Enzymatic reduction of α-substituted ketones with concomitant dynamic kinetic resolution

Article abstract of DOI:10.1016/j.molcatb.2016.07.008

Racemic α-substituted ketones were converted to the corresponding chiral alcohols with high diastereo- and enantioselectivities using enzymatic reduction with concomitant dynamic kinetic resolution. Reductions of N-protected α-amino ketones by microorgani

Full text of DOI:10.1016/j.molcatb.2016.07.008

Accepted Manuscript
Title: Enzymatic reduction of ␣-substituted ketones with
concomitant dynamic kinetic resolution
Author: Ronald L. Hanson Zhiwei Guo Francisco
´
¨
Gonzalez-Bobes Michael D.B. Fenster Animesh Goswami
PII:
S1381-1177(16)30131-X
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.molcatb.2016.07.008
MOLCAB 3399
To appear in:
Journal of Molecular Catalysis B: Enzymatic
Received date:
Revised date:
Accepted date:
6-6-2016
14-7-2016
15-7-2016
´
Please cite this article as: Ronald L.Hanson, Zhiwei Guo, Francisco Gonzalez-Bobes,
¨
MichaelD.B.Fenster, AnimeshGoswami, Enzymaticreductionof␣-substitutedketones
with concomitant dynamic kinetic resolution, Journal of Molecular Catalysis B:
Enzymatic http://dx.doi.org/10.1016/j.molcatb.2016.07.008
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Enzymatic reduction of α-substituted ketones with concomitant dynamic kinetic resolution
Ronald L. Hanson, Zhiwei Guo, Francisco González-Bobes, Michaël D. B. Fenster, Animesh
Goswami*
Chemical and Synthetic Development, Bristol-Myers Squibb, One Squibb Drive, New
Brunswick, NJ 08903, USA

Graphical abstract

Highlights
-
Enzymatic reduction of racemic α-substituted ketones and simultaneous dynamic kinetic
resolution
-
-
Stereoselective preparation of homochiral α-substituted chiral alcohols
Reductions of N-protected α-amino ketones by microorganisms and commercial enzymes
provided N-protected α-amino alcohols in high de and ee
Choice of buffer is a crucial factor for the reduction of an α-methyl ketone to the desired
chiral alcohol in high de and ee
-
-
Glucose dehydrogenase is acting as a ketoreductase
Abstract
Racemic α-substituted ketones were converted to the corresponding chiral alcohols with
high diastereo- and enantioselectivities using enzymatic reduction with concomitant dynamic
kinetic resolution. Reductions of N-protected α-amino ketones by microorganisms and
commercial enzymes provided N-protected α-amino alcohols. Choice of buffer was found to be a
crucial factor for the successful reduction and simultaneous dynamic resolution of an α-methyl
ketone to the corresponding chiral alcohol.
Keywords
Enzymatic reduction, dynamic kinetic resolution, α-substituted ketones, chiral alcohols
1. Introduction

Ketoreductases are frequently used to reduce ketones to chiral alcohols [1]. With an
enolizable group at the α-position the stereochemistry at the adjacent carbon can often be
established due to keto-enol tautomerism and selection of only one of the ketone enantiomers for
reduction by the enzyme leading to reduction with concomitant dynamic kinetic resolution
(DKR) [2-5]. Although there are many examples of such ketone reduction with DKR, there are
only a few reports on synthesis of α-alkyl or N-protected α-amino hydroxy esters from the
corresponding ketones using enzymatic reduction of ketone and DKR [6-8].
Compound 1 (Fig 1) is a positive allosteric modulator for the metabotropic glutamate
receptor (mGluR5) considered for development as a possible treatment of neurological disorders
[9]. The initial synthetic route to 1 used an intermediate syn-(1R,2R) N-protected amino alcohol 2
prepared by an aminohydroxylation catalyzed by OsO4 which was further converted to a key
intermediate 3 for the synthesis of 1 [10]. The aminohydroxylation reaction had only moderate
regioselectivity and enantioselectivity, and required careful attention to downstream removal of
osmium. Using alternative chemistry, both the syn- 2, anti-4 protected amino alcohols were
further converted to 3 the penultimate for the initial synthetic route.
The alcohol 8 (Fig 2) is a key chiral intermediate for a GPR40 modulator drug candidate
10 for the treatment of diabetes [11, 12]. Initial synthesis of 8 was carried out in seven steps with
three isolations involving Evan’s auxiliary to create chiral centers, and finally hydrogenation of
benzyl amine 9 to 8 with an overall yield of 12% [13].
This report describes the preparation of both syn- and anti N-protected amino alcohols 2
and 4 from ketone 5 (Fig 1), and anti alcohols 9 and 12 from ketone 11 (Fig 2). In both cases, an
α-substituted racemic ketone was enzymatically reduced to a chiral alcohol with the
stereochemistry at the second chiral center (α-carbon) set via a dynamic resolution. Summaries

