Process development on the enantioselective enzymatic hydrolysis of S-ethyl 2-ethoxy-3-(4-hydroxyphenyl)propanoate

Article abstract of DOI:10.1021/op0256035

A novel biocatalytic approach for the large-scale production of S-2-ethoxy-3-(4-hydroxyphenyl)propanoic acid S-1 from its racemic ethylester rac-2 by enantioselective hydrolysis has been developed. S-1 is an important building block in the synthesis of PPARα and -γ agonists such as Ragaglitazar [NNC 61-0029 ((-)DRF2725)]. The development history comprises enzyme screening, biocatalyst and process optimization, and scale-up to pilot plant. The project was thereby highly interdisciplinary by combining biotechnology and chemistry technologies. The final process was successfully run on a 44-kg pilot scale in 43-48% yields and with high enantiomeric purities (98.4-99.6% ee).

Full text of DOI:10.1021/op0256035

Organic Process Research & Development 2003, 7, 82−88
Lecture Transcripts
Process Development on the Enantioselective Enzymatic Hydrolysis of S-Ethyl
2
-Ethoxy-3-(4-hydroxyphenyl)propanoate
,
†
†
†
†
‡
Heinz-Josef Deussen,* Magali Zundel, Marine Valdois, Søren Vig Lehmann, Volker Weil, Carsten Mailand Hjort,
‡
‡
†
Peter Rahbek Østergaard, Erik Marcussen, and Søren Ebdrup
NoVo Nordisk A/S, Research & DeVelopment, NoVo Nordisk Park, 2760 MaaloeV, Denmark, and NoVozymes A/S,
Molecular Biotechnology, 2880 BagsVærd, Denmark
Abstract:
Scheme 1. Synthesis of Ragaglitazar [NNC 61-0029
(-)DRF2725)]
(
A novel biocatalytic approach for the large-scale production
of S-2-ethoxy-3-(4-hydroxyphenyl)propanoic acid S-1 from its
racemic ethylester rac-2 by enantioselective hydrolysis has been
developed. S-1 is an important building block in the synthesis
of PPARr and -γ agonists such as Ragaglitazar [NNC 61-0029
((-)DRF2725)]. The development history comprises enzyme
screening, biocatalyst and process optimization, and scale-up
to pilot plant. The project was thereby highly interdisciplinary
by combining biotechnology and chemistry technologies. The
final process was successfully run on a 44-kg pilot scale in 43-
48% yields and with high enantiomeric purities (98.4-99.6%
ee).
marketed PPAR agonists and is a member of a new class of
insulin sensitisers. Ragaglitazar [NNC 61-0029] has currently
entered phase III clinical trials for treatment of diabetes
mellitus type 2.
Introduction
S-2-Ethoxy-3-(4-hydroxyphenyl)propanoic acid S-1 is a
key intermediate in the synthesis of the new antidiabetic drug
Ragaglitazar [NNC 61-0029 ((-)DRF2725)].1 Ragaglitazar
Ragaglitazar [NNC 61-0029] can be prepared by the
condensation of 2-[phenoxazin-10-yl]ethylmethane sulpho-
,2
6
nate 3 with the 2-propylester of S-1 in the presence of
[NNC 61-0029] belongs to a novel therapeutic class of
potassium carbonate in refluxing toluene. The acid ragagli-
tazar [NNC 61-0029] can then be obtained by basic hydroly-
compounds of dual acting PPARR and -γ agonists aimed
for treatment of type-2 diabetes. Ragaglitazar is designed
to restore insulin sensitivity and to correct the dyslipidaemic
disorders in people with type-2 diabetes.4 The compound,
in-licensed by Novo Nordisk from Dr. Reddy’s Laboratories,
is chemically and pharmacologically different from presently
3
4
,7
sis of the coupled 2-propylester (Scheme 1). The devel-
opment for the synthesis of the key intermediate, enantio-
,5
8
9,10
merically pure S-1, will be described in this work.
A reported scalable synthetic route for the synthesis of
enantiomerically pure S-1 or its esters uses diazotation of
O-benzylated L-tyrosine to afford its R-hydroxy acid, fol-
lowed by alkylation, with high retention of the chirality.1
However, this process had in the earlier preparations certain
*
Author for correspondence. Telephone: +45 4443 7202. Fax: +45 4443
4
742. E-mail: hjdu@novonordisk.com.
†
Novo Nordisk A/S.
Novozymes A/S.
‡
(1) Siripragada, M. R.; Chebiyyam, P.; Potlapally, R. K.; Batchu, C. S.;
Mamillapally, R. S. PCT Appl. WO 00/26200.
drawbacks such as moderate yields, the formation of about
11
2
0% bi-product, and up to 4% racemisation.
