A Chemoenzymatic Route to Chiral Intermediates Used in the Multikilogram Synthesis of a Gamma Secretase Inhibitor

Article abstract of DOI:10.1021/acs.oprd.7b00096

A chemoenzymatic route for the production of an intermediate to a gamma secretase inhibitor is described. The route is robust and was run at multikilogram scale. The process employs both a transaminase catalyzed reductive amination of a substituted tetralone and an alcohol dehydrogenase catalyzed reduction of an α-ketoester to generate the two chiral centers in the molecule, with nearly perfect stereoselectivity. The process also features simple isolation schemes, including a direct drop isolation of the aminotetralin phosphate salt.

Full text of DOI:10.1021/acs.oprd.7b00096

Article
pubs.acs.org/OPRD
A Chemoenzymatic Route to Chiral Intermediates Used in the
Multikilogram Synthesis of a Gamma Secretase Inhibitor
Michael Burns,*,† Carlos A. Martinez,† Brian Vanderplas,† Richard Wisdom,‡ Shu Yu,†
and Robert A. Singer*,†
†Chemical Research and Development, Pfizer Inc., Eastern Point Road, Groton, Connecticut 06340, United States
‡Euticals GmbH, Industriepark Hoechst, D569, 65926, Frankfurt, Germany
ABSTRACT: A chemoenzymatic route for the production of an intermediate to a gamma secretase inhibitor is described. The
route is robust and was run at multikilogram scale. The process employs both a transaminase catalyzed reductive amination of a
substituted tetralone and an alcohol dehydrogenase catalyzed reduction of an α-ketoester to generate the two chiral centers in the
molecule, with nearly perfect stereoselectivity. The process also features simple isolation schemes, including a direct drop
isolation of the aminotetralin phosphate salt.
retrosynthetic analysis (Scheme 1) the molecule can be made
using various disconnections. Two approaches considered were
the reductive amination of tetralone 1 with norvaline or the
reductive amination of ketoester 3 with tetralin 2. Both
approaches were attempted but were low yielding, especially
using hydrogenation methods. A limited panel of commercial
imine reductase enzymes were screened for the ability to
catalyze the reductive amination, but no reaction was
observed.8 Alternately, the coupling of an activated alcohol,
such as 5, with aminotetralin 2 was investigated.
The option of using biocatalysis to produce 2 and
hydroxyester 4 (Scheme 2) was considered attractive due to
the possibility of achieving high stereoselectivity. The
enzymatic amination of tetralones has previously been reported
using a variety of transaminases. Both (S)-selective and (R)-
selective activity has been demonstrated using transaminases
from Vibrio fluvialis, Chromobacterium violaceum, and Para-
coccus fluorescens, as well as a variant from Arthrobacter citreus.9
Additionally, the asymmetric enzymatic reduction of α-
ketoesters is well-known.10
We favored using biocatalysis because it delivered 2 as a
stable intermediate while the tetralone was prone to oxidation.
The reduction of 3 worked well with alcohol dehydrogenases
and asymmetric chemical reductions (hydrogenation) to afford
4. With 4 in hand we were able to demonstrate alkylation with
the triflate which was successful.
Here we describe the use of a transaminase to convert 1 to 2
and an alcohol dehydrogenase to convert 3 to 4. After
activating 4 as a triflate, these chiral intermediates are coupled
to form 6 which is the key building block in a route for the
synthesis of 7, a selective gamma secretase inhibitor with
potential antitumor activity (Scheme 2).
INTRODUCTION
■
Over the past two decades, biocatalysis has become an
important technology for the pharmaceutical, fine chemical,
and agro industries. Many examples of robust and scalable
enzymatic processes have enabled process chemists to gain a
level of familiarity with, and trust in biocatalytic reactions that
was previously absent. Biocatalysis has been demonstrated in
the commercial scale production of several blockbuster
pharmaceuticals such as Lipitor, Lyrica, Xalkori, and Januvia.1−4
Biocatalysis is a useful tool for chemical synthesis due to the
many attractive properties of enzymes, including the ability to
operate under mild reaction conditions, and chemo- and regio-
and stereoselectivity. Enzymes are also highly regarded as green
reagents and are readily produced from renewable starting
materials and are biodegradable.
Many shortcomings of biocatalysts have recently been
alleviated. These include the availability of enzymes in
commercial quantities, the increased number of enzyme classes
available with a broad substrate specificity, and the ability to
engineer enzymes that are highly stable under desirable process
conditions.
Advances in related fields have dramatically improved the
utility of enzymes in process chemistry. Efficient and low-cost
rapid DNA sequencing and DNA synthesis, as well as novel
bioinformatics and protein engineering techniques, have
enabled access to a wide selection of robust enzymes that can
readily be used in process chemistry.