of our work on enzymatic reductions and the detailed synthesis of drug candidates have been
reported by us in separate publications [13, 14].
2. Experimental
Starting materials and product markers were prepared as described [9-14].
Enzymes were purchased from commercial sources. Microorganisms were purchased from
ATCC (ATCC#) or obtained from our own culture collection (SC#). Reaction conversion and
purity of isolated product were determined from achiral HPLC area percentage (AP).
Diastereromeic excess (de) and enantiomeric excess (ee) were determined from chiral HPLC.
2.1. Screening for reduction with commercial enzymes:
Ketoreductases (1 mg) were screened in 1-mL reactions in 24-well plates
containing 1 mg ketone dissolved in 0.02 mL DMSO, 1 mM NADP or NAD, 5 mg glucose, 0.05
mg glucose dehydrogenase (3.2 U, Amano), 1 mM dithiothreitol and 0.98 ml 0.1 M potassium
phosphate buffer pH 7. After 20 to 87 h at 30oC and 400 rpm, reactions were quenched with 1
ml methanol and analyzed by HPLC. For the Codexis screening kit and KRED panels the
manufacturer’s protocol was followed.
2.2. Screening for reduction with microorganisms:
Microbial strains were screened in 24-well plates containing inoculum and 2 ml F7
medium (1% malt extract, 1% yeast extract, 0.1% peptone and 2% dextrose adjusted to pH 7) per
well. Plates were incubated for 1 day at 28°C, 600 rpm then 2 mg ketone in 0.04 mL DMSO was
added and the incubation was continued for 2 days. The incubations were quenched by addition
of 2 mL methanol per well and analyzed by HPLC.
2.3. Reduction of ketone 5

2.3.1. Analysis of ketone 5 and N-protected alcohols 2, 4, 6, 7:
2.3.1.1. Achiral HPLC:
Samples of 0.25 or 0.5 mL were diluted to 1 mL with methanol, filtered and analyzed
with a Luna C18(2) (15x0.46 cm, 5) column. The mobile phase solvents were A water:
CH3CN (95:5) and B water: CH3CN (5:95) each containing 10 mM ammonium acetate with a
gradient of 50 to 70% B in 12 min, flow rate was 1 mL/min, detection was by UV at 270 nm,
temperature was 40 °C, and injection volume was 5 µL. Retention times for Boc-protected
compounds were: 6.89 min anti-alcohol (4a/7a), 7.38 min syn-alcohol (2a/6a), ketone (5a) 9.85
min. Retention times for CBz-protected compounds were: 7.15 min anti-alcohol (4b/7b), 7.36
min syn-alcohol (2b/6b), ketone (5b) 9.81 min.
2.3.1.2. Chiral HPLC for Boc-amino alcohols:
Samples were analyzed with a Chiralpak IC (25 x 0.46 cm, 5µ) column. The mobile
phase was water: CH3CN (1:1) containing10 mM ammonium acetate, flow rate was 0.5 mL/min,
detection was by UV at 270 nm, temperature was ambient, and injection volume was 20 µL.
Retention times were: 4a anti 1(1R,2S) 14.6 min, 2a syn 1(1R,2R) 16.2 min, 6a syn 2 (1S,2S)
18.5 min, 7a anti 2(1S,2R) 19.8 min.
2.3.1.3. Chiral HPLC for CBz-amino alcohols:
Samples were analyzed with a Phenomenex Lux Cellulose-3 (15 x 0.46-cm, 3) column.
The mobile phase solvents were CH3CN and water each containing 0.05% trifluoroacetic acid
with a gradient of 35 to 50%, flow rate was 1 mL/min, detection was by UV at 270 nm,
temperature was 25 °C, and injection volume was 20 µL. Retention times were: 4b anti 1(1R,2S)
15.7 min, 2b syn 1(1R,2R) 17.1 min, 7b anti 2 (1S,2R) 21.7 min, 6b syn 2 (1S,2S) 23.7 min. For

CBz-amino ketone 5b enantiomer separation, the gradient was changed to 35 to 90% CH3CN in
25 min and the retention times for two enantiomers were 13.8 and 14.8 min.
2.3.2. Preparation of anti 1, (1R,2S-) alcohol 4a using Pichia methanolica ATCC56508:
One vial of Pichia methanolica ATCC56508 was added to 100 mL F7 medium in a
500 mL flask and incubated at 28°C, 200 rpm. After 24 h 10 mL portions of this culture were
used to inoculate 1 L F7 medium in three 4-L flasks. The 4-L flasks were incubated at 28°C, 200
rpm for 24 h then 2 g ketone 5a in 50 mL DMSO was added to each of the 1 L cultures and the
incubation was continued for another 31 h. A sample of 0.25 mL was taken from each flask and
diluted with 0.75 mL methanol for HPLC analysis. HPLC showed that the 3 flasks contained a
total solution yield of 4.65 g anti 1 (R,S) alcohol 4a (99% ee, 98% de) and 0.37 g remaining
ketone 5a. The microbial transformation mixtures (3 L in total, pH 7.5) were poured into a 5-L
three neck flask with an overhead stirrer. NaHCO3 (60 g) was added to maintain pH 8~8.5. The
mixture was extracted first with 1.5 L of EtOAc. Celite pad filtration was applied to overcome
the emulsion issue and help phase separation. The cake was extracted with 1 L EtOAc and
filtered. The EtOAc filtrate was used to extract the aqueous phase for a second time. The
combined EtOAc extract was washed with 0.5 L of a mixture of 0.1 L 5% NaHCO3 and 0.4 L
brine. Solvent removal of the combined EtOAc extract gave a solid, which was dried in a
vacuum oven at 30 ºC for 3 h and room temperature overnight to give 6.2 g of anti 1 (1R,2S)-
alcohol 4a as off-white solid, “as is” yield 103%, AP 96.3, ee 99.8%, de 99.0%, NMR (Figs S-1
and S-2).
2.3.3. Preparation of anti 1 (1R,2S)- alcohol 4b using KRED-145:

Glucose (5 g, 27.8 mmoles), 50 mL 1 M potassium phosphate buffer pH 8, 350 mL
water, dithiothreitol (77 mg, 0.50 mmoles), NADP (191 mg, 0.25 mmoles), glucose
dehydrogenase (Amano, 20 mg, 1520 U), KRED-145 (Codexis, 400 mg) ketone 5b (1 g, 2.17
mmoles dissolved in 50 mL DMSO) and 50 mL DMSO were added in that order to a stirred 1-L
reactor and stirring was continued at 30oC. Samples of 0.25 mL taken at various times were
diluted with 0.75 mL for HPLC and indicated little change from 24 h to 41 h of reaction. After
41 h, the conversion was about 98%. The biotransformation mixture was mixed with NaHCO3
(20 g), and extracted with 500 mL of EtOAc. Phase separation was slow. After 3 h, the lower
aqueous phase became clear and was separated. HPLC showed no product left in the aqueous
phase, which was then discarded. The cloudy organic phase was filtered through a pad of celite.
The cake was washed with 200 mL EtOAc and filtered. The combined EtOAc filtrate and
washing was washed with 0.3 L of a mixture of 0.1 L 5% NaHCO3 and 0.2 L brine. Solvent
removal of the EtOAc solution gave a gummy material, which solidified after drying in a
vacuum oven at 30 ºC overnight. The solid was treated with 10 mL of heptanes and filtered. The
solid was set in a hood for 1 h, dried in a vacuum oven at room temperature for four days to give
1.026 g product anti 1, (R,S)- alcohol 4b, AP 94, “as is” yield 102% (corrected yield 96% from
AP) containing the ketone (AP 2.3) and other minor impurities. Chiral HPLC showed one major
enantiomer anti 1, (R,S) alcohol 4b peak with ee 99.7% and de > 99.9%; the syn-isomers were
not detected. Further crystallization from a mixture of ethyl acetate and n-heptane purged the
ketone impurity and provided anti 1, (1R,2S) alcohol 4b as a single pure diastereo-, and
enantiomer in 80% yield, NMR (Figs S-3 and S-4).
2.4. Reduction of ketone 11:
2.4.1. Analysis of ketone 11 and alcohols 9, 12, 13, and 14:

2.4.1.1. Achiral HPLC:
Samples of 0.25 or 0.5 mL were diluted to 1 mL with methanol, filtered and analyzed
with a Luna C18(2) (15x0.46 cm, 3) column. The mobile phase solvents were A water: CH3CN
(95:5) and B water: CH3CN (5:95) each containing 10 mM ammonium acetate with a gradient of
10 to 100% B in 12 min, flow rate was 1 mL/min, detection was by UV at 210 nm, temperature
was 40 °C, and injection volume was 5 µL. Retention times were: 5.1 min syn-alcohol (13/14),
5.5 min anti-alcohol (9/12), ketone (11) 9.5 min.
2.4.1.2. Chiral HPLC for ketone 11 and alcohols 9, 12, 13, and 14:
Samples were analyzed with a Chiralpak IA (15 x 0.46-cm, 5µ) column. The mobile
phase was heptane: ethanol (90:10), flow rate was 1 mL/min, detection was by UV at 220 nm,
temperature was ambient, and injection volume was 10 µL. Ketone and anti-alcohol enantiomers
showed baseline separations with retention times as follows: ketone 11 enantiomers 3.00 and
3.57 min, 12 anti- (3S,4S) alcohol 3.91 min, 9 anti- (3R,4R) alcohol 5.91 min. The syn-alcohols
13 and 14 with retention times of 4.05 and 5.04 min were not well separated.
2.4.2. Rates of racemization and reduction by GDH of ketone 11 in different buffers:
Four 1 mL reactions in 4 mL vials were used for each buffer. The reaction mixtures
were incubated at 30°C at 500 rpm and contained the indicated buffer concentration, 100 µL of
ketone 11, 10 mg glucose, 1mM NAD and 2 mg GDH. A reaction was quenched every 2 hour
with 3 mL isopropanol and a sample was diluted with methanol for HPLC.
2.4.3. Preparation of anti-3R,4R alcohol 9:

To a jacketed reactor, ketone 11 (5 g, 24.60 mmoles), water (24 mL), 1 M potassium
phosphate buffer pH 7 (5 mL), 1 M tris(hydroxymethyl)aminomethane hydrochloride (tris)
buffer pH 7.4 (2.5 mL) were added and the mixture was stirred with a magnetic stirrer. Glucose
(6 g, 33.3 mmoles), a solution of 33.17 mg NAD (33.2 mg, 0.05 mole) in 2.5 ml water, and a
solution of 100 mg Codexis GDH-105 in 5 ml 50 mM potassium phosphate buffer pH 7 were
added to the mixture. The final concentrations of the buffers were: 0.1 M potassium phosphate
buffer, 0.05 M tris buffer. Enzyme was 2% of substrate weight and NAD concentration was 1
mM. The biphasic reaction mixture was stirred at 30oC. The reaction was allowed to continue for
about 24 hours until the ketone 11 was not detected by HPLC. The pH dropped to about 6.65 by
the end of the reaction. Although gluconic acid is a by-product of the reaction, the amine
prevented the pH from dropping too low.
After reaction completion, the pH of the reaction mixture (about 50 mL, pH 6.6) was
adjusted to 11 with 5 M NaOH (6 mL). The aqueous reaction mixture was extracted with MTBE
(2 X 50 mL). Celite pad filtration was applied to overcome emulsion. HPLC of the MTBE phase
showed only the anti-alcohol 11 while that of aqueous waste showed less than 1% product loss.
The MTBE layer was concentrated on a rotary evaporator at 22 °C to a weight of 15 g (about 2
volumes of MTBE left), and 50 mL n-heptane (50 mL) was added. The solution was
concentrated at 22°C to a weight of 40 g whereby product crystallized out (addition of a small
amount of seed facilitated crystallization). The mixture was further concentrated to 28 g at 22 °C.
After stirring for one hour, the mixture was filtered and the solid was dried at ambient
temperature to provide 4.243 g of anti- (3R,4R) alcohol 9 as a crystalline solid, yield 84.0%,

HPLC AP 99.4, de 100%, ee 100%, NMR (Figs S-5 and S-6). Evaporation of the mother liquor
gave 445 mg of additional solid anti-alcohol 9, yield 8.8%, HPLC AP 91.8.
2.4.3. Preparation of anti-(3S,4S) alcohol 12:
To a jacketed reactor, ketone 11 (15 g, 73.8 mmoles), water (161 mL), 1 M potassium
phosphate buffer pH 6 (30 mL), 100mM magnesium sulfate (3.75mL), and isopropanol (60 mL)
were added and the mixture was stirred with a magnetic stirrer. A solution of 229.5 mg NADP in
15 ml water, and a solution of 150 mg Ketoreductase P1-B12 (Codexis) in 15 ml 50 mM
potassium phosphate buffer pH 7 were added to the mixture. The biphasic reaction mixture was
stirred at 30oC. After 15 h about 1% of ketone remained in the reaction mixture and pH went up
to 8.2. The pH was adjusted to 7 with 3 mL of 85% phosphoric acid. The reaction was stopped at
23 h with negligible amount of (0.01% by HPLC) remaining ketone 11.
The above reaction mixture (from 15g ketone 11) was combined with reaction mixture
from another small 1g batch run in the same way (total about 300 mL). The mixture was mixed
with 8 g celite, adjusted to pH 11 with 10 N NaOH (12 mL), stirred at room temperature for 2 h
and filtered through a pad of celite (15 g). The cake was washed with 50 mL water. The
combined filtrate and washing was adjusted to pH 12.0 and stirred in a cold room for 1 h then
filtered through filter paper. The filtrate was clear and adjusted to pH 13.0, stirred in a cold room
overnight, and filtered through filter paper. All solids containing product and the celite cake were
combined, dried in a hood for 2 h, and then dried in a vacuum oven at 30 °C overnight. The
solids were extracted twice each with 200 mL methanol and the cake was washed with 100 mL
methanol. The combined methanol extract (500 mL) was concentrated to about 100 mL, 100 mL
water was added, and further concentrated to 90 g. The concentrated mixture was stirred in an ice
bath for 2 h to allow crystallization. The cold mixture was filtered to provide anti-alcohol 12 as

solid. The solid was dried in a vacuum oven at 30 °C overnight and at room temperature over the
weekend to give 10.7 g solid anti-(3S,4S) alcohol 12, yield 66%, slightly pink color, HPLC AP
99.2, de~100%, ee~100%, NMR (Figs S-7 and S-8).
3. Results and Discussion
3.1. Reduction of ketone 5
3.1.1. Reduction of N-Boc protected ketone 5a: Preparation of Boc-amino alcohol
Five hundred twenty-seven commercial ketoreductases, eight BMS cloned
ketoreductases, and two hundred forty-one microbial strains were screened for reduction of the
Boc-amino ketone 5a. Several commercial ketoreductases were identified which gave
predominantly a single stereoisomer for three of the four possible Boc-amino alcohols (Table 1).
The desired syn1(1R,2R)- isomer 2a was obtained with >75% de and >95% ee from two
commercial ketoreductases. The desired anti 1(1R,2S)- isomer 4a was obtained with >90% de
and >99% ee with two commercial ketoreductases. The syn1(1R,2R)- isomer 2a and anti
1(1R,2S)- isomer 4a producing enzymes identified in initial screening were further evaluated at
different pHs and enzyme loading to optimize reaction condition. Though diastereo- and
enantioselectivities were high, all enzymes required very high loading for the reduction of Boc-
protected ketone 5a. Enzyme activities were too low and these enzymes will be impractical for
preparative purposes unless their activities are improved by directed evolution [15]. Four
commercial enzymes produced the undesired anti 2(1S,2R)- isomer 7a with >99% ee; two with
about 95% de while the other two with >99% de. Interestingly, the undesired syn2 (1S,2S)-
isomer 6a was not produced by any commercial ketoreductase enzyme. Several yeast strains
gave a high ee but low de for the desired syn1(1R,2R)- isomer 2a (Table 2); other yeast strains