(
2) Lohray, B. B.; Lohray, V. B.; Bajji, A. C.; Kalchar, S.; Poondra, R. R.;
Padakanti, S.; Chakrabarti, R.; Vikramadithyan, R. K.; Misra, P.; Juluri,
S.; Mamidi, N. V. S. R.; Rajagopalan, R. J. Med. Chem. 2002, 44 2675-
(6) Batchu, C.; Mamillapalli, R. S.; Gaddam, O. R. PCT Appl. WO 02/34733.
(7) Despite the excellent results of the process described in this article, another
synthetic pathway using different intermediates was chosen for further large-
scale production.
(8) Other antidiabetic drugs in development use the same building block,
stressing the importance of a straightforward synthesis of the title
compound: Thalen, P.; Jacinto, S.; Hultstrand, T.; Amrot-Fors, L.; Camejo,
G.; Wallin, B.; Svensson, L.; Brockenhuus, H.; Ljung, B.; Oakes, N.
Diabetes 2001, 50 (Suppl. 2), A122; Murphy, G. J.; Holder, J. C. Trends
Pharmacol. Sci. 2000, 21, 469-474.
2
678.
(
3) Sauerberg, P.; Pettersson, I.; Jeppesen, L.; Bury, P. S.; Mogensen, J. P.;
Wassermann, K.; Brand, C. L.; Sturis, J.; W o¨ ldike, H. F.; Fleckner, J.;
Andersen, S.-S.; Mortensen, S. B.; Svensson, L. A.; Rasmussen, H. B.;
Lehmann, S. V.; Polivka, Z.; Sindelar, K.; Panajotova, V.; Ynddal, L.;
Wulff, E. M. J. Med. Chem. 2002, 45 789-804.
(
4) Ebdrup, S.; Pettersson, I.; Rasmussen, H. B.; Deussen, H.-J.; Jensen, A.
F.; Mortensen, S. B.; Fleckner, J.; Pridal, L.; Nygaard, L.; Sauerberg, P. J.
Med. Chem. 2003. In press.
(5) Wulff, E. M.; Jeppesen, L.; Bury, P. S.; Mogensen, J. P.; Fleckner, J.;
Andersen, A.-S. T.; Wassermann, K.; Sauerberg, P. Diabetes 2001, 50
(9) Ebdrup, S.; Deussen, H.-J.; Zundel, M.; Bury, P. S. PCT-Appl. WO 01/
11073.
(Suppl. 2), A524.
(10) Ebdrup, S.; Deussen, H.-J.; Zundel, M. PCT-Appl. WO 01/11072.
8
2
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development
10.1021/op0256035 CCC: $25.00 © 2003 American Chemical Society
Published on Web 12/13/2002

Scheme 2. Pilot-plant synthesis of rac-2
Scheme 3. Enantioselective enzymatic hydrolysis of rac-2
The reaction mixture was then heated with triethyl
1
7
phosphite to 150 °C to give triethyl 2-ethoxyphosphono-
acetate 6 in 96-99% yield (over both steps). 6 was reacted
with 4-benzyloxybenzaldehyde in a Horner-Emmons-
18
Wadsworth reaction using t-BuOK as a base. 3-(4-Benzyl-
oxyphenyl)-2-ethoxyacrylic acid ethyl ester 7 could be
obtained by crystallization from the reaction mixture as an
E/Z-mixture. Hydrogenation of 7 at 2-6 bar yielded rac-2
4
,19
in 77-88% overall yield in >98% purity. All steps were
successfully scaled-up to pilot plant.
Biocatalysis has become a widely used tool in the
pharmaceutical industry to the synthesis of optically active
intermediates.12 Enantioselective hydrolysis of esters and
enantioselective esterification of carboxylic acids using
hydrolases are among the most commonly used methods for
the preparation of optically pure alcohols and esters. This is
due to the fact that hydrolases do not require cofactors, are
Enzyme Screening. More than 80 hydrolases (lipases,
esterases, cutinases, proteases) were screened for enantio-
2
0
selectivity on 2 as a substrate in two screening rounds
(Scheme 3). The initial screening revealed three enzyme
2
1
samples to be highly enantioselective for the S-enantiomer
22
with an enantiomeric ratio E > 200: protease 2 (aspergil-
lopepsin I) from Aspergillus aculeatus, alp protease (Oryzin)
from Aspergillus oryzae, and proteinase 2A from Aspergil-
lus oryzae (Fluka).
13
simple to use, and are now widely commercially available.