Historically, because of their availability, hydrolase enzymes
produced for the textile, paper, and food industries were the
most commonly used biocatalysts. Over the past two decades
new classes of enzymes such as nitrilases, aldolases, alcohol
dehydrogenases, and transaminases have become reliable tools.
Thanks to the continuing advances in the field, we can look
forward to industrially useful enzymes such as imine reductases,
amine dehydrogenases, and oxidases. Some recent reviews offer
an excellent description of the current state of the field of
biocatalysis.5−7
This paper is focused on the synthesis of compound 6, an
intermediate for the gamma secretase inhibitor 7. From a
Received: March 13, 2017
© XXXX American Chemical Society
A
DOI: 10.1021/acs.oprd.7b00096
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
Scheme 1. Retrosynthetic Analysis
Scheme 2. Route to 6 Using Transaminase and Alcohol Dehydrogenase Enzymes
RESULTS AND DISCUSSION
Table 1. Results from a Screen of 224 Transaminase
Enzymes for the Conversion of 1 to 2
■
a
Based on our experience with using transaminases to produce
chiral amines with excellent stereoselectivity, we decided to
screen our enzyme collection for the conversion of 1 to 2.
% ee
vendor
c-LEcta
Johnson Matthey JM-TA-120
enzyme name
(S)-2
% conversion
ATA-47
>99%
>99%
>99%
>99%
>99%
>99%
>99%
>99%
>99%
32.9
21.1
18.4
14.2
14.0
11.9
6.9
evocatal
Asymchem
Codexis
Almac
1.2.104
116890
ATA-256
TAm-105
Asymchem
Codexis
Pfizer
116889
The ATA-47 enzyme from c-LEcta showed the most promise
in the initial screen (Table 1). A rescreen with five additional
transaminases from c-LEcta that were not included in the initial
screen, but were known to have selectivity similar to ATA-47,
were tested with a higher substrate concentration. In this
screen, the substrate was dissolved in methanol to facilitate
addition to the reactions. Alternatively, a more enzyme friendly
cosolvent such as DMSO could have been used. The results are
shown in Table 2.
ATA-47 showed the best enantioselectivity, whereas ATA-69
showed the highest activity. Further tests with these two
enzymes were carried out to confirm their activities and
selectivities at different enzyme−substrate ratios and with
varying amounts of t-amyl alcohol added as described in Table
3.
ATA-260
6.0
A. denitrificans (AAP92672)
enzyme name
4.7
vendor
% ee (R)-2 % conversion
Pfizer
A. terreus (XP_001209325.1)
ATA-72
>99%
>99%
>99%
>99%
>99%
>99%
>99%
>99%
>99%
>99%
29.4
16.4
15.4
12.7
9.0
c-LEcta
c-LEcta
Codexis
Almac
ATA-71
ATA-P2-A01
TAm-901
Codexis
c-LEcta
Pfizer
ATA-117
8.9
ATA-73
8.8
A. fumigatus (XP_748821)
ATA-P2-A01
7.7
Codexis
Pfizer
7.1
A. oryzae (XP_001818566)
5.6
a
Identities of the transaminases that are not from vendors are
indicated by their accession numbers. Conditions: ketone, 2 mg/mL;
enzyme, 2 mg/mL, DMSO, 2%; isopropylamine hydrochloride, 2
equiv; 100 mM potassium phosphate with 3 mM pyridoxal phosphate,
pH 7.5, 40 °C, 18 h.
The reactivity in the presence of cosolvents was evaluated
due to the poor solubility of 1 in aqueous reaction buffer.
Addition of toluene, cyclohexane, isopropyl acetate, t-amyl
alcohol, dimethylformamide (DMF), or methanol (at 3%) was
tested. Another solvent, DMSO, which is commonly used in
B
DOI: 10.1021/acs.oprd.7b00096
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
a
The transamination reaction was run using ATA-47 to
produce the desired (S)-2 as a phosphate salt. The reaction was
conducted in an aqueous isopropylamine phosphate buffer
supplemented with the required pyridoxal phosphate cofactor
without any added cosolvent. A key finding during process
development was the need for a stream of nitrogen overlaying
the reaction. This prevented oxidation of the tetralone and also
removed the acetone coproduct, thus ensuring the reaction
proceeded to completion. As described above, the aminotetralin
phosphate salt precipitated as it was formed; therefore, product
isolation was achieved by a simple filtration followed by water
washes to remove the enzyme, pyridoxal phosphate, and
isopropylamine buffer. Acetone washes were then carried out to
remove soluble organic impurities and traces of unreacted 1.
After drying, the aminotetralin phosphate salt was isolated with
95% yield and >99% purity and purity (HPLC, 210 nm a/a).
The undesired (R)-2 was not detected.
Table 2. Screen of Additional Transaminase Enzymes
enzyme
mass
conversion at
16 h (HPLC a/
a %)
enzyme
ATA-40
(mg)
ee (%)
comments
30
68
97 on running a further 24 h
went to 98%
conversion
ATA-47
30
58
>99% on running a further 24 h
went to 98%
conversion
ATA-50
ATA-64
20
20
12
12
43 not suitable
98 minimal further
conversion with
increased reaction
time.
