoreductase. The Keywords: alcohols; enzyme catalysis; gene expres-
cloned ketoreductase required NADP(H), had a sub- sion; Hansenula polymorpha; ketoreductase; NADPH
unit molecular weight of 29,220 and a native mo- regeneration
Introduction
Chiral alcohols are frequently required as synthons for
drugs or drug candidate compounds. Preparation by en-
zymatic resolution has been often described, but unless a
dynamic resolution process is developed, this approach
is limited to a maximum 50% yield. Reduction of ke-
tones to chiral alcohols by dehydrogenases has also
been widely used,[1] and has the advantage of possible
100% yields. Ketoreductases are found in many micro-
bial strains and can give either enantiomer of the alcohol
but whole cells may not give high ee because of compet-
ing enzymes with lower or opposite enantioselectivity.
Usually purification, cloning and overexpression of the
desired ketoreductase is required to achieve high ee
and productivity. The synthesis of the leading candidate
compound in an anticancer program required (S)-2-
chloro-1-(3-chlorophenyl)-ethanol (2 in Scheme 1) as
an intermediate. Other possible candidate compounds
used analogues of the S-alcohol. We report here on the
identification of strains which reduce the precursor ke-
tone to the desired S-alcohol and on the purification,
cloning and overexpression of the corresponding keto-
reductase.
Scheme 1. Ketones reduced by cloned ketoreductase.
Adv. Synth. Catal. 2005, 347, 1073–1080
DOI: 10.1002/adsc.200505045
ꢀ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1073

Ronald L. Hanson et al.
FULL PAPERS
Cofactor regeneration is a requirement to reduce cost the highest ee. Almost all strains screened reduced the
and allow the use of a physiological concentration of the ketone to the alcohol, and even excess glucose dehydro-
NAD(P)H cofactor. Yeast formate dehydrogenase is genase reduced the ketone to the alcohol. The coupled
commonly used for NADH regeneration, and glucose enzyme assay and preparative procedures use glucose
dehydrogenase from Bacillus species may be used for ei- 6-phosphate dehydrogenase to regenerate the NADPH
ther NADH or NADPH regeneration. Mutated formate required for the reduction to avoid the ketone reduction
dehydrogenases from Pseudomonas[2] and Saccharomy- by glucose dehydrogenase (which decreased the ee).
ces cerevisiae[3] that are more active with NADPH have
also been described. It is advantageous to clone and ex-
press both the ketoreductase and cofactor regeneration Enzyme Purification and Sequencing
enzymes in the same cell. Auseful system for production
of chiral alcohols has been developed by Kataoka et al.[4] Hansenula polymorpha ATCC 58401 cells (2.088 kg)
These workers expressed carbonyl reductases together were obtained from a 50-L fermentation. Using a whole
with glucose dehydrogenase from Bacillus megaterium cell assay with 1 mg/mL chloroketone, activity was high-
in E. coli to produce various chiral alcohols. Glucose est 37 h after the start of the fermentation (72% conver-
6-phosphate dehydrogenase can also be used for sion in 1 h with 80% ee). Activity when cells were har-
NADPH regeneration, and indeed that is one of the ma- vested after 48 h was 32% conversion in 1 h with 85%
jor physiological functions of the enzyme. The crystal- ee. Purification of the Hansenula ketoreductase was car-
line enzyme from Saccharomyces carlsbergensis was re- ried out to obtain amino acid sequences necessary for
ported to have a specific activity of 676 mmoles NADPH polymerase chain reaction (PCR) amplification of the
formed per minute per mg protein[5] which is considera- corresponding gene. During initial purification attempts
bly higher than other enzymes used for cofactor regener- the activity was not stable, despite the addition of yeast
ation. Synthesis of chiral alcohols using glucose 6-phos- protease inhibitor cocktail, 10% glycerol and 1 mM di-
phate dehydrogenase for NADPH regeneration has thiothreitol. Of various additives tried for stabilization,
been reported,[6,7] but because of the high cost of glucose only NADP and NADPH were effective (NAD was not
6-phosphate, the use of this method is usually limited to effective). Therefore 0.5 mM NADP was added to col-
small-scale reactions such as the production of drug me- umn buffers. The ketoreductase from Hansenula poly-
tabolites by P450 monooxygenases. Since glucose is morpha SC 13824 was purified 112-fold from an extract
transported into E. coli via a phosphotransferase system using Phenylsepharose, Q-Sepharose and Sephacryl S-
that uses phosphoenol pyruvate as a phosphoryl donor 100. FPLC with a unoQ column was used as a final
to convert glucose to glucose 6-phosphate,[8] we have step to obtain a single band on an SDS gel that was blot-
tried to exploit this system to use glucose for NADPH ted to PVDF for sequencing. A summary of the purifica-
regeneration by overexpressing glucose6-phosphate de- tion is shown in Table 1. Although the initial extract
hydrogenase from Saccharomyces cerevisiae in E. coli gave S-alcohol with 90% ee, after the Q-Sepharose chro-
along with the cloned ketoreductase.