It is known from the literature that substituted R-benzyl-
23
14
15
oxypropanoates and R-methoxypropanoates can be enan-
tioselectively hydrolyzed to their corresponding carboxylic
acids by biocatalysis. Athough the first method uses growing
cells and the second method only discloses moderate
enantioselectivities, these results encouraged us to search for
an enzymatic kinetic resolution for the large-scale preparation
of enantiomerically pure S-1 or an ester of S-1 with a target
enantiomeric excess of >98% ee. This work includes
screening for enantioselective enzymatic activity, process and
biocatalyst optimization, as well as scale-up to pilot plant,
which all will be described herein.
However, when the hydrolysis was repeated using highly
purified samples of protease 2, no activity at all was found.
The active enzyme preparation used in the first screening
was shown to contain traces of other secreted proteins from
the production organism, suggesting that one of these proteins
was the enzyme responsible for the observed catalytic
activity. Therefore, a number of additional enzymes, with a
focus on commercially available industrial enzyme prepara-
tions produced by the two Aspergillus strains including
Aspergillus niger were tested in a second screening round.
The results (selected are listed in Table 1) revealed the
desired enzymatic activity to be present in all Aspergillus
Results and Discussion
Synthesis of rac-Ethyl 2-Ethoxy-3-(4-hydroxyphenyl)-
propanoate (rac-2). The synthesis of racemic ethyl 2-ethoxy-
2
4
preparations tested. Although the enantioselectivities of
many of the enzyme preparations were very high, the specific
activities (hydrolytic activity towards S-2/g of enzyme
3
2
-(4-hydroxyphenyl)propanoate rac-2 is depicted in Scheme
. Commercially available 2,2-diethoxyacetate 4 was chlo-
(16) Bach, K. K.; El-Seedi, H. R.; Jensen, H. M.; Nielsen, H. B.; Thomsen, I.;
rinated with 1.2 equiv of acetylchlorid using a catalytical
Torssell, B. G. Tetrahedron 1994, 50, 7543-7556.
(17) Grell, W.; Machleidt, H. Justus Liebigs Ann. Chem. 1966, 53-67.
1
6
amount of iodine to give 2-chloro-2-ethoxyacetate 5.
(
18) Wadsworth, Jr. Org. React. 1977, 25, 73-253.
Toluene was added to the reaction mixture, and low-boiling
components were distilled off together with the toluene.
(19) Enantioselective hydrogenation of a 36/64 E/Z mixture resulted in only
45% ee (100% Z: 85% ee) for the best catalyst tested.
(
20) Other esters of S-1 (2-propyl, hexyl, 2-ethoxyethyl) were included in the
initial screening to elucidate a potential influence of the chain length on
the enantioselectivity. Differences in enantioselectivity found were only
within experimental and analytical error.
(11) Potlapally, R. K.; Siripragada, M. R.; Kotra, N. M.; Sirisilla, R.; Mamilla-
palli, R. S. PCT Appl. WO 02/24625.
(
(
12) Zaks, A.; Dodds, D. R. Drug DiscoVery Today 1997, 2, 513-531.
13) Bornscheuer, U. T.; Kazlauskas, R. J. Hydrolases in Organic Synthesis;
Wiley-VCH: Weinheim, 1999.
(21) An enzyme which is selective towards the R-enantiomer would be
preferable, but only two enzymes tested were R-selective with an enan-
tiomeric ratio of E < 6 (R-chymotrypsins from bovine pancreas).
(22) Chen, C.-S.; Fujimoto, Y.; Girdaukas, G.; Sih, C. J. J. Am. Chem. Soc.
1982, 104, 7294-7299.
(23) Experimental samples from Novozymes A/S, Denmark.
(24) The low enantiomeric ratio E found for Flavourzyme can be explained by
its high content of substrate unspecific proteases/peptidases.
(
14) Kato, Y.; Ohta, H.; Tsuchihashi, G.-i. Tetrahedron Lett. 1987, 28 (12),
1
303-1306.
(
15) Haigh, D.; Allen, G.; Birrell, H. C.; Buckle, D. R.; Cantello, B. C. C.;
Eggleston, D. S.; Haltiwanger, R. C.; Holder, J. C.; Lister, C. A.; Pinto, I.
L.; Rami, H. K.; Sime, J. T.; Smith, S. A.; Sweeny, J. D. Bioorg. Med.
Chem. 1999, 7, 821-830.