ATA-69
ATA-81
20
20
99
5
86 good activity
74 not suitable
a
Conditions: 10 mg/mL substrate concentration in phosphate buffer,
pH 7.8, 30 °C with 3.3% methanol cosolvent.
To assess the optical purity of (S)-2, the (R)-2 was prepared
as an analytical standard. The transaminase from Aspergillus
terreus was produced through fermentation, and a clarified
lysate was used as the biocatalyst. This transaminase was used
to convert 14 g of 1 to 19 g of (R)-2 under similar conditions to
those used to make the (S)-2. The product was isolated as the
phosphate salt with an isolated yield of 89%. No (S)-2 was
detected.
With the aminotetralin in hand, the next step was to prepare
3 that is the substrate for the chiral reduction using an alcohol
dehydrogenase. 3 was prepared from n-propyl Grignard added
to a solution of di-t-butyl oxalate in MTBE and n-heptane at
−80 °C (Scheme 3). Analysis of the reaction showed that 2.8%
Table 3. Screen of ATA-69 and ATA-47 at a 10 g/L Substrate
Concentration, Varying Enzyme Load, and Varying t-Amyl
Alcohol Concentrations
enzyme ketone
t-amyl
alcohol
mass (mg)
conversion
(HPLC a/a
%)
mass
(mg)
mass
(mg)
ee
enzyme
(%) comments
ATA-
69
1.6
3.2
3.2
3.0
6.9
6.9
29
29
60
60
60
61
100
100
160
80
82 (66 h)
93 (66 h)
low
89 28% conv
16 h
ATA-
69
89 65% conv
16 h
ATA-
69
<1% conv
16 h
ATA-
69
40 (42 h)
58 (42 h)
7 (42 h)
86
Scheme 3. Preparation of 3
ATA-
47
80
100
ATA-
47
160
transaminase reactions is an option that may be investigated in
future development efforts. At this concentration, the addition
of t-amyl alcohol and DMF resulted in a marginal increased
activity of ATA-47 over the use of methanol, whereas toluene
and isopropyl acetate hindered the reaction. In all cases the very
high enantioselectivity of the enzyme was unchanged by the
addition of these solvents. Interestingly, however, further tests
with and without added solvent showed that the 2-phase
(solid/liquid) reaction proceeded efficiently to completion
without the need for the addition of any solvents to help
solubilize the tetralone. Conveniently, it was observed that the
product aminotetralin 2 does precipitate as the phosphate salt
as it was formed, resulting in a slurry to slurry transamination
process step.
Despite a higher activity, the suboptimal enantioselectivity of
ATA-69 was confirmed, and this enzyme was not evaluated in
further studies. Further work was continued with ATA-47 and
ATA-40.
Further studies with ATA-47 and ATA-40 included an
examination of enzyme loading, reaction temperature, pH, and
isopropylamine concentration (up to 1.8 M). The performance
of the two enzymes were comparable in except that ATA-47
consistently gave product with >99% ee, whereas ATA-40
produced a lower quality product having an ee of 97%. Based
on the desire to maintain a high % ee, ATA-47 was chosen for
scale-up.
of the racemic 4 was formed as a byproduct. This racemic 4 is
believed to be formed by reduction of the enol form of 3 to the
racemate 4 during quench. The origin of the hydride could be
from a beta-hydride elimination of excess propylmagnesium
chloride from which the magnesium delivers the hydride to the
ketone or in a Michael type fashion to the enol form of the
ketone. The level of the racemic 4 formed was greatly reduced
by switching the quench from a slow addition of the aqueous
solution to a fast addition of the reaction mixture into aqueous
citric acid to immediately neutralize anionic species and avoid
transfer of a hydride to the ketone. This impurity was
impossible to avoid and its separation from the 3 by fractional
distillation also proved to be very difficult. However, since it
had been observed that low concentrations of the undesired
diastereomer could be readily removed during the crystal-
lization of 6, the ketoester oil was used directly in the next step
without further purification.
A total of 190 alcohol dehydrogenases were screened for the
reduction of 3 to 4. These were all obtained from commercial
sources. Enzymes that produced (S)-4 or (R)-4 with high
selectivity were identified. The best hits are listed in Table 4. In
this screen, the NAD(P)H cofactor was recycled using glucose
and glucose dehydrogenase.