matography step the enzyme preparation produced S-
alcohol with 100% ee. The purified protein required
NADP and was not active with NAD. The subunit mo-
lecular weight determined with an Agilent Bioanalyzer
2100 Protein 200 Labchip was 27 kd.
Results and Discussion
An N-terminal sequence (H2N-DKLPTEAPQLP)
was obtained which showed sequence homology with
sorbitol dehydrogenase from Candida albicans. Esti-
Screening
119 strains were screened for conversion of chloroke- mated protein concentration from the Agilent Bioana-
tone 1 to (S)- chloroalcohol 2. Most of the cultures lyzer 2100 indicated that 200 pmoles of protein per
gave predominantly the R-alcohol with the highest ee lane were provided for sequencing. However, the N-ter-
being 97.6%. Of 23 strains that produced the S-alcohol minal sequence contained only about 2.5 pmoles, indi-
Hansenula polymorpha ATCC 58401 (73.8% ee) and cating most of the N-terminal amino acid may be
Rhodococcus globerulus ATCC 21505 (71.8% ee) gave blocked. A tryptic digest gave several peaks which
Table 1. Purification of the ketoreductase from extract of Hansenula polymorpha ATCC 58401.
Step
Volume [mL]
Activity [units]
Protein [mg]
Specific activity [U/mg]
Extract
380
48
24
72.2
31.3
15.0
6.6
629.3
22.8
2.04
0.114
1.37
7.35
Phenyl Sepharose
Q Sepharose
Sephacryl S-100
24
0.516
12.79
1074
ꢀ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 1073–1080

Preparation of Chiral Alcohols
FULL PAPERS
contained an estimated 50–100 pmoles of peptide cluded NdeI and SmaI sites before the start and stop
and sequences of two of the internal peptides sites of the insert, respectively, were used to amplify
(AIMNTNLDGVYYCAK and SLAMEWVGFAR) the ketoreductase gene. The amplified fragment was li-
were obtained. These sequences also showed homology gated into pBMS2000 for expression in E. coli
with dehydrogenases in the data bases.
BL21Star(DE3) under the control of the tac promoter.
After growth of the cells and induction with 1 mM iso-
propyl b-d-thiogalactoside (IPTG), a sonicated extract
was prepared that contained a predominant 27 kd pro-
tein (Figure 2, lanes 4 and 9). In an activity assay, the ex-
Cloning
Degenerate primers for the N-terminal and internal se- tract gave complete conversion of chloroketone 1 to the
quences were used in PCR reactions to clone 400 and chloro-S-alcohol 2 with 100% ee. An extract of E. coli
600 base pair (bp) fragments of the ketoreductase BL21Star(DE3) cells lacking the plasmid produced
gene from Hansenula polymorpha DNA. The 400 bp only a trace of the chloroalcohol HPLC peak.
fragment was used to prepare a digoxigenin-labeled
probe which bound to a 4 kilobase (kb) band from a
XhoI digest of Hansenula polymorpha ATCC 58401 Molecular Weight of Native Ketoreductase
chromosomal DNA. A library was prepared in E. coli
containing DNA from the 4 kb region of the XhoI digest The molecular weight of the expressed ketoreductase in
ligated tothe pZErO-2 vector, and the probe was used to the extract was estimated as 88,094 by Superdex-200 gel
identify E. coli colonies carrying the homologous se- filtration chromatography. The peak fraction from the
quence. Sequencing of the plasmid prepared from one
of the hybridizing colonies showed that the insert con-
tained an open reading frame homologous to known ke-
toreductases and that the sequences corresponding to
the amino terminal and two internal peptides of the pu-
rified enzyme were present (Figure 1). Primers that in-
Figure 2. Analysis of expression of cloned enzymes in E. coli
extracts. Lanes1, 6, 2, 7: extract of strain with ketoreductase
(27 kd) and glucose 6-phosphate dehydrogenase (55 kd).
Lanes 1, 6 were induced with 50 mM IPTG. Lanes 2, 7 were
induced with 1 mM IPTG. Lanes 4, 9: extract of strain with
ketoreductase only induced with 1 mM IPTG. Lanes 5, 10:
ketoreductase peak from Superdex-200 chromatography.
