Enzymatic process for the synthesis of cis/trans-(1R,5R)-bicyclo[3.2.0] hept-6-ylidene-acetate: Solvent effect and NMR study

Article abstract of DOI:10.1021/op0600409

An efficient enzymatic process has been developed to resolve a diastereomeric mixture of racemic ethyl bicyclo[3.2.0]hept-6-ylidene-acetate (1). Using 40% acetone, not only was the enantioselectivity of Candida antarctica lipase B (CAL-B) significantly improved to E > 200 from E = 2.7, remarkably the enzyme is able to maintain low diastereoselectivity for the Z (cis)- and E (trans)-isomers leading to an overall high isolated yield (40-45%, vs maximum theoretical yield 50%) and excellent enantiomeric excess (>98.5% ee). Preliminary studies using 2D TROSY shows that there is probably a global conformational change in the N¹⁵-labeled enzyme CAL-B when the content of organic cosolvents increases from 0% to 30%. The route was successfully scaled to 63 kg for the synthesis of GABA analogues.

Full text of DOI:10.1021/op0600409

Organic Process Research & Development 2006, 10, 655−660
Enzymatic Process for the Synthesis of cis/trans-(1R,5R)-Bicyclo[3.2.0]hept-
-ylidene-acetate: Solvent Effect and NMR Study
6
†
‡
‡
†
†
,†
Daniel Yazbeck, Andrew Derrick, Maninder Panesar, Alan Deese, Anita Gujral, and Junhua Tao*
Biotransformations Group, Chemical R&D, Pfizer Global R&D, 10578 Science Center DriVe,
San Diego, California 92121, U.S.A., and Chemical R&D, Pfizer Global R&D, Ramsgate Road, Sandwich, England
Abstract:
Scheme 1
An efficient enzymatic process has been developed to resolve a
diastereomeric mixture of racemic ethyl bicyclo[3.2.0]hept-6-
ylidene-acetate (1). Using 40% acetone, not only was the
enantioselectivity of Candida antarctica lipase B (CAL-B)
significantly improved to E > 200 from E ) 2.7, remarkably
the enzyme is able to maintain low diastereoselectivity for the
Z (cis)- and E (trans)-isomers leading to an overall high isolated
yield (40-45%, vs maximum theoretical yield 50%) and
excellent enantiomeric excess (>98.5% ee). Preliminary studies
using 2D TROSY shows that there is probably a global
15
conformational change in the N -labeled enzyme CAL-B when
the content of organic cosolvents increases from 0% to 30%.
The route was successfully scaled to 63 kg for the synthesis of
GABA analogues.
Introduction
candidates identified that was able to catalyze the hydrolysis
of the ethyl ester with some selectivity and high reactivity.
Unfortunately, the enantioselectivity of the enzyme was quite
low (E ) 2.7, or 33% ee at a conversion of 50%) under
screening conditions with a pH 7 buffer and 10% acetonitrile
(1R,5R)-Bicyclo[3.2.0]hept-6-ylidene-acetic acid (2) is a
key intermediate toward the synthesis of GABA (γ-ami-
nobutyric acid) analogues such as 3, therapeutically useful
agents for the treatment of epilepsy and neuropathic pain
1
(
Scheme 1). To our knowledge, no asymmetric synthesis
(ACN). Previous experience in working with CAL-B,
of this molecule has been reported prior to this study. We
describe here the development of an efficient and cost-
effective enzymatic process to prepare enantiomerically pure
E/Z-2 by the resolution of racemic and E/Z-ethyl bicyclo-
however, gave us hope that we might be able to improve
the selectivity towards this substrate via medium engineering.