C
DOI: 10.1021/acs.oprd.7b00096
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(Scheme 4). The reaction was nearly complete (>95%) within
6 h but was left to stir overnight at which point the residual 3
Table 4. Results of the Screen of 190 Alcohol
Dehydrogenases for the Conversion of 3 to 4
a
vendor
enzyme
% ee (R)-4
% conversion
Scheme 4. Reduction of 3 to 4
Codexis
c-LEcta
c-LEcta
Almac
KRED-P1-B10
ADH-107L
ADH-108L
A161
>99
100
100
>99
>99
100
>99
100
Almac
A411
>99
100
Almac
A631
>99
100
Almac
A671
>99
100
Almac
AR-108
>99
100
Codexis
Codexis
KRED-NADH-110
KRED−P1-B12
enzyme
95.3
95.2
% ee (S)-4
100
100
vendor
% conversion
was <0.1%. After extraction, washing, and solvent stripping, the
4 recovered as an oil (29.2 kg, 88% yield, 96.7% ee). Since the
starting ketoester had 2.8% racemic-4, this represents an
enantioselectivity over the enzymatic reaction of 99.5%.
To assess the optical purity of (R)-4, (S)-4 was also
produced as an analytical standard. One gram of 3 was reduced
using Almac CRED A641 to produce (S)-4 with an isolated
yield of 77% and a stereopurity of 92.1% ee. The NADPH
cofactor was recycled using glucose dehydrogenase. Control
reactions showed that background KRED activity in the glucose
dehydrogenase preparation was responsible for the production
of (R/S)-4, and this accounted for the lower than expected
stereopurity.
The free amine from 2 phosphate salt was prepared by
suspension in aqueous sodium hydroxide and extraction with
MTBE. The MTBE solvent was displaced with DCM, and the
solution of 2 was stored cold under nitrogen until required. The
triflate 5 was formed by slowly adding trifluoromethanesulfonic
anhydride to a chilled solution of 4 in DCM and
diisopropylethylamine. The free 2 was charged to the solution
of 5 and the mixture was allowed to warm to 30 °C. This
reaction was quenched with potassium bicarbonate and the
phases separated. The solvent was stripped under vacuum and
displaced with 1,4-dioxane to precipitate the diisopropylethyl-
amine triflate salt which was removed by filtration. The HCl
salt of the coupled product 6 in 1,4-dioxane was formed by the
addition of water and HCl gas. Prior to HCl addition, water was
added to the 1,4-dioxane to a concentration of 2.5% as it had
been shown that this assisted in solubilizing impurities and
enabled a purer product to be formed during crystallization.
The stereochemical integrity of 2 and 4 was verified upon
intercepting 6 which is an intermediate common to both the
current route as well as the original early development/
Medchem route.11
Almac
A641
>99
91.9
88.8
88.4
88.1
86.3
100
100
100
100
100
100
Codexis
KRED-P3-G09
JM-ADH-104
KRED−P1-B04
KRED−P3-C09
117297
Johnson Matthey
Codexis
Codexis
Asymchem
a
Conditions: ketone, 2 mg/mL; enzyme, 2 mg/mL, DMSO, 2%,
glucose, 1.2 equiv, NAD+, 0.1 mg/mL, NADP+, 0.1 mg/mL, glucose
dehydrogenase, 0.1 mg/mL; 100 mM potassium phosphate buffer with
2 mM MgCl2, pH 7.0, 30 °C, 18 h.
On the basis of cost, availability, and performance, the c-
LEcta enzymes, ADH-107L and ADH-108L, were selected for
further development. In initial tests, isopropanol was used for
recycling the cofactor, which has advantages over the use of
glucose dehydrogenase, since there is no need for a second
enzyme and there is no requirement for pH control. With
isopropanol, ADH-107L proved the most active; however, at
increased concentrations of 3 and isopropanol it was found that
increasing amounts of an undesired late eluting (on GC)
impurity were formed and the selectivity of the enzyme was
marginally reduced. The use of ADH-108L with isopropanol
was unsuitable as it was unstable to higher concentrations of
this solvent. Since the decision to switch to a glucose/glucose
dehydrogenase cofactor recycle system was chosen, the identity
of the late eluting peak was not pursued. An option that was
not explored was the rescreening of enzymes using substrate
coupled cofactor recycle with isopropanol serving as the
reductant. This may have identified enzymes that were more
tolerant of increased concentrations of isopropanol.
Further evaluation of ADH-108L with a glucose/glucose
dehydrogenase cofactor recycle system demonstrated higher
activity and stability when compared to ADH-107L with the
isopropanol system. Under these conditions, it was found to be
7 fold more active per unit mass than ADH-107L. ADH-108L
with the use of glucose/glucose dehydrogenase for cofactor
recycle was selected for scale-up development. The conditions
of temperature, cofactor concentration, ketoester/glucose
concentration, pH range, and enzyme loading were investigated
with the aim of developing a reaction that was robust and
reproducible with respect to both conversion and selectivity
with minimal enzyme loading.