Lanes 3, 8: extract of control E. coli (no plasmid). Extracts
Figure 1. Sequence of ketoreductase gene. Peptides from
amino acid sequencing are underlined in bold.
were from 10% w/v sonicates and were analyzed with an Agi-
lent 2100 Bioanalyzer Protein 200 Labchip.
Adv. Synth. Catal. 2005, 347, 1073–1080
asc.wiley-vch.de
ꢀ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1075

Ronald L. Hanson et al.
FULL PAPERS
column contained a single 27 kd subunit when analyzed of the recombinant E. coli, batches of 1, 2, and 20 g ke-
with a Protein 200 Labchip (Figure 2, lanes 5 and 10). tone 3 were reduced. The product alcohol had 100%
From the amino acid composition derived from the ee in each batch. Conditions for the 20 g batch were:
gene sequence, the subunit molecular weight is 29,220, 20 g/L ketone 3 (97 mM), 40 g/L glucose 6-phosphate
suggesting that the native enzyme is a trimer.
(142 mM), 191 mg/L NADP (0.25 mM), 0.1 M potassi-
um phosphate buffer, pH 7, 5% methanol and 5% w/v
E. coli as a microfluidized and centrifuged extract. The
yield of 4 (isolated, but not purified) was 91% with
100% ee. The amount of extract used contained
181,000 units of ketoreductase and 47,000 units of glu-
cose 6-phosphate dehydrogenase and was calculated to
Expression of Hansenula polymorpha Ketoreductase
and Saccharomyces cerevisiae Glucose 6-Phosphate
Dehydrogenase in E. coli
The cloned ketoreductase was copied from the plasmid be capable of reducing the ketone in 2.05 minutes. The
by PCR. SmaI and HindIII sites were added upstream of ketone was 96% depleted in 30 minutes, and when the
the promoter and a SmaI site was added at the 3’ end via reaction was stopped after 5.5 h, 0.5% ketone remained.
the primers. The ketoreductase gene was ligated at a Enzyme assays of samples from a 2-g batch indicated
SmaI site into pBMS2000-SCGD which contains a that both enzymes were inactive 45 minutes after the
cloned S. cerevisiae glucose 6-phosphate dehydrogenase start of the batch, probably from the addition of metha-
gene. Restriction digests with BamHI and NdeI were nol.
used to identify plasmids with the two genes in the
A procedure using intact cells of recombinant E. coli
same orientation, which was thought to be a desired sit- with glucose supplied as a precursor of glucose 6-phos-
uation for a higher transcription rate. A restriction site phate was developed. The final best conditions used
map of the construct is shown in Figure 3. One of the for batches of 2- and 15-g were: 20 mg/mL ketone 3
plasmids with the desired orientation was used to trans- (97 mM), 40 mg/mL glucose (222 mM), 0.153 mg/mL
form BL21Star(DE3) E. coli for expression. This strain NADP (0.2 mM), 0.1 M potassium phosphate buffer,
was designated E. coli SC16529. The cells were induced pH 7, and 10% w/v E. coli. The isolated yield of unpuri-
with either 50 mM or 1 mM IPTG. Expression of both fied 4 was 89% with 100% ee. When the ketone was add-
enzymes was 2- to 3-fold better at 1 mM IPTG (Figure 2, ed in methanol (5% final concentration) or when NADP
lanes 1, 6 versus lanes 2, 7). Sonicated extracts of the cells was omitted, reaction was slow and incomplete.
with either concentration of inducer completely con-
Initially, the product alcohol was isolated using ethyl
verted chloroketone 1 to chloroalcohol 2 in buffer con- acetate extraction, then MTBE extraction was used to
taining NADP and glucose 6-phosphate.
get less emulsion during extraction. Solid phase extrac-
tion using the resin XAD-16 as described in the Exper-
imental Section was a more suitable procedure and
gave good quality product in high yield.