It should be noted that the overall rate of the reaction is
dependent on the ratio of two diastereomers. CAL-B
hydrolyzes the tran-isomer at a rate approximately 2 times
faster than the E-isomer. The Kcat for the Z-(1R,5R)-ethyl
[3.2.0]hept-6-ylidene-acetate (1). In this method, both the
E- and Z-isomers of the (1R,5R)-enantiomer (2) are desired
to be converted into 3 (Scheme 1). While enzymes have been
widely used for the resolution of racemic molecules, this is
a rare example where the resolution involves a mixture of
four stereoisomers rendering the problem much more chal-
lenging and interesting.
-
1
bicyclo[3.2.0]hept-6-ylidene-acetate is about 115 min . In
most batches screened, the ratio of the Z- to E-diastereomers
in the starting material (1) was about 4:1 or greater which
ultimately increased the throughput of the reaction.
Medium Study. Attempts at improving the enantio-
selectivity of the enzyme were made via medium engineer-
ing. After screening a number of representative solvents with
log P values ranging from -1 to 5 and various cosolvent
contents from 20% to 50%, it was found that t-BuOH (log
P 0.35), acetonitrile (log P -0.34), and acetone (log P -0.24)
have the desired physical properties to tune the enantio-
selectivity of the enzyme. Figure 1 and Figure 2 illustrate in
more detail the effect of these solvents on the reactivity and
enantioselectivity of CAL-B towards the substrate 1. As the
cosolvent content was increased from 20% to 50%, the ee’s
increased from 78% to 98.1% for t-BuOH, 85% to 95% for
Results and Discussions
Enzyme Screening. The racemic and E/Z-ethyl bicyclo-
[3.2.0]hept-6-ylidene-acetate (1) was synthesized from bicyclo-
[3.2.0]heptan-6-one according to a literature protocol (see
1
Experimental Section for details). Using an automated
enzyme screening protocol developed in our lab, Candida
antarctica lipase B (CAL-B) was one of the few enzyme
2
*
Author for correspondence. Telephone: (858) 518-4484. Fax: (858) 678-
8
199. E-mail: junhua_tao@yahoo.com.
†
Biotransformations Group, Chemical R&D, Pfizer Global R&D.
‡
Chemical R&D, Pfizer Global R&D.
(
1) (a) Gladwell, I. R.; Pettman, A. J. WO 2004031115. (b) Derrick, A. M.
WO 2004031124. (c) Ghosez, L.; Montaigne, R.; Roussel, A.; Vanlierde,
H.; Mollet, P. Tetrahedron 1971, 27 (3), 615-33.
(2) Yazbeck, D.; Tao, J.; Martinez, C.; Kline, B.; Hu, S. AdV. Synth. Catal.
2003, 345 (4), 524-532.
1
0.1021/op0600409 CCC: $33.50 © 2006 American Chemical Society
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
655
Published on Web 04/04/2006

Figure 1. Effect of solvent and solvent content on the
enantioselectivity of CAL-B towards 1. Reaction conditions:
1
0% enzyme loading (w/w), 100 mg/mL of 1, pH 7.0, rt, 16 h.
Figure 4. Effect of temperature on the reactivity and enan-
tioselectivity of CAL-B towards 1. Reaction conditions: 40%
cosolvent (red line - t-BuOH, blue line - acetone); KPB pH
8.0, 10% enzyme loading (w/w), 100 mg/mL of 1, 16 h.
Figure 2. Effect of solvent and solvent content on the reactivity
of CAL-B towards 1. Reaction conditions: 10% enzyme loading
(w/w), 100 mg/mL of 1, pH 7.0, rt, 16 h.
Figure 5. Effect of cosolvents on enzyme leaching. The residual
activity of recycled Novozyme 435 after a 16 h incubation in
various organic solvents was measured via a subsequent
reaction in KPB buffer at pH 8.0, 10% recycled enzyme (w/w),
100 mg/mL 1, 16 h. Novozyme 435 was recycled via a simple
filtration of the incubated solvent/enzyme solutions.
in ee’s is observed presumably due to background chemical
hydrolysis. The optimal pH was chosen at 8.0.