Vol. 7, No. 1, 2003 / Organic Process Research & Development
•
83

Table 1. Screening results: enantiomeric ratios (E) for
selected enzymes
by filtration through a variety of filter aids or by heating
remained unsuccessful. Nevertheless, filtering of the reaction
mixtures through sand partially broke the emulsions, which
then were further allowed to settle for days. The organic
phases were finally separated and collected. This procedure
was repeated three to four times. The combined organic
enantiomeric
ratio (E)
enzyme
protease 1 (aspergillopepsin II
18
>200
71
a,c,d,f
from A. aculeatus)
protease 2 (aspergillopepsin I
2 4
phases were dried over Na SO , and S-1 could be isolated
a,c,d,f
from A. aculeatus)
in good yields as an oil after evaporation of the solvent. The
experiences from this first scale-up (enzyme batch-to-batch
variations, emulsions during workup caused by the biocata-
lyst preparation, high substrate dilution) suggested that the
biocatalyst preparation should be improved and that develop-
ment work was needed on the process.
Biocatalyst Development. To get a more selective and
active biocatalyst preparation, the most straightforward and
fastest approach was to optimize the fermentation of As-
pergillus oryzae for the production of the enzyme of interest.
The medium pH was found to be a critical process parameter
and had to be kept at 4.5 throughout the fermentation. Using
the improved fermentation process, three fermentations were
run on a pilot scale (1300-L fermentor). The fermentation
supernatant obtained in a filtration step was concentrated by
removal of water in an ultrafiltration. One of the ultrafiltrates
was furthermore freeze-dried followed by homogenization;
NpI protease (neutral proteinase/fungalysin
a,c,f
from A. oryzae)
protease 2A
b,c,f
>200
>200
alp. protease (oryzin from
a,c,f
A. oryzae)
a,c
Flavourzyme
4
104
a,e
Pectinex BE 3L
a,c,d
Pectinex ultra SP-L
Kojizyme 500 mg
Shearzyme 500 L
>200
>200
>200
>200
>200
>200
a,c
a,c
a,c
ferulic acid esterase
a,d
Pectinex SP-L
Rheozymea
a,d,c
(
pectin methyl esterase)
a,d,e
Pectinex AFP L-2
>200
>200
>200
a,e
Novozyme 188
a,,c,d
acetyl xylan esterase
a
Novozymes. b Fluka. c Expressed in Aspergillus oryzae. Expressed in
d
Aspergillus aculeatus. e Expressed in Aspergillus niger. Screened in the first
round (all other enzymes were screened in the second round).
f
1
.9 kg (lot A) of a freeze-dried biocatalyst powder and two
liquid enzyme preparations (lot B, 24 kg; lot C, 20 kg) were
produced, which were used for the further development
described in this article.
preparation) were rather low, strongly suggesting that the
observed activity originated from a secreted enzyme present
in these crude biocatalyst preparations only in small con-
centrations.
27
Process Development. To define the optimum process
conditions, the influence of reaction parameters such as
temperature, pH, co-solvent addition (type of solvent and
phase ratio), substrate concentration, and biocatalyst loading
on the reaction rate and enantioselectivity of the hydrolysis
of S-2 was investigated using the new biocatalyst batches
A-C. It was found to be most advantageous to run the
reaction at a temperature of 20-30 °C (preferably 25 °C),
First Laboratory Scale-Up. The first scale-up of the
enzymatic hydrolysis was performed using Protease 2A from
Fluka. This enzyme showed the best compromise between
activity and enantioselectivity among the first enzymes
25
screened, which were available on a larger scale. A number
of small-scale experiments aimed at optimizing the enzymatic
hydrolysis on a milligram-gram-scale of the enzymatic
hydrolysis were performed.26 These experiments revealed
large batch-to-batch variations in activity and enantioselec-
tivity of the Fluka enzyme. The best procedure was then run
on a 70-g scale in an 18-L separatory funnel with mechanical
stirring; Table 2 lists the characteristics of the reactions.
In both trials, the substrate was dissolved in buffer and
stirred at room temperature with the enzyme. The pH was
kept around 7 by addition of 30% aqueous NaOH, where
the consumed base correlates with the degree of conversion.
The unreacted ester was then removed by extraction after
keeping the pH between 6.5 and 7.5 (automated titration with
28
3
0% NaOH). Outside this range, both activity and enan-
tioselectivity decreased. The addition of cosolvents (to
increase the solubility of the substrate), such as acetone, THF,
ethanol, or 2-propanol resulted in prolonged reaction time
and in reduced enantioselectivity. Instead of cosolvent
addition to increase the poor water solubility of the substrate
(
∼5 g/L), the reaction was run as a biphasic emulsion in
aqueous buffer. It was found that reducing the amount of
buffer negatively affected the reaction rate. The ratio between
substrate and buffer was therefore investigated to optimize
space-time yields and the ease of workup.