CONCLUSION
■
The work described in this paper uses transaminases and
alcohol dehydrogenases to deliver two target chiral compounds
with a high stereopurity. The availability of a large selection of
commercially available enzymes for both of these classes of
enzymes made the identification of potentially useful enzymes
facile. The use of bioinformatics enabled the establishment of
an in-house collection of transmainases, including the
identification of the (R)-selective transaminase that was used
in the production of the undesired (R)-2 which was used for
synthesis of analytical standards.12 The stability of the
commercially available enzymes enabled their use under
process conditions at the multikilogram scale.
Following optimization, 3 was reduced with ADH-108L to
produce the desired (R)-4. The reaction was conducted at 24
°C and at an initial pH of 7.0, in phosphate buffer. The NADH
cofactor was recycled using glucose dehydrogenase which
converted glucose to gluconic acid, and the pH was maintained
by periodic titration with concentrated sodium hydroxide
D
DOI: 10.1021/acs.oprd.7b00096
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
reaction for preparation of ketoester 3. A stainless steel
Grignard reaction vessel was charged with magnesium turnings
(13.4 kg, 1.08 equiv), a solution of n-propylmagnesium chloride
in THF prepared in the laboratory to help with the initiation of
the reaction (0.7 kg) and methyl THF (155 kg). The mixture
was brought to reflux, and 1-chloropropane was charged (2.4
kg). A sample after 2.5 h showed that the reaction had not yet
started, and an extra 0.6 kg of a solution of n-propylmagnesium
chloride in THF prepared in the laboratory was added. After 6
h, the reaction had started, and the remainder of the 1-
chloropropane (37.6 kg) was slowly added over 90 min. The
reaction was held at reflux for a further 2.5 h and then cooled to
20−25 °C. A sample the next morning showed nondetectable
levels of 1-chloropropane, and the solution was transferred via a
tube with a filter to a low temperature charging vessel.
EXPERIMENTAL SECTION
■
Analytical Methods. Methods were as follows: chiral GC
for 3 and 4; column, Agilent CP-cyclodextrin-β-2,3,6-M-19 (25
m × 0.25 mm × 0.25 μm); constant flow with helium at 1.2
mL/min; gradient, 60−150 °C at 10 °C/min; FID detection.
UPLC Method for 1 and 2 proceeded as follows. Before
analysis, the quenched, clarified samples were derivatized with
Marfey’s Reagent (CAS 95713-52-3).13 A 50 μL sample was
mixed with 10 μL of aqueous 1 M sodium bicarbonate and 200
μL of Marfey’s Reagent (5 g/L in acetonitrile) for 60 min at 40
°C. The reaction was quenched with 10 μL of 1 M HCl and
diluted with 230 μL of acetonitrile. Column, Waters BEH C18,
2.1 × 50 mm, 1.7 μm, part no. 186002350; column
temperature, 45 °C; flow rate, 0.4 mL/min; mobile phase,
0.1% aqueous TFA/ACN; gradient, t = 0 min 95:5, t = 5 min
5:95; UV detection at 340 nm.
Enzymes and Reagents. Aminotransaminase (ATA-47),
alcohol dehydrogenase (ADH-108L), and glucose dehydrogen-
ase (GDH-03L) were from c-LEcta GmbH, Leipzig, Germany.
Pyridoxal phosphate monohydrate (CAS 41468-25-1) was from
Boaray Ltd., Shanghai China. NAD hydrate (CAS 53-84-9,
product code OR301) was supplied by Europa Bioproducts,
Ely, Cambridge, UK. 6,8-Difluoro-2-tetralone (CAS 843644-23-
5) was prepared to order by Usun Fine Chemical Products Ltd.,
Nanjing, China and by Ash Ingredients Inc., GlenRock, NY,
USA. Di-t-butyl oxalate (CAS 691-64-5) was from Aceto
France SAS, Paris, France. 1-Chloropropane (CAS 540-54-5,
product code C0266) was purchased from TCI Deutschland
GmbH, Eschborn, Germany. Trifluoromethanesulfonic anhy-
dride (CAS 358-23-6) was from Central Glass, Halle/
Westfallen, Germany. Glycerin (CAS 56-81-5, Gycamed
99.7% grade) was obtained through Univar GmbH, Essen,
Germany. Glucose monohydrate (CAS 50-99-7) was produced
by Cargill (grade C*Dex 02001) and obtained through
Chemische Fabrik GmbH, Moenchengladbach, Germany.
Isopropylamine (CAS 75-31-0, produced by Oxea) was
obtained from IMCD Deutschland GmbH, Koeln, Germany.
N,N-Diisopropylethylamine (CAS 7087-68-5, 98% technical
grade) was obtained from BASF, Ludwigshafen, Germany. All
other materials were standard plant quality.
The low temperature vessel was charged with di-t-butyl
oxalate (37.5 kg, 1 equiv), MTBE (167 kg), and n-heptane (154
kg). The solution was cooled to −83 °C, and the freshly
prepared Grignard solution (1.05 equiv) was added over 105
min, maintaining the temperature between −79 and −83 °C. A
sample taken 1 h after complete addition showed 96%
conversion to the desired 3. The reaction was quenched by
the direct addition of a 33% solution of acetic acid (2 equiv) in
MTBE maintaining the reaction mixture at ≤ −80 °C. The
reaction mixture was then warmed to −20 °C and transferred
to another (glass-lined) reactor. The low temperature vessel
and transfer lines were washed through with 30 kg MTBE.