Temperature Effect. Temperature optimization studies
were carried out from 30 to 60 °C. As seen from Figure 4,
temperature has a minor effect on both the reactivity and
enantioselectivity of the reaction. Therefore, all subsequent
reactions were run at room temperature for higher process
efficiency.
Enzyme Sources. To identify the ideal CAL-B form for
process development, different forms of the enzyme from
various suppliers were selected for comparison including
Chirazyme L2 solution from Roche, Novozyme 435 from
Novozyme, cross-linked enzyme crystal (CLEC) CAL-B
from Altus, Chirazyme L2 Carrier Fixed (C.F.) from Roche,
and Chirazyme L2 Lyo powder from Roche). All forms of
CAL-B showed similar reactivity and price profile except
for the CLEC and the Chirazyme L2 C.F., which are less
reactive. In general, the lypophilized forms are substantially
more expensive than the crude extracts. Ultimately, No-
vozyme 435 was selected, as it can be readily recycled to
further enhance the process efficiency, resulting in simpler
workup and reduction in the cost of goods.
Figure 3. Effect of pH on the reactivity and enantioselectivity
of CAL-B towards 1. Reaction conditions: 40% cosolvent (red
line - t-BuOH, blue line - acetone); potassium phosphate
buffers (KPB) with varying pH, 10% enzyme loading (w/w),
100 mg/mL of 1, rt, 16 h.
acetone, and 92.4% to 98.5% for ACN (Figure 2). At the
same time, the rate of the reaction decreased for all three
cosolvents (Figure 3). Therefore, a dramatic solvent effect
was observed, where not only was the enantioselectivity
significantly improved to E > 200, but remarkably the
enzyme was also able to maintain low diastereoselectivity
for the E- and Z-isomers as well. This was essential to
maximize the overall yield of the reaction because both the
E- and Z-(1R,5R)-2 will be turned into the same intermediate
for the synthesis of 3. Acetone (40%) is ideal for further
process development since it allows us to achieve the desired
enantioselectivity and maintains a high reactivity (98.1% ee,
Enzyme Recycling Studies. Novozyme 435 is an im-
mobilized form of CAL-B through physical absorption to
the macroporous acrylic resin. Therefore, solvents will have
a strong effect on enzyme leaching and its residual activity
after solvent incubation. Among the eight solvents studied,
acetonitrile has the strongest negative effect on leaching
(Figure 5). Over 70% of the activity was lost after incubation
in 40% acetonitrile for 16 h compared to the nonincubated
37.7% conversion at pH7 after 16 h). Acetone is also
inexpensive and easier to handle at large scale for both the
reaction and downstream separation.
pH Effect. As the pH changes from 5.0 to 8.0, the rate
of the reaction increases without any sacrifice to enantio-
selectivity (Figure 3). At pH’s higher than 8.5, a decrease
656
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development

the synthesis of 3 without further purification. The im-
mobilized enzyme was recycled for the next batch of
hydrolytic resolution, leading to lower material cost. Sub-
sequently, the enzymatic process was successfully scaled up
to 62.8 kg scale with a reaction volume of 630 L in a 1450-L
glass-lined reactor.
NMR Study of the Solvent Effect. Due to the dramatic
solvent effect observed in this case, and the dynamic nature
15
of enzymatic catalysis, we decided to prepare the N -labeled
CAL-B and use NMR to monitor the change of the protein
backbone as the content of solvent is increasing. While it is
known that solvents have strong effects on both reactivity
and enantioselectivity, the fundamental mechanism is still
Figure 6. Effect of pH on enzyme leaching. The residual
activity of recycled Novozyme 435 after a 16 h incubation in
various buffer pH’s was measured via a subsequent reaction
in KPB buffer at pH 8.0, 10% recycled enzyme (w/w), 100 mg/
mL 1, 16 h. Novozyme 435 was recycled via a simple filtration
of the incubated buffer/enzyme solutions.