Alternative Substrates
Two other ketones (3 and 5 in Scheme 1, intermediates
for potential drug candidates) were reduced equally as Conclusion
fast as 1 by the 1 mM IPTG-induced E. coli extract
and gave single peaks with the chiral HPLC method A ketoreductase from Hansenula polymorpha was puri-
used for 2. The alcohol 4 produced by the enzymatic re- fied, cloned and overexpressed in E. coli, and was used
duction was also shown to be the S-enantiomer by com- to prepare (S)-2-chloro-1-(3-chloro-4-fluorophenyl)-
parison with an authentic standard. No chiral marker ethanol in high yield and ee. Glucose 6-phosphate dehy-
was available for the product alcohol 6.
drogenase from Saccharomyces cereviseae was also ex-
pressed in the same strain to allow glucose to be used
in whole cells for NADPH regeneration. Further meta-
bolic engineering of the strain could be used to increase
the production of phosphoenol pyruvate needed for glu-
Preparative Batches of Alcohol 4
E. coli SC16529 was grown in a fermentor and the ex- cose phosphorylation and to enhance the remaining en-
pression of the enzymes was induced with 1 mM IPTG zymes needed for the phosphogluconate oxidative path-
as described in the Experimental Section to provide an way.
enzyme source for preparative batches. Using an extract
Figure 3. Ketoreductase (KR)/glucose-6-phosphate dehydrogenase (SCGD) construct.
1076
ꢀ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 1073–1080

Preparation of Chiral Alcohols
FULL PAPERS
resis and blotted onto a PVDF membrane. Sequencing of the
N-terminus and two internal peptides was done by Argo Bio-
analytica. Purity and molecular weight of the enzyme subunit
were determined with an Agilent Bioanalyzer 2100 using aPro-
tein 200 Labchip.
Experimental Section
Screening
Z49 medium contained 3% trypticase soy broth powder (BBL
brand). F7 medium contained 1% malt extract, 1% yeast ex-
tract, 0.1% peptone and 2% dextrose adjusted to pH 7. Cul-
tures were grown in 2 mL medium (16 strains on Z49, 103
strains on F7) in multiwell plates shaken at room temperature.
After 24 h, 20 mL of a 100 mg/mL solution of chloroketone 1 in
methanol were added to each well and the incubation was con-
tinued for 3 days. Samples were extracted with 2 mL ethyl ace-
tate per mL broth. The extract (1 mL) was dried at 408C under
a nitrogen stream and the residue was dissolved in 1 mL 2-
propanol for HPLC analysis of ketone and alcohol concentra-
tion and ee of S-alcohol.
Enzyme Assays
Prior to cloning the ketoreductase, enzyme assays were per-
formed using HPLC. To evaluate activities of whole cells dur-
ing fermentations, cells were pelleted from 0.5 mL broth with
a microfuge and resuspended in 0.5 mL 50 mM potassium
phosphate buffer pH 7 containing 10 mg/mL glucose and
0.5 mg 1 dissolved in 5 mL methanol. For cell extracts and col-
umn fractions, ketoreductase assays contained in 0.5 mL:
40 mM potassium phosphate buffer pH 7, 0.8 mM NADP,
8.5 mM sodium glucose 6-phosphate, 1.6 units glucose 6-phos-
phate dehydrogenase from bakerꢂs yeast, 0.5 mg 1 dissolved in
5 mL methanol and ketoreductase sample. Assays were carried
out in microfuge tubes at room temperature on a 12 rpm end-
over-end rotator. Reactions were terminated (typically after
30 min) by addition of 0.5 mL acetonitrile for quantitation by
reverse phase HPLC or by extraction with 1 mL ethyl acetate
for ee measurements by chiral HPLC. Better results were ob-
tained for chiral chromatography when the samples were dis-
solved in mobile phase instead of 2-propanol. Activities were
calculated after HPLC analysis as 1 mmole/min¼1 unit.
After cloning, spectrophotometric enzyme assays were used
to measure activity in extracts of recombinant E. coli. Ketore-
ductase assays contained in 1 mL: 0.1 M potassium phosphate
buffer pH 7, 0.3 mM NADPH, and 0.1 mg ketone 3 in 0.05 mL
methanol. The reaction was started with cell extract (diluted
into 50 mM potassium phosphate buffer, pH 7, containing
1 mM dithiothreitol). Absorbance decrease at 340 nm was
used to calculate the enzyme activity as 1 mmole/min¼1 unit.
Glucose 6-phosphate dehydrogenase assays contained in
1 mL: 0.1 M potassium phosphate buffer, pH 7, 0.3 mM
NADP and 0.5 mM sodium glucose 6-phosphate. The reaction
was started with cell extract (diluted into 50 mM potassium
phosphate buffer, pH 7, containing 1 mM dithiothreitol). Ab-
sorbance increase at 340 nm was used to calculate the enzyme
activity as 1 mmole/min¼1 unit.