and two Aspergillus strains gave both high de and ee for the desired anti 1(1R,2S)- isomer 4a
which could be used for an alternate route. Microbial strains giving the undesired syn 2(1S,2S)-
6a and anti 2(1S,2R)- 7a isomers were also identified. Several strains of Pichia methanolica
produced the anti 1(1R,2S)- alcohol 4a almost exclusively (both ee and de >99%). After further
comparison of the strains giving the desired anti 1(1R,2S)- 4a, Pichia methanolica ATCC 56508
was chosen for a preparative batch. Three 1-L cultures were each incubated with 2 g of the
ketone 5a for 31 h. After work up 6.2 g of the anti 1(1R,2S- isomer) 4a was obtained as an off-
white solid in almost quantitative yield with excellent purity (HPLC AP 96.3) and high ee 99.8%
and de 99.0%.
3.1.2. Reduction of N-CBz protected ketone 5b: Preparation of CBz-amino alcohol
Of the commercial enzymes active with the N-Boc-amino ketone 5a, only KRED-145
(Codexis) had substantial activity for reduction of the CBz-amino ketone 5b, giving the desired
anti 1(1R,2S)- alcohol 4b. Most importantly, activity of KRED-145 for the reduction of CBz-
amino ketone 5b was much higher than that for Boc-amino ketone 5a. After reaction
optimization, 40 wt% of KRED-145 was found to complete the reduction of 5b in 24 hours.
Ninety-three strains of yeast were screened for reduction of the CBz-amino ketone 5b and gave
mostly the same hits and misses as for the Boc-amino ketone but with lower conversions.
KRED-145 was used for reduction of 1 g of the CBz-amino ketone 5b with glucose
dehydrogenase used for NADPH regeneration. After work up, almost quantitative yield of 1.03 g
of anti 1(1R,2S)- alcohol 4b was obtained as a solid with excellent purity (HPLC AP 94) and
excellent ee 99.7 and de >99.9%. Further crystallization from a mixture of ethyl acetate and n-
heptane provided anti 1, (1R,2S) alcohol 4b as a single pure diastereo-, and enantiomer in 80%
yield.

The remaining Boc- or CBz-amino ketone was racemic during the reactions and the
almost complete reduction to predominantly one alcohol isomer indicated a dynamic kinetic
resolution with racemization of the adjacent carbon (C*H-NHP) during enzymatic reduction of
the ketone 5. The chiral alcohols obtained by biocatalytic transformations have been successfully
used for the synthesis of the drug candidate 1 [14].
3.2. Reduction of ketone 11
Twenty four keto reductase enzymes from Codexis were screened. The first five used
glucose dehydrogenase and the remainder used isopropanol for cofactor regeneration. Ketone 11
was almost completely reduced under the screening conditions by the ketoreductase enzymes.
Results showed various levels of anti-/syn- ratio for the reduction by ketoreductase enzymes
from about 1:6 to 6:1 spread of diastereoselectivity. None of the KRED enzymes showed very
high diastereoselectivity. Surprisingly, the best ratio of anti to syn product was found in the
control sample containing glucose dehydrogenase (GDH) but no ketoreductase. GDH served as
a ketoreductase as well as for the oxidation of glucose to regenerate the cofactor for the reaction.
Chiral HPLC indicated that the product from reduction by GDH was almost exclusively the
desired anti (3R,4R) -alcohol 9. Results are shown in Table 3. Reduction with GDH under
screening condition was not complete and optimization studies were carried out to improve the
conversion and diastereoselectivity.
The activity of GDH enzyme for reduction of ketone is 30% higher in tris-buffer than in
potassium phosphate buffer at the same pH 7. With tris buffer alone the ketone 11 does not
racemize quickly and one enantiomer of unreacted ketone remains predominantly in the reaction
mixture. Although the initial rate of the enzyme reaction was faster in tris buffer alone than in
phosphate buffer alone, racemization was very slow in the absence of the phosphate buffer and

the reaction took a long time to complete in tris buffer alone. The potassium phosphate buffer
was necessary to racemize ketone 11 during the reaction (Fig 3). The addition of tris buffer
together with the phosphate buffer gave a somewhat higher reaction rate. Therefore the
potassium phosphate tris buffer combination was used for preparative batches. (Fig 4). Jones et
al reported [16, 17] reduction of 2-methyl cyclohexanone, 2-ethylcyclohexanone and 3-
ethyltetrahydrothiopyran-4-one by horse liver alcohol dehydrogenase and reported the formation
of predominantly the anti-alcohol with as high as 100% ee for 1R,2R isomer for the 2-
methylcyclohexanone. They reported slow reaction with phosphate buffer and preferred faster
reacting tris-buffer media. The recovered ketone was not racemic and the reduction was less than
50% even after a long time.
GDH enzymes from various sources were examined for the reduction of ketone 11. For
Codexis GDH-105, the pH optimum for ketone reduction is 6.2 - 6.5, apparent Km for NADH is
0.07 mM, and apparent Vmax at pH 7 is 0.1 U/mg. Either 1 mM NAD or NADP could be used in
the reaction by Codexis GDH-105, and NAD was used for preparative lab batches. Codexis
CDX-901 GDH gave 100% ee of anti-alcohol 9 but showed slow and incomplete conversion
using either NAD or NADP. With Amano GDH, pH optimum for ketone reduction is 7,
apparent Km for NADPH is 0.25 mM, and apparent Vmax at pH 7.6 is 0.43 U/mg. Reduction
using NAD was much slower than with NADP for Amano GDH, and NADP was used for lab
batches. An in-house GDH recombinant enzyme [18] from Gluconobacter oxydans required
NADP, gave fair diastereoselectivity for anti- alcohol but only 51 % ee for the anti (3R,4R)-
alcohol enantiomer 9.
Preparative 5-g conversions of ketone 11 (100 mg/mL) to anti (3R,4R)- alcohol 9 were
carried out using Codexis GDH-105 (2% of substrate weight) and 1 mM NAD. Ketone 11 has