3,4
not clear, and no NMR studies have been reported to study
15
5
the solvent effects of N -CAL-B. The 2D TROSY experi-
ment affords a fingerprint of the protein amide region that
is highly sensitive to changes in the protein environment. It
is, therefore, a many-parameter NMR probe for studying
intermolecular interactions of the protein. Structural and
functional changes within a protein created by a chemical
or physical event, such as solvation or binding to another
molecule, can be probed by monitoring the chemical shift
6
,7
15
changes in a TROSY spectrum. To prepare N -CAL-B,
the CAL-B gene was cloned from the genomic DNA isolated
from Candida antarctica (ATCC 32657) and amplified by
8
15
PCR. Subsequently, the N -CAL-B was expressed in Pichia
Figure 7. The effect of temperature on enzyme leaching. The
residual activity of recycled Novozyme 435 after a 16 h
incubation in KPB buffer at various temperatures was mea-
sured via a subsequent reaction in KPB buffer at pH 8.0, 10%
recycled enzyme (w/w), 100 mg/mL 1, 16 h. Novozyme 435 was
recycled via a simple filtration of the incubated buffer/enzyme
solutions.
pastoris using a pPIC9 secretion vector (Invitrogen, CA)
15
under a minimal medium supplemented with NH
4
Cl. The
expressed recombinant protein was purified by Sepharose
9
,10
1
15
column. Figure 8a shows the 2D, H/ N TROSY spectra
for CAL-B in the absence (black) and presence (red) of
acetonitrile (30 vol %) (for details see Experimental Section).
It can be seen that many of the CAL-B amide resonances
are shifted upon the addition of acetonitrile. A possible
interpretation of these data is that CAL-B adopts a very
different conformation in the presence of acetonitrile as
control (Novo 435 Ctl, Figure 5). In contrast, in 40% acetone,
not only does CAL-B retain most of its activity (70%) after
one recycle run, it is also more active than other solvents
under the same cosolvent content (see Medium Study
section). Essentially no loss in activity was observed in 100%
t-BuOH for this form of CAL-B. It should also be noted
that the same enzyme retains almost 87% of the activity after
incubation and recycling from 100% aqueous phosphate
buffer. On the other hand, pH does not have a significant
effect on the recycling of Novozyme 435, and pH 8.0 was
chosen due to the high activity of CAL-B under the condition
2
compared to the normal H O/buffer state. It should be noted
that the TROSY spectrum from the normal state was
observed again when the 30% acetonitrile was removed from
the mixture (Figure 8b). We are currently preparing an
inactive mutant of CAL-B (Ser105Ala in the catalytic triad)
which eventually shall allow an in situ monitoring of the
backbone change during reaction when solvent contents
change. The inactive form may also help to understand the
substrate solvation using microcalorimetry. Solvation studies
of CAL-B crystals in various solvent contents may also
(
Figure 6). Temperature has a strong effect on the leaching
and residual activity of Novozyme 435, which lost almost
5% of the activity after incubation in a buffer at 50-60 °C
7
for 16 h (Figure 7). Therefore, the reaction should be
conducted at room temperature.
(
3) Schmitke, J. L.; Stern, L. J.; Klibanov, A. M. Proc. Natl. Acad. Sci. U.S.A.
998, 95, 12918-12923.
(4) Laane, C.; Boeren, S.; Vos, K.; Veeger, C. Biotechnol. Bioeng. 1987, 81-
7.