45% conversion. The aqueous reaction mixture left was then
acidified to pH 2-3 by addition of hydrochloric acid,
followed by extraction of S-1. During those extractions
emulsions formed (due to the high content of biocatalyst),
which resulted in incomplete phase separations (even after
allowing to settle for days). Attempts to break the emulsions
Even though the amount of biocatalyst could be signifi-
cantly reduced compared to that used in the first laboratory
scale-up reactions, due to higher enzymatic activities, emul-
sions still formed at acidification to pH 2-3 during work-
up. Again, all commonly employed methods, such as the
usage of different filter aids to break the emulsions and to
(
25) Only the results from the first screening round were known at this time.
Due to the urgent demand of the project S-1 was produced on the basis of
the best knowledge available. The results obtained by the second screening
round together with additional lab-scale experiments obtained later suggest
Pectinex Ultra SP-L as the most favorable enzyme mixture.
(27) The relative activities/g of enzyme preparation were determined by activity
assay as following: lot A/B/C: 8.4/1.3/1.0. The enzymatic activity present
in each batch hydrolyzed rac-2 with an enantiomeric ratio E > 200.
(28) Nonselective chemical hydrolysis was <0.1% after 4 days while stirring
in buffer without enzyme applying these conditions.
(26) Biocatalysis for Fine Chemical Synthesis; Roberts, S. M., Ed.; John Wiley
&
Sons: Chichester, 1999.
84
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development

Table 2. Characteristics of the first laboratory scale-up using proteinase 2A from Fluka
first trial
second trial
scale (rac-2)
70 g
70 g
10 g
proteinase 2A
52 g
phosphate buffer pH ) 7/0.1 M
200 mL/g, rac-2 (solution)
8 h
45%
200 mL/g, rac-2 (solution)
48 h
45%
reaction time
conversion
total solvent used during
pH 7 extractions
total solvent used
during pH 3 extraction
comments
57 mL AcOEt/g rac-2
(3 extractions needed)
51 mL AcOEt/g rac-2
(3 extractions needed)
emulsion at pH 2-3
drying of combined organic phases over
71 mL MTBE/g rac-2
(4 extractions needed)
83 mL MTBE/g rac-2
(3 extractions needed)
emulsion at pH 2-3
drying of combined organic phases over
product isolation
Na
2
SO
4
followed by filtration and
2 4
Na SO followed by filtration and
evaporation of the solvent
evaporation of the solvent
isolated yield S-1
43%
>96%
96.8
43%
>96%
94.3
75
HPLC purity
%
ee
calculated E
160
Table 3. Characteristics of the improved laboratory procedure using the three hydrolytic enzyme mixtures (lots A-C) obtained
by an optimized fermentation of Aspergillus oryzae
lot A
lot B
lot C
scale (rac-2)
50 g
50 g
50 g
Aspergillus oryzae
hydrolytic enzyme mixture
phosphate buffer
pH ) 7/0.1 M
0.8 g, lot A
8 g, lot B
15 g, lot C
2 mL/g rac-2
(biphasic mixture)
30 h
48%
75%
2 mL/g rac-2
(biphasic mixture)
23 h
45%
75%
2 mL/g rac-2
(biphasic mixture)
25 h
41%
75%
reaction time
conversion
water removed
by distillation
methanol added
2.5 times the volume
of the removed water
precipitation and removal
of the biocatalyst in aqueous
methanol prior to extraction
expeditious phase
2.5 times the volume
of the removed water
precipitation and removal
of the biocatalyst in aqueous
methanol prior to extraction
expeditious phase
2.5 times the volume
of the removed water
precipitation and removal
of the biocatalyst in aqueous
methanol prior to extraction
expeditious phase
comments
separation at pH 3.
separation at pH 3.
separation at pH 3.
total solvent used
5 mL MTBE/g rac-2
(3 extractions needed)
4 mL MTBE/g rac-2
(3 extractions needed)
drying of the combined
5 mL MTBE/g rac-2
(3 extractions needed)
4 mL MTBE/g rac-2
(2 extractions needed)
drying of the combined
6 mL MTBE/g rac-2
(3 extractions needed)
4 mL MTBE/g rac-2
(2 extractions needed)
drying of the combined
during pH 7 extractions
total solvent used during
pH 3 extraction
product isolation
organic phases over Na
followed by filtration
2
SO
4
organic phases over Na
followed by filtration
2
SO
4
organic phases over Na
followed by filtration
2 4
SO
and evaporation of the solvent
and evaporation of the solvent
and evaporation of the solvent
isolated yield S-1
48%
40%
41%
HPLC purity
>99%
>98%
>98%
%
ee
99.1
99.3
99.6
calculated E
>200
>200
>200
improve phase separations, were unsuccessful. We then
discovered that both the acid 1 and the ester 2 were soluble
in aqueous methanol29 (pH 7), while the biocatalyst formed
a filterable precipitate under the same conditions. In a typical
process, 75% of the water from the reaction mixture was
evaporated, and methanol was then added (volume of
methanol added: 2.5-4.0 times the volume of the water
precipitate had been removed, the methanol was distilled off,
and the original amount of water was added again. The
following extractions with MTBE, at neutral pH and after
acidification to pH 2-3, did not give emulsions and resulted
in expeditious phase separations. A series of experiments
were run on a 50-100-g scale to adjust the reaction
parameters and the workup procedure. Table 3 lists typical
characteristics of the process using the three different enzyme
lots A-C.