little solubility in the aqueous reaction but amino alcohol 9 is soluble. Although gluconic acid is
a by-product of the reaction, no pH control was necessary. The formation of the soluble amino
alcohol 9 prevented the pH from dropping below about 6.7. The isolated yield of 9 was 84% with
>99% de and >99% ee. A total of 6.2 kg of anti (3R,4R)- amino alcohol 9 was produced using
this procedure in four batches in 71-73% yield with excellent purity (HPLC AP > 99.7) and
stereoselectivity (both ee and de > 99.9%) which has been used for the synthesis of drug
candidate 10. The ketoreduction with concomitant dynamic resolution resulted in the production
of the desired synthetic intermediate 8 in 42% overall yield in three synthetic steps and three
isolations compared to a 12% overall yield in seven steps with two isolations of the previous
Evan’s auxiliary route to create the two chiral centers [13].
The undesired anti (3S,4S)- alcohol enantiomer 12 was also prepared. Although initial
screening of Codexis KRED P1-B10 and P1-B12 at pH 7 gave alcohols with anti:syn ratios of
only about 6, the ratios improved to > 40 at pH 6 with ee >99 (no detectable alcohol 9).
Preparative reduction of ketone 11 by KRED P1-B12 (1% of substrate weight) provided anti
93S,4S)- alcohol 12 in 66% isolated yield with excellent purity (HPLC AP 99.2) and
stereoselectivity (both de and ee >99%).
4. Conclusions
Enzymatic reduction of two racemic α-substituted ketones with simultaneous
dynamic kinetic resolution was utilized to prepare two chiral alcohols with adjacent stereocenters
in high diastereo- and enantioselectivities. The choice of buffer has a dramatic and opposing
effect on the reduction and racemization rates in one case, and the product alcohol was obtained
with high stereoselectivity by appropriate choice of reaction media.
5. Acknowledgments

The authors thank Dr. Charles Pathirana for some NMR spectra.
Funding:
This research did not receive any specific grant from funding agencies in public,
commercial, or not-for-profit sectors.
Appendix A. Supplementary Data
Supplementary data associated with this article can be found in the online version at ….

6. References
[1]
[2]
[3]
[4]
T. S. Moody, J. D. Rozzell, Modern biocatalytic ketone reduction, in: A. Goswami, J. D.
Stewart (Eds.), Organic synthesis using biocatalysis, Elsevier, Amsterdam, 2016, pp.
149-185.
Y. Yang, M. Drolet, M. M. Kayser, The dynamic kinetic resolution of 3-oxo-4-phenyl-β-
lactam by recombinant E. coli overexpressing yeast reductase Ara1p, Tetrahedron:
Asymm. 16 (2005) 2748-2753.
S. Lüdeke, M. Richter, M. Müller, Stereoselective synthesis of three isomers of tert-butyl
5-hydroxy-4-methyl-3-oxohexanoate through alcohol dehydrogenase –catalyzed dynamic
kinetic resolution, Adv. Synth. Catal. 351 (2009) 253-259.
A. Cuetos, A. Rioz-Martinez, F. R. Bisogno, B. Grischek, I. Lavandara, G. de Gonzalo,
W. Kroutil, V. Gotor, Access to enantiopure α-alkyl-β-hydroxy esters through dynamic
kinetic resolutions employing purified/overexpressed alcohol dehydrogenases, Adv.
Synth. Catal. 354 (2012) 1743-1749.
[5]
[6]
Z. Guo, B. P. Patel, R. M. Corbett, A. Goswami, R. N. Patel, Stereospecific microbial
reduction of ethyl 1-benzyl-3-oxo-piperidine-4-carboxylate,Tetrahedron: Asymm. 17
(2006) 2015-2020.
R. N. Patel, A. Banerjee, J. M. Howell, C. G. McNamee, D. Brzozowski, D. Mirfakhrae,
V. Nanduri, J. K. Thottathil, L. J. Szarka, Microbial synthesis of (2R,3S)-(-)-N-benzoyl-
3-phenyl isoserine ethyl ester – a taxol side-chain synthon, Tetrahedron: Asymm. 4
(1993) 2069-2084.
[7]
M. H. Yasuda, M. Ueda, K. Okano, H. Mihara, N. Esaki, Enzymatic synthesis of
unnatural amino acids, in: V. A. Soloshonok and K. Izawa (Eds.), Asymmetric synthesis