1
Process Description. Under the optimal reaction condi-
tions (Novozyme 435, 40% acetone, pH 8.0, rt), the reaction
was initially demonstrated at 100-g scale. With a substrate
loading of 100 g/L and an enzyme loading of 25 g/L, the
reaction reached 45-50% conversion in a batch reactor
controlled by an auto pH titrator within 24 h. After filtration
of the immobilized enzyme, the desired mixture of E/Z-
8
(
5) For NMR studies using nonlabeled CAL-B, see: (a) Vallikivi, I.; J a¨ rving,
I.; Pehk, T.; Samel, N.; T o˜ ugu, V.; Parve, O. J. Mol. Cat. B 2004, 32, 15-
19. (b) Hansen, T. V.; Waagen, V.; Partali, V.; Anthonsen, H. W.;
Anthonsen, T. Tetrahedron: Asymmetry 1995, 6, 499-504.
6) Pervushin, K.; Wider, G.; Wuethrich, K. J. Biomol. NMR 1998, 1, 345-
348.
(
(
(
(
7) Rance, M.; Loria, J. P.; Palmer, A. G., III. J. Magn. Reson. 1999, 136,
91-101.
(2R,5R)-bicyclo[3.2.0]hept-6-ylidene-acetate (2) was isolated
8) Uppenberg, J.; Hansen, M. T.; Patkar, S.; Jone, T. A. Structure 1994, 2,
293-308.
with a yield of 40-45% and 98.5% ee (E > 200 at a
conversion of 49%) after removal of the leftover E/Z-
enantiomer 4 (Scheme 2) (for details see Experimental
Section). Both the E/Z-2 was carried on to the next step for
9) For a literature protocol, see: Rotticci-Mulder, J. C.; Gustavsson, M.;
Holmquist, M.; Hult, K.; Martinelle, M. Protein Express. Purif. 2001, 21,
386-392.
(10) Pickford, A. R.; O’Leary, J. M. Methods Mol. Biol. 2004, 278, 17-33.
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
657

Scheme 2
provide useful information on the solvent effect at the
molecular level.3
flow rate was 0.7 mL/min. Sample concentration was 1 mg/
mL, and the retention times were observed as follows;
(
(
(
(
1R,5R,E)-bicyclo[3.2.0]hept-6-ylideneacetic acid 12.1 min;
1R,5R,Z)-bicyclo[3.2.0]hept-6-ylideneacetic acid 12.7 min;
1S,5S,E)-bicyclo[3.2.0]hept-6-ylideneacetic acid 13.5 min;
1S,5S,Z)-bicyclo[3.2.0]hept-6-ylideneacetic acid 18.0 min.
Ethyl Bicyclo[3.2.0]hept-6-ylideneacetate (1). Bicyclo-
3.2.0]hept-2-en-6-one (42 kg, 388 mol) was dissolved in
Conclusions
The challenge in this process was to identify an enzyme
that is selective towards both the E- and Z- diastereomers of
one enantiomer leaving behind the E/Z isomers of the other
enantiomer. The only enzyme identified upon screening was
Candida antarctica lipase B, but it exhibited an E value of
only 2.7 under screening conditions with 10% organic
cosolvent. Comprehensive medium engineering yielded a
dramatic effect, improving the E value of CAL-B to >200
towards this substrate using 40% acetone. This process was
further optimized using an immobilized CAL-B to reduce
the cost of goods by enzyme recycling and to enhance
process efficiency through simpler workup. Finally the
enzymatic transformation was successfully scaled up to 62.8
kg scale in a batch reactor. This process represents the first
synthesis of enantiomerically pure E/Z-(2R,5R)-bicyclo[3.2.0]-
hept-6-ylidene-acetic acid, a key intermediate needed for the
preparation of GABA analogues. This is a rare example
where enzymatic resolution involved four stereo-
isomers and the success resulted from a strong solvent effect.