When the mixture content had warmed to 5 °C, water was
added during which the temperature rose to 16 °C. The phases
were separated at 19 °C. The aqueous phase (pH 4.5) was
discarded, and the organic phase was further washed with 10%
aqueous sodium chloride which was separated and discarded.
The organic phase was given a final wash with 7.5 L of water,
which was also separated and discarded. The organic phase was
stripped under vacuum to give 33.8 kg of oil with a combined
purity of 85.7% (GC a/a), based on 3 (including the enol ester)
plus racemic-4. This equates to a 90.7% yield from starting di-t-
butyl oxalate. 13C NMR (100 MHz, CDCl3): δ 195.7, 160.8,
1
83.8, 40.9, 27.7, 16.6, 13.5. H NMR (CDCl3, 400 MHz): δ
2.67 (t, J = 8.0 Hz, 2H), 1.61−1.54 (m, 2H), 1.48 (s, 9H), 0.90
(t, J = 7.1 Hz, 3H).
Screening of Alcohol Dehydrogenase Enzymes. The
screening of the enzymes was carried out in 96-well format.
Plates contained 1 mg of enzyme in 20 uL of phosphate buffer.
Each well was charged with 100 mM potassium phosphate, 2
mM MgCl2 buffer, pH 7.0 (500 uL); ketoester 3 (1 mg) in
DMSO (10 uL); glucose dehydrogenase Codexis CDX901
(0.05 mg); NADP+ (0.05 mg); NAD+ (0.05 mg); and glucose
(1.3 mg). The plates were incubated shaking at 30 °C overnight
and quenched with 1.0 mL of ethyl acetate. The organic phase
was analyzed using chiral GC.
Screening of Transaminase Enzymes. The screening of
the enzymes was carried out in 96-well format. Plates contained
1 mg of enzyme in 20 μL of phosphate buffer. Each well was
charged with 100 mM potassium phosphate buffer, pH 7.5 (500
μL); 1 (1 mg) in DMSO (10 μL); pyridoxal phosphate (0.37
mg); and isopropylamine hydrochloride (1.05 mg). The plates
were incubated shaking at 40 °C overnight and quenched with
0.5 mL of acetonitrile. The mixtures were centrifuged to clarify
and analyzed using uPLC after derivatization with Marfey’s
Reagent (CAS 95713-52-3).
Preparation of (S)-6,8-Difluoro-1,2,3,4-tetrahydro-
naphthalen-2-amine 2. Aqueous isopropylamine/potassium
phosphate buffer, pH 7.3, was prepared by adding water (718
kg), 85% phosphoric acid, (107.2 kg), isopropylamine (93 kg,
11.6 equiv), and 45% potassium hydroxide (4.5 kg) to the tank.
8 L of this buffer was removed, and 6,8-difluorotetralone (24.8
kg, 1 equiv) was added. The transaminase enzyme, c-LEcta,
ATA-47 (110 g), and pyridoxal phosphate monohydrate (861
g) were each prepared as two separate solutions/suspensions in
4 L of buffer and added to the reactor at room temperature
after the 6,8-difluorotetralone addition. The biotransformation
was run at 37−39 °C with a constant stream of nitrogen
sparged through the reaction suspension to remove the
acetone. After 3 days the reaction showed <0.5% starting 1
and was cooled to 20−25 °C, filtered, and washed four times
with water (1000 kg in total), followed by three washes with
acetone (480 kg in total). The amine salt was then dried on the
filter by passing a stream of nitrogen through the bed at room
temperature.
A total of 30.9 kg of salt was recovered. The isolated yield
over this step was 94%. An analysis showed nondetectable
levels of (R)-2 (after derivatization with Marfey’s reagent). The
Preparation of tert-Butyl 2-oxopentanoate 3. The
Grignard solution was prepared and used directly in the
E
DOI: 10.1021/acs.oprd.7b00096
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
1
HPLC purity (a/a 210 nm) was 99.5%. H NMR (DMSO-d6,
ature below −20 °C. After complete addition, the reaction was
stirred for 1 h after which a sample showed complete
conversion of the added trifluoromethanesulfonic anhydride.