Growth of Hansenula polymorpha ATCC 58401 and
Preparation of Cell-Free Extract
Note: Strains deposited as Hansenula polymorpha are now
called Pichia angusta by the American Type Culture Collec-
tion. Cells were grown in 50 liters F7 medium for 48 h starting
from a 0.5 liter inoculum. Airflow was 1 vvm, temperature was
288C, agitation was 500 rpm and SAG antifoam was used to
control foaming. Cell paste (2.088 kg) was recovered by centri-
fugation and stored at ꢀ708C. Cells (20% w/v) were suspend-
ed in 50 mM potassium phosphate buffer pH 7, containing
10% glycerol, 1 mM dithiothreitol, 1% yeast protease inhibitor
cocktail (Sigma) and disrupted by 10 passages through a micro-
fluidizer. Unbroken cells and debris were removed by centrifu-
gation of the extract at 20434ꢁg for 20 min. The extract was
stored at ꢀ208C until used for purification.
Purification of Ketoreductase
Extract (380 mL) was applied to 50 mL Amersham/Pharmacia
Phenyl Sepharose 6 fast flow in a 26 mm diameter column at
3 mL/min. The column was packed in 20 mm tris (chloride)
pH 7.4 containing 10% glycerol, 1 mM dithiothreitol and
0.5 mM NADP (buffer A). The column was washed at 1 mL/
min with 120 mL buffer A, then 300 mL buffer A containing
1% triton X-100 (Pierce Surfact-Ampsꢁ X-100). The 48 mL
peak of activity was applied to 30 mL Amersham/Pharmacia
Q Sepharose fast flow in a 16-mm diameter column packed
in buffer A and eluted with a 240-mL 0 to 0.2 M NaCl gradient
in buffer A at a flow rate of 2 mL/min. The activity peak frac-
tions containing 24 mL were concentrated to about 6 mL using
anAmicon YM10 membrane in a stirred cell. The concentrated
enzyme was passed at a flow rate of 1 mL/min through 450 mL
Amersham/Pharmacia Sephacryl S-100 packed in a 2.6ꢁ
100 cm column with 50 mM potassium phosphate buffer
pH 7 containing 10% glycerol, 1 mM dithiothreitol, 0.5 mM
NADP and 0.15 M NaCl. The peak two fractions containing
24 mL were concentrated and diafiltered into buffer A with
an Amicon Centricon-10, loaded onto a Biorad Uno Q1 col-
umn (1.3 mL), then eluted with a 16-mL 0 to 0.25 M NaCl gra-
dient in buffer A at a flow rate of 1 mL/min. The activity peak
fractions containing 1 mL were concentrated with an Amicon
Microcon-10 to 0.1 mL then subjected to SDS gel electropho-
HPLC Analysis
Quantitation was performed with a YMC Pak ODS A 3 m 15ꢁ
0.6 cm column. The mobile phase was a gradient of 20% aceto-
nitrile/80% water to 90% acetonitrile/10% water from 0 to
12 min, flow rate was 1 mL/min, column temperature was
408C, detection was with a DAD set at 220 nm and injection
volume was 5 mL. Retention times were: chloroalcohol 2
9.9 min, chloroketone 1 11.1 min.
Chiral separations were performed with a Chiralpak AD-H
25ꢁ0.46 cm column (Daicel Chemical Industries, Ltd.). The
mobile phase was 95% hexane/5% ethanol, flow rate was
1 mL/min, column temperature was 188C, detection was with
a DAD set at 220 nm and injection volume was 10 mL. Reten-
tion times were: S-enantiomer of 2 10.4 min, R-enantiomer of
2 13.1 min, chloroketone 11.0 min. Authentic chiral markers
Adv. Synth. Catal. 2005, 347, 1073–1080
asc.wiley-vch.de
ꢀ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1077

Ronald L. Hanson et al.
FULL PAPERS
were prepared by Bristol-Myers Squibb chemists by Corey re-
duction of the ketones.
(10ꢁ300 mm) and run at 0.5 mL/min using an AKTA FPLC.
The column buffer was 50 mM potassium phosphate pH 7 con-
taining 10% glycerol, 0.1 M NaCl and 1 mM dithiothreitol.
Protein molecular weight standards were aldolase (158000),
bovine serum albumin (67000), ovalbumin (43000), chymo-
trypsinogen A (25000) and ribonuclease (13700).