[
n-heptane (82 L), and 5% Pd/C (50% wet, 0.12 kg) was
added. The reaction mixture was hydrogenated at 30 psi
(1551 Torr) of hydrogen and 30 °C for 8 h. The reaction
mixture was filtered, and the vessel and catalyst residue were
washed with n-heptane (20 L). The filtrate was then used
directly in the next step without further purification. Sepa-
rately, to a suspension of sodium ethoxide (27.2 kg, 400 mol)
in n-heptane (107 L) was added triethylphosphonoacetate
(95.9 kg, 428 mol) over 1 h, maintaining the temperature at
r45 °C. The reaction mixture was then stirred at -45 °C
for 75 min. The solution of bicyclo[3.2.0]heptan-6-one in
n-heptane (388 mol) from the previous step described above
was added over 2.5 h, maintaining the temperature at -48
to -38 °C. The reaction mixture was then stirred at - 45
°
C for 2 h and quenched with a 25% aqueous solution of
15
A preliminary 2D NMR study using N -CAL-B indicates
the organic solvent probably has a global effect on the change
of protein conformation.
hydrochloric acid (80 L). The phases were allowed to warm
to room temperature and then separated. To the n-heptane
phase was added a 17.6% Na
kg) prepared by dissolving Na CO
309 L). The mixture was stirred for 10 min, and the phases
were separated. To the n-heptane phase was added water
202 L), the mixture was stirred for 10 min, and the phases
2 3
CO solution in water (375
(66 kg, 623 mol) in water
Experimental Section
2
3
(
General Procedure. For a general procedure of enzyme
screening in a 96-well plate please see ref 2. The majority
of enzymes utilized in the preparation of screening kits were
obtained from various enzyme suppliers including Amano
(
were separated. The organic phase was concentrated in vacuo
to give the product (60.8 kg, 337 mol) in 87% yield as a
(Nagoya, Japan), Roche (Basel, Switzerland), Novozymes
1
slightly yellowish oil (Z:E ratio typically >4:1): H NMR
A/S (Bagsvaerd, Denmark), Altus Biologics Inc. (Cambridge,
MA), Biocatalytics (Pasadena, CA), Toyobo (Osaka, Japan),
Sigma-Aldrich (St. Louis, MO). Reactions were performed
in an Eppendorf thermomixer-R (VWR), and HPLC analysis
was carried out by an Agilent 220 HPLC autosampler.
Solvents were obtained from EM Science (Gibbstown, NJ)
and were of the highest purity available. All reactions
involving air-sensitive reagents were performed under dry
nitrogen. All materials obtained from commercial suppliers
3
(400 MHz, CDCl ) δ 5.58-5.55 (m, 1H, Z isomer), 5.54-
5
2
1
1
2
1
.52 (m, 1H, E isomer), 4.20 (q, 2H, E isomer), 4.14 (q,
H, Z isomer), 3.76-3.70 (m, 1H, E isomer), 3.41-3.35 (m,
H, Z isomer), 3.25-3.15 (m, 1H, Z isomer), 2.90-2.82 (m,
H, E + Z isomers), 2.60-2.50 (m, 1H, Z isomer), 2.30-
.20 (m, 1H, E isomer), 2.05-1.97 (m, 1H, E isomer), 1.80-
.50 (m, 6H, E + Z isomers), 1.28 (t, 3H, E + Z isomers)
1
in agreement with literature data.
1
were used without further purification. H NMR spectra were
Kcat Measurement. The Z-(1R,5R)-ethyl bicyclo[3.2.0]-
recorded at 400 MHz on a Varian Mercury spectrometer with
hept-6-ylidene-acetate was prepared by prep-HPLC, and its
concentrations were varied from 8 to 5000 µM in the
presence of 5 µL of CAL-B (20.6 µM). After 3 min, the
reaction was quenched by the addition of a 10% 3.2%-TFA
the solvent as an internal standard (δ H: CDCl
3
7.26 ppm).
C NMR spectra were recorded at 75 MHz on a Varian
Inova spectrometer with the solvent as the internal standard
δ C: CDCl 77.03 ppm). Chiral purity was measured by
1
3
(
3
solution. Kcat and K
3.5 program. In the presence of 10% acetonitrile at room
temperature, Vmax ) 59.4 µM/min or K ) 303 µM and Kcat
m
were calculated using the Kaleidagraph
HPLC employing a Chiralpak AS-RH (4.6 mm × 150 mm)
at a column temperature of 20 °C and UV wavelength of
m
-
1
225 nm. The gradient elution is tabulated below, and the
) 115 min .