30
previously removed ) to solubilize both substrate and
product3 while precipitating the biocatalyst. Cellulose
powder from spruce proved to be a very efficient filter aid
to accelerate the following filtration. After the biocatalyst
1
In comparison to the first scale-up, the new laboratory
procedures were now significantly improved: The reagent
concentration was significantly higher, less solvent was used
(29) Ethanol or 2-propanol can be used as well.
(
30) The ease of filtration of the precipitate improved, when the amount of
methanol was increased. Usually the phase separations in the following
steps were also faster with increasing amounts of methanol.
(31) The conversion was most precisely determined by HPLC analysis of this
solution.
Vol. 7, No. 1, 2003 / Organic Process Research & Development
•
85

Table 4. Characteristics of the semi-pilot batches
first semi-pilot batch
second semi-pilot batch
scale (rac-2)
15.1 kg
0.27 kg, lot A
(additional 84 g enzyme added after 22 h)
2.5 mL/g rac-2
(biphasic mixture)
27 h
13.8 kg
0.22 kg, lot A
Aspergillus oryzae
hydrolytic enzyme mixture
phosphate buffer
pH ) 7/0.1 M
2.9 mL/g rac-2
(biphasic mixture)
30 h
reaction time
conversion
water removed by distillation
methanol added
42%
90%
44%
85%
2.0 times the volume
of the removed water
filter (4.5 h)
expeditious phase separation at pH 2-3.
6 mL MTBE/g rac-2
(4 extractions needed)
3 mL MTBE/g rac-2
(2 extractions needed)
concentrated under vacuum,
2.3 times the volume
of the removed water
(centrifuge 40 min)
expeditious phase separation at pH 2-3.
4 mL MTBE/g rac-2
(3 extractions needed)
3 mL MTBE/g rac-2
(2 extractions needed)
concentrated under vacuum,
2-propanol added and distilled
off until <0.2% water
42%
removal of biocatalyst
comments
total solvent used
during pH 7 extractions
total solvent used
during pH 3 extraction
product isolation
2
-propanol added and distilled
off until <0.2% water
isolated yield S-1
41%
HPLC purity
97.8%
98.6
1
0
.3% methyl ester
.7% 2-propyl ester
0.6% methyl ester
0.8% 2-propyl ester
98.6
%
ee
98.7
calculated E
>200
>200
for the extraction procedures, the workup was now scalable
to pilot-plant scale (batch reactor), and chemical and enan-
tiomeric purities were improved. The results also showed
that enantiomeric purities >99.0% ee should be obtainable
at 45% conversion with these enzyme batches. Two kilo-
lab batches were produced using enzyme lot A, on a 500-
step), storing of S-1 in solution (2-propanol is the solvent to
be used in the next reaction step) avoided the risk of
3
4
undesired crystallization. Two semi-pilot batches were
produced by applying these modifications, and the results
are listed in Table 4. Recycled solvents were used for the
second batch.
3
2
and 1000-g scale, showing the robustness of the process.
The product quality was not significantly changed in
comparison to that in the laboratory trials. Some S-1 methyl
and 2-propyl ester were formed during the removal/distil-
lation of the alcohols. As S-1 will be transformed into the
2-propyl ester in the following reaction anyway, the presence
of ester impurities was not an issue. The enantiomeric purities
were somehow lower in comparison to the values usually
observed during the laboratory trials but still above our target
range (>98.0% ee).
The process was finally run in pilot-plant using all three
enzyme lots. Table 5 lists the results of the pilot campaign.
Except that slightly more water was removed prior to the
biocatalyst precipitation, the pilot-process conditions re-
mained basically unchanged in comparison to those of the
semi-pilot process. The reaction times were shorter in
comparison to those in the laboratory and semi-pilot trials,
probably due to a better mixing in the pilot reactor. While
the phase separations during the MTBE extractions normally
took between 30 and 60 min in all the earlier trials,
unsatisfactory phase separations were observed in batches
1, 2, and 4 during the first pH-7 extractions (see Table 5).