Cloning of the Ketoreductase Gene
Degenerate primers were prepared based on the N-terminal
(sense) and two internal amino acid sequences (antisense) of
the ketoreductase. They were used in PCR with chromosomal
DNA from Hansenula polymorpha SC 13824 and gave 400-bp
and 600-bp products. The 400-bp product was ligated to PCR4-
TOPO as recommended by the manufacturer (Invitrogen) and
transformed into TOP10 E. coli (Invitrogen) by electropora-
tion. The sequence of the insert was obtained from the Bris-
tol-Myers Squibb sequencing facility and showed homology
with short chain dehydrogenases/reductases. The 400-bp
PCR product was used as template to prepare a digoxigenin-la-
beled probe (Roche Biochemicals) to screen five restriction
enzyme digests of chromosomal DNA from Hansenula poly-
morpha ATCC 58401. All five digests contained a band that
bound the probe. The probe hybridized to a XhoI fragment
of about 4 kbp which was thought large enough to contain
the full-length ketoreductase gene. DNA from this region
was purified and inserted into the pZErO-2 vector (Invitro-
gen), then used to transform DH10B (Invitrogen) cells by elec-
troporation. Two colonies that contained the insert were iden-
tified by hybridization with the probe and confirmed to contain
the 400 bp sequence by colony PCR. A pZErO-2 plasmid that
contained the insert was prepared, and random primers were
inserted with the GPS-1 Tn7 transposon-based procedure
(New England Biolabs) for use in sequencing the plasmid. Se-
quencing showed that the insert contained the full length gene
for the ketoreductase, and the sequences corresponding to the
amino terminal and two internal peptides obtained from the
purified enzyme were present. Using PCR with the appropri-
ate primers and the plasmid containing the full length gene
as template, an NdeI site was introduced at the start of the
gene and an SmaI site was introduced after the stop codon.
The ends of the PCR product were digested with these two re-
striction enzymes then ligated into pBMS2000 cut with the
same enzymes. The ligation product was used to transform
DH5a-T1 (Invitrogen) chemically-competent E. coli cells,
and colonies containing the ketoreductase gene insert were
identified by PCR. Plasmids were isolated from two of the col-
onies and one of the plasmids was used to transform
BL21Star(DE3) E. coli cells (Invitrogen) for expression. One
of the colonies was grown at 308C in MT5 medium (composi-
tion given below), and induced at OD600¼0.99 with 1 mM
IPTG. After 4 hours induction, cells were harvested and stored
frozen. Extracts (10% w/v) were prepared by sonication and
found to contain a predominant 27 kd protein with an Agilent
2100 Bioanalyzer Protein 200 Labchip. In an activity assay, the
extract gave complete conversion of chloroketone 1 to the S-al-
cohol 2 with 100% ee.
Construction of Strain with Ketoreductase and
Glucose 6-Phosphate Dehydrogenase
The cloned ketoreductase gene was transferred to pBMS2000-
SCGD which contains a glucose 6-phosphate dehydrogenase
gene for NADPH regeneration. The glucose 6-phosphate de-
hydrogenase gene was PCR cloned from Saccharomyces cere-
visiae DNA using primers based on a published sequence[9] as
an NdeI-EcoRI fragment and expressed in pBMS2000 by Tho-
mas Franceschini and Li You (Bristol-Myers Squibb). At-
tempts to cut the ketoreductase gene from plasmid
pBMS2000 with HindIII and SmaI were not successful, appa-
rently because the SmaI site was lost upon cloning. Therefore,
the gene and promoter were copied from the plasmid by PCR.
SmaI and HindIII sites were added upstream of the promoter
and an SmaI site was incorporated into the primer homologous
to the 3’ end of the ketoreductase gene. The amplified product
was trimmed with SmaI then ligated into pBMS2000-SCGD
(plasmid containing cloned glucose 6-phosphate dehydrogen-
ase from Saccharomyces cerevisiae) cut with SmaI. The ligation
product was used to transform DH5a-T1 E. coli cells. Plasmids
were isolated from seven of the transformants and orientations
of the two cloned genes were checked by cutting with HindIII,
BamH1, or NdeI. One of the HindIII sites was apparently lost
and no conclusion could be drawn from that digestion. Howev-
er the BamH1 and NdeI digestions were consistent with three
of the plasmids having the two genes in the same orientation
(desired situation for higher transcription rates) and one of
the plasmids having the genes with opposite orientation (unde-
sired). One of the plasmids with the desired orientation was
used to transform BL21Star(DE3) E. coli for expression.
This strain was designated E. coli SC16529.
E. coli SC16529 (pBMS2000-SCGD-KR)
Fermentation in 15-Liter Tanks
MT5 inoculum medium contained:2.0%Yeastamin; 4.0% glyc-
erol; 0.6% sodium phosphate, dibasic; 0.3% potassium phos-
phate, monobasic; 0.125% ammonium sulfate; 0.0246% mag-
nesium sulfate heptahydrate (added post-autoclaving from a
filter-sterilized solution); 0.005% kanamycin (added post-au-
toclaving from a filter-sterilized solution); pH was adjusted to
7.2 prior to autoclaving at 1218C.