658
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development

1
15
15
15
Figure 8. (a) 2D H/ N TROSY spectra. The N -CAL-B control spectrum is displayed in black, and the N -CAL-B containing
1
15
15
3
0% acetonitrile is shown in red. (b) 2D H/ N TROSY spectra. The N -CAL-B control spectrum is displayed in black, and the
N -CAL-B after acetonitrile removal is shown in red.
1
5
(
1R,5R,E)-Bicyclo[3.2.0]hept-6-ylideneacetic Acid (Z-
). To a reaction flask equipped with a pH electrode, an
overhead stirrer, and a base addition line controlled by a 718
Stat titrino-metrohm pH titrator was added acetone (400 mL)
and potassium phosphate buffer (pH ) 7.2, 600 mL). The
pH of the resulting mixture was adjusted to pH 8.0, and 25
2
Vol. 10, No. 3, 2006 / Organic Process Research & Development
•
659

Table 1
chain reaction) using primers CTTCGAATTCCTACCTTC-
CGGTTCGGACC and GCGGATACAGCGGCCGCGGG-
GGTGACGATGCCGGAG (see ref 8). The protein was
cultured and expressed in Pichia pastoris using a pPIC9
vector. The CAL-B construct was transformed into GS115
strain. Transformants were screened for mut+ phenotype
following manufacturer’s recommended protocol (Invitrogen
product manual, CA). In general, a single colony was
inoculated into 25 mL of culture media and grown until OD
time (min) 0.1% TFA/acetonitrile (%) 0.1% TFA/water (%)
0
.00
30
30
10
30
30
70
70
90
70
70
1
2
2
3
5.00
2.00
3.00
3.00
2
-6 (18-24 h). The cells were then spun down, and the
g of Novozyme 435 was added. The suspension was then
stirred at room temperature for 1 min, and the racemic ethyl
ester 1 (100 g) was added to the mixture. The reaction was
monitored by RP-HPLC looking at both the conversion and
ee’s of the product, and was stopped after about 45-50%
of the starting material had been consumed (approximately
pellet was resuspended to an OD of 1.0 in minimal media
to induce expression. Methanol was then added every 24 h
(1%), and the culture was induced for 72 h (for details, see
the Invitrogen manual). The media supernatant was tested
for the activity using the ethyl ester 1. Once the activity of
the CLA-B was confirmed, steps were taken to minimize
degradation by changing growth media to BMG media
24 h). After the reaction was complete, the heterogeneous
mixture was filtered to recover the immobilized enzyme. The
reaction mixture was concentrated under vacuum to remove
the 400 mL of acetone. The aqueous solution was then
acidified to pH 4.0 with 1 N HCl, and extracted three times
with MTBE. The acid and ester fractions were pooled, dried
with sodium sulfate, and concentrated by rotary evaporation.
A vacuum distillation was then carried out to separate the
(phosphate buffer, pH6.0). One set was grown in the absence
15
4
of exgenous isotopes, and the other was grown with NH -
Cl (3 g/L) (see ref 10). Isotope labeled cultures demonstrated
comparable expression (data not shown). The protein content
was determined using a Bradford assay and was estimated
to have an expression of 15 mg/L of cultures. Using a 1-mL
butyl sepharose column, 500 mL of the filtered protein-
containing media was purified. For purification, ammonium
acetate was added to the supernatant to reach final concen-
tration of 0.8 M. Onto a 1-mL butyl sepharose column, was
loaded 500 mL of the sample without concentration. The
supernatant was applied to the butyl sepharose column and
eluted with a gradient of 0.8 M ammonium acetate in
phosphate buffer (50 mM) and water. The eluted fractions
were tested for activity, and fractions containing the desired
purity (B10-A11) were pooled to get a total of 2 mg of N -
CAL-B.