In semi-pilot the same enzyme was used as in the pilot first
batch without any phase-separation problems. We do not see
a reason for this behaviour in the pilot trial. More experience
would be needed to identify critical parameters if the process
were to be continued with similar enzyme batches. Chemical
Scale-Up to Pilot. Some changes were introduced during
the technology transfer to semi-pilot to adapt the process to
plant requirements: The reaction mixture was heated to 80-
90 °C after the desired conversion of 45%. At that temper-
ature the biocatalyst irreversibly deactivated. This excluded
the risk of exceeding hydrolysis above the desired conversion
in case of prolonged workup and resulted in a more robust
33
process. It was technically possible to distill off more water
from the reaction mixture than as initially judged during the
laboratory optimization. Consequently, lower amounts of
methanol were necessary to solubilize the product and the
substrate, without influence on the filtration step using semi-
pilot equipment. To accelerate the filtration, a centrifuge was
used instead of a filter. The solvent volumes (MTBE) used
for the extractions could be reduced without a negative
impact on the phase separations. Finally, the combined
MTBE solutions containing S-1 were not treated with Na
2
-
SO followed by filtration as done in the laboratory
4
procedures. Instead, the water was removed by azeotropic
distillation with most of the MTBE. 2-Propanol was added
to the reaction mixture, and the remaining MTBE/water was
distilled off with some of the 2-propanol to give S-1 as a
50-70% (weight) solution in 2-propanol. In addition to the
advantage of shortening the process (omitting a filtration
(
32) kg batch: 41% hydrolysis, 39% isolated yield, >99% HPLC purity, 98.9%
ee; 500 g batch: 42% hydrolysis, 41% isolated yield, >99% HPLC purity,
(34) Pure S-1 is a crystalline solid (mp ) 105 °C). Usually the S-1 obtained
after the processes was obtained as an oil, which then slowly crystallized
on standing.
9
8.8% ee.
(33) Tested on 50-g scale first.
86
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development

Table 5. Characteristics of the pilot batches
first pilot batch
second pilot batch
43.9 kg
third pilot batch
43.9 kg
fourth pilot batch
43.9 kg
scale (rac-2)
43.9 kg
Aspergillus oryzae
hydrolytic enzyme mixture
phosphate buffer
pH ) 7/0.1 M
0.74 kg, lot A
7.5 kg, lot B
7.5 kg, lot B
7.5 kg, lot C
2.5 mL/g rac-2
(biphasic mixture)
21 h
2.5 mL/g rac-2
(biphasic mixture)
20 h
2.5 mL/g rac-2
(biphasic mixture)
16 h
2.5 mL/g rac-2
(biphasic mixture)
16 h
reaction time
conversion
water removed
by distillation
43%
93%
47%
93%
44%
95%
48%
94%
methanol added
2.3 times the volume
of the removed water
extraction at pH7:
first layer
2.3 times the volume
of the removed water
extraction at pH7:
first layer
2.3 times the volume of
the removed water
extraction at pH7:
first layer
2.3 times the volume
of the removed water
extraction at pH7:
first layer
comments
separated overnight.
second layer separated
overnight only with
additional MTBE.
extraction at pH3:
separated overnight.
second and third layer
separated within
30-60 min
extraction at pH3:
30-60 min for all
phase separations
5 mL MTBE/g rac-2
(3 extractions needed)
3 mL MTBE/g rac-2
(2 extractions needed)
as in semi-pilot
44%
separated within 60 min
second layer
separated over 2 days.
second layer
separated within
30-60 min
extraction at pH3:
30-60 min for all
phase separations
3 mL MTBE/g rac-2
(2 extractions needed)
3 mL MTBE/g rac-2
(2 extractions needed)
as in semi-pilot
48%
separated within
30 min
extraction at pH3:
30-60 min for all
phase separations
3 mL MTBE/g rac-2
(2 extractions needed)
3 mL MTBE/g rac-2
(2 extractions needed)
as in semi-pilot
44%
3
0-60 min for all
phase separations
5 mL MTBE/g rac-2
(2 extractions needed)
3 mL MTBE/g rac-2
(2 extractions needed)
as in semi-pilot
43%
total solvent used
during pH 7 extractions
total solvent used
during pH 3 extraction
product isolation
isolated yield S-1
HPLC purity
97.7%
98.4%
0.7% 2-propyl ester
98.4
98.8%
0.3% 2-propyl ester
99.0
99.3%
0.1% 2-propyl ester
99.6
1
.7% 2-propyl ester
%
ee
99.4
calculated E
>200
>200
>200
>200
Table 6. Comparison of the initial laboratory procedure, the pilot plant process described in this article, and a potential future
process using the cloned enzyme
possible process using
the cloned enzyme
first laboratory scale-up
pilot-plant process
enzyme/kg S-1
143-743 g (freeze-dried)
17 g (freeze-dried) or
2 g (ultra filtrate)
1
70-340 g (ultra filtrate)
MTBE/kg S-1
MeOH/kg S-1
water/kg S-1
154 l
none
200 l
ca. 1 kg
none
2
6-8 l
5.8 l
5.0 l
none
130 g
5
6-8 l
none
2.5 l
none
none
3
2 4
Na SO /kg S-1
filter aid (cellulose)
distillation steps
filtration steps
2
1
0
purities and isolated yields of the product S-2 were again
excellent. The enantiomeric purities were higher than in the
semi-pilot campaign despite the fact that the hydrolyses were
run to higher conversions.