Then free aminotetralin 2/DCM solution (145 kg, 1.0 equiv)
was added; the reaction was warmed to 30 °C, stirred
overnight, and continued until a sample showed <2% residual
free 2 remaining. At this point, HPLC showed 94.4% product 6
and 1.9% incorrect diastereomer (ratio 98.0:2.0). The reaction
was quenched by the addition of a 12% potassium bicarbonate
aqueous solution (210 kg) after which the phases were
separated. The DCM/product containing phase was stripped
under vacuum and the solvent exchanged to 1,4-dioxane at a
sump temperature ≤30 °C. The diisopropylethylamine triflate
salt precipitated and was filtered at 20−25 °C. The product 6 in
1,4-dioxane was returned to the vessel and the water content
adjusted to 2.5%. HCl gas (3.6 kg) was then added, maintaining
the temperature between 24 and 29 °C, until a sample showed
a pH of approximately 3. The temperature was then adjusted to
25 °C and the product filtered. The filter cake was washed 4
times with 1,4-dioxane (196 kg) and partially dried under a
stream of nitrogen to give 41.8 kg damp product. HPLC
analysis showed 99.3% and 99.4% purity by HPLC (a/a; 210
nm) with 0.1% of the incorrect diastereomer. The coupled
product 6 was dried in a tray drier at 40−45 °C, high vacuum,
and under a stream of nitrogen to yield a total of 32.1 kg of 6
(80% overall yield on tetralone 1 and 61% overall yield on di-t-
butyl oxalate). 1H NMR (400 MHz, DMSO-d6) δ: 9.95 (s, 1H),
9.36 (s, 1H), 7.07−7.00 (m, 1H), 6.90−6.85 (m, 1H), 4.13 (s,
1H), 3.48−3.35 (m, 1H), 3.27−3.18, (m, 1H), 2.97−2.75 (m,
3H), 2.30−2.18 (m, 1H), 2.00−1.78 (m,3H), 1.45 (s, 9H),
1.44−1.25 (m, 2H), 0.91 (t, 3H).
400 MHz): δ 6.99−6.75 (m, 2H), 3.10−2.95 (m, 2H), 2.90−
2.65 (m, 4H), 2.30−2.15 (m, 1H), 1.90−1.75 (m, 1H), 1.50−
1.35 (m, 1H).
Fermentation To Produce (R)-Selective Transaminase.
The gene for a transaminase from Aspergillus terreus was
synthesized, cloned into the pJExpress411 vector, and trans-
formed into E. coli BL21(DE3) cells. These cells were cultured
in Terrific Broth and induced with IPTG in order to express
soluble transaminase enzyme. The cells were harvested by
centrifugation, suspended in buffer, and lysed using homoge-
nization. The homogenate was clarified by centrifugation, and
the resulting crude lysate was used as the biocatalyst to prepare
(R)-2.
Preparation of tert-Butyl-(R)-2-hydroxypentanoate 4.
The reaction was carried out in an aqueous sodium phosphate
buffered solution without the addition of any cosolvent. The
formed gluconic acid was neutralized by the periodic addition
of sodium hydroxide as required. To the vessel was added 0.3
M phosphate buffer, pH 7.0 (300 kg), followed by glucose
monohydrate (61 kg, 1.8 equiv) and glycerol (135 kg). The
alcohol dehydrogenase, c-LEcta ADH-108L (56 g), glucose
dehydrogenase, c-LEcta GDH-03L (66 g), and NAD+ hydrate
(130 g) were then added as a preprepared solution in buffer.
The starting 1 (35 kg, 1 equiv) was added, and the reaction was
performed at 22−25 °C. Samples were taken every 30−60 min,
the pH measured, and 33% sodium hydroxide added as
required. The reactions were nearly complete after 5−6 h, after
which the reactions were stirred overnight for complete
conversion. The reaction was worked up by extraction into
MTBE (three extractions with a total of 550 kg MTBE)
followed by two water back-washes with 40 L of water each
wash. The MTBE solution was then stripped under vacuum to
give an oil. The final distillation conditions were 10−20 mbar
vacuum, with a sump temperature 36−40 °C. The oil was
purified using wiped-film distillation. The wall temperature was
90−93 °C; the vacuum was 6−8 mbar, and the cooling bath
temperature was 15−20 °C. A total of 29.1 kg distilled oil was
obtained with a corrected yield of 88%. The % ee of the t-butyl
AUTHOR INFORMATION
Corresponding Authors
*E-mail: Michael.p.burns@pfizer.com.
*E-mail: Robert.a.singer@pfizer.com.
■
ORCID
Michael Burns: 0000-0002-7414-3704
Notes
The authors declare no competing financial interest.
1
(R)-2-hydroxypentanoate was 96.7%. H NMR (400 MHz,
CDCl3) δ: 4.20−4.00 (m, 1H), 2.80 (s, 1H), 1.85−1.65 (m,
2H), 1.46 (s, 9H), 1.45−1.31 (m, 2H), 0.85 (s, 3H).