NMR Analysis. NMR spectra were recorded at 30 °C
using a Bruker-Biospin AV700 spectrometer operating at 700
MHz for H and equipped with a 5 mm TCI z-gradient
Cryoprobe. TROSY (transverse relaxation optimized spec-
troscopy) data were acquired using the Bruker pulse sequence
acid product 2 as a E/trans mixture (>98.5% ee by chiral
1
HPLC with 40-45% yields) from the residual ester:
H
NMR (400 MHz, CDCl ) δ 5.58-5.55 (m, 1H, Z isomer),
3
5
3
2
2
.54-5.52 (m, 1H, E isomer), 3.77-3.70 (m, 1H, E isomer),
.43-3.35 (m, 1H, Z isomer), 3.26-3.16 (m, 1H, Z isomer),
.91-2.82 (m, 1H, Z isomer), 2.61-2.51 (m, 1H, Z isomer),
.32-2.22 (m, 1H, E isomer), 1.83-1.51 (m, 6H, E + Z
13
isomers). C NMR (75 MHz, CDCl
8.9, 37.5, 35.7, 33.1, 32.5, 24.6.
Alternatively, the product can be obtained by an acid/
3
) δ 175.2, 172.5, 112.5,
4
15
base workup (62.8-kg scale): the enzyme was removed by
filtration using a centrifuge after the reaction was finished
from an initial set up with 62.8 kg of racemic E/trans-ethyl
bicyclo[3.2.0]hept-6-ylideneacetate in 248 L of acetone and
1
512 L of buffer. The enzyme was washed with n-heptane
(313 L), and the pH of the filtrate was adjusted to pH 2
trosyf3gpphsi19 (see refs 6-7); 256 t
collected using 128 acquisitions per increment, and 4096 t
data points. Prior to Fourier transformation, each dimension
was apodized by a 90 degree shifted sine-bell function and
zero-filled in t
Data processing and acquisition were performed using the
TOPSPIN 1.3 software package (Bruker-Biospin). Proton
chemical shifts are referenced to internal 3-(methylsilyl)
1
increments were
using concentrated hydrochloric acid (15.5 L). The aqueous
layer was separated and extracted with MTBE (150 L). The
combined organic layers were extracted with 1 N NaOH (121
L) and the layers separated. The organic phase was extracted
a second time with 1 N NaOH (50 L), and the layers were
separated. The combined basic aqueous layers were washed
twice with n-heptane (2 × 54 L) to remove the starting
material. The pH of the aqueous phase was then adjusted to
pH 1-2 using 36% commercial grade concentrated hydro-
chloric acid (30 L) and extracted twice with MTBE (2 ×
2
1
to form a final matrix size of 2048 × 512.
propane-1,1,2,2,3,3,-d
6
-sulfonic acid sodium salt (DSS). The
1
5
N chemical shifts are referenced indirectly to DSS using
the absolute frequency ratios.
155 L). The combined MTBE extracts were then washed
with water (100 L) and concentrated in vacuo to give the
crude product (17.2 kg) raw as a pale-yellow oil, which can
be further purified as a white powder by crystallization in
ethanol-water (2.2:3.4 v/v) with a melting point of 68 °C
by DSC.
Acknowledgment
We thank Carlos Martinez, Jean Xie, and Matthew Bernett
for helpful discussions and Dr. K. Fabian and Lanxess for
help in process developing and scaling up.
15
N -CAL-B Cloning, Expression, and Purification. The
genomic DNA was isolated from Candida antarctica (ATCC
Received for review February 20, 2006.
OP0600409
32657) and the CAL-B gene amplified by PCR (polymerase
660
•
Vol. 10, No. 3, 2006 / Organic Process Research & Development