MT5-M1 fermentor medium contained: 2.0% Quest Hy-
Pea; 1.85% Tastone-154; 4.0% glycerol; 0.6% sodium phos-
phate, dibasic; 0.125% ammonium sulfate; 0.04% UCON anti-
foam; 0.0246% magnesium sulfate heptahydrate (added post-
sterilization from a filter-sterilized solution). Medium in the
fermentor was batched with deionized water to yield a final
volume of 15 liters after post-sterilization additions and feeds.
The pH was adjusted to 7.4 prior to sterilization.
Estimation of the Molecular Weight of the
Ketoreductase Expressed in E. coli
A sample of the extract (0.1 mL) prepared above in 50 mM po-
tassium phosphate buffer pH 7 containing 10% glycerol and
1 mM dithiothreitol was applied to a Superdex 200 column
A frozen vial of E. coli SC16529 was thawed and transferred
to a 4-liter flask containing 1 liter of MT5 medium. The flask
1078
ꢀ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 1073–1080

Preparation of Chiral Alcohols
FULL PAPERS
was incubated at 308C and 250 rpm for ca. 24 h, at which time
the OD600 of the F1 stage growth was 6.4 units/cm. The fermen-
tor, containing15 liters of MT5-M1 medium, was then inoculat-
ed with sufficient broth (ca. 820 mL) to yield an optical density
in the tank of 0.35 units/cm. Induction was performed approx-
imately 2.5 hpost-inoculation. IPTG, which had beendissolved
in 500 mL water and filter-sterilized through a 0.2 um filter, was
added aseptically to the tank to yield a final concentration of
1.0 mM. The tank was run at 308C with an agitator speed of
500 rpm and airflow of 1 vvm. Ammonium hydroxide was
used to maintain pH at 7.0 throughout the run. The tank was
harvested 21 h after inoculation. At this time, the broth was
subcooled and centrifuged, followed by a wash with 10 mM
pH 7.0 sodium phosphate buffer and a second centrifugation.
The final yield was 296 g cell paste which was frozen and stored
at ꢀ708C until needed for preparative biotransformation.
NMR of Alcohol
NMR spectra were recorded in CDCl3 solution with TMS as in-
ternal standard using Jeol NMR spectrophotometer
a
(400 MHz for 1H and 100 MHz for 13C) and d values are report-
ed. 1H NMR: d¼2.86 (1 H, br s, OH), 3.68 (2 H, m, H-2’), 4.84
(1 H, m, H-1’), 7.13 (1 H, m, H-5), 7.22 (1 H, m, H-6), 7.44 (1 H,
m, H-2); 13C NMR: d¼50.99 (C-2’), 73.13 (C-1’), 116.57 (C-6),
116.79 (C-3), 125.78 (C-5), 128.29 (C-2), 136.74 (C-1), 156.27
and 158.74 (C-4).
Preparation of 4 using Whole Cells
Preparation of 4 using Cell Extract
E. coli SC16529 (75 g wet cell weight) was dispersed in a solu-
tion containing glucose (30 g, 167 mmoles), 675 mL water and
75 mL 1 M potassium phosphate buffer pH 7 by using an Ultra-
turrax homogenizer. Dry cell weight was 34.6% of the wet cell
weightof the E. coli cell paste. Thecell suspension was added to
a 1-L jacketed reactor, mechanically stirred and maintained at
308C. NADP (115 mg, 0.15 mmoles) and ketone 3 (15 g, 72.4
mmoles, finely ground with a mortar and pestle) were added
to start the reaction. pH 7 was maintained by periodic addi-
tions of 10 N NaOH during the first two hours, and later by ad-
dition of 1 N NaOH with a pH stat.
After 22 hreactiontime, 150 g of washed XAD-16 were add-
ed and the mixture was stirred with a mechanical stirrer. After
21 h, HPLC showed about 3% of the alcohol 4 remaining in the
aqueous layer. After 22 h, the mixture was filtered through a 40
mesh stainless steel screen. The XAD-16 was washed with
2000 mL water, then the product alcohol 4 was eluted with eth-
yl acetate (300 mL, 3 times). The 50 mL water layer from the
first desorption was also back-extracted with 100 mL ethyl ace-
tate. The combined ethyl acetate eluates were dried over anhy-
drous Na2SO4, filtered, and solvent was removed with a rotary
evaporator. After drying for 20 h in vacuum, 14.8 g of a light
yellowish oil were obtained. HPLC showed that the alcohol
AP was 91.59 and ketone AP was 2.32. The yield of 4 corrected
for AP was 89% and ee was 100%.