and application of amino acids, ACS Symposium Series, Washington DC, 2009, pp. 357-
374.
[8]
[9]
F. Xu, J. Y. L. Chung, J. C. Moore, Z. Liu, N. Yoshikawa, R. S. Hoerrner, J. Lee, M.
Royzzen, E. Cleator, A. G. Gibson, R. Dunn, K. M. Maloney, A. Mahbub, A. Goodyear,
J. Lynch, N. Yasuda, P. N. Devine, Asymmetric synthesis of cis-2,5-disubstituted
pyrolidine, the core scaffold of β3-AR agonists, Organic Letters 15 (2013) 1342-1345.
A. P. Degnan, H. Huang, L. B. Snyder, F. Yang, K. W. Gillman, M. F. Parker,
Preparation of oxazolidinone derivatives as modulators of mGluR5, PCT Int. Pat.
Appl. WO 2012064603 A1 20120518.
[10] F. Yang; L. B. Snyder; A. Balakrishnan; J. M. Brown; D. V. Sivarao; A. Easton; A.
Fernandes; M. Gulianello; U. M. Hanumegowda; H. Huang; Y. Huang; K. M. Jones; Y-
U. Li; M. Matchet; G. Mattson; R. Miller; K. S. Santone; A. Senapati; E. E. Shields; F. J.
Simutis; R. Westphal; V. J. Whiterock; J. J. Bronson; J. E. Macor; A. P. Degnan,
Discovery and preclinical evaluation of BMS-955829, a potent positive allosteric
modulator of mGluR5, ACS Med. Chem. Letters, 7 (2016) 289-293.
[11] B. A. Ellsworth, J. Shi, W. R. Ewing, E. A. Jurica, A. S. Hernandez, X. Wu, Z. Gu, Z.
Hong, S. P. O’Connor, Preparation of dihydropyrzole GPR40 modulators for treating
diabetes and related conditions and other diseases, PCT Int Pat Appl. WO 2014078611
A1, 22 May 2014.
[12] B. A. Ellsworth, J. Shi, E. A. Jurica, L. L. Nelson, X. Wu, A. H. Hernanendez, Z. Wang,
Z. Gu, K. N. Williams, B. Chen, E. C. Cherney, X. –Y. Ye, Y. Wang, M. Zhou, G. Cao,
C. Xie, J. J. Wilkes, H. Liu, L. K. Kunselman, A. K. Gupta, R. Jayarama, T. Ramar, J. P.
Rao, B. A. Zinker, Q. Sun, E. A. Dierks, K. A. Foster, T. Wang, M. E. Cvijic, J. M.

Whaley, J. A. Robl, W. R. Ewing, Discovery of BMS-986118, a dual MOA GPR40
agonist that produces glucose-dependent insulin and GLP-1 secretion, MEDI-31, 248th
ACS National Meeting, San Francisco, USA, August 10-14, 2014.
[13] M. D. Fenster, R. L. Hanson, K. Katipally, T. Paul, R. Rao, E. M. Simmons, I. S. Young,
J. J. Zhu, Scalable synthesis of a GPR40 agonist, ORGN-119, 248th ACS National
Meeting, San Francisco, USA, August 10-14, 2014.
[14] F. Gonzalez-Bobes, R. Hanson, N. Strotman, Z. Guo, A. Goswami Enantioselective
synthesis of a positive allosteric modulator of the mGluR5 receptor via dynamic kinetic
resolution of α-aminoketones Adv. Synth. Catal. (in press).
[15] U. T. Bornscheuer, G. W. Huisman, R. J. Kazlauskas, S. Lutz, J. C. Moore, K. Robbins,
Engineering the third wave of biocatalysis, Nature, 485 (2012) 185-194.
[16] J. B. Jones, T. Takemura, Enzymes in organic synthesis. 28. Reinvestigation of the horse
liver alcohol dehydrogenase-catalyzed reductions of 2-alkylcycolhexanaones.
Identification of cis-alkylcyclohexanols as minor products, Canadian J. Chem. 60 (1982)
2950-2956.
[17] J. B. Jones, T. Takemura, Enzymes in organic synthesis. 30. Reaction conditions –
control of enantiomeric purities. Horse liver alcohol dehydrogenase-catalyzed reductions
of 2-alkylcyclohexanones and their thiopyran analogs, Canadian J. Chem. 62 (1984) 77-
80.
[18] R. L. Hanson, R. M. Johnston, S. L. Goldberg, W. L. Parker, A. Goswami, Organic
Process Research and Development, Enzymatic preparation of an R-amino acid
intermediate for a γ-secretase inhibitor, 17 (2013) 693-700.