responsible for the desired catalytic activity on S-2 could
be isolated, and the gene encoding it was cloned. A much
purer and more active enzyme preparation was then pro-
3
5
duced by transforming a suitable host cell with the DNA
sequence encoding the esterase, cultivating the transformed
organism under conditions permitting the production of the
Outlook
36
The combination of biocatalyst and process development
described in this article allowed quick progress from labora-
tory scale to pilot production. Parallel, a molecular biotech-
nology project (cloning and expression of the enzyme) was
initiated, aiming at the production of a highly concentrated
and purer biocatalyst. Such an improved enzyme preparation
was expected to greatly facilitate the downstream processing.
By using a genetically engineered production organism the
biocatalyst production should also become more cost-efficient
due to higher fermentation yields of the desired enzymatic
activity. Shortly after the pilot campaign, the enzyme
enzyme, and recovering the enzyme from the culture. The
enzyme preparation thereby obtained was successfully tested
in laboratory-scale experiments, where the removal of the
biocatalyst before workup was not necessary. The final
enzymatic process to be used in future production is thereby
expected to be more economical and superior to the pilot-
(
35) An enantiomeric ratio of E > 200 together with a significant higher specific
activity was found (>80 times the activity of lot A). This allowed such a
low biocatalyst dosage, that no emulsions formed during workup.
36) Østergaard, P. R.; Mailand Hjort, C.; Deussen, H.-J.; Zundel, M.; Ebdrup,
S.; Christensen, S.; Patkar, S. PCT-Appl. WO 02/12472.
(
Vol. 7, No. 1, 2003 / Organic Process Research & Development
•
87

plant process described in this article. Table 6 compares the
major differences which have an economic impact of the
initial laboratory procedure (first laboratory scale-up), the
pilot-plant process described in this article, and a potential
future production process, Furthermore, a laboratory proce-
dure has been developed for the racemization of the unreacted
R-2, which would allow the recycling of the unwanted
procedure. The whole process was optimized and scaled up
to pilot scale (43.9-kg scale). The isolated yields (44-48%),
chemical purities (97.7-99.3%), and enantiomeric purities
(98.4-99.6% ee) of the product S-1 obtained from the pilot
trials were excellent and consistent with the previous
experiments. The only drawback was the occurrence of
unexpected slow phase separations not seen previously during
extraction. Nevertheless, the pilot trials were still technically
feasible.
37
enantiomer in a future production.
Summary
Experimental Section
Enzymatic activity for the enantioselective hydrolysis of
S-2 was identified by screening of over 80 hydrolytic
enzymes. It was found that almost all enzyme preparations
derived from Aspergillus fungi were highly enantioselective
Experimental details about the enzyme screening results,
the analytical methods, the chemoenzymatic process, and the
Aspergillus oryzae fermentation have been published ear-
9,10
lier.
(E > 200). However, the activity of the enzyme preparations
had to be improved for the desired catalytic activity to
facilitate a good process. Increased enzymatic activity per
volume culture broth of the enzyme of interest was obtained
by optimizing the parameters of the fermentation of As-
pergillus oryzae. Three pilot-scale fermentation batches were
produced. A scalable chemoenzymatic process was devel-
oped for the enantioselective hydrolysis of S-2 using these
hydrolytic enzyme mixtures. A major process hurdle, the
formation of emulsions during workup caused by the
biocatalyst, could be circumvented by a simple precipitation
Acknowledgment
We wish to thank Novozymes A/S for enzyme samples
and fermentation technology support and Siegfried CMS for
technical support. We especially thank L. Højlund Chris-
tensen, R. von Gr u¨ nigen, S. A. Patkar, T. Muxoll Fatum,
M. W u¨ rtz Christensen, and S. Christensen for their scientific
assistance and for fruitful discussions. We thank M. Schmel-
ling Kure, B. Bentzen, and A. Jørgensen Winsløv for
excellent technical assistance.
Received for review October 21, 2002.
OP0256035
(
37) R-2 can be racemized with sodium ethoxide in ethanol or sodium hydride
in THF.
88
•
Vol. 7, No. 1, 2003 / Organic Process Research & Development