E. coli SC16529 (80 g wet cell weight) wassuspended to 400 mL
in 50 mM potassium phosphate buffer pH 7 containing 1 mM
dithiothreitol, passed twice through a microfluidizer at 12000
psi and centrifuged for 15 min at 23,424ꢁg. The supernatant
was used as a source of ketoreductase and glucose 6-phosphate
dehydrogenase. To a 2-L jacketed reactor magnetically stirred
and maintained at 308C were added: sodium glucose 6-phos-
phate (40 g, 142 mmoles), 650 mL water and 100 mL 1 M po-
tassium phosphate buffer pH 7. The solution was adjusted to
pH 7 by addition of about 12.5 mL of 10 N NaOH. Cell extract
(200 mL containing 724 U/mL ketoreductase and 188 U/mL
glucose 6-phosphate dehydrogenase) and NADP (153 mg, 0.2
mmoles) were added to the solution. Ketone 3 (20 g, 97
mmoles) dissolved in 50 mL methanol was slowly pipetted
into the stirred solution. The progress of the reduction was fol-
lowed by HPLC. After 4.5 h the pH was adjusted from 6.35 to 7
by addition of about 7 mL 10 N NaOH and an additional 50 mL
cell extract and 38 mg NADP (0.05 mmoles) were added.
After 5.5 h total reaction time, 200 g of washed XAD-16 res-
in were added and the mixture was stirred with a mechanical
stirrer overnight (16 h). The adsorption was followed by
HPLC. After 16 hours only about 1.6% of alcohol 4 remained
in the aqueous layer. The mixture was filtered through a 40
mesh stainless steel screen to collect the XAD-16. The XAD-
16 was washed with about 3000 mL water, then ethyl acetate
(500 mL) was added, the mixture was stirred with a mechanical
stirrer for 60 minutes, then filtered through the stainless steel
screen mesh. The filtrate contained about 150 mL water and
ethyl acetate (500 mL). The two layers were separated. Only
1% of the total alcohol was in the water, which was discarded.
The desorption of XAD-16 was nearly complete with two more
500-mL portions of ethyl acetate. There was no water layer af-
ter the first desorption. The combined ethyl acetate solution
was dried over anhydrous Na2SO4 and filtered (glass fritted
disc, medium porosity). Removal of solvent in the rotary evap-
orator provided 20 g of an oil. After drying for 24 h in vacuum,
19.6 g of a light yellowish oil were obtained. HPLC showed that
the alcohol area percent (AP) was 94.01 and ketone AP was
1.26. The yield of 4 corrected for AP was 91%, and ee was
100%.
References
[1] M.-R. Kula, in: Stereoselective Biocatalysis, (Ed.: R. N. Pa-
tel), Marcel Dekker, New York, 2000, pp. 839–866.
[2] V. I. Tishkov, A. G. Galkin, V. V. Fedorchuk, P. A. Savit-
sky, A. M. Rojkova, H. Gieren, M.-R. Kula, Biotechnol.
Bioeng. 1999, 64, 187–193.
[3] A. E. Serov, A. S. Popova, V. V. Fedorchuk, V. I. Tishkov,
Biochem. J. 2002, 367, 841–847.
[4] M. Kataoka, K. Kita, M. Wada, Y. Yasohara, J. Hasegawa,
S. Shimizizu, Appl. Microbiol. Biotechnol. 2003, 62, 437–
445.
Adv. Synth. Catal. 2005, 347, 1073–1080
asc.wiley-vch.de
ꢀ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
1079

Ronald L. Hanson et al.
FULL PAPERS
[5] E. A. Noltmann, C. J. Gubler, S. A. Kuby, J. Biol. Chem.
1961, 236, 1225–1230.
[8] N. D. Meadow, D. K. Fox, S. Roseman, Annu. Rev. Bio-
chem. 1990, 59, 497–542.
[6] C.-H. Wong, G. M. Whitesides, J. Am. Chem. Soc. 1981,
103, 4890–4899.
[9] D. Thomas, H. Cherest, Y. Surdin-Kerjan, EMBO J. 1991,
10, 547–553.
[7] W. Hummel, Appl. Microbiol. Biotechnol. 1990, 34, 15–
19.
1080
ꢀ 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
Adv. Synth. Catal. 2005, 347, 1073–1080