Fig 1. Reduction of ketone 5

Fig 2: Reduction of ketone 11

Fig 3: Enantiomeric excess of the remaining ketone 11 in GDH reduction carried out in different
buffers
0.1M Phos
0.1M tris
0.1M Phos + 0.05M tris
70
60
50
40
30
20
10
0
2
4
6
8
Time, h

Fig 4: Conversion of ketone 11 (100 mg/mL) to anti-alcohol 9 by GDH in different buffers
0.1M Phos
0.1M tris
0.1M Phos + 0.05M tris
100
80
60
40
20
0
2
4
6
8
Time, h

Table 1: Boc-amino alcohols produced from reduction of ketone 5a by commercial
ketoreductases
Product
Ketoreductase
Source
Enzyme Conversion
de
ee
wt%
100
(%)
45
(%)
82.7
(%)
97.2
2a, syn 1
(1R,2R)-
2a, syn 1
(1R,2R)-
4a, anti 1
(1R,2S)-
4a, anti 1
(1R,2S)-
7a, anti 2
(1S,2R)-
7a, anti 2
(1S,2R)-
7a, anti 2
(1S,2R)-
7a, anti 2
(1S,2R)-
KRED-NADH-107 Codexis
KRED-148
Codexis
500
100
500
100
100
500
500
65
97
77.3
91.2
95.9
95.1
95.3
96.6
>99*
>99*
>99*
>99*
ES KRED-112
KRED-145
Enzysource
Codexis
91
KRED-NADH-112 Codexis
100
93
CRED A-131
P1-B05
Almac
Codexis
Codexis
95
>99* >99*
>99* >99*
P2-D11
84
* Product concentration was low in the screening experiments leading to small peak area.
Though other isomers were not seen in HPLC, de and ee values can be reliably estimated to be
>99%.

Table 2. Boc-amino alcohols produced from reduction of ketone 5a by microbial strains
Product
Microorganism
Hansenula anomola
Hansenula anomola
Pichia anomola
Strain ID
Conversion
(%)
de
(%)*
30.1
ee
(%)*
>99*
2a, syn 1
(1R,2R)-
2a, syn 1
(3R,4R)-
2a, syn 1
(1R,2R)-
2a, syn 1
(1R,2R)-
2a, syn 1
(1R,2R)-
2a, syn 1
(1R,2R)-
2a, syn 1
(1R,2R)-
2a, syn 1
(1R,2R)-
6a, syn 2
(1S,2S)-
4a, anti 1
(1R,2S)-
4a, anti 1
(1R,2S)-
4a, anti 1
(1R,2S)-
4a, anti 1
(1R,2S)-
4a, anti 1
(1R,2S)-
4a, anti 1
(1R,2S)-
4a, anti 1
(1R,2S)-
7a, anti 2
(1S,2R)-
ATCC20211
SC16158
100
100
90
87
85
90
91
63
59
85
95
92
93
93
88
83
68
38.2
41.9
26.8
18.5
21.2
19.9
16.2
91.9
>99*
>99*
>99*
>99*
>99*
96
>99*
>99*
>99*
>99*
>99*
>99*
>99*
97.1
ATCC8168
ATCC20029
ATCC36995
ATCC66094
ATCC66346
ATCC20144
ATCC13259
ATCC56510
ATCC56508
ATCC58372
ATCC56509
ATCC46071
SC2164
Pichia anomola
Pichia anomola
Pichia anomola
Pichia anomola
Pichia anomola
Rhodococcus sp.
Pichia methanolica
Pichia methanolica
Pichia methanolica
Pichia methanolica
Pichia methanolica
Aspergillus niger
Aspergillus terreus
>99*
>99*
>99*
>99*
>99*
>99*
>99*
>99*
SC13803
96
Actinosynnema
pretiosum
ATCC31309
>99*
* Product concentration was low in the screening experiments leading to small peak area.
Though other isomers were not seen in HPLC, de and ee values can be reliably estimated to be
>99%.

Table 3: Screening results for reduction of ketone 11
Enzyme*
Ketone
11,
%
syn-
Alcohol
(13+14)
%
anti-
alcohol
(8 + 12)
%
anti- /
syn-
Ratio
anti –
alcohol, 12
3S,4S
%
anti-alcohol,
8
3R,4R
%
KRED 101
KRED 119
KRED 130
KRED NADH-
101
0
64
64
74
28
36
36
26
72
0.57
0.55
0.36
2.56
ND
ND
ND
Undetected
ND
ND
ND
> 99.5%**
< 0.5%
0
< 0.5%
KRED NADH-
110
0
66
34
0.52
ND
ND
KRED P1-A04
KRED P1-B02
KRED P1-B05
KRED P1-B10
KRED P1-B12
KRED P1-C01
KRED P1-H08
KRED P1-H10
KRED P2-B02
KRED P2-C02
KRED P2-C11
KRED P2-D03
KRED P2-D11
KRED P2-D12
KRED P2-G03
KRED P2-H07
KRED P3-B03
KRED P3-G09
KRED P3-H12
GDH
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
16
80
51
27
15
14
54
53
68
60
59
86
56
60
60
87
79
16
37
32
2
20
49
73
85
86
46
47
32
40
41
14
44
40
40
13
21
84
63
68
82
0.26
0.96
2.64
5.77
6.06
0.86
0.89
0.47
0.65
0.70
0.16
0.80
0.68
0.66
0.15
0.26
5.12
1.71
2.11
37.5
ND
ND
> 99.5**
> 99.5**
> 99.5**
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Undetected
Undetected
Undetected
Undetected
ND
ND
Undetected
Undetected
Undetected
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
> 99.5**
> 99.5**
> 99.5**
> 99.5**
ND = Not determined,
* Enzymes are from Codexis,
** Only this enantiomer peak was seen in chiral HPLC