REFERENCES
Coupling of (S)-6,8-Difluoro-1,2,3,4-tetrahydronaph-
thalen-2-amine 2 and tert-Butyl-(R)-2-hydroxypenta-
noate 4. The free amine (24.7 kg) was released from the
phosphate salt by suspending it in water (237 kg) plus MTBE
(690 kg) and then adding 33% sodium hydroxide (14.9 kg) to
pH 10.8−11.2. The phases were separated and the aqueous
phase re-extracted two more times with MTBE. The combined
MTBE phases were back-washed with a little water, and the
phases were separated and then dried by the addition of sodium
sulfate (38 kg). The sodium sulfate was filtered off and washed
with a little MTBE. The combined MTBE phases were stripped
under vacuum and at a temperature <30 °C to an oil, and then
dichloromethane (DCM) was added. The aminotetralin 2/
DCM solution was stored at 0−5 °C, under nitrogen until
required. The amine solution, had an HPLC purity of 99.5% a/
a (210 nm), and the DCM:MTBE ratio was 95:5.
■
(1) Martinez, C. A.; Hu, S.; Dumond, Y.; Tao, J.; Kelleher, P.; Tully,
L. Development of a chemoenzymatic manufacturing process for
pregabalin. Org. Process Res. Dev. 2008, 12 (3), 392−398.
(2) Patel, R. N. Green Biocatalysis; John Wiley & Sons, 2016.
(3) Savile, C. K.; Janey, J. M.; Mundorff, E. C.; Moore, J. C.; Tam, S.;
Jarvis, W. R.; Colbeck, J. C.; Krebber, A.; Fleitz, F. J.; Brands, J.
Biocatalytic asymmetric synthesis of chiral amines from ketones
applied to sitagliptin manufacture. Science 2010, 329 (5989), 305−309.
(4) Debarge, S.; Erdman, D. T.; O’Neill, P. M.; Kumar, R.;
Karmilowicz, M. J. Process and intermediates for the preparation of
pregabalin. WO2014155291A1, 2014.
(5) Torrelo, G.; Hanefeld, U.; Hollmann, F. Biocatalysis. Catal. Lett.
2015, 145 (1), 309−345.
(6) Reetz, M. T. Biocatalysis in organic chemistry and biotechnology:
past, present, and future. J. Am. Chem. Soc. 2013, 135 (34), 12480−
12496.
(7) Bornscheuer, U.; Huisman, G.; Kazlauskas, R.; Lutz, S.; Moore, J.;
Robins, K. Engineering the third wave of biocatalysis. Nature 2012,
485 (7397), 185−194.
(8) Schrittwieser, J. H.; Velikogne, S.; Kroutil, W. Biocatalytic imine
reduction and reductive amination of ketones. Adv. Synth. Catal. 2015,
357 (8), 1655−1685.
Hydroxyester 4 (21.6 kg, 1.11 equiv corrected), DCM (570
kg), and diisopropylethylamine (29.1 kg, 2.18 equiv) were
charged to a low temperature vessel and the contents cooled to
−20 to −25 °C. Trifluoromethanesulfonic anhydride (31.1 kg,
1.07 equiv) was then slowly added, maintaining the temper-
F
DOI: 10.1021/acs.oprd.7b00096
Org. Process Res. Dev. XXXX, XXX, XXX−XXX

Organic Process Research & Development
Article
(9) Pressnitz, D.; Fuchs, C. S.; Sattler, J. H.; Knaus, T.; Macheroux,
P.; Mutti, F. G.; Kroutil, W. Asymmetric Amination of Tetralone and
Chromanone Derivatives Employing ω-Transaminases. ACS Catal.
2013, 3 (4), 555−559.
(10) Pennacchio, A.; Giordano, A.; Rossi, M.; Raia, C. A. Asymmetric
Reduction of α-Keto Esters with Thermus thermophilus NADH-
Dependent Carbonyl Reductase using Glucose Dehydrogenase and
Alcohol Dehydrogenase for Cofactor Regeneration. Eur. J. Org. Chem.
2011, 2011 (23), 4361−4366.
(11) Brodney, M. A.; Auperin, D. D.; Becker, S. L.; Bronk, B. S.;
Brown, T. M.; Coffman, K. J.; Finley, J. E.; Hicks, C. D.; Karmilowicz,
M. J.; Lanz, T. A. Design, synthesis, and in vivo characterization of a
novel series of tetralin amino imidazoles as γ-secretase inhibitors:
Discovery of PF-3084014. Bioorg. med. chem. lett. 2011, 21 (9), 2637−
2640.
(12) Hohne, M.; Schatzle, S.; Jochens, H.; Robins, K.; Bornscheuer,
̈
̈
U. T. Rational assignment of key motifs for function guides in silico
enzyme identification. Nat. Chem. Biol. 2010, 6 (11), 807−813.
(13) Bhushan, R.; Bruckner, H. Marfey’s reagent for chiral amino acid
̈
analysis: a review. Amino Acids 2004, 27 (3−4), 231−247.
G
DOI: 10.1021/acs.oprd.7b00096
Org. Process Res. Dev. XXXX, XXX, XXX−XXX